<%BANNER%>

Using Stable Isotopes to Characterize Ornamental Aquaculture Pond Trophic Dynamics and Maximize Fish Production

Permanent Link: http://ufdc.ufl.edu/UFE0041253/00001

Material Information

Title: Using Stable Isotopes to Characterize Ornamental Aquaculture Pond Trophic Dynamics and Maximize Fish Production
Physical Description: 1 online resource (700 p.)
Language: english
Creator: KAO,JON
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2011

Subjects

Subjects / Keywords: AQUACULTURE -- ISOTOPE -- MANAGEMENT -- NUTRITION -- ORNAMENTAL -- POND -- PRODUCTION -- STABLE -- SWORDTAIL -- TROPHIC -- TROPICAL -- XIPHOPHORUS
Forest Resources and Conservation -- Dissertations, Academic -- UF
Genre: Fisheries and Aquatic Sciences thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: USING STABLE ISOTOPES TO CHARACTERIZE ORNAMENTAL AQUACULTURE POND TROPHIC DYNAMICS AND MAXIMIZE FISH PRODUCTION Fish production in pond aquaculture systems is highly variable among individual ponds; an obvious area of improvement would consist of maintaining uniformly high levels of production in ponds through the proper understanding and manipulation of pond trophic dynamics. One method used for understanding the flows of energy and matter within ponds is by tracing and quantifying the flow of energy and matter via the movement of stable isotopes, from allochthonous feed inputs, and live foods derived from allochthonous fertilizer inputs and autochthonous energy and matter sources within the pond's trophic system. Using the common swordtail (Xiphophorus hellerii) live-bearer tropical fish as a model animal, indoor feeding trials were successfully used to validate carbon and nitrogen isotope tracing methods, and obtain mean carbon (delta^13^C) and nitrogen (delta^15^N) isotope enrichment magnitudes for use in outdoor pond trials. Four nutrient types (commercial processed feed, commercial unprocessed feed, organic cottonseed meal fertilizer, and inorganic liquid fertilizer) that are widely used within the aquaculture industry, were applied to 24 ponds (6 ponds per treatment) evenly stocked with swordtail broodstock. At harvest, fish biomass production was greatest within the two feed treatments after the 12-week trial. Inorganic fertilizer produced the lowest overall fish biomass, and least number of marketable swordtails (standard length > 31 mm). Stable isotope analyses revealed that distinct nutrient movement patterns were present among pond treatments: (1) nutrients were moving in parallel through feed pond food webs, moving directly into fish and plankton (to varying degrees), (2) nutrients were moving in serial through inorganic fertilizer pond food webs: fertilizer ? phytoplankton ? small zooplankton ? macrozooplankton ? swordtails. The organic cottonseed fertilizer treatment was a special case, cottonseed meal performed as both a fertilizer and a directly ingestible high protein food source, trophic utilization and nutrient movement results from this treatment, were intermediate between parallel and serial nutrient movement observed within the other three treatments. Large plankton (> 200 ?m) assemblages were analyzed for taxa and abundance, and no qualitative assemblage differences were found that could be associated with high or low fish production performance. Temporal differences in plankton assemblage compositions and abundances were clearly present, likely due to plankton community succession resulting from competition and predation, but assemblages did not largely differ among pond nutrient treatment type. The generalist food habits of swordtails imply that swordtails are capable of efficiently preying upon large macrozooplankton, but due to the pharyngeal jaw arrangement within the swordtail buccal cavity, gut content analysis was not a viable method of determining swordtail diet or predation effects. Twenty-four hour captive feeding trials demonstrated that swordtails were capable of reducing large plankton biomass and altering large plankton assemblage composition. Swordtails preferentially predated upon Moina daphnids relative to calanoid copepods in 24-hour incubation trials.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by JON KAO.
Thesis: Thesis (Ph.D.)--University of Florida, 2011.
Local: Adviser: Cichra, Charles E.
Local: Co-adviser: Yanong, Roy P.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2012-04-30

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2011
System ID: UFE0041253:00001

Permanent Link: http://ufdc.ufl.edu/UFE0041253/00001

Material Information

Title: Using Stable Isotopes to Characterize Ornamental Aquaculture Pond Trophic Dynamics and Maximize Fish Production
Physical Description: 1 online resource (700 p.)
Language: english
Creator: KAO,JON
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2011

Subjects

Subjects / Keywords: AQUACULTURE -- ISOTOPE -- MANAGEMENT -- NUTRITION -- ORNAMENTAL -- POND -- PRODUCTION -- STABLE -- SWORDTAIL -- TROPHIC -- TROPICAL -- XIPHOPHORUS
Forest Resources and Conservation -- Dissertations, Academic -- UF
Genre: Fisheries and Aquatic Sciences thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: USING STABLE ISOTOPES TO CHARACTERIZE ORNAMENTAL AQUACULTURE POND TROPHIC DYNAMICS AND MAXIMIZE FISH PRODUCTION Fish production in pond aquaculture systems is highly variable among individual ponds; an obvious area of improvement would consist of maintaining uniformly high levels of production in ponds through the proper understanding and manipulation of pond trophic dynamics. One method used for understanding the flows of energy and matter within ponds is by tracing and quantifying the flow of energy and matter via the movement of stable isotopes, from allochthonous feed inputs, and live foods derived from allochthonous fertilizer inputs and autochthonous energy and matter sources within the pond's trophic system. Using the common swordtail (Xiphophorus hellerii) live-bearer tropical fish as a model animal, indoor feeding trials were successfully used to validate carbon and nitrogen isotope tracing methods, and obtain mean carbon (delta^13^C) and nitrogen (delta^15^N) isotope enrichment magnitudes for use in outdoor pond trials. Four nutrient types (commercial processed feed, commercial unprocessed feed, organic cottonseed meal fertilizer, and inorganic liquid fertilizer) that are widely used within the aquaculture industry, were applied to 24 ponds (6 ponds per treatment) evenly stocked with swordtail broodstock. At harvest, fish biomass production was greatest within the two feed treatments after the 12-week trial. Inorganic fertilizer produced the lowest overall fish biomass, and least number of marketable swordtails (standard length > 31 mm). Stable isotope analyses revealed that distinct nutrient movement patterns were present among pond treatments: (1) nutrients were moving in parallel through feed pond food webs, moving directly into fish and plankton (to varying degrees), (2) nutrients were moving in serial through inorganic fertilizer pond food webs: fertilizer ? phytoplankton ? small zooplankton ? macrozooplankton ? swordtails. The organic cottonseed fertilizer treatment was a special case, cottonseed meal performed as both a fertilizer and a directly ingestible high protein food source, trophic utilization and nutrient movement results from this treatment, were intermediate between parallel and serial nutrient movement observed within the other three treatments. Large plankton (> 200 ?m) assemblages were analyzed for taxa and abundance, and no qualitative assemblage differences were found that could be associated with high or low fish production performance. Temporal differences in plankton assemblage compositions and abundances were clearly present, likely due to plankton community succession resulting from competition and predation, but assemblages did not largely differ among pond nutrient treatment type. The generalist food habits of swordtails imply that swordtails are capable of efficiently preying upon large macrozooplankton, but due to the pharyngeal jaw arrangement within the swordtail buccal cavity, gut content analysis was not a viable method of determining swordtail diet or predation effects. Twenty-four hour captive feeding trials demonstrated that swordtails were capable of reducing large plankton biomass and altering large plankton assemblage composition. Swordtails preferentially predated upon Moina daphnids relative to calanoid copepods in 24-hour incubation trials.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by JON KAO.
Thesis: Thesis (Ph.D.)--University of Florida, 2011.
Local: Adviser: Cichra, Charles E.
Local: Co-adviser: Yanong, Roy P.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2012-04-30

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2011
System ID: UFE0041253:00001


This item has the following downloads:


Full Text

PAGE 1

USING STABLE I SO TOPES TO CHARACTERIZE ORNAMENTAL AQUACULTURE POND TROPHIC DYNAMICS AND MAXIMIZE FISH PRODUCTION By JON S KAO A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2011

PAGE 2

2 2011 Jon S. Kao

PAGE 3

3 To Shari, my sister Jaylynn, and my parents, without whom, this would not have been possible

PAGE 4

4 ACKNOWLEDGMENTS T hanks go out to committee chairs Drs. Charles E. Cichra and Roy P.E. Yanong, and committee members Drs. Thomas K. Frazer, Richard D. Miles and Joseph J. Delfino, for whose patience and understanding I will forever be in their gratitude. S pecial thanks go out to Craig A. Watson the director of the University of Floridas Tropical Aquaculture Laboratory in Ruskin, Florida, who generously provided funds and facilities that made this work possible. Additional thanks to the staff of the Fisheries Program at the University of Florida and staff and interns at the Tropical Aquaculture Laboratory who are too numerous to name in their entirety: Dr. Jeff Hill, Sonya Sampson, Ryan Schelb, Christopher Wilkinson, Tyler Pavlowich, Sherry Giardini, Mary Cichra, Carlos Martinez, Michael K rasilovsky, Scott Graves, Debra Pouder, Chris Daniels, and Beth Privett, and Julie Anne and Ricardo Russo. Thanks go to my good friends Micah Alo, Richard Kline and Chris Tilghman who were always there to pitch in, bounce ideas off, or share a few beers. I especially thank my late parents Ruth and ShihChien Dick Kao whose love, affection and support will always be with me Thanks go to my sister Jaylynn Milstein and her husband Michael for their love and generous financial assistance, and to my brother Jeff, whose eyes still glaze over at the mention of technical aspects of my research. Lastly, I thank Shari Hanson, whose love and support are beyond the measure of any scientific metric.

PAGE 5

5 TABLE OF CONTENTS P age ACKNOWLEDGMENTS ...............................................................................................................4 TABLE OF CONTENTS .................................................................................................................5 LIST OF TABLES .........................................................................................................................10 LIST OF FIGURES .......................................................................................................................12 LIST OF ABBREVIATIONS ........................................................................................................36 ABSTRACT ...................................................................................................................................37 CHAPTER 1 INTRODUCTION ..................................................................................................................39 Background .............................................................................................................................39 Tropical Fish Pond Aquaculture ......................................................................................39 Aq uaculture Pond Trophic Ecology ................................................................................40 Aquaculture Pond Applied Nutrient Management Practices ...........................................42 Trophic Dynamics Investigations Using Stable Isotopes ................................................45 2 INDOOR SWORDTAIL FEEDING AND ISOTOPE METHODOLOGY VALIDATION TRIAL ...........................................................................................................47 Materials and Methods ...........................................................................................................48 Experimental Feed Composition and Isotopic Signature Profiles ...................................48 Broodstock and Fry Production .......................................................................................49 Manufactured Feeds ........................................................................................................51 Experimental Design .......................................................................................................51 Freeze Drying and Sample Preparation ...........................................................................53 Isotopic Signature ............................................................................................................54 Elemental Carbon to Nitrogen Ratio ...............................................................................55 Statistical Analysis ..........................................................................................................55 Fry Growth and Survival ..........................................................................................55 Isotopic Analyses .....................................................................................................56 Results .....................................................................................................................................56 Fry Growth ......................................................................................................................56 Fry Survival Rates ...........................................................................................................57 Feed Isotopic Signatures ..................................................................................................57 Swordtail Isotopic Signatures ..........................................................................................57 Elemental Carbon to Nitrogen Ratio Analysis ................................................................59 Discussion ...............................................................................................................................60 Fry Growth ......................................................................................................................60

PAGE 6

6 Fry Survival Rates ...........................................................................................................61 Indoor Trial Fry Isotope Signature ..................................................................................61 Elemental Carbon to Nitrogen Ratios ..............................................................................63 3 OUTDOOR POND TRIAL SWORDTAIL BIOMASS PRODUCTION ..............................84 Methods ..................................................................................................................................85 Study Site .........................................................................................................................85 Water Chemistry and Physical Environment ..................................................................87 Pre Harvest Fry Collection ..............................................................................................88 Harvest Fish Collection ...................................................................................................89 Statistical Analyses ..........................................................................................................89 Results .....................................................................................................................................90 Pre harvest Fry Growth ...................................................................................................90 Fish Production at Harvest ..............................................................................................90 Discussion ...............................................................................................................................92 Pre harvest Fry Growth ...................................................................................................92 Harvest Fish .....................................................................................................................94 Water Chemistry and Physical Environment ..................................................................96 4 OUTDOOR POND APPLIED NUTRIENT ISOTOPE TRACING TRIAL .......................113 Methods ................................................................................................................................117 Plankton Sample Collection ..........................................................................................117 Isotopic Analyses ...........................................................................................................119 Time Averaged Plankton Assemblage Isotopic Analyses ............................................120 Trophic Group Isotopic Signature Differences .............................................................120 Trophic Level Determination ........................................................................................121 Isot opic Determination of Consumer Diet ....................................................................125 Results ...................................................................................................................................128 Pond Nutrient Isotope Signatures ..................................................................................128 Pre harvest Fry Isotopic Signatures ...............................................................................129 Harvest Fish Isotope Signature Differences ..................................................................131 Large Zooplankton Isotope Signatures ..........................................................................133 Medium Plankton Isotope Signatures ............................................................................138 Small Plankton Isotope Signatures ................................................................................141 Pond Nutrient and Pond Fish Isotope Signature Differences ........................................144 Pond Fish and Large Zooplankton Isotope Signature Differences ................................146 Large Zooplankton and Pond Nutrient Isotope Signature Differences .........................148 Small Plankton and Pond Nutrient Isotope Signature Differences ...............................149 Isotopically Derived Large Harvest Fish Diet ...............................................................150 Isotopically Derived Small Harvest Fish Diet ...............................................................151 Isotopically Derived Pre Harvest Fry Diet ....................................................................152 Discussion .............................................................................................................................153 Pond Nutrient Carbon Isotope Signatures .....................................................................153 Pond Nutrient Nitrogen Isotope Signatures ...................................................................154 Pre harvest Fry Isotope Signatures ................................................................................155

PAGE 7

7 Pre Harvest fry Carbon Isotope Signatures ...................................................................156 Pre Harvest Fry Nitrogen Isotope Signatures ................................................................157 Large Harvest Fish Carbon Isotope Signature Differences ...........................................158 Large Harvest Fish Nitrogen Isotope Differences .........................................................161 Small Harvest Fish Carbon Isotope Signatures .............................................................164 Small Harvest Fish Nitrogen Isotope Signatures ...........................................................164 Harvest Fish Size Class Isotope Signature Differences ................................................167 Large Zooplankton Carbon Isotope Differences ...........................................................167 Time Averaged Large Zooplankton Carbon Isotope Signatures ...................................171 Large Zooplankton Nitrogen Isotope Signature Differences ........................................175 Time Averaged Large Zooplankton Nitrogen Isotope Differences ..............................176 Medium Plankton Assemblage Isotopic Signatures ......................................................178 Time Averaged Medium Plankton Carbon Isotope Differences ...................................180 Medium Plankton Assemblage Nitrogen Isotope Signatures ........................................181 Time Averaged Medium Plankton Nitrogen Isotope Differences ................................184 Small Plankton Carbon Isotope Differences .................................................................184 Time Averaged Small Plankton Carbon Isotope Differences .......................................185 Small Plankton Assemblage Nitrogen Isotope Signature Differences ..........................186 Time Averaged Small Plankton Nitrogen Isotope Differences .....................................188 Nutrient Movement within Fish and Plankton Size Assemblages ................................188 Pond Nutrient and Harvest Fish Carbon Isot ope Signature Differences .......................189 Large Harvest Fish and Applied Nutrient Nitrogen Isotope Differences ......................194 Small Harvest Fish and Pond Nutrient Isotopic Differences .........................................197 Small Harvest Fish and Applied Nutrient Nitrogen Isotope Differences ......................202 Diet Estimates Using Both Carbon and Nitrogen Isotope Signatures ...........................204 Large Zooplankton and Harvest Fish Isotope Differences ............................................207 Large Harvest Fish and Large Zooplankton Assemblage Nitrogen Isotope Differences .................................................................................................................213 Small Harvest Fish and Large Zooplankton Isotope Differences .................................216 Large Zooplankton and Pond Nutrient Isotopic Differences ........................................219 Small Plankton and Nutrient Isotope Differences .........................................................222 Small Plankton Assemblage Applied Nutrient Nitrogen Isotope Differences ..............226 5 LARGE ZOOPLANKTON ASSEMBLAGE TAXA IDENTIFICATION AND ENUMERATION .................................................................................................................296 Methods ................................................................................................................................298 Plankton Sample Collection ..........................................................................................298 Large Plankton Assemblage Examination and Processing ...........................................298 Large Plankton Assemblage Taxa Identification ..........................................................299 Large Plankton Community Analyses ...........................................................................300 Percent Number, Volume, and Frequency of Occurrence Graphs ................................303 Major Large Plankton Assemblage Taxa Analyses .......................................................305 Large Plankton Assemblage Similarity Indices ............................................................305 Statistical Analyses ........................................................................................................307 Results ...................................................................................................................................308 Large Zooplankton Produc tion Analyses ......................................................................308

PAGE 8

8 Time Averaged Large Plankton Volumes .....................................................................309 Time Averaged Large Plankton Densities ....................................................................310 Outdoor Pond Trial Major Large Plankton Groups .......................................................310 Copepod Densities for 10 Weekly Sampling Periods ...................................................310 Large Plankton Assemblage Simplified Morisitas an d Percent Similarity Indices .....314 Deterministic Morisitas Similarity and Percent Similarity Indices ..............................314 Bootstrap Morisitas Similarity and Percent Similarity Indices ....................................318 Discussion .............................................................................................................................324 Large Zooplankton Assemblage Production Volumes ..................................................324 Large Zooplankton Assemblage Production Numbers ..................................................326 Outdoor Pond Trial Major Large Zooplankton Groups ................................................329 Copepod Density for 10 Weekly Sampling Periods ......................................................330 Rotifer Density for 10 Weekly Sampling Periods .........................................................330 Daphnid Density for 10 Weekly Sampling Periods ......................................................331 Pond Large Zooplankton Assemblage %IRI Trends .....................................................331 Deterministic Similarity Indices ....................................................................................333 6 24HOUR SWORDTAIL CAPTIVE FEEDING EXPERIMENTS .....................................414 Methods ................................................................................................................................415 24Hour Captive Feeding/ Plankton Grazing Experiment .............................................415 Descriptive Parameters, Indices and Statistical Analyses .............................................417 Results ...................................................................................................................................418 24Hour Feeding Trial Large Plankton Volume Differences ........................................418 24Hour Feeding Trial Large Plankton Densities .........................................................420 24Hour Feeding Trial Major Zooplankton Taxa Differences ......................................422 24Hour Feeding Trial Major Taxa %IRI .....................................................................427 24Hour Feeding Trial Similarity Indices .....................................................................428 Bootstrap Large Plankton Assemblage Similarity Indices ............................................429 Discussion .............................................................................................................................432 Large Zooplankton Assemblages ..................................................................................432 24Hour Captive Feeding Trials ....................................................................................433 Major Taxa Densities among 24 Hour Trial Treatments ..............................................435 24Hour Trial Community %IRI Values .......................................................................439 24Hour Trial Simplified Morisitas and Percent Similarity Indices ............................440 Bootstrap Initial, NF, and Fish MSI and PSI Estimates ................................................440 7 SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS .......................................515 Indoor Fry Growth and Isotopic Tracing Validation Trial ............................................515 Indoor Fry Growth .........................................................................................................515 Initial Fry Isotope Signature ..........................................................................................517 Fry Growth and Survival Rates .....................................................................................518 Indoor Trial Experimental Design Improvements ........................................................519 Outdoor Pond Production Trial ............................................................................................521 Pre harvest Fry Growth .................................................................................................521 Fish at Harvest ...............................................................................................................522

PAGE 9

9 Water Chemistry and Pond Physical Environment .......................................................524 Experimental Design Considerations and Further Research .........................................524 Isotopic Investigation of Pond Trophic Dynamics ...............................................................526 Pond Nutrient Isotope Signatures ..................................................................................526 Pre Harvest Fish Isotopic Signatures ............................................................................531 Harvest Fish Isotopic Signatures ...................................................................................533 Small Harvest Fish Isotope Differences ........................................................................537 Swordtail Ontogenetic Dietary Differences ..................................................................537 Plankton Isotope Signatures ..........................................................................................540 Harv est Fish and Applied Nutrient Isotope Differences ...............................................542 Isotopically Derived Fish Diet Scenario Issues .............................................................545 Plankton Assemblage Trophic Identity Issues ..............................................................546 Juvenile Swordtail Live Food Dependence ...................................................................548 Applied Nutrient Isotopic Signature Influences upon Pond Taxa .................................550 Biases Using the ED and IS Dietary Estimate Methods ................................................551 Small Harvest Fish ED and IS Diet Estimation Issues ..................................................553 Small Harvest Fish Dietary Estimate Issues ..................................................................555 Plankton Size Assemblage and Trophic Position/Isotopic Signature Issues .................555 Small Plankton and Nutrient Isotope Signature Differences .........................................558 Plankton Assemblage Experimental Artifact and Model Species Issues ......................564 Large Zooplankton Assemblage Taxa and Abundances ......................................................566 Large Zooplankton Densities ........................................................................................566 Deterministic Large Plankton Assemblage Similarity Indices ......................................570 Bootstrap Large Plankton Assemblage Similarity Indices ............................................571 24Hour Large Zooplankton Assemblage Captive Feeding Trials .......................................575 Large Plankton Assemblage Biomass ...........................................................................575 APPENDIX A LARGE PLANKTON ASSEMBLAGE %N, %V, AND %FO ...........................................582 B BOOTSTRAPPING SIMILARITY INDEX PROGRAM ...................................................663 LIST OF REFERENCES .............................................................................................................682 BIOGRAPHICAL SKETCH .......................................................................................................700

PAGE 10

10 LIST OF TABLES Table P age 21 Twelve week indoor swordtail ( Xiphophorus hellerii ) fry feeding trial daily per capita fry ration level as a function of time .......................................................................83 41 Dietary estimates of harvest fish derived from isotopic results from IsoSource linear mixing model computer program ( 95% CI) and Euclidean distance measurements ....289 42 Dietary estimates of preharvest fish derived from isotopic results from IsoSource linear mixing model computer program ( 95 % CI) and Euclidean distance measurements ...................................................................................................................290 43 Back calculated primary producer carbon isotope signature ( 13C) estimates based upon actual (*) and hypothetical pond guild carbon signature results and trophic level estimations ...............................................................................................................291 44 Isotopically derived trophic differences among harvest fish predators, applied nutrients and plankton size assemblages (via 15N) ........................................................292 45 Processed feed (PRO) pond treatment dietary components ranked in order of percentage importance (1 highest, 4 lowest) for the Euclidean distance and IsoSource linear algebra computer program. ....................................................................................293 46 Unprocessed feed (UNP) pond treatment dietary components ranked in order of percentage imp ortance (1 highest, 4 lowest) for the Euclidean distance and IsoSource linear algebra computer program. ....................................................................................293 47 Cottonseed meal (CSM) pond treatment dietary components ranked in order of percentage importance (1 highest, 4 lowest) for the Euclidean distance and IsoSource linear algebra computer program. ....................................................................................293 48 INO pond treatment dietary components ranked in order of percentage importance (1 highest, 4 lowest) for the Euclidean distance and IsoSource linear algebra computer program. ...........................................................................................................................293 49 Food source inclusion rates for large zooplankton derived from isotopic results, IsoSource program, linear mixing models, and Euclidean distance measurements. .......294 410 Trophic level estimates of plankton assemblages and harvest fish within treatments, trophic levels relative to treatment small plankton assemblage 15N values only. ..........295 51 30 most common large zooplankton assemblage components and their approximate volumes calculated via photomicrographs. ......................................................................412

PAGE 11

11 52 Frequency of similar large plankton assemblages between treatment pairs according to the simplified Morisitas similarity index (MSI) and percent similarity index (PSI) using deterministic (%IRI) and resampling (%N) methods. ............................................413

PAGE 12

12 LIST OF FIGURES Figure P age 21 Greenhouse physical layout, location of concrete burial vaults and life support system for swordtail fry production and broodstock growout. ..........................................65 22 Epoxy coated concrete burial vault used to culture swordtail fry and broodstock in greenhouse .........................................................................................................................65 23 Cage housing broodstock to minimize fry cannibalism .....................................................66 24 Concrete vault freshwater recirculation/partial flow through swordtail fry production and growout system in greenhouse ....................................................................................66 25 Five isonitrogenous and isocalorific experimental flake feeds ..........................................67 26 Physical layout and experiment al design of the indoor feeding trial study. ......................68 27 Individual replicate aquarium (approx. 6.4 L) for indoor feeding and isotopic me thodology validation trial. .............................................................................................69 28 Indoor fry feeding and isotope methodology validation trial aquaria batteries .................69 29 Virtis Sentry freeze dryer. ................................................................................................70 210 Swordtails after 48 hrs in freeze dryer ...............................................................................70 211 Desiccator used to store samples prior to weighing for dry weight and isotopic analysis sample preparation ...............................................................................................71 212 Mean swordtail fry (10 subsample fry / replicate aquaria) weight of randomly selected single replicate aquaria from each of five experimental feed treatments ............72 213 Mean swordtail fry weight ( 95 % CI) of five experimental feed treatments after 90 days ....................................................................................................................................73 214 Swordtail fry survival rates for five feed treatments after 12 weeks .................................74 215 Isotopic signatures ( 95 % CI ) of swordtail broodstock and initial fry for indoor feeding trial ........................................................................................................................75 216 Six flake feed 13C (absolute value) and 15N signature values ( 95 % CI) ....................76 217 Mean pre harvest swordtail fry isotope signature [ 13C (abs), 15N] trajectories over 90 day indoor trial ..............................................................................................................77 218 Indoor trial fry isotope signatures [ 13C(abs), 15N 95 % CI] at harvest (90 days) ...................................................................................................................................78

PAGE 13

13 219 Carbon isotope signature differences ( 13C 95 % CI) between harvest fry and respective experimental feed treatments after 90 days ......................................................79 220 Nitrogen isotope signature differences ( 15N 95 % CI) between harvest fry and respective experimental feed treatments after 90 days ......................................................80 221 Cartesian plot of indoor feeding trial swordtail juveniles (3 months), initial fry, male (BSM) and female broodstock (BSF), and five experimental feeds ..................................81 222 Indoor fry elemental [C]/[N] ratios ( 95 % CI) at harvest after 90 days on five experimental feeds .............................................................................................................82 31 Clockwise from top left: (a) drained and washed pond, hydrated lime applied to substrate visible as white powder, (b) diesel tractor engine pump used to drain ponds in background, smaller wash pump in foreground.............................................................97 32 Clockwise from top: (a) adult female and male redwag swordtails ( Xiphophorus hellerii ) held within large polyethylene tubs (~ 280 L) within a large greenhouse ...........98 33 Pond harvest equipment .....................................................................................................99 34 Pre harvest fry mean dry weights, periodically sampled over 12 week pond trial ..........100 35 Processed feed (PRO) treatment mean fry dry weight ( 95% CI) sampled over 12 week trial ..........................................................................................................................101 36 Unprocessed feed (UNP) treatment mean fry dry weights ( 95% CI) sampled over 12 week trial .....................................................................................................................102 37 Cottonseed meal (CSM) treatment fry mean dry weights ( 95% CI) sampled over 12 week trial ..........................................................................................................................103 38 Inorganic fertilizer (INO) tre atment mean fry dry weights ( 95% CI) sampled over 12 week trial .....................................................................................................................104 39 Mean fish biomass (g 95% CI) at harvest .....................................................................105 310 Mean fish numbers (#s 95% CI) at harvest .................................................................106 311 Mean marketable (> 31 mm SL) fish numbers ( 95% CI) at harvest ...........................107 312 Mean unmarketable ( ........................108 313 Mean individual marketable (> 31 mm SL) fish weights ( 9 5 % CI) at harvest ............109 314 Mean individual unmarketable ( .......110 315 Large harvest fish (SL > 31 mm) individual fish wet weight coefficient of variation (sd/mean) values among applied pond nutrient treatments ..............................................111

PAGE 14

14 316 Air temperature daily maximum and minimum (C) from 1 March to 30 June 2006 for Ruskin, Florida ...........................................................................................................112 41 Average large harvest fish (swordtail) biomass ( 95 % CI) as a function of nitrogen isotope signature differences (semi log plot) between large harve st fish and applied nutrient treatments ...........................................................................................................237 42 Integrated depth pond water column sampler. .................................................................238 43 Clockwise from top left: (a) freeze dried large zooplankton size assemblage on Whatman GF/C pre combusted glass filter disc ..........................................................239 44 Mean carbon isotope signatures ( 13C 95 % CI) differences among pond nutrient treatments ...........................................................................................................240 45 Mean nitrogen isotope signatures ( 15N 95 % CI) of applied pond nutrient treatments .........................................................................................................................241 46 Average pre harvest pond swordtail fry carbon (absolute value) and nitrogen isotope signatures ( 13C, 15N) for three weekly sampling dates among four nutrient treatments .........................................................................................................................242 47 Average processed feed (PRO) treatment preharvest swordtail fry carbon and nitrogen isotope signatures [ 13C (abs. value), 15N 95 % CI] for three weekly sampling dates ..................................................................................................................243 48 Average unprocessed feed (UNP) pond treatment pre harvest swordtail fry ca rbon and nitrogen isotope signatures [ 13C (abs. value), 15N 95 % CI] for three weekly sampling dates .....................................................................................................244 49 Average cottonseed meal (CSM) pond treatment preharvest swordtail fry carbon and nitrogen isotope signatures [ 13C (abs. value), 15N 95 % CI] for thre e weekly sampling dates .....................................................................................................245 410 Average inorganic fertilizer (INO) treatment pre harvest swordtail fry carbon and nitrogen isotope signatures [ 13C (abs. value), 15N 95 % CI] for three weekly sampling dates ..................................................................................................................246 411 Time averaged pond harvest large swordtail (> 31 mm SL) carbon and nitrogen isotope signatures [ 13C (abs. value), 15N 95 % CI] among four nutrient treatments .........................................................................................................................247 412 Average pond harvest small fish (SL signatures [ 13C (abs. valve), 15N 95 % CI] among four pond nutrient treatments .........................................................................................................................248

PAGE 15

15 413 Pond trial large zooplankton (> 200 m) carbon [ 13C (abs. value)] and nitrogen ( 15N) isotope signatures for 11 weekly sampling periods among four pond nutrient treatments .........................................................................................................................249 414 Processed feed (PRO) treatment large zooplankton assemblage carbon and nitrogen isotope signatures [ 13C (abs. value), 15N 95 % CI] among 11 weekly sampling periods ..............................................................................................................................250 415 Unprocessed feed (UNP) treatment large zooplankton assemblage carbon and nitrogen isotope signatures [ 13C (abs. value), 15N 95 % CI] among 11 weekly sampling periods ..............................................................................................................251 416 Cottonseed meal (CSM) treatment large zooplankton (> 200 m) assemblage carbon and nitrogen isotope signatures [ 13C (abs. value), 15N 95 % CI] among 11 weekly sampling periods ..................................................................................................252 417 Inorganic fertilizer treatment large zooplankton (> 200 m) assemblage carbon and nitrogen isotope signatures [ 13C (ab s. value), 15N 95 % CI] among 11 weekly sampling periods ..............................................................................................................253 418 Time averaged large zooplankton (> 200 m) carbon and nitrogen isotope signatures [ 13C (absolute value), 15N 95 % CI] among four pond nutrient treatm ents .........254 419 Processed feed (PRO) medium zooplankton (32 200 m) assemblage carbon and nitrogen isotope signatures [ 13C (absolute value), 15N 95 % CI] among 12 weekly sampling periods ..................................................................................................255 420 Unprocessed (UNP) feed medium plankton assemblage c arbon and nitrogen isotope signatures [ 13C (absolute value), 15N 95 % CI] among 12 weekly sampling periods ..............................................................................................................................256 421 Cottonsee d meal (CSM) medium plankton assemblage carbon and nitrogen isotope signatures [ 13C (absolute value), 15N 95% CI] a mong 12 weekly sampling periods ..............................................................................................................................257 422 Inorganic fertilizer (INO) medium plankton (32 200 m) assemblage carbon and nitrogen isotope signatures [ 13C (absolute value), 15N 95 % CI] among 12 weekly sampling periods ..................................................................................................258 423 Time averaged medium plankton (32 200 m) carbon and nitrogen isotope signatures [ 13C (absolute value), 15N 95 % CI] among 12 weekly sampling periods ..............................................................................................................................259 424 Processed feed (PRO) small plankton (1 32 m) ass emblage carbon and nitrogen isotope signatures [ 13C (absolute value), 15N 95 % CI] among 12 weekly sampling periods ..............................................................................................................260

PAGE 16

16 425 Unprocessed feed (UNP) small plankton (1 32 m) assemblage carbon and nitrogen isotope signatures [ 13C (absolute value), 15N 95 % CI] among 12 week ly sampling periods ..................................................................................................261 426 Cottonseed meal (CSM) small plankton (1 32 m) assemblage carbon and nitrogen isotope signatures [ 13C (absolute value), 15N 95 % CI] among 12 weekly sampling periods ..............................................................................................................262 427 Inorganic fertilizer (INO) small plankton (1 32 m) assemblage carbon and nitrogen isotope signatures [ 13C (absolute value), 15N 95 % CI] among 12 weekly sampling periods ..................................................................................................263 428 Time averaged small plankton (1 32 m) carbon and nitrogen isotope signatures [ 13C (absolute value), 15N 95 % CI] among 12 weekly sampling periods ...........264 429 Average pond trial large (> 31 mm SL) swordtail and nutrient source isotope signature difference magnitudes ( 13C, 15N 95% CI) at harvest .......................265 430 Average pond trial small ( 1 mm SL) swordtail and nutrient source isotope signature difference magnitudes ( 13C, 15N 95 % CI) at harvest ......................266 431 Average pond trial large (> 31 mm SL) swordtail and large zooplankton carbon isotope signature difference magnitudes (sign neutral) at harvest (13C 9 5 % CI) ....................................................................................................................................267 432 Average pond trial large (> 31 mm SL) swordtail and large zooplankton nitrogen isotope difference magnitudes (sign neutral) at harvest (15N 95 % CI) ..............268 433 Average pond trial small ( isotope difference magnitudes (sign neutral) at harvest (13C 95 % CI) ..............269 434 Average pond trial small ( isotope difference magnitudes (sign neutral) a t harvest ( 15N 95 % CI) ..............270 435 Average pond trial large zooplankton and applied nutrient carbon isotope signature (sign neutral) difference magnitudes ( 13C 95 % CI) ...........................................271 436 Average pond trial large zooplankton and applied nutrient nitrogen isotope signature (sign neutral) difference magnitudes ( 15N 95 % CI) ...........................................272 437 Average pond trial small plankton and applied nutrient carbon isotope signature (sign neutral) difference magnitudes ( 13C 95 % CI) ....................................................273 438 Average pond trial small plankton and applied nutrient nitrogen isotope signature (sign neutral) difference magnitudes ( 15N 95 % CI) .......................................... 274

PAGE 17

17 439 Cartesian plot of processed feed (PRO) treatment harvest swordtail, processed feed, and three plankton size assemblage 13C (absolute value) and 15N value s ( 95 % CI) ....................................................................................................................................275 440 Cartesian plot of unprocessed feed (UNP) treatment harvest swordtail, processed feed, and three plankton size assemblage 13C (absolute value) and 15N values ( 95 % CI) ...........................................................................................................................276 441 Cartesian plot of cottonseed meal (CSM) treatment harvest swordtail, cottonseed meal, and three plankton size assemblage 13C (absolute value) and 15N values ( 95 % CI) ........................................................................................................................277 442 Cartesian plot of inorganic fertilizer (INO) treatment harvest swordtail, and three plankton size assemblage 13C (absolute value) and 15N values ( 95 % CI) ..........278 443 Isotopic variation among five pond groups (small, medium, and large plankton assemblages, small and large harvest swo rdtails) for four applied pond nutrients ..........279 444 Postulated cottonseed meal (CSM) food web based upon isotopically derived Euclide an distance and IsoSource program swordtail diet estimates ..............................280 445 Postulated inorganic fertilizer (INO) food web based upon isotopically derived Euclidean distance and IsoSource program swordtail diet estimates ..............................281 446 Average total ammonia nitrogen (TAN ) concentrations measured for 12 weekly sampling periods (mg/L 95 % CI) ................................................................................282 447 Processed feed (PRO) treatment average pond chlor ophyll [a] among 12 weekly sampling periods (g/L 95 % CI) .................................................................................283 448 Unprocessed feed (UNP) treatment average pond chlorophyll [a] among 12 weekly sampling periods (g/L 95 % CI) .................................................................................284 449 Cottonseed meal fertilizer (CSM) treatment average pond chlorophyll [a] among 12 weekly sampling periods (g/L 95 % CI) ....................................................................285 450 Inorganic fertilizer (INO) treatment average pond chlorophyll [a] among 12 weekly sampling periods (g/L 95 % CI) .................................................................................286 451 Time averaged mean chlorophyll [a] concentrations (g/L 95% CI) measured over 12 weekly sampling periods among four pond treatments ..............................................287 452 Diagram illustrating potential isotopic collinearity (13C, 15N) of two hypothetical composite diets (diet 1 and 2), and associated dietary prey inclusion rate bias ...............288 51 Large plankton (> 200 m) assemblage mean volumes for 10 weekly sampling periods for four pond nutrient treatments: processed feed (PRO), unprocessed feed (UNP), cotton seed meal (CSM), and inorganic liquid fertilizer (INO) ..........................338

PAGE 18

18 52 Processed feed (PRO) treatment mean large plankton (> 200 m) assemblage volumes (mm3/L 95% CI) for 10 weekly sampling periods .........................................339 53 Unprocessed feed (UNP) treatment large plankton (> 200 m) assemblage mean volumes (mm3/L 95% CI) for 10 weekly sampling periods .........................................340 54 Cottonseed meal (CSM) treatment large plankton (> 200 m) assemblage mean volumes (mm3/L 95% CI) for 10 weekly sampling periods .........................................341 55 Inorganic fertilizer (INO) treatment large plankton (> 200 m) assemblage mean volumes (mm3/L 95 % CI) for 10 weekly sampling periods ........................................342 56 Time averaged large plankton (> 200 m) assemblage mean volumes (mm3/L 95% CI) for four nutrient treatments ........................................................................................343 57 Large plankton (> 200 m) assemblage mean densities (number/L) among 10 weekly sampling periods for four nutrient treatments ..................................................................344 58 Processed feed (PRO) treatment large plankton (> 200 m) assemblage mean densities (number/L 95% CI) among 10 weekly sampling periods ..............................345 59 Unprocessed feed (UNP) treatment large plankton (> 200 m) assemblage mean densities (number/L 95 % CI) among 10 weekly sa mpling periods .............................346 510 Cottonseed meal (CSM) treatment large plankton (> 200 m) assemblage mean densities (number/L 95 % CI) am ong 10 weekly sampling periods .............................347 511 Inorganic fertilizer (INO) treatment large plankton (> 200 m) assemblage mean densities (n umber/L 95 % CI) among 10 weekly sampling periods .............................348 512 Time averaged large plankton (> 200 m) assemblage mean densities (number/L 95% CI) for four nutrient treatments ...............................................................................349 513 Processed feed (PRO) treatment large (> 200 m) plankton assemblage pooled %IRI (index of relative importance) for 10 weekly sampling periods ......................................350 514 Unprocessed feed (UNP) treatment large (> 200 m) plankton assemblage %IRI (index of relative importance) for 10 weekly sampling periods ......................................351 515 Cottonseed meal (CSM) treatment large (> 200 m) plankton assemblage %IRI (index of relative importance) for 10 weekly sampling periods ......................................352 516 Inorganic fertilizer (INO) treatment large (> 200 m) plankton assemblage %IRI (index of relative importance) for 10 weekly sampling periods ......................................353 517 Processed feed (PRO) treatment large (> 200 m) plankton assemblage unpooled %IRI (index of relative importance) for 10 weekly sampling periods ............................354

PAGE 19

19 518 Unprocessed feed (UNP) treatment large (> 200 m) plankton assemblage unpooled %IRI (index of relative importance) for 10 weekly sampling periods ............................355 519 Cottonseed meal (CSM) treatment large (> 200 m) plankton assemblage unpooled %IRI (index of relative importance) for 10 weekly sampling periods ............................356 520 Inorganic fertilizer (INO) treatment large (> 200 m) plankton assemblage unpooled %IRI (index of relative importance) for 10 weekly sampling periods ............................357 521 Processed feed (PRO) treatment major (highest %IRI) large (> 200 m) plankton assemblage taxa [ Diaptomous copepods (LHC), Filinia rotifers (CLD), Moina mac rocopa (DAP; Daphnid)] densities for 10 weekly sampling periods ........................358 522 Processed feed (PRO) treatment Diaptomous copepod (LHC) densities (number/L 95 % CI) for 10 weekly sampling periods, differing letters denote statistical differences between sampling periods .............................................................................359 523 Processed feed (PRO) treatment Filinia rotifer (CLD) densities (number/L 95 % CI) for 10 weekly sampling periods, unshared letters denote statistical differences between sampling periods ................................................................................................360 524 Processed feed (PRO) treatment Moina macrocopa (DAP; Daphnid) densities (number/L 95 % CI) for 10 weekly sampling periods ..................................................361 525 Unprocessed feed (UNP) treatment major (highest %IRI) large plankton assemblage taxa [ Diaptomous copepods (LHC), Filinia rotifers (CLD), Moina macrocopa (DAP; Daphnid)] densities for 10 weekly sampling periods ......................................................362 526 Unprocessed feed (UNP) treatment Diaptomous copepod (LHC) densiti es (number/L 95% CI) for 10 weekly sampling periods, unshared letters denote statistical differences between sampling periods .............................................................................363 527 Unprocessed feed (UNP) treatment Filinia rotifer (CLD) taxa densities (number/L 95% CI) for 10 weekly sampling periods, unshared letters denote statistical differences between sampling periods .............................................................................364 528 Unprocessed feed (UNP) treatment Moina macrocopa (DAP; Daphnid) taxa densities (number/L 95% CI) for 10 weekly sampling periods, unshared letters denote statistical differences between sampling periods .................................................365 529 Cottonseed meal fertilizer (CSM) treatment major (highest %IRI) large (> 200 m) plankton assemblage taxa [ Diaptomous copepods (LHC), Filinia rotifers (CLD), Moina macrocopa (DAP; Daphnid)] densities for 10 weekly sampling periods .............366 530 Cottonseed meal fertilizer (CSM) treatment Diaptomous copepod taxa densities (number/L 95 % CI) for 10 weekly sampling periods ..................................................367

PAGE 20

20 531 Cottonseed meal fertilizer (CSM) treatment Filinia rotifer taxa densities (number/L 95 % CI) for 10 wee kly sampling periods .......................................................................368 532 Cottonseed meal fertilizer (CSM) treatment Moina macrocopa (DAP; Daphnid) taxa densities (number/L 95 % CI) for 10 weekly sampling periods ...................................369 533 Inorganic fertilizer (INO) treatment major (highest %IRI large (> 200 m) plankton assemblage taxa [ Diaptomous copepods (LHC), Filinia rotifers (CLD), Moina macrocopa (DAP; Daphnid)] densities for 10 weekly sampling periods ........................370 534 Inorganic fertilizer (INO) treatment Diaptomous copepod (LHC) taxa densities (number/L 95% CI) for 10 weekly sampling periods ...................................................371 535 Inorganic fertilizer (INO) treatment Filinia rotifer (CLD) taxa densities (number/L 95% CI) for 10 weekly sampling periods ........................................................................372 536 Inorganic fertilizer (INO) treatment Moina macrocopa (DAP; Daphnid) taxa densities (number/L 95% CI) for 10 weekly sampling periods ....................................373 537 Large (> 200 m) plankton assemblage taxa, clockwise from upper right: (a) Calanoid copepod Diaptomous spp., (b) daphnid Moina macrocopa, (c) rotifers mainly Filinia spp. (d) aquatic insect (Ephemer optera or Odonata) ...........................374 537 Large plankton assemblage taxa, clockwise from upper right: (e) spherical rotifer, (f) rotifer Hexarthra spp., (g) rotifer Brachionus spp., (h) unknown ostracod ...............375 538 Large (> 200 m) plankton assemblage treatment pair deterministic MSI (simplified Morisitas similarity index) va lues, generated from taxa pooled %IRI values for 10 weekly sampling periods ..................................................................................................376 539 Processed feed (PRO) and unprocessed feed (PRO) treatment deterministic MSI values for 10 weekly sampling periods ............................................................................377 540 Processed feed (PRO) and cottonseed meal (CSM) fertiliz er treatment deterministic MSI values for 10 weekly sampling periods ....................................................................378 541 Processed feed (PRO) and inorganic (INO) fertilizer treatment deterministic MSI values for 10 weekly sampling periods ............................................................................379 542 Unprocessed feed (UNP) and cottonseed meal (CSM) fertilizer treatment deterministic MSI values for 10 weekly sampling periods ..............................................380 543 Unprocessed feed (UNP) and inorganic fertilizer (INO) treatment deterministic MSI values for 10 weekly sampling periods ............................................................................381 544 Cottonseed meal (CSM) and inorganic (INO) fertilizer treatment deterministic MSI values for 10 weekly sampling periods ............................................................................382

PAGE 21

21 545 Large (> 200 m) plankton assemblage treatment pair deterministic PSI (percent similarity index) values, generated from taxa pooled % IRI values for 10 wee kly sampling periods ..............................................................................................................383 546 Processed (PRO) and unprocessed (UNP) feed treatment deterministic PSI (percent similarity index) values for 10 weekly sampling periods ................................................384 547 Processed feed (PRO) and cottonseed meal fertilizer (CSM) treatment deterministic PSI (percent similarity index) values for 10 weekly sampling periods ...........................385 548 Processed feed (PRO) and inorganic fertilizer (INO) treatment deterministic PSI (percent similarity index) values for 10 weekly sampling periods ..................................386 549 Unprocessed feed (UNP) and cottonse ed meal (CSM) fertilizer treatment deterministic PSI (percent similarity index) values for 10 weekly sampling periods .....387 550 Unprocessed feed (UNP) and inorganic fertilizer (INO) treatment deterministic PSI (Percent Similarity Index) values for 10 weekly sampling periods .................................388 551 Cottonseed meal (CSM) and inorganic fertilizer (INO) treatment deterministic PSI (Percent Similarity Index) values for 10 weekly sampling periods .................................389 552 Processed (PRO) and unprocessed (UNP) feed treatment bootstrap MSI (Simplified Morisitas Index) values ..................................................................................................390 553 Scale adjusted processed (PRO) and unprocessed (UNP) feed treatment bootstrap MSI (simplified Morisitas Index) values ........................................................................391 554 Processed feed (PRO) and inorganic fertilizer (INO) treatment bootstrap PSI (Percent Similarity Index) values .....................................................................................392 555 Processed (PRO) and unprocessed (UNP) feed treatment bootstrap PSI (Percent Similarity Index) values ...................................................................................................393 556 Processed (PRO) and unprocessed (UNP) feed treatment bootstrap MSI (simplified Morisitas similarity index ...............................................................................................394 557 Processed feed (PRO) and cottonseed meal fertilizer (CSM) treatment bootstrap MSI (simplified Morisitas similarity index ............................................................................395 558 Processed feed (PRO) and inorganic fertilizer (INO) treatment bootstrap MSI (simplified Morisitas similarity index; critical value CH ........................396 559 Unprocessed feed (UNP) and cottonseed meal fertilizer (CSM) treatment bootstrap MSI (simplified Morisitas similarity index ; critical value CH ................397 560 Unprocessed feed (UNP) and inorganic fertilizer (INO) treatment bootstrap MSI (simplified Morisitas similarity index; critical value CH ........................398

PAGE 22

22 561 Cottonseed meal (CSM) and inorganic fertilizer (INO) treatment bootstrap MSI (simplified Morisitas similarity index; critical value CH ........................399 562 Processed (PRO) and unprocessed (UNP) feed treatment bootstrap PSI (percent similarity index; critical value PSI ...........................................................400 563 Processed feed (PRO) and cottonseed meal fertilizer (CSM) treatment bootstrap PSI (percent similarity index; critical value PSI .............................................401 564 Processed feed (PRO) and inorganic fertilizer (INO) treatment bootstrap PSI (percent similarity index; critical value PSI ...........................................................402 565 Unprocessed feed (UNP) and cottonseed meal fertilizer (CSM) treatment bootstrap PSI (percent similarity index; critical value PSI ......................................403 566 Unprocessed feed (UNP) and inorganic fertilizer (INO) treatment bootstrap PSI (percent similarity index; critical value PSI .............................................404 567 Cottonseed meal (CSM) and inorganic fertilizer (INO) treatment bootstrap PSI (percent similarity index; critical value PSI .............................................405 568 Processed (PRO) and unprocessed (UNP) feed treatment bootstrap PSI (percent similarity index; critical value PSI ...........................................................406 569 Processed feed (PRO) and cottonseed meal (CSM) fertilizer treatment bootstrap PSI (percent similarity index; critical value PSI .............................................407 570 Processed feed (PRO) and inorganic fertilizer (INO) treatment bootstrap PSI (percent similarity index; critical value PSI values ...........................................................408 571 Unprocessed feed (UNP) and cottonseed meal fertilizer (CSM) treatment bootstrap PSI (percent similarity index; critical value PSI .................................................409 572 Unprocessed feed (UNP) and inorganic fertilizer (INO) treatment bootstrap PSI (percent similarity index; critical value PSI .............................................410 573 Cottonseed meal (CSM) and inorganic fertilizer (INO) treatment bootstrap PSI (p ercent similarity index; critical value PSI .............................................411 61 Initial (0 hrs) processed feed (PRO) treatment large zooplankton community pooled taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for 24 hour feeding trial .....................................................................................443 62 NF (no fish 24 hr incubation) processed feed treatment (PRO) large zooplankton pooled community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for 24hour feeding trial ..............................................................444

PAGE 23

23 63 Fish (single fish 24 hr incubation) processed feed (PRO) treatment large zooplankton pooled community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%F O)] for 24 hour feeding trial ..............................................................445 64 Initial (0 hrs) unprocessed feed (UNP) treatment large zooplankton pooled community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for 24 hour feeding trial ...................................................................446 65 NF (no fish 24 hr incubation) unprocessed feed (UNP) treatment large zooplankton pooled community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for 24hour feeding trial ..............................................................447 66 Fish (single fish 24 hr incubation) processed feed (PRO) treatment large zooplankton pooled community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for 24hour feeding trial ..............................................................448 67 Initial (0 hrs) cottonseed meal (CSM) treatment large zooplankton pooled community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for 24 hour feeding trial ...................................................................449 68 NF (no fish 24 hr incubation) cottonseed meal (C SM) treatment large zooplankton pooled community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for 24hour feeding trial ..............................................................450 69 Fish (single fish 24 hr incubation) cottonseed meal (CSM) treatment large zooplankton pooled community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for 24hour feeding trial .........................................451 610 Initial (0 hrs) inorganic fertilizer (INO) treatment large zooplankton pooled community taxa parameters [% number (%N), % volume (%V), % frequency of occur rence (%FO)] for 24 hour feeding trial, taxa %N, %V, or %FO L water sample per pond, n = 4 ponds. ...............................................................................452 611 NF (no fish 24 hr incubation) inorganic fertilizer (INO) treatment large zooplankton pooled community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for 24hour feeding trial ..............................................................453 612 Fish (no fish 24 hr incubation) inorganic fertilizer treatment large zooplankton pooled community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for 24hour feeding trial ..............................................................454 613 Initial (0 hrs) processed feed (PRO) treatment large zooplankton unpooled community taxa parameters: % number (% N 1 SE), % volume (% V 1 SE), and frequency of occurrence (% FO) for 24hour feeding trial ..............................................455 614 NF (no fish 24 hr incubation) processed feed (PRO) treatment large zooplankton unpooled community taxa parameters: % number (%N 1 SE), % volume (%V 1 SE), and % frequency of occurrence (%FO)] for 24hour feeding trial ..........................456

PAGE 24

24 615 Fish (single fish 24 hr incubation) processed feed (PRO) treatment large zooplankton unpooled community taxa parameters: % number (%N 1 SE), % volume (%V 1 SE), and % frequency of occurrence (%FO) for 24hour feeding trial ............................457 616 Initial (0 hrs) unprocessed feed (UNP) treatment large zooplankton unpooled community taxa parameters: % number (N 1 SE), % volume (%V 1 SE), and frequency of occurrence (%FO) for 24hour feeding trial ...............................................458 617 NF (no fish 24 hr incubation) unprocessed feed (UNP) treatment large zooplankton unpooled community taxa parameters: % number (%N 1 SE), % volume (%V 1 SE), and % frequency of occurrence (%FO) for 24hour feeding trial ............................459 618 Fish (single fish 24 hr incubation) unprocessed feed (UNP) treatment large zooplankton unpooled community taxa parameters: % number (%N 1 SE), % volume (%V 1 SE), and % frequency of occurrence (%FO) for 24hour feeding trial ...................................................................................................................................460 619 Initial (0 hrs) cottonseed meal f ertilizer (CSM) treatment large zooplankton unpooled community taxa parameters: % number (N 1 SE), % volume (%V 1 SE), and frequency of occurrence (%FO) for 24hour feeding trial ...............................................461 620 NF (no fish 24 hr incubation) cottonseed meal (CSM) treatment large zooplankton unpooled community taxa parameters: % number (%N 1 SE), % volume (%V 1 SE), and % frequency of occurrence (%FO) for 24hour feeding trial ............................462 621 Fish (single fish 24 hr incubation) cottonseed meal (CSM) treatment large zooplankton unpooled community taxa parameters: % number (%N 1 SE), % volume (%V 1 SE), and % frequency of occurrence (%FO) for 24hour feeding trial ...................................................................................................................................463 622 Initial (0 hrs) inorganic fertilizer (INO) treatment large zooplankton unpooled community taxa parameters: % number (N 1 SE), % volume (%V 1 SE), and frequency of occurrence (%FO) for 24hour feeding trial ...............................................464 623 NF (no fish 24 hr incubation) inorganic fertilizer (INO) treatment large zooplankton unpooled community taxa parameters: % number (%N 1 SE), % volume (%V 1 SE), and frequency of occurrence (%FO) for 24hour feeding trial ................................465 624 Fish (single fish 24 hr incubation) inorganic fertilizer (INO) treatment large zooplankton unpooled community taxa parameters: % number (%N 1 SE), % volume (%V 1 SE), and % frequency of occurrence (%FO) for 24hour feeding trial ...................................................................................................................................466 625 24hour feeding trial large ( > 200 m) plankton assemblage taxa total volumes (mm3/L) among four pond nutrient treatments ................................................................467

PAGE 25

25 626 Processed feed (PRO) treatment 24 hour feeding trial large (> 200 m) plankton assemblage volumes (mm3/L 95% CI) within three treatments [Initial, NF (no fish), Fish] ........................................................................................................................468 627 Unprocessed feed (UNP) treatment 24 hour feeding trial large (> 200 m) plankton assemblage volumes (mm3/L 95% CI) within three treatments [Initial, NF (no fish), Fish] .................................................................................................................................469 628 Cottonseed meal fertilizer (CSM) treatment 24 hour feeding trial large (> 200 m) plankton assemblage volumes (mm3/L 95% CI) within three treatments [Initial, NF (no fish), Fish] ..................................................................................................................470 629 Inorganic fertilizer (INO) treatment 24 hour feeding trial large (> 200 m) plankton assemblage volumes (mm3/L 95% CI) within three treatments [Initial, NF (no fish), Fish] ........................................................................................................................471 630 Pooled (among nutrient treatment) 24 hour feeding trial large (> 200 m) plankton assemblage volumes (mm3/L 95% CI) within three treatments [Initial, NF (no fish), Fish], 16 pooled 1 L replicate pond water samples per treatment ..........................472 631 24hour feeding trial large (> 200 m) plankton assemblage densities (plankton number/L) among four pond nutrient treatments .............................................................473 632 Processed feed (PRO) treatment 24 hour feeding trial large (> 200 m) plankton assemblage densities (plankter number/L 95% CI) within three treatments [Initial, NF (no fish), Fish] ............................................................................................................474 633 Unprocessed feed (UNP) treatment 24 hour feeding trial large (> 200 m) plankton assemblage densities (plankter number/L 95% CI) within three treatments [Initial, NF (no fish), Fish] ............................................................................................................475 634 Cottonseed meal fertilizer (CSM) treatment 24 hour feeding trial large (> 200 m) plankton as semblage densities (plankter number/L 95% CI) within three treatments [Initial, NF (no fish), Fish] ...............................................................................................476 635 Inorganic fertilizer (INO) treatment 24 hour feeding trial large (> 200 m) plankton assemblage densities (plankter number/L 95% CI) within three treatments [Initial, NF (no fish), Fish] ............................................................................................................477 636 Pooled (among nutrient treatment) 24 hour feeding trial large (> 200 m) plankton assemblage densities (plankter number/L 95% CI) within three treatments [Initial, NF (no fish), Fish] ............................................................................................................478 637 Processed feed (PRO) treatment 24 hour feeding trial large (> 200 m) plankton assemblage Diaptomous copepod densities (copepods/L 95% CI) within three treatments [Initial, NF (no fish), Fish] .............................................................................479

PAGE 26

26 638 Unprocessed feed (UNP) treatment 24 hour feeding trial large (> 200 m) pla nkton assemblage Diaptomous copepod densities (copepods/L 95% CI) within three treatments [Initial, NF (no fish), Fish] .............................................................................480 639 P rocessed feed (PRO) treatment 24 hour feeding trial large (> 200 m) plankton assemblage Diaptomous copepod densities (copepods/L 95% CI) within three treatments [Initial, NF (no fish), Fish] .............................................................................481 640 Inorganic fertilizer (INO) treatment 24 hour feeding trial large (> 200 m) plankton assemblage Diaptomous copepod densities (copepods/L 95% CI) within three treatments [Initial, NF (no fish), Fish] .............................................................................482 641 Pooled (among nutrient treatment) 24 hour feeding trial large (> 200 m) plankton assemblage Diaptomous c opepod densities (copepods/L 95% CI) within three treatments [Initial, NF (no fish), Fish] .............................................................................483 642 Processed feed (PRO) treatment 24 hour feeding trial large (> 200 m) plankton assemblage Filinia rotifer densities (rotifers/L 95% CI) within three treatments [Initial, NF (no fish), Fish] ...............................................................................................484 643 Unprocessed feed (UNP) treatment 24 hour feeding trial large (> 200 m) plankton assemblage Filinia rotifer densities (rotifers/L 95% CI) within three treatments [Initial, NF (no fish), Fish] ...............................................................................................485 644 Cottonseed meal fertilizer (CSM) treatment 24 hour feeding trial large (> 200 m) plankton assemblage Filinia rotifer densities (rotifers/L 95% CI) within three treatments [Initial, NF (no fish), Fish] .............................................................................486 645 Inorganic fertilizer (INO) treatment 24hour feeding trial large (> 200 m) plankton assemblage Filinia rotifer densities (rotifers/L 95% CI) within three treatments [Initial, NF (no fish), Fish] ...............................................................................................487 646 Pooled (among nutrient treatment) 24 hour feeding trial large (> 200 m) plankton assemblage Filinia rotifer densities (rotifers/L 95% CI) within three treatments [Initial, NF (no fish), Fish] ...............................................................................................488 647 Processed feed (PRO) treatment 24 hour feeding trial large (> 200 m) plankton assemblage Moina macrocopa (Daphnid) densiti es (daphnids/L 95% CI) within three treatments [Initial, NF (no fish), Fish] ....................................................................489 648 Unprocessed feed (UNP) treatment 24 hour feeding trial large (> 200 m) plankton assemblage Moina macrocopa (Daphnid) densities (daphnids/L 95% CI) within three treatments [Initial, NF (no fish), Fish] ....................................................................490 649 Cottonseed meal fertilizer (CSM) treatment 24 hour feeding trial large (> 200 m) plankton assemblage Moina macrocopa (Daphnid) densities (daphnids/L 95% CI) within three treatments [Initial, NF (no fish), Fish] .........................................................491

PAGE 27

27 650 Inorganic fertilizer (INO) treatment 24 hour feeding trial large (> 200 m) plankton assemblage Moina macrocopa (Daphnid) densities (daphnids/L 95% CI) within three treatments [Initial, NF (no fish), Fish] ....................................................................492 651 Pooled (among nutrient treatment) 24 hour feeding trial large (> 200 m) plankton assemblage Moina macrocopa (Daphnid) densities (daphnids/L 95% CI) within three experimental treatments [Initial, NF (no fish), Fish] ..............................................493 652 24hour feeding trial Initial control treatment major taxa densities (plankter number/L) within four pond nutrient treatments .............................................................494 653 24 hr feeding trial Initial treatment Diaptomous copepod densities (copepods/L 95% CI) among four treatments .......................................................................................495 654 24 hr feeding trial Initial treatment Filinia spp. rotifer densities (rotifers/L 95% CI) within large (> 200 um) plankton assemblages ...............................................................496 655 24 hr feeding trial Initial treatment Moina macrocopa (Daphnid) densities (daphnids/L 95% CI) within four nutrient treatments ..................................................497 656 24hour feeding trial Initial (0 hrs) treatment major (taxa %IRI > 0.5%; IRI: index of relative importance) large (> 200 m) plankton assemblage taxa %IRI values pooled among four nutrient treatments ........................................................................................498 657 24hour feeding trial NF (no fish + 24 hrs) treatment major (taxa %IRI > 0.5%; IRI: index of relative importance) large (> 200 m) plankton assemblage taxa %IRI values pooled among four pond nutrient treatments ........................................................499 658 24hour feeding trial Fish (one fish + 24 hrs) treatment major (taxa %IRI > 0.5%; IRI: index of relative importance) large (> 200 m) plankton assemblage taxa %IRI values pooled among four pond nutrient treatments ........................................................500 659 24hour feeding trial Initial (no fish + 0 hrs) treatment major (taxa %IRI > 0.5%; IRI: index of relative importance) large (> 200 m) plankton assemblage taxa unpooled %IRI values among four pond nutrient treatments ..........................................................501 660 24hour feeding trial NF (no fish + 24 hrs) treatment major (taxa %IRI > 0.5%; IRI: index of relative importance) large (> 200 m) plankton assemblage taxa unpooled %IRI values among four pond nutrient treatments ..........................................................502 661 24hour feeding trial Fish (one fish + 24 hrs) treatment major (t axa %IRI > 0.5%; IRI: index of relative importance) large (> 200 m) plankton assemblage taxa unpooled %IRI values among four pond nutrient treatments ..........................................503 662 Deterministic large plankton assemblage simplified Morisitas similarity index values (using taxa %IRI) between 24 hour feeding trial treatments within four pond nutrient treatments ...........................................................................................................504

PAGE 28

28 663 Deterministic large plankton assemblage percent similarity index values [using taxa %IRI (index of relative importance)] between 24 hour feeding trial trea tments within four pond nutrient treatments ...........................................................................................505 664 24hour feeding trial large (> 200 m) plankton assemblage bootstrap (taxa % nu mber; 1000 iterations) simplified Morisitas similarity index (MSI 95% CI) between treatments within the processed feed (PRO) treatment .....................................506 665 24hour feeding trial large (> 200 m) plankton assemblage bootstrap (taxa % number; 1000 iterations) simplified Morisitas similarity index (MSI 95% CI) between treatments within the unprocessed feed (UNP) treatment .................................507 666 24hour feeding trial large (> 200 m) plankton assemblage bootstrap (taxa % number; 1000 iterations) simplified Morisitas similarity index (MSI 95% CI) between treatments within the cottonseed meal (CSM) treatment ..................................508 667 24hour feeding trial large (> 200 m) plankton assemblage bootstrap (taxa % number: 1000 iterations) simplified Morisitas similarity index (MSI 95% CI) between treatments within the inorganic fertilizer (INO) treatment ................................509 668 24hour feeding trial large (> 200 m) plankton assemblage bootstrap (taxa % number; 1000 iterations) percent similarity index (PSI 95% CI) between treatments within the processed feed (PRO) pond nutrient treatment ..............................................510 669 24hour feeding trial large (> 200 m) plankton assemblage bootstrap (taxa % number; 1000 iterations) percent similarity index (PSI 95% CI) between treatments within the unprocessed feed (UNP) treatment .................................................................511 670 24hour feeding trial large (> 200 m) plankton assemblage bootstrap (taxa % number; 1000 iterations) percent similarity index (PSI 95% CI) between treatments within the cottonseed meal fertilizer (CSM) treatment ....................................................512 671 24hour feeding trial large (> 200 m) plankton assemblage bootstrap (taxa % number; 1000 iterations) percent similarity index (PSI 95% CI) between treatments within the inorganic fertilizer (INO) treatment, four replicates per treatment; unshared letters denote statistical differences (P < 0.05, Tukeys multiple comparison test). ..................................................................................................................................513 672 24hour feeding trial large (> 200 m) plankton assemblage bootstrap (taxa % number; 1000 iterations) percent similarity index (PSI 95% CI) between treatments within pooled pond nutrient treatments ...........................................................................514 5A 1 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment ...............................................................................................................583

PAGE 29

29 5A 2 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP ) treatment ...............................................................................................................584 5A 3 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment ................................................................................................585 5A 4 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment ................................................................................................................586 5A 5 Large zooplankton (size > 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment .......................................................................................................587 5A 6 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP) treatment ...............................................................................................................588 5A 7 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment ................................................................................................589 5A 8 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of o ccurrence (%FO)] for inorganic fertilizer (INO) treatment ................................................................................................................590 5A 9 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment ...............................................................................................................591 5A 10 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP) treatment ...............................................................................................................592 5A 11 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment ................................................................................................593 5A 12 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment ................................................................................................................594 5A 13 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment ...............................................................................................................595

PAGE 30

30 5A 14 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP) treatment ...............................................................................................................596 5A 15 Large zooplankton (> 200 m) ass emblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment ................................................................................................597 5A 16 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment ................................................................................................................598 5A 17 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequ ency of occurrence (%FO)] for processed feed (PRO) treatment ...............................................................................................................599 5A 18 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP) treatment ...............................................................................................................600 5A 19 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment ................................................................................................601 5A 20 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment ................................................................................................................602 5A 21 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment ...............................................................................................................603 5A 22 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP) treatment ...............................................................................................................604 5A 23 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment ................................................................................................605 5A 24 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment ................................................................................................................606 5A 25 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment ...............................................................................................................607

PAGE 31

31 5A 26 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP) treatment ...............................................................................................................608 5A 27 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment ................................................................................................609 5A 28 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment ................................................................................................................610 5A 29 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment ...............................................................................................................611 5A 30 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP) treatment ...............................................................................................................612 5A 31 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment ................................................................................................613 5A 32 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment ................................................................................................................614 5A 33 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment ...............................................................................................................615 5A 34 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP) treatment ...............................................................................................................616 5A 35 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment ................................................................................................617 5A 36 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency o f occurrence (%FO)] for inorganic fertilizer (INO) treatment ................................................................................................................618 5A 37 Large zooplankton (> 200 m) assemblage community taxa param eters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment ...............................................................................................................619

PAGE 32

32 5A 38 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP) treatment ...............................................................................................................620 5A 39 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment ................................................................................................621 5A 40 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment ................................................................................................................622 5A 41 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment ...............................................................................................................623 5A 42 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP) treatment ...............................................................................................................624 5A 43 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment ................................................................................................625 5A 44 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment ................................................................................................................626 5A 45 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequ ency of occurrence (%FO)] for processed feed (PRO) treatment ...............................................................................................................627 5A 46 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP) treatment ...............................................................................................................628 5A 47 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment ................................................................................................629 5A 48 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment ................................................................................................................630 5A 49 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment ...............................................................................................................631

PAGE 33

33 5A 50 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP) treatment ...............................................................................................................632 5A 51 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment ................................................................................................633 5A 52 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment ................................................................................................................634 5A 53 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment ...............................................................................................................635 5A 54 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP) treatment ...............................................................................................................636 5A 55 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment ................................................................................................637 5A 56 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment ................................................................................................................638 5A 57 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment ...............................................................................................................639 5A 58 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP) treatment ...............................................................................................................640 5A 59 Large zooplankton (size > 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % freq uency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment .......................................................................................641 5A 60 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment ................................................................................................................642 5A 61 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment ...............................................................................................................643

PAGE 34

34 5A 62 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP) treatment ...............................................................................................................644 5A 63 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment ................................................................................................645 5A 64 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V ), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment ................................................................................................................646 5A 65 Large zooplankton (> 200 m) assemblage comm unity taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment ...............................................................................................................647 5A 66 Large zooplankton (size > 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed treatment ...................................................................................................................648 5A 67 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment ................................................................................................649 5A 68 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer ( INO) treatment ................................................................................................................650 5A 69 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment ...............................................................................................................651 5A 70 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP) treatment ...............................................................................................................652 5A 71 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment ................................................................................................653 5A 72 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment ................................................................................................................654 5A 73 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequ ency of occurrence (%FO)] for processed feed (PRO) treatment ...............................................................................................................655

PAGE 35

35 5A.74 Large zooplankton (size > 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed treatment ...................................................................................................................656 5A 75 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fert ilizer (CSM) treatment ................................................................................................657 5A 76 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment ................................................................................................................658 5A 77 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment ...............................................................................................................659 5A 78 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (PRO) treatment ...............................................................................................................660 5A 79 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment ................................................................................................661 5A 80 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment ................................................................................................................662

PAGE 36

36 LIST OF ABBREVIATIONS 13C stable carbon isotope value 15N stable nitrogen isotope value 13C change (enrichment or depletion) in carbon isotope signature value 15N change (enrichment or depletion) in nitrogen isotope signature value CH or MSI Simplified Morisitas index of similarity DIC dissolved inorganic carbon DIN dissolved inorganic nitrogen DOM dissolved organic material POM particulate organic material PSI percent similarity index %N percent number %V percent volume %FO percent frequency of occurrence PRO processed commercial feed UNP unprocessed commercial feed (ground meal) CSM cottonseed meal (ground cottonseed) INO inorganic liquid nitrogen fertilizer ANOVA analysis of variance statistical test IRI index of relative importance %IRI index of relative importance standardized to percentage

PAGE 37

37 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy USING STABLE I S OTOPES TO CHARACTERIZE ORNAMENTAL AQUACULTURE POND TROPHIC DYNAMICS AND MAXIMIZE FISH PRODUCTION By Jon S. Kao May 2011 Chair: Charles E. Cichra Cochair: Roy P.E. Yanong Major: Fisheries and Aquatic Sciences Fish production in pond aquaculture systems is highly variable among individual ponds; a n obvious area of improvement would consis t of maintaining uniformly high levels of production in ponds through the proper understanding and manipulation of pond trophic dynamics. One method used for understanding the flows of energy and matter within ponds is by tracing and quantifying the flow of energy and matter via the movement of stable isotopes, from allochthonous feed inputs, and live foods derived from allochthonous fertilizer inputs and autochthonous energy and matter sources within the ponds trophic system. Using the common swordtail ( Xiphophorus hellerii ) live bearer tropical fish as a model animal, indoor feeding trials were successfully used to validate carbon and nitrogen isotope tracing methods, and obtain mean carbon ( 13C) and nitrogen ( 15N) isotope enrichment magnitudes for use in outdoor pond trials. Four nutrient types ( commercial processed feed, commercial unprocessed feed, organic cottonseed meal fertilizer, and inorganic liquid fertilizer) that are widely used within the aquaculture industry, were applied to 24 ponds (6 ponds per treatment) evenly stocked with swordtail broodstock. At harvest, fish biomass production was greatest within the two feed

PAGE 38

38 treatments after the 12 week trial. I norganic fertilizer produced the lowest overall fish bio mass, and least number of marketable swordtails (standard length > 31 mm). Stable isotope analyses revealed that d istinct nutrient movement patterns were present among pond treatments: (1) nutrients were moving in parallel through feed pond food webs, moving directly into fish and plankton (to varying degrees), (2) nutrients were moving in serial through inorganic fertilizer pond food webs: fertilizer phytoplankton small zooplankton macrozooplankton swordtails. The organic cotto nseed fertilizer treatment was a special case, cottonseed meal performed as both a fertilizer and a directly ingestible high protein food source, trophic utilization and nutrient movement results from this treatment, were intermediate between parallel and serial nutrient movement observed within the other three treatments. Large plankton (> 200 m) assemblages were analyzed for taxa and abundance, and no qualitative assemblage differences were found that could be associated with high or low fish production performance. Temporal differences in plankton assemblage compositions and abundances were clearly present, likely due to plankton community succession resulting from competition and predation, but assemblages did not largely differ among pond nutrient tr eatment type. The generalist food habits of swordtails imply that swordtails are capable of efficiently preying upon large macrozooplankton, but due to the pharyngeal jaw arrangement within the swordtail buccal cavity, gut content analysis was not a viabl e method of determining swordtail diet or predation effects. Twenty four hour captive feeding trials demonstrated that swordtails were capable of reducing large plankton biomass and altering large plankton assemblage composition. Swordtails preferentially predated upon Moina daphnids relative to calanoid copepods in 24hour incubation trials.

PAGE 39

39 CHAPTER 1 INTRODUCTION Background Ornamental finfish aquaculture is the largest revenue component ( 65%) of Floridas aquaculture industry, and totaled over $33 million in farm gate sales in 2005 ( FASS/FDACS 2005) down from a near industry record of over $42 million in 2001 (FASS/FDACS 2002) Over 5.45 million square meters of water surface were under tropical ornamental fish aquac ulture cultivation in the state of Florida in 2001 (FASS/FDACS 2002). The ornamental fish trade in the U.S. was valued at $660 million in foreign imports and $182 million in exports (Adams et al. 2001, Larkin et al. 2001), not including domestic trade. B y increasing domestic production in a more efficient and environmentally benign manner, fishing pressure on stocks of wild caught species and a half billion dollar foreign trade deficit could be reduced, while increasing employment and economic revenue in local production areas and potentially diversifying income sources for local farmers (Chapman et al. 1997) Tropical Fish Pond Aquaculture Ornamental finfish pond production methods are conceptually simple and typically consist of either periodic stocking of aerated ponds with fry or eggs produced by brood stock held in tank systems, or from brood stock left to reproduce freely within a newly prepared pond (Tamaru et al. 2001). Ponds are broadcast fed formulated commercial feeds daily, and often fertilizers are periodically applied to maintain phytoplankton bloom conditions to provide supplemental live feeds until fish are of marketable size. At harvest, fish are collected using nets or traps, graded for size and appearance, and sold. Typically, at the beginning of a production cycle, ponds are drained, de mucked, and excavated if heavily eroded. Ponds are then sterilized via pH shock by the application of hydrated lime [Ca(OH)2] to remove or reduce potential predators,

PAGE 40

40 competitors, pathogenic organis ms, and pest species (including undesirable fish species) that can interfere with harvest and grading, lime application also has the added benefit of raising pond total alkalinity; fertilizer (e.g., cottonseed meal) is often added prior to filling and stoc king to initiate and sustain a phytoplankton bloom (Boyd 1995, 1997, Kurten et al. 1999). Florida Aquaculture Best Management Practices ( FASS/ FDACS 2001) are legally binding statutes which have been established in an attempt to improve aquaculture facili ties design, management, and operations to better address growing public and government concerns with such issues as water usage, nutrient discharges to surface and ground waters, prevention of exotic species release, and general environmental degradation (FDACS 2000). Improved pond aquaculture management practices, such as the use of feeds with better growth efficiencies may reduce effluent nutrient levels (Boyd 1997, Cho and Bureau 1997), while simultaneously increasing production rates and production ef ficiencies, and therefore increase profitability given comparable feed prices. N utrient/feed quality evaluations are based upon the amount of growth in weight (protein and/ or energy content as opposed to gain in water weight ) of an organism as a funct ion of the amount of food consumed ( usually as bulk weight ) The highe r the observed growth rate for a specific amount of feed or live food, the higher the quality of that feed or live food (Halver 1989). A major question in intensive pond aquaculture systems is the fate and efficacy of relatively expensive commercial feeds and relatively inexpensive fertilizers once they are introduced into the aquaculture pond ecosystem and food web; employing stable isotope s as trac ers provides a promising approach in determining the ultimate fate of these nutrients. Aquaculture Pond Trophic Ecology Trophic ecology is the study of feeding relationships among organisms. These relationships are also impacted by biophysical, geophysical, and ecological processes. Crucial

PAGE 41

41 processes such as when autotrophic producer organisms (primarily photosynthetic plants and bacteria) are able to utilize the energy in sunlight (photosynthetic organisms) or a reduced inorganic substrate (e.g., hydrogen sulfide), and fix inorganic nitrogen and carbon (in the soil, water, or air) into complex organic compounds (carbohydrates, lipids, and proteins) that contain chemical energy stored within their chemical bonds ( NRC 1993, Jobling 2001). Heterotrophic consumer organisms may then consume these producer organisms and/or their byproducts, and they in turn may be consumed by predator organisms at higher trophic levels. Detritovores and heterotrophic decomposers (bacteria, fungi) ultimately return these organisms, their products, and wa stes back into inorganic compounds (remineralization). These inorganic compounds can then again be utilized by autotrophic organisms under the proper conditions. The study of the flow of energy and the cycling of matter (nutrients) among trophic levels, and the factors that influence these trophic interactions are the major components of trophic ecology. Trophic ecology makes up one of, if not the, fundamental basis of ecology. Obtaining sufficient energy and matter (nutrients) is the primary objective of every organism. Only by obtaining sufficient energy and nutrients are organisms able to maintain their physical and chemical structures, while perform ing necessary activities to grow, reproduce, and ultimately persist through ecological and evolutiona ry time The production and storage of biochemical energy and biomass from solar energy and inorganic nutrients by producer organisms drives and sustains almost all natural ecosystems. N otable exceptions include chemosynthetic sulfur bacteria in anoxic muds common in salt marshes and other estuarine environments, and chemosynthetic hydrogen sulfide bacteria in deep ocean cold black smoker seeps and in deep ocean geothermal hot seeps and other aphotic environments ( Corliss et al. 1979, Grassle 1985, Jannasch and Mottl 1985).

PAGE 42

42 In contrast to natural aquatic ecosystems and extensive aquaculture systems (no intentional anthropogenic nutrient input), semi intensive and intensive aquaculture systems rely upon human intervention in the form of supplemental nutrient inputs. These inputs may take the form of natural or formulated feeds, and organic or inorganic fertilizers, which are used singly or in combination to increase production of the desired species beyond that which could be supported naturally. Co mmercial formulated feeds are the most common form of supplemental nutrients used in the American aquaculture industry and come in two basic forms: nutritionally complete and incomplete feeds (Lovell 1998, Egna and Boyd 1997, Halver 2003). Complete feeds provide all of the nutritional needs of the target species, providing balanced, metabolically readily available, high quality and highdensity nutrients (protein, lipids, carbohydrates, minerals, and vitamins) to produce high growth and survival rates, and promote good fish health (Royes 2003). Incomplete feeds are generally cheaper and supplement the nutritional needs of the target species, which must obtain a portion of their nutritional needs from live feeds or digestible organic matter within their env ironment. In addition to complete and incomplete feeds, two forms of feeds: cooked (properly heated) and extruded feeds, and ground meals (ingredients mechanically ground into small pieces or fine powders) are typically used in the American pond aquaculture industry. Heating (cooking) and extruding feed ingredients generally increases the digestibility and nutritional availability of some ingredients (starches, proteins), but may reduce or destroy other nutrients (nonheat stable proteins, amino acids, an tioxidants and vitamins), which may need to be reapplied (top dressed) following extrusion (Royes 2003, Barrows et al. 2008). Aquaculture Pond Applied Nutrient Management Practices Energy and matter, in the form of carbohydrates and lipids, must be pre sent in sufficient quantities and forms nutritionally available to the culture animal (Lovell 1998) I f energy in the

PAGE 43

43 form of carbohydrates and lipids are insufficient for the needs of the fish, dietary proteins (amino acids) may be metabolized for energ y. This is both expensive and wasteful, as protein is generally the greatest feed component expense and provides less energy on a per weight basis than lipids and carbohydrates. Additionally, high levels of amino acid catabolism are potentially lethal du e to increased ammonia production as a direct byproduct. Ammonia can quickly degrade water quality and become toxic under the proper conditions and in sufficient concentrations. Furthermore, energy to protein ratios must be properly balanced to promote good growth and prevent the development of nutritionally related pathological conditions and/or unmarketable physical characteristics ( e.g., fatty fish fillets, fatty livers dysfunctional liver diseases, etc.). Historically, fish meal has been the primar y source of protein in fish feed formulations due to the high growth rates and high palatability observed with its use F ish meal protein is a high quality protein for fish having an amino acid composition in good agreement with the amino acid requirement of most fish species (Halver 1989, Lim et al. 2004). However, for the foreseeable future the use of plant based protein sources will continue to increase as limited availability, increasing cost, and potential nonsustainability of fish meal use ma kes plant based alternatives more attractive (Hardy 2003, Forster 2004, Barrows et al. 2008). Although plant based proteins are generally cheaper and more readily available than fish meal, their effectiveness in fish feed formulations is often inferior to fish meal based feeds due to lower palatability, suboptimal amino acid composition, and antinutritional factors (e.g., phytate, gossypol) when used at high concentrations; therefore, plant protein based feed formulations frequently need to be tested in re search trials prior to their use in practical feeds (Day and PlascenciaGonzalez. 2000, Lim et al. 2004, Yamamoto et al. 2007, Barrows et al 2008).

PAGE 44

44 Fertilizer alone is often applied to fish ponds, but more frequently it is applied in conjunction with feed for aquaculture pond production. Inorganic fertilizers in chemical forms available to phytoplankton, provide additional nitrogen, phosphorous, potassium (N P K ) and trace elements necessary for enhanced plant growth depending upon fertilizer costs and l ocal needs (regional geochemistry and water chemistry) required by the farmer. Organic fertilizers, in the form of plant and animal silage, manures, plant and animal meal byproducts from oil processing and the food industry, also are often used to stimul ate phytoplankton, protozoan, and bacterial/fungal detrital production (Schroeder 1987, FAO 1995, Halver 2002). Both inorganic and organic fertilizers boost living food web component production, increasing natural phytoplankton, zooplankton, and invertebrate biomass for the ultimate consumption and increased production of the ponds target species. These relatively low cost fertilizers utilize the ponds ecosystem to transfer nutrients to fish, and may prove to be a cost effective alternative to more expensive commercial feeds (Schroeder 1987, Barkoh 1996). Additionally, live foods may provide trace elements not present or in sufficient quantities in manufactured feeds, thereby enhancing growth, health, and appearance (marketability). Periodic sampling of fish and examination of their gut contents for diet and trophic investigations is useful for describing short time scale nutritional processes, but is often temporally and spatially limited ; due to the large amounts of time and labor involved i n fish collection, processing, and analyzing and recording individual fish gut content items. Additionally, gut content analysis can bias diet composition data by overestimating the importance of prey components that resist digestion (e.g., chitinous crus taceans, shelled mollusks) and time of sampling in relation to a given fish species feeding behavior (diurnal, nocturnal, crepuscular, seasonal differences, ontogenetic dietary changes, continuous feeding,

PAGE 45

45 gorge and digest, etc .); furthermore, unidentifie d prey remains and empty stomachs, which are often a sampling artifact of regurgitation brought about by capture stress, are commonly observed during gut content analysis (Cailliet et al. 1986). Due to the logistical and budgetary constraints required to properly implement gut content analysis, it is usually limited temporally and spatially to a single season and/or geographical area. A major shortcoming of gut content analyses is that it does not address cumulative long time scale questions regarding the overall nutritional needs and predatory impact of fish upon their environment and local food web. Trophic Dynamics Investigations Using Stable Isotopes A longtime scale analytical methodology is necessary to complement the short time scale analytical m ethod of gut content analysis. Stable isotope tracer techniques have been successfully used to describe trophic processes in various ecosystems (Frazer 1996, Hansson et al. 1997, Velinsky and Fogel 1999, Gu et al. 2001, Lochmann et al. 2001, Dawson et al. 2002) I sotope tracer methods could potentially provide less biased trophic information than gut content analysis due to the cumulative sources of nutrition that are contained within the isotopic profile of the whole fish (DeNiro and Epstein 1978, 1981a Saito et al. 2001, McCutchan et al. 2003, Olive et al. 2003). Due to the longterm relationship between dietary history and consumer organism tissue isotope signature, the advantages of using isotopic signature analyses in diet studies also can be proble matic. Short term dietary changes such as intraseasonal diet switching and heavy consumption of highly abundant, but ephemeral prey pulses may be important to the life history of the fish and the trophic ecology of the habitat, but remain undetected usi ng stable isotope dietary analysis techniques. However, gut content analysis methods also can fail to properly evaluate prey that is only briefly abundant and heavily exploited by fish predators. Depending upon the questions being asked and the time scal es involved, solely using either gut content analysis or stable

PAGE 46

46 isotope techniques may make it difficult or impossible to answer the questions being asked by the researcher. Ideally, the complementary strengths of the two diet analysis methods would be em ployed simultaneously to investigate fish dietary habits. However, using both diet analysis techniques is not always possible due to logistical, budgetary or other constraints F or example, gut content analysis may not be possible when a high frequency of regurgitation from capture stress occurs or when fish employ pharyngeal teeth, detritovorous feeding habits, etc A lternately, isotope methods may provide little or no information regarding short term dietary habits (seasonally ephemeral prey, daily pre y or fish migrations, etc .) and feeding chronology patterns (diurnal, nocturnal, crepuscular). By comparing and contrasting the trophic ecologies of aquaculture ponds that are undergoing different management techniques (feeds and/or fertilizers), it may be possible to identify management strategies (feeds or fertilizers, processed or unprocessed feed, complete or incomplete feeds) that are more cost effective for pond aquaculture practitioners, increasing the profitability and longterm economic viability of their operations. Unfortunately, nutrients introduced into pond ecosystems from commercial feeds and fertilizers may ultimately end up as undesirable bacterial, fungal, algal, or animal biomass (i.e., predatory and pest insects, crayfish, frogs, turtles, etc) a s opposed to increased fish biomass (desired nutrient sink) F eed and fertilizer inputs also may be lost to the sediments, and become unavailable for biological production (Boyd 1995, Boyd 1997, Tepe and Boyd 2003). Empl oying stable isotopes as tracers in pond ecosystems may allow estimation of the relative efficiencies of fish production from commercial fish feeds and fertilizers, e nabling producers to determin e if higher commercial feed costs, versus relatively inexpens ive fertilizer are justified.

PAGE 47

47 CHAPTER 2 INDOOR SWORDTAIL FEE DING AND ISOTOPE MET HODOLOGY VALIDATION TRIAL The use of stable isotopes, within applied nutrients to investigate pond trophic dynamics, was first investigated using stable isotopes as tracers in an indoor study under controlled laboratory conditions. The purpose of the indoor trial was to determine the feasibility of using carbon and nitrogen isotopes as trac ers of the transfer of manufactured feeds to their fish consumers in an indoor laborator y setting as an analog for outdoor pond food web processes. If juvenile fish from different feed (isotopically distinct) group treatment s were isotopically distinct at the conclusion of the three month trial, then these results could be taken as validatio n of the feed/nutrient isotope tracing methodology (DeNiro and Epstein 1978, 1981a Fry 1991). Another important purpose of the indoor isotope tracing methodology validation trial was to determine the mean magnitude and sign ( +, ) of isotopic enrichment for both carbon and nitrogen over a single trophic level in a simplified food web involving swordtails (Xiphophorus heller i i ) in a controlled laboratory setting. Conducting indoor laboratory experiment al trials allowed for the complete control of what growing fry consumed. Food web component carbon and nitrogen isotope signatures are used for contrasting and complimentary purposes (Hobson and Wassennaar 1999). Carbon signatures are primarily used to trace nutritional sources within a food web due to ca rbons highly conserved isotopic signature, as carbon in a consumer is only minimally enriched relative to its food (~ 0.5 1 13C ; DeNiro and Epstein 1978, 1981b, Fry 1991). In contrast, nitrogen exhibits a trophic level enrichment; consumers are enri ched in 15N relative to their food at a rough ly constant level (~ 3.4 15N ; DeNiro and Epstein 1981, Fry 1991), allowing trophic position within a food web to be predicted with a high degree of confidence (Kling and Fry 1992, Vander Zanden and

PAGE 48

48 Rasmussen 1999, Vander Zanden and Vadeboncoeur 2002) given that basal nutrient nitrogen signature is known with a fair degree of certainty. Carbon and nitrogen isotope signatures tend to complement each other, in that nutrient/food sources believed to be primary s ources of nutrition for a predator must have both plausible carbon and nitrogen signature magnitudes and mathematical signs (+, ). T rophic position of an organism can be estimated by comparing its isotopic signature relative to other organisms in the sys tem T he expected magnitude of isotopic fractionation occurring between a consumer organism and lower trophic levels ( ~ 1.0 C, ~ 3.4 N per trophic level), believed to support the organism within the food web, will determine whether an organism or nutrient source is a plausible prey item or nutrient source for th at organism, or at the very least, potentially within the trophic food web supporting the organism (Carpenter et al. 1985, Phillips and Gregg 2003.) Materials and Methods Experimental Feed Composition and Isotopic Signature Profiles Five isonitrogenous and isocalorific feed formulat ions with varying proportions of animal (fish meal) and plant (wheat protein isolate) based proteins were fed to sword tail fry ( Xiphophorus hellerii : Family Poeciliidae ) in indoor tanks. Because feed formulations differed in their protein source inclusion rates, they also differed in their stable carbon and nitrogen isotope compositions. Stable isotopes of carbon (12C, 13C) and nitrogen (14N, 15N) were used as elemental tracers to characterize nutrient transfer from experimental flake feeds to ornamental fish fry. Additionally, carbon and nitrogen isotope fractionation magnitudes between feed and fry for a single trophi c level were measured for each of the five experimental flake feed treatments.

PAGE 49

49 The equation for determining carbon isotope signature ( 13C) is given as (Anderson et al. 1988, Knowles and Blackburn (eds.) 1993, Cook et al. 1998): 1313 13 13C = 1000samplestd stdRR R where: 13C is the carbon delta signature; (), 13Rsample is the ratio of 13 12C C in the sample material and 13Rstd is the ratio of 13 12C C in the standard material. Similarly for nitrogen isotope signature determination ( 15N): 1515 15 15N = 1000samplestd stdRR R where: 15N is the nitrogen delta signature; (), 15Rsample is the ratio of 15 14N N in the sample material and 15Rstd is the ratio of 15 14N N in the standard material Broodstock and Fry Production A total of 500 broodstock swordtails were used during the indoor portion of the study (400 female: 100 male) to produce fry for this trial. In a large greenhouse ( Figure 2 1; 22 m length 9.1 m width), 10 epoxycoated concrete burial vaults ( Figure 22; 240 cm length 89.5 cm width 73.6 cm height; approximately 1,665 L volume) each held a group of broodstock ( 75 female: 25 male) in a cubeshaped broodstock confinement basket ( Figure 23; 66 cm length 66 cm width 49.5 cm height) lined with polyethylene mesh (0.635 cm 0.635 cm square plastic

PAGE 50

50 mesh). As swordtails are primarily surface feeders, b roodstock were fed TetraMin commercial flake food as their staple diet. Floating flake feed was delivered to broodstock via automated belt feeders on 12hour timers, feeders were refilled every morning [between 09:0011:00 hrs (EST)]. Newly spawned fry dropped through the mesh basket where water currents within the vault moved fry away from confined broodstock, minimizi ng cannibalism. The vaults were supplied with filtered freshwater from a partial recirculation filter system that had a large drop bead filter, fluidized sand filter bed, UV sterilizer and degassing/bio media trickle bed (Figures 21, 24); when temperatures were over 26 C, cooler (~ 22 C), degassed well water was automatically [nominal inch (1.9 cm) solenoid valve and submersible thermostat] shunted directly into the vats and excess water left the system through a screened drain. Twenty randomly sel ected broodstock (10 female: 10 male) and 60 fry (trial day 0) were sacrificed for morphometric (standard length and weight), and stable isotop e analyses (5 female, 5 male and 10 fry) immediately prior to the initiation of the indoor feeding trial; sacrificed broodstock and fry were not fed for 16 hrs prior to euthanasia to prevent biased weight measurement and isotopic contamination from ingested feed. Examination of excised gastrointestinal (GI) tracts from sacrificed broodstock (n = 8) not used for isotopic analysis revealed that no undigested matter remained in the GI tract. Sacrificed broodstock were euthanized with an overdose of tricaine methanesulfonate (MS222) at a dose of 2.5 g/L, their length (SL mm) and wet weight (g) were recorded, and then f rozen ( 74 C 2 C) for several months prior to freeze drying Samples were freeze dried at 57 C under vacuum (1 104 Torr; Welch Chemstar vacuum pump model 1402N 60) for 2448 hrs depending on sample wet weight) for further morphometric analysis (dry weight) and processing/preparation for isotopic analysis.

PAGE 51

51 Remaining fry (0 14 days old) were transferred indoors and held within a n elliptical polyethylene tank ( 103 cm length x 81 cm height x 61 cm width ; approx. 265 L volume) on the same freshwater recirculating filter system as the indoor trial experimental test chambers (Figures 2 6 28; same room and water temperature). After a oneweek acclimation period, fry were randomly allocated to one of five experimental flake feed treatments beginning on 12 August 2005. E xperimental feed regimes began the following day. When necessary, animals used within this study were humanely sacrificed according to the current animal use guidelines of the University of Florida Institutio nal Animal Care and Use Committee (IACUC) The details of animal use for this study are on file at the University of Florida IACUC office (IACUC # D591). Manufactured Feeds Five flake feeds, containing wheat protein isolate (courtesy of Honeyville Grains Inc.) and fishmeal were manufactured at the Fish Technology Center (U.S. Fish and Wildlife Service) in Bozeman, Montana using a Buflovak Double Drum Dryer (30.5 cm diameter 45.7 cm length drum) running at 5.46 kg/cm2 (78 psi) steam pressure and at temperatures between 68.3 81.1 C. Each feed contained approximately 40% total protein by dry wt., but differed in their proportions of proteins derived from animal (FM fish meal ) and plant (WP wheat protein isolate ) protein sources : 0% F M and 100% W P %, 25% F M and 75% WP 50% F M and 50% WP 75% F M and 25% WP, and 100% FM and 0% WP ( Figure 25). For the duration of the study, f lake feeds were stored in air tight polyethylene buckets at approximately 18 C 3 C (~ 65 F). Experimental Design I ndoor trial experimental design consisted of five isocalorific and isonitrogenous experimental feeds fed to randomly selected individual fish, housed in 105 replicate aquaria

PAGE 52

52 (Figures 2 6 28; 6.4L rectangular polystyrene tanks; 21 aquaria per feed treatment ) for a period of 90 days. Each tank ( Figure 2 7) initially held 10 fry (114 days old) that were randomly allocated to each aquaria from a common tank in groups of five fry (1,050 total fry). Aquaria were held on a recirculation water system that utiliz ed UV sterilization, a drop bead filter and a fluidized sand bed bioreactor to maintain water quality. Local well water (~ 11 mg/L CaCO3 alkalinity, ~ 22 mg/L CaCO3 hardness) was used to fill and maintain system volume; water lost from daily bead filter b ack flushing was automatically replaced from a gravity fed head tank reservoir and float valve contained in the system sump. A standard photoperiod (14 hrs light: 10 hrs dark) was maintained (Lovell 1998). Water temperature within the experimental system was maintained by setting the room thermostat to maintain a constant temperature of 24C 2 C. Fry were fed at an initial daily ration of 15% mean wet body weight based upon initial mean fry wet weight. Subsequent rations were adjusted based upon weight measurements occurring at 10day intervals for the first thirty days, and weight measurements made every 15 days thereafter un til termination of the trial. T o measure fry weight change to calculate feed weight adjustments, one replicate (10 fry) from ea ch of the five treatments was chosen at random at a predetermined time interval and euthanized for weight and standard length measurement (and potential isotopic analysis). Sacrificed fry were not fed for 16 hrs prior to euthanasia to ensure ingested fo od would not bias weight measurements and future isotopic analysis. Following weight and standard length measurement, fry were stored frozen ( 74 2 C) for a period from 12 18 months then freeze dried and individually stored in polystyrene vials (32ml volume ) in 23L air tight polyethylene buckets that contained 250 g of desiccant (Dri rite: calcium sulfate/cobaltous chloride mixture). Daily ration level (% wet body w eight ) for all

PAGE 53

53 replicates within a given time interval was based upon the euthanized replicate with the highest mean fry weight. Ration for a given replicate was adjusted on a per capita basis (initial 10 fry per replicate) following a daily morning mortality check prior to feeding. Daily ration was reduced to 10% mean body weight f or days 3190, with weighing adjustments occurring at days 45, 60 and 75. Freeze Drying and Sample Preparation Samples prepared for isotopic analysis were stored frozen as mentioned above, at 74 C, and then freeze dried ( Figu res 2 9 211; Virtis Sentry) at 57C for 24 48 hrs (16 20 hrs for smaller specimens); a freeze drying pilot study was performed with swordtails of comparable weight and size to those in the main study to determine freeze drying durations after whi ch no further sample weight loss occurred. Following freeze drying, whole fish samples were ball milled into a fine homogenous powder (Spex 5100 ball mill). During the milling process, forceps were used to place fry samples into small stainless steel capsules each containing a stainless steel ball vibrated at a high speed from 2080 seconds depending on fry weight (3500 rpm, 1/15 hp motor) M illing capsules, milling balls and forceps were sprayed with 70% isopropyl alcohol and swabbed with sterilize d co tton swabs (Q tips Unilever Corp.) and laboratory wipes (KimWipes Kimberly Clark Corp.) to remove trace contaminants and prevent carry over between samples. As an added precaution, latex gloves were worn during the sample preparation process and were sp rayed with 70% isopropyl alcohol and wiped with laboratory wipes between samples. A small amount of powdered fry material (1501,300 g) or entire fry if fry dry weight was less than 1,000 g, were placed into preformed 7 mm x 4 mm tin sample capsules (Costech Analytical Industries, Inc.) and weighed to the nearest g using a Cahn 25 Automatic Electrobalance laboratory scale. Alcohol swabbed forceps were used to fold capsules into a tight bal l to prevent loss of sample material, folded capsules were then placed

PAGE 54

54 into pre cleaned and pre sterilized polystyrene well plates and shipped to the University of Florida Center for Isotope Geoscience Laboratory (Department of Geology: UFCIGL) or to the N ort hern Arizona University Colorado Plateau Stable Isotope Laboratory (Department of Biological Sciences: NAUCPSIL) for stable isotope analyses. Samples were analyzed for carbon and nitrogen isotope composition ( 13C and 15N) with an isotope ratio mass spectrometer [VG Prism (series II) mass spectrometer interfaced with a Carlo Erba CNS (carbon, nitrogen, sulfur) elemental analyzer (UFCIGL) or a ThermoFinnegan Delta Plus Advantage gas isotope ratio mass spectrometer interfaced with a Costech Analytical ECS4010 elemental analyzer ( NAUCPSIL) Triplicate weighed subsample isotopic analyses were made for ten randomly selected broodstock (five male, five female) to determine within fish variation in isotopic signature values. Isotopic Signature I sotopic sig nature s of sampled material were reported in standard delta notation ( ) where : 1515 15 15N = 1000samplestd stdRR R 15Rsample is the ratio of 15 14N N in the sample material and 15Rstd is the ratio of 15 14N N in the standard material 1313 13 13C = 1000samplestd stdRR R 13Rsample is the ratio of 13 12C C in the sample material and13Rstd is the ratio of 13 12C C in the standard material

PAGE 55

55 The analytical standard for nitrogen is atmospheric nitrogen ( 15Nair = 0 ) and the standard for carbon, historically is the Pee Dee Belemnite (PDB) formation carbon ( 13CPeeDee = 0 ); Belemnitella americana ( Cephalo poda) an extinct Cretaceous limestone marine fossil from the Pee Dee formatio n in South Carolina, U.S.A Unfortunately, the PDB formation has been exhausted and this carbon standard is no longer available C arbon source s isotopically s tandardized to the PDB standard were used Elemental Carbon to Nitrogen Ratio Elemental carbon to n itrogen ratio determination w as performed on a limited basis during isotopic analysis at the NAU isotope analysis laboratory. Elemental carbon to nitrogen ratio w as determined using a CNS (carbon, nitrogen, sulfur) elemental analyzer downstream from the isotope ratio mass spectrometer. Statistical Analysis Statistical analys e s w ere performed using Microsoft Excel and Prism statistical software package version 4.03 (GraphPad Software Inc. 2005) Microsoft Excel was used to calculate basic descriptive statistics (mean, standard deviation, standard error, 95% confidence intervals, etc.). Prism was used to conduct parametric tests such as analysis of variance (ANOVA) and post hoc comparison tests, unless otherwise noted. Fry Growth and Survival Final (24 36 hrs. postprandial) wet weights of fry were compared using a one way analysis of variance (ANOVA) statistical test 24 36 hours postprandial was deemed to be a s ufficient time interval to ensure that most, if not all, swordtail gastrointestinal tracts were empty. Average fry survival rates among treatments also were compared using a one way ANOVA test, after first performing an arcsine square root data transformation upon the survival

PAGE 56

56 rate percentage data. Percentage data is typically binomial in distribution, requiring transformation to a normal distribution prior to ANOVA. Isotopic Analyses C arbon and nitrogen isotope signatures of randomly selected subsamples of broodstock females, males and initial fry used during the three month trial were compared using a one way ANOVA. C arbon and nitrogen isotope signatures of experimental feeds, and surviving harvest fry from the five treatment groups, also were compared using ANOVA (Zar 1984, Ott 2000) Results Fry Growth Fry weight trajectories did not appear to qualitatively differ among treatments for randomly selected single replicates for the first three weighing periods (days 10, 20, 30) F ry growth at day 30 corresponded to a n increase of 146 % from their initial mean weight X = 0.0089 g ( 0.0082 to 0.0097 g 95 % CI n = 60). Fry subsample mean weights began to visibly diverge among the five experimental treatments (Figure 212) on day 45; the mean weight of the 0% fishmeal: 100% wheat protein diet fry having a mean weight slightly less X = 0.021 g ( 0.012 to 0.030 g 95 % CI n = 10) than the mean treatment group with the most similar diet composition and near est mean weight (25% fishmeal: 75% wheat protein: X = 0.046 g, 0.045 to 0.050 g 95 % CI n = 10). However, these differences were not statistical differen ces due to the lack of replication between the start and end of the experime ntal trial. These single tank replicates were periodically sacrificed (one per treatment) to determine fry weight changes over the twelve week course of the trial for daily feed ration adjustment s and to track isotopic changes in fry among treatments over time (Table 2 1)

PAGE 57

57 Similar to the lower growth trajectory exhibited by the all plant diet pre harvest fry (single replicates; Figure 212), all plant diet harvest fry average weight also w as lower (Figure 2 13). At the conclusion of the 90day trial, the 0% fishmeal: 100% wheat protein diet treatment fish ( X = 0.136 g, 0.1160 to 0.1566 g 95 % CI n = 12) had a significantly lower average weight than those of the other diet treatments (Figure 2 13; P < 0.01, Tukeys multiple comparison test ). Fry Survival Rates Indoor fry percent survival rates did not differ among experimental feed treatments at twelve week trial termination [ Figure 214; one way ANOVA (arcsin square root data transformation: P = 0.1475, F = 1.765(2,4 ) = 0.05] Mean survival rates within replicates for the five experimental feed treatments were: 100%FM:0%WP X = 84.6% ( 76.9 to 92.3 % 95 % CI n= 14), 75%FM:25%WP X = 82.3% ( 71.5 to 93.1 % 95 % CI n =13 ), 50%FM:50%WP X = 82.3% ( 74.8 to 89.8 % 95 % CI n = 13), 25%FM:75%WP X = 88.0% ( 80.4 to 95.6 % 95 % CI n = 15), and 0%FM:100%WP X = 94.2% ( 89.1 to 99.2 % 95 % CI n =12). Feed Isotopic Signatures The experimental feeds differed in their carbon and nitrogen isotope signatures along a regular (monotonic) sequence (Figure 216), as would be expected by stepwise inclusion rates of plant and animal protein in their manufacture. The TetraMin tropical fish staple flake control diet that was fed to female and male broodstock, isotopically differed from the five experimental feed diets for either 13C or 15N, and frequently both. Swordtail Isotopic Signature s Initial male brood stock, female broodstock and (0 to 14dayold) fry mean carbon isotope signatures C were : 21.63 ( 21.80 to 21.45 95 % CI, n=10), 21.57 ( 21.67 to 21.47 95 % CI, n=11), and 21.42 ( 21.42 to 21.41 95 % CI, n=21), respectively. C

PAGE 58

58 signatures did not differ significantly (Figure 2 15; one way ANOVA: P = 0.6053, F = 0.5085, (2,2) = 0.05). Initial male broodstock, female broodstock, and (0 to 14day old) fry mean nitrogen isotope signatures ( N) were 10.81 (8.91 to 12.70 95 % CI n=10), 11.28 (9.54 to 13.01 95 % CI n=11), and 11.63 (8.72 to 14.54 95 % CI n=21), respectively. Only male broodstock and fry nitrogen signatures differed significantly ( Figure 2 15; one way (2,2) = 0.05). Si milar to the 13C and 15N signature pattern of the five experimental feeds (Figure 2 16), pre harvest fry 13C and 15N signatures changed among treatment fry groups in a regular trajectory during the course of the trial (Figures 2 17 218). Following diet switching, isotopically similar fry within different treatments isotopically diverged and came to resemble their respective feeds over the twelve week s of the trial. These pre harvest differences were not statistical differences as only single replic ates (one aquari um ) from each treatment were analyzed for each of the periodic sampling periods prior to trial completion. At harvest, carbon isotope ( C) signature differences among treatment groups were readily apparent ( Figure 218). However, fo ur pairwise comparisons of treatment fry C signatures did not significantly differ: (100FM:0WP and 75FM:25WP), (75FM:25WP and 50FM:50WP), (50FM:50WP and 25FM:75WP), and (75FM:25WP and 25FM:75WP), ( Figure 2 18; one (2,4) = 0.05). Fry 13C signatures, reflected the carbon isotope signatures of their respective food sources. Nitrogen isotope signatures also differed among treatment groups ( Figure 218 ; one (2,4) = 0.05).

PAGE 59

59 At harvest, C signatures of fry differed from their respective feeds by an average of 0.64 (0.31 to 1.51 C 95 % CI ) and nitrogen signatures differed by an average of 3.03 (1.82 to 8.09 N 95 % CI ). The magnitudes of the differences in C values betwe en fry and feed did not appear to change with plant protein inclusion level in the experimental feeds ( Figure 2 19); however, differences in 15N signatures increased as plant protein inclusion increased, but these differences asymptote for the three exper imental feeds with the highest plant content ( Figure 2 20). Nitrogen isotope signature differences between harvest fry and their respective experimental feeds were the lowest for the two feeds that contained the most fish protein (100FM:0WP and 75FM:25WP) the fry and feed nitrogen isotope signature differences of these two groups significantly differed from those of the two treatments that contained the most wheat protein ( Figure 2 20; P < 0.05 Tukeys multiple comparison test). A dual ( Cartesian ) plot o f C and N values for experimental feeds, control feed, broodstock, initial fry and juvenile swordtails at harvest was created (Melville and Connolly 2003) to provide a visual representation of the contrasting carbon and nitrogen isotope signatures present among different groups ( Figure 2 21) Monotonous changes in harvest fish isotopic signature, closely follow monotonous isotopic signature differences present within experimental feeds. Elemental Carbon to Nitrogen Ratio Analysis Elemental carbon to nitrogen ratios for indoor trial harvest fry did not significantly differ among treatments ( Figure 222; one (2,4) = 0.05). Elemental carbon to nitrogen ratio values did not differ greatly among fry treatment s, actual mean elemental carbon to nitrogen ratio values for the indoor trial fry at harvest were: 100FM:0WP X = 4.31 ( 0 to 8.83 95 % CI), 75FM:25WP X = 4.26 ( 0 to 8.73 95 % CI),

PAGE 60

60 50FM:50WP X = 4.33 ( 0 to 8.88 95 % CI), 25FM:75WP X = 4.74 ( 0 to 10.64 95 % CI), and 0FM:100WP X = 4.90 ( 0 to 10.04 95 % CI). Discussion Fry Growth M ean pre harvest weights from different experimental feed treatments were almost identical for the first 30 days of the feeding trial. Fry growth follow ed an exponential growth trajectory, typical for juvenile fish es ( Cailliet et al. 1986 Tamaru et al. 2001). After trial day 30, single replicate fry weigh ts diverged and were often erratic among feed treatments (Figure 212) The 100% wheat protein diet fry growth rate lagged behind the other fry groups after day 30 and fish in this treatment group were smaller at the end of the 90day trial. Although pre harvest growth data were derived from single replicate subsample fry it seems plausible that from day 30 onward, a missing dietary factor(s) limited the growth of the 100%plant protein diet fry and prevented them from attaining the size of fry from the other experimental feed treatments. This conclusion is supported by the finding that the 100% plant protein diet fry were significantly smaller than the other four experimental feed treatments at the end of the twelve week trial ( Figure 2 13). Wheat protein isolate was intentionally selected as a low quality diet in that it is deficient in the essential amino acid lysine (1.6% by weight; NRC 1993). The two experimental diets ( 0% fishmeal: 100% wheat protein; 25% fishmeal : 75% wheat protein) with the lowest lysine content had 0.57%, and 1.03% lysine respectively by weight (certified analyses; ABC Laboratories Gainesville, Florida); lysine content of both diets were lower than the 1.43% lysine by weight minimum requirement established for tilapia (Aca nthopterygii: closest related species listed in Fish Nutritional Requirements, N RC 1993 ) be low which reduced growth occurs

PAGE 61

61 Final fry weights at the end of the 12week indoor feeding trial did not differ among the four diets that contained fishmeal as a feed ingredient; only the 100% wheat protein diet fry group exhibited an overall smaller size Although r educed feed palatability may have been a factor in the reduced growth of the 100% wheat protein fed treatment group, a lack of differences in fry survival among treatment groups (Figure 2 14) tends to contradict lower palatability as a possible reason for the reduced growth observed for the 100% wheat protein fed treatment group. Fry Survival Rates The absence of fry survival rate differences among f eed treatment groups at harvest suggests that overall nutritional status of fry did not differ among treatments. Although lower feed palatability and/or a mino acid deficiencies within the 100% plantprotein diet may have produced suboptimal growth of fr y within this treatment, these deficiencies were sub lethal. Indoor Trial Fry Isotope Signature Carbon isotope signatures of male broodstock, female broodstock, and 0 to 14day old fry did not significantly differ (Figure 2 15) This was to be expected as all three groups were fed the same commercial feed (TetraMin) and due to the typically low magnitude of carbon isotope fractionation on the order of 0.25 to 1.0 per trophic level ( DeNiro and Epstein 1978, 1981b, Gearing 1991, Fry 1991, Vander Zanden and Rasmussen 2001, Post 2002aa ). Fry were slightly enriched in C relative to the control feed ( C between feed and fry = 0.91 ) than both male ( C feed and male broodstock: 0.70 ) and female ( C feed and female broodstock: 0.75 ) broods tock. This was probably due to fry receiving a majority of their nutrition (lipids and carbohydrates) from maternally derived egg nutrients from their mothers (ovoviviparous

PAGE 62

62 species); this may have placed fry in to a hemi trophic level above their mother s, producing fry slightly enriched in their carbon isotope signature relative to both female and male broodstock. Only male broodstock and 0 to 14day old fry nitrogen isotope ( N ) signatures significantly differed, female broodstock did not differ from either male broodstock or 0 to 14dayold fry (Figure 215) Of the three groups, fry were the most isotopically enriched in 1 5N and were significantly enriched in 1 5N relative to the male broodstock group. Again, this was possibly due to the fry being a hemi trophic level higher than their mothers as they gained maternal nutrition in the form of nitrogenous proteins and amino acids. Follicle development likely involves the same isotopic fractionation processes (e.g. enrichment due to molecular diffus ion, enzyme kinetics, etc ) that occur during digestion, and may have caused fry to become isotopically enriched in the heavier nitrogen isotope relative to their mothers and the male broodstock (Vallowe 1957, Chong et al. 2004). 13C signatures of pre harvest fry from different treatments diverged ( Figure 2 17) as of the first ration adjustment weighing; fry continued to isotopically diverge and more closely resemble their respective feeds ( Figure 216) as the trial continued. At harvest, three fry pair wise comparisons did not differ in C signature ( Figure 218). The latter finding demonstrated that fry groups that had the same initial carbon isotope signature became isotopically distinct over time when fed isotopically distinct feeds; and that 13C signature s can be used to identify the nutritional source of a consumer group for a single trophic level (swordtails ). These results validated the isotope tracing methodology as a means of identifying the nutritional source of a particular fish group. 1 5N signatures of fry also diverged ( Figure 217) after the first ration adjustment period (day 10). Isotopically, fry increasingly resembled their respective feeds ( Figure 2 16), while

PAGE 63

63 becoming more distinct from each other. At harvest, all fry groups s ignificantly differed in N signature ( Figure 218) except for the same two treatments that did not differ for C signature, (75FM:25WP and 50FM:50WP), and (50FM:50WP and 25FM:75WP). This was further evidence that the indoor trial was successful in validating the stable isotope tracing method for identifying the trophic linkages between fish and their nutritional sources. At harvest, fry carbon isotope signatures differed from their respective feeds by an average of 0.64 13C ( Figure 219; range: 0.26 to 0.86 13C) This difference was well within the range of carbon isotope enrichment values reported in the literature (typically ~ 0.4 1.0 13C ) for a single upward step in trophic level (DeNiro and Epstein 1978, 1981b, Rodelli et al. 1984, Fry 1991). Fry nitrogen isotope signatures differed from their respective feeds by an mean of 3.03 15N ( Figure 2 20; range: 1.84 to 3.95 ), which also was well within the range of nitrogen isotope enrichment values ( 3.04.0 ) reported in the literat ur e for a single upward step in trophic level ( DeNiro and Epstein 1978, 1981a Minigawa and Wada 1984, Schoeninger and DeNiro 1984, Fry 1991). Both the observed carbon and nitrogen isotope enrichment results from the indoor twelve week swordtail fry feeding trial support the validity of using the stable isotope tracing method as a means of investigating trophic dynamics in an aquatic ecosystem (Glibert and Capone 1993). Elemental Carbon to Nitrogen Ratios Elemental Carbon to Nitrogen ratios have been used as a rough measure of an organisms health and nutritional state, with lower ratios tending to indicate that nitrogen (and protein) are not presently limiting to the organism (Val iela 199 5, Williams et al. 2007 ). Elemental carbon to nitrogen ratios of indoor fry at harvest did not differ among the experimental feed treatment

PAGE 64

64 groups ( Figure 2 22), indicating that fry nutritional states probably did not biologically differ among treatment groups and that all fish groups had approximately the same protein and energy [ lipid and carbohydrate ( glycogen) ] content. S upport ing the contention that all fry treatment groups were roughly consuming the same relative quantities of protein, carbohydrates and lipids (rations provided in excess of satiation ad libitum ) re gardless of the potentially lower palatability of the 100% wheat protein isolate experimental feed (Arndt et al. 1999, Yamamoto et al. 2007, Barrows et al. 2008). Lower growth of the 100% plant protein fed fry treatment group was likely due to a lysine deficiency present within the all plant flake feed diet (NRC 1993).

PAGE 65

65 Figure 21. G reenhouse physical layout, location of concrete burial vaults and life support system for swordtail fry production and broodstock growout. Figure 22. Epoxy coated concrete burial vault used to culture swordtail fry and broodstock in greenhouse (240 cm length 89.5 cm width 73.6 cm height: approx. 1,665 L vol ume ) Sump Vat Bead Filter Fluidized Bed (2.3 m height)Vat 1 Vat 11 Fan Fan Approximately 8500 Liter capacityRecirculated Well Water System UV Sterilizer Door Door Greenhouse Dimensions : 22 X 9.1 meters N pump 1 pump 2 70% shade cloth

PAGE 66

66 Figure 23. C age housing broodstock to minimize fry cannibalism newly spawned fry fell through mesh. A single c age holding 25 male and 75 female broodstock fish w as held within each epoxy coated concrete burial vault. Cage dimensions: 66 cm length 66 cm width 49.5 cm height Figure 24. C oncrete vault freshwater recirculation/partial flow through swordtail fry production and growout system in greenhouse

PAGE 67

67 Figure 25. Five isonitrogenous and isocalorific experimental flake feeds used in the indoor feeding trial (from left to right: 100% fish meal protein: 0% whe at protein ; 75% FM : 25% WP ; 50% FM : 50% WP ; 25% FP: 75% WP ; 0% FP: 100% WP ).

PAGE 68

68 Figure 26. P hysical layout and experimental design of the i ndoor feeding trial study. 1 2 3 4 8 8 8 9 1 ten fry per tank Indoor 3 month fry grow out trial: 5 treatments, 21 replicates, 105 tanks 4 fry (subsamples) used for isotopic analysis sump fluidized sand bed head tank bead filter

PAGE 69

69 Figure 27. I ndividual replicate aquarium (approx. 6.4 L) for indoor feeding and isotopic methodology validation trial. Figure 28. Indoor fry feeding and isotope methodology validation trial aquaria batteries. Each aquaria wa s approximately 6.4 L in volume and held 10 fry at trial inception.

PAGE 70

70 Figure 29. Virtis Sentry freeze dryer. Figure 210. S wordtails after 48 hrs in freeze dryer ( 57 C 1 104 Torr vacuum).

PAGE 71

71 Figure 211. D esi c cator used to store samples prior to weighing for dry weight and isotopic analysis sample preparation ( attachment hardware not present in photo) ; approxima tely 0.5 kg Drierite (calcium sulfate) dessicant lining bottom

PAGE 72

72 Figure 212. Mean swordtail fry ( 10 subsample fry / replicate aquaria) weight of randomly selected single replicate aquaria from each of five experimental feed treatments [% FM (fishmeal protein): % WP (wheat protein)] sacrificed at regular time intervals for daily ration feeding adjustment ; treatment mean of two replicates for 25 % FM: 75 % WP treatment. 0.00 0.05 0.10 0.15 0.20 0.25 Wet Weight (g) 100 FM:0 WP 75 FM: 25 WP 50 FM: 50 WP 25 FM: 75 WP 0 FM: 100 WP

PAGE 73

73 Figure 213. Mean swordtail fry weight ( 95 % CI) of five experimental feed treatments after 90 days (% fishmeal: % wheat protein isolate); differing letters denote significant differences between groups (P < 0.05, Tukeys multiple comparison test) 0.00 0.05 0.10 0.15 0.20 0.25 0.30 100%FM:0%WP 75%FM:25%WP 50%FM:50%WP 25%FM:75%WP 0%FM:100%WPWet Weight (g)Feed Composition b b b b a

PAGE 74

74 Figure 214. Swordtail f ry survival rates for five feed treatments after 12 weeks ( % survival 95 % CI), differing letters denote significant differences between groups (P < 0.05, Tukeys multiple comparison test) ANOVA statistical test was performed upon arcsine transformed data. Treatment s ample sizes (from left): 14, 13, 1 3, 15, and12. 0 10 20 30 40 50 60 70 80 90 100 100%FM:0%WP 75%FM:25%WP 50%FM:50%WP 25%FM:75%WP 0%FM:100%WP Survival Rate (%)Feed Composition a a a a a

PAGE 75

75 F igure 215. Isotopic signature s ( 95 % CI ) of swordtail broodstock and initial fry for indoor feeding trial; BSF (female broodstock), BSM (male broodstock), C = absolute value s ; differing letters denote significant differences between groups (P < 0.05, Tukeys multiple comparison test) Treatment sample sizes (from left): 11, 10, and 21. 0 5 10 15 20 25 BSF BSM initial fry 15N and 13C(abs) d13C (abs) d15N avg c c c ab a b

PAGE 76

76 Figure 216. Six flake feed 13C ( absolute value ) and 15N signature value s ( 95 % CI) ; differing letters denote statistical differences between treatments ( P < 0.05 Tukeys multiple comparison test ) within elements, C or N. TetraMin r egistered trademark of the Tetra Werke Company Treatment sample sizes (from left): 4, 4, 4, 4, 4, and 2. 0 5 10 15 20 25 100FP:0WP 75FP:25WP 50FP:50WP 25FP:75WP 0FP:100WP TetraMin 15N & 13C(abs) Feed Composition dC abs dN d d a b c e a b c d e f

PAGE 77

77 Figure 217. Mean pre harvest swordtail f ry isotope signature [ 13C (abs), 15N ] trajectories over 90 day indoor trial, single replicates for each date and treatment combination ( FM % fishmeal : WP % wheat protein isolate). 6 8 10 12 14 16 18 20 22 24 C(abs) and N 100FM:0WP dN 75FM:25WP dN 50FM:50WP dN 25FM:75WP dN 0FM:100WP dN 100FM:0WP dC 75FM:25WP dC 50FM:50WP dC 25FM:75WP dC 0FM:100WP dC

PAGE 78

78 Figure 218. Indoor trial fry isotope signatures [ 13C (abs) 15N 95 % CI ] at harvest (90 days) fry isotope signature values (day zero) ; differing letters denote significant differences between treatments (P < 0.05, Tukeys multiple comparison test) within elements, C or N Treatment sample sizes (from left): 6, 6, 6, 5, and 6. 0 5 10 15 20 25 100FM:0WP 75FM:25WP 50FM:50WP 25FM:75WP 0FM:100WP initial day0 13C (abs) & 15N feed composition d13C (abs) d15N avg a b bc d d c bc b a ab

PAGE 79

79 Figure 219. Carbon isotope signature differences ( 13C 95 % CI) between harvest fry and respective experimental feed treatments after 90 days; differing letters de note significant differences between groups (P < 0.05, Tukeys multiple comparison test). Treatment sample sizes (from left): 6, 6, 6, 5, and 6. 0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 100FM:0WP 75FM:25WP 50FM:50WP 25FM:75WP 0FM:100WP 13C (abs value) feed composition dC diff a a a a a

PAGE 80

80 Figure 220. Nitrogen isotope signature differences ( 15N 95 % CI) between harvest fry and respective experimental feed treatments after 90 days, differing letters denote significant differences between groups (P < 0.05, Tukeys multiple comparison test). Treatment sample sizes (from left): 6, 6, 6, 5, and 6. 0 1 2 3 4 5 6 100FM:0WP 75FM:25WP 50FM:50WP 25FM:75WP 0FM:100WP 15N dN diff dN diff a ab bc c c

PAGE 81

81 Figure 221. Cartesian plot of indoor fe eding trial swordtail juveniles (3 months) initial fry, male (BSM) and female broodstock (BSF) and five experimental feeds (% fm: % wp, fm fishmeal, wp wheat protein) and control feed (Tetra Min) 13C (abs) and 15N signature values ( 95 % CI) ; confi dence intervals omitted for swordtail initial fry, a nd broodstock for clarity. 15N 13C15 27 15 BSF BSM TetraMin 0 3 6 9 12 19 23 25 initial fry 100fm:0wp fry 75fm:25wp fry 50fm:50wp fry 25fm:75wp fry 0fm:100wp fry 100fm:0 wp 75fm:25 wp 50fm:50 wp 25fm:25 wp 0fm:100 wp Indoor Harvest Fry and Experimenta Feed Isotopic Plot

PAGE 82

82 Figure 222. Indoor fry elemental [C]/[N] ratios ( 95 % CI) at harvest after 90 days on five experimental feed s (FM % fishmeal, WP % wheat protein) treatments; differing letters denote significant differences between treatments (P < 0.05, Tukeys multiple comparison test). Treatment sample sizes (from left): 6, 6, 6, 5, and 6. 2 0 2 4 6 8 10 12 100FM:0WP 75FM:25WP 50FM:50WP 25FM:75WP 0FM:100WP C/Nfeed composition C/N a a a a a

PAGE 83

83 Table 2 1. Twelve week indoor swordtail (Xiphophorus hellerii ) fry feeding trial daily per capita fry ration level as a function of time day 0 to 10 ration based upon initial fry size at start of trial Days 0 10 Days 11 20 Days 21 30 Days 31 45 Days 46 60 Days 61 75 Days 75 90 15% ration (wet wt.) 15% r ation (wet wt.) 15% r ation (wet wt.) 10% r ation (wet wt.) 10% r ation (wet wt.) 10% ration (wet wt.) 10% ration (wet wt.) 0.00134 g feed/ fish/ day 0.0185 g feed/fish/day 0.0261 g feed/fish/day 0.0250 g feed/fish/day 0.0513 g feed/fish/day 0.0977 g feed/fish/day 0.1208 g feed/fish/day T reatment highest fry wt. 25%FM:75%WP 0%FM:100%WP 25%FM:75%WP 100%FM:0%WP 50%FM:50%WP 25%FM:75%WP calculation error resulted in lower feeding ration (still greater than 30 minute satiation level).

PAGE 84

84 CHAPTER 3 OUTDOOR POND TRIAL S WORDTAIL BIOMASS PRO DUCTION Consistently high production rates are the goal of all aquaculture producers. Only by producing large numbe rs of highquality livestock can fish farmers sustain profitable operations when faced with low profit margins, potentially catastrophic livestock losses due to inclement weather (outdoor operations) and overseas competition. Increasing labor costs, ope rations overhead (electricity, feed, and transportation fuel costs), land prices, and consolidation of the retail pet store market have reduced the profit margin for ornamental fish producers. The s wordtail (Xiphophorus helleri i ) is a popular ornamental fish in the freshwater tropical fish pet trade ( Figure s 32a c). In the United States, swordtails and other ornamental livebearers are typically raised within earthen ponds in southern Florida, primarily centered within Polk, Hillsborough, Miami Dade and Palm Beach counties ( FASS/ FDACS 2005 ), where air temperatures rarely drop below 16C (60C) and groundwater from surficial aquifers is relatively close to the surface. Profit margins for producers are typically small and large numbers of fish of acceptable appearance and health are required to provide sufficient profits for producers. A series of outdoor pond trials were performed to evaluate four common freshwater tropical fish species cultured using four different aquaculture pond management pract ices: processed commercial feed, unprocessed commercial feed, cottonseed meal ( organic fertilizer), and an inorganic fertilizer (M Krasilovsky unpublished data). In conjunction with one of these series of pond trials, involving the red wag swordtail (Fig ure s 32a c), isotopic analyses of planktonic assemblages within the ponds, as well as the applied feeds, fertilizers and fish were conducted to determine the fate of applied nutrients (feeds and fertilizers) and if pond treatment trophic efficiencies an d fish production were correlated

PAGE 85

85 Fish production rates were used to evaluate the efficacy of the four different pond (nutrient) management practices. Fish production differences among replicate ponds within management treatments, also were used to ev aluate management techniques as lower variation among replicate ponds is desirable. Production efficiencies of fish fed manufactured feeds and live feeds derived from increased phytoplankton production through the addition of organic and inorganic ferti lizers were measured by harvesting ponds after a threemonth grow out period. Actual production efficiencies also were compared with production estimat es predicted from trophic efficiencies determin ed from the isotopic signatures of the major food web components (Chapter 4). Methods Study Site The study was conducted at the Tropical Aquaculture Laboratory (TAL), ( University of Florida Institute of Food and Agricultural Sciences [UF IFAS] School of Forest Resources and Conservation), a satellite facility located in Ruskin, Florida, U.S.A. Twenty four adjoining rectangular experimental ponds were used for this study. E ach pond had similar physical dimensions ( Figure 3 1a d; 17.8 m length 8.6 m width 2 m avg. depth), surface area ( 0.015 ha ) and water volumes [~ 153,000 L ( 40,370 gallons ) per pond ] Due to the high water table (typically 50 cm below grade) and permeable soil, ponds were filled primarily from a surficial aquifer with only sporadic inflow of supplemental well water nec essary to maintain pond volume during the hotter summer months when higher rates of evaporation occur red The pond study began on 13 March, 2006, when ponds were initially drained and excess sediment and organic material was washed and removed from the substrate over a period of three days. Immediately following draining and washing, the pond substrate was sterilized with hydrated lime [Ca(OH)2] hand broadcast onto the pond bottom at approximately 800 grams per

PAGE 86

86 pond ( Figure 31a) ; lime application quickly raise s the pH of any remaining pond water ( Boyd 1974, 1995, 1997, Lin et al. 1997) and the pond substrate, killing many of the remaining organisms (potential predators, competitors and pathogens). Additionally, increased al kalinity provided by lime addition stimulates algal growth that support s the pond food web, supplement ing the nutritional needs of broodstock and fry ( Boyd 1995, 1997). Stocking occurred 14 days later on 27 March, 2006, when lower pond pH levels (< 8.0 pH ) were deemed amenable to fish stocking and sufficient manpower was available. Each pond was stocked with 250 (200 female: 50 male) redwag swordtail ( Xiphophorus helleri i ) broodstock ( Figure s 32a b) Due to the large number of broodstock required for t he study (6000), broodstock (> 4 cm S.L.) were obtained via seining and trapping from outside ponds that were initially stocked with red wag broodstock, and broodstock grown from fry kept in a greenhouse recirculation system A ll broodstock were obtained from the same genetic stock, but ages and nutritional histories differed substantially. T o account for these differences, prior to stocking, broodstock fish from the two sources were separated by sex and evenly distributed within ponds, to block for the e ffects of broodstock from different sources within the treatment ponds (Smart et al. 1998, 2001) The 24 ponds were subjected to four different nutrient treatments with six replicate ponds per treatment: (1) processed feed (PRO) (2) unprocessed feed (U NP) (3) cottonseed meal organic fertilizer (CSM ) and (4) nitrogen only (ammonium nitrate) inorganic fertilizer ( INO). U nprocessed feed was partially processed by milling, but differed from processed feed by not being heated during extrusion and further homogenized by regr inding. B y heating the processed feed during extrusion, starches within the feeds bec ome slightly gelatinized, increasing their digestibility and available caloric energy In addition, anti nutritional compounds (e.g., phytate,

PAGE 87

87 lectins, and trypsin inhibitors) become inactivated when carefully heated (Halver 1989 Brown et al. 1997) The two commercial feeds were formulated using identical ingredients which were chosen and present in quantities deemed typical of high quality commercial tropical ornamental fish feeds utilized by local ornamental fish farmers in the area. T hese feeds were specially manufactured for TAL by Purina Animal Feeds Corporation ( Table 3.1). Ponds designated as feed treatment ponds were broadcast fed t wice daily at dawn (approx. 300 g ~ 07:00 EST) and late afternoon (approx. 300 g ~ 17:00 EST), for a total of 600 g of feed per pond per day. Cottonseed meal fertilizer also was hand broadcast twice daily to their respective treatment ponds at the same ti mes that processed and unprocessed feeds were applied. Inorganic liquid fertilizer (inorganic liquid fertilizer 11: 0 : 37 N:P:K ) w as administered according to weekly Secchi disk readings and additional fertilizer w as applied if Secchi depths were less than 45 cm (Boyd 1979, 1995, Lin et al. 1997, Tepe and Boyd 2003) just after midday (13:00 EST). When dictated by Secchi depth values, i norganic fertilizer was applied at a rate of 4 kg/ha NH4N03: 8 kg/ha P2O5 Fertilizers were applied to ponds to ensure the maintenance of a green water algal bloom to support pond food webs, as Secchi depths have been shown to correlate well with chlorophyll [a] concentrations in low turbidity/low color waters and were used to monitor phytoplankton abundance (Florida Lakewatch 2001). Water Chemistry and Physical Environment Temperature and d issolved oxygen concentrations were measured for each pond prior to t he t wice daily feedings (dawn and late afternoon) throughout the t welve weeks of the outdoor feeding trial Standard water chemistry parameters: t otal alkalinity (mg/L CaCO3 equivalent), total hardness (mg CaCO3 equivalent), pH, total ammonia nitrogen, nitrite, nitrate, total phosphorous and chlorophyll [a] were measure d weekly as part of another study ( M. Krasilovsky unpublished data).

PAGE 88

88 Pre Harvest Fry Collection Fry were collected weekly during the first three weeks (21 days) of the study, and were collected biweekly thereafter; fry were collected on days 7, 14, 21, 35, 49, 63, 77, and 84 of the 90day experimental trial. Fry were collected from within the pond with a 74 cm (width) x 5 1 cm (length) 38 cm (bag depth) rectangular dip net with 1 mm square mesh. Fry were collected beginning at 10:00 hrs on the day of collection, fry were collected while walking through the shallow water and emergent weeds growing along the perimeter of each pond while periodically thrusting the net into the shallow weeds where fry were known to congregate. For each pond, a minimum of ten fry were placed into pre sterilized (Whirlpak 2 ounce nominal size) polyethylene bags and placed into an insulated cooler lined with in an ice water slurry U pon return to the laboratory f ry were frozen and stored at ( 4 C) Fry were stored frozen in an ice block (distilled water added) within their polyethylene sample bags to prevent dehydration from water ice sublimation freezer burn for six to eight months prior to length and weight measurements. Fry were thawed prior to length measurement an d gastrointestinal tracts and all internal organs were removed prior to being weighed (wet weight to the nearest 0.001 gram ) to reduce the bias of recently ingested prey organisms during isotopic analyses of fry. Fry length and weight measurements were m ade for a minimum of six fry from each pond for each of the eight sampling dates (~ 1,152 samples). After GI tract removal, and length and weight measurements, fry were individually placed upon preweighed aluminum foil weighing boats and placed into a 70 C ( 4 C) ultra cold freezer (Revco Elite ultra low temperature freezer ) for 12 to 24 hrs prior to being freeze dried at 57 C (Virtis corporation model Sentry 16 freeze dryer) for 16 to 48 hrs depending upon specimen size. Fry dry weights were measured and fry were then stored in sterile, pre cleaned, 0.7ml polyethylene microcentrifuge tubes.

PAGE 89

89 Harvest Fish Collection The pond fry grow out trial lasted for 90 days, and was terminated beginning the week of 18 June, 2006, with the initiation of pond harvesting. The application of all pond treatments ceased two days prior to the initiation of harvest. Swordtails were harvested over the course of four days using a 7.6 m length 1.8 m height 4 mm 4 mm square mesh multifilament nylon beach seine with weighted lead line ( Figure 3 3a). During harvest, two people on opposite banks dragged the seine along the long axis of the pond to the opposite bank where fish were concentrated and placed into 19L plastic buckets approximately half filled with pond water. Swordtails were immediately transferred to large insulated coolers filled with aerated well water and transported to the TAL quarantine building, where fish were placed into large epoxy coated concrete vaults (240 cm length 89.5 cm width 73.6 cm height; approx. 1,665 L volume) approximately filled with aerated well water; fish were then separated from debris and pest species (crayfish, tadpoles, insects, undesirable fish) and sorted into two size classes (small mm SL and large > 31 mm SL) using a hand operated floating grate sorter ( Figure 3 3b). Three subsamples (when possible) of 300 fish were taken from each of the two size classes for each pond, and subsample live wet weights, this allowed subsample and treatment mean fish live wet weight to be calculated ( X = subsample fish live wet weight grams/300 fish) for each fish size class. Additionally, smaller subsamples were taken from pond subsamples to determine individual morphometric mean swordtail standard lengths (SL) and live wet weights of the two size classes for each pond; small numbers (~ 10 20) of fish were also frozen for eviscerated dry weight and isotopic determination. Statistical Analyses Comparisons among treatments and s ampling dates were performed using twoway repeated measures ANOVA due to the nonindependent nature of repeated samplings from

PAGE 90

90 replicate ponds ( 2 tailed critical = 0.05). Comparisons among treatments for time averaged data were performed using one way ANOVA ( 2 tailed critical = 0.05). Post hoc tests were performed using the Bonferroni post test (twoway repeated measures ANOVA; 2 tailed critical = 0.05) and Tukeys multiple comparison test (one way ANOVA; 2 tailed critical = 0.05). Results Pre harv est Fry Growth Pre harvest pond fry were sampled periodically during the 12 week outdoor grow out trial: on days 7, 14, 21, 35, 50, 63, and 77 post stocking. Eviscerated fry dry weights generally increased over the 12 week trial (Figures 3 4 3 8). Nutrient treatment was not a significant factor in determining fry dry weight (2way repeated measures ANOVA: P(nutrient) = 0.2362, F(nutrient) (2,3) = 0.05), but time was significant ( Figures 34 3 8; 2 way ANOVA: P(time) < 0.0001, F(time) (2,6) = 0.05); to perform repeated measures analyses, two missing data points of 168, were replaced with their respective treatment and date mean values in order (PRO pond B10, 17 May 2006, and INO pond B9, 17 May 2006). Significant pairwise differences in dry fry weight and sampling date were found within individual pond nutrient treatments, (Figures 35 38; P < 0.05, Bonferroni post hoc test). Fish Production at Harvest Total fish biomass at harvest differed greatly among treatments ( Figure 3 9; 1way (2,3) = 0.05). PRO pond mean fish biomass was the highest X = 8.56 kg ( 6.41 to 10.71 kg 95 % CI ), followed in order from highest to lowest by UNP ponds X = 7.25 kg ( 6.29 to 8.20 kg 95 % CI), CSM X = 5.25 kg ( 4.79 to 5.70 kg 95 % CI ) and INO X = 3.75 kg ( 2.82 to 4.67 kg 95 % CI) ponds. PRO treatment ponds did not differ significantly

PAGE 91

91 from UNP ponds, but both di ffered significantly from CSM and INO ponds, which did not significantly differ from each other. Mean fish numbers at harvest significantly differed among treatments (1 way ANOVA, P (2,3) = 0.05). PRO pond mean fish numbers were the highest X = 20,290 ( 15,940 to 24,650 95 % CI ), UNP X = 17,490 ( 14,870 to 20,110 95 % CI ) and INO X = 17,460 ( 13,750 to 21,180 95 % CI ) ponds had the next highest and CSM treatment ponds had the lowest mean number X = 13,180 ( 9,990 to 16,370 95 % CI ). However, due to large within treatment variations in pond total fish numbers, post hoc pairwise comparison tests found only one significant pa irwise difference in total fish numbers among nutrient treatments, PRO v. CSM pond treatments ( Figure 3 10) Mean marketable ( SL > 31 mm) fish numbers differed markedly and significantly among treatments (Figure 311; 1 (2,3) = 0.05). PRO treatment ponds had the highest mean number of large fish X = 2,446 ( 1790 to 3100 95 % CI ), followed in order by UNP treatment ponds X = 1,919 ( 1540 to 2300 95 % CI ), CSM ponds X = 1,506 ( 1360 to 165 0 95 % CI ) and INO ponds X = 424 ( 277 to 570 95 % CI ). Post hoc pairwise comparisons indicated that PRO treatment ponds had significantly more marketable fish than either of the fertilizer pond treatments but did not differ from the UNP treatment, which had more marketable fish than the INO treatment but did not differ from the CSM treatment ; INO treatment numbers were significantly lower than those of the other treatments (Figure 3 11; P < 0.05). Mean unmarketable ( SL ( Figure 3 12; 1 (2,3) = 0.05). PRO ponds X = 17,850 ( 14,190 to 21,510 95 % CI ) and INO ponds X = 17,040 ( 13,490 to 20,590 95 % CI ) had the

PAGE 92

92 highest numbers of small fish, followed by the UNP ponds X = 15,570 ( 12,700 to 18,450 95 % CI ) and lastly by the CSM ponds X = 11,670 ( 8,410 to 14,940 95 % CI ). Post hoc pairwise tests indicated that t he PRO and INO ponds had significantly greater numbers of small fish than the CSM ponds, and the UNP pond numbers did not significantly differ from any other treatment ( Figure 3 12; P < 0.05) Large fish wet weight (SL > 31 mm) mean s at harvest did not significantly differ among treatments (Figure 313; 1(2,3) = 0.05). PRO pond large fish mean wet weight X = 1.82 g ( 1.62 to 2.02 g 95 % CI ) was similar to UNP X = 1.71 g ( 1.52 to 1.90 g 95 % CI ), CSM X = 1.75 g ( 1.52 to 1.90 g 95 % CI ) and INO X = 1.87 g ( 1.39 to 2.36 g 95 % CI ) large fish mean weights W eight variation ( 95 % CI ) was greatest for the INO pond large harvest fish than for those of the other three treatments ( Figure 313 ). The individual large harvest fish (SL > 31 mm) wet weight coefficient of variation values differed markedly among tr eatments (Figure 3 15). Small fish wet weight (SL mean s at harvest also did not significantly differ among treatments ( Figure 314: 1 (2,3) = 0.05). PRO pond small fish at harvest had a mean wet weight X = 0.23 g ( 0.33 to 0.14 g 95 % CI ) slightly lower than the UNP pond small fish X = 0.28 g ( 0.36 to 0.20 g 95 % CI ) and CSM pond small fish X = 0.28 g ( 0.36 to 0.21 g 95 % CI ), which had identical mean weights and the INO pond small fish had the lowest mean wet weight X = 0.18 g ( 0.25 to 0.12 g 95 % CI ). Discussion Pre harvest Fry Growth Mean fry weights increased over the 12week duration of the outdoor pond trial for all four pond nutrient treatments ( Figure 34). However, fry mean weight trajectories were uneven

PAGE 93

93 and mean fry weights dropped between the fourth and fifth weeks for the two fertilizer treatments and the fifth and seventh weeks for the two feed treatments (Figures 3 4 38). Th ese drops in mean fry weight w ere likely due to an increase in overall reproductive output of adult females. Although fecundity increases as food supplies and temperatures increase (Milton and Arthington 1983, Dawes 1991, Hutchings 1993, Hopkins and Unwin 1997, Tamaru et al. 2001), the most likely factor impacting fry recruitment is th at large numbers of older female fry had become sexually mature during these time periods. In outdoor farm ponds in Hawaii, female fry produced at the beginning of the grow out season reached first age of reproduction at approximately 68 weeks of age (Tamaru et al. 2001); depending upon stocking density, food availability, photoperiod, temperature and other environmental factors A ge at first reproduction for swordtails has been reported by other investigators to occur between six to twelve weeks of age ( Sohn 1977, Dawes 1991, Dussault and Kramer 1981, McKenzie et al 1983, Milton and Arthington 1983, Campton and Gall 1988, Tamaru et al. 2001). These reported time frames (6 8 weeks) coincide d with the decrease in pre harvest fry average weight observe d within th is study ( Figure 34). Pre harvest fry samples were made of numerous cohorts as fry were continuously produced throughout the 12 week trial ; based upon the observation of large variations in fry sizes (and presumably ages) within the periodic fry samples, and extensive knowledge of swordtail reproductive biology ( Dawes 1991, Axelrod and Wischnath 1991, Tamaru et al. 2001 ). Mean fry weight increase was consistent with a logistic or Gompertz trajectory, but actual fry growth patterns (and instan taneous growth rates) could not be determined due to continuous spawning within the ponds that resulted in large numbers of fry of differing weights and ages (Cailliet et al. 1996).

PAGE 94

94 Adult female reproductive output increases with i ncreasing water tempe ratures (Dawes 1991, Tamaru et al. 2001). Increased water temperature decreases the age at first reproduction of female fry, and increases adult female fecundity with increasing water temperature s up to 29C ( Tamaru et al. 2001) Both air and pond surface water ( 17 31C ) temperatures rose dramatically as the trial progressed from mid March to mid June (Figure 3 16) Total pond producti on also was likely enhanced by the effects of higher temperatures and longer photoperiods on phytoplankton production rates (Fogg and Thake 1987, Boyd 1979); increased primary production has been shown to le a d directly to increased zooplankton production, both higher phytoplankton and zooplankton prey standing stock s are factors known to increase adult female swor dtail fecundity (Kruger et al. 2001, Chong et al. 2004). Harvest Fish The two feed treatments had the highest mean harvest biomass yields, and had significantly higher fish yields than the two fertilizer pond treatments (Figure 3 9). Additionally, th e two feed treatments individual large harvest fish weight coefficient of variation values were much lower than those of the two fertilizer treatments (Figure 3 13), indicating that individual fish weight variations were much higher within the two fertiliz er treatments. High total fish biomass at harvest is a desirable performance aspect for a conveniently applied pond nutrient. Another is large numbers of commercially marketable fish of sufficient and uniform size (usually total length), and appearance High variation in size often indicates food limitation and/or overstocking ( Figure 315), which may result in increased intraspecies aggression and social hierarchies that may create and amplify size disparities among or within cohorts ( Hannes and Franc k 1983, Hannes et al. 1984, Franck et al. 1985 ) Additionally, from a marketing perspective size and color inconsistencies are extremely undesirable to potential

PAGE 95

95 ornamental tropical fish buyers and retailers and additional stress from handling and sorti ng is detrimental to appearance and increases fish mortality ( Karplus et al. 2003). Total fish numbers at harvest differed only between the processed feed and cottonseed meal treatments ( Figure 3 10). I norganic treatment total harvest did not significantly differ from the processed feed treatment which had the highest number of fish among the four treatments. T he INO treatment produced high fish numbers but had the lowest mean fish biomass yield of the four treatments, resulting in a general ly stunted appearance of the fish within this treatment. PRO ponds had the highest number of large marketable fish (SL > 31 mm), PRO treatment mean large harvest fish numbers were significantly greater than those of the two fertilizer treatments UNP pond marketable fish numbers only differed significantly from the INO ponds, which had significantly fewer marketable fish than the other pond treatments (Figure 311). PRO ponds also had the highest number of unmarketable fish ( followed cl osely by the UNP and INO treatments ; the CSM ponds had significantly fewer smaller fish than both the PRO and INO ponds ( Figure 312). The fact that INO ponds had the lowest total fish biomass, fewest large marketable fish and high numbers of small unmar ketable fish again attests to the generally stunted small fish that were harvested from these ponds. Large fish ( > 31 mm SL) mean weights did not differ among treatments at harvest (Figure 313). Small fish ( mean weights also did not differ among treatments at harvest ( Figure 3 14). However, s mall fish mean weights varied among treatments more than their large fish counterparts (Figures 313 314) Greater weight variation among individual small harvest fish within the INO treatment ponds, relative to those of the other three treatments, may have been due to lower food availability within INO treatment ponds Large numbers of small fry

PAGE 96

96 were able to escape through the 4 mm 4 mm square mesh of the seine which may have slightly skewed the size frequency data results. Water Chemistry and Physical Environment Water t emperature increased steadily throughout the 12 weeks of this outdoor pond feeding trial. Morning oxygen readings decreased slightly and lat e afternoon oxygen readings increased as pond heterotroph biomass, algal biomass, photoperiod, and water temperature increased from early spring to late spring/early summer (March June). Oxygen levels were typically at their daily minimums during the earl y pre dawn hours due to the respiration of pond organisms and lack of photosynthesis during nighttime hours, and increased to saturation ( oxygen) or near saturation levels during the late afternoon (~ 16:00 hrs EST) depending on cloud cove r. Feeding was postponed when oxygen concentration levels were below ~ 2.0 mg/L to prevent stress from potential ly hypoxic conditions. Depending upon duration and severity of hypoxic and anoxic conditions pond fish may become stressed and susceptible to disease or suffer mortality. Low oxygen conditions were only present on four occasions and did not appear to cause any noticeable disease or mortality events. Water chemistry parameters such as pH, hardness and alkalinity were relatively stable during the trial. This was not unexpected as the surficial aquifer ground water source in this region flows through a karst geology that imparts high alkalinity to the water due to high concentrations of dissolved CaCO3 and MgCO3 Total ammonia nitrogen (TAN: N H3, NH4 +) and nitrite levels also were unremarkable (TAN typically < 2.5 mg/L NO2 < 0.05 mg/L ), but were sporadically higher within the INO treatment (Michael Krasilovsky unpublished data) ; most likely due to the weekly introduction of ammonium nitrate f ertilizer, which immediately ionize s to ammonium (NH4 +) and is subsequently oxidized to nitrate.

PAGE 97

97 Figure 31a d. Clockwise from top left: (a) drained and washed pond, hydrated lime applied to substrate visible as white powder, (b) diesel tractor engine pump used to drain ponds in background, smaller wash pump in foreground (c) pond filled with a combination of both groundwater and well water, (d) ponds have been refilled and are ready for stocking once pH levels have stabilized (~7 8 pH) and plankton blooms have become established, steel frame for bird netting/predator control. d c b a

PAGE 98

98 Figure 32a c Clockwise from top: (a) adult female and male redwag swordtails ( Xiphophorus hellerii ) held within large polyethylene tubs (~ 280 L) within a large greenhouse, (b) adult female and male red wag swordtails (note conspicuous sword fin protrusion on ventr al portion of caudal fin), (c) male swordtail ( Xiphophorus hellerii ) internal anatomy (Source: H.E. Evans in Aquariology: A natomy, Genetics, and Breeding, Tetra Press 199 2). b c a

PAGE 99

99 Figure 33a b. Pond harvest equipment: (a) close up of 7.6 m length 1.8 m height beach seine (4 mm 4 mm square mesh nylon multifilament), (b) floating fish grader, variety of removable grates insert into floating frame, inter tine distance determines fish size/weight. b a

PAGE 100

100 Figure 34. Pre harvest fry mean dry weight s, periodically sampled over 12 week pond trial (2006), n= 6 ponds per treatment. 0.00 0.01 0.02 0.03 0.04 0.05 Eviscerated Dry Weight (g) pro avg. unpro avg. cotton avg. inorg avg.

PAGE 101

101 Figure 35. P rocessed feed (PRO) treatment mean fry dry weight ( 95% CI) sampled over 12 week trial (2006) six replicate ponds for each data point ; differing letters deno te statistical differences (P < 0.05). 0.00 0.01 0.02 0.03 0.04 0.05 0.06 Eviscerated Dry Weight (g)a ab b bc ab bc d

PAGE 102

102 Figure 36. Un processed feed (UNP) treatment mean fry dry weight s ( 95% CI) sampled over 12 week trial (2006) six replicate ponds for each data point ; differing letters denote statistical differences ( P < 0.05). 0.00 0.01 0.02 0.03 0.04 0.05 0.06 Eviscerated Dry Wight (g)a ab b c bc bc c

PAGE 103

103 Figure 37. Cottonseed meal (CSM) treatment fry mean dry weight s ( 95% CI) sampled over 12 week trial (2006) six replicate ponds for each data point ; differing letters denote statistical differences ( P < 0.05). 0.00 0.01 0.02 0.03 0.04 0.05 0.06 Eviscerated Dry Weight (g)d ab c bc bc cd a

PAGE 104

104 Figure 38. Inorganic fertil izer (INO) treatment mean fry dry weight s ( 95% CI) sampled over 12 week trial (2006) six replicate ponds for each data point ; differing letters denote statistical differences ( P < 0.05). 0.00 0.01 0.02 0.03 0.04 0.05 0.06 Eviscerated Dry Weight (g)a abc bc ab bc c c

PAGE 105

105 Figure 39. Mean fish biomass (g 95% CI) at harvest, differing letters denote significant differences between treatments (P < 0.05, Tukeys multiple comparison test ). Sample sizes: PRO (5), UNP (6), CSM (6), and INO (6). 0 2000 4000 6000 8000 10000 12000 PRO UNP CSM INOFish Biomass (g)a a b b

PAGE 106

106 Figure 310. Mean fish numbers (#s 95% CI) at harvest differing letters denote s ignificant differences between treatments (P < 0.05, Tukeys multiple comparison test ). Sample sizes: PRO (5), UNP (6), CSM (6), and INO (6). 0 5000 10000 15000 20000 25000 30000 PRO UNP CSM INO Fish Numbersa ab ab b

PAGE 107

107 F igure 311. Mean marketable (> 31 mm SL) fish numbers ( 95% CI) at harvest differing letters denote significant differences between treatments (P < 0.05, Tukeys multiple comparison test). Sample sizes: PRO (5), UNP (6), CSM (6), and INO (6). 0 500 1000 1500 2000 2500 3000 3500 4000 PRO UNP CSM INOFish Numbersa b bc c

PAGE 108

108 Figure 312. Mean unmarketable ( 31 mm SL) fish numbers ( 95% CI) at harvest dif fering letters denote significant differences between treatments (P < 0.05, Tukeys multiple comparison test). Sample sizes: PRO (5), UNP (6), CSM (6), and INO (6). 0 5000 10000 15000 20000 25000 PRO UNP CSM INOFish Numbersab b b a

PAGE 109

109 Figure 313. Mean individual marketable ( > 31 mm SL) fish weight s ( 95 % CI) at harvest minimum 25 sub sample fish from each replicate pond (data courtesy of M. Krasilovsky); differing letters denote significant differences between groups (P < 0.05, Tukeys multiple comparison test ). Sample sizes: PRO (5), UNP (6), CSM (6), and INO (6). 0 0 0 5 1 0 1 5 2 0 2 5 PRO UNP CSM INO Wet Weight (g)a a a a

PAGE 110

110 Figure 314. Mean individual unmarketable ( 31 mm SL) fish weight s ( 95 % CI) at harvest minimum 25 fish from each replicate pond (raw data courtesy of M. Krasilovsky); differing letters denote significant differences between groups (P < 0.05, T ukeys multiple comparison test ). Sample sizes: PRO (5), UNP (6), CSM (6), and INO (6). 0 0 0 1 0 1 0 2 0 2 0 3 0 3 0 4 0 4 PRO UNP CSM INO Wet Weight (g)a a a a

PAGE 111

111 Figure 315. Large harvest fish (SL > 31 mm) individual fish wet weight coefficient of variation (sd/mean) values among applied pond nutrient treatments minimum 25 subsample fish morphometric values from each replicate pond ( data courtesy of M. Krasilovsky ) Sample sizes: PRO (5), UNP (6), CSM (6), and INO (6). 0 00 0 05 0 10 0 15 0 20 0 25 0 30 PRO UNP CSM INO Coefficient of Variation

PAGE 112

112 Figure 316. A ir temperature daily maximum and minimum (C) from 1 March to 30 June 2006 for Ruskin, Florida, USA ( 27 4257.72 N, 82 15.70 W ); data provided courtesy of the National Weather Service, Tampa Bay Area Station, Ruskin, Florida. 0 5 10 15 20 25 30 35 40 Daily Air Temp C max min

PAGE 113

113 CHAPTER 4 OUTDOOR POND APPLIED NUTRIENT ISOTOPE TRA CING TRIAL In outdoor pond trial s, commercial feed and fertilizer stable isotope signatures were used to follow nutrient flows within aquaculture pond trophic systems. Fish and plankton assemblage stable isotope signature determinations allowed the identification of potential nutritiona l sources utilized by fish and various planktonic taxa, evaluate the relative importance of potential nutrient sources, and describe trophic processes occurring within the different treatments ( Harrigan et al. 1989, Fry 1991, Kling and Fry 1992, Gu et al. 1996, Lochmann and Phillips 1996, Hansson et al. 1997, Hoyer et al. 1998, Jones and Waldron 2003, Grey et al. 2004, GregoryEaves et al. 2007 ) C ommonly applied nutrients (feeds, organic and inorganic fertilizers ) were applied to four pond treatments in a manner amenable to investigating research hypotheses rather than designed to maximize fish production and/or mimic typical commercial fish produc tion methods Fertilizers and feeds are typically applied in tandem to ornamental fish production ponds but to reduce confounding factors in this study, fertilizers and feeds were used individually within designated treatment ponds after stocking brood fish Feeds and meals are commonly applied once/twice daily to commercial ponds near the study site (centr al and southern Florida); liquid fertilizers are typically applied either on a periodic basis (e.g., bi weekly, weekly, semi weekly), and/or on an asneeded basis depending upon general pond appearance (absence, magnitude or even color of a conspicuous al gal bloom). Cumulative labor expenses from daily applications of feeds and meals can become a major expense over the course of a single production cycle (M. Krasilovsky unpublished data). Cotton ( Gossypium hirsutum ) is a broadleaved, semi tropical, C3 di cotyledonous plant commonly grown within the southern and western United States (National Cotton Council of America website: cottonusa.org). Cottonseed meal is a high protein byproduct of cotton

PAGE 114

114 processing, derived from cotton seeds and is a commonly appl ied organic fertilizer in aquaculture ponds within the southern United States ( Teicher Coddington et al. 1997 ). In outdoor pond trials, isotopic compositions of fish, zooplankton, phytoplankton, and applied nutrients (feeds and fertilizers) were evaluate d to determine nutrient fates (nutrient sinks) following nutrient application/inputs (two feeds, two fertilizers). In ponds that received commercial feeds, the lack of matching isotope signatures between fish and commercial fish feed inputs would indicate a lack of trophic coupling (0 % energy/nutrient transfer), and completely matching isotope signatures would indicate an ideal 100 % energy/nutrient transfer between the fish and applied feed ( Lochmann and Phillips 1996, Lochmann et al. 2001). In ponds that only receive fertilizer closer isotopic signatures between fish and fertilizer should indicate closer trophic coupling of fish to fertilizers. If fewer trophic levels separate fertilizers and fish, less undesirable organism production ( e.g., crayfish, insects, tadpoles, and snails) and/or nutrient losses (e.g., mineralization, sediment losses, chemical precipitation due to redox conditions, and atmospheric outgassing) would be expected. This would likely indicate that applied nutrients assimilated by fishes were moving through fewer trophic intermediaries. With higher energy/nutrient transfer efficiencies, greater fish production should be observed relative to systems with lower energy/nutrient transfer efficiencies due to greater numbers of trophic intermediaries and/or sources of energy/nutrient loss (high pest species production, sediment losses, atmospheric volatilization, etc .), barring unforeseen losses from dise ase, predation, catastrophic weather events, etc. In addition to pond fish and applied nutrient isotope signature measurements, periodic isotopic signature measurements of pond plankton fractions (small, medium and large plankton assemblages) also were m ade. Isotope signatures were used to determine trophic positions of

PAGE 115

115 plankton assemblages (potential fish prey) fish fry and small and large harvest fish with in the different treatments. In addition to determining trophic positions, primary sources of n utrition for harvest ed fish and plankton size assemblages were identified, as well as the relative contributions (% basis) of plankton assemblages and applied nutrients to swordtail diets within the respective treatments. The majority of aquatic, marine, a nd terrestrial food web studies that have utilized stable isotope tracers, have traced two biologically important elements, carbon and nitrogen within the ecosystem(s) of interest (Minigawa and Wada 1984, Fry et al. 1999, Frazer 19 96, Gu et al. 2001). Nat ural abundances vary ( DeNiro and Eptein 1978, 1981a, Fry 1991) for the stable isotopes of carbon [ 98.89 % [12C] and 1.11 % [13C]) and nitrogen (99.63 % [14N] and 0.37 % [15N]). The relative abundances of carbon and nitrogen isotopes vary within ecosystems according to primary producer origins (plant species/physiology), ecosystem type (terrestrial/aquatic/marine/estuarine), inorganic nutrient form [atmospheric carbon and nitrogen, or DIC and DIN in solution] and organism/taxa trophic position (1 producer, 1 consumer, 2 consumer, 3 consumer, herbivore, carnivore, omnivore) among others ( Rounick and Winterbourn 1986, Peterson and Fry 1987, Takahashi et al. 1990, Fry 1991, Vander Zanden and Rasmussen 2001, Hall 2004, GregoryEaves et al. 2007). Carbon isot ope signatures are primarily used to identify food web base nutrient source(s) due to the high degree of carbon isotope signature conservation among primary producer(s) and their consumers within an ecosystem. Carbon isotope signatures typically undergo l ittle isotopic change between trophic levels (enrichment ~ 0.0 1.0 13C), allowing basal nutrient sources within an ecosystem to be fairly easily identified (DeNiro and Epstein 1978, 1981b, Estep and Vigg 1985, GregoryEaves et al. 2007).

PAGE 116

116 Nitrogen iso tope signatures are primarily used to determine trophic relationships among organisms within an ecosystem ( Peterson and Fry 1987, Wada 1980, Hideshige and Wada 1990, Fry et al. 1999, Vander Zanden and Rasmussen 1999, Yoshii et al. 1999, Frazer 1996 Gu et al. 2001, Fry 2006, De Brabandere et al. 2007). Within an ecosystem, nitrogen isotope enrichment magnitudes (typically 3.4 15N) between trophic levels are typically much greater than those occurring for carbon (DeNiro and Epstein 1978, 1981a ). This allows nitrogen isotope signatures (measured via 1 4N and 1 5N content) to be used to determine an organisms relative trophic position (1 producer, 1, 2, and 3 consumer, etc .) as well as the number and identity of trophic level intermediaries between a n organism or biological material of interest ( e.g. excreted urine or feces) and the ecosystems basal nutrient(s); if a thorough survey of ecosystem taxa (e.g., fish, zooplankton, phytoplankton, applied nutrients, etc .) nitrogen isotope signatures is undertaken ( DeNiro and Epstein 1981, Minigawa and Wada 1984, Estep and Vigg 1985). Similar to carbon isotope tracers, nitrogen isotope tracers can also be use d to identify presumed nutrient base (s) within a food web, but with a much lower degree of accuracy Within the outdoor pond trial, t he primary working hypothesis regarding the two commercial feed treatments, was that fish within these ponds were consuming and assimilating applied feeds largely to the exclusion of available l 4ive foods. Carbon and nitrogen contained within feeds would become sequestered within their respective treatment fish and the movements of these nutrients could be traced and their assimilation verified using stable isotope analyses techniques. The primary working hypothesi s regarding the two fertilizer treatments was that nutrients containing carbon and nitrogen would pass through a number of trophic guilds/intermediaries before becoming sequestered within the fish.

PAGE 117

117 A primary goal of the study was to examine the relative magnitudes of enrichment occurring within ponds from applied nutrients to major planktonic guilds and ultimately fish, not the actual isotope signature values ( C and N) of the taxa within each pond treatment. Larger isotopic differences between fertilizer nutrients and fertilizer treatment fish relative to commercial feeds and feed treatment fish were expected, and these isotopic differences also were expected to be inversely related to target species ( Xiphophorus hellerii ) biomass producti on at harvest Reasonable intermediate isotopic signature values for the postulated food chain guilds were viewed as validation of the primary working hypotheses concerning food web/chain structure, food chain length, and fish production efficiencies among the four pond nutrient treatments. Methods Plankton Sample Collection Weekly water samples were taken from each of 24 aquaculture ponds (Chapter 3 methods) for 12 weeks and analyzed colorimetrically for basic water chemistry (pH, NO2, TAN, hardness alkalinity ; Hach Laboratories: www.hach.com ), and chlorophyll [a] concentrations. Duplicate weekly water samples were collected to determine the isotopic signatures (carbon and nitrogen) of three plankton size assemblages [ large ( > 200 m ) ; medium (32 200 m ) ; small (1 32 m ) ; DeVries and Stein 1991] and taxonomic composition and relative abundances of the large plankton (> 200 m) assemblages within the ponds of the four (processed feed, unprocessed feed, cottonseed meal organic fertilizer, and liquid inorganic fertilizer) applied nutrient treatments (Chapter 5). An integrated water column sampler ( Figure 42; 35 mm I.D. schedule 40 PVC pipe x 3.05 m length) with a gravityoperated one way flap valve at the anterior to prevent sample loss was used to combine four pond subsamples (one from each bank) into one aggregate sample for each pond. The sampler was inserted (valve end first) into each

PAGE 118

118 pond obliquely from the bank and care was taken to not touch the bottom in order to avoid entraining bott om sediment, detritus and benthic organisms S amplers and collection buckets were rinsed with an initial sample of pond water from the target pond which was then discarded to prevent cross contamination of samples from previously sampled ponds. Collected water was mixed vigorously and two one liter samples were drawn from the aggregate sample for each pond for isotopic and taxonomic plankton analyses (Chapter 5) additionally, one 100ml sample was taken for chlorophyll [ a ] and total nitrogen and phosphorous content analys e s ( M. Krasilovsky unpublished data) One set of one liter pond water samples was immediately fixed with Lugols solution preservative (APHA /AWWA 1999) and stored within light excluding containers for later filtering into plankton size assemblage fractions for microscopic analysis. The remaining unfixed one liter pond water samples were kept refrigerated (approx. 4 C) after collection and seria lly filtered within 24 hrs through the following mesh sizes : 200 m, 35 m and 1 m R etained particles from each sieve were vacuum filtered onto a Whatman GF/C pre combusted glass fiber filter disk (47mm diameter), placed into individual 58 ml ( 2oz. nominal Whirlpak) presterilized polyethylene bags and immediately frozen at 4 C ; sample disks were typically frozen for eight to twelve months prior to freeze drying. Prior to freeze drying, frozen s amples were transferred to individual conical tubes and frozen in an ultra low temperature freezer (~ 74 2 C) for 1624 hours Individual glass fiber filter disks were freeze dried for 24 hrs at approximately 56 to 60 C. After freeze drying, individual filter disks were carefully folded with forcep s and placed into sterile 1.7 ml microcentrifuge tubes ; forceps were swabbed with alcohol (70% ethanol) and low lint laboratory wipes (Kimwipes Kimberly Clark Corp.) between samples to minimize cross

PAGE 119

119 contamination. Due to the variable amount and small pa rticle size of organic material on the two smallest size fraction disks (retained particles: 35 200 m and 1 35 m), these disks were sent directly to N orthern A rizona U niversity C olorado P lateau S table I sotope L aboratory (NAUCPSIL). O nce at NAUCPSIL, technicians with previous experience using the stable isotope mass spectrometer e stimated how much filter material and associated particulate matter to use in sample analysis (Richard Doucett pers. c omm.). The l arge ( > 200 m ) plankton (primarily zooplan kton; Chapter 5) fraction sample preparation procedure consisted of : (1) removing the freeze dried sample disks from their presterilized microcentrifuge storage capsules, (2) sample disks were then carefully unfolded and attached to a small vibrating motor (3) as the sample disk vibrated, its surface was lightly brushed using a small nylon bristle brush ( Figure 43) to separate and concentrate retained particles into preweighed (0.001 mg) tin combustion capsules, and (4) sample and capsule were then weighed to the nearest (0.001 mg) using a Cahn Electrobalance (model 25). Tin sample capsules were carefully folded and packed with forceps for shipment to NAUCPSIL for analysis. Instruments (forceps aluminum funnel, acrylic capsule holder ) were sprayed w ith 70% isopropyl alcohol and cleaned with low lint laboratory wipes, or had particles removed (brushes) using f iltered and dried canned compressed air (Innovera canned air duster) between samples to minimize possible cross contamination. Isotopic Analys e s Carbon and nitrogen isotopic analyses ( C and N) were performed upon pre harvest fry samples and plankton collected at regular intervals during the outdoor pond trial ( Chapter 3; methods ). Weekly plankton samples were divided into three siz e fractions and presumed trophic assemblages consisting of a large (primarily) zooplankton fraction (> 200 m), a medium mixed plankton (rotifers and algae) fraction (32 200 m), and a small (primarily) phytoplankton

PAGE 120

120 fraction (1 32 m). Plankton size assemblage isotopic signatures were compared among treatments and sampling dates (a maximum of 12 weekly samplings for plankton) using twoway repeated measures ANOVA P airwise differences were determined using Bonferroni post comparison tests (P < 0.05), ( Zar 19 84, Ott 2000). Time Averaged Plankton Assemblage Isotopic Analyses Plankton size assemblage isotopic data were time averaged ( C and N) for each treatment. Time averaged isotopic data were used for comparisons with other trophic groups (s ize assemblages) within the same treatment and within the same trophic group (e.g. large zooplankton assemblage) among the other pond nutrient treatment groups. Comparisons were primarily made using one way ANOVA, with pairwise differences made using Tu keys multiple comparison test (P < 0.05), ( Zar 1994, Ott 2000). Trophic Group Isotopic Signature Differences Carbon and nitrogen isotopic signature differences among presumed trophic groups and applied nutrient treatments were calculated as simple arithmetic differences, and analyzed. For example, isotopic differences between small harvest fish and large zooplankton were derived as: C( sm fish zoo) = C ( sm harvest fish) C ( lg zoo plankton); per pond treatment N( sm fish zoo) = N ( sm harvest fish) N (lg zooplankton); per pond treatment Due to the small magnitude of carbon isotope differences between fish, nutrients and plankton groups which often resulted in somewhat a rbitrary positive and negative arithmetic signs for carbon isotope signature differences between groups, for statistical analyses involving carbon isotope signature differences, absolute values of carbon isotope signature differences were used ( Motulsky 1995). For comp a risons of trophic position differences between groups, simple arithmetic differences in their native values were used (non absolute isotopic signature

PAGE 121

121 values). For a given element, 95% confidence intervals for the differences between the isotopic signature means of different taxa were calculated using pooled standard error (pooled SE; Parker 1979). Trophic Level Determination Carbon isotope signature determinations are primarily used for identifying the basal nutrient source(s) for consumer organism(s) within a food web ( DeNiro and Epstein 1978, Harrigan et al. 1989, Fry 1991, Zohary et al. 1994, Vander Zanden and Rasmussen 1999, Saito et al. 2000). Based upon the carbon isotope signature of a applied nutrients, primary producers, a nd consumer organism(s), t he following equation was used to back calculate and identify, the carbon isotope signature of the primary producer or anthropogenic nutrient source within a pond treatments food web (modified from Saito et al. 2000): 13Cprim, i = 13Ci (TLi 1) C (Equation 4.1) where: 13Cprim, i = est imated av erage carbon signature of primary producer/nutrient sources () for consumer i 13Ci = measured carbon signature for consumer i () TLi = estimated trophic level of consumer i ; (TLi C = estimated trophic enrichment of carbon signature per trophic level For this study, the estimated carbon isotope signature trophic enrichment value per trophic level ( Cor C) was derived from the mean carbon isotope signature trophic enrichment rate observed in the indoor feeding trial (Chapter 2; C = 0.64 ).

PAGE 122

122 With slight modification, this equation also was used to determine the trophic difference between different consumer groups (e.g., primary, secondary, and tertiary consumers). Simple arithmetic differences in carbon and nitrogen isotope signatures were used t o determine trophic distances between groups: TDCij = [( 13Cj 13Ci) / C] 1) (Equation 4.2) where: TDCij = carbon isotope signature derived trophic distance (levels) estimate between taxon i j 13Ci = (abs. value) carbon signature consumer i () 13Cj = (abs. value) carbon signature for consumer j () (assuming trophic level j > trophic level i ) C = estimated trophic enrichment of carbon signature per trophic level Although carbon isotope signatures were used to calculate trophic distances, this was primarily performed as a redundant check upon trophic distances estimated using nitrogen isotope signatures. Carbon isotope signatur es are primarily used to identify the basal nutrient source(s) within an ecosystem, and are considered superior to nitrogen for this purpose ( Fry and Parker 1979, Haines and Montague 1979, Gu et al. 2001, Vander Zanden and Rasmussen 2001, McCutchan et al. 2003, Fry 2006). Nitrogen isotope signatures are primarily used to estimate the trophic position of consumer organisms within a putative food web, and are considered a better determinant of trophic position than carbon isotope signatures ( Peterson and Fry 1987, Vander Zanden et al. 1997, Vander Zanden and Rasmussen 1999). An equation identical to that given for trophic distance calculation based upon carbon isotope signature was used for trophic distance calculations using nitrogen isotope signatures:

PAGE 123

123 TDNij = ( 1 5Nj 1 5Ni) / (Equation 4.3) where: 1 5Ni = nitrogen signature consumer i () 1 5Nj = nitrogen signature for consumer j () (assuming trophic level j > trophic level i ) TDNij = nitrogen isotope derived estimated trophic distance (levels) between taxon i j = estimated trophic enrichment of nitrogen isotope signature per trophic level The estimated nitrogen isotope signature trophic enrichment rate ( ) was derived from the indoor feeding trial (Chapter 2; = 3.03 ). The higher trophic enrichment rates observed for nitrogen isotope signatures ( typically ~ 3.4 per trophic level) relative to carbon signatures ( typically ~ 0.01.0 C per trophic level) and associated high potential for overlap of carbon isotope signature error measures, is the primary reason that nitrogen isotope signatures rather than carbon isotope signatures are used to describe trophic positions within hypothetic al food webs ( DeNiro and Epstein 1979, 1981a Minigawa and Wada 1984, Vander Zanden and Rasmussen 2001, McCutchan et al. 2003, Fry 2006). Carbon and nitrogen isotope signature ba sed trophic source (primarily C) and trophic position (primarily N) estimates were made using the above equations to determine relative trophic distances from presumed basal nutrient sources and o ther trophic groups (Levine 1980). The same equation (based upon Equation 4.3) was used to estimate a consumer organism s trophic position relative to primary producers on a continuous trophic scale (Vander Zanden et al. 1997):

PAGE 124

124 1515 15()i i pp iTPNNN ( Equation 4.4) where: iTP = estimated trophic position of consumer i N i= nitrogen isotope signature of consumer i N pp= nitrogen isotope signature of presumed primary producer pp. N = nitrogen isotope signature enrichment rate Due to the plankton sampling and plankton size fractionation (sieve choice) experimenta l design methodologies, plankton size assemblage trophic position ( TPi) and function (primary producer, primary consumer, secondary consumer, etc.) estimates were based upon trophic position assumptions based upon nominal plankton taxa size Assuming that large zooplankton were more likely predate upon smaller zooplankton than vice versa ( Carpenter et al. 1985, 1987, Briand and Cohen 1987, Cohen et al. 1993, Brett et al. 1994 Warren and Spencer 1996, Vander Zanden et al. 1999, Post 2002b, Hoeinghaus et al. 2008) The role of fish and applied nutrients within pond ecosystems were assumed to be the top and bottom of the ir food chains, respectively Feeds (p rocessed and unprocessed ) and cottonseed meal were considered to be equivalent to primary producer s within living food webs Cottonseed meal was a special case in that it is believed to promote algal growth as a plant or green manure, but is also a highly nutritious seed grain when consumed directly ( Fowler 1980, Teichert Coddington 1997). As oppos ed to producing discretely defined trophic level/position estimates (positive integer values), Equation 4.4 can produce a continuum of trophic position values (continuous variable), which are better suited for a highly omnivorous and potentially cannibali stic species such as the swordtail (Dionne 1985, Jones et al. 1998a 1998b, 2007, Kruger et al. 2001, Saito et al. 2000, Vander Zanden and Rasmussen 1997).

PAGE 125

125 Isotopic Determination of Consumer Diet Two diet estimation methods (Euclidean distance and linear mixing model) that utilize stable isotope signatures were employed to determine both large and small harvest able fish diets as percentages of introduced nutrients and potential plankton prey size assemblages. T he Euclidean distance method is a determinist ic method and therefore, error measures were not generated, the linear mixing method utilized a linear algebra software program (IsoSource) written for IBM PC compatible computers (IsoSource computer program, D. Phillips USEPA). The Euclidean distance method utilized isotopic signature vector distances between presumed predator/consumer and prey/nutrient sources along the X ( 1 5N, 13C). The adjusted Euclidean distance method is given as: dij = 13'132 2()()iij jCC (Equation 4.5) where: dij = Euclidean isotopic signature distance between nutrient/organisms i j 13Ci = adj usted carbon isotope signature of predator/consumer i 1 5Ni = adj usted nitrogen isotope signature of predator/consumer i 13Cj = carbon isotope signature of producer/prey/nutrient j 1 5Nj = nitrogen isotope signature of producer/prey/nutrient j Adjusted predator carbon and nitrogen isotope signatures ( 13Ci, 1 5Ni) were obtained by subtracting the mean trophic enrichment values ( C and N; Equations 4.6 4.7) obtained in the indoor feeding trials (Chapter 2) from the una djusted predator carbon and nitrogen isotope signatures ( 13Ci, 1 5Ni): 13Ci = 13Ci C Equation 4.6 1 5Ni = 1 5Ni N Equation 4.7

PAGE 126

126 Euclidean distance measures were then used to compute percent prey/nutrient contributions (percent C, N basis) to the predator (e.g., harvest fish) diet. Calculation of prey utilization u sing Euclidean distances used the reciprocal distance (% prey utilization inversely proportional to Euclidean trophic distance) equation given as (Kline et al. 1993, Ben David and Schell 2001, Phillips 2001): %X = 1/' 1/'1/'1/'...1/' DX DADBDCDn X 100 (Equation 4.8) where: %X = percent inclusion of prey/nutrient X in diet (% C, N basis) DX = adjusted Euclidean trophic distance predator/consumer to prey/nutrient X ( where X = prey/nutrient A, B, C n; adjusted for trophic enrichment C, N ) DA = adjusted Euclidean trophic distance prey/nutrient A to predator/consumer DB = adjusted Euclidean trophic distance prey/nutrient B to predator/consumer DC = adjusted Euclidean tr ophic distance prey/nutrient C to predator/consumer Dn = adjusted Euclidean trophic distance prey/nutrient n to predator/consumer The linear mixing model utilizes a series of mass balance equations that are simultaneously solved for three unknown vari ables when two isotope tracing elements are used ( Schwarcz 1991 Phillips 2001, Phillips and Koch 2002, Phillips and Gregg 2003). A Microsoft Visual Basic computer program IsoSource (ver. 1.3.1) was utilized to iteratively determine percent food source usage and error estimates based upon two isotope signature values and a three foodsource maximum (e.g., C, N; Phillips 2003 and Gregg, Phillips 2005, Benstead et al. 2006). The basic equations for the linear mixing model are given as (Harrigan et al. 1989): 1 3Cpred = a 1 3Ci + b 1 3Cj + c 1 3Ck (Equation 4.9) 1 5Npred = a 1 5Ni + b 1 5Nj + c 1 5Nk (Equation 4.10)

PAGE 127

127 1 = a + b + c (Equation 4.11) where: 1 3Cpred = adjusted carbon isotope signature of the predator/consumer 1 5Npred = adjusted nitrogen isotope signature of the predator/consumer 1 3Ci = carbon isotope signature of prey i 1 3Cj = carbon isotope signature of prey j 1 3Ck = carbon isotope signature of prey k 1 5Ni = nitrogen isotope signature of prey i 1 5Nj = nitrogen isotope signature of prey j 1 5Nk = nitrogen isotope signature of prey k and a, b, c are proportions of diet consisting of prey i j and k respectively Dual element ( N, C) isotopic signature Cartesian plots ( 1 3C 1 5N ) of male and female broodstock, initial fry, harvest ed fish, and experimental feeds were made for the indoor trial isotopic data (Chapter 2; Figure 221). Similarly, isotopic plots for ea ch outdoor pond nutrient treatment that included small ( 31 mm SL) and large harvest fish ( > 31 mm SL), applied pond nutrients (except inorganic fertilizer), and the three plankton size/trophic assemblages were also generated. Once target organisms (swor dtail s ) and nutrient source candidates (feeds, plankton assemblages, etc.) were plotted on a carbon and nitrogen isotope signature Cartesian plane, potential nutrient source candidates isotopic signature point co ordinates were connected to describe a prey/food resource polygon. Polygons created by connecting the (X: 1 3C, Y: 1 5N) coordinates of each potential nutrient/prey group defined the potential trophic space for each fish predator group (small and large ha rvest fish) within a given pond nutrient treatment. If a target species/ consumer s isotopic coordinates are found to lie within the circumscribed resource

PAGE 128

128 polygon, then a plausible dietary solution is possible (Kline et al. 1993, BenDavid et al. 1997, Be nDavid and Schell 2001, Phillips 2001) Euclidean distance calculations and linear algebra techniques are then able to estimate relative use (% diet inclusion) among the isotopically analyzed potential food sources. Results Pond Nutrient Isotope Signatur es Pond nutrient carbon isotope signature differences were significant among the three pond nutrients analyzed [processed feed (PRO), unprocessed feed (UNP), and cottonseed meal (CSM) treatments, Figure 44; 1 (2,2) = 0.05] the inorganic fertilizer (INO ) pond nutrient did not contain carbon and was therefore excluded from the analysis. P rocessed feed 13C averaged 23.74 ( 23.91 to 23.56 95 % CI), unprocessed feed 13C averaged 23.95 ( 24.36 to 23.54 95 % CI), and cottonseed meal fertilizer 13C averaged 26.43, ( 26.54 to 26.32 95 % CI). Both processed and unprocessed feeds differed from cottonseed meal 13C values, but did not differ from each other when analyzed with post hoc pairwise testing (P < 0.05, Tukeys multiple comparison test). N itrogen isotope signatures also differed among pond treatments ( Figure 4 5; 1way (2,3) = 0.05). The p rocessed feed 15N average of 4.69 (4.26 to 5.11 95 % CI) and unprocessed feed 15N average of 4.80 (3.75 to 5.86 95 % CI) were similar, the cottonseed meal 15N average of 3.79 (3.72 to 3.86 95 % CI) was lower than both feeds, while the inorganic fertilizer 15N average of 0.08 ( 0.55 to 0.39 95 % CI) was the lowest of the four applied nutrients. Applied nutrient nitrogen signatures differed, except for the PRO and UNP treatments, which was expected given that both feeds were manufactured from identical ingredients. Significant pairwise differen ces were: PRO and CSM,

PAGE 129

129 PRO and INO, UNP and CSM, UNP and INO, and CSM and INO, (P < 0.05, Tukeys multiple comparison test). Pre harvest Fry Isotopic Signatures Limited carbon and nitrogen isotope signature analyses were performed for periodically sampled pre harvest pond fry (Chapter 3; days 7, 14 and 21 post stocking). Pre harvest fry carbon isotope signatures quickly diverged among treatments as the trial progressed ( Figure 46). Carbon isotope signatures significantly differed for both nutrient treatment and sampling date (rep. meas. 2 way ANOVA: P(nutrient) = 0.0004, F(nutrient) (2,3) = 0.05; P(time) < 0.0001, F(time) (2,2) = 0.05, P(interaction) = 0.0058, F(interaction) (2,6) = 0.05). Repeated measures ANOVA analyses requires a fully balanced experimental design with no missing data values (Zar 19 84, Ott 2000, Motulsky 1995), so a single pseudovalue was generated for a missing PRO pond replicate 13C value (pond A11) by inserting the mean of the remaining PRO pond treatment replicates for the given sampling date (4 April 2006) into the data array. Fry were unavailable from this pond and sampling date due to the low standing stock of fry within this pond following the recent introduction of broodstock. Several differences in preharvest fry carbon isotope signature were found among sampling dates within nutrient treatments (Figures 4 7 4 10), two treatments developed within treatment differences between fry after only seven days (Figures 4 7, 4 9; PRO day 7 and day 14, CSM day 7 and day 14), whereas it took 14 days for fry 13C differences to develop for the remaining two treatments (Figures 4 8, 410; UNP day 7 and day 21, INO day 7 and day 21; P < 0.05, Bonferroni post hoc test). Withinnutrient trea tment pairwise differences were not significant between day 14 and day 21 fry for any of the four nutrient treatments. Pairwise differences in pre harvest fry 13C between nutrient treatments for given sampling dates were:

PAGE 130

130 PRO and CSM (day 14, 21), PRO a nd INO (day 21), UNP and CSM (day 14, 21), UNP and INO (day 21), CSM and INO (day 7, 14, 21); no pairwise differences between PRO and UNP treatments were present for any of the three sampling dates (P < 0.05, Bonferroni post hoc test). Although, pond nut rient and sampling date were both significant factors determining pre harvest pond fry 13C isotopic signature, interaction was also significant between the two factors, P(interaction) = 0.0058 ( C), F(interaction) = 3.624. Even though significant interac tion occurred between factors, due to the high F values reported for the two factors in the analyses both nutrient and sampling date factors likely had strong influences upon fry C signature a lthough nutrient and sampling date factor effects were potentially non linear across levels ( Motulsky 1995). Pre harvest fry nitrogen isotope signature values also diverged as the trial progressed (Figures 4 7 410). Nitrogen isotope signatures significantly differed for both nutrient treatment and sampling date (rep. meas. 2 way ANOVA: P(nutrient) < 0.0001, F(nutrient) (2,3) = 0.05; P(time) < 0.0001, F(time) = 21.59, (2,2) = 0.05; P(interaction) < 0.0001, F(interaction) (2,6) = 0.05). T o perform repeated meas ures ANOVA, a single pseudovalue replicate was generated for this data set ( 15N) due to the unavailability of fry for pond A11 for 4 April 2006; pseudovalue generation for nitrogen signature was identical to carbon signature pseudovalue generation describ ed previously. Within nutrient treatment differences in pre harvest fry nitrogen isotope signature among sampling dates were: CSM day 7 and day 14, INO day 7 and day 14, INO day 7 and day 21; no pairwise differences were found between day 14 and day 21 fry within any of the four nutrient treatments (P < 0.05, Bonferroni post hoc test). Pairwise fry 15N differences between nutrient treatments for given sampling dates were: PRO and CSM differed for all three dates (days 7, 14,

PAGE 131

131 21), PRO and INO (days 14, 21), UNP and CSM (days 14, 21), UNP and INO (days 14, 21), and CSM and INO (day 7). PRO and UNP pre harvest fish 15N signatures did not differ for any of the three sampling dates Although pond treatment and time were both significant factors determining 15N signature in pre harvest pond fry, interaction was again significant between the two factors (P(interaction) < 0.0001); resulting in significant factor effects (nutrient treatm ent, sampling date), and potentially nonlinear interactions between the levels of the two factors (Motulsky 1995). Harvest Fish Isotope Signature Differences Large marketable fish ( > 31 mm SL ) carbon isotope ( 13C) signatures differed among treatment groups at harvest ( Figure 4 11; 1(2,3) = 0.05). PRO treatment large fish 13C mean 23.39 ( 24.20 to 22.58 95 % CI), UNP treatment 13C mean 23.26 ( 23.93 to 22.58 95 % CI), and the CSM treatment 13C mean 24.46 ( 24.92 to 24.00 95 % CI) were similar, but INO treatment large fish 13C mean 20.15 ( 21.22 to 19.09 95 % CI) was markedly higher than that of the other three nutrient groups. Pairwise post hoc testing determined that only the INO treatment large fish 13C significantly differed from those of the other three treatments (Figure 4 11; P < 0.001, Tukeys multipl e comparison test). Small unmarketable fish ( 31 mm SL ) carbon isotope ( 13C) signatures also differed among treatments ( Figure 412; 1 (2,3) = 0.05). The PRO treatment small fish 13C signature mean 24.31 ( 24.92 to 23.70 95 % CI) was slightly less enriched than for larger ( > 31 mm SL ) fish 23.39 ( 24.20 to 22.58 95 % CI ) from the same ponds, but was similar to both the UNP and CSM small harvest fish 13C means (UNP: 24.12 24.50 to 23.74 95 % CI; CSM 25.25 26.22 to 24.28 95 % CI), whereas

PAGE 132

132 the INO small fish 13C signature mean 20.29 ( 21.29 to 19.30 95 % CI) was the highest among all treatments. Pairwise difference analyses found that INO small fish C was significantly higher than the C values of the other three treatments, which did not significantly differ from each other (Figure 412; P < 0.05, Tukeys multiple comparison test). Large and small harvest fish carbon isotope signatures significan tly differed within individual pond nutrient treatments (rep. meas. 2way ANOVA: P(fish size) = 0.0087, F(fish size) = 8.645; P(nutrient) < 0.0001, F(nutrient) = 101.0; P(interaction) = 0.5927, F(interaction) = 0.6507). Mean c arbon isotope signature diffe rence between small and large fish within each nutrient treatment: PRO 13C = 0.917 ( 1.761 to 0.072 95 % CI), UNP 13C = 0.137 ( 0.779 to 0.505 95 % CI), CSM 13C = 0.503 ( 1.433 to 0.427 95 % CI), and INO 13C = 1.194 ( 0.068 to 2.456 95 % CI). Although significant differences in carbon signature were present between small and large harvest fish within individual pond treatments, post hoc comparison tests were not able to identify significant individual pairwise treatment diffe rences (P > 0.05, Bonferroni post hoc tests). Repeated measures analysis was performed due to the non independent nature of small and large harvest fish sampled from the same replicate ponds. Large fish ( > 31 mm SL ) nitrogen isotope ( 15N) signatures di d not differ among treatment groups at harvest (Figure 4 11; 1(2,3) = 0.05). PRO treatment fish 15N signature mean 7.62 ( 6.78 to 8.46 95%CI) was similar to the remaining three pond treatment large fish 15N mean s : UNP 7.82 ( 7.39 to 8.25 95 % CI ), CSM 8.26 ( 7.80 to 8.73 95 % CI ), and INO 7.44 ( 6.81 to 8.07 95 % CI). In contrast, small harvest fish ( 31 mm SL ) nitrogen isotope signatures significantly differed among pond nutrient treatments ( Figure 4 12; 1 (2,3) = 0.05). PRO pond treatment small fish 15N signature mean 7.96 ( 7.59 to 8.33 95 %

PAGE 133

133 CI), UNP small fish 15N mean 8.18 ( 7.51 to 8.85 95 % CI), and CSM small fish 15N mean 8.85 ( 7.98 to 9.71 95 % CI) largely overlapped, while the INO small fish 15N mean 6.33 ( 5.87 to 6.79 95 % CI) was considerably lower. Pairwise comparisons found that only the INO treatment small fish 15N signature mean differed from those of the other three groups ( Figure 412; P < 0.05, Tukeys multiple comparison test). Large and small harvest fish nitrogen isotope signature differences within individual pond treatments (e.g., PRO large fish N PRO small fish N) did not significantly differ among the four pond nutrient treatments (rep. meas. 2 way ANOVA: P(fish size) = 0.7731, F(fish size) = 0.08564; P(nutrient) < 0.0001, F(nutrient) = 14.67; P(interaction) = 0.0012, F(interaction) = 8.165). Small and large fish nitrogen isotope differences for each nutrient treatment averaged : PRO 15N = 0.339 ( 1.861 to 1.183 95 % CI), UNP 15N = 0.137 ( 1.589 to 1.315 95 % CI), CSM 15N = 0.503 2.105 to 1.099 95 % CI ) and IN O 15N = 1.194 ( 0.648 to 1.740 95 % CI). Although significant differences in small and large harvest fish nitrogen isotope signature within individual pond treatments were not detected using repeated measures 2way ANOVA, post hoc comparison tests we re able to identify a single significant small and large fish 15N signature difference within the INO pond treatment ( H. Motulsky pers comm.), ( P < 0.05, Bonferroni posttests). R epeated measures analysis was performed due to the non independent nature of small and large harvest fish sampled from the same replicate ponds. Large Zooplankton Isotope Signatures C) significantly differed among nutrient treatments and sampling dates ( Figure 413; re p. meas. 2way ANOVA, P(nutrient) < 0.0001, F(nutrient) (2,3) = 0.05; P(time) < 0.0001, F(time) (2,10) = 0.05, P(interaction) = 0.0189, F(interaction) (2,30) = 0.05). Large zooplankton carbon isotope

PAGE 134

134 signatures ( 13C) only differed between the INO treatment and the other three pond nutrient treatments within some of the eleven weekly sampling periods. Furthermore, significant differences were only present for a small number of sampling periods depending upon which tre atments were compared (P < 0.05, Bonferroni post hoc comparison test). The majority of pairwise differences in large zooplankton carbon signature between nutrients for a given sampling date occurred between the INO and CSM nutrient treatments (5 of 11 sam pling periods), all of which occurred during the latter half of the trial. Large zooplankton carbon isotope signatures differed among sampling dates within each of the four pond nutrient treatments (Figures 414 417). Pairwise differences in large z ooplankton 13C signature between sampling dates for the PRO nutrient treatment (three differences) all involved a single sampling date (25 May 2006) that had a lower mean 13C signature value and relatively small variation among replicate ponds ( Figure 4 14; lower C values higher on yaxis); UNP large zooplankton C signatures differed for a single pair of sampling dates ( Figure 4 15; 4 May 2006 and 25 May 2006); CSM large zooplankton C differences (nine) among sampling dates all included a single sampling date ( Figure 4 16; 25 May 2006), which again had a lower C value and relatively low variation among pond replicates. There were several INO large zooplankton C signature differences among sampling dates ( Figure 4 17), the majority of these differences involved the first sampling date (6 April 2006) ; this date h ad a l ow mean C value and greater variation among replicate ponds relative to samples collected near the end of the study (P < 0.05, Bonferroni post test). Due to insufficient lar ge zooplankton biomass for isotopic analyses, primarily at the beginning of the trial, s ixteen missing C values (single date and pond replicate) out of 264 date and replicate data points were replaced with pseudovalues generated by replacing missing v alue s with mean

PAGE 135

135 value s calculated from the remaining replicates for the given pond treatment and sampling date. T o perform a twoway repeated measures ANOVA, a fully balanced experimental design is required that does not allow missing data Although large F values for each factor indicated that both factors influenced large zooplankton 13C values, factor interactions (treatment time) also were significant ; t his indicated that factor effects may have been nonlinear across levels within each factor ( Motulsky 1995). Large zooplankton nitrogen isotope signatures ( N) significantly differed among pond treatments and sampling dates for the last eight weeks of the study ( Figure 4 13; rep. meas. 2 way ANOVA, P(nutrient) < 0.0001, F(nutrient) (2,3) = 0.05; P(time) < 0.0001, F(time) (2,7) = 0.05; P(interaction) = 0.0002, F(interaction) (2,21) = 0.05). Samples from the first four weeks of the study were omitted as insufficient large zooplankton biomass had developed withi n the ponds to provide adequate sample material for nitrogen isotope analyses given that only a single liter of water was filtered from each replicate pond for isotopic analysis. PRO and UNP large zooplankton N values did not differ for any of the eight sampling periods. Both PRO and UNP large zooplankton N values differed with CSM values, but only for three and two sampling dates, respectively; additionally, these dates occurred in the middle of the experimental trial and N isotopic differences were no longer detected during the last few weeks of the trial. INO l arge zooplankton N signature values differed more frequently with those of the other three treatments (PRO, UNP and CSM) than the three nonINO treatments differed from each other. PRO and UNP large zooplankton N signature values differed with INO large zooplankton N signature values (6 and 5 sampling dates, respectively) more frequently than did the CSM and INO large zooplankton N signature values (3 sampling dat es) Interestingly most of these differences occurred within the last four to six weeks of the three month trial (P <

PAGE 136

136 0.05, Bonferroni post hoc comparison test). Thirty five missing values of 264 data points (date and treatment replicate pond) were replaced by pseudovalues generated by replacing the missing values with mean values calculated for the remaining replicates for the given nutrient treatment and sampling date (identical process for missing C values). The greater number of missing zooplankton N signature values was due to the higher detection threshold for nitrogen for the Thermo Finnigan D ELTA plus Advantage gas isotoperatio mass spectrometer used to determine carbon and nitrogen isotope ratios (R. D oucett, NAUCPSIL, pers. comm.). Hi gher detection thresholds require more biological material to generate reliable nitrogen isotope results. Additionally, biological materials typically have lower nitrogen content relative to carbon (% dry w eight basis) Unfortunately, samples taken early in the study contained too little nitrogen to provide usable results, due to the low zooplankton standing stock s in these newly filled ponds. Large zooplankton nitrogen isotope signature differences among sampling dates were present wit hin the PRO, CSM and INO treatments (Figures 4 14, 416 417). N o large zooplankton nitrogen isotope signature differences among sampling dates occurred within the UNP ponds ( Figure 4 15). Pairwise differences between sampling dates for the PRO treatment (three differences) all involved a single sampling date (27 April 2006) that had a higher mean N signature value, and latter dates that had lower N signatures and lower variation among replicates ( Figure 414). Within the CSM treatment, pairwise differences among sampling dates occurred due to two consecutive sampling dates (1, 8 June 2006) near the end of the study that had higher N signature values than earlier samples ( Figure 4 16). Within the INO treatment, pairwise differences among sampl ing dates primarily occurred due to two consecutive samples (27 April 2006, 4 June 2006) collected earlier in the study that had higher N signatures and

PAGE 137

137 greater variation in N signatures among pond replicates ( Figure 417). Although a later date wit hin the INO treatment also had a higher 15N signature value (1 June 2006). Time averaged large zooplankton C and N isotopic signatures also were compared among pond nutrient treatments using 1way ANOVA. Time averaged large zooplankton carbon isotope signature s ( C) differed among treatments, agreeing with the 2 way repeated measures ANOVA results for nutrient treatments and sampling dates ( Figure 418; 1 way ANOVA, P < (2,3) = 0.05). PRO pond l arge zooplankton C signature mean 23.10 ( 23.61 to 22.59 95 % CI) and UNP mean 23.40 ( 24.21 to 22.58 95 % CI) were roughly equal, and CSM 24.39 ( 25.48 to 23.30 95 % CI) and UNP values also overlapped, but INO C mean 20.82 ( 21.49 to 20.16 95 % CI) was higher than all other treatments. Pairwise testing found two treatment pairs that did not differ for large zooplankton C signature values: PRO and UNP, and UNP and CSM (P > 0.05, Tukeys multiple comparison test). The INO large zooplankton C signature value was significantly higher (13C enriched) than the other three treatments. Although the CSM treatment large zooplankton assemblage was the most depleted in 13C, it did not significantly differ from the PRO and INO treatments, but did differ from the UNP large zooplankton C values. Time averaged large zooplankton N signatures also significantly differed among pond treatments (Figure 418; 1 way ANOVA, P < 0.0001(2,3) = 0.05). PRO treatment 15N mean 7.13 ( 6.89 to 7.37 95 % CI), and UNP treatment 15N mean 7.35 ( 6.77 to 7.93 95 % CI) confidence intervals largely overlapped. The CSM treatment N mean 6.29 ( 5.72 to 6.85 95 % CI) was lo wer than both the PRO and UNP mean s, but higher than the INO m ean 4.95 ( 4.60 to 5.31 95 % CI), which was considerably lower than the other three treatments. Post hoc significance tests found that large zooplankton 15N signatures differed

PAGE 138

138 from each other among all treatments except for the PRO and UNP treatments ( Figure 4 18; P < 0.05, Tukeys multiple comparison test). Medium Plankton Isotope Signatures Medium size plankton assemblage ( 32 among pond nutrient treatments and sampling periods (Figures 419 422; 2 way rep. meas. ANOVA, P(nutrient) < 0.0001, F(nutrient) (2,3) = 0.05; P(time) < 0.0001, F(time) (2,11) = 0.05; P(interaction) < 0.0001, F(interaction) (2,33) = 0.05). PRO and UNP and PRO and CSM medium plankton carbon signatures did not differ within any of the twelve sampling periods (P > 0.05, Bonferroni posttests). PRO and INO signatures differed for the latter half (six) of the twelve week study, UNP and CSM treatments differed for two consecutive sampling dates mid trial (18 May 2006, 25 May 2006), UNP and INO treatments differed for four of the last six weeks of the study (11 May 2006, 1, 8 and 15 June 2006), and CSM and INO signatures differed for six of the last seven weeks of the study (4, 11, 18 and 25 May 2006, 1 and 8 June 2006; P < 0.05, Bonferroni post hoc tests). Within treatment differences in medium size plankton carbon isotope signature occurred among sampling dates for all four pond nutri ent treatments (Figures 4 19 422; P < 0.05, Bonferroni posttests). PRO treatment signatures did not vary greatly among sampling dates, most differences were between high carbon isotope signatures present during the first sampling dates and lower carbon isotope signatures later in the study; the lowest signature occurred during the last week of the study (15 June 2006; Figure 419). UNP treatment medium plankton carbon isotope signatures followed a time trajectory nearly identical to that of the PRO treatment, with the highest measured signat ure found during the first week of sampling, followed by a steady downward (lower C values progressing upward from abscissa) progression till trial

PAGE 139

139 termination (Figures 4 19 420). CSM signatures also did not vary greatly over the twelve weeks, agai n the highest carbon isotope signature was during the first week of the study but remained fairly constant thereafter (Figure 421). INO signatures varied less than those of the other three treatments, fluctuating briefly during the first few weeks and r emaining fairly constant thereafter ( Figure 4 22). A single pseudovalue out of 288 samples (4 treatments x 12 sampling dates x 6 replicate ponds) was generated for a missing signature (CSM pond A4 20 April 2006) ; sample was mistakenly discarded due to hum an error. Medium size plankton assemblage nitrogen isotope signatures differed among treatments, but sampling period was not a significant factor within treatments (Figures 4 19 4 22; 2way rep. meas. ANOVA, P(nutrient) = 0.0003, F(nutrient) (2,3) = 0.05; P(time) = 0.3836, F(time) = (2,11) = 0.05; P(interaction) = 0.0450, F(interaction) (2,33) = 0.05). PRO and UNP signatures differed for only one of the twelve sampling periods (1 June 2006), PRO and CSM treatments did not differ for any of the twelve sampling dates, and PRO and INO differed for only two sampling periods during the first half of the twelve week trial (6 April 2006 and 27 April 2006). UNP and CSM signature s did not differ for any of the twelve sampling peri ods, UNP and INO signatures differed for a single sampling period (1 June 2006) and CSM and INO signatures differed most frequently with four pairwise differences (6 and 27 April 2006, 4 May 2006, and 1 June 2006; P < 0.05, Bonferroni posttests). Although it was determined that time was not a significant factor in explaining differences in medium size plankton nitrogen isotope signature among sampling dates within applied nutrient treatments, post hoc pairwise difference analyses did find (rare) significa nt differences in signature between sampling dates (P < 0.05, Bonferroni posttests). In the PRO treatment, differences among sampling dates were infrequent and did not follow any discernable pattern ( Figure 4 19), only a single date (1 June 2006) was

PAGE 140

140 resp onsible for the two significant pairwise differences observed among sampling dates. Within the UNP treatment, pairwise differences in signature also were infrequent and due to a single pair of sampling dates ( Figure 420; 4 May 2006 and 1 June 2006). CSM signatures did not differ among sampling dates ( Figure 421). A t the start of the trial (31 March 2006), a single sampling date with a high medium plankton nitrogen signature was responsible for the three pairwise differences (31 March 2006 and 27 April 2006; 31 March 2006 and 4 May 2006; 31 March 2006 and 1 June 2006) observed within the INO ponds ( Figure 4 22). Out of 288 small plankton samples (4 treatments x 12 sampling dates x 6 replicate ponds), t welve pseudovalues were generated for missing (UNP ponds: A3 1 June 2006, A5 6 April 2006, B8 31 March, 18 May 2006; CSM ponds: A1 1 June 2006, B12 20 April 2006; INO ponds: B7 6 April, 1 June 2006) and outlier (UNP ponds: B8 20 April, 4 May 2006; CSM ponds: A6 13 April 2006, B12 27 April 2006) medium plankton nitrogen isotope signatures. Outlier values were defined as values that were more than two standard deviations away from the mean value of the remaining replicates (Zar 1984). Time averaged mediumplankton carbon and nitrogen isotope signatures also were compared using 1way ANOVA. Time averaged medium size plankton carbon signature differences were found among pond nutrient treatments ( Figure 423; 1 way ANOVA, P(nutrient) < 0.0001, F(nutrient) = (2,3) = 0.05). PRO 1 3C mean 24.15 ( 24.15 to 23.86 95 % CI) was similar to both UNP C mean 23.86 ( 24.41 to 23.31 95 % CI) and CSM C mean 24.76 ( 25.59 to 23.93 95 % CI), but the INO C mean 22.19 ( 22.72 to 21.66 95 % CI), had the highest C signature, differing from a ll three non INO treatments. Significant differences were not found between the signatures of the PRO and UNP, and PRO and CSM treatments, but pairwise differences in carbon isotope signature were present between

PAGE 141

141 the following pond nutrient treatment pairs: PRO and INO, UNP and CSM, UNP and INO, and CSM and INO treatments (P < 0.05, Tukeys multiple comparison test). Only one treatment pair (UNP and CSM) did not include the INO treatment medium plankton assemblage, which had a markedly higher carbon isotope signature ( Figure 423; lower position on graph). Time averaged mediumsize plankton assemblage nitrogen isotope signatures also differed among pond nutrient treatments ( Figure 4 23; 1way ANOVA, P(nutrient) = 0.0003, F(nutrient) (2,3) = 0.05). PRO 15N mean 5.95 ( 5.04 to 6.85 95 % CI) was similar to both UNP 15N mean 5.66 ( 5.28 to 6.15 95 % CI) and CSM 15N mean 5.80 ( 4.71 to 6.90 95 % CI), but all three were significantly higher than the INO 15N mean 3.51 ( 2.45 to 4.58 95 % CI ). D ifferences onl y occurred between the INO and three nonINO treatments, no significant differences were detected between the remaining three nutrient treatments ( Figure 423; P > 0.05, Tukeys multiple comparison test). Small Plankton Isotope Signatures Small size plankton assemblage ( 1 pond nutrient treatments and sampling periods within treatments (Figures 424 427; 2way rep. meas. ANOVA, P(nutrient) < 0.0001, F(nutrient) (2,3) = 0.05; P(time) < 0.0001, F(time) = (2,11) = 0.05; P(interaction) = 0.0003, F(interaction) (2,33) = 0.05). PRO and UNP small plankton carbon isotope signatures did not differ for any of the twelve sampling periods and PRO and CSM treatment signatures differed for only two sampling dates at the start of the outdoor pond trial (31 March 2006 and 6 April 2006) I n contrast, PRO and INO signatures differed for the last nine sampling periods (P < 0.05, Bonferroni posttests). Similar to the PRO and CSM tr eatment comparison, UNP and CSM signature differences were present (to a slightly greater degree for the first five weeks) at the beginning of the study, but were absent in the latter

PAGE 142

142 half of the study; UNP and INO signature differences followed a pattern similar to that of the PRO and INO treatments, initial similarities (four of first five sampling dates) were absent in the last seven weeks of the study. The two fertilizer (CSM and INO) treatment signatures differed for all twelve sampling dates (P < 0.0 5, Bonferroni posttests). Within pond nutrient treatment differences in small plankton carbon isotope signature among sampling dates occurred within all four pond nutrient treatments (P < 0.05, Bonferroni posttests). PRO treatment 13C signature was hig hest during the first week of the study (31 March 2006), and significantly differed from the signatures measured during the remaining eleven weeks of the study. PRO treatment 13C signatures slowly oscillated over the twelve weeks of the study, and did not appear to stabilize about a constant value ( Figure 424; P < 0.05, Bonferroni posttests). The UNP signatures ( Figure 425) followed a trajectory intermediate to that observed for the PRO small (a gradual oscillation) and the UNP medium plankton 13C signatures (high initial carbon signature then gradual decrease) CSM signatures oscillated from a high initial carbon signature (31 March 2006) gradually decreasing before finally increasing to 13C values (last three sampling periods) that did not stat istically different from the initial value ( Figure 4 26). INO signatures differed among sampling dates due to three sampling dates (6, 13 April 2006 and 8 June 2006), but otherwise did not vary greatly ( Figure 4 27). A single pseudovalue was generated f or one missing sample (UNP pond B5 31 March 2006) out of 288 samples (4 treatments x 12 sampling dates x 6 replicate ponds) that was lost due to human error. Small size plankton assemblage nitrogen isotope signatures also differed among pond nutrient treatments and among sampling periods within pond treatments (Figures 424 427; 2way rep. meas. ANOVA, P(nutrient) < 0.0001, F(nutrient) (2,3) = 0.05; P(time) < 0.0001, F(time) (2,11) = 0.05; P(interaction) < 0.0001, F(interaction ) (2,33) = 0.05). Among the three

PAGE 143

143 nonINO treatments, 15N signatures did not differ among treatments for any of the twelve sampling periods. PRO and INO signatures differed for five of twelve sampling periods during the twelve week trial, bu t differences did not appear to follow any temporal pattern with differences occurring at the beginning, middle and end of the study with equal likelihood. UNP and INO treatment signatures only differed twice among the twelve sampling dates and differences were non consecutive and therefore not part of a possible trend. The two fertilizer treatments had the highest frequency of significant differences in signature (five of twelve sampling dates), but again, differences were present at the beginning, middl e and end of the study with no discernable temporal pattern. Small plankton nitrogen isotope signature differences among sampling dates within nutrient treatments were not present for either the PRO or UNP treatments (Figures 4 24 425; P > 0.05, Bonfe rroni posttests). CSM treatment small plankton nitrogen isotope signatures significantly differed among sampling dates, a gradual oscillation and overall increase occurred during the twelve weeks of the study ( Figure 4 26). Within the INO treatment, 15N signature s were fairly constant (Figure 4 27), except for two notable sampling periods (6 April 2006 and 18 May 20006) with low signatures which were not consecutive and were not part of any apparent trend in signature values. Two signature pseudovalues (PRO pond A8 31 March 2006 and UNP pond B5 31 March 2006) were generated for two missing samples lost due to human error. Time averaged small plankton assemblage carbon isotope signature values significantly differed among pond nutrient treatments ( Fig ure 4 28; 1way ANOVA, P(nutrient) < 0.0001, F(nutrient) (2,3) = 0.05). The PRO 13C mean 27.18 ( 27.87 to 26.49 95 % CI), and UNP mean 26.53 ( 27.22 to 25.84 95 % CI) were similar. The CSM C signature mean 28.64 ( 29.26 to 28.01 95 % CI) was the lowest, and the INO C mean 23.26

PAGE 144

144 ( 23.78 to 22.73 95 % CI) was the highest of the four pond treatments. Pairwise comparisons among the four treatments found that the PRO and UNP treatments did not differ, and that the two fertilizer treatments significantly differed from both feed treatments as well as from each other ( Figure 428; P < 0.05, Tukeys multiple comparison test). Time averaged small plankton assemblage nitrogen isotope signatures also differed among pond nutrient treatments (Figure 428; 1 way ANOVA, P(nutrient) < 0.0001, F(nutrient) = (2,3) = 0.05). The PRO 15N mean 4.38 ( 3.59 to 5.18 95 % CI), UNP 15N mean 4.47 ( 3.82 to 5.13 95 % CI), and CSM 15N mean 4.29 ( 3.84 to 4.73 95 % CI) w ere similar, but the INO 15N mean 1.19 ( 0.75 to 1.62 95 % CI) was markedly lower Pairwise difference testing found that only the INO treatment differed from the other three treatments ( Figure 4 28; P < 0.05, Tukeys multiple comparison test). Pond Nutrient and Pond Fish Isotope Signature Differences Time averaged carbon isotope signature ( C) difference magnitudes (absolute value) between large harvest fish and their respective pond treatment nutrients were calculated and compared among treatments and significant differences were found ( Figure 429; 1 way (2,2) = 0.05). PRO large harvest fish and processed feed carbon isotope differences [ C (fish) C (nutrient)] averaged 0.46 13C ( 0.23 to 1.15 13C 95 % CI ), UNP large fish and unprocessed feed C differences averaged 0.63 13C ( 1.32 to 0.06 13C 95 % CI), and CSM large fish and cottonseed meal carbon isotope differences averaged 1.97 13C ( 1.51 to 2.43 13C 95 % CI), INO large harvest fish and inorganic fertilizer carbon isotope difference comparisons were not possible as the inorganic fertilizer did not contain carbon. Pairwise comparison tests found two significant differences in fish and nutrient carbon isotope difference ma gnitude; CSM large harvest fish and cottonseed

PAGE 145

145 meal differed in carbon isotope signature values to a much greater degree than either the PRO or UNP large harvest fish, and their respective feed nutrients ( Figure 4 29; P < 0.05, Tukeys multiple comparison test). Time averaged nitrogen isotope signature ( 15N) differences between large harvest fish and their respective pond treatment nutrients were calculated and compared among treatments and significant differences were found ( Figure 4 29: 1way ANOVA, P (2,3) = 0.05). PRO treatment large harvest fish and processed feed 15N differences ( N(fish) N(nutrient)) averaged 2.94 1 5N ( 2.10 to 3.78 1 5N 95 % CI), UNP large harvest fish and unprocessed feed N differences averaged 3.05 1 5N ( 2.64 to 3.46 1 5N 95 % CI), CSM large harvest fish and cottonseed meal N differences averaged 4.48 1 5N ( 4.01 to 4.94 1 5N 95 % CI), and INO large harvest fish and inorganic fertilizer N differences averaged 7.52 1 5N ( 6.89 to 8.15 1 5N 95 % CI). Comparisons of nitrogen isotope difference magnitudes between large harvest fish and their respective pond nutrient among the different pond nutrient treatments were significant between all treatment pairs, except for th e PRO and UNP treatments ( Figure 4 29; P < 0.05, Tukeys multiple comparison test). Time averaged carbon isotope signature difference magnitudes between small harvest fish and their respective pond treatment nutrients were compared among treatments and no significant differences were found ( Figure 430: 1 (2,2) = 0.05). PRO treatment small fish and processed feed 13C difference averaged 0.64 1 3C ( 0.17 to 1.11 1 3C 95 % CI), UNP small harvest fish and unprocessed feed 13C difference averaged 0.29 1 3C ( 0.09 to 0.49 1 3C 95 % CI), and CSM small harvest fish and cottonseed meal 13C averaged 1.18 1 3C ( 0.22 to 2.15 1 3C 95 % CI). The magnitude of differences between feed nutrients and their respective small harvest fish carbon isotopes were

PAGE 146

146 not significant among the three nonINO treatments ( Figure 4 30; P > 0.05, Tukeys multiple comparison test). Again, the INO treatment small fish and applied inorganic fertilizer C signature difference analysis was not possible due to the absence of carbon within the applied inorganic fertilizer. Time averaged nitrogen isotope signature differences between small harvest fish and their respective pond nutrients were compared among treatments and significant differences were found ( Figure 430: 1(2,3) = 0.05). Nitrogen isotope signature ( 15N) differences between small fish and their respective nutrients averaged : PRO: 3.27 1 5N (2.91 to 3.64 1 5N 95 % CI), UNP: 3.37 1 5N (2.70 to 4.04 1 5N 95 % CI), CSM: 5.06 1 5N (4.19 to 5.92 1 5N 95 % CI), and INO: 6.41 1 5N (5.95 to 6.87 1 5N 95 % CI ). The pattern of 15N signature differences between small harvest fish and their respective nutrient s was identical to that found for large harvest fish and their respective feeds, differences were found among all treatment pair combinations except the PRO and UNP treatments ( Figure 4 30; P < 0.05, Tukeys multiple comparison test). Pond Fish and Large Zooplankton Isotope Signature Differences Carbon isotope signature difference magnitudes betw een large ( > 31 mm SL) harvest fish and their respective large zooplankton assemblages [ C (fish) C (lg.zooplankton)] were calculated and compared among treatments, but no significant differences in the magnitude (abs. value sign neutral) of C signature differences were found ( Figure 4 31; 1way ANOVA, P = 0.0723, F = (2,3) = 0.05). Large harvest fish and time averaged large zooplankton C signature differences averaged : PRO 0.65 1 3C ( 0.94 to 2.24 1 3C 95 % CI), UNP 0.41 1 3C ( 0.97 to 1.79 1 3C 95%CI), CSM 1.01 1 3C ( 0.72 to 2.74 1 3C 95%CI), and INO 1.13 1 3C ( 0.70 to 2.95 1 3C 95%CI) treatments.

PAGE 147

147 Nitrogen isotope signature differences between large harvest fish and their respective large zooplankton as semblages [ N(fish) N(lg.zooplankton)] were calculated and compared among treatments and significant differences were found ( Figure 432; 1 way ANOVA, P = (2,3) = 0.05). Nitrogen isotope signature differences between large harvest fish and time averaged large zooplankton assemblages within the treatments were: PRO 0.70 1 5N ( 1.04 to 2.44 1 5N 95 % CI), UNP 0.75 1 5N ( 0.38 to 1.89 1 5N 95 % CI), CSM 1.98 1 5N ( 0.57 to 3.39 1 5N 95 % CI), and INO 2.49 1 5N ( 1.11 to 3.87 1 5N 95 % CI). Large harvest fish and large zooplankton nitrogen isotope signature difference magnitudes did not differ between the two feed treatments (PRO and UNP), nor between the two fertilizer treatments (CSM and INO), but the tw o feed treatments did significantly differ from those of the two fertilizer treatments (Figure 4 32; P < 0.05, Tukeys multiple comparison test). Carbon isotope signature differences between small harvest fish and their respective treatments time averaged large zooplankton assemblage [ absolute value ( C (fish) C (lg.zooplankton))] were calculated and compared among treatments and were not significant ( Figure 433; 1(2,3) = 0.05). Time averaged small harvest fish and large zooplankton 13C signature differences averaged : PRO 1.07 1 3C ( 0.39 to 2.54 1 3C 95 % CI), UNP 1.03 1 3C ( 0.12 to 2.18 1 3C 95 % CI), CSM 1.12 1 3C ( 0.88 to 3.11 1 3C 95 % CI), and INO 0.73 1 3C ( 1.06 to 2.52 1 3C 95 % CI). Nitrogen isotope signature differences between small fish and their respective treatment large zooplankton [ N (fish) N (lg.zooplankton)] were calculated and significant differences were found among treatments ( Figure 4 34; 1 way ANOVA, P (2,3) = 0.05). Nitrogen isotope signature differences between small fish at harvest and time averaged large

PAGE 148

148 zooplankton averaged : PRO 0.65 1 5N ( 0.80 to 2.11 1 5N 95 % CI), UNP 1.11 1 5N ( 0.21 to 2.43 1 5N 95 % CI), CSM 2.56 1 5N (0 .90 to 4.22 1 5N 95 % CI), and INO 1.38 1 5N ( 0.12 to 2.63 1 5N 95 % CI). The magnitude of nitrogen isotope signature differences [ 15N(fish) 15N(lg.zooplankton)] between CSM small fish and large plankton were significantly greater than the differences between feed treatment small harvest fish and the respective large zooplankton assemblages, but INO treatment small harvest fish and large zooplankton nitrogen isotope signature differences magnitudes did not significantly differ from any other pond treatment ( Figure 4 34; P < 0.05, Tukeys multiple comparison test). Large Zooplankton and Pond Nutrient Isotope Signature Differences Carbon isotope signature ( C) difference magnitudes between timeaveraged large z ooplankton and their respective pond nutrients were calculated and compared among treatments and differences were found ( Figure 4 35: 1(2,2) = 0.05). Carbon isotope signature difference magnitudes between large zooplankton and their respective pond nutrient treatments [ C (large zooplankton) C (pond nutrient)] averaged : PRO 0.71 1 3C ( 0.71 to 2.13 1 3C 95 % CI), UNP 0.88 1 3C ( 0.65 to 2.40 1 3C 95 % CI), and CSM 3.03 1 3C ( 0.79 to 5.28 1 3C 95 % CI) T he INO pond nutrient treatment (inorganic fertilizer) did not contain carbon and was omitted from analysis. Pairwise comparisons of carbon signature magnitude differences between large zooplankton and their respective pond nutrients found signif icant differences in isotope signature difference magnitudes between the PRO and CSM, and UNP and CSM treatments ( Figure 435; P < 0.05, Tukeys multiple comparison test). A gain, isotopic signature difference magnitudes between the isotopically examined t axa ( i.e. large zooplankton C) and their applied nutrients (feed C) did not differ between the PRO and UNP treatments.

PAGE 149

149 Nitrogen isotope signature (N) differences between timeaveraged large zooplankton and their respective pond nutrients were calculated and compared among treatments, and significant differences among treatments were found ( Figure 436: 1way ANOVA, P < 0.0001, (2,3) = 0.05). Nitrogen isotope signature differences between large zooplankton and their respective pond nutrient treatments [ N (large zooplankton) N (pond nutrient)] averaged : PRO 2.66 1 5N ( 0.76 to 4.57 1 5N 95 % CI), UNP 2.32 1 5N ( 0.70 to 3.95 1 5N 95 % CI), CSM 2.50 1 5N ( 0.68 to 4.31 1 5N 95 % CI), and INO 5.03 1 5N ( 3.39 to 6.67 1 5N 95 % CI). Pairwise treatment comparisons of large zooplankton and their respective pond nutrient nitrogen isotope signature differences only found significant differences between the INO and other three nutrient treatments, which did not differ from one another ( Figure 4 36; P < 0.05, Tukeys multiple comparison test). Small Plankton and Pond Nutrient Isotope Signature Differences Time averaged carbon isotope signature ( 13C) difference magnitudes between small plankton assemblages (1 and applied nutrients differed among pond treatments ( Figure 437; 1way ANOVA, P = 0.0124, F = 5.962, (2,2) = 0.05). S mall plankton and feed nutrient carbon isotope difference averaged : PRO 3.44 13C ( 2.75 to 4.13 13C 95 %CI), UNP 2.58 13C ( 1.89 to 3.27 13C 95 % CI), and CSM 2.21 13C ( 1.58 to 2.83 13C 95 % CI). Again, due to the lack of carbon in the inorganic fertilizer, the INO treatment was omitted from this analysis. Pairwise difference analyses found that the small plankton and pond nutrient carbon isotope differences of the PRO and CSM treatments were significantly different, and again, the PRO and UNP treatments d id not differ (Figure 437; P < 0.05, Tukeys multiple comparison test).

PAGE 150

150 Time averaged nitrogen isotope signature ( 15N) difference magnitudes between small plankton assemblages and pond nutrient s, also differed among pond treatments ( Figure 4 38; 1way ANOVA, P = 0.0178, F = 4.251, (2,3) = 0.05). Sm all plankton and pond nutrient nitrogen isotope differences averaged: PRO 0.58 N ( 0.04 to 1.13 N 95 % CI), UNP 0.58 N ( 0.23 to 0.94 N 95 % CI), CSM 0.55 N ( 0.20 to 0.90 N 95%CI) and INO 1.26 N ( 0.83 to 1.70 N 95 % CI) Non INO treatment small plankton and applied nutrient 5N difference magnitudes were similar and markedly lower than the INO treatment. Not surprisingly, pairwise treatment comparisons found that the INO treatment small plankton assemblage and inorganic fertilizer nitrogen isotope signature difference was significantly greater than those of the other three treatments, which did not differ from each another ( Figure 4 38; P < 0.05, Tuke ys multiple comparison test). Isotopically Derived Large H arvest F ish D iet Euclidean distance estimates of PRO treatment large harvest fish diet contribut ions ( Figure 4 39, Table 4 1 ) were : 70% processed feed, 7% large zooplankton, 16% medium plankton, and 7% small plankton. IsoSource model estimates of PRO large harvest fish diet ( Table 4 1) were : 90.1% processed feed ( 87.74 to 92.46 % 95 % CI), 0.2% large zooplankton ( 0.0 to 0.70 % 95 % CI), 0.8% medium plankton ( 0.0 to 1.7 9% 95 % CI), and 8.9% small plankton ( 6.79 to 11.01 % 95 % CI). Euclidean distance estimates of UNP treatment large harvest fish diet ( Figure 4 40, Table 41) were : 90% processed feed, 2% large zooplankton, 6% m edium plankton, and 2% small plankton. IsoSource model derived estimates of UNP large harvest fish diet (T able 4 1) were : 91.1% processed feed ( 86.17 to 96.03 % 95 % CI), 1.0% large zooplankton ( 0.0 to 2.05 % 95 %

PAGE 151

151 CI), 6.8% medium plankton ( 1.55 to 12.05 % 95 % CI), and 1.0% small plankton ( 0.16 to 1.84 % 95 % CI). Euclidean distance estimates of CSM treatment large harvest fish diet (Figure 4 41; Table 41) were: 17% cottonseed meal, 23% large zooplankton, 51% medium plankton, and 9% small plankton. IsoSource model derived estimates of CSM large harvest fish diet ( Table 4 1) were : 21.4% cottonseed meal ( 10.28 to 31.68 % 95 % CI), 22.0% large zooplankton ( 6.05 to 37.95 % 95 % CI), 53.6% medium plankton ( 29.25 to 77.95 % 95 % CI), and 1.1% small plankton ( 0.0 to 3.62 % 95 % CI). Euclidean distance estimates of INO treatment large harvest fish diet ( Figure 4 42, Table 41) were : 65 % large zooplankton, 25 % medium plankton, and 10 % small plankton. IsoSource model derived estimates of INO large harvest fis h ( Table 4 1) were : 86.5 % large zooplankton ( 82.72 to 90.28 % 95 % CI), 8.6 % medium plankton ( 2.41 to 14.79 % 95 % CI), and 4.9 % small plankton ( 1.75 to 8.05 % 95 % CI). Isotopically Derived Small Harvest Fish Diet Euclidean distance estimates of PRO t reatment small harvest fish diet contributions ( Figure 4 38, Table 4 1) were : 35 % processed feed, 14 % large zooplankton, 33 % medium plankton, and 19 % small plankton. IsoSource model derived estimates of PRO small harvest fish diet (Table 4 1) were : 43.5 % processed feed ( 37.79 to 49.21 % 95 % CI), 7.0 % large zooplankton ( 1.41 to 12.59 % 95 % CI), 14.8 % medium plankton ( 3.38 to 26.22 % 95 % CI), and 34.7 % small plankton ( 31.60 to 37.80 % 95 % CI). Euclidean distance estimates of UNP treatmen t small harvest fish diet ( Figure 439, Table 41) were : 37 % processed feed, 14 % large zooplankton, 32 % medium plankton, and 17 % small plankton. IsoSource model derived estimates of UNP small harvest fish diet ( Table 4 1 ) were : 22.7 % processed feed ( 8.43 to 36.97 % 95 % CI), 13.4 % large zooplankton ( 8.36 to 18.44

PAGE 152

152 % 95 % CI), 29.4 % medium plankton ( 10.93 to 47.87 % 95 % CI), and 34.5 % small plankton ( 31.98 to 37.02 % 95 % CI). Euclidean distance estimates of CSM treatment small harvest fish diet ( Figure 441, Table 4 1) were : 21% cottonseed meal, 25% large zooplankton, 40% medium plankton, and 14% small plankton. IsoSource model derived estimates of CSM small harvest fish diet ( Table 4 1) were : 0.6 % cottonseed meal ( 0.0 to 1.44% 95 % CI ), 63.3 % large zooplankton ( 61.73 to 64.87 % 95 % CI), 2.0% medium plankton ( 0.01 to 3.99 % 95 % CI), and 34.1% small plankton ( 33.16 to 35.04 % 95 % CI). Euclidean distance estimates of INO treatment small harvest fish diet ( Figure 442, Table 41) were : 34 % large zooplankton, 47 % medium plankton, and 19 % small plankton. IsoSource model derived estimates of INO treatment small harvest fish ( Table 4 1) were : 67.3 % large zooplankton ( 64.05 to 70.55 % 95 % CI), 7.5 % medium plankton ( 1.52 to 13.48 % 95 % CI), and 25.3 % small plankton ( 22.36 to 28.24 % 95 % CI). Isotopically Derived Pre Harvest Fry Diet Euclidean distance estimates of PRO treatment pre harvest fry diet (Table 4 2) were : 30 % processed feed, 23 % large zooplankton, 38 % medium plankton, and 9 % small plankton. IsoSource model derived estimates of PRO treatment pre harvest fish ( Table 4 2) were : 49.8 % processed feed ( 45.71 to 53.89 % 95 % CI), 46.9 % large zooplankton ( 43.96 to 49.84 % 95 % CI), 3 % medium plankton ( 0.17 to 5.83 % 95 % CI), and 0.3 % small plankton ( 0.0 to 0.82 % 95 % CI). Euclidean distance estimates of UNP treatment pre harvest fry diet (Table 4 2) were : 24 % unprocessed feed, 19 % large zooplankton, 50 % medium plankton, and 8 % small plankton. IsoSource model derived estim ates of UNP treatment pre harvest fish ( Table 4 2) were : 33.1 % unprocessed feed ( 19.35 to 46.85 % 95 % CI), 42.4 % large zooplankton ( 35.37 to 49.43 % 95 %

PAGE 153

153 CI), 23.7 % medium plankton ( 4.81 to 42.59 % 95 % CI), and 0.7 % small plankton ( 0.0 to 1.64 % 95 % CI). Euclidean distance estimates of CSM treatment pre harvest fry diet (Table 4 2) were : 34 % cottonseed meal, 21% large zooplankton, 31% medium plankton, and 14% small plankton. IsoSource model derived estimates of CSM treatment pre harvest fish ( Tabl e 4 2) were : 54.6 % cottonseed meal ( 50.30 to 58.90 % 95 % CI), 20.9 % large zooplankton ( 8.94 to 32.86 % 95 % CI), 23.2 % medium plankton ( 7.56 to 38.84 % 95 % CI), and 1.3% small plankton ( 0.0 to 2.66 % 95 % CI). Euclidean distance estimates of INO treatment pre harvest fry diet (Table 4 2) were : 32% large zooplankton, 55% medium plankton, and 13% small plankton. IsoSource model derived estimates of INO treatment pre harvest fish ( Table 4 2) were : 34.1 % large zooplankton ( 32.74 to 35.46 % 95 % CI), 65 % medium plankton ( 63.69 to 67.31 % 95 % CI), and 0.9 % small plankton ( 0.0 to 1.84 % 95 % CI). Discussion Pond Nutrient Carbon Isotope Signatures Commercial feed (PRO and UNP) carbon isotope signatures differed from the applied cottonseed meal (CSM), but did not differ from each other due to their manufacture from identical ingredients (Figure 4 4). Commercial feeds were an indeterminate mixture of C3 and C4 crop plants and animal byproducts, the primary protein sources within the two commercial feeds were soybean meal, wheat middlings and fishmeal. Unfortunately, actual ingredient inclusion rates were not available from the manufacturer due to the proprietary nature of the commercial feeds [ J. Herlyn ( Purina Feed LLC ) pers comm. ] Cottonseed meal is a high protein, commonly applied organic fertilizer derived from the harvesting and processing of cotton, a broadleaved dicotyledonous (C3) crop plant ( Fowler 1980,

PAGE 154

154 Teichert Coddington 1997). The C signature measured for cottonseed meal (C3) in this study ( X = 26.43 13C ) agreed well with the mean C signature value reported in the literature for other terrestrial dicotyledonous plants (C3 dicot ~ 27.00 13C ; OLeary 1988, Boutton 1991). In contrast, terrestrial monocots typically have much higher carbon isotope signatures ( C4 monocot ~ 13 13C ; OLeary 1988, Boutton 1991, Fry 2006). The inorganic fertilizer (ammonium nitrate solution) used in the study, did not contain carbon, and therefore did not produce a carbon signature reading during carbon and nitrogen isotope analyses. For this reason, applied inorganic fertilizer nutrient and various INO treatment taxa carbon isotope signature difference calcul ations were not performed. Pond Nutrient Nitrogen Isotope Signatures Commercial feed (PRO and UNP) nitrogen isotope signatures differed from the two fertilizers (CSM and INO), which differed from each other ( Figure 45). Again, the PRO and UNP commerc ial feeds did not differ due to their manufacture from identical ingredients. Although cottonseed meals nitrogen isotope signature significantly differed from that of the two feed treatments, it was only slightly lower than that of the two applied feed s ( Figure 45). The INO fertilizer nutrient nitrogen isotope signature ( X = 0.08 N) was markedly lower than those of the other three nutrient treatments ( N range: 3.79 to 4.80 ), due to its manufacture from atmospheric n itrogen ( N(atm N) = 0.00 ). The extremely low inorganic fertilizer N proved useful as a tracer within the inorganic fertilizer treatment ponds. Subsequent trophic levels/plankton size assemblages within the INO treatment ponds had markedly lower nitrogen isotope signature values than their commercial feed and organic fertilizer treatment counterparts (Figures 4 18, 4 23, 428).

PAGE 155

155 Pre harvest Fry Isotope Signatures Due to budget and time constraints, pre harvest fry from only three sampling int ervals were analyzed for carbon and nitrogen isotope signatures to determine if, when and at what weight fry carbon and nitrogen isotopic signatures significantly diverged from presumably isotopically similar stock. Fry were not obtained for pond A11 (P RO) for one sampling date, 4 April 2006, due to low fry standing stock at the onset of the study; this may have been a result of the presence of invasive killifish, which ultimately necessitated the omission of this pond from data analyses at harvest Man y ponds had f ew fry at the beginning of the trial and acquiring even a small number of fry took much more time than later in the study when standing stocks of fry were higher. Broodstock for the outdoor pond trials were obtained from different sources ( greenhouse vats and outdoor ponds) and had different nutritional histories (i.e., indoor fish: TetraMin Staple Flake; outdoor fish: Purina Aquamax). Isotopically, offspring of mothers from different sources initially resembled their mothers and likely differed from fry of mothers from other sources. However, the initial mean isotopic signature of all broodstock and fry in the ponds should have been equal due to the random allocation of broodstock among the 24 replicate ponds of the four pond nutrient tr eatments ( Zar 1984, Riley and Edawards 1998, Montgomery 2001). Following diet switching by broodstock and as nascent fry consumed similar food items, it was expected that isotopic signatures among fry within ponds and treatments would converge (DeNiro and Epstein 1978, 1981a Min a gawa and Wada 1984, Fry 1991,2006, Fry et al. 1996), and diverge among fry from differing nutrient treatment ponds over time This was the observed pattern as fry isotope signatures increasingly diverged over time (Figures 4 4 410).

PAGE 156

156 Pre Harvest fry Carbon Isotope Signatures Initial fry carbon isotope signatures frequently differed with subsequent sampling dates within a given treatment: day 7 and day 14 fry (PRO and CSM treatments), and day 7 and 21 day fry (PRO, UNP and CSM treatments). However, no differences in carbon isotope signature were present between day 14 and day 21 fry within any of the four pond nutrient treatments, indicating that fry had reached carbon isotope equilibrium with their new nutrient sources by day 14. Pre harvest fry carbon isotope signature differences between treatments were also examined. Only the PRO and UNP fry did not differ at day 21, nor did they differ for the earlier day 7 and day 14 sampling periods. The lack of a C signature difference between the PRO and UNP pre harvest fry was not unexpected as the PRO and UNP feeds were manufactured from the same ingredients. The finding that the two feed treatment fry groups did not differ from each other in carbon signature for any of the three sampling periods, and that both feed treatment fry groups differed from the two fertilizer fry groups by day 21 further supports the hypothesis that feeds applied to ponds were being consumed and assimilated by their intended target groups (swordta ils). Another explanation for the observed feed treatment fry isotope signature similarities was that fry were utilizing live feeds (large zooplankton, medium and small plankton) that were obtaining their nutrition from the applied feeds, and that the exp loited live food bases within the feed treatment ponds had similar isotopic signatures. Cottonseed meal and inorganic fertilizer treatment pre harvest fry carbon isotope signatures differed for all three sampling dates. Likely a result of differing carbo n sources for primary producers within the CSM and INO treatment ponds (cottonseed meal decomposition, and abiotic carbon compounds, respectively)

PAGE 157

157 Pre harvest fry carbon isotope signature differences that developed over time (~ 1421 days) from isotopical ly similar pre harvest pond fry, indicate that nutrients derived from applied feeds and fertilizer nutrients resulted in isotopically distinct fry treatment groups over time. This finding was identical to that observed among indoor study fry fed five expe rimental feeds (Chapter 2; Figure 2 17). Pre Harvest Fry Nitrogen Isotope Signatures Similar to pre harvest fry carbon isotope signature results, pre harvest fry nitrogen isotope signatures differed among sampling dates and pond nutrient treatments (Figures 4 4 47). The pattern of nitrogen isotope differences between fry collected on different sampling dates, but within a given pond nutrient treatment resembled that for carbon isotopes. If present, nitrogen signature differences were only between day 7 and day 14 (CSM and INO treatments) or day 7 and day 21 (INO treatment), no differences were present between day 14 and day 21. Within their respective treatments, PRO and UNP pond nutrient treatment pre harvest fry did not isotopically differ for days 7, 14, or 21. Nitrogen isotope signature differences between fry from different pond nutrient treatments collected on the same sampling date, also followed a pattern similar to those found for carbon isotope signatures ( Figure 4 6). Is oto pic differences, t hat may not have been initially present (day 7) developed over time between fry of different pond nutrient treatments ( day 14, day 21 and at harvest day 90). PRO and UNP treatment pre harvest fry did not differ in nitrogen isotopic signa ture for any of the three time periods (day 7, 14, 21), which again was not surprising, as the two commercial feeds were manufactured from the same ingredients. Again, supporting the contention that fry within the PRO and UNP treatment ponds were either f eeding upon and assimilating the applied commercial feeds, and/or were utilizing live prey that were

PAGE 158

158 utilizing the applied feeds, or at the very least, utilizing live prey that were isotopically influenced by the application of the applied feeds. Large Har vest Fish Carbon Isotope Signature Differences PRO and UNP treatment large harvest fish carbon isotope signatures did not differ, which was not surprising as these two groups consumed feeds that were manufactured using identical ingredients. Although the mixture of C3 and C4 terrestrial plant crops ( C3( dicots) ~ 27.00 C C 4 ( monocots) ~ 13.00 C ; Boutton 1991, Fry 2006) fishmeal, and their respective inclusion rates in the commercial feeds produced a composite C signature that differed from that of the CSM fertilizer (Figure 4 4), CSM treatment large harvest fish carbon isotope signatures did not differ from those of the two feed treatments ( Figure 412). The finding and reason for why CSM treatment large harvest fish carbon isotope s ignatures did not differ from both the PRO and UNP fish is less clear. Cottonseed meal is a readily available and inexpensive terrestrial plant by product from the cotton industry that is high in protein with an attractive amino acid profile for use as an animal feed ( NRC 1993). It is likely that cottonseed meal acted as both an organic fertilizer that stimulated the production of living organisms (potential live feeds) within the ponds and as a high quality meal feed that was directly consumed (pers. obs.) and partially assimilated by the fish ( Fowler 1980, Teichert Coddington et al. 1997 ). Isotopic signature differences among the applied nutrients ( Figure 43; PRO, UNP and CSM) and their respective plankton assemblage C signatures (Figures 4 18, 423, 428; large, medium, and small plankton) were no longer present at the highest trophic levels small and large harvest fish (Figures 4 11 412). Carbon isotope signature differences present between the two commercial feeds and cottonseed meal and their respective small plankton assemblages

PAGE 159

159 (Figures 4 3, 4 28) were only partially present within the medium plankton (Figure 423; PRO and CSM medium plankton no C difference) and large zooplankton assemblages (Figure 4 18; UNP and CSM large zooplankton no C difference) ; and as mentioned previously, were no longer present within the two harvest fish size classes (Figures 4 11 412). The carbon isotope enrichment magnitude present between the applied cottonseed meal and CSM treatment large harvest fish indicate s that interstitial trophic groups were present between the applied cottonseed meal and CSM harvest fish, which was the expected food chain/web pattern within the CSM treatment. A plausible explanation for w hy small plankton assemblage carbon isotope signature differences among the PRO, UNP and CSM treatments were no longer present among the small and large harvest fish groups of these treatments was due to the highly catholic and omnivorous feeding habits of the swordtail ( Axelrod 1991, Tamaru et al. 2001). Swordtail diets (especially within the CSM treatment) likely consisted of numerous plant and animal sources from different trophic levels, which may have obscured carbon isotope signature differences present at the base and lower trophic levels ( Figure 4 28) of the food web/chain, by the point that these nutrients had reached higher trophic levels. INO large harvest fish carbon isotope signatures differed from those of the other three treatment groups ( F igure 411) l ikely due to the fact that INO fish were only able to consume live plants and animals that ultimately obtained their carbon from autochthonous sources ( Boyd 1979, 1995) within these ponds (remineralized inorganic carbon derived from bottom substrates, pond respiration, decaying organic matter), entering from the atmosphere (atmospheric CO2, dust particles, insects, organic debris, etc ) or carried in by ground water movements (aqueous carbonic acid aqueous CO2, carbonates, etc .).

PAGE 160

160 These largely inorganic forms (with the exception of insects and plant material entering the ponds) of carbon could not be directly utilized by INO pond treatment fish, requiring inorganic carbon to be fixed, assimilated or liberated by other pond organism (photosynthetic autotrophs, bacterial, and fungal decomposers) before entering the pond food web and becoming available to the target species ( Boyd 1979, 1995,Schroeder 1987, Diana et al. 1991, Hall 2004, De Brabandere et al. 2007 ). The INO small plankto n assemblage (primary phytoplankton) utilized this inorganic carbon pool(s), producing a composite carbon isotope signature differing from those of the other three treatments ( Figure 428). Due to the C signature differences present between the INO and nonINO small plankton assemblages, it is also assumed that the carbon sources utilized by 1 producers also differed within the INO and nonINO treatments. Assuming that nonanthropogenic or unintentional anthropogenic allochthonous and autochthonous c arbon sources were equally available to all twenty four treatment ponds, small plankton assemblage carbon isotope signature differences among treatments likely arose from autotroph assimilation of dissolved carbon compounds (DIC/DOC) leached, and/or remine ralized from the applied nutrients following consumption, assimilation, respiration, and excretion by (or death and decay of) pond heterotrophs ( Boutton 1991, Diana et al. 1991, Fry 1991, Boyd 1997, Hoyer et al. 1998, GregoryEaves et al. 2007 ). Isotopic ally differing inorganic carbon pools should result in isotopically differing primary producer (small plankton assemblage) carbon isotope signatures within the feed and organic and inorganic fertilizer treatment ponds ( Keeley et al. 1986, Kline et al. 1993, Fry 1999, Fry et al. 1999). Unfortunately, inorganic carbon pool (DIC: aqueous CO2, 3HCO and 2 3CO ) carbon isotope signatures were not measured within the different treatment ponds and their contribution to phytoplankton carbon isotope signatures relative to atmospheric carbon remains unknown.

PAGE 161

161 Additionally, nonanthropogenic autochthonous and allochth onous inorganic carbon input magnitudes also were unknown. With each additional trophic level occurring between an inorganic carbon source and the target species, an observable increase in the carbon isotope signature ( C enrichment) of the target specie s occurs (DeNiro and Epstein 1978, Fry and Parker 1979). The carbon isotope signature difference between INO harvest fish and their presumptive basal carbon source(s) was likely greater in magnitude than that of the other three treatments, reflecting the greater number of trophic levels carbon was believed to have passed through before being assimilated by the fish. Unfortunately, the INO treatment DIC pool C signature was not isotopically measured, making it necessary to use the INO small plankton assemblage C signature in food web/chain length comparisons (Vander Zanden and Rasmussen 1999). In addition to small plankton assemblage carbon isotope signature differences among the four treatments resulting from possible leaching and remineralization of the different applied nutrients, fundamentally different mechanisms by which carbon entered the two swordtail size classes likely occurred between the two feed treatments (via feed ingestion) and the two fertilizer treatments (via food web utilization, es pecially within the INO treatment). Primary producer carbon isotope signature differences among treatments ( Figure 428) and/or direct ingestion of applied nutrients versus live food web utilization differences in swordtail nutrition were likely responsi ble for the large carbon isotope signature difference s observed between INO large harvest fish and those within the other three treatments (Figure 411). Large Harvest Fish Nitrogen Isotope Differences Large harvest fish nitrogen isotope signatures did not differ among pond nutrient treatments (Figure 411). It was somewhat surprising that INO treatment large harvest fish

PAGE 162

162 nitrogen isotope signatures did not differ from that of the other three treatments, given that the three INO plankton assemblages (l arge, medium, small) and INO small harvest fish nitrogen isotope signatures were significantly lower than their counterparts within the other three treatments (Figures 4 12, 418, 423, 428). INO treatment plankton assemblages were dependent upon applied inorganic fertilizer as their primary nitrogen source (Boyd 1974, 1979, Schroeder 1978, Yusoff and McNabb 1989, KnudHansen et al. 1993, KnudHansen 1998, Kastner and Boyd 1996, Ludwig et al. 1998) and had much lower N signatures than their nonINO treatment counterparts. Because atmospheric nitrogen was the feedstock used in the industrial manufacture of the inorganic fertilizer, this made the inorganic fertilizers nitrogen signature [ by definition N2(atm) 0.00 N] much lower than that of the next lowest applied nutrient ( Figure 4 5; cottonseed meal = 3.79 N ). Although the lack of large harvest fish N signature differences among treatments seems puzzling in light of the carbon isotope signature result s, it was not surprising when it is known that nitrogen isotope signatures are less useful as nutrient source tracers than as a means of providing relative trophic position information ( Hideshige and Wada 1990, Kling and Fry 1992, Minagawa and Wada 1984, H ansson et al. 1997, Fry et al. 1999, Fry 2006, Michener and Kaufman 2007) Nitrogen isotope signature enrichment rates can be highly variable between trophic levels and enrichment rates of 1.5 to 4.0 for a single trophic level have been reported ( DeNir o and Epstein 1981, Fry 2006). Application of the four different pond nutrient regimes was expected to have resulted in large harvest fish nitrogen isotope signature differences among the four pond treatments. Omnivory by swordtails may explain the lack of nitrogen isotope signature differences among large harvest fish within pond treatments (Figure 4 11). If swordtails were consuming plant and

PAGE 163

163 animal prey, as well as animal prey from different trophic levels, this more trophically varied food web die t would result in a greater loss of the basal nutrients isotopic and trophic identity among their higher trophic level consumers, than would have resulted from a less trophically varied (e.g., strict zooplanktivore) and more simplistic dietary food chain (1 producer 1 consumer 2 consumer 3 consumer etc. ) originally postulated for swordtails within the fertilizer treatment ponds. Food chains originally postulated for feed treatment ponds were even more simplistic (applied feed swordtails). These simple food chains were somewhat unrealistic given what is know n of swordtail dietary habits and managed pond food web structure. A more intriguing explanation for the lack of differences in large harvest fish N signature among the four pond treatments and the INO treatment large harvest fish in particular, is the possibility of high rates of cannibalism of small fry and juveniles by larger fish within the ponds (Dione 1985, Jones et al. 1998a, 1998b, 2007, Beaudoin et al. 1999, Saito et al. 2000, Kruger et al. 2001). Cannibalism seems plausible in light of the lower nitrogen isotope signature of the INO taxa (small, medium, and large plankton assemblages, and small harvest fish) relative to their counterparts in the other three treatments and the lack of large harvest fish N signature differences among the four pond treatments (Figures 411 412, 418, 423, 428). Significant cannibalism by large harvest fish within the INO treatment, would place INO large harvest fish at a higher trophic level and increase their nitrogen isotope signatures to a greater degree than if they were strictly consuming nonfish live foods. Additionally, small fish and fry ( SL) made up a large percentage of the fish within the INO treat ment ponds (Figures 3 10 312), large numbers of size stunted fish ha ve been cited as being an indicator of food limitation ( Yusoff and McNabb 1989, Osenberg et al. 1992, Hopkins and Unwin 1997, Ingram

PAGE 164

164 2009). Evidence of food limitation for INO treatment harvest fish was provided by the large variation (Chapter 3; Figure s 313, 3 15) in INO treatment individual large fish weights and coefficient of variation (weight) values relative to those of the other t hree treatments, also which are often indicators of food limitation (Ali 1990, Diana et al. 1991). Small Harvest Fish Carbon Isotope Signatures Small harvest fish carbon isotope signatures followed a trajectory similar to that of large harvest fish ; INO t reatment small harvest fish were more enriched in 13C relative to those of the other three treatments ( Figure 411 412; note: higher C values, closer to abscissa). Again, the higher INO small fish carbon isotope signature was likely due to a combina tion of the greater number of trophic levels believed to be separating INO small harvest fish from their basal nutrient source and the higher carbon isotope signature of the INO treatment small plankton assemblage and its possible contribution to swordtai l diet (Figures 4 18, 4 23, 428). Another possible factor contributing to higher INO small and large harvest fish C signatures is that the INO treatment basal inorganic carbon pool 13C signature was considerably higher than those of the other treatments. This may have been the case if carbon from the nonINO treatment nutrients were present at high enough concentrations as remineralized DIC to influence small phytoplankton 13C within their respective ponds during carbon fixation; r esult ing i n primary producers that isotopically resembled ( C) their respective applied nutrients. Unfortunately, due to budget, time and logistical constraints, C signatures of basal inorganic carbon pools within the various treatment ponds were not measured. Small Harvest Fish Nitrogen Isotope Signatures In contrast to INO large harvest fish nitrogen isotope signatures, small harvest fish nitrogen isotope signatures differed from those of the other three treatments (Figure 4 12). Only

PAGE 165

165 INO pond treatment fish were entirely dependent upon live food generated within the pond ecosystem as no directly ingestible nutrients were applied to the INO treatment ponds. Fish within the PRO, UNP, and to a lesser degree CSM pond treatments may have been partly or eve n largely independent of any reliance upon live foods. Small harvest fish within the PRO, UNP and CSM treatments were consuming applied nutrients, and/or pond organisms that were consuming, or at a minimum isotopically influenced by the applied nutrients ( Figure 4 5; INO treatment lowestN signature) in quantities sufficient to raise their nitrogen isotope signatures to values greater than those within the INO treatment. The pattern of higher N signatures of the nonINO treatments large, medium a nd small plankton assemblages compared to their INO treatment counterparts (Figures 4 18, 423, 428), was identical to that found for plankton assemblage C signatures (i.e. INO C and N differing from those of the nonINO treatments). As previously suggested nonpiscine heterotroph s (e.g., zooplankton) may have been consuming applied nutrients, and autotrophic organisms (e.g., phytoplankton) may have been absorbing organic compounds leached from the applied nutrients or utilizing remineralized inor ganic carbon and nitrogen derived from other sources D issolved inorganic nitrogen (DIN) may have been derived from remineralized applied nutrients, primarily in the form of excreted ammonia and urea, or as free amino acids, or remineralized nitrite and nitrate. This was likely the reason that different plankton size assemblages (including phytoplankton) had similar nitrogen isotope signatures within treatments (PRO, UNP and CSM), and why patterns of nitrogen isotope signature differences for different ta xa groups (e.g., small plankton or small harvest fish) were similar among treatments ( Figure 412, 4.28; i.e., INO treatment taxa consistently had lowest N).

PAGE 166

166 Although INO small fish mean weight did not statistically differ from those of the three other t reatments, INO small fish mean weight (0.18 g) was lower ( 22 % less; Figure 314) than that of the next lowest mean (PRO: 0.23 g) Because fish size is generally proportional to fish mouth gape, which is directly correlated with potential prey size, it is likely that INO treatment small harvest fish fed on smaller prey from lower trophic levels than their counterparts within the nonINO treatments (Werner and Hall 1979, Dussault and Kramer 1981, Lemly and Dimmick 1982, Schmitt and Holbrook 1984a, b, Post and McQueen 1987, Arthington 1989, Holbrook and Schmitt 1992, 1994, Osenberg et al. 1992, Mittelbach and Osenberg 1993, Wainwright and Richard 1995, DeVries et al. 1998, Cailliet et al. 1996 Gu et al. 1996a, b, Mittlebach et al. 1999 ). T he combination of lower INO small harvest fish weight and higher INO treatment small fish numbers ( Chapter 3; Figures 3 12, 3 14) may have caused INO small fish diet s to shift in favor of smaller prey due to both mouth gape and/or prey abundance limitations due to increased i ntraspecific competition Smaller, lower trophic level prey also wa s a possible explanation for the lower N signature of the INO treatment small harvest fish relative to their counterparts in the other three treatments. As previously mentioned, the complete dependence of INO treatment small harvest fish upon live foods produced within the ponds and the lower N signatures observed for the INO treatment plankton assemblages (Figures 4 18, 423, 428), may have manifested itself as the lower N signature of INO small harvest fish. The lower nitrogen isotope signature (tracer signal) of the INO treatments inorganic fertilizer basal nutrient relative to the nonINO treatment basal nutrients appears to have propagated throughout the INO treatment food web.

PAGE 167

16 7 Harvest Fish Size Class Isotope Signature Differences As previously mentioned, t he small carbon isotope signature difference observed between small and large harvest fish within a given treatment may have been due to ontogenetic differences between small and large fish ; with smaller fish eating smaller food items from lower trophic levels relative to larger fish due to gape size limitations and beha vioral differences in microhabitat preference ( Schmitt and Holbrook 1984a, Holbrook and Schmitt 1992, Cailliet et al. 1996, Mittlebach and Osenberg 1993 ). Ontogenetic dietary differences are consistent with the observation that smaller fish and fry were closely associated with the submerged and emergent vegetation present in the shallow water areas adjacent to pond perimeters. Aggregations of presumably higher trophic level large macroscopic zooplankton were frequently observed at the surface in deeper po rtions of the pond, where larger fish were known to congregate and smaller fish were absent. Large Zooplankton Carbon Isotope Differences Large zooplankton carbon isotope signatures did not differ among the PRO, UNP and CSM treatments for any of the 10 weekly sampling dates, only the INO treatment large zooplankton carbon isotope signatures differed from those of the other treatments for any of the 10 sampling dates. During the last three weeks of the study, INO tre atment large zooplankton C (range: 19.76 to 19.88) signatures were considerably higher than those of the other three treatments ( mean values: 24.25 CSM 23.16 UNP 22.59 PRO). The frequency with which the INO treatment large zooplankton C signatures differed from those of the other three treatments varied: PRO and INO treatments (one sampling period), UNP and INO (two periods), and CSM and INO (five periods). The majority of betweentreatment differences in large zooplankton C occurred during the last five weekly sampling periods. All five CSM and

PAGE 168

168 INO large zooplankton C isotopic signature differences occurred during the last five sampling dates. A decrease in variation and stabilization about a mean C value ( range: 19.76 to 19.88 ) was observed for the INO treatment large zooplankton, and was most likely responsible for the observed differences between the INO and CSM treatments near the end of the trial ( Figure 417). The frequency and temporal pattern of large zooplankton carbon isotope signature differences between the two fertilizer treatments suggests that isotopic equilibrium between the CSM and INO large zooplankton assemblages and their respective applied nutrients was not reached until approximately half way into th e pond trial. This supports the contention that carbon with in the cottonseed meal and inorganic fertilizer ponds were derived from different sources, which was fortunate in that this likely produced the isotopically contrasting CSM and INO small phytoplankton assemblage, and subsequent t rophic level taxa C differences ( Figures 4 18, 424, 428) The absence of large zooplankton carbon isotope signature differences between the PRO and UNP treatments for the 10 sampling dates was not surpri sing as both presumably received a large portion of their nutrition from applied feeds that had identical chemical compositions. The fact that the PRO and UNP large zooplankton carbon isotope signatures did not differ from each other, but differed from those of the two fertilizer treatments, supports the hypothesis that pond organisms (not limited to fish) in the feed treatment ponds were either assimilating, or at a minimum, isotopically influenced by the applied feeds. Additionally, both feeds were ground to a fine powder so the possibility of consumption and assimilation by heterotrophic macrozooplankton seems plausible. Large zooplankton carbon isotope signatures differed among sampling dates within each of the four pond nutrient treatments. Howeve r, within treatments, large zooplankton carbon

PAGE 169

169 isotope signature differences among sampling dates did not appear to follow any discernable pattern (Figures 4 14 417). Withintreatment carbon isotope signature differences among sampling dates were more frequent within the two fertilizer treatments T he CSM treatment had the greatest number of pairwise differences ( 10), followed by the INO treatment (7). O nly three differences were present within the PRO treatment and only one within the UNP treatment. Although the CSM treatment had the greatest number of pairwise differences in large zooplankton C signature among sampling dates, all significant pairwise differences involved a single date that had an unusually low carbon signature (25 May 2006, 28.92 C). INO treatment large zooplankton C signature differences among sampling dates, also only involved a single date w ith a high carbon signature ( 4 April 2006, 18.95 C) and the three earliest sampling dates which had relatively low carbon signatures ( Figure 417) These differences were likely due to decreasing C signature variation among sampling date mean s as the trial progressed; large zooplankton C signature 95 % CI near the end of the trial were noticeably narrower than those at the start of the trial (Figures 4 16 417). Although variation decreased with time for both large zooplankton carbon and nitrogen signatures for the PRO, UNP and CSM treatments, this decrease was greatest within the INO pond nutrient treatment. The h igher frequency of large zooplankton carbon isotope signature differences between treatments during latter sampling periods suggests that large zooplankton (and other pond taxa) had come to isotopic equilibrium with their prey organisms and applied nutrients ( Figure 4 134.17). The smaller number of isotopic signature differences among sampling dates within the INO treatment, may have been due to higher levels of chemical and isotopic homogeneity of carbon and nitrogen pools within the INO treatment. The a tmospheric nitrogen used in the manufacture of the liquid inorganic fertilizer (Haber Bosch process) wa s chemically and

PAGE 170

170 isotopically homogenous ( Haber 1920). Presumably, the carbon pools utilized by primary producers within the INO pond treatment also were relatively large, well mixed and isotopically homogenous and stable over the three month study (atmospheric carbon dioxide, dissolved ground water carbon dioxide and carbonates; Boutton 1991). Although a tmospheric carbon isotope signatures undergo s easonal and latitudinal variations, primarily due to vegetative growing seasons and carbon fixation due to photosynthesis ( Estep and Vigg 1985, Takahashi et al. 1990, Zohary et al. 1994, Hall 2004) these changes should have been relatively minor during th e short duration of this study. In contrast to the isotopic homogeneity of the carbon pool utilized by the INO pond treatment primary producers, the potential for carbon isotope variation among the commercial feed (e.g., animal and plant material), and cottonseed meal particles was relatively high. Organisms consuming these nutrients may not be assimilating particles of uniform chemical and isotopic composition (more so within the UNP and CSM treatments), and therefore, individual consumer organisms may vary slightly in their chemical and isotopic makeup (C. Watson pers comm. ). Isotopic changes also may have been occurring as these perishable materials (feeds and cottonseed meal) aged and oxidized, were subjected to changes in humidity and temperature, a nd underwent bacterial and fungal decomposition. Feeds, cottonseed meal and liquid fertilizers were stored in a climate controlled room at a temperature of approximately 21C (+/4C) H owever, humidity and temperature levels noticeably fluctuated throughout the day, depending on ambient temperature and weather conditions. The inorganic fertilizer treatment is believed to have been relatively stable due to its simple chemical composition, storage conditions (cool, dark

PAGE 171

171 room) and highly concentrated for m that would tend to inhibit microbial growth, in comparison to the applied organic nutrients Unfortunately, to perform repeated measures ANOVA, it was necessary to replace 16 missing large zooplankton carbon isotope signature values (of 264 possible single date and pond replicate combinations) with pseudovalues. Missing carbon isotope signature values were the result of insufficient quantities of biological material required for isotope ratio mass spectrometry analysis. This usually occurred at th e beginning of the trial when recently drained and sterilized ponds were still in the initial stages of colonization by aquatic organisms ; cooler weather and shorter photoperiods also contributed to lower standing stock s of phytoplankton and zooplankton ( Fogg and Thake 1987, Malin and Paerl 1994, Hall 2004, Hoeinghaus et al. 2008) Lower plankton standing stocks early in the study relative to later sampling periods were almost certainly responsible for the significan t tim e factor, in the twofactor repeate dmeasures ANOVA. It was not clear how the significant time and nutrient treatment factor interactions were auto correlated to produce any obvious trends ( Zar 1984, Sokal and Rohlf 1981). Time Averaged Large Zooplankton Carbon Isotope Signatures Ti me averaged large zooplankton C signature differences among the four treatments found that only two of the six possible treatment pair combinations did not significantly differ: PRO and UNP, and UNP and CSM ( Figure 4 18 ). PRO and UNP large zooplankton C isotope signatures did not differ, possibly due to direct consumption of the isotopically identical PRO and UNP feeds by these relatively large macrozooplankton (> 200 m), which may have been large enough to directly consume applied feeds (Chapter 5). The reason why time averaged UNP and CSM large zooplankton carbon isotope signatures did not differ is less clear ( Figure 4 18), and somewhat puzzling given the presence of C signature differences among

PAGE 172

172 the applied feed and cottonseed meal ( Figure 4 4) and their respective small plankton assemblages ( Figure 4 28). Within the feed and cottonseed meal treatments, carbon isotope signature differences present at the base of the food chain/web may have became attenuated as carbon progressed to higher t rophic levels before becoming statistically insignificant (PRO, UNP and CSM treatments) within the two top trophic levels examined (Figures 4 11 412, 418; harvest fish, large zooplankton). As previously mentioned, this may have occurred because plankton size assemblages only approximately consisted of functional trophic groups (i.e. 1 producer, 1 consumer, 2 consumer), and each plankton assemblage li kely contained a mixture of autotrophs and heterotrophs (Chapter 5) as well as heterotrophic organisms (bacteria, fungi, protists, zooplankton) from different trophic levels (e.g., the medium and large plankton assemblages). Potential large zooplankt e r o mnivory and carnivory from different trophic levels may have further dampened large zooplankton assemblage carbon isotope signature differences among treatments and among trophic guilds/plankton assemblages within treatments ( Kling and Fry 1992, Vander Za nden and Rasmussen 2001). The low magnitude of carbon signature enrichment typically observed ( ~ 0.4 1.0 13C ) with each successive trophic level, also likely contributed to the lack of carbon isotope signature differences at higher trophic levels that were present at lower trophic levels ( Fry 2006). The INO large zooplankton assemblage carbon isotope signature was significantly higher than those of the nonINO treatments. Additionally, INO treatment small (primarily phytoplankton), and medium (phytopl ankton and rotifers) plankton assemblages also had carbon isotope signatures higher than their counterparts in the three other treatments (Figures 4 23, 428). This was the carbon isotope signature pattern expected for the INO plankton assemblages relative to those within the nonINO treatments; given the primary alternate hypothesis of

PAGE 173

173 parallel nutrient movement within the two feed treatments, and serial nutrient movement within the two fertilizer treatments (more so within the INO than CSM treatment). T he prediction that the INO large plankton assemblage carbon isotope signature would be higher than those of the nonINO treatments was based upon the assumption that PRO, UNP and CSM treatment plankton assemblage organisms (small, medium and large) either directly consumed applied nutrients (heterotrophs), indirectly utilized applied nutrients via prey that assimilated applied nutrients (heterotrophs), and/or that primary producers within the nonINO treatments utilized dissolved inorganic carbon (DIC) pa rtially derived from remineralized applied nutrient carbon. Applied nutrients were likely remineralized into DIC following consumption and assimilation by heterotrophic organisms, which could subsequently be utilized by photosynthetic autotrophs ( Rounick and Winterbourn 1986, Cabana and Rasmussen 1994, Valiela 199 5, Grey et al. 2005, Hall 2004, GregoryEaves 2007). If nonINO treatment taxa (plankton and fish) were isotopically influenced by their respective applied nutrients (including via remineralization) to a greater degree than by nutrients derived from food chain/production resulting from autochthonous and nonanthropogenic allochthonous inorganic nutrient sources, these taxa (fish and plankton) should have had carbon isotope signature values more si milar to one another ( Figure 4 43 ) compared to taxa within the INO treatment (e.g., greater processed feed treatment C similarity between large fish and large zooplankton, compared to their inorganic fertilizer treatment counterparts). In contrast, INO pond treatment organisms were predicted to have a longer food web/chain between them and their basal autochthonous carbon source(s), becoming more enriched in 13C with each trophic level, and having higher C signatures relative to their counterparts in the other three treatments. This carbon isotope signature difference pattern among the four

PAGE 174

174 treatments was not limited to the large zooplankton assemblages. Among the four treatments, carbon isotope signature differences remained the sam e within all of the postulated trophic groups (Figures 411 4 12, 418, 423, 428; small and large harvest fish, large, medium and small plankton size assemblages): PRO and UNP groups did not differ in C signatures, CSM group C signatures were slightly lower than other groups, and INO group C signatures were significantly higher. PRO and UNP taxa carbon isotope signature similarities were likely due to the manufacture of both feeds from identical ingredients (identical batch/supplier), and the direct (fish, large zooplankton) and indirect (smaller zooplankton and phytoplankton) utilization of the applied feeds by the PRO and UNP treatment pond taxa. A s previously mentioned, a n alternate explanation for higher INO taxa C signatures relative to their non INO counterparts is that the DIC pool utilized by INO primary producers had a higher C signature than those of the three nonINO treatments; and this higher carbon isotope signature signal was propagated throughout the INO treatment food we b taxa resulting in their higher carbon isotope signatures relative to their non INO taxa counterparts (Figures 411 412, 418, 423, 428). Additionally, both processes (greater carbon enrichment due to trophic processes due to longer food chain, and higher INO DIC C signature) may have been occurring simultaneously within the INO pond treatment. However, a higher INO DIC pool C signature relative to the non INO DIC pool C signatures scenario seems unlikely due to the large and relatively uniform (isotopically) inorganic carbon pool(s) typically utilized by freshwater autotrophs ( Zohary et al. 1994, Yoshii et al. 1999, Fry 2006). The same non anthropogenic inorganic carbon pool was likely present within all experimental ponds to some degree, and its influence upon pond taxa carbon isotope signatures likely did not differ among treatments.

PAGE 175

175 Baseline DIC pool carbon isotope signatures for the four pond treatments would have been extremely useful in determining which of t he above explanations (if any) was correct, or at least more likely. Unfortunately, DIC carbon isotope signatures were not measured within the four pond treatments due to budget, labor logistical and time constraints. Large Zooplankton Nitrogen Isotope Signature Differences Large zooplankton nitrogen isotope signature values also differed among sampling dates and treatments. Similar to large zooplankton carbon isotope signature results, large zooplankton nitrogen isotope signatures did not differ bet ween PRO and UNP treatments for any of the eight sampling dates. The higher N signatures of the two feed treatment large zooplankton assemblages differed less frequently with the CSM fertilizer treatment than they did with the INO fertilizer treatment: PRO and CSM (3 differences), UNP and CSM (2 differences), PRO and INO (6 differences), and UNP and INO (5 differences). The two fertilizer treatment large zooplankton N signatures (CSM and INO) only differed three times during the study. Differences in large zooplankton N signature between treatments did not follow any discernable temporal pattern, as differences between treatments were present at the beginning, middle and end of the study with equal likelihood, and were frequently not consecutive. Large zooplankton N signature differences among sampling dates within individual treatments were found for the PRO, CSM and INO trea tments, with no differences among sampling dates present for the UNP treatment. The majority of these pairwise differences involved latter dates and their smaller confidence intervals, and earlier dates when more variation among replicates occurred, espec ially within the INO treatment (Figures 4 14 417). Large zooplankton N signature variations were markedly lower than C signature variations for the same sampling dates and treatments. Although N isotope signature variations about the

PAGE 176

176 means were lower (narrower error margins), significant differences in large zooplankton N signature among dates were less frequent due to the flatter (more uniform) time series trajectories of N signatures relative to C signatures (Figures 4 13 417). The lower frequency of large zooplankton N signature differences relative to C signature differences, between and within treatments among the eight sampling dates was likely a function of both the fewer sampling dates analyzed and the smaller differences among sampling date mean s (Figures 413 417). Unfortunately, it was necessary to omit large zooplankton N signature results from the first three weeks of the study due to insufficient large zooplankton biomass in the ponds Additionally, it was necessary to replace 34 missing sampling date and replicate pond N values (of 192 combinations) with pseudovalues generated in the same manner as missing C signature values before conducting repeated measures ANOVA. A gre ater number of missing 15N data occurred than for 13C data. Due to the lower molar content of nitrogen relative to carbon within biological materials, nitrogen has a higher detection threshold than carbon for the isotope ratio mass spectrometer used in this study; hence a greater quantity of biological material is required to adequately perform nitrogen isotope signature analysis (R. Doucett pers. comm.). Time A veraged Large Zooplankton Nitrogen Isotope Differences Time a veraged large zooplankton N signature differences were significant among treatments (Figure 418). Similar to time averaged large zooplankton C signature results, time averaged large zooplankton N signatures for the PRO and UNP treatments did not differ. Time averaged CSM large zooplankton N signature differed from all other groups, as did the INO large zooplankton N signature, which had the lowest N isotope signature of all four treatments (Figure 418).

PAGE 177

177 The lack of large zooplankton nitrogen isotope signat ure differences between the PRO and UNP treatments was likely due to many heterotrophs ( including, but not limited to large zooplankton, i.e., fish, etc .) in these ponds either receiving direct nutrition from the isotopically identical applied feeds (Figu res 4 11 4.12, 418, 423, 428), indirectly by the consumption of a lower trophic group(s) that depended heavily upon the applied feeds, or by consuming primary producers which had utilized and become isotopically influenced by remineralized carbon and nitrogen originally derived from the applied feeds. Direct consumption of the two commercial feeds and to a lesser extent cottonseed meal in the non INO treatment ponds by pond zooplankton should have produced shorter food chains/ webs than those occurring within the INO fertilizer treatment (less trophic distance to the applied and presumed basal nutrients within the INO treatment). Shorter food chains/webs were the observed pattern for time averaged nitrogen isotope differences between large zooplankton and their applied pond nutrients ( Figure 4 36) within the non INO treatments Time averaged CSM large zooplankton nitrogen isotope signatures were intermediate between those of the two feed and inorganic fertilizer treatments ( Figure 418), indicating tha t CSM treatment pond trophic processes may have been intermediate to those occurring within the feed and inorganic treatment ponds. In that cottonseed meal was being utilized by direct consumption as a high protein seed meal, and as a green manure that releases nutrients to stimulate primary production during bacterial and fungal decomposition. CSM food chain/web trophic distances may have been longer than those occurring within the two feed treatments, but short in comparison to trophic distances prese nt between the inorganic fertilizer basal nutrient and INO treatment heterotrophs However, food chain/web lengths did not appear to be significantly greater within the CSM treatment in comparison to the two feed treatments, as

PAGE 178

178 nitrogen isotope signature differences between CSM large zooplankton and the applied cottonseed meal did not significantly differ from the nitrogen isotope signature differences between feed treatment large zooplankton and applied feeds within the two feed treatments ( Figure 4 36). INO treatment pond organisms were wholly dependent upon the applied inorganic fertilizer and likely limited autochthonous nutrients as their basal nitrogen sources. The longer INO pond food web/chain produced greater nitrogen isotope enrichment between the INO treatment nitrogen pool (fertilizer plus autochthonous and nonanthropogenic allochthonous nitrogen) and its dependent organisms. D ue to the extremely low N signature of the basal INO fertilizer (inorganic fertilizer ~ 0.00 N), isotopic enrichment did not increase the N signature of the various INO pond treatment trophic guilds/size assemblages to values equal to or greater than their counterparts in the other three treatments (Figures 4 11412, 418, 423, 428). Although INO large zooplankton 15N signatures were lower than their counterparts within the nonINO treatments, the N signature difference between the applied basal nutrient and large zooplankton was much greater within the INO treatment than within the non INO treatments This indicated that applied inorganic fertilizer nitrogen had passed through a greater number of trophic intermediaries, and had become isotopically enriched before contributing to somatic growth within the INO treatment taxa (Figures 4 18, 436, 438). Medium Plankton Assemblage Isotopic Signatures Medium plankton assemblage carbon isotope signatures differed among pond nutrient treatments and among sampling dates within pond treatments. Again, PRO and UNP treatment medium plankton carbon signatur es did not differ during the 10 weekly sampling periods, supporting the contention that commercial feed treatment plankton assemblages (large

PAGE 179

179 zooplankton, medium and small plankton) either directly, or indirectly derived the majority of their nutrients fro m the applied feeds as opposed to autochthonous or nonanthropogenic allochthonous nutrient sources. Although the PRO and CSM treatment applied nutrient compositions (Chapter 3; Table 31) and carbon isotope signatures differed ( Figure 4 4), PRO and CSM medium plankton assemblage carbon signatures did not differ for any of the 10 sampling periods analyzed. Cottonseed meal was believed to function as a combination of a directly consumable nutrient and organic fertilizer for stimulating phy toplankton production. Heterotroph consumption of both cottonseed meal and live foods within the CSM ponds may have obliterated any isotopic differences (within a given taxa) with the other three treatments that would have been imparted by consumption of the applied cottonseed meal nutrient alone. Additionally, the cottonseed meal carbon isotope signature only marginally differed from that of the two commercial feeds (Figure 44; note truncated y axis scale). Only two sampling period differences in medium plankton assemblage carbon isotope signature occurred between the UNP and CSM treatments near the second half of the 12week trial (consecutive sampling dates: 18 and 25 June 2006). Infrequent UNP and CSM medium plankton carbon isotope signature differences may have occurred for the same reason the PRO and CSM treatments lacked differences, namely that the two feed treatments had identical isotopic compositions. Only the INO treatment medium plankton carbon i sotope signatures consistently differed from those of the other three pond treatments, the majority of differences occurred during the latter sampling periods of the trial. This indicated that isotopic equilibrium had become established within the different treatment ponds during this time.

PAGE 180

180 Medium plankton assemblage carbon isotope signatures differed among sampling periods within each treatment, but significant differences in medium plankton carbon isotope signatures measured at the start and end of th e pond trial only occurred within the nonINO treatments. Additionally, medium plankton assemblage carbon isotope signatures followed a similar pattern within all three non INO treatments initially high C values declined to lower values at the end of the trial ( Figures 419 422). This decrease was less pronounced within the CSM treatment and as mentioned above, nonexistent within the INO treatment ( Figure 422). The lack of INO medium plankton C differences at the start and end of the trial m ay have been due to the isotopic homogeneity of the carbon pool utilized by INO treatment autotrophs (primarily small plankton assemblage). Potential isotopic homogeneity may have lead to a more rapid establishment of isotopic equilibrium ( C) of pond a utotrophs (small plankton assemblage) and heterotrophs (medium and large plankton assemblages) within the INO treatment. The conclusion that carbon isotope signatures reached isotopic equilibrium over time also was supported by the observation that when significant pairwise differences among sampling periods were present within a treatment, differences were generally between isotopically higher (lower on graph) mean s at the beginning of the trial and isotopically lower mean s later in the trial (Figures 4 19 422). The lack of medium plankton assemblage carbon isotope signature differences among consecutive sampling periods in the latter half of the study, indicates that medium plankton assemblage carbon isotope signatures had stabilized about an mean eq uilibrium value. Time Averaged Medium Plankton Carbon Isotope Differences Time averaged medium plankton assemblage carbon isotope signatures differed among treatments (Figure 423). Again, PRO and UNP medium plankton assemblage carbon isotope

PAGE 181

181 signatures did not differ, nor did PRO and CSM treatment medium plankton assemblage carbon isotope signatures differ. The INO medium plankton assemblage carbon isotope signature was significantly higher than those of the other three treatments. This indicated that the carbon incorporated by the INO treatment medium plankton assemblage may have moved through more trophic intermediaries, becoming more enriched in the heavier carbon isotope (13C) than the medium plankton assemblages of the other treatments. As stated previously, an alternate or simultaneous process occurring within the INO treatment ponds was that carbon fixed by primary producers within the INO ponds was from an inorganic carbon pool that had a much higher carbon isotope signature. Unfortunately, due to the lack of baseline carbon isotope signatures for the DIC pools being utilized by primary producer organisms within the different treatment ponds, a starting point for measuring carbon isotope fractionation occurring within the different treatment pond ecosystems was unavailable. This makes any inferences about the actual length of the food chain in these systems partially supposition. Medium Plankton Assemblage Nitrogen Isotope Signatures Medium size plankton assemblage nitrogen isotope signatur es differed among pond nutrient treatments, but did not differ among sampling periods within nutrient treatments. Medium plankton assemblage nitrogen isotope signature differences between pond treatments did not appear to follow any temporal pattern among the 12 sampling dates, and the number of nitrogen isotope signature differences was small compared to that found for carbon signature differences. The reason for the lower frequency of medium plankton assemblage N signature differences between treatments among sampling dates was likely due to the higher variation of nitrogen isotope signatures relative to those for carbon isotope signatures (Figures 419 422).

PAGE 182

182 This was somewhat unexpected, as large plankton assemblage N signatures generally had lower variation among replicates than their C signature counterparts (Figures 4 14 417). Greater N signature variation among replicates within the medium plankton assemblages, relative to replicates within the large plankton assemblages may have been due to greater trophic diversity of organisms within the medium plankton assemblage relative to those within the large plankton assemblages. Large plankton assemblage taxa primarily consisted of large zooplankton (Chapter 5), whereas the medium plankton assemblage was a trophically diverse mixture of phytoplankton and zooplankton (pers. obs.), consisting of a diverse mixture of herbivores, carnivores, and omnivores. This artificial and unrealistic characterization of the medium plankton size assembl age as a discrete trophic group/unit was due to it consisting of trophically diverse organisms, primarily mixed phytoplankton and zooplankton (e.g. rotifers, eggs, copepod nauplii, protists). The more trophically diverse medium plankton assemblage may have produced more nitrogen isotope signature variation among its replicates than that produced by a more trophically discrete and trophically functional group such as the large plankton assemblage ( Figure 4 14 4.17; Chapter 5, primarily zooplankton). PRO a nd UNP medium plankton assemblage nitrogen isotope signatures differed infrequently, differing for only one of the twelve weekly sampling dates. As previously mentioned, this was likely due to the two feeds being manufactured from identical ingredients an d the large influence of the applied feeds upon the isotopic signatures of the taxa within their respective treatment ponds. The medium plankton assemblage was a mixture of smaller zooplankton (pers. obs.; rotifers, nauplii, protists, ciliates) and phytoplankton, and was likely isotopically influenced by heterotroph consumption (feeds, small zooplankton and phytoplankton) and phytoplankton that utilized DIN derived from remineralized applied feeds.

PAGE 183

183 Cottonseed meal treatment medium plankton nitrogen isotope signatures did not differ from those within either feed treatment for any of the 12 weekly sampl es. This was likely due to the greater variation in medium plankton nitrogen isotope signature enrichment magnitudes frequently observed among replicate sampl es, relative to carbon isotope enrichment values ( Rounick and Winterbourn 1986, Post 2002aa ). Additionally, commercial feed nitrogen isotope signatures ( Figure 45) only marginally differed from that of the applied cottonseed meal (~ 1.0 N). The two fertilizer treatment (CSM and INO) medium plankton assemblage N signatures differed most frequently (4 of 12 sampling periods) among the six possible pairwise treatment comparisons ; however, CSM and INO medium plankton N signature differences did not appear to follow any discernable temporal pattern. Although repeated measures ANOVA did not find significant differences in medium plankton nitrogen isotope signatures among sampling dates within treatments, pairwise post hoc tests (Bonferroni, P < 0.05) did find a small number of differences among sampling periods within pond treatments ( Motulsky 1995; H. Motulsky pers com.) that did not appear to follow any chronological pattern (Figures 4 19 420, 422). The INO treatment had the highest frequency of significant differences in medium plankton nitrogen isotope signature among sampling periods, due to a single sampling date (31 March 2006) at the start of the trial that had a higher nitrogen isotope signature than later sampling periods. Again, mor e pseudovalues were necessary for the medium plankton nitrogen isotope analyses (8 of 288) than were required for the carbon isotope analyses (1 of 288) due to the higher nitrogen detection threshold limit of the isotope ratio mass spectrometer. Notably, less pseudovalues were required for the medium plankton assemblage isotopic analysis than for the large plankton assemblage isotopic analysis, presumably due to the greater biomass standing

PAGE 184

184 stocks (~10X biomass increase per trophic level decrease) general ly present at lower trophic levels within an ecosystem ( Paine 1980, Briand and Cohen 1987, Pauly and Christensen 1995, Valiela 19 95 Vander Zanden et al. 1999, Saito et al. 2000, Post 2002b, Hoeinghaus et al. 2008 ). Time Averaged Medium Plankton Nitrogen I sotope Differences Time averaged medium plankton assemblage nitrogen isotope signatures also differed among treatments, with only the INO treatment mean differing from those of the other three treatments (Figure 423). INO medium plankton assemblage nitr ogen isotope signature was markedly lower than the other three treatments, mirroring the pond nutrient ( Figure 4 5) and time averaged large zooplankton ( Figure 418) nitrogen isotope results. Again, this was likely due to the low nitrogen isotope signatur e ( N = 0.08 ) of the applied inorganic nitrogen fertilizer nutrient and presumed basal nitrogen source for the INO treatment ponds. Small Plankton Carbon Isotope Differences Small plankton size assemblage carbon isotope signatures differed among pond nutrient treatments and among sampling periods within pond treatments. Again, PRO and UNP small plankton carbon isotope signatures did not differ for any of the 12 weekly sampling periods. The two feed and cottonseed meal treatments differed for only a small number of early sampling intervals : PRO and CSM (2), and UNP and CSM (5). In contrast, pairwise treatment comparisons involving the INO treatment, differed for the majority of the latter sampling periods: PRO and INO (last 9 periods), UNP and INO (8 of last 9 periods), and the CSM and INO treatments differed for all 12 weekly sampling periods. The two differing trends in small plankton carbon isotope signatures, diverging (P RO and INO, UNP and INO) and converging (PRO and CSM, UNP and CSM) values among treatments, took from two to five weeks before small plankton isotopic equilibriums were established with their presumed primary carbon

PAGE 185

185 source(s) applied nutrients or autocht honous carbon source in the case of the INO treatment ponds. Small plankton assemblage carbon isotope signature differences among sampling dates occurred within all treatments (Figures 4 24 4 27). Small plankton carbon isotope signatures followed sim ilar trajectories within all pond treatments. Initially high small plankton C signature values gradually oscillated before ending at a slightly lower carbon isotope signature value. Final small plankton carbon isotope signature values differed from initial values in the two commercial feeds, but not in the case of the two fertilizer treatments. The two fertilizer treatment small plankton carbon isotope signature s were more constant and had flatter trajectories compared to the two feed treatments (Fig ures 4 24 427), resulting in no carbon isotope signature differences between the initial and final sampling periods. This may have been due to the slow and even release of DIC from the applied organic cottonseed meal (bacterial and fungal mediated remi neralization/decay) within the CSM treatment ponds, and a large, isotopically homogenous and unmeasured DIC source(s) within the INO treatment ponds (e.g., atmospheric CO2, and groundwater carbonates). Time Averaged Small Plankton Carbon Isotope Differenc es Time averaged small plankton assemblage carbon isotope signatures differed among the four treatments (Figure 4 28) O nly the PRO and UNP treatment time averaged small plankton assemblage carbon isotope signatures did not differ. However, it remai ns unclear how the small plankton assemblage (primarily small phytoplankton) could obtain carbon (and nitrogen) from applied feeds designed for consumption by relatively large heterotrophs (Figure 4 4). It is possible that carbon (and nitrogen) compounds from fine feed particles applied to ponds could have leached directly into the water where they

PAGE 186

186 were assimilated by pond phytoplankton ( Schroeder 1983, Lochmann and Phillip 1996, Lovell 1998, Lochmann et al. 2001). Although it is more likely that dissolve d inorganic carbon and nitrogen compounds (DIC and DIN) originally derived from applied feeds (remineralized) entered aqueous solution and became available to primary producers following consumption, assimilation, decomposition, respiration and excretion of feed nutrients by the various heterotrophic pond taxa (Boyd 1979, 1997, Coman et al. 2003, Grey et al. 2004). If sufficient applied feed nutrient leaching and remineralization occurred to isotopically influence the dissolved inorganic carbon (and nitrogen) nutrient pools available to primary producers within the feed ponds, taxa that utilized primary production within these ponds also would have isotopic compositions influenced by the applied feeds, even if these taxa were unable to utilize the fee ds via direct consumption. The similarity of the PRO and UNP plankton (large, medium and small) assemblage carbon isotope signatures, and the finding that the INO plankton assemblage carbon signatures were significantly lower than those of the other three treatments, supports the hypothesis that each of the four pond treatments received the majority of its carbon from the applied nutrient, or a uniform nonanthropogenic (autochthonous and/or allochthonous) carbon pool in the case of the INO treatment. Eve n at the lowest trophic levels (small plankton/phytoplankton), feed nutrients appear to have isotopically influenced feed treatment pond taxa. Small Plankton Assemblage Nitrogen Isotope Signature Differences Small plankton assemblage nitrogen isotope sign atures differed among pond nutrient treatments and among sampling dates within treatments (Figures 4 24 4 27). Small plankton assemblage nitrogen isotope signature differences between treatments only occurred among the pairwise treatment combination com parisons that involved the INO treatment ( PRO and INO, UNP and INO, and CSM and INO )

PAGE 187

187 Small plankton assemblage nitrogen isotope differences between the three nonINO and INO treatments did not follow any temporal trends with differences occurring at the beginning, middle and end of the study with equal likelihood. Additionally, the number of differences occurring between the nonINO and INO treatments also did not follow any apparent trend: PRO vs. INO (5 differences), UNP vs. INO (2 differences) and CS M vs. INO (5 differences). Again, the reason why INO small plankton nitrogen isotope signatures differed from those of the three nonINO treatments was likely due to the low nitrogen isotope signature of the applied inorganic fertilizer (~ 0.00 15N). Additionally, the applied liquid inorganic fertilizer was formulated to be in a chemical (aqueous ammonium nitrate) form easily assimilated by pond autotrophs. The INO small plankton assemblage consisted primarily of phytoplankton, and should have been t he plankton size fraction most capable of utilizing the applied inorganic fertilizer, and therefore should be the plankton assemblage most isotopically influenced by the applied inorganic fertilizer. Within pond nutrient treatments, s mall plankton nitrogen isotope signature differences among sampling dates only occurred for the two fertilizer treatments (Figures 4 26 4 27) N o within treatment differences among sampling dates occurred for the two feed treatments (Figures 424 425). Small plankton nitrogen isotope trajectories differed slightly within the two fertilizers treatments (Figures 4 26 427). CSM small plankton nitrogen isotope signatures followed a gradual sinusoidal oscillation similar to that observed for CSM small plankton carbon si gnature, whereas INO small plankton nitrogen isotope signature values followed a generally flat trajectory. Within the INO ponds, significant differences in small plankton assemblage nitrogen isotope signatures among sampling dates were due to two markedl y low nitrogen isotope

PAGE 188

188 signature mean s on two nonconsecutive sampling dates ( Figure 4 27; 6 April 2006 and 18 May 2006). Time Averaged Small Plankton Nitrogen Isotope Differences Time averaged small plankton assemblage nitrogen isotope signatures also differed among pond nutrient treatments ( Figure 428). Only the INO small plankton nitrogen isotope signature differed from those of the other three treatments, partially mirroring the nitrogen i sotope signature pattern of the four applied nutrient treatments ( Figure 4 5). Again, t his supports the contention that the nitrogen pools being utilized by the food webs within the different pond treatments were primarily derived from the applied nutrients and not by autochthonous or nonanthropogenic allochthonous nitrogen sources. Fewer pseudovalues were necessary to produce a fully balanced data set for the 2 way repeated measures ANOVA for small plankton assemblage isotope signature difference analyse s than for the medium plankton isotopic analyses, which in turn required fewer pseudovalues than the large plankton isotopic analyses. Again, this was likely due to the greater quantities of biological material commonly observed at lower trophi c levels ~ 10X biomass increase with trophic level decrease ( Paine 1980, Valiela 1995); given that the same volume of pond water was filtered to obtain each of the three plankton size fractions (small, medium and large plankton assemblages/trophic guilds). Nutrient Movement within Fish and Plankton Size Assemblages Parallel movement of isotopically homogenous nutrients within an ecosystems taxa, should produce taxa that have similar isotopic signatures A s all taxa are consuming the same or similar nutr ients and all taxa are located at similar trophic positions within the ecosystems food web.

PAGE 189

189 In contrast, serial movement of nutrients within an ecosystem should produce large isotopic differences between the lowest and highest trophic levels, due to se rial enrichment as nutrients move through the food chain/web As nutrients move from applied basal nutrients to primary producer to primary consumer to secondary consumer and to tertiary consumer, and so forth, the carbon and nitrogen of these nutrients become increasing enriched in the heavier isotopes of carbon and nitrogen (~ 0.64 C, ~ 3.03 N). This n ot only results in increasing isotopic signature with increasing trophic level, but also greater isotopic signature difference (e.g., carbon, nitr ogen, sulfur, etc ) between the presumed basal nutrient and top predator with increasing ecosystem trophic complexity/food chain length (with the exception of high rates of omnivory, cannibalism, or generalist feeding habits). Nutrient movement within po nd treatments appeared to have be en primarily in parallel within the two feed treatments serially within the inorganic fertilizer treatment, and intermediate to parallel and serial within the cottonseed meal treatment. This conclusion is based primarily upon the small range of mean nitrogen isotope signatures among the five taxa/size assemblages (small, medium and large plankton, small and large harvest fish) within the two feed treatments relative to the larger range of values ( Figure 443; greater SD am ong taxa) observed in mean nitrogen isotope signatures within the two fertilizer treatments, and the INO treatment in particular. Taxa carbon isotope signature ranges were largely equal among treatments, demonstrating the highly conserved basal nutrient c arbon isotope signature phenomena which occurs due to the low isotopic enrichment rate of carbon ( Figure 443). Pond Nutrient and Harvest Fish Carbon Isotope Signature Differences Due to the small magnitudes of carbon isotope signature differences betwe en taxa/trophic guilds, statistical comparisons of carbon isotope signature differences among treatments were

PAGE 190

190 sign neutral, and used absolute values of arithmetic differences in taxa C signatures. Actual small harvest fish and applied nutrient carbon isotope signature difference mean value signs were preserved and frequently differed ( Figure 4 30; i.e. PRO and UNP C(small fish nutrient) > 0, INO C(small fish nutrient) < 0) Carbon isotope signature differences between processed and unprocessed feed and their respective large harvest fish were well within published values (0.35 C PRO, 0.69 C UNP) for a single trophic level ( DeNiro and Epstein 19 78, Peterson an d Fry 1987, Fry 2006). The carbon isotope signature difference magnitude between the applied cottonseed meal and CSM large harvest fish was significantly greater than those of the two commercial feeds and their respective large harvest fish ( Figure 4 29). Utilizing the carbon isotope enrichment value measured for an increase of one trophic level obtained from the indoor feeding trial ( X = 0.64 C per trophic level), and the carbon isotope differences found between the two commercial feeds and their respective large harvest fish, the trophic distance estimate derived from carbon isotope signature differences corresponded to roughly one trophic level between large harvest fish and the applied nutrients within the two feed tre atments ( Equation 4.2; 0.54 PRO, 0.99 UNP). These trophic difference magnitudes ( only using carbon isotope data) were roughly those expected assuming that large harvest fish in the feed ponds were receiving the majority of their nutrition from the applie d feeds. Latent carbon isotope signatures from fry and juvenile diets of small and medium plankton may have resulted in the observed feed treatments large harvest fish carbon isotope signatures being slightly less than the expected value (1.0 trophic le vel difference) based solely upon a feed based diet (eq. 4.2; Vander Zanden and Rasmussen 1999, Vander Zanden et al. 1997, 1999, Vander Zanden and Vadeboncoeur 2002, Saito et al. 2000). Overall however, these results indicate that feed treatment large har vest fish were deriving most of their nutrition from applied

PAGE 191

191 feeds rather than live foods within the ponds, but with the possibility that fry and juveniles were deriving large portions of their nutrition from nonfeed (live) food sources before undergoing an ontogenetic shift in diet to applied feeds. The two diet composition estimation methods that utilized both carbon and nitrogen isotope signatures, Euclidean distance (ED) and IsoSource (IS), agreed well with the trophic distance estimates derived solely from carbon isotope signature differences for the two feed treatment large harvest fish groups. Estimated applied feed /nutrient usage: ~ 90% Euclidean distance (ED), and 70% IsoSource (IS) for PRO large harvest fish ~ 91% ED and 87% IS for UNP large h arvest fish. Carbon isotope signature differences between cottonseed meal and their large harvest fish ( 1.97 C CSM) corresponded to approximately three (3.08) trophic levels, based upon a trophic distance estimate derived solely from carbon isotope signature results and indoor carbon isotope enrichment rates ( 0.64 13C ; Chapter 2). Indicating that cottonseed meal served primarily as a nutrient source for the CSM treatment pond food web, and may have made only a minor contri bution to the direct nutrition of CSM treatment fish (based solely upon C results), even though fish were regularly observed feeding upon cottonseed meal broadcast twice daily upon the pond surface (pers. obs.). Three trophic levels (3.08 CSM carbon si gnatures only) separating the CSM treatment large harvest fish and their applied cottonseed meal nutrient seems a reasonable trophic distance estimate given what is known of tropical freshwater lentic ecosystems ( Estep and Vigg 1985, Cabana and Rasmussen 1994, Yoshii et al. 1999) ; i f finely ground cottonseed meal was being directly utilized by rotifers and other small heterotrophs, which were in turn consumed by larger zooplankton, prior to being consumed by CSM treatment fish. However, this simple ( C o nly) interpretation of CSM large harvest fish diet was only

PAGE 192

192 marginally supported by the two diet estimation methods that utilized both carbon and nitrogen isotope signature data ( Table 4 1; CSM large harvest fish diet: cottonseed meal ~ 12% ED and 21.4% IS large zooplankton ~ 16% ED and 22% IS, medium plankton ~ 65% ED and 53.6 IS, small plankton ~ 6% ED, 1.1 IS). The simple food chain initially postulated CSM treatment was: cottonseed meal small plankton medium plankton large zooplankton l arge h arvest fish. A slightly more complex and realistic, but still simplistic CSM treatment food web diagram ( Figure 444) was created that was roughly supported by the two (ED and IS) dual element (C, N) diet estimation methods and observation of CSM treatment fish regularly grazing upon cottonseed meal at the pond surface This alternative scenario to the simple linear food chain described above postulated that CSM harvest fish primarily received their nutrition from the applied cottonseed meal and l arge zooplankton assemblage, and that the large zooplankton assemblage primarily derived their nutrients as described in the simple linear food web/chain scenario: cottonseed meal small plankton medium plankton large zooplankton Nutrients utilized by CSM treatment primary producers were likely remineralized DIC and DIN from respiration and excretion (following CSM consumption), decay of cottonseed meal, or volatile nutrients (free amino acids, sugars) leached directly from the cottonseed meal into a queous solution. The CSM large harvest fish carbon and nitrogen isotopic signature plot position was on a line, roughly midway between the isotopic signature plot positions of the applied cottonseed meal and the CSM large zooplankton assemblage, which would be expected if CSM large harvest fish were primarily deriving their nutrition from these two sources (Figure 4 41). Unfortunately, the isotopic signature plot collinearities of the cottonseed meal, CSM large zooplankton assemblage, and CSM medium plankton assemblages, may have erroneously biased the ED and IS CSM large

PAGE 193

193 harvest fish diet estimates in favor of the CSM medium plankton assemblage (Figures 441, 453). The inorganic fertilizer used for the INO treatment did not contain carbon, preventing a trophic distance estimate based solely upon carbon isotope signatures between the INO fertilizer and INO large harvest fish. Back calculations were made to determine the trophic distance and carbon isotope signature value (and likely identity) of the basa l primary producer within the INO pond food web. It was expected that the INO small plankton assemblage ( 23.26 C ) would be the primary candidate for the basal trophic group (1) within the INO treatment food web. However, the carbon isotope signature distance (3 trophic levels 0.64 13C per trophic level or 1.92 C ) between the INO treatment large harvest fish (Figure 4 11; INO lg fish 20.15 C ) and the primarily mixed rotifer and phytoplankton INO medium plankton assemblage (estimated as 1 consumer; INO med plank 22.19 C ) had the closest isotopic fit to the back calculated primary producer ( Table 4 3; estimated 1 producer ~ mean 22.50 C ) group postulated to be the base of the food web within the INO treatment ponds ( Figure 4 28). The INO treatment medium plankton assemblage was only slightly less enriched in 13C than the back calculated INO treatment 1 producer ( INO med plank 22.19 C INO sm plank 23.26 C ). However, the C signature sign of the INO m edium plankton assemblage relative to the back calculated INO pond primary producer [ C(post 1 prod) < C( med plank)] indicated that it was unlikely that the INO medium plankton assemblage was the base of the food chain within the INO treatment ponds. Whereas the INO small plankton assemblage was of the correct C signature sign and magnitude [ C(post 1 prod) > C(sm plank)] to indicate that it was a possible candidate for the base of the INO treatment pond food chain.

PAGE 194

194 Large Harvest Fish and Applied Nutrient Nitrogen Isotope Differences Nitrogen isotope signature differences between large harvest fish and their respective pond treatment nutrients differed among all groups, except the PRO and UNP treatments. The finding that nitrogen isotope enrichment magnitudes between large harvest fish and their respective applied nutrients did not differ for the two feed treatments provides support for the contention that fish within the applied feed ponds were consuming and assimilating feed nutrients into their bodies for somatic growth, largely to the exclusion of pond live prey. If feed treatment large harvest fish N enrichment levels had differed, it could have been argued that one or both of the feed treatment fish groups were obtaining a significant portion, or even a majority of their nutrition from a nonfeed source ( Figure 429). There is a remote possibility that both PRO and UNP large harvest fish were obtaining most of their nutrition from an isotopically similar, nonanthropogenic basal nutrient source. However, I feel this is unlikely as the most conspicuous potential live food sources within the two feed treatments (small, medium and large plankton) were surveyed for C and N signature profiles, and their profiles did not indicate that live foods were major prey items for large harvest fish within these treatments. Additionally, harvest fish nitrog en isotope signatures differed among treatments (INO fish lower N; Figures 4 11412 ), indicating that some, if not all of the applied nutrients were responsible for the detectable isotopic signature differences of har vest fish within the ir respective p ond treatments N amely, that isotopic signature differences in harvest fish among treatments were an indicator of dietary source differences among treatments (DeNiro and Epstein 197 8, 1981a Fry et al. 1999, Fry 2006). Nitrogen signature differences between feed treatment large harvest fish and their respective feeds were extremely close (2.94 N PRO, 3.05 N UNP) to the mean

PAGE 195

195 nitrogen enrichment magnitude observed for a single trophic level change in the indoor feeding trials (Chapter 2; 3.03 N per trophic level) Which produced e stimated trophic distances (Equation 4.4) of approximately one trophic level separating commercial feeds and large harvest fish within both fee d treatments (Table 4 4; N 0.97 trophic level PRO, N 1.00 trophic level UNP). Nitrogen isotope signature based trophic distance estimates between commercial feeds and their respective large harvest fish agreed well with those estimated solely using carbon isotope data ( C 0.54 trophic level PRO, C 0.99 trophic level UNP), especially for the UNP treatment. Trophic distance estimates derived from nitrogen isotope signature differences between large harvest fish and their applied fertilizer nutrients were: 1.5 trophic levels separat ing cottonseed meal and the CSM large harvest fish ( N 1.48 trophic level CSM), and approximately 2.5 trophic levels separat ing the inorganic fertilizer and INO large harvest fish ( N 2.48 trophic level INO). Based on nitrogen isotope signature differences (Equation 4.4), the greatest trophic distance between swordtail top predators and applied basal nutrients within the four nutrient treatments was estimated to be the inorganic fertilizer treatment, followed by the cottonseed meal treatment, and lastly by the two commercial feed treatments (Figures 4 29 430, Table s 44, 410). N itrogen isotope signature based trophic distance estimates, supported the main alternate hypothesis, that f ish primarily utilized feeds, and that fertilizer treatment fish primarily obtained their nutrients from the applied fertilizers via the pond f ood webs. Carbon and nitrogen isotope signature based trophic distance estimates (Equations 4.2, 4.4) between CSM large harvest fish and their cottonseed meal basal nutrient ( C 3.08 trophic levels, N 1.48 trophic levels) did not agree as closely as their feed treatment counterparts, but

PAGE 196

196 did not dramatically differ (difference ~ 2X). The initially postulated CSM treatment food chain : cottonseed meal small plankton med ium plankton large zooplankton CSM large harvest fish, translated to an expected overall nitrogen isotope signature difference of 12.12 N between CSM large harvest fish and the applied cottonseed meal nutrient, which was much greater than the actu al measured difference (4.48 N ). This indicated that CSM large harvest fish and/or their zooplankton prey were likely consuming organisms from different trophic levels than those strictly postulated by the simple linear food chain model (e.g., omnivory absent). Alternately, or in conjunction with this explanation is that actual trophic assemblages/guilds did not concur with the three plankton size assemblages (experimental artifacts of mesh size selection), which is almost certainly the case. If plankton size assemblages and trophic positions do not directly correspond, plankton size assemblage isotope signatures will not be an accurate representation of distinct trophic levels/guilds within treatment ponds, and trophic distance estimates using Equations 4.2 and 4.4 will deviate from actual trophic distances occurring between trophic groups/guilds. Both carbon and nitrogen diet estimation methods (ED and IS) concluded that CSM large harvest fish diet was a composite of the four isotopically analyzed components ( Table 4 1). This indicated that CSM treatment swordtails were omnivorous generalists and part of a complex food web ( Figure 4 44), as was indicated within the examined literature (Axelrod 1991, Tamaru et al. 2001). Inorganic fertiliz er nitrogen (~ 0.00 N ) was incorporated into primary producers (small plankton assemblage) with a small degree of isotopic nitrogen enrichment (Table 4 4; INO small plankton assemblage and fertilizer: 1.19 N ), which commonly occurs within nitrog en rich (eutrophic) ecosystems (Wada and Hattori 1978, Wada 1980, Evans et al. 1996,

PAGE 197

197 Pennock et al. 1996, Waser et al. 1998, Waser et al. 1999, Grey et al. 2004, Hall 2004 De Brabandere et al. 2007, GregoryEaves 2007). Actual INO large harvest fish and inorganic fertilizer nitrogen isotope differences translated to a trophic distance of ~ 2.5 trophic levels. This total food chain trophic distance was close to the expected number of trophic levels (3) believed to exist within the INO treatment ponds base d upon the simple food chain : inorganic fertilizer small plankton medium plankton large zooplankton INO large harvest fish. The trophic distance between INO large harvest fish and the INO small plankton assemblage (1 producers) calculated solely from nitrogen isotope signature differences (Equation 4.4) and the predicted number of trophic levels present within the simple food web (three trophic levels) differed by only ~ 0.5 trophic levels, assuming small plankton assemblage and fertilizer N signature values were roughly equal. The calculated trophic distance of 2.5 is even more plausible than the estimated 3.0 given the omnivorous and possible juvenile ontogenetic dietary habits of swordtails ( Table s 44, 410). Small Harvest Fish and Pond Nutrient Isotopic Differences Unlike large harvest fish, carbon isotope signature difference magnitudes between small harvest fish and pond nutrients did not differ among pond treatments ( Figure 4 30). As was previously mentioned, statistical comparisons among treatments were sign neutral and used absolute value differences. A ctual carbon isotope signature difference value signs between groups (e.g., PRO small fish C and processed feed C) frequently differed (Figure 430). Carbon isotope signatur e differences between the two commercial feeds and their respective small harvest fish translated to roughly one negative trophic level shift ( 0.89 PRO, 0.27 UNP) T hese nonsensical results (negative enrichment sign s ) indicated that commercial feeds were not as predominant a dietary component for feedtreatment small harvest fish as feeds

PAGE 198

198 were for large harvest fish ( Table 4 1). This conclusion also was supported by the PRO and UNP small harvest fish ED and diet estimates. When comparing the diets o f PRO large harvest fish ( Table 4 1; PRO(lg fish): processed feed ~ 90% ED and 70% IS ) and small harvest fish, feed only marginally remained the primary dietary component of small harvest fish ( Table 4 1; PRO(sm fish): processed feed 35% ED and 43.5% IS). The lesser importance of applied feed for small harvest fish, relative to their large fish counterparts also occurred within the UNP treatment ED and IS diet estimates ( Table 4 1; UNP(sm fish): unprocessed feed 37% ED and 22.7% IS). Dietary (trophic) e stimates based solely upon carbon isotope signature differences ( i.e., a single element ; Figure 452) are inferior to estimates based solely upon nitrogen isotopes or estimates based upon both carbon and nitrogen signatures ( DeNiro and Epstein 1978, 1981a, Fry 2006). It is likely that large proportions of other dietary items (live feeds), were obscuring the carbon isotope signatures of ingested feeds within PRO and UNP small harvest fish ( Table 4 1), possibly due to a greater juvenile dependence upon live foods relative to older fishes. Roughly two trophic levels ( 1.85; Table 41, based on 13C ) separated the CSM small harvest fish and the cottonseed meal pond nutrient This was roughly one trophic level less than the calculated trophic distance betwe en the CSM treatment large harvest fish and the cottonseed meal pond nutrient (Table 4 1, based on 13C) and approximately 2 trophic levels less than that predicted by the simple, initially postulated linear food web ( cottonseed meal small plankton medium plankton large zooplankton s mall fish). It is much more likely that actual CSM treatment pond trophic interactions resembled a complex food web ( Figure 444) as supported by the isotopic signature data and the ED and IS CSM harvest fish diet estimates ( Table 4 1), rather than the initially postulated simple but longer food chain.

PAGE 199

199 Unlike small harvest fish f rom the two commercial feed treatments, CSM small harvest fish may actually have been utilizing the applied nutrient to a greater degree t han large harvest fish, lessening the isotopic distance separating CSM small fish and their applied CSM nutrient. Alternately or in addition to this possibility, CSM small harvest fish may have been utilizing the small and medium plankton assemblages to a greater degree than their large harvest fish counterparts. Large and small CSM harvest fish were probably exploiting the different plankton trophic guilds to differing degrees (ontogenetic dietary habit shift, trophic specialization, microhabitat utiliza tion ), including the three plankton size assemblages and through direct consumption of cottonseed meal ( Lemly and Dimmick 1982, Holbrook and Schmitt 1992, Jones and Waldron 2003, Melville and Connolly 2003 ). This seems entirely possible for an omnivorous, generalist diet species such as the swordtail, especially if mouth gape and microhabitat usage differences between small fry/juveniles and adult fish size classes were present. Omnivory and carnivorous predation from multiple trophic levels for fish and larger invertebrates were likely responsible for the shorter trophic distance actually measured (1.85 trophic levels ; Table 4 1 C), versus the four trophic levels/distance that were predicted to exist between the CSM small harvest fish and the applied cottonseed meal (Briand and Cohen 1987, Post 2002a 2002b, Cohen et al. 2002) Cottonseed meal may have been the best basal nutrient source candidate for CSM treatment small harvest fish based solely on arithmetic C signature differences, and the indoor trial carbon isotope enrichment rate (Chapter 2; 0.64 C per trophic level). However, the two diet estimation methods (ED and IS) that utilized both carbon and nitrogen isotope data to determine fish diet, came to highly divergent es timates, both from each other and the simple carbon isotope signature difference based trophic distance estimation method ( Table

PAGE 200

200 41; CSM(sm fish): cottonseed meal 22% ED and 0.6% IS, large zooplankton 25% ED and 63.3% IS, medium plankton 39% ED and 2.0% I S, small plankton 14% ED and 34.1% IS). Conspicuous differences between the CSM small harvest fish ED and IS diet estimations, and the high diet inclusion (percentage) estimate for medium plankton (ED) may have been due to the CSM medium plankton assemblages isotopic composition being intermediate (isotopically collinear) to CSM large zooplankton and the applied cottonseed meal nutrient ( Figure s 4 41 453; Schwarcz 1991, Phillips 2001). This may have produced an overestimation of the importance of the CSM medium plankton assemblage within the CSM small harvest fish diet. Additionally, fry and juvenile ontogenetic dietary habits and the lower body weights of small harvest fish relative to large harv est fish may have contributed to greater isotopic variation of small fish (relative to large fish) for both carbon and nitrogen. Due to the possible effects of greater juvenile herbivory and omnivory (ontogenetic dietary habit differences ), a more trophi cally diverse and lower trophic level diet (planktonic and epiphytic plant and animal prey) along the pond perimeter, changes in diet due to seasonal plankton community changes/succession, and the greater effects any dietary changes would produce within a smaller fishs isotopic profile because of lower fish to prey weight ratios. Greater small harvest fish isotopic variation, due to greater dietary niche breadth and lower fish mass relative to potentially shifting food source availabilities, would likely contribute to greater errors in small harvest fish diet estimates derived from methods that utilize organism isotope signatures. A roughly 4X initial individual organism weight gain is generally considered the minimum weight increase necessary to properl y describe and differentiate organism dietary shifts in diet studies which utilize stable isotope techniques ( Fry 2006).

PAGE 201

201 INO small harvest fish carbon isotope signature (INO sm fish 20.29 C ) was used to trophically back calculate the carbon isotope signature of the basal plankton assemblage within the INO treatment food web (small fish estimated 1 producer 22.21 C), which differed only slightly from the INO large harvest fish estimate d basal trophic group carbon isotope signature (large fish estimated 1 producer 22.07 C). The INO treatment plankton assemblage with the C signature closest in value to the estimated basal plankton group was the small plankton assemblage ( 23.26 C). This relatively large isotopic difference (1.05 C) between the actual and estimated small plankton carbon isotope signatures was likely due to plankton size assemblages being som ewhat erroneously a ttributed to different functional trophic a ssemblages due to plankton sieve selections rather tha n reflect ing actual trophic divisions within the pond ecosystem. Unfortunately, the large variations observed in C signature data relative to the typically low 13C enrichment rates (Chapter 2; 13C = 0.64 ) that occur between trophic levels, made predictions of prey carbon isotope signature values via back calculation of 13C signatures of limited value within the INO treatment, relative to N signature back calculation analysis ( Smith and Wal ker 1980, Peterson and Fry 1987, Post 2002a Fry 2006). Similar to what may have occurred within the CSM treatment, the lower average body weight of INO small harvest fish (Chapter 3; Figure 3 14) relative to INO large harvest fish may have been a factor contributing to the large ED and IS diet estimate differences between INO small and large harvest fish. In that, low average body weight s of INO small harvest fish and dietary differences among individual fish, may have inflated variation in dietary estimates due to the greater isotopic influence of prey and potential diet differences upon smaller fish. Additionally, higher INO swordtail cannibalism rates due to food limitation, which may be

PAGE 202

202 indicated by the relatively greater population percentage of small fish (Chapter 3; Figure 3 12) within the INO treatment, and size variation among individual INO large harvest swordtails (Chapter 3; Figure 3 15) may have further increased INO large harvest fish isotopic signatures ( 13C, 15N) relative to their small harvest fish counterparts. Small Harvest Fish and Applied Nutrient Nitrogen Isotope Differences N itrogen isotope signature differences between small harvest fish and their respective nutrients differed among all pairwise treatment combinations, e xcept for the PRO and UNP treatments. Trophic distance estimates based upon nitrogen isotope signature differences between small harvest fish and their respective applied nutrients, and the mean rate of nitrogen enrichment (Chapter 2; 3.03 N ) per trophic level derived from the indoor trials were approximately: a single trophic level within the two feed treatments (1.08 PRO, 1.11 UNP), ~ 1.7 trophic level (1.67 CSM) within the CSM treatment and roughly two trophic levels (2.12 INO) within the INO treatment Trophic position/distance estimates ( N signature only) for groups believed to be trophic intermediaries between small harvest fish and pond nutrient treatments or directly driving fish production in the case of applied feeds, were well wit hin expected values. One trophic level between the two commercial feeds and their respective small harvest fish indicated that these fish were likely obtaining the majority of their nutrition from the applied feeds. In contrast to the C signature trop hic distance estimates, the estimated number of trophic levels for feed treatment small harvest fish was fractionally higher than an integer value of one unexpectedly m aking small harvest fish within the feed treatment ponds appear to be trophically higher than their large harvest fish counterparts ( Table 4 4).

PAGE 203

203 A possible explanation for why feed treatment small harvest fish had higher estimated trophic positions than their large harvest fish counterparts was that smaller fish were obtaining a larger po rtion of their nutrition from live foods than larger fish. This is a reasonable assumption if smaller fish were feeding in the shallow, weedier areas of the pond. Also, their smaller mouth gape may have forced them to utiliz e planktonic and epiphytic fa una and flora to a higher degree than larger fish. Zooplankton consuming applied feeds would tend to have nitrogen isotope signatures on par with fish that directly consumed applied feeds (Figures 411 412, 418), and phytoplankton that were utilizing remineralized DIN derived from applied feeds also would have nitrogen isotope signatures similar or closer to that of the applied feeds ( Figure 4 28). If feed treatment zooplankton and phytoplankton were being heavily grazed upon by small juveniles and f ry to a greater degree than their large harvest fish counterparts these small fish would be come further enriched in 15N, and tend to appear to be at a higher trophic level than their large fish counterparts (Figures 4 11 412; small feed fish N > lar ge feed fish N). Larger fish (pers. obs.) were more often in the deeper areas of the pond, where heavy grazing and lack of cover may have reduced zooplankton standing stock s and reduced their importance as a food source for larger fish within the two feed treatments. These processes may have been occurring to a lesser extent within the CSM treatment (Figures 4 11 4 12; small CSM fish N > large CSM fish N), but did not appear to be occurring within the INO pond treatment (small INO fish N < large INO fish N). The limitations of only using simple arithmetic differences in the isotopic signature s of a single element to evaluate dietary candidates was apparent as isotopic distance information of a single element only evaluates potential food source s along a single dimension. Different food source candidates m ay have similar isotopic signatures for one element (e.g. carbon) but differ

PAGE 204

204 for another (e.g., nitrogen) I f only a single element (carbon) is examined, two (or more) potential dietary candidates may appear to be isotopically similar or even identical ( Figure 452). If this is the case, no conclusions can be drawn as to which of the food sources is being utilized to a greater degree if differential prey/nutrient utilization is present ( Schwarcz 1991, Ben David and Schell 2001, Phillips 2001, Fry 2006). When carbon and nitrogen isotope enrichment differences between feed treatment small harvest fish and their applied feeds were examined together, it was clear that carbon isotope signature data alone could not adequately determine trophic position (Table 4 4; 0.89 PRO, 0.27 UNP resulting in negative trophic levels between small feed fish and applied feeds). When carbon and nitrogen isotope signatures were used in conjunction, more differentiation was possible among potential prey candidates m aking determinations of which prey candidates were isotopically plausible and which were not, easier, and in some cases, possible (Figures 4 39 442). Diet Estimates Using Both Carbon an d Nitrogen Isotope Signatures Although the Euclidean distance (ED) and IsoSource linear algebra program (IS) fish diet estimates were previously mentioned, a more thorough examination and interpretation of those results are given here. Both the ED and IS PRO and UNP large harvest fish diet estimates (Table 41) agreed well with simple arithmetic differences that used either carbon or nitrogen signature s (~ 1 trophic level) in concluding that applied feeds were the primary source of nutrition for large harvest fish within the two feed treatments. This supported the primary alternate hypothesis that large feed treatment fish primarily utilized feed for their nutritional needs. PRO feed treatment large fish diet was estimated to be primarily composed of the applied feed (70% ED, 90.1% IS), as was the UNP treatment large fish diet (90% ED, 91.1% IS). PRO and UNP small harvest fish

PAGE 205

205 diet estimates were much more catholic, consisting of roughly equal proportions of applied feeds and the three plankton size assemblages ( Table 4 1). CSM large harvest fish ED and IS diet estimates were roughly similar and both concluded that the medium plankton assemblage was the primary nutrient source (51%ED, 53.6% IS). CSM small harvest fish ED and IS diet estimates differed greatly, with the ED method estimat ing medium plankton (40%) assemblage as the primary dietary component with nearly equal percentages of the applied cottonseed meal and two remaining plankton assemblages, whereas the IS method estimated CSM small harvest fish diet as being roughly two thirds large zooplankton to one third small plankton (Table 4 1). As previously mentioned the CSM medium plankton assemblages isotopic coordinates lie roughly midway and almost directly upon a line connecting the isotopic coordinates of the CSM large zooplankton assemblage and the applied cottonseed meal nutrient (Figure 441). This has bee n cited as a major confounding factor in delineating diets based upon isotopic signatures, as a diet consisting of equal proportions of CSM large zooplankton and cottonseed meal will produce an isotopic signature similar to a diet consisting primarily of C SM medium plankton ( Figure 452 ) within a consumer organism (such as the swordtail) given equal assimilation efficiencies ( Schwarcz 1991, BenDavid and Schell 2001, Phillips 2001, Fry 2006). For INO treatment large harvest fish, both the ED and IS meth ods estimated that large zooplankton was the primary dietary component (65% ED, 86.5% IS) with smaller contributions by medium (25% ED, 8.6% IS) and small (10% ED, 4.9% IS) plankton assemblages. INO treatment small harvest fish dietary estimates somewhat differed for the two methods, medium plankton (47% ED) followed by large zooplankton (34% ED) were the primary food sources according to the ED method, whereas large zooplankton was the primary food source (67.3% IS)

PAGE 206

206 followed by small phytoplankton (25.3% IS) for the IS method. Again, isotopically collinear N and C signature plots for INO large medium and small plankton may have biased the ED INO small harvest fish diet estimates in favor of INO medium plankton. Large and small harvest fish lined u p vertically (similar C dissimilar N : large fish enriched ~ 1.1 15N relative to small fish) within the INO treatment isotopic plot ( Figure 442), but lined up horizontally within the isotopic plots of the other three treatments (Figures 439 442). This indicated there was a greater trophic difference between the INO small and large harvest fish groups relative to their nonINO treatment counterparts, whereas there was greater trophic similarity between small and large harvest fish within the nonINO treatment groups relative to those within the INO treatment. These findings suggest that limited, direct consumption of smaller fish by larger fish (cannibalism) may have occurred within the INO treatment ponds a lthough INO treatment fry and small harvest fish were not included as a potential prey for INO large harvest fish using the ED and IS diet estimation methods. Although ED and IS large harvest fish diet estimates differed slightly in their predicted in clusion rates (% of diet), dietary importance rankings were similar between the two methods. Both methods determined that feeds were the primary nutrients for feed treatment large harvest fish, and that medium plankton and large zooplankton were the prima ry food sources for CSM and INO large harvest fish, respectively. Simple arithmetic differences in the isotopic signatures of nutrients/taxa for a single element ( C or N) also suggested (via trophic distance estimates) that large harvest fish de rived most of their nutrition from applied feeds within the two feed treatments and from the pond food web in the two fertilizer treatments. Although it was an expected outcome of the trial

PAGE 207

207 that fish ate applied fish feeds and that fish fertilized ponds u tilized live foods, actual confirmation or rejection of the primary alternate hypothesis was desirable. Dietary importance rank agreement between the Euclidean distance and IsoSource computer program was much lower for small harvest fish compared to larg e harvest fish within all four treatments (Tables 4.5 4.8). When small harvest fish dietary component estimates were ranked in order of importance from highest to lowest, no rank orders completely concurred between the two diet estimation methods for any of the four pond nutrient treatments. Again, this may have been due to the more generalized diet believed to occur for smaller fish, as well as their lower predator to prey weight ratios. This would tend to increase small harvest fish isotopic variation within treatments due to dietary differences of individual fish (Estep and Vigg 1985, Beaudoin et al. 1999). Large Zooplankton and Harvest Fish Isotope Differences Carbon isotope signature differences (signed values) between large harvest fish and their r espective large zooplankton assemblages ( Chapter 2; 0.64 C per trophic level) translated to roughly 0.41 PRO, 0.21 UNP, 0.44 CSM and 1.18 INO trophic levels. As previously mentioned, negative trophic distance estimates denote isotopic depletion of a presumed predator group relative to a presumed prey group (i.e., presumed predator has a lighter isotope signature than presumed prey), which indicates a lack of trophic coupling (resource utilization) between groups (e.g., large zooplankton and large harvest fish). These results indicate that large harvest fish in the PRO and CSM ponds were not heavily utilizing the large zooplankton standing stock s within their ponds, according to carbon isotopic signature differences. PRO and CSM large zooplankton assemblages likely utilized processed feed and cottonseed meal directly, accounting for the lower carbon isotope signature difference

PAGE 208

208 magnitudes and negative trophic distance (unutilized resource) estimates between large harvest fish and large zooplankton within these treatments (Figures 4 11, 418, 431). Within their respective treatments, processed feed and cottonseed meal nutrients appeared to be moving directly into large harvest swordtails and the large zooplankton assemblages (parallel), indicating a lack of a strong and direct trophic coupling between these two taxa. CSM large harvest fish and large zooplankton carbon isotope signature difference results were more difficult to interpret. Although cottonseed meal was applied as an organic fertilize r, larger fish were regularly observed ingesting and may have been partially assimilating finely ground cottonseed meal (pers. obs.), which contains high amounts of crude protein. Contradicting this observation, both the ED and IS diet estimations indicat ed that CSM large harvest fish primarily consumed medium plankton assemblage prey ( Table 4 7), which may have been an erroneous estimate caused by the isotopic composition collinearities of the applied cottonseed meal, CSM large zooplankton, and medium pla nkton assemblages (Figure s 441, 4 53). Being visual predators ( Axelrod 1991, Jones et al. 1998a, 1998b, 2007), i t is more likely that CSM large harvest fish were consuming both cottonseed meal and large zooplankton ( Figure 444), which c ould produce a n i sotopic signature profile similar to that of a diet consisting primarily of CSM medium plankton within a fish consumer ( Figure 441 ). UNP large harvest fish and large plankton carbon isotope signature difference results were less clear, and appeared to i ndicate that UNP large harvest fish were primarily consuming applied feeds and limited amounts of large zooplankton. Alternately, large UNP fish may have been more dependent upon live feeds when smaller relative to their PRO large fish counterparts, but no obvious cause for such a scenario makes this seem unlikely Regardless, the positive trophic distance estimate ( Table 4 4) and low carbon isotope signature difference between UNP

PAGE 209

209 large harvest fish and large zooplankton indicated that (based on C) th e large zooplankton assemblage was probably a minor UNP large harvest fish nutrient source. As expected, INO large harvest fish and large zooplankton C signature differences indicated that fish were likely utilizing large zooplankton as a major food s ource within the INO ponds. INO large harvest fish carbon isotope signatures were higher (positive trophic distance/proper magnitude ~ 1.0 trophic level) than their large zooplankton assemblages (Figures 411, 418; Table 4 4 ), making large zooplankton a plausible prey item candidate for INO large harvest fish. Within their respective pond treatments, small harvest fish and large zooplankton carbon isotope signature difference magnitudes corresponded to trophic distance estimates of 1.84 PRO, 1.14 UNP, 1.67 CSM, and 0.95 INO (Chapter 2; 0.64 C per trophic leve l ). These trophic distance estimates ( Table 4 4) indicated that feed and CSM treatment small harvest fish groups were not (wrong sign, C < 0) obtaining the majority of their nutrition f rom the large plankton assemblage. In contrast, the IS and ED diet estimation methods predicted that feed treatment small harvest fish were obtaining most of their nutrition from the applied feeds, except for the UNP small harvest fish IS diet estimate w hich indicated that small plankton were their primary nutrient source ~ 34.5% ( Table 4 1). IS and ED diet estimates also determined that feed treatment small harvest fish were receiving large portions of their nutrition from the medium and small plankton assemblages: (PRO) ~15 33% medium and ~2035% small plankton, and (UNP) ~ 29 32% medium and ~ 1735% small plankton ( Table 4 1). Interestingly, feed treatment small harvest fish were estimated to be consuming relatively minor amounts of large zooplankton ( Table 4 1; PRO 0.2 14%, UNP 1 14%; IS and ED). This may have been due to the tendency

PAGE 210

210 for large zooplankton to congregate at the surface in deeper areas of the pond, and small fish to inhabit the shallower, weedier areas along the pond perimeters (pers. obs). Unfortunately, a thorough survey of large zooplankton densities in relation to distance from shore and water depth was not conducted due to time constraints. Based on carbon isotope signature differences ( Table 44) CSM small harvest fish did not appear to be heavily utilizing large zooplankton, but may have been consuming prey from other trophic levels or directly ingesting cottonseed meal as was postulated for CSM large harvest fish. In contrast, the IsoSource diet estimate determined large zoop lankton to be the primary nutrient source for CSM small harvest fish, and the ED method ranked large zooplankton as their second most important food item ( Table 4 7 ). Interestingly, although the IS method ranked large zooplankton as the primary food sourc e for CSM small harvest fish, the IS method ranked medium plankton as the primary prey of CSM large harvest fish ( Table 4 7). Given what is know n of predator feeding behavior (mouth gape and maximum prey particle size ) from optimal foraging theory ( Schmitt and Holbrook 1984a, b, Osenberg et al. 1992, Mittlebach et al. 1999) and microhabitat differences between small and large CSM treatment fish and large zooplankton assemblages (deeper water) it seems unlikely that CSM large harvest fish would be ut ilizing more medium than large zooplankton, while CSM small harvest fish would be consuming more large plankton than medium plankton. Again, this may be due to the collinear isotopic signature plot positions ( C and N) of medium and large plankton, a nd cottonseed meal within the CSM treatment (Figures 441, 444, 453). The ED diet estimation method ranked plankton assemblages and applied cottonseed meal in the same increasing order of dietary importance for CSM small and large harvest fish ( Table 47; medium plankton, large zooplankton, cottonseed meal, and small plankton, respectively ).

PAGE 211

211 M edium plankton were ranked higher in dietary importance than large plankton for CSM large harvest fish, which again may have be en due to the collinear isotopic plot positions of the potential food items ( Figures 441, 4 53) and associated potential diet estimate bias. Regardless of these potential issues, because the IS and ED methods use both carbon and nitrogen isotope data, more confidence can be attributed to their estimates due to the simultaneous use of isotopic signatures two element rather than that of only a single element. INO small harvest fish appeared to be primarily utilizing large zooplankton, based on trophic distances derived from carbon isotope signature differences ( Table 4 4; ~ 0.95 trophic levels) between INO small harvest fish and large zooplankton. This was not surprising given the lack of an ingestible applied organic nutrient and the visually mediated feeding behavior of swordtails. However, the two methods used to estimate INO small harvest fish diets (IS and ED) both utilize carbon and nitrogen isotope signatures, and produced different dietary importance rankings for the three plankton assemblages that were isotopically analyzed ( Table 4 8). The IS diet estimate predicted that large zooplankton were the major nutrient source for INO small harvest fish, whereas the ED method predicted medium plankton to be their major food source ( Table 4 8). Similar to the CSM treatment results, Euclidean distance INO small harvest fish diet estimation may be biased in favor of the INO medium plankton assemblage, due to the carbon and nitrogen isotope signature plot position of the medium plankton assemblage being roughly midway, and nearly collin ear to the INO small and large plankton plot points (Figures 4 42, 453). If small phytoplankton and epiphytic algae are important food sources for INO small fry and juveniles living within the physical cover (predator avoidance) of emergent and submerge d

PAGE 212

212 plants in the shallows along the pond perimeter, and large zooplankton become important adult prey items as juvenile fish grow and move to deeper water, a diet ary history consisting primarily of mixed small phytoplankton (juvenile diet) and large zooplankton (adult diet), could isotopically resemble a diet consisting primarily of medium plankton using the Euclidean distance diet estimation method and INO plankton isotopic signature values (Figures 442, 445). INO pre harvest fry ED and IS diet estimates ( Table 4 2) closely agreed with INO small harvest fish ED diet estimates ( Table 4 1), with the exception of a small increase in the diet inclusion rate of the small plankton assemblage (19%) in the INO small harvest fish ED diet estimate. The INO preharvest fry (ED) and small harvest fish (ED and IS) diet estimations indicated that medium plankton were a more important food item than small plankton (largely autotrophs) for this group of fish ( Table 4 2). The INO pre harvest fry IS estimate had much higher dietary inclusion rates for large and small plankton, at the expense of a lower inclusion rate for the medium plankton assemblage [ Table 4 1: INO small harvest fish (medium plankton 55% ED, 65% IS); Table 4 2: INO pre harvest fish (7.5% IS medium pl ankton)]. Fish size series (fry, juvenile and adult) isotopic profiles proved useful within the different treatment ponds as a means of determining if natal fry and small juvenile fish were obtaining most, or at least a detectable portion of their nutrit ion from small phytoplankton and epiphytic algae. INO pre harvest fish (fry and juvenile fish ) isotope signature s, indicated that it was more likely that INO small harvest fish were heavily predating upon the INO medium plankton assemblage, than an isotop ically similar composite diet of INO small and large zooplankton ( Figure 4 42). This dietary scenario is suggested by the direct observation of large macrozooplankton assemblages congregating near the water surface in the deeper, central portions of the ponds where smaller swordtails were largely absent. A more thorough survey of

PAGE 213

213 fry isotopic signatures would have been useful in determining the role, if any, of small phytoplankton in the direct nutrition (cannibalism) of larger swordtails within the INO treatment ponds. Large Harvest Fish and Large Zooplankton Assemblage Nitrogen Isotope Differences The INO large harvest fish and large zooplankton nitrogen isotope signature difference was significantly greater than those of the other three groups. As a general visual trend (not statistical), harvest fish and plankton assemblage nitrogen isotope signature differences were much lower within the two feed treatments than those within the two fertilizer treatments ( Figure 4 32 ; Table 4 4). Nitrogen isoto pe signature differences between large harvest fish and their respective large zooplankton assemblages corresponded to trophic distances of 0.08 PRO, 0.22 UNP, 0.65 CSM and 0.82 INO (Chapter 2; 3.03 N per trophic level). Nitrogen isotope signature derived trophic distance estimates between large harvest fish and large zooplankton assemblages for the PRO, UNP and INO treatments agreed well with those derived from carbon isotope signatures ( Table 4 4). Specifically, it seems unlikely that large zoopla nkton were a major food item for feed treatment large harvest fish, as much higher trophic distance s ( ~ 1.0 ) would have been expected between predator s and prey. It seems likely that INO large harvest fish were receiving a large portion of their nutrition from the large zooplankton assemblage according to nitrogen isotope signature differences between the two taxa. In addition to nitrogen isotope signature difference based trophic distance estimates, and as stated earlier, PRO and UNP large harvest fish IS and ED diet estimates indicate that feeds were their primary nutrients sources ( Tables 4 1, 45 46, 4 10).

PAGE 214

214 Nitrogen isotope signature based trophic distance estimates, indicated that large harvest fish from the two fertilizer treatments may have utilized large zooplankton standing stocks to a greater degree than their counterparts within the two feed treatments, and more so within the INO treatment ponds than within the CSM treatment ponds ( Table s 44, 4 10) Nitrogen isotope signature difference derived trophic distance estimates did not eliminate CSM large zooplankton as a potential major dietary component for CSM large harvest fish, unlike the trophic distance derived from carbon isotope signature differences ( Table 4 4; C 0.44 trophic leve ls). Again, this suggests that nitrogen isotope signature based trophic distance estimates are superior to those provided by carbon isotope signatures. When CSM taxa carbon and nitrogen isotope signatures were considered simultaneously for diet estimation (via the ED and IS methods), several plausible CSM large harvest fish dietary scenarios occurred. One possibility was that CSM large harvest fish were subsisting primarily upon the medium plankton assemblage, which was predicted by the ED and IS diet est imation methods ( Table 4 1). Another possible scenario was that CSM large harvest fish were primarily consuming cottonseed meal and large zooplankton in roughly equal amounts ( Figure 4 44). As previously mentioned, both of these dietary scenarios were is otopically valid, due to the CSM medium plankton assemblages carbon and nitrogen isotope signature plot point being roughly midway and collinear upon the line connecting the carbon and nitrogen isotope signature plot points of the cottonseed meal and the CSM large zooplankton assemblage ( Figure 4 41). CSM large harvest fish IS and ED diet estimates may be inaccurately biased in favor of the CSM medium plankton assemblage ( Schwarcz 1991, BenDavid and Schell 2001, Phillips 2001, Fry 2006), again, due to the isotopic signature collinearity of the applied cottonseed meal, and CSM medium and large plankton assemblages ( Figures 441, 445, 453).

PAGE 215

215 However, the CSM large harvest fish ED and IS dietary estimates indicate that the large zooplankton assemblage was not their primary dietary component ( Table 4 1). Both CSM pre harvest fish ED and IS diet estimates ranked the medium plankton assemblage second in prey importance only to cottonseed meal ( Table 4 2). The CSM small harvest fish ED diet estimate ranked m edium plankton as the primary dietary source, and only the small harvest fish IS diet estimate indicated that large zooplankton were their primary prey item ( Table 4 1). Even though the ED and IS CSM fish diet estimates (except IS small harvest fish) did not indicate that the CSM large zooplankton assemblage was the primary prey group for CSM fish size classes (fry, small and large), in the opinion of this researcher, larg e zooplankton did play a major role in the CSM fish diets based upon direct observation of cottonseed meal consumption and the general observation of large zooplanktons microhabitat preference for deeper water. Of the two isotopically plausible CSM fish diets [(a) cottonseed meal and large zooplankton or (b) medium plankton], a diet of cottonseed meal and large zooplankton appeared to be a more plausible diet for CSM treatment fish (especially large harvest fish) than one composed primarily of prey derive d from the medium plankton assemblage. The INO large harvest fish and large zooplankton nitrogen isotope signature difference derived trophic distance estimate (0.80 trophic levels), was close to the theoretical value expected for a difference of one trophic level ( Table s 44, 4 10 ). Combined with the INO large harvest fish and large zooplankton carbon isotope signature difference derived trophic distance estimate [ 1.18 trophic levels ( C ) ], nitrogen and carbon derived trophic distance estimates between INO large harvest fish and large zooplankton closely corroborated each other (correct sign and magnitudes), and were both close to the values expected for a direct predator and prey relationship.

PAGE 216

216 INO large harvest fish IS and ED diet estimates ( Table 4 1) agreed well with the simple trophic distance estimates derived from carbon and nitrogen isotope signature differences between INO large harvest fish and large zooplankton ( Table s 44, 410 ). Both INO large harvest fish IS and ED diet estimates ranked large zooplankton as the primary food source ( Table 48) of INO large harvest fish by wide margins ( Table 4 1; 86.5% IS, 65% ED). Small Harvest Fish and Large Zooplankton Isotope Differences Nitrogen isotope signature difference magnitudes between small harv est fish and large zooplankton differed among treatments ( Figure 4 34). Again, nitrogen isotope signature differences between small harvest fish and their respective large zooplankton assemblages d id not differ within the two feed treatments The CSM small harvest fish and large zooplankton nitrogen isotope signature difference did not significantly differ from those of the other three treatment groups. Only the INO treatment small harvest fish and large zooplankton isotope signature difference was great er than those of the two feed treatments. Nitrogen isotope signature differences between small harvest fry and their respective large zooplankton correspond ed to trophic distances of 0.18 PRO, 0.34 UNP, 0.81 CSM and 0.43 INO (Chapter 2; 15N 3.03 per trophic level). The nitrogen isotope signature based trophic distance estimates for the two feed treatments were small, making it unlikely that feed treatment small harvest fish were heavily utilizing their respective large zooplankton assemblages. Thi s conclusion was supported by the feed treatment small harvest fish ED and IS diet estimates, which were trophically broader than their large fish counterparts. The feed treatment small harvest fish ED and IS diet estimates indicated lower consumption of feed and higher consumption of the three plankton assemblages, compared to feed treatment large harvest fish diet estimates ( Table 4 1). Although feed marginally remained

PAGE 217

217 the most important small harvest fish dietary component, according to the PRO small harvest fish ED and IS, and UNP small harvest fish ED diet estimates. The UNP small harvest fish IS diet estimate ranked medium and small plankton as their primary nutrient sources, but only slightly higher than feed and large zooplankton ( Table 4 1). Nitrogen isotope signature based trophic distance estimates between small harvest fish and large zooplankton calculated for the two fertilizer treatments indicate that CSM and INO small harvest fish likely obtained a major portion of their nutrition from large zooplankton, INO small fish to a greater degree than CSM small fish. Latent isotope signals resulting from natal diets that consisted largely of lower trophic level phytoplankton and zooplankton (e.g., rotifers, protists, ba cterial flocs) may have been present within all pond treatments; resulting in somewhat indeterminate trophic level/position estimations (including omnivory) for small size class fish among treatments, individual fish within treatments, and fractional (noninteger) trophic level/position determinations for large size class fish from the two fertilizer treatments. The CSM small harvest fish and large zooplankton assemblage nitrogen isotope signature difference derived trophic distance estimate, concurred wit h the IS CSM small fish diet estimate, which indicated that large zooplankton were the primary food source of CSM small harvest fish ( Table 4 1). However, the carbon based CSM small harvest fish and large zooplankton trophic distance estimate was of the i ncorrect sign to indicate large zooplankton were the major food source for CSM small harvest fish ( Table 4 4; 13C 0.44 trophic levels). The carbon isotope signature difference derived CSM small harvest fish and large zooplankton assemblage trophic dis tance ( 13C incorrect sign) and ED diet estimation both indicated that large zooplankton were not the primary prey of CSM small harvest fish According to the ED diet estimate, CSM small harvest fish were consuming almost equal

PAGE 218

218 proportions of medium pla nkton, large zooplankton, cottonseed meal and small plankton ( Table 41; in increasing order of importance). This seems to be the more likely scenario, given that the CSM small harvest fish ED diet estimate ranked potential prey/food sources in the same order of importance as the CSM large harvest fish ED and IS diet estimates ( Table 4 7). Additionally, when carbon and nitrogen isotope signature difference based trophic distance estimates for the CSM small harvest fish were considered simultaneously, lar ge zooplankton did not appear to be the primary dietary component ( Table 4 4; incorrect 13C sign). Again, CSM small harvest fish diet estimates (ED and IS) may have been incorrectly biased in favor of the medium plankton assemblage due to its intermediate carbon and nitrogen isotopic signatures relative to cottonseed meal and the CSM large zooplankton assemblage (Figures 4 41, 453). Only the INO small harvest fish and large zooplankton assemblage carbon and nitrogen isotope signature difference derived trophic distance estimates were of the correct signs and magnitudes to indicate that INO large zooplankton were a plausible primary prey item for INO small harvest fish ( Table 4 1). INO small harvest fish ED and IS diet estimates gave conflicting results. T he ED estimate indicated that medium plankton were the primary dietary item followed by large zooplankton and small plankton, whereas the IS estimate, indicated that large zooplankton were the primary prey by a fairly large margin (~ 67%), foll owed by small plankton (~ 25%), and little medium plankton (Tables 4 1, 4 8). Unfortunately, and similar to what may have been occurring within the CSM treatment, isotopic collinearities of the INO small, medium and large plankton assemblages (Figures 4 42, 453), may have biased the INO small harvest fish ED diet estimate in favor of the INO medium plankton assemblage. However, from the INO large harvest fish ED and IS diet estimates ( Table

PAGE 219

219 41; large zooplankton primary prey) and what is know n from opti mal foraging theory that visual predators tend to preferentially select large common prey ( Brooks 1968), it would seem much more likely that large and medium plankton would be more important nutritional sources for INO small harvest fish. However, small p lankton may be an important natal food for INO treatment swordtails and may still influence their overall isotopic composition even after swordtails switch diets as they mature; a lthough this did not appear to be indicated by the INO pre harvest fry isot ope ED and IS diet estimate results ( Table 4 2). A n additional and major issue that may have contributed to increasing INO swordtail diet estimate errors is that the INO treatment taxa isotopic signature plot (Figure 442) may have omitted a major food s ource(s) from the diet of the INO swordtails. T he food resource polygon drawn within the INO treatment 13C and 15N isotopic signature plot, does not include the INO small and large harvest fish isotopic signature plots within its boundaries (Figure 4 42 ), which indicates that a major food source may have been omitted from isotopic analyses. Small epiphytic plants and animals may have been important prey for INO fry and young juveniles, but due to time and budgetary constraints samples of this potential prey component were not collected for isotopic analysis. L arge Zooplankton and Pond Nutrient Isotopic Differences Isotopic signature differences between large zooplankton and their respective pond nutrients differed among treatments for both carbon and nitrogen (Figures 435 436). Carbon isotope signature differences between large zooplankton and pond nutrients translated to approximately one trophic level ( Table 4 4; 0.99 PRO, 0.87 UNP) for the two feed treatments, and over three trophic levels (3.19 CSM) for the CSM treatment. I t was likely that feed nutrients were being directly utilized by the large zooplankton assemblage within the two feed treatments

PAGE 220

220 and that cottonseed meal nutrients were moving through a number of trophic levels before becoming sequestered within the CSM treatment large zooplankton assemblage. Due to the absence of carbon within the applied inorganic fertilizer, the trophic distance between the INO treatment large zooplankton and small plankton assemblages (primarily phytoplankton) was estimated to be slightly over one trophic level (1.24 INO), which was somewhat less than expected (~ 2.0). The expected trophic distance was based upon a sim ple and somewhat unrealistic, postulated INO treatment food chain model: small plankton assemblage (phytoplankton) medium plankton assemblage (rotifers and other heterotrophs) large zooplankton. The shorter than estimated trophic distance between the I NO small plankton and large zooplankton assemblages, was likely due to a combination of experimental artifacts resulting from sieve size selections that resulted in plankton size assemblages that did not coincide with functional trophic groups (Chapter 5; e.g., large phytoplankton within the large zooplankton assemblage), and omnivory and predation from multiple trophic levels ( Figure 4 45). Large zooplankton and applied nutrient trophic distance estimates derived from nitrogen isotope signature differences were approximately one trophic level for the two feed (0.88 PRO, 0.77 UNP) and CSM fertilizer (0.82 CSM) treatments. Nitrogen isotope signature differences indicated that almost two trophic levels (1.66 INO) separated the INO treatment large z ooplankton and applied inorganic fertilizer, and slightly more than one trophic level (1.24 INO) separated the INO treatment large and small plankton assemblages. This indicates that the applied INO fertilizer was being assimilated by phytoplankton with a detectable degree of nitrogen isotope signature enrichment (0.44 N) impl ying that nitrogen was not limiting within the INO treatment ponds (Wada and Hattori 1978, Wada 1980, Peterson and Fry 1987,

PAGE 221

221 Pennock et al. 1996, Waser et al. 1998, 1999). The similar nitrogen isotope signatures of the applied inorganic fertilizer and the INO small plankton assemblage indicate s that fertilizer nitrogen was being assimilated by phytoplankton before being consumed and assimilated by the large zooplankton assembla ge and possible smaller (< 200 m) zooplankton intermediaries ( Table 4 4). This seems a reasonable assumption as the applied inorganic fertilizer had to be assimilated by the ponds autotrophic community before entering the food web in the INO treatment p onds. Both the ED and IS diet estimations indicated that commercial feeds were the primary nutrient sources for large zooplankton assemblages within both feed treatments ( Table 4 9). Within the CSM treatments, the IS method ranked cottonseed meal as th e primary nutrient, whereas the ED method ranked medium plankton followed closely by cottonseed meal as the primary nutrients of the large zooplankton assemblage. INO treatment large zooplankton diets were estimated to consist primarily of medium plankton followed closely by small plankton using the ED method. S mall plankton was identified to be the primary dietary component of INO large zooplankton using the binomial linear algebra (Table 4 9; non IsoSource) dietary estimation method ( Ben David a nd Schell 2001, Fry 2006). The role of autochthonous and nonhuman mediated allochthonous carbon in phytoplankton nutrition was unknown due to the lack of baseline (no applied nutrient control treatment) inorganic carbon pool isotope signature information for the different pond treatments Due to the absence of carbon within the INO treatments basal nutrient the INO small plankton assemblage (phytoplankton) w as analyzed as the nutrient base of the INO treatment food web ( Zohary et al. 1994, Yoshii et a l. 1999, Hall 2004, GregoryEaves 2007)

PAGE 222

222 Unfortunately, plankton assemblages were artificially separated on the basis of size rather than as discrete functional trophic groups (1 producer, 1 and 2 consumers, etc .). The se artificial and somewhat arbitra ry plankton size divisions and t rophic interactions (trophic positions/feeding habits e.g., omnivory) likely contributed to trophic positions that were not as clearly defined as proposed within the highly oversimplified INO treatment food chain: small pla nkton (primary producers ) medium plankton (rotifers other heterotrophs) large zooplankton fish This also may have contributed to the noninteger carbon isotope enrichment magnitudes observed among the postulated trophic guilds ( Table s 44, 4 10 ). Small Plankton and Nutrient Isotope Differences From a limited, qualitative microscopic examination of the preserved small plankton assemblage (1 32 m) samples, small plankton assemblages were found to consist primarily of small green algae. This con clusion also was supported by the vivid green color of the small plankton filtrate (refrigerated/unpreserved) when present upon the white Whatman GF/C filter disks in comparison to the brown to bright red color of the medium and large plankton filtrates. Unlike the medium plankton assemblage, which was largely a mixture of phytoplankton (e.g., green algae, diatoms) and zooplankton (e.g., rotifers, nauplii), the small plankton assemblage appeared to largely coincide with an actual functional trophic group (1 producers). Because the small plankton assemblage coincided with the trophic group that converts solar energy and inorganic carbon (DIC) and nitrogen (DIN) into energy containing organic compounds, a more thorough examination of this group is warranted. Carbon isotope signature differences between small plankton and pond nutrients differed only between the PRO and CSM treatments ( Figure 437). The large differences were likely due to phytoplankton within these assemblages utilizing autochthonous and r emineralized applied nutrient DIC rather than direct utilization of applied feeds, as would have occurred for

PAGE 223

223 heterotrophic organisms. The large carbon isotope signature differences, between the two feed and CSM small plankton assemblages and their respec tive applied nutrients were equivalent to 0.86 (PRO), 0.64 (UNP), and 0.55 (CSM) photosynthetic inorganic carbon fixation isotopic signature enrichment levels ( Table 4 4), assuming a primary producer isotopic enrichment rate of 19 C typical for C3 aquatic plants when [CO2] is high ( Peterson and Fry 1987, OLeary 1988). These values seem plausible given what is currently know n of aquatic photosynthetic carbon fixation dynamics ( Finlay and Kendall 2007) A lack of direct trophic coupling between the small plankton assemblages and applied feed nutrients w as expected within the PRO and UNP pond treatments, as these plankton assemblages consisted primarily of phytoplankton that are likely unable to directly utilize nutrients (organic compounds: starche s, lipids, proteins) designed for animal consumption ( Lochmann and Phillips 1996, Lochmann et al. 2001, Grey et al. 2004 ) Within the CSM treatment, the limited ability of phytoplankton to directly utilize applied cottonseed meal also was expected, as the nutrients are primarily in the form of starches and proteins that are largely unavailable to autotrophs. Cottonseed meal nutrients were believed to be unavailable to primary producers prior to their remineralization to DIC and DIN via bacterial and f ungal decomposition, and following ingestion and digestion by heterotrophic organisms ( Landau 1992, Teicher Coddington et al. 1997). Although nonINO treatment primary producers were unable to directly utilize feed and cottonseed meal nutrients directly, b ased on the carbon isotope signature differences between nonINO treatment small plankton assemblages and applied nutrients it seems that primary producers were able to utilize remineralized DIC derived from recycled applied nutrients in addition to isotopically unmeasured autochthonous and non human mediated allochthonous DIC

PAGE 224

224 and atmospheric carbon sources. Because small plankton assemblages were composed primarily of phytoplankton autotrophs, they appear to have been isotopically influenced by applied nutrients, even when these nutrients were intended for large heterotrophs. By altering the carbon isotope signatures of the inorganic carbon pools utilized by autotrophs, applied nutrients within the nonINO treatments may have influenced small plankton assemblage carbon isotope signatures. This may explain why feed treatment small plankton carbon isotope signatures did not differ from each other, but differed from those of the two fertilizer treatments, and why each of the two fertilizer treatments small plankton carbon isotope signatures differed from those of the other three treatments ( Figure 4 28). If nonanthropogenic DIC was the predominant form of DIC available to pond primary producers, then no small plankton assemblage carbon isotope signature di fferences would be expected among treatments; additionally, all small plankton assemblage carbon isotope signature values would be expected to have been near that measured for the INO treatment (Figure 428), due to the absence of applied carbon within the INO treatment ponds. After several weeks of twice daily feed and once daily cottonseed meal additions, applied nutrient carbon had apparently been recycled by repeated mineralization and photosynthetic carbon fixation of applied nutrients within the PRO UNP and CSM treatments. Inorganic carbon originally derived from applied nutrients was utilized by pond primary producers, which was reflected in the small plankton assemblage carbon isotope signature differences observed among treatments ( Figure 428). The range of carbon isotope signature values observed among the three nonINO treatment small plankton assemblages ( 28.64 to 26.53 C) were in the upper ranges reported for phytoplankton assemblages in the literature ( 42 to 26 C Deines 1 980; 28 to 27 C

PAGE 225

225 Yoshioka et al. 1994), but less enriched than other studies ( 23.2 to 17.9 C Zohary et al. 1994). The INO small plankton assemblage carbon isotope signature w as higher than that of the other three treatments ( Figure 428 ) This was probably due to autochthonous (mineralization of pond sediments, decaying organic matter, organism respiration, etc .) and allochthonous inorganic carbon from nonhuman mediated sources (groundwater, atmospheric, etc .) which resulted in a labile i norganic carbon pool with a composite isotopic signature higher than those of the nonINO treatment ponds. Isotopic enrichment processes for inorganic carbon uptake by aquatic autotrophs are strongly biased against the heavier 13C due to diffusion rate di fferences and carbon fixation enzyme kinetics (RuBisCo; ~ 20 C, Peterson and Fry 1987), unless DIC availability is extremely low relative to photosynthetic demand. Typically, algal fractionation values are heavily influenced by the isotopic signature of the available DIC (primarily aqueous CO2 and HCO3 -), and the relative availability of these two chemical species ( Rounick and Winterbourn 1986, Peterson and Fry 1987, Fogel et al. 1992 ). Generally, when aqueous [CO2] is high and/or phytoplank ton growth rates are low, isotopic carbon fractionation between the DIC pool and the algae is highest (Finlay and Kendall 2007). Isotopic discrimination during the fixation of carbon dioxide and carbonate during photosynthesis (i.e. RuBisCo, C3, PEP carb oxylase, C4 metabolic differences) tends to mask or obliterate the isotopic signature identity of the carbon pool being utilized if the carbon pool is sufficiently large relative to carbon demand ( Smith and Walker 1980, Yoshii et al. 1999, Finlay and Kendall 2007). Additionally, seasonal changes as well as latitudinal gradients in the atmospheric carbon dioxide pool carbon isotope signature can occur due to terrestrial photosynthetic carbon demands ( Estep and Vigg 1985, Fogel et al. 1992,

PAGE 226

226 Cifuentes et al. 1998, Takahashi et al. 1990, Zohary 1994, Grey et al. 2001, Fry 2006) which can add to carbon isotope signature variability and carbon pool identification uncertainty. Unfortunately, dissolved inorganic carbon pool isotope signatures of the four treatm ents were not measured during the study due to logistical, time and budgetary constraints. Additionally, due to the lack of baseline carbon isotope signature s for the DIC pool and pond taxa (phytoplankton, zooplankton, fish) in the absence of anthropogeni cally applied nutrients (no applied nutrient control treatment), the causes of carbon isotope differences among the taxa in the various treatments are partially conjecture, especially in comparisons involving primary producers (small plankton assemblage). A remote possibility remains that the carbon isotope signatures of the DIC pools naturally differed among ponds, and that these differences resulted in observable isotopic differences among the respective small phytoplankton assemblages, and taxa that tr ophically depended upon them. However, due to the random assignment of experimental ponds among treatments, it would be unlikely that these differences would occur to a greater degree within a given treatment. Small Plankton Assemblage Applied Nutrient Nitrogen Isotope Differences The INO small plankton and inorganic fertilizer nitrogen isotope signature difference ( INO 1.26 N) was higher than that of the other three treatments but was still much lower than the 3.03 N nitrogen enrichment magnitude observed in the indoor trial (Chapter 2), and the mean enrichment magnitude levels reported in the literature for heterotroph predation for one trophic level (~ 3.4 N ; Peterson and Fry 1987, Kling and Fry 1992, Gu et al. 1996, Post 2002a ).

PAGE 227

227 The INO small plankton assemblage nitrogen isotope signature was lower than those of the other three treatments, which did not differ from one another ( Figure 4 28). Again, this was likely due to the lower nitrogen isotope signature of the INO treatment food web/chains presumed basal nutrient ( Figure 45; inorganic fertilizer 0.08 N) and the small initial N difference among the three nonINO applied nutrients ( Figure 4 5 cottonseed meal and feeds N < 1.0 ); assuming that non INO treatment small plankton assemblages heavily utilized remineralized feed and cottonseed meal nitrogen ( Figure 4 38). Small plankton assemblage N signature differences among treatments indicated that small plankton assemblages were significantly influenced by the addition of applied nutrients, as were their respective food webs (Figures 4 11 412, 418, 423). However, whether (1) only the nonINO treatment phytoplankton assemblage nitrogen signatures were influenced by applied feed and cottonseed meal nutrient additions, or (2) only the INO phytoplankton assemblage nitrogen signature was altered by the application of the applied nutrient (i.e., inorganic fertilizer), or (3) all four treatments small plankton assemblages were isotopically influenced by their re spective applied nutrients, was unknown due to the lack of baseline small plankton assemblage isotopic signature data in the absence of applied nutrients (no applied nutrient control treatment), and the lack of aqueous inorganic (DIN: NH4 +, NO2 -, NO3 -) a nd organic nitrogen chemical species (urea, labile free amino acids) isotope data within the INO and non INO treatment ponds. Regardless, the presence of nitrogen isotope signature differences among the four small plankton assemblages strongly suggests that small plankton assemblages were utilizing inorganic nitrogen pools (DIN) with differing isotopic signatures (Figure 4 28). L ower nitrogen isotope enrichment magnitudes between small plankton assemblages and their applied nutrients were not surprising given that small plankton assemblages consisted

PAGE 228

228 primarily of autotrophs, which typically exhibit low nitrogen isotope enrichment rates (01 N) relative to their inorganic nitrogen source s ( Finlay and Kendall 2007). Because nitrogen is often a limiti ng nutrient for plants within an ecosystem, plant nitrogen uptake typically result s in little to no nitrogen isotope enrichment Large nitrogen isotope enrichment magnitudes between photosynthetic autotrophs and their DIN pools tend to occur when availabl e nitrogen pools are extremely large and nitrogen uptake is low, usually resulting from low algal growth and reproductive rates ( Cifuentes et al. 1988, Needoba 2004, Finlay and Kendall 2007). Because phytoplankton nitrogen uptake mechanisms differ from nitrogen uptake processes occurring during prey/particle ingestion by animals ( Montoya and McCarthy 1995, Cowe y and Walton 1989), phytoplankton are not able to directly utilize applied feed nutrients (possible exception of leaching) formulated for carnivorous or omnivorous fish. Due to feed nitrogen being primarily in the form of proteins, applied feed nitrogen was largely unavailable to primary producers within the feed treatment ponds ( Tyler et al. 2003). Phytoplankton and other primary producers obtain labile nitrogen (urea, NH4 +, NO3 -, NO2 -, N2 [ cyanobacteria, legumes via mutualistic bacteria]) via diffusion across cell membranes (Fogg and Thake 1987, OLeary 1988, Valiela 199 5) U ptake kinetics of these different nitrogenous chemical species, and their typically low (limiting nutrient) concentrations often result in fairly low nitrogen isotope signature enrichment magnitudes relative to those observed between heterotrophs and their plant or animal prey ( Wada 1980, Wada and Hattori 1978, DeNir o and Epstein 1981, Vander Zanden et al. 1997, Waser et al. 1998). Additional factors known to influence phytoplankton nitrogen isotope enrichment magnitudes relative to their primary nitrogen sources are: inorganic nitrogen pool characteristics (molecula r size/chemical species, ionic charge, isotopic signature, concentration/availability as mentioned previously), phytoplankton characteristics (species, cell

PAGE 229

229 size, somatic growth rate, senescence, nutritional history, population growth rate, etc), temperature (via effects on photosynthesis and respiration), photoperiod, light levels (via effects on photosynthesis e.g., photoinhibition, algal self shading), water chemistry (turbidity, pH, redox), ecosystem trophic state, among others (Fogg and Thake 1987, Vander Zanden and Rasmussen 1997, Vander Zanden et al. 1997, 1999). Although feed treatment small phytoplankton assemblages and applied feeds were not trophically coupled via direct consumption, they seemed to be isotopically coupled, resulting in small isotopic signature differences (~ 0.60 N) between feed treatment small plankton assemblages and the heavily recycled and remineralized feed nitrogen they were presumably utilizing (Figure 438). Phytoplankton within the small plankton assemblages of the two feed treatments were utilizing DIC and DIN derived from applied feeds, possibly due to decomposition of uneaten feed and from respiration, egestion and excretion of pond taxa that had consumed applied feeds. The nitrogen isotope signature diffe rence between the CSM small plankton assemblage and applied cottonseed meal also was small (~ 0.55 N), and much lower than that observed within the INO treatment (1.26 N ). These small nitrogen isotope signature differences indicated probable t rophic and isotopic coupling. The most plausible explanation for the se small differences ( Figure 438 ) is that processes believed to be occurring within ponds for applied nutrient carbon, were also occurring for nitrogen. F eed and cottonseed meal nitrogen were being remineralized and recycled within the nonINO pond food webs into DIN that was then available for assimilation by phytoplankton. DIN may have be en a limiting nutrient for phytoplankton within these ponds, resulting in low nitrogen isotope si gnature enrichment between the nonINO treatment small plankton assemblages and their respective inorganic

PAGE 230

230 nitrogen pools. Proteins within applied feeds and cottonseed meal were eventually metabolized to ammonium (NH4 +) by pond organisms and transformed by bacteria in to nitrite and nitrate. Subsequently, primary producers within the small plankton assemblage may have assimilated these nutrients for amino acid synthesis, protein synthesis, growth, and reproduction ( Bilb y et al. 1996, Pennock et al. 1996, Needoba et al. 2004 ). Even largely unpalatable blue green algae ha s been shown to influence the nitrogen isotope signature of higher trophic levels by fixing atmospheric nitrogen before eventually under going remineraliz ation and recycling into more palatable green algae, which then entered the food web ( Estep and Vigg 1985, Valiela 1995, Piola et al. 2008). The larger nitrogen isotope signature difference between the INO small phytoplankton and inorganic fertilizer treatment, relative to those of the two feed treatments (Figure 4 38 ), may have been due to higher DIN concentrations within the INO treatment ponds ( Figure 446). H igher algal nitrogen turnover rates due to greater nitrogen availability (luxury uptake ) w ithin the INO ponds, may have produced greater N signature enrichment magnitudes within INO pond algal cells relative to their nonINO primary producer counterparts ( Fogel and Cifuentes 1993, Pennock et al. 1996, Needoba et al. 2004 ). However, the much lower nitrogen isotope signature difference (and N based trophic distance estimate) between the CSM small phytoplankton and applied cottonseed meal fertilizer (~ 0.55 ) compared to that occurring within the INO treatment (Figure 438; Table 44 ) was somewhat puzzling given that nitrogen availability within the two fertilizer treatments did not differ ( Figure 446 ). A possible explanation for the lower nitrogen isotope signature difference between the CSM small plankton and applied cottonseed meal in the presence of high DIN, was that higher algal standing stocks within the CSM treatment ponds ( Figure 4 49;

PAGE 231

231 chlorophyll [a]), may have reduced nitrogen availability due to resource competition among phytoplankton and other aquatic and semi aquat ic primary producers. P er capita phytoplankton nitrogen availability (DIN/algal biomass) was lower within the CSM ponds relative to the INO ponds which may have reduced the nitrogen isotope signature enrichment differences between phytoplankton and their DIN pool due to greater competition among CSM pond phytoplankton for available nitrogen and lower overall nitrogen turnover rates within the algal cells ( Fogg and Thake 1987, Fogel and Cifuentes 1993, Pennock et al. 1996, Needoba et al. 2004). Interestin gly, the nitrogen isotope signature difference derived trophic distance estimate (0.55 trophic levels) between CSM small plankton and the applied cottonseed meal was identical to that calculated from carbon isotope signature differences ( Table s 44, 410; Equations 4.24.3) This close agreement between two trophic distance estimates derived from the isotopic differences of two elements (C, N), implies that the theoretically predicted carbon and nitrogen isotopic enrichment rates ( 19 C, ~ 0 N ) for primary producers and their inorganic carbon and nitrogen pools were very close to the actual isotopic enrichment rates occurring within the cottonseed meal treatment ponds (OLeary 1988) As previously stated, isotopic enrichment magnitudes for inorganic nitrogen uptake by aquatic primary producers are typically low, on the order of 01 N depending primarily upon inorganic nitrogen availability and phytoplankton nitrogen demands ( Pennock et al. 1996, Needoba et al. 2004, Finlay and Kenda ll 2007 ). This was supported by the low nitrogen isotope signature difference between the INO small plankton assemblage and applied inorganic nitrogen fertilizer (Figure 438; 1.26 N), which was the t reatment in which the greatest confidence was present that the applied nutrient was a major if not primary phytoplankton nitrogen source. Recycl e d nitrogen, from applied nutrients within the nonINO treatments w as the likely primary

PAGE 232

232 nitrogen source for autotrophs as indicated by the small nitrogen isotope differences measured between the nonINO treatment small plankton assemblages and their respective applied nutrients ( Figure 438; 0.550.58 N ). However, nonanthropogenic all ochthonous nitrogen input magnitudes were not measured nor 1 producer N signatures in the absence of anthropogenic nitrogen input, therefore it is not certain that applied nutrients were the primary sources of nitrogen within the ponds. H owever unlikely, the presence of a major nonanthropogenic nitrogen source(s) could not be eliminated as baseline fish production, DIN pool size and taxa isotopic signatures in the absence of applied nutrients could not be measured due to the lack of a no applied nu trient control treatment. Because baseline fish production in the absence of nutrient application could not be compared to fish production resulting from anthropogenic nutrient addition, it is not entirely certain that applied nutrients were responsible for observed fish production or observed differences in fish production among the four treatments. In the absence of anthropogenic nutrient input, it was assum ed that nitrogen availability is the primary limiting factor in phytoplankton production, and ul timately fish production within aquaculture ponds in the area This is a reasonable assumption, given that phosphorous, which is the usual limiting nutrient for freshwater primary production, is generally not limiting within freshwater ponds in the study a rea; due to large phosphate deposits within the limestone Karst geology of central and southern Florida (Lakewatch #103). Due to the lack of a no applied nutrient control treatment, and difficulty of measur ing DIN fluxes and DIN pool isotopic signatures occurring within the different treatment ponds, their potential contribution to pond taxa N signatures and taxa production are unknown. Because atmospheric nitrogen and the applied inorganic fertilizer (INO) nitrogen isotope signatures were

PAGE 233

233 nearly identical (~ 0.00 N ), it was not possible to eliminate atmospheric nitrogen as the primary nitrogen source responsible for biological production within the INO treatment ponds. However, the lack of a large cya nobacteria /blue green algae ( which can fix atmospheric nitrogen into plant biomass ) bloom s within any of the 24 treatment ponds ( pers. obs.) made this scenario unlikely. Algal standing stocks were estimated using chlorophyll [a] concentration data measur ed weekly within each pond ( Figures 4 47 450). Time averaged chlorophyll [a] concentrations only differed for the CSM treatment, which was significantly higher than those within the other three treatments (Figure 4 51). This indicated that cottonseed meal ponds had higher phytoplankton standing stocks ( Florida Lakewatch 2001), but due to the lack of primary consumer grazing rate information or direct measurements of primary production rates, it remains unknown whether primary production rates were greater within the CSM treatment ponds relative to the other treatments. Primary production rate information would have been extremely useful as evidence of whether higher algal production was responsible for the higher, but not significantly greater total fish production rate mean observed for the CSM treatment relative to the INO treatment (Figure 39). Time averaged total ammonia nitrogen concentrations were significantly higher in the two fertilizer treatments than within the two feed treatments (Figure 446). This was expected and may have been responsible for the higher phytoplankton standing stock observed within the CSM ponds, although a higher algal standing stock was not observed within the INO ponds. This may have been due to higher, unmeasured pri mary consumer grazing rates within the INO ponds and/or due to the more labile and possibly volatile nature of inorganic fertilizer nutrients being unable to maintain high phytoplankton growth rates and standing stocks relative to the

PAGE 234

234 slow nutrient release nature of the cottonseed meal fertilizer ( Kastner and Boyd 1996) However, higher zooplankton grazing rates within the INO ponds did not appear to be occurring, as large zooplankton standing stocks did not differ among ponds for the four treatments ( Chap ter 5; Figures 5 6, 5.12). Although higher predation upon zooplankton by swordtails within the two fertilizer treatments may have kept zooplankton standing stocks low (trophic cascades), masking the presence of greater zooplankton grazing pressures upon algae within the CSM and INO ponds ( Mallin and Paerl 1994, Hansson and Tranvik 1997, Tranvik and Hansson 1997, Cohen et al. 2002). Again, although there was weekly algal standing stock (Figure 4 47 4.51) and treatment ammonia nitrogen levels (Figure 4 46) information phytoplankton production rate and zooplankton grazing rate information was lacking Due to this lack of information, the quantitative contribution primary producers may have made to food webs of the two fertilizer treatments, as well as any supplemental contributions that phytoplankton may have made to the food webs or to fish production within the ponds of the two commercial feed treatments remains unknown ( Diana et al. 1991, KnudHansen et al. 1993). I n conjunction with an absence of DIN pool N signature information a lack of primary production rate information, means that t he effects of primary production rates upon the magnitude of primary producer nitrogen isotope signature fractionation within the treatments is also unknown. A s higher phytoplankton growth rates typically result in lower N signature fractionation magnitudes between 1 producers and the exploited DIN pool ( Evans et al. 1997, Finlay and Kendall 2007). The INO small plankton assemblage (primarily phytoplankton) was slightly more enriched (1.26 N) than the applied inorganic fertilizer, at an enrichment rate similar to that found

PAGE 235

235 for marine phytoplankton (1.015 N ; Wada 1980). Due to the isotopic similarity of the applied inorganic nitrogen fertilizer and atmospheric nitrogen (~ 0.00 N ), it was not unexpected that nitrogen isotope signature enrichment between the INO small plankton assemblage and applied inorganic nitrogen fertilizer was similar to that measured between atmospheric nitrogen and ni trogen fixing cyanobacteria mats (1.51.7 N ) in a coastal lake in New South Wales, Australia (Piola et al. 2008 ). If similar nitrogen isotope enrichment rates between phytoplankton and their respective inorganic nitrogen pools occurred within the non INO treatments (~ 1.26 N enrichment rate), this m ight have allowed back calculat ion of the nitrogen isotope signa tures of the inorganic nitrogen pools (DIN) utilized by the nonINO treatment small plankton assemblages. However, lower inorganic nitrogen concentrations within the two feed treatment ponds ( Figure 4 46) may have reduced nitrogen isotope fractionation ra tes [ N N ( INO enrichment rate ] between inorganic nitrogen pools, and primary producers within the feed treatment ponds ( Figure 4 38). Making back calculation of the N signature of the nonINO inorganic nitrogen pools (DIN), unreliable ( M ichener and Lajtha 2077 ). Again the presence of a no applied nutrient control pond treatment would have been useful in that it would have provided baseline isotopic profile information for algal, zooplankton and fish taxa, as well as baseline production rate information for these taxa in the absence of applied nutrients. Baseline no applied nutrient small plankton/phytoplankton isotope signatures would have been useful in back calculating the isotopic signature profiles of nonanthropogenic DIC and DI N pools being utilized by pond autotrophs at the base of the pond food webs/chain. With an additional expenditure of time, money and labor, actual pond DIC and DIN pool isotopic signature information could have been measured with in no applied nutrient

PAGE 236

236 c ontrol ponds. This information could have been used to verify the DIC and DIN pool isotopic signature estimates derived from back calculation of no applied nutrient pond taxa isotope signatures. This would have allowed c alculat ion of the contribution, if any, made by autochthonous and nonhuman mediated allochthonous (groundwater DIN, wind blow dust, leaf litter, insects, etc .) nutrients, as well as that of applied nutrients to primary production within the respective treatment ponds a llowing rudiment ary quantitative and improved qualitative estimates of primary production originating from applied nutrient contributions, and therefore, rough estimates of potential live food contributions to fish production among the four treatments. Making it easier to evaluate any increase in fish biomass and economic benefits (if any) provided by the different applied nutrients, and the general efficacy of their use.

PAGE 237

237 Figure 41. Average l arge harvest fish (swordtail) biomass ( 95 % CI) as a function of nitrogen isotope signature differences (semi log plot) between large harvest fish and applied nutrient treatments; differing letters denote significant differences between groups (P < 0.05, Tukeys multiple comparison test) Sample sizes: PRO (5), UNP (6), CSM (6), and INO (6). y = 921. 56x + 10342 R = 0 8537 100 1000 10000 100000 2 0 3 0 4 0 5 0 6 0 7 0 8 0Total Fish Biomass (g) N difference between large harvest fish and applied nutrient fish biomass Linear (fish biomass) PRO UNP CSM INO a a b b

PAGE 238

238 Figure 42. Integrated depth pond water column sampler. one-way flapper valve opening into sampler 3.05-meter length schedule 40 PVC (1.25-inch nominal) H2O H2OFigure 4.2: Integrated pond water sampler pond water line flapper valve detail (neoprene rubber)

PAGE 239

239 Figure 43a d. Clockwise from top left: (a ) freeze dried large zooplankton size assemblage on Whatman GF/C pre combusted glass filter disc (47 mm dia.), (b) filter disc attached to small vibrating motor while particles are brushed into funnel, (c ) open tin foil sample capsule containing plankton sample, ( d) tin capsule with sample prior to being closed, folded and placed into analysis tray. a b c d

PAGE 240

240 Figure 44. Mean carbon isotope signatures ( 13C 95 % CI) of three pond nutrient treatments [processed feed (PRO), unprocessed feed (UNP), and cottonseed meal fertilizer (CSM)] n = 3 (PRO), 3 (UNP), and 2 (CSM); unshared letters denote statistical differences between treatments (P < 0.05, Tukeys multiple comparison test ). 21 22 23 24 25 26 27 28 PRO UNP CSM C(abs) a a b

PAGE 241

241 Figure 45. Mean nitrogen isotope signature s ( 15N 95 % CI) of applied pond nutrient treatments [processed feed (PR O), unprocessed feed, cottonseed meal (CSM), and inorganic fertilizer (INO)] n = 3 (PRO), 3 (UNP), 2 (CSM), and 2 (INO); unshared letters denote statistical differences between treatments (P < 0.05, Tukeys multiple comparison test ). 2 1 0 1 2 3 4 5 6 7 PRO UNP CSM INO N c c b a

PAGE 242

242 Figure 46. Average p re harvest pond swordtail fry carbon (absolute value) and nitrogen isotope signatures ( 13C, 15N) for three weekly sampling dates among four nutrient treatments [processed (PRO) and unprocessed (UNP) feed, cottonseed meal (CSM) and inorganic (INO ) fertilizer]; n = 6 replicate ponds per treatment 0 5 10 15 20 25 30 day 7 day 14 day 21 13C(abs) and N PRO d 13 C PRO d 15 N UNPRO d 13 C UNPRO d 15N CSM d 13 C CSM d 15 N INO d 13C INO d 15 N 15N 13C

PAGE 243

243 Figure 47. Average p rocessed feed (PRO) treatment pre harvest swordtail fry carbon and nitrogen isotope signatures [ 13C (abs. value), 15N 95 % CI ] for three weekly sampling dates ; n = 6 replicate ponds per treatment, unshared letters denote statistical differences within element, C or N (P < 0.05, Tukeys multiple comparison test ). 0 5 10 15 20 25 day 7 day 14 day 21 13C(abs) and N PRO d 13 C PRO d 15 N a b b a a a

PAGE 244

244 Figure 48. Average u nprocessed feed (UNP) pond treatment pre harvest swordtail fry carbon and nitrogen i sotope signatures [ 13C (abs. value), 15N 95 % CI ] for three weekly sampling dates ; n = 6 replicate ponds per treatment, unshared letters denote statistical differences within elements, C or N, (P < 0.05, Tukeys multiple comparison test ). 0 5 10 15 20 25 30 day 7 day 14 day 21 13C(abs) and N UNP d 13 C UNP d 15 N a ab b a a a

PAGE 245

245 Figure 49. Average cottonseed meal (CSM) pond treatment pre harvest swordtail fry carbon and nitrogen isotope signatures [ 13C (abs. value), 15N 95 % CI ] for three weekly sampling dates ; n = 6 replicate ponds per treatment, unshared letters denote statistica l differences within elements, C or N (P < 0.05, Tukeys multiple comparison test ). 0 5 10 15 20 25 30 day 7 day 14 day 21 13C(abs) and N CSM d 13 C CSM d 15 N a b b a b ab

PAGE 246

246 Figure 410. Average i norganic fertilizer (INO) treatment pre harvest swordtail fry carbon and nitrogen isotope signatures [ 13C (abs. value), 15N 95 % CI ] for three weekly sampling dates ; n = 6 replicate ponds per treatment, unshared letters denote statistical differences within elements, C or N (P < 0.05, Tukeys multiple comparison test ). 0 5 10 15 20 25 day 7 day 14 day 21 13C(abs) and N INO d 13C INO d 15 N a a a a b ab

PAGE 247

247 Figure 411. Time a verage d pond harvest large swordtail ( > 31 mm SL) carbon and nitrogen isotope signatures [ 13C (abs. value) 15N 95 % CI ] among four nutrient treatments numbers in parentheses denote sample size ; unshared letters denote statistical differences within elements, C or N, (P < 0.05, Tukeys multiple comparison test ). 0 5 10 15 20 25 30 PRO ( 5 ) UNP ( 5 ) CSM ( 6 ) INO ( 6 ) C(abs) and N mean dN mean dC b a a a a a a a

PAGE 248

248 F igure 412. Average p ond harvest small fish (SL 31 mm) carbon and nitrogen isotope signatures [ 13C (abs. valve) 15N 95 % CI ] among four pond nutrient t reatments numbers in parentheses denote sample size; unshared letters denote statistical differences (P < 0.05, Tukeys multiple comparison test). 0 5 10 15 20 25 30 PRO ( 5 ) UNP ( 5 ) CSM ( 6 ) INO ( 6 ) C(abs) and N mean dN mean dC a a a b a b b b

PAGE 249

249 Figure 413. Pond trial large zooplankton (> 200 m) carbon [ 13C (abs. value ) ] and nitrogen ( 15N) isotope signatures for 11 weekly sampling periods among four pond nutrient treatments n = 6 ponds per treatment 0 5 10 15 20 25 30 35 d13C (abs) and 5N PRO d 13C PRO d 15 N UNP d 13 C UNP d 15 N CSM d 13 C CSM d 15N INO d 13C INO d 15 N

PAGE 250

250 Figure 414. P rocessed feed (PRO) treatment large zooplankton assemblage carbon and nitrogen isotope signatures [ 13C (abs. value) N 95 % CI ] among 11 weekly sampling periods n = 6 ponds per treatment ; unshared letters denote statistical differences within elements, C or N, (P < 0.05, Bonferroni post test ), too few replicates for statistical analysis 0 5 10 15 20 25 30 35 13C (abs) and 15N PRO d 13 C PRO d 15N ab ab ab ab a ab b a ab a ab a b a ab ab ab ab a *

PAGE 251

251 Figure 415. U nprocessed feed (UNP) treatment large zooplankton assem blage carbon and nitrogen isotope signatures [ C (abs. value) 15N 95 % CI ] among 11 weekly sampling periods n = 6 ponds per treatment ; unshared letters denote statistical differences within elements, C or N, (P < 0.05, Bonferroni post test ) too few replicates for statistical analysis. 0 5 10 15 20 25 30 35 13C (abs) and 15N UNP d 13 C UNP d 15 N a ab ab ab ab ab ab b ab ab ab a a a a a a a a *

PAGE 252

252 Figure 416. C ottonseed meal (CSM) treatment large zooplankton (> 200 m) assemblage carbon and nitrogen isotope signatures [ 13C (abs. value) 15N 95 % CI ] among 11 weekly sampling periods n = 6 ponds per treatment ; unshared letters denote statistical differences within elements, C or N, (P < 0.05, Bonferroni post test ), too few replicates for analysis. 0 5 10 15 20 25 30 35 13C (abs) and N CSM d 13C CSM d 15 N a a a a a a a b a a a a b a a a a b a

PAGE 253

253 Figure 417. I norganic fertilizer treatment large zooplankton (> 200 m) assemblage carbon and nitrogen isotope signatures [ 13C (abs. value) 15N 95 % CI ] among 11 weekly sampling periods n = 6 ponds per treatment ; unshared letters denote statistical differences within elements, C or N, (P < 0.05, Bonferroni post test ), too few replicates for statistical analysis. 0 5 10 15 20 25 30 35 13C (abs) and N INO d 13C INO d 15 N b b b ab a ab ab ab ab ab ab a a bc e ab c ab ab

PAGE 254

2 54 Figure 418. T ime averaged large zooplankton (> 200 m) carbon and nitrogen isotope signatures [ 13C (abs olute value) 15N 95 % CI ] among four pond nutrient treatments n = 6 ponds per treatment ; unshared letters denote statistical differences within elements, C or N, (P < 0.05, Tukeys multiple comparison test ). 0 5 10 15 20 25 30 PRO UNP CSM INO 13C(abs) & 15N dC dN b bc c a c c b a

PAGE 255

255 Figure 419. P rocessed feed (PRO) medium zooplankton (32 200 m) assemblage carbon and nitrogen isotope signatures [ 13C (abs olute value) 15N 95 % CI ] among 12 weekly sampling periods n = 6 ponds per treatment ; unshared letters denote statistical differences within elements, C or N, (P < 0.05, Bonferroni post test ). 0 5 10 15 20 25 30 13C(abs) and 15N dC dN a bcd abcd abc ab abc bcd bcd cd bcd bcd d a b ab ab ab b ab ab ab ab ab ab

PAGE 256

256 Figure 420. U nprocessed (UNP) feed medium plankton assemblage carbon and nitrogen isotope signatures [ 13C (abs olute value) 15N 95 % CI ] among 12 weekly sampling periods n = 6 ponds per treatment ; unshared letters denote statistical differences within elements, C or N, (P < 0.05, Bonferroni post test ). 0 5 10 15 20 25 30 13C(abs) and 15N bc b bc bc cd bc bc bcd bcd d a bcd a ab ab ab ab ab ab ab ab b ab ab

PAGE 257

257 Figure 421. C ottonseed meal (CSM) medium plankton assemblage carbon and nitrogen isotope signatures [ 13C (abs olute value) 15N 95% CI ] among 12 weekly sampling periods n = 6 ponds per treatment ; unshared letters denote statistical differences within elements, C or N, (P < 0.05, Bonferroni post test ). 0 5 10 15 20 25 30 35 13C(abs) and 15N dC dN a bcd abc ab abc bc cde de e bc abc bc a a a a a a a a a a a a

PAGE 258

258 Figure 422. I norganic fertilizer (INO) medium plankton (32 200 m) assemblage carbon and nitrogen isotope signatures [ 13C (abs olute value) 15N 95 % CI ] among 12 weekly sampling periods n = 6 ponds per treatment ; unshared letters denote statistical differences within elements, C or N, (P < 0.05, Bonferroni post test ). 0 5 10 15 20 25 30 13C(abs) and 15N dC dN abc bc c abc abc abc abc abc ab ab a abc a a ab ab ab a a ab ab ab ab ab

PAGE 259

259 Figure 423. T ime averaged medium plankton (32 200 m) carbon and nitrogen isotope signatures [ 13C (abs olute value) 15N 95 % CI ] among 12 weekly sampling periods n = 6 ponds per treatment ; unshared letters denote statistical differences within elements, C or N, (P < 0.05, Tukeys multiple comparison test ). 0 5 10 15 20 25 30 PRO UNP CSM INO C (abs) & N dC avg dN avg a b c bc a b b b

PAGE 260

260 Figure 424. P rocessed feed (PRO) small plankton (1 32 m) assemblage carbon and nitrogen isotope signatures [ 13C (absolute value) 15N 95 % CI ] among 12 weekly sampling periods n = 6 ponds per treatment ; unshared letters denote statistical diffe rences within elements, C or N, (P < 0.05, Bonferroni post test ). 0 5 10 15 20 25 30 35 13C(abs) and 15N dC dN bcde cdefg b bcd bcdef efg g fg bcde bc bcdef a a a a a a a a a a a a a

PAGE 261

261 Figure 425. U nprocessed feed (UNP) small plankton (1 32 m) assemblage carbon and nitrogen isotope signatures [ 13C (absolute value) 15N 95 % CI ] among 12 weekly sampling periods n = 6 pond per treatment ; unshared letters denote statistical differences within elements, C or N, (P < 0.05, Bonferroni post test ). 0 5 10 15 20 25 30 35 13C(abs) and 15N dC dN ab bc ab ab bc c c c bc bc bc a a a a a a a a a a a a a

PAGE 262

262 Figure 426. C ottonseed meal (CSM) small plankton (1 32 m) assemblage carbon and nitrogen isotope signatures [ 13C (absolute value) 15N 95 % CI ] among 12 weekly sampling periods n = 6 ponds per treatment ; unshared letters denote statistical differences within elements, C or N, (P < 0.05, Bonferroni post test ). 0 5 10 15 20 25 30 35 13C(abs) and 15N dC dN cd d abcd bcd bcd abcd cd d abc a ab bcd ab abc a cd d abcd abcd abc abcd bcd d a

PAGE 263

263 Figure 427. I norganic fertilizer (INO) small plankton (1 32 m) assemblage carbon and nitrogen isotope signatures [ 13C (absolute value) 15N 95 % CI ] among 12 weekly sampling periods n = 6 ponds per treatment ; unshared letters denote statistical differences within elements, C or N, (P < 0.05, Bonferroni post test ). 10 5 0 5 10 15 20 25 30 35 C(abs) and N dC dN cd d ab abcd abc abcd bcd ab ab a abcd c b c c c c c bc a c c c ab

PAGE 264

264 Figure 428. T ime averaged small plankton (1 32 m) carbon and nitrogen isotope signatures [ 13C (absolute value) 15N 95 % CI ] among 12 weekly sampling periods n = 6 ponds per treatment ; unshared letters denote statistical differences within elements, C or N, (P < 0.05, Tukeys multiple comparison test). 0 5 10 15 20 25 30 35 PRO UNP CSM INO C (abs) & N dC avg dN avg a b b c a b b b

PAGE 265

265 Figure 429. Average p ond trial large ( > 31 mm SL) swordtail and nutrient source isotope signature difference magnitudes ( 13C, 1 5N 95% CI ) at harvest, n = 6 ponds per treatment ; unshared letters denote statistical differences within elements, C or N, (P < 0.05, Tukeys multiple comparison test). 3 1 1 3 5 7 9 N C N C N C N C PRO UNP CSM INO 13C (abs) & 15N Difference a b b a a b c

PAGE 266

266 Figure 430. Average p ond trial small ( SL) swordtail and nutrient source isotope signature difference magnitudes ( 13C 15N 95 % CI) at harvest n = 6 ponds per treatment ; unshared letters denote statistical differences within elements, C or N, (P < 0.05, Tukeys multiple comparison test ). 3 1 1 3 5 7 9 N C N C N C N C PRO UNP CSM INO 13C & 15N Difference a a a a a b c

PAGE 267

267 Figure 431. Average p ond trial large ( > 31 mm SL) swordtail and large zooplankton carbon isotope signature difference magnitudes (sign neutral) at harvest ( 13C 95 % CI), n = 6 ponds per treatment ; unshared letters denote statistical differences (P < 0.05, Tukeys multiple comparison test). 1 5 1 0 0 5 0 0 0 5 1 0 1 5 2 0 2 5 3 0 3 5 PRO UNP CSM INO C Difference a a a a

PAGE 268

268 Figure 432. Average p ond trial large ( > 31 mm SL) swordtail and large zooplankton nitrogen isotope difference magnitudes (sign neutral) at harvest ( 1 5N 95 % CI), n = 6 ponds per treatment ; unshared letters denote statistical differences (P < 0.05, Tukeys multiple comparison test). 2 0 1 0 0 0 1 0 2 0 3 0 4 0 5 0 PRO UNP CSM INO 15N Difference a b a b

PAGE 269

269 Figure 433. Average p ond trial small ( SL) swordtail and large zooplankton carbon isotope difference magnitudes (sign neutral) at harvest ( 13C 95 % CI), n = 6 ponds per treatment ; unshared letters denote statistical differences (P < 0.05, Tukeys multiple comparison test ). 1 5 1 0 0 5 0 0 0 5 1 0 1 5 2 0 2 5 3 0 3 5 PRO UNP CSM INO 13C Difference a a a a

PAGE 270

270 Figure 434. Average p ond trial small ( SL) swordtail and large zooplankton nitrogen isotope difference magnitudes (sign neutral) at harvest ( 1 5N 95 % CI), n = 6 ponds per treatment ; unshared letters denote statistical differences (P < 0.05, Tukeys multiple comparison test ). 2 1 0 1 2 3 4 5 PRO UNP CSM INO 15N Difference a b a ab

PAGE 271

271 Figure 435. Average p ond trial large zooplankton and applied nutrient carbon isotope signature (sign neutral) difference magnitudes ( 13C 95 % CI), n = 6 ponds per treatment ; unshared letters denote statistical differences (P < 0.05 Tukeys multiple comparison test ). 3 5 3 0 2 5 2 0 1 5 1 0 0 5 0 0 0 5 PRO UNP CSM 13C Difference a b b

PAGE 272

272 Fi gure 436. Average p ond trial large zooplankton and applied nutrient nitrogen isotope signature (sign neutral) difference magnitudes ( 1 5N 95 % CI), n = 6 ponds per treatment ; unshared letters denote statistical differences (P < 0.05 Tukeys multiple comparison test ). 0 1 2 3 4 5 6 7 8 PRO UNP CSM INO 15N Difference a a a b

PAGE 273

273 Figure 437. Average p ond trial small plankton and applied nutrient carbon isotope signature (sign neutral) difference magnitudes ( 13C 95 % CI), n = 6 ponds per treatment ; unshared letters denote statistical differences (P < 0.05, Tukeys multiple comparison test ). 0 0 0 5 1 0 1 5 2 0 2 5 3 0 3 5 4 0 4 5 PRO UNP CSM 13C Difference b a ab

PAGE 274

274 Figure 438. Average p ond trial small plankton and applied nutrient nitrogen isotope signature (sign neutral) difference magnitudes ( 1 5N 95 % CI), n = 6 ponds per treatment ; unshared letters denote statistical differences (P < 0.05 Tukeys multiple comparison test ). 0 0 0 2 0 4 0 6 0 8 1 0 1 2 1 4 1 6 1 8 2 0 PRO UNP CSM INO 15N Difference a a a b

PAGE 275

275 Figure 439. C artesian plot of processed feed ( PRO) treatment harvest swordtail, processed feed, and three plankton size assemblage 13C (absolute value) and 1 5N values ( 95 % CI) ; n = 6. Broodstock (BSF Female, BSM Male), and initial fry 13C and 1 5N value error measures omitted for clarity; n = 11, 10, and 21, respectively. 15N C15 BSF BSM 0 3 6 9 12 initial fry PRO lg fish harvest PRO sm fish harvest PRO feed PRO lg plankton PRO med plankton PRO sm plankton PRO lg fish harvest adj. PRO sm fish harvest adj. 13 27 19 23 25 15

PAGE 276

276 Figure 440. Cartesian plot of unprocessed feed ( UNP ) treatment harvest swordtail, processed feed, and three plankton size assemblage 13C (absolute value) and 1 5N values ( 95 % CI) ; n = 6. Broodstock (BSF Female, BSM Male), and initial fry 13C and 1 5N value error measures omitted for clarity; n = 11, 10, and 21, respectively 15N 13C15 BSF BSM 3 6 9 12 initial fry UNP lg fish harvest UNP sm fish harvest UNP feed UNP lg plankton UNP med plankton UNP sm plankton (abs. value) UNP lg fish harvest adj. UNP sm fish harvest adj. 27 19 23 25 15 0

PAGE 277

277 Figure 441. Cartesian plot of cottonseed meal ( CSM ) treatment harvest swordtail, cottonseed meal, and three plankton size assemblage 13C (absolute value) and 1 5N values ( 95 % CI) ; n = 6. Broodstock (BSF Female, BSM Male), and initial fry 13C and 1 5N value error measures omitted for clarity; n = 11, 10, and 21, respectively 15N 13C15 BSF BSM 3 6 9 12 initial fry CSM lg fish harvest CSM sm fish harvest CSM fertilizer CSM lg plankton CSM med plankton CSM sm plankto (abs. value) CSM lg fish harvest adj. CSM sm fish harvest adj. 27 19 23 25 15 0

PAGE 278

278 Figure 442. Cartesian plot of inorganic fertilizer (INO) treatment harvest swordtail, and three plankton size assemblage 13C (absolute value) and 1 5N values ( 95 % CI) ; n = 6. Broodstock (BSF Female, BSM Male), and initial fry 13C and 1 5N value error measures omitted for clarity; n = 11, 10, and 21, respectively 15N 13C15 BSF BSM 15 3 6 9 12 initial fry INO lg fish harvest INO sm fish harvest INO lg plankton INO med plankton INO sm plankton (abs. value) INO lg fish harvest adj. INO sm fish harvest adj. 27 19 23 25 0

PAGE 279

279 Figure 443. Isotopic variation among five pond groups (small, medium, and large plankton assemblages, small and large harvest swordtails ) for four applied pond nutrients n = 6 ponds per treatment 0 0 0 5 1 0 1 5 2 0 2 5 3 0 PRO UNP CSM INO Cumulative Taxa SD dC SD dN SD

PAGE 280

280 Figure 444. Postulated cottonseed meal ( CSM ) food web based upon isotopically derived Euclidean distance and IsoSource program s wordtail diet estimates (Table 4 1) ; small phytoplankton (small plankton assemblage: 132 m), medium zooplankton (medium plankton assemblage: 32 200 m), and large zooplankton (large plankton assemblage: 200+ m) large fish >31 mm SL small fish <31 mm SLlarge zooplankton medium zooplankton small phytoplankton applied cottonseed meal nutrient uptake via remineralized CSM DIC and DIN growth transition other DIC and DIN sources ? (taxa respiration/excretion)/organic matter decompos

PAGE 281

281 Figure 445. Postulated inorganic fertilizer ( INO) food web based upon isotopically derived Euclidean distance and IsoSource program swordtail diet estimates (Table 4 1) ; small phytoplankton (small plankton assemblage: 132 m), medium zooplankton (medium plankton assemblage: 32200 m), and l arge zooplankton (large plankton assemblage: 200+ m) large fish >31 mm SL small fish <31 mm SLlarge zooplankton medium zooplankton small phytoplankton applied inorganic nitrogen fertilizer growth transition cannibalism??

PAGE 282

282 Figure 446. Average total ammonia nitrogen (TAN) concentrations measured for 12 weekly sampling periods (mg/L 95 % CI), n = 6 ponds per treatment ; unshared letters denote statistical differences (P < 0.05, Tukeys multiple c omparison test). 0 0 0 2 0 4 0 6 0 8 1 0 1 2 1 4 1 6 PRO UNP CSM INO TAN mg/La a b b

PAGE 283

283 Figure 447. P rocessed feed (PRO) treatment average p ond c hl orophyll [a] among 12 weekly sampling periods (g/L 95 % CI), n = 6 ponds per treatment ; unshar ed letters denote statistical differences (P < 0.05, Bonferroni post test). 0 20 40 60 80 100 120 140 160 180 200 Chlorophyll [a] g/L a bcd cd d abc ab abc ab ab abc abc abc

PAGE 284

284 Figure 448. Unprocessed feed (UNP) treatment average pond chlorophyll [a] among 12 weekly sampling periods ( g/L 95 % CI), n = 6 ponds per treatment ; unshared letters denote statistical differences (P < 0.05, Bonferroni post test). 0 20 40 60 80 100 120 140 160 180 200 Chlorophyll [a] ug/La a c bc abc abc a a a abc ab abc

PAGE 285

285 F igure 449. Cottonseed meal fertilizer (CSM) treatment average pond chlorophyll [a] among 12 weekly sampling periods ( g/L 95 % CI), n = 6 ponds per treatment ; unshared letters denote statistical differences (P < 0.05, Bonferroni post test). 0 20 40 60 80 100 120 140 160 180 200 Chlorophyll [a] ug/La bc bc bc abc ab bc ab bc c c ab

PAGE 286

286 Figure 450. I norganic fertilizer (INO) treatment average pond chlorophyll [a] among 12 weekly sampling periods ( g/L 95 % CI), n = 6 ponds per treatment ; unshared letters denote statistical di fferences (P < 0.05, Bonferroni post test). 0 20 40 60 80 100 120 140 160 180 200 Chlorophyll [a] ug/La a a a a a a a a a a a

PAGE 287

287 Figure 451. Time averaged m ean c hl orophyll [a] concentrations (g/L 95% CI) measured over 12 week ly sampling periods among four pond treatments n = 6 ponds per treatment ; unshared letters denote statistical differences (P <0.05, Tukeys multiple comparison test). 0 10 20 30 40 50 60 70 80 90 100 PRO UNP CSM INO Chlorophyll [a] ug/L Chl [a] a a a b

PAGE 288

288 Figure 452. D iagram illustrating potential isotopic collinearity (13C 1 5N ) of two hypothetical composite diets (diet 1 and 2), and associated dietary prey inclusion rate bias ; given isotopic signatures of predator and three potential prey (A, B, C) diet 1 and diet 2 could not be isotopically discernable a s sep a rate diets. Therefore the probabilities of diet 1 or diet 2 being the more likely true diet can not be determined. A B Ccomposite diets isotopic similarities 33% 66% Diet 1: 100% Prey B Diet 2: 33% Prey A and 66% Prey C (0% Prey B) A B C D E predator ^^C N

PAGE 289

289 Table 4 1. Dietary estimates of harvest fish derived from isotopic results from IsoSource linear mixing model computer program ( 95% CI) and Euclidean distance measurements; raw Euclidean distance measure values also included. Presump tive Nutrient Source Target Species Feed/Fertilizer Large P lankton Med ium Plankton Small Plankton PRO Large Fish 90.1 % IS (87.7-92.5%) 70 % EDIM 0.31 EDM 0.2 % IS (0.0 -0.7%) 7 % EDIM 2.96 EDM 0.8 % IS (0.0 -1.8%) 16 % EDIM 1.36 EDM 8.9 % IS (6.8 -11.0%) 7 % EDIM 3.16 EDM PRO Small Fish 43.5 % IS (37.8-49.2%) 3 4 % EDIM 1.24 EDM 7.0 % IS (1.4 -12.6%) 14 % EDIM 3.07 EDM 14.8 % IS (3.4 -26.2%) 33 % EDIM 1.29 EDM 34.7 % IS (31.6-37.8%) 1 9 % EDIM 2.31 EDM UNP Large Fish 91.1 % IS (86.2-96.0%) 90 % EDIM 0.06 EDM 1.0 % IS (0.0 -2.1%) 2 % EDIM 2.40 EDM 6.8 % IS (1.6 -12.1%) 6 % EDIM 0.51 EDM 1.0 % IS (0.2 -1.8%) 2 % EDIM 2.71 EDM UNP Small Fish 22.7 % IS (8.4 -37.0%) 3 7 % EDIM 0.88 EDM 13.4 % IS (8.4 -18.4%) 1 4 % EDIM 2.42 EDM 29.4 % IS (10.9-47.9%) 3 2 % EDIM 1.04 EDM 34.5 % IS (32.0-37.0%) 1 7 % EDIM 1. 89 EDM CSM Large Fish 21.4 % IS (11.1-31.7%) 1 7 % EDIM 1.96 EDM 22.0 % IS (6.1 -38.0%) 23 % EDIM 1.48 EDM 53.6 % IS (29.3-78.0%) 51 % EDIM 0. 6 6 EDM 1.1 % IS (0.0 -3.6%.) 9 % EDIM 3.66 EDM CSM Small Fish 0.6 % IS (0.0 -1.44%) 2 1 % EDIM 2.10 EDM 63.3 % IS (64.9-61.7%) 25 % EDIM 1.80 EDM 2.0 % IS (0.0 -4.0%) 40 % EDIM 1.1 3 EDM 34.1 % IS (33.2-35.0%) 14 % EDIM 3.15 EDM INO Large Fish ------------------------------36.07* EDM 86.5 % IS (82.7-90.3%) 65 % EDIM 0.63 EDM 8.6 % IS (2.4 -14.8%) 25 % EDIM 1.6 6 EDM 4.9 % IS (1.8 -8.1%) 10 % EDIM 4.06 EDM INO Small Fish ------------------------------6.83* EDM 67.3 % IS (64.1-70.6%) 34 % EDIM 1.73 EDM 7.5 % IS (1.5 -13.5%) 47 % EDIM 1.26 EDM 25.3 % IS (22.4-28.2%) 19 % EDIM 3.14 EDM IS: IsoSource Program; ( X 95% CI) (http://www.epa.gov/wed/pages/models/stableIsotopes/isotopes.htm) EDIM: Euclidean Distance Inclusion Measurement EDM: Euclidean Distance Measure inorganic fertilizer tre atment 13C ~ 15 (aqueous HCO3; Boutton 1991)

PAGE 290

290 Table 4 2. Dietary estimates of preharvest fish derived from isotopic results from IsoSource linear mixing model computer program ( 95 % CI) and Euclidean distance measurements; raw Euclidean distance measure values also included. Presumptive Nutrient Source Target Species Feed/Fertilizer Large Zooplankton Medium Plankton Small Plankton PRO Pre -Harvest Fry 49.8 % IS (45.7-53.9%) 30 % EDIM 1.24 EDM 46.9 % IS (44.0-49.8%) 23 % EDIM 1.60 EDM 3 % IS (0.2 -5.8%) 38 % EDIM 0.97 EDM 0.3 % IS (0.0 -0.8%) 9 % EDIM 4.23 EDM UNP Pre -Harvest Fry 33.1 % IS (19.4-46.9%) 24 % EDIM 1.10 EDM 42.4 % IS (35.4-49.4%) 19 % EDIM 1.42 EDM 23.7 % IS (4.8 -42.6%) 50 % EDIM 0.52 EDM 0.7 % IS (0.0 -1.6%) 8 % EDIM 3.43 EDM CSM Pre -Harvest Fry 54.6 % IS (50.3-58.9%) 34 % EDIM 1.46 EDM 20.9 % IS (8.9 -32.9%) 21 % EDIM 2.30 EDM 23.2 % IS (7.6 -38.8%) 31 % EDIM 1.57 EDM 1.3 % IS (0.0 -2.7%) 14 % EDIM 3.59 EDM INO Pre -Harvest Fry ----------------------------------------34.1 % IS (32.7-35.5%) 32 % EDIM 1.37 EDM 65 % IS (62.7-67.3%) 55% EDIM 0.79 EDM 0.9 % IS (0.0 -1.8%) 13 % EDIM 3.33 EDM IS: IsoSource Program; ( X 95 % CI) (http://www.epa.gov/wed/pages/models/stableIsotopes/isotopes.htm) EDIM: Euclidean Distance Inclusion Measurement EDM: Euclidean Distance Measure inorganic fertilizer treatment 13C ~ 15 (aqueous HCO3; Boutton 1991)

PAGE 291

291 Table 4 3. Back calculated primary producer carbon isotope signature ( 13C) estimates based upon actual (*) and hypothetical pond guild carbon signature results and trophic level estimations (assuming 0.64 13C enrichment per trophic level). Taxon Estimated t rophic l evel PRO UNP CSM INO Small Plankton 1 -32 m 1.0 27.18 26.53 28.64 23.26 Medium Plankton 32-200 m 2.0 -24.79 -24.50 -25.40 -22.83 Large Zooplankton 3.0 -24.34 -24.34 -25.88 -22.47 Small Fish 3.5 25.91 25.72 26.85 21.89 Large Fish 4.0 25.31 25.18 22.07 22.07 Mean estimated 1 producer 13C 25.51 25.25 26.63 22.50 Applied nutrient 13 C signature for reference 23.74 23.95 26.43 n/a actual small plankton assemblage carbon isotope signature results.

PAGE 292

292 Table 4 4. Isotopically derived trophic differences among harvest fish predators, applied nutrients and plankton size assemblages (via 1 5N); based upon 1 3C provided as supplemental information. Assuming 0.64 13C, 3.03 15N, enrichment rates from indoor feeding trial (Chapter 2). Trophic d istance estimate PRO UNP CSM INO Lg Fish / Nutrient 15N 13 C 0.97 0.54 1.01 1.09 1.48 3.08 2.48 n/a Lg Fish / Lg Zooplankton 1 5N 1 3 C 0.07 0.41 0.23 0.21 0.62 0.44 0.80 1.18 Lg Fish / Med Plankton 1 5N 1 3 C 0.60 1.28 0.66 0.80 0.81 0.47 1.30 3.18 Lg Fish / Sm Plankton 1 5N 1 3 C 1.04 5.92 1.10 5.20 1.31 6.53 2.06 4.85 Sm Fish / Nutrient 1 5N 1 3 C 1.08 0.89 1.11 0.27 1.67 1.85 2.11 n/a Sm Fish / Lg Zooplankton 1 5N 1 3 C 0.18 1.84 0.34 1.14 0.81 1.67 0.43 0.96 Sm Fish / Med Plankton 1 5N 1 3 C 0.71 0.15 0.78 0.56 1.01 0.76 0.93 2.96 Sm Fish / Sm Plankton 1 5N 1 3 C 1.15 4.48 1.22 3.77 1.51 5.30 1.70 4.63 Lg Zoo / Nutrient 15N 13 C 0.88 0.99 0.77 0.87 0.82 3.19 1.66 n/a Lg Zoo / Med Plank 1 5N 1 3 C 0.46 1.64 0.48 0.72 0.16 0.58 0.48 2.14 Lg Zoo / Sm Plank 15N 13 C 0.98 6.37 0.88 4.90 0.66 6.63 1.24 3.80 Med Plank / Nutrient 1 5N 1 3 C 0.42 0.65 0.28 0.14 0.66 2.61 1.18 n/a Med Plank / Sm Plankton 1 5N 1 3 C 0.52 4.73 0.39 4.17 0.50 6.06 0.77 1.67 Sm Plank / Nutrient 15N 13 C 0.58 0.86 0.58 0.64 0.55 0.55 1.26 n/a Negative values denote probable lack of trophic linkage, values near unity denote roughly one trophic level difference and high probability of direct trophic linkage, values greater than one denote probable distant linkage or lack of trophic linkage. as suming 19 13C enrichment rate between primary producers (C3) and carbon pool 13C (Fry 2006), and ~ 0.0 15N enrichment rate between primary producers and exploited nitrogen pool N (Fogel and Cifuentes 1993, Fry 2006).

PAGE 293

293 Table 4 5. Processed feed (PRO) pond treatment dietary components ranked in order of percentage importance (1 highest, 4 lowest) for the Euclidean distance and IsoSource linear algebra computer program. Processed Feed Lg. Plankton Med. Plankton Sm. Plankton Large Fish IS 1 4 3 2 Large Fish ED 1 3.5 2 3.5 Small Fish IS 1 4 3 2 Small Fish ED 1 4 2 3 Table 4 6. Unprocessed feed (UNP) pond treatment dietary components ranked in order of percentage importance (1 highest, 4 lowest) for the Euclidean distance and IsoSource linear algebra computer program. Unprocessed Feed Lg. Plankton Med. Plankton Sm. Plankton Large Fish IS 1 3.5 2 3.5 Large Fish ED 1 3.5 2 3.5 Small Fish IS 3 4 2 1 Small Fish ED 1 4 2 3 Table 4 7. Cottonseed meal (CSM) pond treatment dietary components ranked in order of percentage importance (1 highest, 4 lowest) for the Euclidean distance and IsoSource linear algebra computer program. Table 4 8. INO pond treatment dietary components ranked in order of percentage importance (1 highest, 4 lowest) for the Euclidean distance and IsoSource linear algebra computer program. Large. Plankton Medium Plankton Small Plankton L arge Fish IS 1 2 3 Large Fish ED 1 2 3 Small Fish IS 1 3 2 Small Fish ED 2 1 3 Cottonseed Meal Lg. Plankton Med. Plankton Sm. Plankton Large Fish IS 3 2 1 4 Large Fish ED 3 2 1 4 Small Fish IS 4 1 3 2 Small Fish ED 3 2 1 4

PAGE 294

294 Table 4 9. Food source inclusion rates for large zooplankton derived from isotopic signature results, IsoSource program, linear mixing models and Euclidean distance measurements. Presumptive Nutrient Source Target Species Feed/Fertilizer Medium Plankton Small Plankton PRO Large Zooplankton 93.1 % IS (90.7-95.5%) 54 % EDIM 1.29 EDM 0.3 % IS (0.0 -0.8%) 31 % EDIM 2.27 EDM 6.6 % IS (4.4 -8.8%) 15 % EDIM 4.69 EDM UNP Large Zooplankton 73.8 % IS (62.6-85.0%) 45 % EDIM 1.39 EDM 6.4 % IS (1.3 -11.5%) 39 % EDIM 1.55 EDM 19.8 % IS (12.1-27.5%) 16 % EDIM 3.85 EDM CSM Large Zooplankton 59.2 % IS (55.1-63.3%) 38 % EDIM 2.92 EDM 38.5 % IS (35.6-41.4%) 41 % EDIM 2.57 EDM 2.2 % IS (0.2 -4.2%) 21 % EDIM 5.18 EDM INO Large Zooplankton ------------------------------36.61* EDM 33 % Linear Model ( N only) 53 % EDIM 1.69 EDM 67% Linear Model ( N only) 47 % EDIM 1.89 EDM IS: IsoSource Program; X (95%CI lower limit 95%CI upper limit) ( http://www.epa.gov/wed/pages/models/stableIsotopes/isotopes.htm ) Linear Model Using Binomial Solution (INO treatment only) EDIM: Euclidean Distance Inclusion Measurement EDM: Euclidean Distance Measure inorganic fertilizer ~ 15 13C (aqueous HCO3; Boutton 1991)

PAGE 295

295 Table 410. Trophic level estimat es of plankton assemblages and harvest fish within treatments, trophic levels relative to treatment small plankton assemblage 15N values only. Taxon 15 N difference ( 15 N) Hypothe tical t rophic l evel PRO UNP CSM INO Small plankton 1 32 m 15 N 1.0 4.38 1.0 4.47 1.0 4.29 1.0 1.19 1.0 Medium plankton 32200 m 15 N 2.0 5.95 1.5 5.66 1.4 5.80 1.5 3.51 1.8 Large plankton 15 N 3.0 7.35 2.0 7.13 1.9 6.29 1.7 4.95 2.2 Small f ish SL mm 15 N 3.5 7.96 2.2 8.18 2.2 8.85 2.5 6.33 2.7 Large f ish SL > 31 mm 15 N 4.0 7.62 2.1 7.82 2.1 8.26 2.3 7.44 3.1 Taxa t rophic position relative to primary producer (Vander Zanden and Rasmussen 1999); e.g., PRO Small Fish trophic position: (7.96 4.38)/3.03 + 1.0 (autotroph) trophic levels. Nitrogen isotope enrichment magnitude value from Chapter 2 ( 15N = 3.03 per trophic level).

PAGE 296

296 CHAPTER 5 LARGE ZOOPLANKTON ASSEMBLAGE TAXA IDEN TIFICATION AND ENUMERATION L arge (> 200 m) plankton assemblage taxa were examined and compared among pond nutrient treatments (primarily zooplankton) in an attempt to identify and enumerate major taxa associated with differing pond ma nagement methods (feeds versus fertilizers), and fish production rates. Zooplankton taxa and numbers also were compared among weekly sampling dates to characterize large size assemblage zooplankton community change over time. Large plankton assemblages within the different ponds treatments were compared using the simplified Morisitas similarity index (MSI) and the percent similarity index (PSI), to determine if plankton assemblage commonalities occurred among the different treatments ( Krebs 1998). Als o, if common plankton assemblage characteristics occurred within treatments: (1) were assemblage differences present among treatments? (2) were these differences along a predictable nutritional/nutrient type grade ?; i.e. were feed treatment large plankton community taxa more similar to each other than either feed treatment large plankton community was to those of the two fertilizers and vice versa ? Because direct microscopic analyses of swordtail gut contents were not possible due to the finely ground and l argely unidentifiable prey items that resulted from the action of the swordtails pharyngeal teeth, more indirect dietary habit investigation methods were employed. One indirect method was the use of large plankton assemblage communit y analys e s [index of relative importance (%IRI)], which were performed as proxies for potential swordtail prey/diets. Potential large plankton prey taxa abundances (% number and % volume) and indices of relative importance were assumed to be correlated with their potential im portance as prey items for swordtails. Although this indirect method of characterizing possible swordtail diets was used, additional lines of evidence and more direct dietary habit analys i s methods other than gut

PAGE 297

297 content analysis also were utilized (Chapter 4: fish and pond taxa isotopic signature analyses; Chapter 6: 24hour captive feeding trials). Isotopic signature data for applied pond nutrients (inorganic fertilizer nitrogen only), plankton assemblages and swordtails were used to generate isoto pically derived swordtail diet estimates (Chapter 4) which were used to roughly quantify potential predation pressure occurring for the different plankton size assemblages and dietary inclusion rates for nonINO applied nutrients. Additionally, 24 hour c ontrolled feeding trials were performed with individual swordtails incubated with pond water to determine potential swordtail predation effects upon the large zooplankton communit ies within the various treatment ponds (Chapter 6). Deterministic large pla nkton assemblage similarity estimates (MSI and PSI) were calculated for pairwise treatment comparisons among the four applied pond nutrient treatments using mean %IRI values calculated for each treatment for each of the 10 analyzed weekly sampling dates (Krebs 19 98, Cortes 1997). Due to the deterministic nature of the similarity (MSI and PSI) indices generated using large plankton assemblage taxa %IRI values, error measures were lacking for these similarity value estimates and conclusions regarding large plankton community similarities were based upon arbitrarily chosen threshold critical values ( Za ret and Rand 1971, Cailliet and Barry 1978, Lindquist 199 8) A Bayesian resampling bootstrap computer program was developed to generate similarity index values (MSI and PSI) and error measures from major large plankton assemblage taxa %N within the large zooplankton assemblages occurring within the various treatment ponds. Resampled indices and error measures were produced using replicate pond large plankton assemblage datasets (Appendix B: MSI/ PSI/% N bootstrap computer program ) that were repeatedly sampl ed over numerous iterations. Error measure calculations allowed statistical comparisons of large zooplankton

PAGE 298

298 assemblage MSI and PSI to be made among sampling dates within pond nutrient treatment pair compari s ons Methods Plankton Sample Collection A more detailed description of methods used for weekly pond plankton sample collections for stable isotope analyses and microscopic examination is given in Chapter 4 (Methods) A brief summary of the methods used to collect plankton for microscopic examination i s given below. Weekly pond water samples (1 L/pond) were taken from twentyfour replicate ponds (13:0015:00 EST ), that were randomly and evenly assigned into four applied pond nutrient treatments (6 replicate ponds per treatment) for a period of twelve weeks. Samples w ere fixed within one hour of collection using Lugols preservative ( APHA 19 89) and stored in light proof containers. At a later date preserved samples w ere serially filtered through three sieves ( 200m, 35m and 1m square mesh ) ; r esult ing in three size fraction s (identical size fractions as for isotope analyses; Chapter 4) that were stored in light proof storage boxes in 9ml brown flint glass vials containing ~ 610 drops of Lugols solution. Sample water filtered through the final sieve (< 1 m ) was discarded Large Plankton Assemblage Examination and Processing Preserved large plankton assemblages were examined for zooplankton, phytoplankton, and incidental organisms by identifying and enumerating all taxa via light microscopy (Olympus BH2 binocular microscope with 4X, 10X, 20X, 40X objectives and 10X ocular for 40X, 100X, 200X and 400X total magnification) L arge plankton assemblage samples were serially examined using a SedgewickRafter plankton co unting cell in 1ml aliquots until the entire sample was surveyed. The SedgewickRafter counting cell slide ha d a 50 mm x 20 mm

PAGE 299

299 graduated grid base and depth of 1 mm for a total volume of exactly one milliliter, the graduated volume and grid base allows f or quantitative assessment of organism densities and measurement of organism linear dimensions, respectively. Microscopically examined (processed) samples were then returned to their original flint glass sample vials for archival purposes. Large plankton samples from the first two weeks of the 12 week trial were discarded due to a general lack of organisms present within pond water samples during the first two weeks of the 12week study (31 March 2006 and 6 April 2006). Absence of large plankton were li kely due to pond sterilization with hydrated lime prior to stocking cooler water temperatures, and shorter photoperiods associated with late winter/early spring at the start of the outdoor pond trial. Microscopically examined large plankton assemblage samples thus consisted of weekly samples from four pond nutrient treatments with six replicate ponds per treatment over a course of ten weeks (4610 =240) samples. Unfortunately, one sample was lost due to human error ( from pond B1 13 April 2006, UNP feed treatment). Limited samples for the two smaller size fractions : medium ( 35 200 um ) and small ( 135 um) were examined for randomly selected ponds and dates Unfortunately, due to the large number of individual organisms per sample, their morphological similarity and large amounts of flocculent material in the sample only a cursory examination was feasible making organism identification and enumeration extremely imprac t ical and time consuming for the medium (mixed phytoplankton and small zooplankton) and small plankton (primarily phytoplankton) size fractions T he two smaller plankton size assemblages were ultimately excluded from quantitative microscopic analyses. Large Plankton Assemblage Taxa Identification Large plankton (primarily zooplankton) assemblage fractions were analyzed to determine assemblage organisms present by taxon, number and estimated volume. All large plankton were

PAGE 300

300 identified to order and lower when possible and notable plankton organisms were photographed depending on it s commonality /novelty, possession of distinguishing characteristics, and determination of linear dimensions for volumetric estimation Twenty randomly selected digital photomicrographs of each ma jor plankton ic item were measured using the 1.0 mm 1.0 mm grid of the SedgwickRafter cell as a size reference under known magnification. The determination of linear dimensions a llowed rough volumetric approximations (length x width x height) of items based upon simple geometric solids (sphere, cube, cone, etc .) or combination of solids that roughly corresponded to the geometry of the item. With minor differences depending upon taxon shape, the general taxon volume estimate equation is given as ( as an example, cumulative volume of cube hexahedron taxon): l w h n (Equation 5.1) where: l = length in m w = width in m h = height in m n = number of a given taxon. Taxon percent volume (%V) and number (%N) were calculated as the cum ulative taxon volume or number as a percentage of the total volume or number of large zooplankton items (all taxa) within a sample. Taxon percent frequency of occurrence (%FO) was calculated as the percentage of treatment replicates where a given taxon was present F or the present study, the maximum number of replicates was six ponds for each of the four pond nutrient treatments. Large Plankton Community Analyses were perfo rmed using the index of relative importance (IRI), which is a compound index that combines metrics of taxon number, weight or volume (physical size) and frequency of

PAGE 301

301 occurrence within the sample population, and by inference the community population under investigation ( Pinkas et al. 1971, Cortes 1997 ) Specifically, percent index of relative importance (%IRI) analyses were used to evaluate taxa importance within the plankton community to facilitate comparisons among replica tes, treatments and potential comparisons with past and future studies and differing ecosystems ( Cortes 1997). In addition to taxa %IRI values, zooplankton taxa densities (taxa number/L water sample) and zooplankton taxa volume (ml zooplankton taxa/L wat er sample) estimates also were calculated. Advantages of using %IRI are that a single value is used to describe multiple community metrics, greatly simplifying calculations, interpretation of data, and comparisons among studies. For example, if a numeri cally dominant taxon has a small individual volume or biomass and is found in just a few samples within the community, this numerical dominance (and associated bias) is offset by its small size and limited distribution within the community, thereby reducin g its community IRI value and giving a more realistic, less biased measure of the taxons importance within the community. This is an advantage when evaluating a community component as a potential prey item for a predator species when direct stomach content data are unavailable. The index of relative importance for a plankton community (IRI) is given by the formula (Pinkas et al. 1971): IRI(%%)%iiiiNVFO ( Equation 5.2) where: %Ni = the numerical % fraction of a single plankton community compone nt/taxa ( i ) among the total number of plankton community items within a sample %Vi = the percentage volume fraction of a single plankton community component/taxa ( i ) among the total community volume within a sample

PAGE 302

302 %FOi = the fre quency that a single plankton community component/taxa ( i ) occurs within replicate samples from a given population. IRI values range from a maximal value of 200,000, when a single community component is contained within all samples to a value of zero when a component is absent from the community. Due to the extremely large range of IRI values that a given community component can have, Cortes has advocated %IRI as a transformation of raw IRI values in order to put the relative importance index conce pt into a more intuitively meaningful context ( Cortes 1997). Percent IRI is given as: 1IRI % IRI = 100 IRIi i n i i ( Equation 5.3) where: IRIi = the raw IRI value for the i th community component %IRIi = the percentage of the total IRI contributed by the i th community component Percent IRI (%IRI) is a data transform ation of raw IRI values standardized to the total IRI values calculated for a given sample. Again, advantages of using %IRI over raw IRI values is that transform ed %IRI values allow for easier interpretation of the community components relative importance to the community and potential predator species I t also facilitates comparisons among different samples within and among different studies occurring in different geographical areas and time periods. An arbitrary %IRI cutoff value of 0.5% was used to limit plankton analyses to those assemblage components with the greatest potential n utritional impact on pond fry and broodstock. Thirty large plankton assemblage components regularly had %IRI values equal to or greater than 0.5%, thereby meeting the cutoff criterion (Table 5.1 ).

PAGE 303

303 Two different methods of calculating %IRI were employed, the first consisted of using large zooplankton assemblage taxa data from pooled replicates (pooled %IRI), and the second consisted of using individual replicate data (unpooled %IRI). Large zooplankton assemblage taxa and potential swordtail prey pooled % IRI values (Equations 5.25.3) are useful in describ ing which taxa were present within the different pond nutrient treatments as a whole. Unpooled large zooplankton taxa %IRI values (Equations 5.4) are useful in that they describe the taxa present within the different pond nutrient treatments on a n average basis among replicate ponds within the different pond nutrient treatments. Unpooled average relative importance values allow variation of taxa importance among ponds within treatments to be examined, a nd how these variations might affect potential predators within different ponds and among the different pond treatments. Large plankton assemblage taxa unpooled %IRI values were calculated as: ()%%IRI()%iunpooled NV iiiXXFO (Equation 5.4) where: % N iX is the mean %N for taxon i among replicate ponds % V iX is the mean %V for taxon i among replicate ponds %iFO is the %FO for taxon i among the replicate pond population, ()IRIunpooledi is the pooled IRI for taxon i Percent Number, Volume, and Frequency of Occurrence Graphs Using a compound index such as the IRI had the advantage of reducing the numerical bias of a small and highly numerous community or prey item by incorporati ng the item s relative size into the index, or conversely, a large, but numerically rare item. Unfortunately, IRI and %IRI also have the disadvantage of eliminating information present in individual metrics such as the relative number, relative volume, and the prevalence of an item (frequency of

PAGE 304

304 occurrence) in the community or predators diet (rare or common). Variations in potential prey availability (as %N and %V) among individual prey assemblage populations or among individual predators diets cannot b e examined when metrics are combined into a single compound metric such as IRI and %IRI. T o address the loss of information that occurs when individual large plankton assemblage metrics (%N, %V, %FO) are combined into a single compound index, t hree vector graphs of the se metrics were produced for each sampling date analyzed within each of the four pond nutrient treatments (Cailliet et al. 1996). Large plankton assemblage three vector graphs consisted of the percent number of an assemblage component as a percentage of all components on the positive Y axis, percent volume of an assemblage component as a percent age of the total assemblage volume displayed on the negative Y axis, and the frequency of occurrence (%FO) of the assemblage component among the six pond replicates along the positive X axis The relative size of the parallelogram created on the graph utilizing percent number (%N), percent volume (%V), and percent frequency of occurrence (%FO) data for an assemblage component denoted the taxons importance (positively correlated) in relation to the sum total area prescribed by all large plankton assemblage components within replicate samples for a given treatment ( Table 5.1, 30 large plankton assemblage taxa, each %IRI > 0.5%). G raphical plots of %N %V, and %FO also were calculated in two different ways using pooled and unpooled values. Pooled values were created by summing the counts of a given assemblage component (taxon) among all replicates within a given pond nutrient treatment and calculating its percentage of all organisms counted within all replicates for a given treatment ( pooled %N ), or as the total volume of a given assemblage component (taxon) summed among all replicates as a percentage of the volume of all large zooplankton assemblage components

PAGE 305

305 summe d among all replicates within a given treatment (pooled %V) ; %FO calculation was identical in the case of pooled and unpooled values. Unpooled values were calculated as the mean %N or %V of a given assemblage component among the six replica tes of each pond nutrient treatment; the advantage of unpooled values was that it allowed the calculation of error measures for either %N or %V for a given assemblage component and allowed evaluation of individual varia tion among replicates for given plankton assemblage component s Major Large Plankton Assemblage Taxa Analyses Detailed analyses [weekly large plankton densit y values ( plankton number / L )] of the three large plankton assemblage components that regularly displayed the highest %IRI values (major prey) among pond nutrient treatments and weekly sampling dates also were performed. Major large plankton assemblage taxa analyses were performed to determine if different patterns in availability of these potential major prey occurred among the pond t reatments and/or weekly sampling dates. Large Plankton Assemblage Similarity Indices In an attempt to determine if the magnitude of large zooplankton community similarities/dissimilarities between different treatment ponds were correlated with fish produc tion rates, large plankton assemblage taxa similarities among pond nutrient treatments were examined using large plankton assemblage taxa %IRI, %N and two qualitative similarity indices. The similarity indices used to compare large plankton assemblages between treatments and among sampling dates within treatments were the simplified Morisitas similarity index and percent similarity index (Horn 196 6, Krebs 1998). The simplified Morisitas index (MSI) is given as (Krebs 1998):

PAGE 306

306 Simplified Morisitas Index 1 22 112n ijik i H nn ijik iipp C pp (Equation 5.5) where : pij = the proportion of taxon i of treatment assemblage j pik = the proportion of taxon i of treatment assemblage k n = the number of taxa categories for assemblages j and k A widely used but arbitrary significance level of 0.65 was defined as the critical value for the simplified Morisitas index (Equation 5.5) denoting as semblage taxa similarity ( Zaret and Rand 1971, Cailliet and Barry 1978). The percent similarity index (PSI) is given as (Krebs 1998): (min.)100ijikPSIpp (Equation 5.6) where : pij = the proportion of taxon i of treatment assemblage j pik = the proportion of taxon i of treatment assemblage k n = t he total number of taxa categories for assemblages j and k min.ijikpp = either ijp or ikp which ever is smaller Again, an arbitrary critical value of 60% was chosen to denote similarity between large plankton assemblage between treatments using the percent similarity index (PSI; Equation 5.6). T his value was based upon PSI critical values chosen by previous researchers (Horn 1966, Cailliet and Barry 1978, Lindquist 1998). Higher PSI critical values have been chosen by others in describing plankton assemblage communities ( e.g., 80%; Silver 1971, 1975), but these value s was deemed too high to be useful within the current study.

PAGE 307

307 Each of the two similarity indices (MSI and PSI) was calculated using two different methods: deterministic and Bayesian bootstrap similarity indices values. Large plankton assemblage MSI and PSI were estimated for each pond nutrient treatment pair and weekly sampling date. Deterministic MSI and PSI values were calculated using %IRI values (%IRI 0.5%) of the major large plankton assemblage taxa. Bootstrap resampling simil arity ind ex values were estimated using taxon counts within each replicate pond which were then transformed to %N values (%N program ( Appendix B) written in BASIC to estimate resampled MSI and PS I mean s and their error measures for the ten sampling periods during which large plankton assemblage taxa were examined. Bootstrap MSI and PSI values were estimated by resampling from dataset pools composed of the actual replicate pond large zooplankton assemblage taxa (%N) data, error measures were generated based upon numerous draws (iterations) using t scores, draw pool degrees of freedom, and standard error measurements of the individual iterations to produce 95% confidence intervals for the mean boot strap MSI and PSI values estimated for each treatment pair and sampling date (Zar 19 84, Ott 2000). Additionally, bootstrap percent similarity index (PSI) mean values were generated with varying numbers of replicate data sets randomly pooled (draw size) for analyses (4, 6, 8, 15, and 20) and the number of iterations performed (10, 50, 100, 500, 1000, and 5000) to observe the behavior of the various index estimates and error measures under differing conditions Statistical Analyses Total large zooplankton assemblage densities (numbers/L) and volumes (m m3/L) were compared among treatments and 10 weekly sampling dates using repeated measures two way ANOVA statistical analyses. Time averaged differences in large zooplankton assemblage total densit ies and volume s were compared among treatments using oneway ANOVA. Densities of

PAGE 308

308 the three major large zooplankton groups (consistently highest %IRI values) were compared among treatments and among 10 weekly sampling dates using ANOVA. Treatment pair large zooplankton assemblages were determined to be similar when deterministic large zooplankton assemblage taxa simplified Morisitas and percent similarity indices were above an arbitrarily chosen critical threshold value (0.65 MSI, 60% PSI). Bootstrap MSI and PSI val ues generated using large zooplankton assemblage taxa %N r andomly selected from replicate dataset pairs (6 pairs per iteration) were compared using 95 % CI overlaps among sampling dates within treatment pair comparisons (Liberty BASIC 4.0 computer program; Appendix B ). This allowed statistical comparisons and conclusions to be drawn regarding differences in indices similarity values among sampling dates (e .g., were large zooplankton assemblages of the PRO and CSM treatments more similar for latter sampling dates than they were for earlier sampling dates?). Results Large Zooplankton Production Analyses Differences in large plankton volume s (mm3/ L pond wat er sample) among the four pond nutrient treatments and 10 weekly sampling periods w ere compared ( Figures 5 1 55 ; rep. meas. 2way ANOVA: P(nutrient) = 0.1305, F(nutrient) (2,3) = 0.05; P(time) < 0.0001, F(time) = 8.327, (2,9) = 0.05; P(interaction) = 0.2118, F(interaction) (2,27) = 0.05). A lthough the nutrient treatment was not significant, post hoc pairwise testing found a sin gle pairwise difference among the four nutrient treatments and ten sampling dates for total large plankton volume, CSM 8 June 2006 and INO 8 June 2006 ( P < 0.05, Bonferroni post hoc multiple comparison test ) Time was highly significant and numerous pairwise differences in large plankton volume were found among sampling dates within individual pond nutrient t reatments ( Figures 5 2 53, 55); these differences usually occurred between samples collected early in the trial when large plankton

PAGE 309

309 volumes were low and mid and late samples when plankton standing stocks were considerably higher ( Figure 5 1). Only t he CSM pond nutrient treatment did not have any differences among sampling dates ( Figure 5 4). Time Averaged Large Plankton Volumes Time averaged large (> 200 m) plankton volumes did not differ among the four pond nutrient treatments (Figure 56; rep. meas. 1 way ANOVA, P = 0.1812, F = 1.746, (2,3 ) = 0.05). PRO pond treatment large plankton volume mean X = 3.64 mm3/L ( 2.33 to 4.94 mm3/L 95 % CI ), UNP mean X = 4.08 mm3/L ( 2.47 to 5.70 mm3/L 95 % CI ), CSM mean X = 2.55 mm3/L ( 2.09 to 3.01 mm3/L 95 % CI ), and INO mean X = 3.08 mm3/L ( 2.05 to 4.10 mm3/L 95 % CI ) were all of similar magnitudes. Interestingly, the CSM treatment exhibited much lower variation among its six replicate ponds than the other three treatments (Figure 56). Mean large zooplankton density (number/L) differences among the four pond nutrient treatments and 10 weekly sampling periods also were compared ( Figures 5 7 511; rep. meas. 2way ANOVA, P(nutrient) = 0.4978, F(nutrient) (2,3) = 0.05; P(time) < 0.0001, F(time) = (2,9) = 0.05; P(interaction) = 0.5084, F(interaction) (2,27) = 0.05). Again, although the nutrient factor was found to be nons ignificant, a single significant difference in large zooplankton numbers were found during pairwise comparisons of pond treatments CSM 8 June 2006 and INO 8 June 2006 (P < 0.05, Bonferroni post hoc multiple comparison test). This was the same post hoc pa irwise difference observed for zooplankton volume Several differences in mean large zooplankton density were found among sampling periods within each pond nutrient treatments, except for the PRO treatment (Figure 5 8). The UNP treatment had several pairwise differences that resulted from low densities at the start of the trial and much higher densities mid and late trial ( Figure 5 9). CSM pairwise differences all involved a single sampling date, due to

PAGE 310

310 a single pond that had Filinia rotifer densities two orders of magnitude higher than all other replicates ( Figure 510; 1,644 rotifers in pond A6 on 11 May 2006; mean of 23.4 rotifers in five remaining ponds). Two nonconsecutive sampling dates (11 May 2006, 8 June 2006) were involved in all pairwise differences for the INO treatment (Figure 511). Time Averaged Large Plankton Densit ies Time averaged large plankton assemblage density differences among treatments were not present ( Figure 5 12; rep. meas. 1 way ANOVA, P = 0.4424, F = 0.9244, (2,3 ) = 0.05). PRO pond treatment large zooplankton density mean X = 143.6 plankt on/L ( 202.1 to 85.1 plankt on /L 95 % CI ), UNP mean X = 183.0 plankt on/L ( 256.0 to 110.1 plankton/L 95 % CI ), CSM mean X = 132.8 plankton/L ( 211.1 to 54.5 plankton /L 95 % CI ), and INO mean X = 153.3 plankton /L ( 255.0 to 51.7 mm3/L 95 % CI ) were all of similar magnitudes. Outdoor Pond Trial Major Large Plankton Groups Major large plankton groups were defined as the three assemblage components/taxa which consistently had the highest %IRI values among the thirty commonly found assemblage taxa (%IRI > 0.5 %) among the four nutrient treatments (Figures 5 13 520). The three highest IRI values among the thirty most common large plankton assemblage components/taxa ( Figures 5 215.36, Table 5.1), consistently included: large Calanoid copepods ( Diaptomous spp.; Figure 5 37a ), daphnids ( Moina macrocopa; Figure 537b ), and rotifers ( Filinia spp.; Figure 5 37 c ). Other rotifers and ostracods made sizable contributions to the large plankton assemblage ( Figure 538e h) Copepod Densit ies for 10 Weekly Sampling Periods Weekly Diaptomous copepod (LHC) densities (nu mber/L) differed significantly among sampling dates within nutrient treatments, but not among nutrient treatments ( Figures 5 22, 526,

PAGE 311

311 530, 534: rep. meas. 2P(time) < 0.0001, F(time) any of the sampling dates (P > 0.05, Bonferroni post hoc test). Within individual pond tre atments, copepod density differences among dates were found for the PRO, UNP and INO nutrient treatments, but were not found within the CSM nutrient treatment ( Figures 522, 526, 530, 534). Within the two feed treatments, differences in copepod density among the weekly periods were between samples collected early in the study and samples gathered mid and late trial when copepod standing stock s were considerably higher ( Figures 522, 526). Extremely large densities of copepods for pond A5 (UNP) on 4 M ay 2006 (A5: 297 copepods/L ; remaining replicates mean: X = 33 copepods/L ) and pond A12 (UNP) on 18 May 2006 (A12: 685 copepods /L ; remaining replicates mean: X = 118 copepods /L ) contributed to the extremely large variation for these dates ( Figure 526). Copepod densities among weekly sampling periods for the INO treatment roughly followed the same trajectories as the PRO and UNP treatments; low copepod densities occurred at the start of the trial, steadily increasing before reaching a maximum mid trial, which then slowly tapered off declined toward the end of the trial (Figure 5 34). Of the four treatments, only the CSM weekly copepod densities did not differ among sampling periods ( Figure 530). A single UNP sampling date and pond large zooplankton assemblage sample [13 April 2006/pond (B1)] was lost due to human error. T o perform repeated a measures 2 way ANOVA, a pseudovalue was generated by replacing the missing value with the mean copepod count of the five remaining replicates. Due to the early date of the missing sample (first week of 12 week

PAGE 312

312 trial), low plankton standing stocks had developed, potentially minimizing the impact of the missing value on overall plankton density mean s. Mean rotifer ( Filinia spp.) densities did not differ among pond nutrient treatments, but did differ among sampling intervals within treatments (rep. meas. 2way ANOVA, P(nutrient) = ) = 0.05; Filinia spp. densities did not significantly differ among weekly sampling dates within either feed treatment (Figures 523, 527). In contrast to the PRO and UNP pond treatments, Filinia spp. densities differed among sampling dates for CSM and INO nutrient treatments ( Figures 5 31, 535). The extremely large variation about the mean observed on 11 May 2006 for the CSM treatment was due to a single pond having a rotifer density two orders of magnitude higher than the overall mean of the remaining ponds for this sampling date ( Figure 5 31; pond A6: 1644 rotifers /L ; remaining replicate mean X = 23.4 rotifers /L ). The INO treatment also had large variation about a single sampling date due to the presence of high densities of Filinia spp. for a single pond ( Figure 535; 11 May 2006 pond B7: 1128 rotifers /L ; remaining replicate mean X = 42 rotifers /L ). As stated previously, a single replicate for the UNP nutrient treatment large zooplankton assemblage was lost and a pseudovalue was generated for the missing Filinia spp. density (13 April 2006 UNP). Daphnids did not appear in substantial num bers until well into the 12week trial, typically mid trial (~ week six) or later (Figures 5 24, 528, 532, 536). Daphnid frequencies were compared among nutrient treatments and sampling dates within individual treatments (rep. meas. 2way ANOVA, P(nutr 0.05). Although pond nutrient treatment was not found to be a significant factor signifi cant

PAGE 313

313 differences between pond t reatments were detected using post hoc pairwise comparisons (Bonferroni posttests ; Motulsky pers comm.). Only three pairwise differences in daphnid density were found between pond nutrient treatments, all involving the INO t reatment for a single sampling date (8 June 2006, PRO and INO, UNP and INO, and CSM and INO); during the sampling date of 8 June 2006, INO treatment daphnid density was considerably lower than mean daphnid densities of the three other treatments ( Figure 536). Daphnid density differences among sampling dates occurred within each of the four treatments, with among sampling date differences typically occurr ing between early sampling periods when daphnids were absent and mid and late sampling periods when daphnids were present in substantial numbers ( Figures 524, 528, 532, 536). Daphnid frequencies in the PRO treatment varied significantly among sampling dates, with daphnids largely absent from ponds until midtrial, with significantly higher numbers in the latter half of the trial; interestingly, daphnids were present within the PRO treatment ponds sooner and in larger numbers than in the other three treatments, first appearing in the 18 May 2006 samples ( Figure 5 24), approximately two weeks prior to the remaining three treatments ( Figures 528, 5 32, 536). Daphnid densities in the UNP treatment differed significantly among sampling dates and followed a pattern similar to the PRO treatment (Figure 5 28). Daphnid densities in the CSM treatment did not significantly differ among sampling dates, probably due to the low daphnid density mean s measured during the trial and high variations observed during the last weeks of the trial ( Figure 532). INO daphnid frequencies also did not differ among weekly samp ling intervals, although means increased during the latter weeks of the study. INO daphnid densities followed a trajectory similar to the other treatments, with daphnids absent during the first half to two thirds of the trial and a steady increase thereaf ter until trial termination. Within

PAGE 314

314 the INO treatment, all pairwise differences in daphnid density involved a single sampling date (INO 8 June 2006) that was significantly higher than all other sampling dates ( Figure 536). Large Plankton Assemblage Simpl ified Morisitas and Percent Similarity Indices Similarity indices were calculated as deterministic values and using bootstrap Monte Carlo resampling estimates. D eterministic estimate were calculated using weekly large plankton assemblage %IRI values ( Fi gures 5 38 551), and bootstrap estimates were calculated using weekly large plankton assemblage taxa %N values ( Figures 552 573). Deterministic Morisitas Similarity and Percent Similarity Indices Deterministic simplified Morisitas index value (M SI) comparisons between treatments did not appear to dramatically differ between treatment pairs ( Figure 538). MSI values followed similar temporal patterns with high similarity among treatments at the start of the trial, likely due to the low large plan kton abundances and limited taxonomic range present following pond sterilization and cooler, shorter photoperiod days early in the study; MSI values dipped sharply and oscillated before steadily increasing to high similarity levels between treatments before trial termination. Of the six possible pairwise treatment comparisons of large plankton assemblage MSI similarities, the PRO and INO (Figure 542; exception 18 May 2006), and UNP and CSM ( Figure 5 42) treatments appeared to have the most similar large p lankton assemblages over the 12week course of the trial. Large plankton assemblages were examined for only ten of the twelve weeks of the trial due to the lack of organisms present during the first two weeks of the trial following pond sterilization and the colder, shorter photoperiod days associated with late winter/early spring. These factors likely reduced phytoplankton production to levels that only allowed negligible zooplankton production, and/or insufficient time had passed for phytoplankton and z ooplankton colonization and reproduction to occur within the ponds. Unfortunately, only five UNP large plankton assemblages were present within the dataset for

PAGE 315

315 MSI and PSI estimation on 13 April 2006 due to the loss of a replicate due to human error (pond B1). PRO and UNP large plankton assemblages had similar MSI values (MSI of the ten weeks that were examined ( Figure 5 39). Additionally, of the three weekly sampling periods that were dissimilar (MSI < 0.65), all occurred during the first two thirds of the 12week trial, and only two were for consecutive sampling periods (18 May and 25 May 2006). PRO and CSM large plankton assemblages had similar MSI values for eight of the ten weeks that were examined ( Figure 540). The two sampling pe riods when PRO and CSM treatment large plankton assemblages were not similar, also were nonconsecutive (27 April 2006 and 18 May 2006). PRO and INO large plankton assemblage MSI values were similar for nine of the ten weekly sampling dates examined ( Figur e 5 41). The single midtrial sampling date that the PRO and INO treatments large plankton assemblages were dissimilar, again was 18 May 2006. UNP and CSM large plankton assemblage MSI values were similar for all ten of the weekly sampling dates examine d ( Figure 542). Additionally, the UNP and CSM MSI trajectories only oscillated between 0.67 and 1.0 compared to the erratic MSI value oscillations occurring between the other pairwise treatment combinations (e.g., Figure 543, UNP and INO treatments). UNP and INO large plankton assemblage MSI values were similar for seven of the ten weekly sampling dates examined, although two large declines in MSI values occurred ( Figure 5 43). Of the three dissimilar sampling periods (27 April, 18 May and 25 May 2006) only two were consecutive and all three roughly occurred during the middle of the 12 week trial.

PAGE 316

316 CSM and INO large plankton assemblage MSI values were similar for seven of the ten weeks examined ( Figure 5 44), and none of the three dissimilar sampling da tes (27 April, 25 May, and 8 June 2006) were consecutive. A qualitative examination of the CSM and INO MSI value time series data indicates that these two treatments had the most dissimilar large plankton assemblages over the 12 week outdoor trial, as indicated by the large oscillations and three low MSI minima observed during the trial ( Figure 5 44, 27 April 2006 MSI = 0.24). Large MSI value oscillations occurred among five of the six pairwise treatment combinations, with the single exception of the UNP and CSM treatment comparison. However, MSI values were uniformly high ( Figures 538 543) during the last few weeks (1, 8, 15 June 2006) of the trial, with the exception of the CSM and INO treatment pair ( Figure 544). Percent similarity index (PSI) values exhibited less variation among treatment pair comparisons than their MSI counterparts ( Figures 538, 545). Of the six possible pairwise treatment comparisons, the PRO and INO, UNP and CSM, and UNP and INO appeared to display the greatest large plankt on assemblage similarities ( Figures 5 48 5 50). Surprisingly, only one sampling date and treatment pair (PRO and CSM treatments) comparison displayed large plankton assemblage taxa PSI dissimilarity (PSI < 0.6 or 60%) among the four pond nutrient trea tments and ten weekly samples that were examined and compared for the 12week trial (Figures 545, 547). PRO and UNP large plankton assemblage PSI values were significant (PSI ten of the sampling dates that were examined ( Figure 546), all P SI values oscillated between 61 and 91 percent similar ity No PSI value trends appeared to be occurring during the twelve weeks other than minor oscillations resulting from slight differences in PSI values among consecutive weekly sampling periods.

PAGE 317

317 The PR O and two fertilizer treatment comparisons (PRO and CSM, PRO and INO) PSI value trajectories were unremarkable (Figures 5 47 548). PRO and CSM large plankton assemblage PSI values were similar for all but one sampling period (27 April 2006, PSI = 55%), the remaining sampling periods did not differ dramatically ( Figure 547). PRO and INO large plankton assemblage PSI values were similar for all ten weekly sampling periods examined ( Figure 5 48). PRO and INO PSI values displayed gradual oscillations tha t were centered between 7080%, with values rising slightly toward the end of the trial. UNP and CSM l arge plankton assemblage PSI values also were similar for all ten of the weekly sampling periods ( Figure 5 49). Aside from a low initial PSI value at the start of the trial (13 April 2006, PSI = 62%), UNP and CSM PSI values were high (9095%) before declining slightly and stabilizing (~ 80%) during the last few weeks of the trial. A smooth PSI value oscillation occurred between the UNP and CSM treatment pair during the ten weeks that the large plankton assemblage taxa were examined ( Figure 549). UNP and INO large plankton assemblage PSI values were similar for all ten weekly sampling periods ( Figure 550). PSI values slowly increased to a maximum value mid trial (18 May 2006) before declining only slightly during the last half of the trial. L ittle variation occurred in UNP and INO treatment large plankton PSI values during the ten weekly sampling periods producing a smooth PSI time series trajectory ( F igure 550). CSM and INO large plankton taxa PSI values were similar among all ten weekly sampling periods ( Figure 5 51). Two PSI minima occurred during the trial (27 April and 8 June 2006) when PSI values dropped to the low 60s; however six of the ten sa mpling periods had PSI values well over 80%.

PAGE 318

318 Bootstrap Morisitas Similarity and Percent Similarity Indices Bootstrap resampling estimates of MSI and PSI were performed for taxon counts (%N) between different treatments for differ ent draw pool sizes (3, 4, 5, 6, 8, 15 and 20 replicates), and iteration rates (10, 50, 100, 500 and 1000 iterations). Draw pool sizes and iteration rates were varied to observe how these indices reacted to different resampling parameters (Figures 5 52A 555A ; Appendix A ). MSI and PSI values generally exhibited lower variation at higher draw and iteration rates (Figures 5 53 555, 568 573). Unfortunately, only five UNP treatment large plankton assemblages were present within the dataset for resampling estimation on 13 April 2006 because a replicate was lost due to human error (pond B1). Bootstrap generated large plankton community MSI and PSI mean s and 95% confidence interval estimates among pond nutrient treatment pairs and sampling dates are summarized in Table 5.2. Within pairwise treatment comparisons, differences in weekly large plankton assemblage MSI values (%N) occurred for all comparisons except between the CSM and INO treatments (Figures 5 56 561). MSI treatment pair values followed a similar trajectory pattern: a gradual increase from initial values, followed by a steep decline, followed by a gradual increase before declin ing to near constant values during the last few weeks of the 12week trial. The CSM and INO MSI values displayed a more gradual oscillation than those of their five treatment pair counterparts ( Figure 5 61). PRO and UNP treatment pair bootstrap large plankton assemblage MSI values differed for nine of the pairwise sampling date combinations (45 total comparisons) among the ten weekly sampling periods examined ( Figure 556; P < 0.05, Bonferroni post test). The majority of these differences were due to a single period early in the study (27 April 2006), which had a low MSI value (MSI = 0.09) between the two feed treatments. MSI values leveled off during the latter half of the 12week trial, possibly as a result of a more stable plankton community composition

PAGE 319

319 with greater abundance within feed treatment ponds. T axa were similar (%N MSI > 0.65) between the PRO and UNP treatments for six of the ten sampling dates ( Figure 556), and the majority of these similar a ssemblage sampling periods occurred during the latter two thirds of the trial. PRO and CSM large plankton assemblage bootstrap MSI values followed a time series trajectory similar to those of the PRO and UNP MSI values (Figures 5 56 557). Of the nine PRO and CSM MSI pairwise sampling date differences that occurred among the ten sampling dates examined, seven were due to a single date (27 April 2006, MSI = 0.06) with a noticeably lower MSI value (P < 0.05, Bonferroni post test). Again, MSI slowly incr eased from a minimal value early in the trial (27 April 2006), reached a maximum at mid trial, before declining slightly to a fairly constant value for the last few weeks. During only two of the ten weekly sampling periods examined were PRO and CSM treatm ent large plankton taxa significantly similar, although three additional sampling dates were extremely close (0.630.64) to the arbitrarily chosen MSI critical value (CH(critical) = 0.65). PRO and INO large plankton assemblage taxa time series bootstrap MSI values ( Figure 5 58) also followed a pattern similar to those of the three other PRO treatment pairwise comparisons ( Figure s 5 55 557). Of the eight pairwise differences in PRO and INO MSI values that occurred among the ten sampling dates (P< 0.05, Bonferroni post test), five occurred due to a single sampling date (27 April 2006). MSI values gradually increased from the 27 April 2006 minimum before reaching a maximum four weeks later and then slightly decreas ed to a plateau for the last few weeks ( Figure 5 58). Large plankton taxa of the PRO and INO treatments were found to be similar (MSI examined, all of which were in the last five weeks of the trial ( Figure 558).

PAGE 320

320 UNP and CSM large plankton asse mblage time series bootstrap MSI values ( Figure 559) followed a trajectory slightly different from the pairwise PRO treatment similarity comparisons ( Figures 5 56 558; PRO and UNP, PRO and CSM, PRO and INO). In contrast to the PRO treatments pairwise MSI comparisons, the UNP and CSM MSI minimum value occurred at 4 May 2006 rather than 27 April 2006, slowly rising to a maximum on 25 May 2006 before decreasing to a plateau for the last few weeks of the trial. Large plankton taxa of the UNP and CSM treatments were found to be similar for only three (two consecutive) of the ten sampling periods examined (P < 0.05, Bonferroni post test), although two sampling dates (1 June and 15 June 2006, MSI = 0.64) were near the critical value ( Figure 559). UNP and IN O large plankton assemblage time series bootstrap MSI value time series trajectory also resembled that of the UNP and CSM trajectory ( Figure 5 60) T he UNP and INO MSI minima occurred on 4 May 2006, before slowly rising to a plateau for the last four weeks. Of the seven pairwise differences that occurred among the ten sampling dates (P < 0.05, Bonferroni post test), all were due to a single sampling date (4 May 2006) that occurred fairly early in th e 12week trial. UNP and INO assemblages were found to be similar for seven of the ten sampling periods (MSI occurred in the last five weeks of the trial ( Figure 5 60). Only the CSM and INO large plankton assemblage time series boot strap MSI values did not differ (overlapping 95 % CI ) among the ten weekly sampling periods ( Figure 5 61; P > 0.05, Bonferroni post test). CSM and INO MSI values smoothly oscillated over the ten weeks for which large plankton assemblage taxa were examined Six of the ten weekly sampl es had similar large plankton assemblages (MSI One additional date had an MSI close to the critical value (8 June 2006, MSI = 0.64).

PAGE 321

321 D ifferences in weekly large plankton assemblage bootstrap PSI values occurred for all treatment pair combinations except the CSM and INO combination ( Figures 5 62 5 67). A ll other treatment combinations displayed similar bootstrap PSI temporal patterns: initial values were followed by a gradual incre ase, then by a sharp decrease that slowly increased to a maximal level before decreasing to a nearly constant level for the last few weeks. T he CSM and INO treatment pair displayed more gradual oscillations in PSI values than those of their five treatmen t pair counterparts. These sharp decreases in large plankton assemblage MSI and PSI values for 27 April and/or 4 May 2006, were most likely due to the application of diuron herbicide (Karmax; DuPont Chemical Corp.) on 19 April 2006 to control a filament ous algae outbreak in pond A7 (processed feed). All ponds were treated with diuron to control any potential filamentous algae that had not yet been visually detected and to eliminate potential experimental bias associated with differing replicate pond management practices. PRO and UNP large plankton assemblage bootstrap PSI values among the ten weekly sampling periods followed a trajectory similar to that estimated for their bootstrap MSI values ( Figures 5 56, 562). PSI minima early in the study gradually increased to a maximum four weeks later before declining to a stable plateau for the last three week s Of the nine pairwise differences in PSI observed during the 12week trial (P < 0.05, Bonferroni post test), six were due to a single sampling dat e with a low PSI value (27 April 2006). On only one of the ten sampling dates (25 May 2006) did the PSI value denote significant similarity (PSI between the PRO and UNP large plankton assemblage taxa ( Figure 562). This was in stark contrast to t he deterministic PRO and UNP treatment pair large plankton assemblage PSI time series values ( Figure 5 46)

PAGE 322

322 The PRO and CSM large plankton taxa bootstrap PSI value trajectory was similar to that observed for the MSI and PSI value trajectories of the PRO and UNP treatments ( Figures 5 56, 562 563). A PSI minimum early in the trial (27 April 2006), gradually increased before falling to a stable value for the last few weeks. Of the eleven pairwise sampling period differences in large plankton taxa PSI values for the PRO and CSM treatments (P < 0.05, Bonferroni post test), seven were due to a single date (27 April 2006). PRO and CSM large plankton assemblages were found to be similar (PSI > 0.60) for only a single date (25 May 2006) among the ten dates examined. The PRO and INO treatment pair large plankton assemblage bootstrap PSI value trajectory differed somewhat from that of the other pairwise PRO treatment comparisons ( Figures 562 564; PRO and UNP, PRO and CSM). The PRO and INO treatment pair PSI minimum observed early in the study (27 April 2006) gradually increased to a plateau earlier an d without the maximum peak and slight decline observed within the other PRO treatment PSI comparisons ( Figures 5 62 563). Of the 12 pairwise differences in PRO and INO PSI values observed among the ten sampling dates ( Figure 564; P < 0.05, Bonferroni post test), seven were due to a single low PSI value (27 April 2006) and the remaining five were due to a consecutive samp l e (4 May 2006) with a relatively low PSI value. Within the PRO and INO time series comparisons, only two nonconsecutive dates (25 May 2006 and 15 June 2006) were found to be similar (PSI Figure 564). UNP and CSM large plankton assemblage bootstrap PSI values followed a trajectory similar to those produced by the PRO pairwise treatment comparisons ( Figures 562 565; PRO and UNP, PRO and CSM, PRO and INO). A PSI minimu m slightly later than that encountered within the PRO treatment comparisons (5 May versus 27 April 2006) gradually increased to a

PAGE 323

323 maximum that sharply decreased before settling to a plateau for the last few weeks. UNP and CSM assemblages were dissimilar for most of the ten weeks (P < 0.05, Bonferroni post test) B ootstrap PSI values were similar (PSI es (18 May and 25 May 2006). UNP and INO larg e plankton taxa bootstrap PSI values also had their minimum value on 4 April 2006 ( Figure 566), but followed a PSI trajectory more similar to that of the PRO and INO treatment PSI values (Figure 564). PSI smoothly increased from a minimum value before reaching a plateau earlier than within the other treatment pair comparisons (Figures 5 62 563, 565). Within the UNP and INO comparisons, eight pairwise PSI value differences that occurred among the ten dates (P < 0.05, Bonferroni post test), were due to a single date (4 May 2006) which had a low PSI value estimate. UNP and INO assemblages were similar (PSI only two nonconsecutive sampl e s (25 May and 15 June 2006) of the ten sampling dates, although two dates (18 May and 1 June 2006) had PS I values near the critical value ( Figure 5 66). The CSM and INO large plankton taxa bootstrap PSI values displayed a smoothly oscillating trajectory similar to their MSI counterpart (Figures 561, 567). Again, similar to the CSM and INO MSI time series v alues, no significant differences in MSI occurred among the ten dates (P > 0.05, Bonferroni post test). On only two nonconsecutive dates were the CSM and INO assemblages similar ( Figure 5 67; 25 May and 15 June 2006). Additionally, the PSI values measured for these two samp les were only marginally higher than the arbitrarily chosen 60% critical value (63% PSI on 25 May 2006, and 60% PSI on 15 June 2006). Bootstrap PSI estimates also were performed with 20 resampled replicates drawn randomly with replace ment (from the actual six replicate pond sample pool) to determine the effects of increased draws per iteration ( with 1000 iterations per weekly PSI value estimate) would have

PAGE 324

324 upon the similarity index ( Figures 568 573). Overall large plankton assemblage PSI trajectory patterns remained the same among pairwise treatment combination s ( Figures 562 5 73), but the overall variation in mean values was much lower and greater curve smoothing occurred as draw numbers were increased (e.g. Figures 567, 573). Again, only five UNP treatment large plankton assemblages were present within the dataset for resampling estimation on 13 April 2006 because a replicate was lost due to human error (pond B1). Discussion Large Zooplankton Assemblage Production Volumes Only one large zooplankton volume (mm3/L of pond water) difference occurred among the four pond nutrient treatments and ten weekly pond water sampling dates ( CSM and INO 8 June 2006; Figure 51). This indicated that large zooplankton standing stocks did not differ among treatments, and that presumably, prey availabilities also did not differ for potential large zooplankton predators (e.g., swordtails) within the different treatment ponds. Although large zooplankton standing stocks did not differ among treatments (except CSM and INO treatments on 8 June 2006), large zooplankton production rates When volume comparisons were performed among sampling dates within treatments, differences in large zooplankton volume were found within all nutrient treatments except the CSM treatment ( Figure 5 2 55). Differences followed similar patterns in the PRO, UNP and INO treatments (Figures 5 2 53, 55). Low volumes / standing stocks were associated with the cooler, shorter photoperiod of late winter/early spring s amples which significantly differed from higher large zooplankton volumes / standing stocks that occurred during the warmer, longer may have differed among treatments if grazing rates differed sufficiently and occurred at high enough magnitudes among treatments to mask differential zoopl ankton production rates (trophic cascades) that may have occurred among treatment ponds ( Carpenter et al. 1985, 1987, Post and McQueen 1987, Cohen et al. 2002)

PAGE 325

325 photoperiod late spring s amples. Large zooplankton assemblage volume time series trajectories followed pond phytoplankton standing stocks as measured via chlorophyll [a] concentrations ( Chapter 4; Figures 448 452). Although large zooplankton volumes increased from early to late spring, volumes decreased prior to the onset of summer. This may have been due to a number of factors, such as temperatures above those optimal for phytoplankton and zooplankton production, algal photoinhibition from increased solar insolation, decreased rainfall, lower oxygen tension due to increased pond metabolism from increasing temperature and increased biomass/organic loading, or other environmental factors, acting either singly or in concert. Alternately, or in addition to the above factors, biological factors such as limited large zooplankton food availability due to increased competition, species succession (competitive exclusion) or increased zooplankton grazing by predators may also have been factors in reducing large zooplankton volume/biomass ( Boyd 1974, Pearl and Slobodkin 1976, Yusoff and McNabb 1989, Diana et al. 1991, KnudHansen et al. 1993, Kastner and Boyd 1996, Ludwig et al. 1998, Brown et al. 2000, Saito et al. 2001, Hall 2004 ). Time averaged large zooplankton volumes (mm3/L pond water) did not differ among nutrient treatments (Figure 56), which indicates that large zooplankton standing stocks, and their potential as a food source for fish did not differ among treatments. However, equal zooplankton standing stocks among treatments does not necessarily indic ate equal zooplankton production rates among treatments i.e., differential zooplanktivore grazing and/or zooplankton reproduction rates may have occurred among treatments. Interestingly, the pattern of small fish ( SL) production numbers at harve st (Chapter 3; Figure 3 10) somewhat resembled the time averaged large zooplankton standing stock volume trend s among the four applied nutrient treatments (Figure 56). Although large zooplankton standing stocks did not significantly differ

PAGE 326

326 among treatme nts, further research with more replicate ponds and/or larger water sample volumes may find a significant correlation between mean fish numbers/biomass at harvest and large zooplankton biomass. Large Zooplankton Assemblage Production Numbers Only one diffe rence in large zooplankton density ( plankton/ L ) was found among the four nutrient treatments and ten sampling periods (CSM on 8 June and INO on 8 June 2006) T his was the same pairwise treatment and sampling date difference found for large zooplankton vol umes ( Figure 57). Within treatment large zooplankton density comparisons among sampling dates found significant pairwise differences for the UNP, CSM, and INO treatments, but no differences among dates were found for the PRO treatment ( Figures 5 8 511). Similar to large zooplankton volume results, pairwise differences in large zooplankton densities among sampling dates within the UNP, CSM and INO pond treatments all involved low zooplankton densities that occurred early in the study, when pond biota were colonizing newly sterilized [hydrated lime application: Ca(OH)2] ponds during the colder, shorter photoperiod days (e.g., Figure 510; 13 April 2006) and for mid trial dates when large zooplankton standing stocks were significantly l arger (e.g., Figures 59 511; 11 May 2006). Pairwise differences among dates for the UNP, CSM and INO treatments all involved a single date with a high zooplankton density (11 May 2006), and earlier dates when large zooplankton densities were low d ue to low water temperature, short photoperiod, etc. As previously mentioned, diuron herbicide (Karmax; DuPont Chemical Corp.) was applied to all ponds on 19 April 2006 to control filamentous algae but did not appear to drastically reduce large zooplan kton densities, as the herbicide was applied early in the trial when standing stocks of planktonic algae and large zooplankton were relatively low ( Figures 51 55, 57 511). A slight decrease in phytoplankton standing stocks may have occurred

PAGE 327

327 immed iately following diuron application, but the relatively rapid increase in chlorophyll [a] and large zooplankton volumes and densities following herbicide application, and the short half life of diuron in aqueous solution (~ 20 hours; EPA 2003), suggests th at this decrease was short lived (Havens 1994, Perschbacher et al. 2002, Perschbacher and Ludwig 2004) Large zooplankton assemblage densities followed similar time trajectories in all four treatments (PRO less pronounced variation) : low numbers occurre d for the first few weeks, numbers then steadily rose to a maximum value before declining slightly near the end of the trial ( Figures 5 7 511). Significant differences in large plankton densities between sampling dates within treatments usually resulted from comparisons between earlier sampling dates when large plankton numbers were extremely low and higher plankton numbers during the middle and latter portions of the trial. Only the PRO treatment large zooplankton densities did not di ffer among sampling dates ( Figure 5 8). The PRO and UNP large zooplankton assemblage density trajectories were somewhat flatter and more constant th a n the inorganic fertilizer treatment, and may have been due to the regular (twice daily) and even applicat ion of feed nutrients in contrast to the large periodic increases of nutrients from weekly applications of the inorganic liquid fertilizer (Boyd 1979, 1995, Kastner and Boyd 1996). CSM treatment large plankton assemblage density was fairly constant among sample dates, with differences due to a single sampl e interval (1 May 2006) when high numbers of large plankton were observed ( Figure 5 10). Cottonseed meal application may have resulted in a slower, controlled release of nutrients within the CSM treatmen t ponds (Kastner and Boyd 1996, Tepe and Boyd 2003) Slower nutrient release and a potential as a directly ingestible food source for fish may make organic plant materials such as cottonseed meal a more attractive fertilizer option than inorganic liquid fertilizers. Slowly released nutrients likely allowed pond food webs to be sustained with a constant supply of

PAGE 328

328 nutrients and potentially under less ecosystem stress, by avoiding large oscillations in plankton numbers that could lead to plankton crashes and potential starvation. Sufficient food availability results in uninterrupted growth, higher reproduction rates, and enhanced immune system function for the target species ( Dussault and Kramer 1981, Diana et al. 1991, Hopkins and Unwin 1997, James and S ampath 1998, KnudHansen 1998, Ludwig et al. 1998, Kruger et al. 2001, Lochmann et al. 2001, Royes and Chapman 2003, Ingram 2009). When zooplankton are heavily preyed upon by fish, high zooplankton densities have been shown to accelerate swordtail growth and recrui tment rates ( Kruger et al. 2001, Tam a r u et al. 2001). Additionally, slow release fertilizers may improve water quality by preventing large spikes in pond ammonia and nitrite levels that can occur with the application of more volatile organic (e.g., animal manures) and inorganic fertilizers, high levels of which are stressors of fish, potentially resulting in lower growth and reproduction rates, increase d disease susceptibility, and increased mortality High pond biomass conditions have a greater potential for anoxic or hypoxic conditions due to a large biomass of respiring organisms that can consume available dissolved oxygen in excess of oxygen production during prolonged overcast conditions and/or anoxic bottom water displacement following heavy rains, or oxygen depletion from a high biomass of decaying organisms following a large algal die off all of which can create the potential for catastrophic production loss es ( Tepe and Boyd 2003). The use of c ontrol ponds with broodstock, but without int ensive nutrient addition, would have been useful in determining zooplankton volumes (standing stocks) and community composition in the presence of a presumed increase in predation pressure (no allochthonous nutrient inputs) from an omnivorous nondemersal visual predator fish species ( Dussault and Kramer 1981, Axelrod 1991, Jones et al. 1998, Tamaru et al. 2001) Additionally, c ontrol ponds

PAGE 329

329 without broodstock fish could have provided useful baseline information regarding large zooplankton volumes and com munity composition in the absence of fish predation. Unfortunately, neither no nutrient nor no broodstock control treatments were incorporated into th is studys experimental design Outdoor Pond Trial Major Large Zooplankton Groups The three major large zooplankton taxa which had the highest %IRI values over the 12 week trial were copepods, rotifers and daphnids ( Figures 513 520, 537). Of the three major large zooplankton taxa, two ( Diaptomous copepods and Filinia rotifers) followed similar population trajectories within the four pond nutrient treatments ( Figures 5 21, 525, 529, 533). Typically, initial zooplankton numbers were low as ponds were slowly colonized following pond draining and sterilization, gradually r ising to a maximum near mid trial, followed by a slight decrease during the latter half of the trial; this pattern was observed for both Diaptomous copepods and Filinia rotifer s ( Figures 5 21 523, 525 527, 529 531, 533 535). Daphnid ( Moina macrocopa) population numbers followed a different pattern, in which daphnids were completely absent from samples until near mid trial, when numbers would slowly increase until trial termination (Figures 524, 528, 532, 536). Among the major large zooplankton ta xa, copepods and rotifers appeared to be early colonizers, and displayed high reproductive potentials typical of an early succession or r selected colonizer species, whereas daphnids did not typically appear until mid trial, and once observed, their numbers slowly and steadily increased, appearing to display lower reproductive potential, possibly through superior competition, typical of a K selected climax species (Pianka 1970, Paine 1980, Post and McQueen 1987).

PAGE 330

330 Copepod Density for 10 Weekly Sampling Periods Copepod densities did not differ among nutrient treatments, but did differ between sampling dates within individual pond nutrient treatments ( Figures 522, 526, 530, 534). Significant differences in copepod densities between sampling dates withi n treatments were usually due to low copepod densities and low variation among replicate ponds at the beginning of the 12week trial and much higher densities and greater variation among replicates at mid and late trial. Copepods appeared to follow an ear ly colonizer or an r species reproductive strategy within the experimental ponds, and followed the same general population growth trajectory regardless of experimental treatment over the twelve week trial. Rotifer Density for 10 Weekly Sampling Periods Rotifer density trajectories were similar to that of copepods in the four pond nutrient treatments and density differences were not significant between treatments for any given sampling period. However, rotifer density differences between sampling periods were significant within nutrient treatments. The general pattern observed for Filinia rotifer densities was similar to that found for Diaptomous copepods: low numbers at the onset, were followed by a gradual increase to a maximum value approximately midw ay through the study, with a slight decline toward the end ( Figures 523, 527, 531, 535). E xtremely high rotifer count s for a single pond within the CSM treatment and a single pond within the INO treatment on the same sampling date (11 May 2006) produced high mean values and large 95% confidence intervals for the CSM and INO treatments for this date (Figures 5 31, 5 35). These high counts also were responsible for the significant pairwise differences in Filinia rotifer densities among sampling da tes within the CSM and INO treatments. A ll pairwise differences in rotifer density between dates involved the 11 May 2006 sample within the CSM and INO treatments.

PAGE 331

331 Daphnid Density for 10 Weekly Sampling Periods Daphnid densities differed among pond nutr ient treatments, but only for a single date ( 8 June 2006), which had relatively high densities for the PRO and INO treatments and relatively low densities for the CSM treatment (Figures 5 24, 532, 536). This single date also generated pairwise differen ces in daphnid density among sampling dates within treatments, all PRO and INO daphnid density differences among sampling periods involved this date ( Figures 524, 536). A different date (15 June 2006) with a high daphnid density, generated all of the p airwise differences among sampling dates detected within the UNP treatment ( Figure 5 28). Interestingly, CSM daphnid densities did not differ among dates, which may have been due to the slow release nature of the cottonseed meal as opposed to the nutrient spikes of the inorganic fertilizer. This possibly resulted in gradual Moina macrocopa population changes within the CSM ponds ( Figure 5 36), due to a constant phytoplankton standing stock prey pool for daphnids. Pond Large Zooplankton Assemblage %IRI Trends As previously mentioned, the three primary large zooplankton groups according to %IRI rankings were copepods, rotifers and daphnids; twentyseven other taxa were routinely found in the samples, but in lower numbers and/or in fewer samples. Copep ods were the most important item within all four pond nutrient treatments (Figures 513 520). Daphnids increased in importance late in the trial and until trial termination a few weeks later in mid June 2006. Interestingly, %IRI trajectories of the th ree major groups were basically similar within all four treatments, copepod numbers underwent gradual fluctuations but maintained their dominance, rotifer numbers underwent large increases and declines, and daphnids underwent a steady increase in abundance after being absent during the first half of the trial. Most interesting was an inverse %IRI trajectory relationship between the copepod and rotifer groups, evident in their

PAGE 332

332 %IRI graphs (Figures 5 17 520); plots of the two trajectories produced roughly inverse images of each other. Also, when rotifers were at a low %IRI, typically near the end of the trial, the daphnid group would appear and mirror the copepod groups %IRI trajectory. The inverse relationship between copepod and rotifer abundances was most likely due to copepod predation on rotifer standing stocks or possibly competition between these groups for a limited phytoplanktonic food source. In the case of the inverse %IRI trajectories for copepods and daphnids, the dominance of daphnids at t he end of the trial was most likely due to daphnids being superior competitors, typical of a K strategy life history species and not due to predation of daphnids upon copepods, based on their similar sizes and feeding habits ( Pianka 1970, Brett et al. 1994, Mallin and Paerl 1994, Burns and Schallenberg 2001a, 2001b). Densities of the three major large zooplankton groups (copepods, rotifers, daphnids) also were individually graphed for each pond nutrient treatment over the 12 weeks of the trial, but the inverse %IRI relationships observed between copepods relative to rotifers and daphnids were not as readily apparent for taxa densities ( Figures 5 21, 525, 529, 533). Arguably, copepod numbers and rotifer numbers were weakly inversely proportional to ea ch other for the PRO and UNP pond nutrient treatments. Daphnids which are considered superior competitors have been known to frequently out compete copepods, which are considered early colonizers (Paine 1980, Gilbert 1988, 1989, Larson et al. 2007). Sim ilar to large plankton assemblage taxa %N and %V estimates ( Appendix 5A), %IRI were calculated using both pooled and unpooled data. Pooled %IRI data displayed large plankton assemblage taxa trends within treatments as a whole ( Figures 513 516), wherea s unpooled %IRI data stressed mean trends occurring among the different large plankton assemblage taxa over time ( Figures 517 520) and individual variation among replicate ponds

PAGE 333

333 within treatments. Unpooled %IRI large plankton assemblage time series pl ots appear to be slightly less erratic than those of pooled %IRI plots (e.g., Figures 5 15, 5 19; CSM treatment), which was expected as these %IRI values were derived from averaged large plankton assemblage taxa %N and %V data. Deterministic Similarity Ind ices Deterministic simplified Morisitas similarity indices (utilizing pooled %IRI) between large plankton assemblage taxa of the four nutrient treatments oscillated erratically between treatment pairs, but generally followed similar trends among the six possible pairwise treatment comparisons ( Figure 5 38). Similarity between the large plankton assemblages of different treatments was high at the start of the trial, probably due to the general low diversity and lack of large zooplankton following recent pond sterilization, cold water temperatures, and short photoperiods associated with late winter/early spring. High water clarity (Secchi disk readings, pers. obs.) and low chlorophyll [a] concentrations (Chapter 4; Figures 448 451) at the beginning of t he trial, indicated that insufficient time had elapsed following pond sterilization, and/or the previously cited factors may have prevented large algal blooms within ponds. Low phytoplankton concentrations would provide insufficient food for m any herbivorous zooplankton, which also would retard herbivorous, carnivorous and omnivorous zooplankton numbers/biomass production. Additionally, colonization of these ponds from neighboring ponds via eggs/spores/cysts/juvenile / adults, attached to insects or water fowl, wind borne, or from resting egg/cysts within the pond sediments may not have hatched/emerged due to unsuitable environmental conditions (cold temperatures, short photoperiod) and/or lack of sufficient time following pond sterilization (pH shock) and subsequent refilling with ground and well water. Pond large plankton assemblages became increasingly diverse over time ( Appendix A; Figures 5A 1 5A 80). Among the majority of pairwise treatment combinations, after a slight

PAGE 334

334 increase in MSI values a sharp decrease occurred on one of two sampling dates (27 April or 4 May 2006), this marked decrease occurred between most treatments pairs ( Figures 5 56 560) but only a slight decrease occurred between the CSM and INO treatments ( Figure 5 61). This sharp decrease in MSI likely occurred due to diuron herbicide application on 19 April 2006, although similarity indices did not decrease uniformly among all pond treatment pairs compared. MSI quickly rebounded, then dropped slightly before reachi ng a plateau of uniformly high values (MSI Figure 5 38 5 44). Of the six possible pairwise treatment combinations, the UNP and CSM treatments were consistently the most similar, with their MSI ranging f rom 0.67 to 1.0, and with much lower fluctuations among sampling periods ( Figure 542). Interestingly, aside from a single sampling date with a low MSI value, the PRO and INO treatments had consistently similar large plankton assemblages (MSI > 0.65) among the ten sampling dates ( Figure 541). The MSI significance level of 0.65 was arbitrary chosen by previous researchers, and does not signify actual functional similarities between a ssemblages, especially when comparing studies and different ecosystems ( Zaret and Rand 1971, Cailliet and Barry 1978, Lindquist 1998). Additionally, the MSI was designed to compare resource utilization differences among predator groups, rather than plankton assemblage community differences among treatments ( Krebs 1998 ). De terministic MSI values between treatment pairs ( Figure 5 38) indicate that large plankton assemblage differences between treatments were more a function of time rather than applied nutrient type. The trajectory that large plankton assemblage similarities followed did not appear to be a function of nutrient type, i.e., the large plankton assemblages of the two commercial feeds did not display a greater similarity to each other than either did to the two

PAGE 335

335 fertilizer treatment, and vice versa. All large zooplankton assemblages appear to have converged over time, becoming more similar over the course of the 12week pond trial (Figures 5 38 544). All treatment pair combinations had significantly similar large plankton assemblage MSI values (MSI > 0.65) for the last two weeks of the trial (Figures 539 544), while for the second to last week of the trial, the CSM and INO treatment MSI w ere just significant at 0.65 ( Figure 5 44). Large zooplankton assemblage taxa deterministi c percent similarity index values (using pooled %IRI) between treatments did not follow any general trends ( Figure 545). PSI values were uniformly high throughout the 12week trial, and did not appear to be noticeably higher between any two treatment pai rs (Figures 5 46 551). Within all six possible treatment pair combinations, only a single large plankton assemblage treatment pair/sampling date dissimilarity occurred, which was between the PRO and CSM treatment pairs (27 April 2006, PSI = 55). The s harp decline observed in MSI observed for 27 April or 4 May 2006 (Figures 538 544), did not translate in to sharp declines in PSI for these dates, as declines were either much lower in magnitude (Figures 546 548, 551) or largely absent (Figures 5 49 550). Unexpectedly, PRO and UNP PSI value trajectories oscillated among sampling dates to the greatest degree among the six pairwise treatment combinations ( Figure 546). Due to the similarity of the two applied feed nutrients, the PRO and UNP time series PSI values were expected to be the most similar (highest PSI values) and least variable among sampling dates within the different treatment combinations. The UNP and CSM, and UNP and INO PSI trajectories appeared to display the smoothest transitio ns among sampling dates (Figures 5 49 550). This may indicate that unprocessed uncooked meal feed may have acted in a manner similar to the two fertilizer nutrients within the UNP pond ecosystem s, rather than the processed (cooked via heat

PAGE 336

336 extrusion) and reground PRO feed, which was designed to be directly ingested by the fish as a highly metabolizable, highquality nutrient source. PSI value trajectories did not appear to indicate increasing large plankton assemblage similarity over time as indicate d by some MSI trajectories (Figures 539 544). The paucity of dissimilar large plankton assemblage PSI sampling date treatment pair combinations, relative to the numerous dissimilar Morisitas index time series large plankton assemblage differences wit hin treatments tends to indicate that MSI is superior to the PSI in detecting large plankton assemblage taxa differences (higher index sensitivity). This also was supported by the working hypothesis that large plankton assemblage community differences bet ween treatments would tend to be present earlier in the study. At the start of the trial, large zooplankton community taxa succession s had not yet occurred and low phytoplankton standing stocks/productivity were well below carrying capacity At the onset of the study even small differences in large zooplankton assemblage composition would be amplified by the similarity indices, due to the low number of zooplankton taxa and their relative abundances (Krebs 1998). Large zooplankton taxa number s and abundances increased later in the study MSI values indicated that large plankton assemblages were more dissimilar earlier in the study, while PSI indicated that large plankton assemblages were similar throughout the study. Additionally, MSI may ha ve a greater ability than PSI to detect major disturbances to a ponds large plankton assemblage, such as when diuron was applied on 19 April 2006, which likely affected large plankton assemblages sampled on 20 and 27 April 2006 (Figures 5 38 544). In contrast to MSI, PSI did not noticeably change during this time (20 and 27 April 2006), with the possible exception of the PRO and CSM PSI values ( Figure 5 47).

PAGE 337

337 The number of similar (MSI pairwise t reatment comparisons among the ten weekly sampling periods differed depending upon whether MSI values were deterministically derived using pooled %IRI values (Figures 5 39 544), or obtained from bootstrap generated MSI values using taxon %N (Figures 5 56 561). Likewise, the number of similar (PSI treatment pairs, also differed greatly depending upon whether PSI was deterministically estimated based upon pooled %IRI (Figures 546 551) or via bootstrapping using taxon %N (Figures 5 62 567). The difference in large plankton assemblage similarity frequencies between treatment pairs ranged between 0 and 7 for MSI, and between 8 and 9 for PSI calculated deterministically using %IRI values or via bootstrap resampling methods using %N (Table 5.2). Closer examination of large plankton assemblage similarity frequencies among pairwise treatment combinations, indicate that MSI frequency differences appeared to be much more consistent and robust than PSI frequency differences (3.33,MSI; 8.33,PSI ) XX for both the deterministic and bootstrap m ethods e ven when considering that some of the discrepa ncies in the number of similar large plankton assemblages observed between the deterministic and bootstrap similarity estimates were due to the different plankton community metrics (%IRI versus %N) used to calculate each of the indices (MSI, PSI).

PAGE 338

338 Figure 51. Large plankton (> 200 m) assemblage mea n volume s for 10 weekly sampling periods for four pond nutrient treatments : processed feed (PRO) unprocessed feed (UNP) cotton seed meal (CSM) and inorganic liquid fertilizer (INO ) ; n = 6 ponds per treatment 0 1 2 3 4 5 6 7 8 9 10 mm3/ L UNP PRO INO CSM

PAGE 339

339 Figure 52. Processed feed (PRO) treatment mean l arge plankton (> 200 m) assemblage volumes (mm3/L 95% CI) for 10 weekly sampling periods, unshared letters denote significant differences between sampling periods (P < 0.05 Bonferroni post test); n = 6 replicate ponds. 0 4 8 12 16 20 24 mm3/ La b a b a ab ab ab ab ab

PAGE 340

340 Figure 53. Unprocessed feed (UNP) treatment l arge plankton (> 200 m) assemblage mean volume s (mm3/L 95% CI) for 10 weekly sampling periods, unshared letters denote statistical differences between sampling periods (P < 0.05, Bonferroni post test) ; n = 6 replicate ponds (* only five replicate samples available for this date) 0 4 8 12 16 20 24 mm3/ La c a bc a ab a abc abc abc

PAGE 341

341 Figure 54. Cottonseed meal (CSM) treatment large plankton (> 200 m) assemblage mean volume s (mm3/L 95% CI) for 10 weekly sampling periods unshared letters denote statistical differences between sampling periods (P < 0.05, Bonferroni post test); n = 6 replicate ponds. 0 4 8 12 16 20 24 mm3/ La a a a a a a a a a

PAGE 342

342 Figure 55. Inorganic fertilizer (INO) treatment l arge plankton (> 200 m) assemblage mean volume s (mm3/L 95 % CI) for 10 weekly sampling periods, unshared letters denote statistical differences between sampling periods (P < 0.05, Bonferroni post test); n = 6 replicate ponds. 0 4 8 12 16 20 24 mm3/ Lab a a a ab ab ab ab c bc

PAGE 343

343 Figure 56. Time averaged large plankton (> 200 m) assemblage mean volumes (mm3/L 95% CI) for four nutrient treatments ; unshared letters denote statistical differences between treatments (P < 0.05, Tukeys multiple comparison test); n = 6 ponds per treatment. 0 1 2 3 4 5 6 PRO UNP CSM INO mm3/La a a a

PAGE 344

344 Figure 57. Large plankton (> 200 m) assemblage mean densities ( number / L ) among 10 weekly sampling periods for four nutrient treatments : processed feed (PRO) unprocessed feed ( UNP ) cottonseed meal ( CSM ) and inorganic fertilizer ( INO) ; n = 6 ponds per treatment 0 200 400 600 800 1000 1200 Number / L CSM N INO N PRO N UNP N

PAGE 345

345 Figure 58. Processed feed (PRO) treatment large plankton (> 200 m) assemblage mean densities ( number / L 95% CI ) among 10 weekly sampling periods unshared letters denote statistical differences between sampling periods (P < 0.05, Bonferroni post test); n = 6 replicate ponds. 0 200 400 600 800 1000 1200 Number / La a a a a a a a a a

PAGE 346

346 Figure 59. Unprocessed feed (UNP) treatment large plankton (> 200 m) assemblage mean densities (number/L 95 % CI ) among 10 weekly sampling periods, unshared letters denote statistical differences between sampling periods (P < 0.05, Bonferroni post test); n = 6 replicate ponds only 5 replicate samples available for this sampling date 0 200 400 600 800 1000 1200 Number / La ab b b a ab ab ab ab ab *

PAGE 347

347 Figure 510. Cottonseed meal (CSM) treatment large plankton (> 200 m) assemblage mean densities (number/L 95 % CI ) among 10 weekly sampling periods, unshared letters denote statistical differences between sampling periods (P < 0.05, B onferroni post test); n = 6 replicate ponds. 0 200 400 600 800 1000 1200 Number / L a b a a ab a ab a a a

PAGE 348

348 Figure 511. Inorganic fertilizer (INO) treatment large plankton (> 200 m) assemblage mean densities (number/L 95 % CI ) among 10 weekly sampling periods, unshared letters denote statistical differences bet ween sampling periods (P < 0.05, Bonferroni post test); n = 6 replicate ponds. 0 200 400 600 800 1000 1200 Number / La a ab bc ab a a c a a

PAGE 349

349 Figure 512. Time averaged large plankton (> 200 m) assemblage mean densities ( number / L 95% CI) for four nutrient treatments : processed feed (PRO), unprocessed feed (UNP), cottonseed meal and inorganic fertilizer (INO), unshared letters denote statistical differences between treatments (P < 0.05, Tukeys multiple comparison test); n = 6 ponds per treatment 0 50 100 150 200 250 300 PRO UNP CSM INO Number / La a a a a

PAGE 350

350 Figure 513. Processed feed (PRO) treatment large (> 200 m) plankton assemblage pooled %IRI (index of relative importance) for 10 weekly sampling periods n = 6 replicate ponds. 0 10 20 30 40 50 60 70 80 90 100 Pooled %IRI COP TRIROT DAP TLR BIN CLR MITE LNE CST ELM VSUKE VLUKE NAUP PHR SEED SPBLL

PAGE 351

351 Figure 514. Unprocessed feed (UNP) treatment large (> 200 m) plankton assemblage %IRI (index of relative importance) for 10 weekly sampling periods ; data pooled from six replicate aquaculture ponds only five replicate samples available for this date 0 10 20 30 40 50 60 70 80 90 100 Pooled %IRI COP TRIROT DAP TLR BIN MITE LNE CST ELM VSUKE VLUKE PHR SPBLL *

PAGE 352

352 Figure 515. Cottonseed meal (CSM) treatment large (> 200 m) plankton assemblage %IRI (index of relative importance) for 10 weekly sampling periods, data pooled from six replicate ponds. 0 10 20 30 40 50 60 70 80 90 100 Pooled %IRI COP TRIROT DAP TLR BIN CLR MITE CST ELM VSUKE VLUKE BBROT SPBLL

PAGE 353

353 Figure 516. Inorganic fertilizer (INO) treatment large (> 200 m) plankton assemblage %IRI (index of relative importance) for 10 weekly sampling periods, data pooled from six replicate ponds. 0 10 20 30 40 50 60 70 80 90 100 Pooled %IRI COP TRIROT DAP TLR BIN MITE LNE OST CST ELM VSUKE VLUKE FLS DPP SEED BBROT NLR

PAGE 354

354 Fi gure 517. Processed feed (PRO) treatment large (> 200 m) plankton assemblage unpooled %IRI (index of relative importance) for 10 weekly sampling periods, n = 6 replicate ponds. 0 10 20 30 40 50 60 70 80 90 100 Unpooled %IRI COP TRIROT DAP TLR BIN CLR MITE LNE OST CST ELM VSUKE VLUKE NAUP PHR SEED BBROT SN SPBLL

PAGE 355

355 Figure 518. Unprocessed feed (UNP) treatment large (> 200 m) plankton assemblage unpooled %IRI (index of relative importance) for 10 weekly sampling periods; n = 6 replicate ponds, only five replicate samples for this sampling period 0 10 20 30 40 50 60 70 80 90 100 Unpooled %IRI COP TRIROT DAP TLR BIN CLR MITE LNE OST CST ELM VSUKE VLUKE PHR SEED SBT BBROT SPBLL

PAGE 356

356 Figure 519. Cottonseed meal (CSM) treatment large (> 200 m) plankton assemb lage unpooled %IRI (index of relative importance) for 10 weekly sampling periods, n = 6 replicate ponds. 0 10 20 30 40 50 60 70 80 90 100 Unpooled %IRI COP TRIROT DAP TLR BIN CLR MITE LNE OST CST ELM VSUKE VLUKE EPH NAUP NON LHC SEED SN SPBLL

PAGE 357

357 Figure 520 Inorganic fertilizer (INO) treatment large (> 200 m) plankton assemblage unpooled %IRI (index of relative importance) for 10 weekly s ampling periods, n = 6 replicate ponds. 0 10 20 30 40 50 60 70 80 90 100 Unpooled %IRI COP TRIROT DAP TLR BIN CLR MITE LNE SCLR OST CST ELM VSUKE VLUKE EPH NAUP ARTH IHT RR GSR PHR NON LHC FLS DPP SEED SBT BBROT SN SPBLL NLR

PAGE 358

358 Figure 521. Processed feed (PRO) treatment major (highest %IRI) large (> 200 m) plankton assemblage taxa [ Diaptomous copepods (LHC), Filinia rotifers (CLD), Moina macrocopa ( DAP; Daphnid)] densities for 10 weekly sampling periods; n = 6 replicate ponds 0 50 100 150 200 250 300 350 Number / L LHC CLD DAP

PAGE 359

359 Figure 522. Processed feed (PRO) treatment Diaptomous copepod (LHC) densities (number/L 95 % CI) for 10 weekly sampling periods, differing letters denote statistical differences between sampling periods (P < 0.05, Bonferroni post test); n = 6 replicate ponds 0 50 100 150 200 250 300 350 400 450 500 Number / Lab ab a abc ab b b ab ab ab

PAGE 360

360 Figure 523. Processed feed (PRO) treatment Filinia rotifer (CLD) densities (number/L 95 % CI) for 10 weekly sampling periods, unshared letters denote statistical differences between sampling p eriods (P < 0.05, Bonferroni post test); n = 6 replicate ponds. 0 100 200 300 400 500 600 700 800 900 1000 Number / La a a a a a a a a a

PAGE 361

361 Figure 524. Processed feed (PRO) treatment Moina macrocopa (DAP; Daphnid) densities (number/L 95 % CI) for 10 weekly sampling periods, unshared letters denote statistical differences bet ween sampling periods (P < 0.05, Bonferroni post test); n = 6 replicate ponds. 0 20 40 60 80 100 120 Number / La a a a a ab ab ab b ab

PAGE 362

362 Figure 525. Unprocessed feed ( UNP ) treatment major (highest %IRI) large plankton assemblage taxa [ Diaptomous copepods (LHC), Filinia rotifers (CLD), Moina macrocopa ( DAP; Daphnid)] densit ies for 10 weekly sampling periods; n = 6 replicate ponds 0 50 100 150 200 250 300 350 Number / L LHC CLD DAP

PAGE 363

363 Figure 526. Unprocessed feed ( UNP ) treatment Diaptomous copepod (LHC) densities (number/L 95% CI) for 10 weekly sampling periods unshared letters denote statistical diff erences between sampling periods (P < 0.05, Bonferroni post test); n = 6 replicate ponds, only five replicate samples available for this date 0 50 100 150 200 250 300 350 400 450 500 Number / La a aaa b b a a a

PAGE 364

364 Figure 527. Unprocessed feed ( UNP ) treatment Filinia rotifer (CLD) taxa densit ies ( number / L 95% CI) for 10 weekly sampling periods, unshared letters denote statistical differences between sampling periods (P < 0.05, Bonferroni post test); n = 6 replicate ponds only five replicate samples available for this sampling date. 0 100 200 300 400 500 600 700 800 900 1000 Number / La a a a a a a a a a *

PAGE 365

365 Figure 528. Unprocessed fe ed (UNP) treatment Moina macrocopa ( DAP; Daphnid) taxa densit ies ( number / L 95% CI) for 10 weekly sampling periods, unshared letters denote statistical differences between sampling periods (P < 0.05, Bonferroni post test); n = 6 ponds per replicate, onl y 5 replicate samples available for this sampling date 0 20 40 60 80 100 120 Number / La a a a a a a ab ab b *

PAGE 366

366 Figure 529. Cottonseed meal fertilizer (CSM) treatment major (highest %IRI) large (> 200 m) plankton assemblage taxa [ Diaptomous copepods (LHC) Filinia rotifers (CLD) Moina macrocopa ( D A P; Daphnid)] densities for 10 weekly sampling periods; n = 6 ponds per treatment. 0 50 100 150 200 250 300 350 Number / L LHC CLD DAP

PAGE 367

367 Figure 530. Cottonseed meal fertilizer (CSM) treatment Diaptomous copepod taxa densities (number/L 95 % CI) for 10 weekly sampling periods, unshared letters denote statist ical differences between sampling periods (P < 0.05, Bonferroni post test); n = 6 replicate ponds 0 50 100 150 200 250 300 350 400 450 500 Number / La a a a a a a a a a

PAGE 368

368 Figure 531. Cottonseed meal fertilizer (CSM) treatment Filinia rotifer taxa densities (number/L 95 % CI) for 10 weekly sampling periods, unshared letters denote statistical differences between sampling periods (P < 0.05, Bonferroni post test); n = 6 replicate ponds. 0 100 200 300 400 500 600 700 800 900 1000 Number / La a a ab b a a a a a

PAGE 369

369 Figure 532. Cottonseed meal fertilizer (CSM) treatment Moina macrocopa (DAP; Daphnid) taxa densities (number/L 95 % CI) for 10 weekly sampling periods, unshared letters denote statistical differences between sampling periods (P < 0.05, Bonferroni post test); n = 6 replicate ponds. 0 20 40 60 80 100 120 Number / La a a a a a a a a a

PAGE 370

370 Figure 533. Inorganic fertilizer (INO ) treatment major (highest %IRI large (> 200 m) plankton assemblage taxa [ Diaptomous copepods (LHC), Filinia rotifers (CLD), Moina macrocopa ( DAP; Daphnid)] densities for 10 weekly sampling periods; n = 6 replicate ponds 0 50 100 150 200 250 300 350 Number / L LHC CLD DAP

PAGE 371

371 Figure 534. Inorganic fertili zer (INO) treatment Diaptomous copepod (LHC) taxa densit ies ( number / L 95% CI) for 10 weekly sampling periods, unshared letters denote statistical differences between sampling periods (P < 0.05, Bonferroni post test) ; n = 6 replicate ponds 0 50 100 150 200 250 300 350 400 450 500 Number / La a a a a ab ab ab b a

PAGE 372

372 Figure 535. Inorganic fertilizer (INO) treatment Filinia rotifer ( CLD) taxa densities (number/ L 95% CI) for 10 weekly sampling periods, unshared letters denote statistical differences between treatments; n = 6 replicate ponds 0 100 200 300 400 500 600 700 800 900 1000 Number / La a a ab b a ab ab ab ab

PAGE 373

373 Figure 536. Inorganic fertilizer (INO) treatment Moina macrocopa (DAP; Daphnid) taxa densities (number/L 95% CI) for 10 weekly sampling periods, unshared letters denote statistical differences between sampling periods (P < 0.05, Bonferroni post test) ; n = 6 replicate ponds 0 20 40 60 80 100 120 140 Number / La a a a a a a a b a

PAGE 374

374 afas Figure 537 a d. Large (> 200 m) plankton assemblage taxa clockwise from upper right : (a) Calanoid copepod Diaptomous spp., (b) daphnid Moina macrocopa, (c) rotifers mainly Filinia spp. (d) aquatic insect (Ephemeroptera or Odonata); grid background (1.0 mm 1.0 mm). a b d c 1.0 mm

PAGE 375

375 Figure 537 e h. Large plankton assemblage taxa clockwise from upper right : (e) spherical rotifer, (f) rotifer Hexarthra spp., (g) rotifer Brachionus spp. (h) unknown ostracod; grid background (1.0 mm 1.0 mm). e f h g 1.0 mm

PAGE 376

376 Figure 538. L arge (> 200 m) plankton assemblage treatment pair deterministic MSI (simplified Morisitas similarity index) values generated from taxa pooled %IRI v alues for 10 weekly sampling periods. F our nutrient treatments : processed feed (PRO) unprocessed feed (UNP) cottonseed meal fertilizer (CSM) and inorganic fertilizer (INO ) 0 0 0 2 0 4 0 6 0 8 1 0 1 2 Similarity ( 0 1 ) PRO_UNP PRO_CSM PRO_INO UNP_CSM UNP_INO CSM_INO

PAGE 377

377 F igure 539. Processed feed (PRO) and unprocessed feed (PRO) treatment deterministic MSI values for 10 weekly sampling periods generated from large (> 200 m) plankton assemblage taxa %IRI values (critical value CH n = 6 replicate ponds per treatment only five UNP treatment replicate samples available f or this date 0 0 0 2 0 4 0 6 0 8 1 0 Similarity ( 0 1 )

PAGE 378

378 Figure 540. Processed feed (PRO) and cottonseed meal (CSM) fertilizer treatment deterministic MSI values for 10 weekly sampling periods generated from large (> 200 m) plankton assemblage taxa %IRI values (critical value CH ponds per treatment. 0 0 0 2 0 4 0 6 0 8 1 0 1 2 Similarity ( 0 1 )

PAGE 379

379 Figure 541. Processed feed (PRO) and inorganic (INO) fertilizer treatment deterministic MSI values for 10 weekly sampling periods generated from large (> 200 m) plankton assemblage taxa %IRI values (criti cal value CH per treatment. 0 0 0 2 0 4 0 6 0 8 1 0 Similarity ( 0 1 )

PAGE 380

380 Figure 542. Unprocessed feed (UNP) and cottonseed meal (CSM) fertilizer treatment deterministic MSI values for 10 weekly sampling periods generated from large (> 200 m) plankton assemblage taxa %IRI values (critical value CH replicate ponds per treatment, only five replicate samples available for this sampling date. 0 0 0 2 0 4 0 6 0 8 1 0 Similarity ( 0 1 )

PAGE 381

381 Figure 543. Unprocessed feed (UNP) and inorganic fertilizer (INO) treatment deterministic MSI values for 10 weekly sampling periods generated from large (> 200 m) plankton assemblage taxa %IRI values (critical value CH ponds per treatment, only five replicate samples available for this sampling date. 0 0 0 2 0 4 0 6 0 8 1 0 Similarity ( 0 1 )

PAGE 382

382 Figure 544. Cottonseed meal (CSM) and inorganic (INO) fertilizer treatment deterministic MSI values for 10 weekly sampling periods generated from large (> 200 m) plankton assemblage taxa %IRI values (critical value CH ponds per treatment. 0 0 0 2 0 4 0 6 0 8 1 0 Similarity ( 0 1 )

PAGE 383

3 83 Figure 545. Large (> 200 m) plankton assemblage treatment pair deterministic PSI (percent similarity index) values, generated fr om taxa pooled % IRI values for 10 weekly sampling periods F our nutrient treatments : processed feed (PRO) unprocessed feed (UNP) cottonseed meal fertilizer (CSM) and inorganic fertilizer (INO ). 0 20 40 60 80 100 PSI ( min%) PRO_UNP PRO_CSM PRO_INO UNP_CSM UNP_INO CSM_INO

PAGE 384

384 Figure 546. Processed (PRO) and unprocessed (UNP) feed treatment deterministic P SI (percent similarity index) values for 10 weekly sampling periods, generated from large (> 200 m) plankton assemblage taxa %IRI values, six replicate ponds per treatment (PSI(critical value) ; five UNP treatment replicate samples, see text. 0 20 40 60 80 100 PSI ( min%)*

PAGE 385

385 Figure 547. Processed feed (PRO) and cottonseed meal fertilizer (CSM) treatment deterministic PSI (percent similarity index) values for 10 weekly sampling periods, generated from large (> 200 m) plankton assemblage taxa %IRI values, six replicate ponds per treatment (PSI(critical value) 0 20 40 60 80 100 PSI ( min%)

PAGE 386

386 Figure 548. Processed feed (PRO) and inorganic f ertilizer (INO) treatment deterministic PSI (percent similarity index) values for 10 weekly sampling periods, generated from large (> 200 m) plankton assemblage taxa %IRI values (PSI(critical value) n = 6 ponds per treatment 0 20 40 60 80 100 PSI ( min%)

PAGE 387

387 Figure 549. Unprocessed feed (UNP) and cottonseed meal (CSM) fertilizer treatment deterministic P SI (percent similarity index) values for 10 weekly sampling periods, generated from large (> 200 m) plankton assemblage taxa %IRI values (critical value PSI n = 6 ponds per treatment. 0 20 40 60 80 100 PSI ( min%)

PAGE 388

388 Figure 550. Unprocessed feed (UNP) and inorganic fertilizer (INO) treatment deterministic PSI (Percent Similarity Index) values for 10 weekly sampling periods, generated from large (> 200 m) plankton assemblage taxa %IRI values (critical value PSI 0.60); n = 6 ponds per treatment only 5 UNP treatment replicate samples available for this date 0 20 40 60 80 100 PSI ( min%)*

PAGE 389

389 Figure 551. Cottonseed meal (CSM) and inorganic fertilizer (INO) treatment deterministic P SI (Percent Similarity Index) values fo r 10 weekly sampling periods, generated from large (> 200 m) plankton assemblage taxa %IRI values (critical value PSI 0.60); n = 6 ponds per treatment ). 0 20 40 60 80 100 PSI ( min%)

PAGE 390

390 Figure 552. Processed (PRO) and unprocessed (UNP) feed treatment bootstrap MSI (Simplified Morisitas Index) values (critical value CH generated from different numbers of randomly drawn replicates and iterations using large (> 200 m) plankton taxa %N values from a single sampling date (25 May 2006), n = 6 replicate ponds 0 0 0 1 0 2 0 3 0 4 0 5 0 6 0 7 0 8 0 9 1 0 0 5 10 15 20 25MSI ( 0 1 )Number of Draws 10 iterations 50 iterations 100 iterations 500 iterations 1000 iterations

PAGE 391

391 Figure 553. Scale adjusted processed (PRO) and unprocessed (UNP) feed treatment bootstrap MSI (simplified Morisitas Index) values (critical value CH generated from different numbers of randomly drawn replicates and iterations using large (> 200 m) plankton assemblage taxa %N values from a single sampling date (25 May 2006), n = 6 replicate ponds. 0 90 0 92 0 94 0 96 0 98 1 00 0 5 10 15 20 25MSI ( 0 1 )Number of Draws 10 iterations 50 iterations 100 iterations 500 iterations 1000 iterations

PAGE 392

392 Figure 554. Processed feed (PRO) and inorganic fertilizer (INO) treatment bootstrap PSI (Percent Similarity Index) values (critical value PSI 0) generated from differ ing numbers of randomly drawn replicates and iterations using large (> 200 m) plankton assemblage taxa %N values from a single sampling date (25 May 2006) ; n = 6 replicate ponds. 52 54 56 58 60 62 64 4 6 8 15 20PSI (%)Number of draws 10 iterations 100 iterations 1000 iterations

PAGE 393

393 Figure 555. Processed (PRO) and unprocessed (UNP) feed treatment bootstrap PSI (Percent Similarity Index) values (critical value PSI generated from differ ing numbers of randomly drawn replicates and iterations using large (> 200 m) plankton assemblage taxa %N values from a single sampling d ate (13 April 2006), n = 6 replicate ponds 31 32 33 34 35 36 37 38 0 4 8 12 16 20PSINumber of Draws 10 iterations 100 iterations 1000 iterations

PAGE 394

394 Figure 556. Processed (PRO) and unprocessed (UNP) feed treatment bootstrap MSI (simplified Morisitas similarity index ; critical value CH ) values ( 95% CI) generated from 1000 iterations per sampling period using l arge (> 200 m) plankton assemblage taxa %N values, unshared letters denote statistical differences between sampling periods (P < 0.05, Bonferroni post test); n = 6 ponds per treatment, only 5 UNP replicate samples available for this date. 0 0 0 2 0 4 0 6 0 8 1 0 1 2 MSI ( 0 1 )abc bcd a ab bcd cd d bcd bcd bcd *

PAGE 395

395 Figure 557. Processed feed (PRO) and cottonseed meal fertilizer (CSM) treatment bootstrap MSI (simplified Morisitas similarity index; critical value CH ) values ( 95% CI) generated from 1000 iterations per sampling period using large (> 200 m) plankton assemblage taxa %N values, unshared letters denote statistical differences between sampling periods (P < 0.05, Bonferroni post test); n = 6 ponds per treatment 0 0 0 2 0 4 0 6 0 8 1 0 1 2 MSI ( 0 1 )a bc bc ab abc c c bc bc bc

PAGE 396

396 Figure 558. Processed feed (PRO) and inorganic fertilizer (INO) treatment bootstrap MSI (simplified Morisitas similarity index ; critical value CH ) values ( 95% CI) generated from randomly drawn replicates at 1000 iterations per sampling period using large (> 200 m) plankton assemblage taxa %N values, unshared letters denote statistical differences between sampling periods (P < 0.05, Bonferroni post test); n = 6 ponds per treatment 0 0 0 2 0 4 0 6 0 8 1 0 1 2 MSI ( 0 1 ) abc abc a ab abc c c bc bc c

PAGE 397

397 Figure 559. Unprocessed feed (UNP) and cottonseed meal fertilizer (CSM) treatment bootstrap MSI (simplified Morisitas similarity index; critical value CH ) values ( 95% CI) generated from 1000 iterations per sampling period using large (> 200 m) plankton assemblage taxa %N values, unshared letters denote statistical differences between sampling dates (P < 0.05, Bonferroni post te st); n = 6 ponds per treatment, only 5 UNP replicate samples available for this date. 0 0 0 2 0 4 0 6 0 8 1 0 1 2 MSI ( 0 1 )a ab abc bc abc bc c abc abc abc *

PAGE 398

398 Figure 560. Unprocessed feed (UNP) and inorganic fertilizer (INO) treatment bootstrap MSI (simplified Morisitas similarity index ; critical value CH ) values ( 95% CI) generated from 1000 iterations per sampling period using large (> 200 m) plankton assemblage taxa %N values, unshared letters denote statistical differences between sampling periods; n = 6 ponds per t reatment; only 5 UNP replicate samples available for this sampling date. 0 0 0 2 0 4 0 6 0 8 1 0 1 2 MSI ( 0 1 )a ab b b b ab b b b b *

PAGE 399

399 Figure 561. Cottonseed meal (CSM) and inorganic fertilizer (INO) treatment bootstrap MSI (simplified Morisitas similarity index ; critical value CH ) values ( 95% CI) generated from randomly drawn replicates at 1000 iterations per sampling period using large (> 200 m) plankton assemblage taxa %N values, unshared letters denote statistical differences between sampling periods (P < 0.05, Bonferroni post test); n = 6 ponds per treatment 0 0 0 2 0 4 0 6 0 8 1 0 1 2 MSI ( 0 1 )a a a a a a a a a a

PAGE 400

400 Figure 562. Processed (PRO) and unprocessed (UNP) feed treatment bootstrap PSI (percent similarity index ; critical value PSI 0.60) values ( 95% CI) generated from 1000 iterations per sampling period using large (> 200 m) plankton assemblage taxa %N values, unshared letters denote statistical differences between sampling periods ( P < 0.05, Bonferroni post test ); n = 6 ponds per treatment, only 5 UNP replicate samples available for this sampling date. 0 20 40 60 80 100 PSI (%)a bc ab ab bc c bc bc ab bc

PAGE 401

401 Figure 563. Processed feed (PRO) and cottonseed meal fertilizer (CSM) treatment bootstrap PSI (percent similarity index ; critical value PSI ) values ( 95% CI) generated from randomly drawn replicates at 1000 iterations per sampling period using large (> 200 m) plankton assemblage taxa %N values, unshared letters deno te statistical differences between sampling periods (P < 0.05, Bonferroni post test); n = 6 ponds per treatment 0 20 40 60 80 100 PSI (%)a bc c ab abc c c bc bc c

PAGE 402

402 Figure 564. Processed feed (PRO) and inorganic fertilizer (INO) treatment bootstrap PSI (percent similarity index ; critical value PSI 0.60) values ( 95% CI) generated from randomly drawn replicates at 1000 iterations per sampling period using large (> 200 m) plankton assemblage taxa %N values, unshared letters denote statistical differences between sampling periods (P < 0.05, Bonferroni post test); n = 6 ponds per treatment 0 20 40 60 80 100 PSI (%)a abc bcd ab bc cd cd cd cd d

PAGE 403

403 Figure 565. Unprocessed feed (UNP) and cottonseed meal fertilizer (CSM) treatment bootstrap PSI (percent similarity index ; critical value PSI ) values ( 95% CI) generated from 1000 iterations per samplin g period using large (> 200 m) plankton assemblage taxa %N values, unshared letters denote statistical differences between sampling periods (P < 0.05, Bonferroni post test); n = 6 ponds per treatment, only 5 UNP replicate samples available for this samp ling date 0 20 40 60 80 100 PSI (%)a c ab abc bc ab bc abc abc abc

PAGE 404

404 Figure 566. Unprocessed feed ( UNP ) and inorganic fertilizer (INO) treatment bootstrap PSI (percent similarity index ; critical value PSI ) values ( 95% CI) generated from 1000 iterations per sampling period using large (> 200 m) plankton assemblage taxa %N values, unshared letters denote statistical differences between sampling periods (P < 0.05, Bonferroni post test); n = 6 ponds per treatment, only 5 UNP replicate samples available for this sampling period 0 20 40 60 80 100 PSI (%)a b b b ab b b b b b *

PAGE 405

405 Figure 567. Cottons eed meal (CSM) and inorganic fertilizer (INO) treatment bootstrap PSI (percent similarity index ; critical value PSI ) values ( 95% CI) generated from randomly drawn replicates at 1000 iterations per sampling period using large (> 200 m) plankton a ssemblage taxa %N values, unshared letters denote statistical differences between sampling periods (P < 0.05, Bonferroni post test); n = 6 ponds per treatment 0 20 40 60 80 100 PSI (%) a a a a a a a a a a

PAGE 406

406 Figure 568. Processed (P RO) and unprocessed (UNP) feed treatment bootstrap PSI (percent similarity index ; critical value PSI 0.60) values generated from 20 randomly drawn replicates at 1000 iterations per sampling period using large plankton assemblage taxa %N values; n = 6 ponds per treatment only 5 UNP replicate samples available for this sampling date.. 0 20 40 60 80 100 PSI (%)

PAGE 407

407 Figure 569. Processed feed (PRO) and cottonseed meal (CSM) fertilizer treatment bootstrap PSI (percent similarity index ; critical value PSI ) values generated from 20 randomly drawn replicates at 1000 iterations per sampling period using large (> 200 m) plankton assemblage taxa %N values ; n = 6 ponds per trea t ment 0 20 40 60 80 100 PSI (%)

PAGE 408

408 Figure 570. Processed feed (PRO) and inorganic fertilizer (INO) treatment bootstrap PSI (percent similarity index ; critical value PSI ) values generated from 20 randomly drawn replicates at 1000 iterations per sampling period using large (> 200 m) plankton assemblage taxa %N values ; n = 6 ponds per treatment. 0 20 40 60 80 100 PSI (%)

PAGE 409

409 Figure 571. Unprocessed feed (UNP) and cottonseed meal fertilizer (CSM) treatment bootstrap PSI (percent similarity index ; critical value PSI ) generated from 20 randomly drawn replicates at 1000 iterations per sampling period using large (> 200 m) pla nkton assemblage taxa %N values; n = 6 ponds per treatment, only 5 UNP repli cate samples available for this date 0 20 40 60 80 100 PSI (%)*

PAGE 410

410 Figure 572. Unprocessed feed (UNP) and inorganic fertilizer (INO) treatment bootstrap PSI (percent similarity index ; critical value PSI ) values generated from 20 randomly drawn replicates at 1000 iterations per sampling period using large (> 200 m) plankton assemblage taxa %N values ; n = 6 ponds per treatment, only 5 UNP replicate samples available for this date 0 20 40 60 80 100 PSI (%)*

PAGE 411

411 Figure 573. Cottonseed meal (CSM) and inorganic fertilizer (INO) treatment bootstrap PSI (percent similarity index ; critical value PSI ) values generated from 20 randomly drawn replicates at 1000 iterations per sampling period using large (> 200 m) plankton assemblage taxa %N values; n = 6 ponds per treatment 0 20 40 60 80 100 PSI (%)

PAGE 412

412 Table 5 1. 30 most common large zooplankton assemblage components and their approximate volumes calculated via photomicrographs. Taxa Length mm Width mm Height mm Diameter mm Volume mm 3 Sample Size Non Diaptomous Copepods 0.4078 0.1429 0.1429 1.5147 5 Diaptomous Copepods 0.7526 0.2486 0.0408 41 Tri Rotifers Filinia spp. 0.1276 0.0521 0.0003 12 Ball Rotifers 0.0351 13 Statoblasts 0.0029 6 Daphnids Moina macrocopa 0.0318 10 Insect Larvae 0.7893 0.1595 0.0152 4 Ostracods 0.5020 0.2969 0.3475 0.0307 9 Tooth Rotifers 0.2312 0.1902 0.0996 0.0025 4 Nematodes 1.5729 0.0290 0.0010 4 Water Mites 0.4102 0.2415 0.2293 0.0201 9 Nauplii like Rotifers 0.0970 0.0656 0.0001 12 True Nauplii 0.1224 0.0659 0.0005 4 Gastropods 1.4457 0.8920 0.8920 0.6695 3 Egg Mass 0.0171 5 Flower Stamens 0.9953 0.3734 0.2073 0.0392 3 Grape Rotifers 0.1513 0.0758 0.0758 0.0005 4 Spike Cup Rotifers 0.0872 0.0513 0.0513 0.0001 1 Ball Fungi 0.0004 2 Chironomids 1.7473 0.1370 0.0471 6 Ephemerata 0.5251 0.1443 0.0029 6 Hymenoptera 1.1285 0.2161 0.2161 0.0305 4 Coleoptera 0.6764 0.1878 0.1646 0.0155 9 Springtail 1.4444 0.3827 0.3827 0.1108 1 Dipteran 1.0590 0.4927 0.4789 0.1299 3 Insect heads 0.3119 0.1889 0.1487 0.0055 3 Pediastrum 0.1407 0.0230 0.0003 0.0003 2 Pope Rotifers 0.0699 0.0568 0.0437 0.0001 1 Grass Seeds 1.9562 0.6737 0.6890 0.4201 4 Misc Rotifers 0.1304 0.0978 0.0978 0.0007 1

PAGE 413

413 Table 5 2. Frequency of similar large plankton assemblages between treatment pairs according to the simplified Morisitas similarity index (MSI) and percent similarity index (PSI) using deterministic (%IRI) and resampling (%N) methods. D etermination M ethod PRO/UNP PRO/ CSM PRO/ INO UNP/ CSM UNP/ INO CSM/ INO MSI D eterministic 7 8 9 10 7 7 PSI D eterministic 10 9 10 10 10 10 MSI B ootstrap 6 2 4 3 7 6 PSI B ootstrap 1 1 1 2 2 2

PAGE 414

414 CHAPTER 6 24HOUR SWORDTAIL CAPTI VE FEEDING EXPERIMEN TS Swordtail (Xiphophorus hellerii ; Family Poeciliidae) dietary habits in semitropical aquaculture ponds were investigated to determine their dependence upon applied anthropogenic nutrients and live foods derived from the plankton community. Swordtail dietary habits w ere investigated using a powerful but indirect method stable isotope signature analysi s (Chapter 4) Carbon and nitrogen stable isotope signatures of swordtail s applied nutrients [two commercial feeds, cottonseed meal fertilizer, liquid inorganic fert ilizer (nitrogen only)], and three pond plankton size fractions (small, medium, large), were measured to determine the contribution of applied nutrients and pond plankton to swordtail diet s within four nutrient treatments Unfortunately, swordtail interna l anatomy ma de conventional gut content analysis impractical. Pharyngeal jaws within the swordtail buccal cavity, grind prey items into a fine, homogenous food bolus. An alternative to gut content analysis is the use of captive feeding trials, consist ing of timed incubation trials in which one or more fish are allowed to feed at will (ad libitum ) upon quantified prey assemblages or known nutrient sources. At trial termination, control treatments incubated without fish, are compared to those incubated with fish and differences in recovered plankton assemblage taxa composition and abundances are compared. A ny differences are attribute d to fish predation. T o compliment the findings from more indirect methods such as stable isotope dietary analyses (Chap ter 4), and potential swordtail prey large plankton assemblage community analyses (Chapter 5), in the absence of direct gut content analysis, 24 hour captive feeding trials were employed to investigate swordtail dietary habits/preferences ( Axelrod 1991, Ta maru et al. 2001, Kruger 2001).

PAGE 415

415 Although captive feeding trials were performed under artificial conditions at fish predator densities that were likely higher than those occurring within the ponds, this was a direct method of evaluating the swordtails abi lity to reduce zooplankton biomass and/or alter the large zooplankton community occurring within the controlled feeding trials, and by proxy, within actual aquaculture ponds. Methods 24Hour Captive Feeding/Plankton Grazing Experiment Twenty four hour captive feeding/plankton grazing experiments were performed (17 18 September 2007) to evaluate the effects of juvenile swordtails (nave to potential live plankton prey) from randomly selected ponds (orthogonally blocked within nutrient treatment/pond wa ter source ) on large zooplankton assemblages (size > 200 m) Three liters of pond water were collected from each of 16 ponds [four replicate ponds from each of four pond nutrient treatments: processed feed (PRO), unprocessed feed (UNP), cottonseed meal f ertilizer (CSM), and inorganic liquid fertilizer (INO)]. Each three liter integrated depth water sample (Chapter 4 Methods) was gently agitated to ensure sample homogeneity and split into three one liter samples, each assigned to one of three treatments: (1) a single one liter pond water sample that was immediately fixed using 10 ml of Lugols solution (APHA /AWWA 1989) following collection (Initial), (2) a flint glass jar containing one liter of pond water and no fry for 24 hrs (NF), and (3) a flint glass jar containing one liter of pond water and a single randomly selected juvenile swordtail for 24 hrs (Fish). Aeration was not provided, but all jars (16 NF: 16 Fish) were exposed to the atmosphere for gas exchange. Experimental jars with fish and control jars without fish were placed onto wooden beams (8.9 cm 8.9 cm 1.8 m) in random order and incubated for 24 hours within a 70% shade cloth covered greenhouse (Chapter 2, Figure 2 1) that was temperature controlled at the upper end (~ 30 C) by thermosta tically controlled ventilation

PAGE 416

416 fans. Natural fluctuations in temperature occur red due to ambient weather and solar insolation levels ( 1516 September 2007: 33.3 C daily high both dates; (NOAA) National Weather Service, Ruskin, Florida ). The purpose of the Initial control treatment was to provide a baseline for the large zooplankton community prior to the 24hour duration of the two experimental treatments (NF and Fish). It was assumed that the zooplankton assemblage of the Initial trea tment would have the greatest plankton community diversity and individual plankton taxa abundances as this plankton assemblage was not subject to potential swordtail predation and capt ivity stressors within a confined space ( i.e., potential suboptimal te mperature regime dissolved oxygen, light levels, etc.) The z ooplankton within the Initial treatment were immediately fixed following collection as Lugols preservative is also a nonvital stain. The purpose of the NF treatment was to provide an in situ baseline control for the Fish treatment. It was assumed that the NF would have a zooplankton community intermediate in complexity and abundance to that of the Initial and Fish treatments due to intra plankton assemblage predation and possible large zooplankton mortality due to stress from experimental conditions. The Fish group simulates potential swordtail predation pressure upon large plankton assemblage communities within the ponds. It was assumed that the Fish treatment plankton community com plexity and taxa abundances would be the lowest due to direct predation from the omnivorous, visually mediated swordtail predators. Randomly chosen juvenile swordtails ( X 16.8 mm SL 6.3 95 % CI ) taken from the greenhouse spawning and grow out systems (Chapter 2, Figures 21 22) were used for the captive feeding experiment. Swordtails had only received commercial flake fish food (Tetra

PAGE 417

417 Min tropical staple flake food) and were nave to the presence and consumption of live planktonic prey as the water source within these culture systems was from a well. J ars were covered with nylon mesh (3 mm square) held in place by rubber bands to permit air exchange with the water surface (~ 2 cm headspace), but not allow juvenile fish to escape C ontrol jars without fish also were covered. Jars were kept in a greenhouse under natural photoperiod and slightly higher than ambient temperatures (2630 C), away from human activity during the trial. No fry mortality occurred during the trial All fish were euthanized for later weight and length measurement N o identifiable prey items were found during gastrointestinal tract examination, but food bolus material was regularly encountered. At trial termination, jars were gently agitated prior to Lugols solution fixation (10 mls), t o collect as much particulate matter as possible. Initial, NF, and Fish large plankton assemblages were filtered, concentrated and sealed into small brown flint glass screw top vials containing 8 9 drops (~ 0.4 ml ) of Lugols preservative (APHA /AWWA 1989; Chapter 5 Methods). Large zooplankton assemblages were microscopically examined using a Sedgewick Rafter counting slide and individual plankton were enumerated by taxa (Chapter 5 Methods). Descriptive Paramete rs, Indices and Statistical Analyses Captive feeding trial large zooplankton assemblage volumes (plankton mm3/liter) and densities (total plankton #/liter; individual taxa #/liter) were calculated and compared among experimental treatments (Initial, NF, Fish) and pond nutrient treatments using 2way repeated measures ANOVA. Methods used for taxa volume estimation (Equation 5.1) and total large zooplankton assemblage volume and density estimates were identical to those used for the (March June 2006) pond trial large zooplankton assemblages (Chapter 5 Methods). Captive feeding trial large zooplankton assemblages were compared using the simplified Morisitas similarity index (MSI) and percent similarity index (PSI) calculated both

PAGE 418

418 deterministically using t axa %IRI and stochastically using taxa %N and bootstrap resampling methods using equations 5.25.6 (Chapter 5 Methods, Appendix 5B). Only large plankton assemblage taxa that had %IRI values > 0.5% were reported or used in the similarity indices analyses. Large zooplankton assemblage taxa communities were deemed similar if indices were greater than arbitrarily chosen threshold values (MSI All descriptive parameter and indice calculations (mean, error measures, %IR I, MSI, PSI) were made using either Microsoft Excel 2003 or the bootstrap Morisitas and percent similarity indices program written in BASIC (Appendix 5B). All statistical comparisons were made using the GraphPad Prism 4.0 statistical package (GraphPad Software, Inc. 2005 ). Results Large Plankton Taxa % Number, % Volume, % Frequency of Occurrence The 30 most important (Chapter 5; highest % IRI) large plankton assemblage taxa were identified and enumerated using the same criteria as within Chapter 5 (Methods, Table 5 1). Large plankton assemblage taxa percent number (%N), volume (%V) and frequency of occurrence (%FO) were measured for pooled and unpooled data (Figures 61 6 24). Qualitative differences among the Initial, NF, and Fish treatments within the four nutrient treatments, were the most conspicuous for daphnids (DAP Moina macrocopa), which were greatly reduced in the presence of swordtails. 24Hour Feeding Tri al Large Plankton Volume Differences Pond nutrient treatment (PRO, UNP, CSM and INO) and experimental treatment (Initial, NF and Fish) large zooplankton volumes (mm3/L) did not significantly differ for the nutrient treatment factor, but significantly dif fered for the experimental treatment factor (Figure 6 25; rep. meas. 2 way ANOVA, P(nutrient) = 0.375, F(nutrient) (2,3) = 0.05; P(experimental treatment) < 0.0001, F(experimental treatment) (2,2) = 0.05; P(interaction) = 0.480, F(int eraction) = 0.949, (2,6) =

PAGE 419

419 0.05). Three pairwise differences within the experimental treatment factor (I nitial, NF, and Fish) were found, PRO I nitial and PRO F ish, UNP I nitial and UNP F ish, and UNP NF and UNP F ish (P < 0.05, Bonferroni posttests). No pairwise differences were found among the nutrient treatments within levels of the experimental treatment (Initial, NF and Fish; P > 0.05, Bonferroni post test). Fish and NF treatments had numerically lower large zooplankton volumes than Initial treatmen ts, and Fish treatments had lower volumes than either the Initial or NF treatments ( Figure 6 25). However, statistical differences were only present within the PRO and UNP treatments due to high levels of variation among replicates. PRO large zooplankton experimental treatment volume averages ( Figure 626) were: Initial X = 11.7 mm3/L (0 to 27.3 mm3/L 95 % CI ), NF X = 8.4 mm3/L ( 0 to 17.8 mm3/L 95 % CI ), and Fish X = 2.9 mm3/L (1.6 to 4.3 mm3/L 95 % CI ). PRO large zooplankton volumes differ ed significantly among the experimental treatment groups (Figure 6 26) : Initial, NF and Fish Post hoc comparison tests found a single pairwise difference between the PRO Initial and Fish experim ental treatment groups (P < 0.05, Bonferroni). UNP large zooplankton volume averages ( Figure 6 27) were: Initial X = 12. 3 mm3/L (0 to 25.3 mm3/L 95 % CI ), NF X = 10.8 mm3/L (0 to 22.2 mm3/L 95 % CI ), and Fish X = 2.5 mm3/L (1.7 to 3.3 mm3/L 95 % CI ). UNP large zooplankton volumes significantly differed among the three treatment groups ( Figure 627 ). Post hoc comparison tests found two significant pairwise treatment differen ces: Initial and Fish and NF and Fish (P < 0.05, Bonferroni post test ). CSM large zooplankton volume averages ( Figure 6 28) were: Initial X = 5.3 mm3/L (0 to 11.2 mm3/L 95 % CI ), NF X = 4.2 mm3/L (0 to 8.6 mm3/L 95 % CI ), and Fish X = 2.1 mm3/L

PAGE 420

420 (0.5 to 3.6 mm3/L 95 % CI ). CSM large zooplankton volumes did not differ significantly among the three experimental treatment groups ( Figure 628; P > 0.05 Bonferroni posttests ). INO large zo oplankton volume averages were: Initial X = 10.2 mm3/L (5.3 to 15.0 mm3/L 95 % CI ), NF X = 8.4 mm3/L (6. 4 to 10.5 mm3/L 95 % CI ), and Fish X = 4.9 mm3/L (2.2 to 7.6 mm3/L 95 % CI ). INO pond treatment large zooplankton volumes did not differ among the three experimental treatment groups ( Figure 6 29; P > 0.05 Bonferroni posttests ). Due to pond nutrient treatment not being a significant factor in the repeated measures 2way ANOVA (P = 0.3753), the four pond nutrient treatments (PRO, UNP, CSM and INO) were pooled among the three experimental treatments ( I nitial, NF, and Fish) to increase the number of replicates from four to sixteen, increasing the statistical power available to analyz e differences among the three levels ( I nitial, NF, and Fish) of the captive feeding trial Following pooling, significant differences in total large zooplankton numbers were found among the three experimental treatment groups (rep. meas. 1way ANOVA, P < 0.0001, F = (2,2 ) = 0.05). Two pairwise differences were found: I nitial and Fish and NF and Fish N o significant difference was found between the NF and Fish groups ( Figure 6 30; P < 0.05, Tukeys Bonferroni post test ). 24Hour Feeding Trial Large Plankton Densities Large zooplankton densities (total numbers/L) did not significantly differ among pond nutrient treatments, but significantly differed for experimental treatments (Initial, NF, and Fish; rep. meas. 2 way ANOVA, P(nutrient) = 0.423, F(nutrient) (2,3) = 0.05; P(experimental treatment) < 0.0001, F(experimental treatment) (2,2) = 0.05; P(interaction) = 0.423, F(interaction) = 1.315, (2,6) = 0.05). Numerically, Initial treatment large zooplankton density averages were highest among the three experimental treatments, followed by NF treatment averages. Fish treatment averages had

PAGE 421

421 the lowest averages ( Figure 6 31). Two pairwise differences were found between the different experimental treatment groups within both the UNP and INO treatments: Initial and Fish, and NF and Fish (P < 0.05, Bonferroni post test). PRO large zooplankton density averages differed among experimental treatment groups ( Figure 6 32). PRO density averages were: Initial X = 354 zooplankton/L (0 to 833 zooplankton/L 95 % CI ), NF X = 2 60 zooplankton/L (49.0 to 470 zooplankton/L 95 % CI ), and Fish X = 85.0 zooplankton/L (46.0 to 124 zooplankton/L 95 % CI ). Within the PRO pond treatment, the Initial zooplankton density was significantly greater than both the NF and Fish treatments, which did not significantly differ from each other ( Figure 632; P < 0.05 Bonferroni post test ). UNP large zooplankton densities differed signi ficantly among the three treatments ( Figure 633) UNP density averages were: Initial X = 365 zooplankton/L ( 0 to 760 zooplankton/L 95 % CI ), NF X = 208 zooplankton/L ( 0 to 646 zooplankton/L 95 % CI ), and Fish X = 142 zooplankton/L ( 0 to 357 zooplankton/L 95 % CI ). Two pairwise differences in density were found among the three experimental treatments: I nitial and Fish, and NF and Fish N o significant difference was found bet ween the I nitial and NF treatments (Figure 633 ; P < 0.05, Bonferroni posttests ). CSM large zooplankton densities did not significantly differ among the three treatments ( Figure 6 34 P > 0.05, Bonferroni post test ). CSM density averages were: Initial X = 149 zooplankton/L ( 0 to 317 zooplankton/L 95 % CI ), NF X = 116 zooplankton/L (0 to 233 zooplankton/L 95 % CI ), and Fish X = 100 zooplankton/L ( 1.0 to 200 zooplankton/L 95 % CI ). INO pond treatment large zooplankton densities did not significantly differ among the three captive feeding trial experimental treatment groups (Figure 635; P > 0.05, Bonferroni post

PAGE 422

422 test ). INO pond treatment large zooplankton density averages among the experimental treatments were: Initial X = 289 zooplankton/L (179 to 399 zooplankton/L 95 % CI ), NF X = 241 zooplankton/L (165 to 316 zooplankton/L 95 % CI ), and Fish X = 114 zooplankton/L (7 4.9 to 153 zooplankton/L 95 % CI ). Similar to large zooplankton volume comparisons, the lack of differences in large zooplankton densities among pond nutrient treatments was used as justification to pool nutrient treatment groups within their respective experimental treatment groups (Initial, NF and Fish), increasing sample size from four to sixteen and i ncreasing the statistical power for experimental treatment comparisons ( Figure 636; rep. meas. 1way ANOVA, P(exp. Treat.) < 0.0001, F(exp. Treat.) = (2,2) = 0.05). Among the three experimental treatment groups, significant differences were found between the I nitial and Fish, and NF and Fish treatment groups, no differences were found between the I nitial and NF groups (P < 0.05, Bonf erroni post test ). 24Hour Feeding Trial Major Zooplankton Taxa Differences The three major zooplankton taxa [Chapter 5; Diaptomous copepods, Filinia rotifers, and daphnids ( Moina macrocopa) ] examined in outdoor pond trials conducted during spring 2006, were individually examined within the large zooplankton assemblages obtained during the 24hour captive feeding trials. Average densities (plankton/L) of the three major taxa generally decreased in a predictable and monotonous manner among the three exper imental treatments (in the following order: Initial, NF, and Fish) within each of the four pond nutrient treatment s ( Figures 6 37 655). Copepod densities did not significantly differ among nutrient treatments, but did significantly differ among 24hour experimental (Initial, NF, and Fish) treatments ( Figures 637 640; rep. meas. 2way ANOVA, P(nutrient) = 0.555, F(nutrient) = 0.730(2,3) = 0.05; P(exp. Treat.) =

PAGE 423

423 0.0005, F(exp. Treat.) (2,2) = 0.05; P(interaction) = 0.939, F(interaction) = 0.285(2,6) = 0.05). Post hoc pairwise comparisons of experimental treatment groups, within individual pond nutrient treatments found only one significant difference in copepod density, PRO I nitial and Fish experimental treatment groups ( Figure 6 37; P < 0.05, Bonferroni post test ). PRO copepod densities among the three experimental treatments (Initial, NF, and Fish) averaged: Initial X = 160 copepods/L (0 to 403 copepods/L 95 % CI ), NF X = 119 cope pods/L ( 0 to 316 copepods/L 95 % CI ), and Fish X = 63.0 copepods/L (29.4 to 96.6 copepods/L 95 % CI ). As previously mentioned, the PRO Initial treatment copepod density was significantly higher than that of the Fish treatment. UNP copepod densities (number/L) did not significantly differ among the feeding trial treatments (Figure 638; P > 0.05 Bonferroni posttests ). UNP copepod densities averaged: Initial X = 117 copepods/L (35.5 to 199 copepods/L 95 % CI ), NF X = 118 copepods/L (16.7 to 2 20 copepods/L 95 % CI ), and Fish X = 49.3 copepods/L (25.9 to 7 2.6 copepods/L 95 % CI ). CSM copepod densities did not significantly differ among the 24hour feeding trial experimental treatments ( Figure 6 39; P > 0.05 Bonferroni posttests ). CSM copepod densities averaged: Initial X = 104 copepods/L ( 1.9 to 206 copepods/L 95 % CI ), NF X = 8 4 copepods/L ( 0 to 182 copepods/L 95 % CI ), and Fish X = 47 copepods/L (12.0 to 82.5 copepods/L 95 % CI ). INO copepod densities did not significantly differ among the feeding trial treatments ( Figure 6 40; P > 0.05 Bonferroni posttests ). INO copepod densities averaged: Initial X = 168 copepods/L (112 to 2236 copepods/L 95 % CI ), NF X = 139 copepods/L (122 to 156 copepods/L 95 % CI ), and Fish X = 102 copepods/L (54 to 150 copepods/L 95 % CI ). Due to the nonsignificance of the pond nutrient factor during the 2way repeated measures ANOVA comparison of copepod density among pond nutrient treatments and experimental

PAGE 424

424 treatments (Initial, NF, and Fish), experimental trea tment groups were pooled across pond nutrient treatments to increase statistical power (unpooled n = 4, pooled n =16) and a 1way repeated measures ANOVA was performed on the pooled data (rep. meas. 1 way ANOVA, P = (2,2) = 0.05). Average copepod density of the pooled experimental treatment groups were: I nitial X (pooled) = 1 37 copepods/L ( 92.3 to 182 copepods/L 95 % CI ), NF X (pooled) = 115 copepods/L (77.1 to 153 copepods/L 95 % CI ), and F ish X (pooled) = 65.4 copepods/L (49.2 to 81.6 copepods/L 95 % CI ). Two significant p airwise differences were found among the three experimental treatments ( I nitial, NF, and Fish) : I nitial and Fish, and NF and Fish ( Figure 641; P < 0.05, Bonferroni post test). Filinia rotifer did not significantly differ among pond nutrients, but did significantly differ among experimental treatments ( Figure 6426.45; rep. meas. 2way ANOVA, P(nutrient) = 0.6958, F(nutrient) ( 2,3) = 0.05; P(exp. Treat.) = 0.0408, F(exp. Treat.) (2,2) = 0.05; P(interaction) = 0.2212, F(interaction) (2,6) = 0.05). Although the nutrient treatment factor F value was not significant, post hoc pairwise comparisons found a single pairwise difference among nutrient treatments (H. Motulsky pers. comm..) Initial PRO and Initial CSM (P < 0.05, Bonferroni posttests). Two pairwise differences were found between experimental groups within the PRO treatment: PRO initial and PRO NF and PRO initial and PRO Fish (P < 0.05, Bonferroni post test). PRO Filinia densities averaged ( Figure 6 42): Initial X = 18.3 rotifers/L (0.0 to 57.4 rotifers/L 95 % CI ), NF X = 2.3 rotifers/L ( 0.0 to 7.3 rotifers/L 95 % CI ), and Fish X = 0.8 rotifers/L (0.0 to 1.5 rotifers/L 95 % CI ). Filinia density within the Initial treatment was significantly higher than for the NF and Fish treatments ( P < 0.05, Bonferroni post tests).

PAGE 425

425 UNP Filinia densit ies did not significantly differ among the three 24hour feeding trial treatments (Figure 643; P > 0.05, Bonferroni post test ). UNP Filinia densities averaged: Initial X = 6.5 rotifers/L ( 0.0 to 21.1 rotifers/L 95 % CI ), NF X = 2.0 rotifers/L ( 0.0 to 6.5 rotifers/L 95 % CI ), and Fish X = 0.8 rotifers/L (0. 0 to 2.3 rotifers/L 95 % CI ). CSM Filinia densit ies did not differ significantly among the 24hour feeding trial treatments (rep. meas. 1 (2,2) = 0.05). CSM Filinia densities ( Figure 644) averaged: Initial X = 1.5 rotifers/L (0.0 to 3.6 rotifers/L 95 % CI ), NF (no fish 24 hours) X = 3.0 rotifers/L ( 0.0 to 8.0 rotifers/L 95 % CI ), and fish (single fish 24 hours) X = 2.5 rotifers/L (0.0 to 8.4 rotifers/L 95 % CI ). INO Filinia densit ies among the 24hour feeding trial treatments did not differ significantly (re p. meas. 1(2,2) = 0.05). INO Filinia densities ( Figure 645) averaged: Initial X = 5.5 rotifers/L ( 0.0 to 21.0 rotifers/L 95 % CI ), NF X = 3.0 rotifers/L ( 0.0 to 10.5 rot ifers/L 95 % CI ), and Fish X = 1.8 rotifers/L (0.0 to 7.3 rotifers/L 95 % CI ). Due to the nonsignificance of the nutrient factor in the 2 way repeated measures ANOVA comparison of Filinia density among pond nutrient and 24hour trial experimental treatments (Initial, NF, and Fish), experimental treatments were pooled across nutrient treatments. As a result of t his increased experimental treatment group (Initial, NF, and Fish) sample size [n = 4(unpooled) to n = 16(pooled)] and associated statistical power, significant differences in experimental treatment Filinia densities were detected (rep. meas. 1 (2,2) = 0.05). Average Filinia densit ies of the pooled treatments were: Initial X (pooled) = 7.9 rotifers/L (0.4 to 15.4 rotifers/L 95 % CI ), NF X (pooled) = 2.6 rotifers/L (0.9 to 4.3 rotifers/L 95 % CI ), and Fish X (pooled) = 1.4 rotifers/L (0.1 to 2.7 rotifers/L 95 % C I ). Although the pooled F

PAGE 426

42 6 statistic was significant, no pairwise differences among experimental treatments (Initial, NF, and Fish) were detected ( Figure 6 46; P < 0.05, Bonferroni post test). Daphnid densities did not significantly differ among nutrients but did differ among f eeding trial treatments (Figures 647 650; rep. meas. 2way ANOVA, P(nutrient) = 0.3380, F(nutrient) (2,3) = 0.05; P(exp. Treat.) = 0.0024, F(exp. Treat.) (2,2) = 0.05; P(interaction) = 0.4674, F(interaction) (2,6) = 0.05). Post hoc comparison tests did not find any differences in daphnid density between nutrient treatments within individual experimental treatments (Initial, NF, Fish), but a single difference in daphnid density betwee n experimental treatments within nutrient treatment was found, UNP Initial and UNP Fish (P< 0.05, Bonferroni posttests). PRO daphnid densit ies among the 24hour feeding trial treatments did not significantly differ ( Figure 647; Bonferroni posttests ). P RO daphnid densities averaged: Initial X = 151 daphnids/L ( 0.0 to 480 daphnids/L 95 % CI ), NF X = 92.8 daphnids/L ( 0.0 to 211 daphnids/L 95 % CI ), and Fish X = 5.5 daphnids/L (0.1 to 10.9 daphnids/L 95 % CI ). UNP daphnid densities averaged: Initial X = 218.5 daphnids/L ( 0 .0 to 601 daphnids/L 95 % CI ), NF X = 153 daphnids/L ( 0.0 to 414 daphnids/L 95 % CI ), and Fish X = 7.0 daphnids/L ( 0.0 to 22.4 daphnids/L 95 % CI ). Within the UNP treatment, Initial average daphnid densit y was significantly higher than th at within the Fish treatment (P < 0.05, Bonferroni post test). CSM daphnid densit ies differences did not signifi cantly differ among the 24 hour feeding trial treatments ( Figure 6 49; P > 0.05, Bonferroni posttests ). CSM daphnid densities averaged: Initial X = 29.5 daphnids/L ( 0.0 to 118 daphnids/L 95 % CI ), NF X = 14.0 daphnids/L ( 0.0 to 55.4 daphnids/L 95 % CI ), and Fish X = 2.0 daphnids/L ( 0.0 to 7.4 daphnids/L 95 % CI ). INO daphnid densit ies did not significantly differ among the 24 hour feeding trial treatments ( Figure 650; P > 0.05, Bonferroni post test ). INO daphnid densities averaged: Initial

PAGE 427

427 X = 85.8 daphnids/L ( 0.0 to 1 75 daphnids/L 95 % CI ), NF X = 73.0 daphnids/L ( 33.2 to 113 daphnids/L 95 % CI ), and Fish X = 8.0 daphnids/L ( 0.0 to 18.7 daphnids/L 95 % CI ). Due to the nonsignificance of the pond nutrient factor, experimental group treatments were pooled across pond nutrient treatment s (unpooled n = 4, pooled n =16) to increase statistical power. This resulted in significant differences in daphnid density among experimental treatments ( Figure 651; rep. meas. 1 way ANOVA, P = 0.00 27, F = 7.276(2,2) = 0.05). Average pooled daphnid densities were: Initial X (pooled) = 1 17 daphnids/L ( 29.2 to 206 daphnids/L 95 % CI ), NF X (pooled) = 8 0.7 daphnids/L ( 28.3 to 13 3 daphnids/L 95 % CI ), and Fish X (pooled) = 5. 3 daphnids/L (2.0 to 8. 7 daphnids/L 95 % CI ). Two pairwise differences in daphnid density were found, Initial and Fish, and NF and Fish ( Figure 6 51; P < 0.05, Bonferroni post test). 24Hour Feeding Trial Major Taxa %IRI Out of the thirty taxa commonly found (%IRI > 0.5%) in the large zooplankton assemblages from the (spring 2006) 12week outdoor pond trials (Chapter 5), only two (Copepod Diaptomous spp. and Daphnid Moina macrocopa; Figure s 537a b) were commonly found within the one liter samples recovered from the 24 hour captive feeding trials. Large plankton assemblage taxa percent number, percent volume, and percent frequency of occurrence graph taxa abbreviations ( Figures 6 1 6 24) are identical to those listed in Appendix A. Among three of the four pond nutrient tre atments, Diaptomous copepods were the most important large plankton community component and potential swordtail prey item group within the Initial 24 hour captive feeding experimental treatment T he single exception was the UNP treatment, which had Moina as the primary taxa ( Figure 656). Moina was the second most important large plankton assemblage component and potential swordtail prey item within the

PAGE 428

428 nonUNP treatments ( Figure 656). NonUNP Initial experimental treatment unpooled %IRI values were sim ilar to pooled %IRI values (Figures 656, 659) except for the UNP pond treatment, which had nearly equal Diaptomous copepod and Moina pooled %IRI values ( Figure 659). The NF treatment large plankton community taxa pattern was identical to that observed within the Initial experimental treatment among the four pond nutrient treatments (Figure 657). Within the large zooplankton assemblage of the NF treatment, Diaptomous copepods were the most important group, followed by Moina daphnids in the nonUNP pond nutrient treatment W ithin the UNP NF experimental treatment group, Moina was again the most important taxon, followed by Diaptomous copepods. Unpooled NF large zooplankton assemblage %IRI values were similar to pooled %IRI values ( Figures 657, 660). The Fish experimental treatment large plankton assemblage community %IRI values were markedly different from either the Initial or NF control treatments (Figure 6 58). Copepod relative %IRI, %N and %V values were much greate r within the Fish treatment compared to within the Initial and NF treatments (Figures 61 624), although actual copepod numbers were dramatically lower (Figures 6 37 6 41). Conversely, daphnid relative %IRI, %N and %V values were markedly lower within the Fish treatment compared to within the Initial and NF treatments (Figures 61 624). Fish treatment unpooled %IRI values did not differ greatly from those of pooled %IRI values, although unpooled Pediastrum algae %IRI values were slightly greater than their pooled %IRI value counterparts ( Figures 6 54, 657). 24Hour Feeding Trial Similarity Indices Large plankton assemblage community similarities among the three 24 hour captive feeding trial treatments were estimated using deterministically derived s implified Morisitas and percent similarity indices (MSI) values using large plankton assemblage taxa %IRI values

PAGE 429

429 (Chapter 5 Methods). Among the four pond nutrient treatments, the 24 hour captive feeding trial large plankton community MSI values were sign ificantly similar (MSI treatment comparisons ( Figure 662; Initial and NF, Initial and Fish, NF and Fish) a lthough the Initial and NF treatment large plankton communities consistently displayed greater similarity to each other (MS I ~ 1.0), than either treatment displayed to the Fish treatment (MSI range: 0.78 0.98), which had lower MSI values as well as greater variation among pond nutrient treatments. UNP Initial and Fish, and NF and Fish pairwise large plankton community simila rities were numerically lower than those of the other three pond nutrient treatments ( Figure 662). Among the 24hour feeding trial treatments, large plankton assemblage PSI values displayed a pattern similar to those of the simplified MSI, except for the PRO pond nutrient treatment (Figure 663). The PRO large plankton community Initial and NF PSI value was markedly lower than those of the other three treatments, although all four pond nutrient treatment Initial and NF PSI values were considered similar (PSI generally lower than their Initial and NF counterparts for all pond nutrient treatments except for the PRO treatment. Similarly, NF and Fish PSI values were generally lower than their Initial and Fish counter parts, except for the UNP treatment. Bootstrap Large Plankton Assemblage Similarity Indices Two way repeated measures ANOVA large plankton assemblage community comparisons, using bootstrapgenerated MSI estimates (using %N) did not find significant diffe rences among the three experimental treatments (Initial, NF, Fish) within given pond nutrient treatments ( Figures 6 64 667), but did find significant differences among the four pond nutrient treatments within experimental treatments (rep. meas. 2 way ANOVA, P(nutrient) = 0.0041, F(nutrient) (2,3 ) = 0.05, P(exp. treat.) = 0.1786, F(exp. treat) (2,2) = 0.05, P(interaction) = 0.3124, F(interaction) (2,6 ) = 0.05).

PAGE 430

430 Numerically, INO Initial and NF large plan kton community similarities (MSI values) were significantly greater than those of the PRO and UNP treatments (P < 0.05, Bonferroni post test). Although large plankton assemblage MSI values did not significantly differ among the three possible pairwise ex perimental treatment combinations (Initial and NF, Initial and Fish, NF and Fish) within any of the four pond nutrient treatments, Initial and NF MSI values were numerically greater than the Initial and Fish, and NF and Fish MSI values for the UNP, CSM and INO treatments (Figures 6 64 667). S uggest ing that the large plankton assemblages of the two control treatments that had not been subjected to fish predation (Initial, NF), had greater large plankton community similarities, at least on a taxa % numbe r basis. However these MSI differences were not statistically significant among the three experimental treatments within each of the four pond nutrient treatments. Bootstrap large plankton assemblage MSI values between the Initial and NF experimental treatments were: PRO X = 0.55 ( 0.16 to 0.95 95 % CI ), UNP X = 0.73 ( 0.36 to 1.11 95 % CI ), CSM X = 0.90 ( 0.7 6 to 1.04 95 % CI ), INO X = 0.92 ( 0.81 to 1.03 95 % CI ) Bootstrap large plankton assemblage MSI values between the Initial and Fish experimental treatments were: PRO X = 0.78 ( 0.42 to 1.14 95 % CI ), UNP X = 0.52 ( 0.05 to 0.99 95 % CI ), CSM X = 0. 69 ( 0.34 to 1 .03 95 % CI ), INO X = 0.88 ( 0.74 to 1.03 95 % CI ) Bootstrap large plankton assemblage MSI values between the NF and Fish experimental treatments were: PRO X = 0.48 ( 0.33 to 0.63 95 % CI ), UNP X = 0.58 ( 0.12 to 1.03 95 % CI ), CSM X = 0.68 ( 0.35 to 1.02 95 % CI ), INO X = 0.83 ( 0.70 to 0.97 95 % CI ) According to the arbitrarily chosen similarity threshold value (MSI Fish, and UNP NF and Fish large plankton communities were no longer similar ( Cailliet and Barry 1978, Lindquist 1998).

PAGE 431

431 Bootstrap generated l arge plankton assemblage PSI value s did not differ among the three experimental treatments within a ny given pond nutrient treatment or among the four pond nutrient treatments within a ny given experimental treatment ( Figures 668 671; rep. meas. 2way ANOVA, P(nutrient) = 0.2348, F(nutrient) (2,3 ) = 0.05, P(exp. treat.) = 0.3540, F(exp. treat) = (2,2) = 0.05, P(interaction) = 0.3540, F(interaction) (2,6 ) = 0.05). T he Initial and NF similarity index (PSI) values were higher than those of the Initial and Fish, and NF and Fish PSI values for the same three pond nutrient treatments as the MSI value comparisons (UNP, CSM and INO). T his indicates that fish presence may h ave reduced large plankton taxa abundance and changed the large plankton community structure to the extent that initially similar large plankton communities were less similar (Initial, NF, Fish). Initial and NF treatment large plankton assemblage bootstra p PSI values were: PRO X = 45.0 ( 13.6 to 76. 4 95 % CI ), UNP X = 60.5 ( 24.8 to 96.2 95 % CI ), CSM X = 71.4 ( 51. 0 to 91.8 95 % CI ), INO X = 75.7 ( 61.2 to 90.2 95 % CI ) Initial and Fish treatment large plankton assemblage bootstrap PSI values were: PRO X = 60.0 ( 32.1 to 88.0 95 % CI ), UNP X = 43.7 ( 12.3 to 75.1 95 % CI ), CSM X = 53.4 ( 22.6 to 84.3 95 % CI ), INO X = 68.8 ( 52.4 to 85.2 95 % CI ) NF and Fish treatment large plankton assemblage bootstrap PSI values were: PRO X = 35.2 ( 25.9 to 44.6 95 % CI ), UNP X = 47.0 ( 14.7 to 79.4 95 % CI ), CSM X = 52.8 ( 23. 8 to 81.9 95 % CI ), INO X = 60.9 ( 46.2 to 75.6 95 % CI ) According to the arbitrarily chosen PSI threshold critical value (PSI hour trial duration without fish predation (NF), and 24hour trial duration with fish predation (Fish), large zooplankton assemblages were no longer similar to the Initial treatment within the UNP and CSM (Initial and NF), and PRO, UNP and CSM ( Initial and Fish) treatments.

PAGE 432

432 T o determine if increasing the number of pond replicates would result in greater sensitivity to differences ( increased statistical power) among the 24hour treatments (Initial, NF, Fish), replicate large plankton assemblages from the f our pond nutrient treatments were pooled among experimental treatments, increasing sample size from four to sixteen replicates for each treatment. Pooled bootstrap PSI values among the feeding trial treatment s, did not significantly differ ( Figure 673). T he same treatment pair comparison PSI value rankings (Initial and NF highest, NF and Fish lowest) occurred for both the pooled and unpooled treatment (nonPRO) data. Pooled average bootstrap PSI values between feeding trial treatment pair values: I nitial vs. NF PSI X = 59.4 ( 48.8 to 70.0 95 % CI ), Initial vs. Fish PSI X = 56.2 ( 44.6 to 67.8 95 % CI ), NF vs. Fish PSI X = 50.2 ( 39.5 to 60.8 95 % CI ). All three pooled bootstrap large zooplankton community PSI similarity comparisons (Initial and NF, Initial and Fish, NF and Fish) were not similar according to the arbitrarily chosen PSI critical value (PSI Due to the nature of resampling statistical techniques, it was no t deemed necessary to perform arcsine square root transformations upon raw percentage data for the PSI comparisons due to the normal distributions that typically result from large numbers of sample iterations (1000 iterations; Simon 1997). Discussion Lar ge Zooplankton Assemblages The majority of taxa and those with the greatest importance (highest %IRI) within the large plankton community, were zooplankton, with minor exceptions (e.g. Pediastrum spp. phytoplankton, Division Chlorophyta green algae). Because the majority of large plankton assemblage taxa were zooplankton, large plankton assemblages are frequently referred to as large zooplankton assemblages within this document

PAGE 433

433 24Hour Captive Feeding Trials Two way repeated measures ANOVA found that large zooplankton assemblage volumes did not differ among pond nutrient treatments within the captive feeding trial experimental treatments, but that large zooplankton volumes did differ among captive feeding trial treatments within given pond nutrient treatments (PRO Initial and Fish, UNP Initial and Fish, and UNP NF and Fish). These results indicate that feeding trial treatment was a factor in determining large zooplankton volume within the PRO and UNP treatments and that both time and/or fish pres ence effected large zooplankton volumes after 24 hours ( Figures 626 627). Fish are likely consuming large zooplankton, at least within the UNP treatment ( Figure 6 27). Fish also may be consuming large zooplankton within the three remaining nonUNP t reatments, as average volumes were numerically lower within the Fish treatment relative to their Initial and NF treatment counterparts within all pond nutrient treatments (Figures 626 630). However, Fish treatment plankton volumes were not statistical ly lower than the Initial and NF feeding trial treatments within the CSM and INO pond nutrient treatments. However, the supposition that fish were grazing upon large zooplankton is reasonable given the limitations of the experimental design utilized for t he 24 hour captive feeding trial, and the fact that direct feeding was not observed due to the desire to avoid observer effects, and that gut content analyses on fish of this size and species had previously proven fruitless. Because pond nutrient treatment was not found to be a significant factor, nutrient treatments were pooled within the three experimental treatments (Initial, NF, and Fish). This increased the sample size from four to sixteen, and increased the statistical power of the analysis. P ooling seemed reasonable in this instance as fish were not obtained from within the ponds that were used for plankton sample collection (feeding history), and the general qualitative similarity of large plankton communities among pond treatments ( Motulsky 1995) Large plankton

PAGE 434

434 communities from each of the four pond nutrient treatments appeared to be qualitatively and quantitatively similar, except for the UNP large plankton community which had higher %IRI, % number and % volume f or daphnids than for copepods ( Figures 61, 64, 67, 610, 652). Pool ing nutrient treatments resulted in two pairwise experimental treatment differences (Initial v. Fish, and NF v. Fish) T trial duration no longer had a n effect upon large zooplankton volume as the I nitial and NF experimental treatment volumes no longer differ ed ( Figure 630) These results indicate that fish presen ce in the experimental treatments w as the only major factor in reducing large zooplankton volume among pooled samples and that fish predation was likely the cause of the observed decrease ( Figure 630). Although time also was a factor in reducing large plankton community volumes, time alone was not a significant factor in reducing large zooplankton volumes among pooled s amples ( Figure 630). Trial duration likely resulted in large plankton volume reduction from numerous causes such as intraplankton predation, food limitation, and environmental stressors (temperature, DO, etc.). However, time was not as strong a factor in reducing large plankton biomass as fish presence, and large plankton volume reductions resulting from trial duration alone were not present among the pooled nutrient treatment data set. Similar to total large plankton volume differences among pond trea tments and 24hour feeding trial treatments, pond nutrient treatment was not a factor in determining large zooplankton densities. However, within a given pond nutrient treatment, total large zooplankton density averages uniformly decreased among feeding t rial treatments in the following order: Initial, NF, and Fish ( Figures 6 31 635). Within pond treatments, total plankton density averages noticeably dropped among the three experimental treatments ( Figures 6 32 635), although statistical differences were not always present due to large variation among replicates.

PAGE 435

435 Again, this suggests that both fish presence and time (24hour trial duration) decrease large plankton abundance due to swordtail planktivory and cooccurring factors unrelated to fish predation (e.g., intra plankton predation, starvation, DO stress, temperature extremes, etc .). Large plankton density differences among the 24hour feeding trial experimental treatments were present only within the UNP and INO pond nutrient treatments, where the presence of a single swordtail (Fish) had lowered large plankton densities relative to either the Initial or NF (no fish) treatments over a 24 hour period ( Figures 6 33, 6 35). As with large plankton volume differences among experimental treatments, l arge plankton density differences among experimental treatments, also indicated that fish presence, and likely fish planktivory, was responsible for the lower large plankton biomasses recovered from these treatments after 24 hours. The finding that pond nu trient was not a significant factor in determining large plankton densities, was used as justification for pooling replicates within experimental treatment groups. Pooled one way repeated measures ANOVA found large plankton density differences between the Fish experimental group and the two remaining treatment groups (Initial and NF), which did not differ from each other (Figure 6 36). Again, these results indicate that fish presence and likely planktivory, but not trial duration (24 hours) had a detectable e ffect upon large plankton assemblage biomass. Although these results indicate that trial duration had a small, but statistically insignificant e ffect upon biomass within the experimental design of the 24 hour captive feeding trial. Major Taxa Densitie s among 24 Hour Trial Treatments Within individual pond nutrient treatments, the three major large zooplankton taxa ( Diaptomous copepods, Filinia rotifers, and daphnids Moina macrocopa) densities decreased within feeding trial treatments in the following order: Initial, NF, and Fish ( Figure 6 376.40).

PAGE 436

436 Filinia rotifer density differences among the three feeding trial treatments were not significant, due to the low densities present within samples, regardless of feeding trial treatment (Initial, NF, and Fish). Diaptomous copepod densities were numerically lo wer within the Fish treatment relative to the two remaining treatments (Initial, NF) in all four pond nutrient treatments (Figure 6376.40), but was only significantly lower within the INO treatment (Figure 640). Similar to large plankton volume and density, when nutrient treatments were pooled, the Fish treatment copepod density was significantly less than that of the Initial and NF treatments ( Figure 641). Within given feeding trial treatments (Initial, NF, or Fish), Filinia densit ies did not differ among pond nutrient treatments, but within nutrient treatments, densities differed among feeding trial treatments. However, a single pairwise difference in Filinia densities was found between the PRO and CSM treatments for the Initial trea tment according to post hoc tests P ost hoc differences can occur among treatments, even when ANOVA does not detect a significant difference among treatments (Bonferroni post test, GraphPad Prism 4.0; Motulsky 1995, Zar 1984). Within pond nutrient treatm ents, significant differences in Filinia density among the feeding trial treatments were only found within the PRO pond treatment (PRO Initial and NF, PRO Initial and Fish). L ack of differences among the Filinia experimental treatments within the nonPRO treatments was puzzling at first, but upon closer examination it can be seen that Initial, NF and Fish treatment rotifer numbers were extremely low (range 2 18 rotifers/L; Figures 6 42 646, 652, 654). Logically, the Initial experimental treatment shou ld have had the highest number of rotifers of the three experimental treatments, which was not the case for the CSM treatment (Figure 644). Initial experimental treatment samples were immediately fixed with Lugols solution upon collection from the pond and had not been subjected to zooplankton or

PAGE 437

437 fish predation, or possible stress and mortality from 24 hours of captivity ( Figures 6 42 646). A major problem with small sample sizes is that significant differences among treatments can easily occur due to random sampling differences (Krebs 1998, Zar 1984). Rotifer densities for the 24hour captive feeding trials (mid September 2007) were e xtremely low relative to the majority of samples acquired during the 12week swordtail pond trials performed the previous year (March June 2006) possibly due to seasonal effects. Even when pond nutrient treatments were pooled, no significant differenc es in Filinia densities among 24hour feeding trial treatments were found ( Figure 646). Again, this may have been due to the low average Filinia densities among the pooled Initial, NF and Fish treatments and large variation among replicates within the Initial treatment. Daphnid densities did not differ among pond nutrient treatments, but did differ among the three feeding trial treatments (Initial, NF, Fish). Within each of the four pond nutrient treatments, Moina macrocopa density averages decreased i n the following order: Initial, NF and Fish ( Figures 647 650). However, only a single significant difference (UNP Initial and UNP Fish) in Moina density was found among the three captive feeding trial experimental treatments ( Figure 6 48). In an atte mpt to determine if daphnid density differences among the three experimental treatments could be better discerned with increased sample size, samples were pooled across the four pond nutrient treatments. Among the pooled data, t wo of the three possible pa irwise treatment comparisons were significant; the Fish treatment had less daphnids than either the Initial or NF treatments, which did not differ from each other ( Figure 6 51). This further supports the contention that fish were grazing on large zooplankton and daphnids in particular, to the extent that the Fish treatment had significantly less daphnids than either of the control treatments (Initial, NF).

PAGE 438

438 Although pooling treatment groups is not ideal, due to the general similarities in the Initial plankton assemblage communities among the four pond nutrient treatments, a strong argument against pooling did not appear to be present ( Figures 6 1, 6 4, 6 7, 610, 656, 659). Additionally, average densities, that did not statistically differ prior to data pooling, differed after pooling, which may indicate the presence of strong experimental treatment effects that were not detected due to inadequate replication (four samples), or experimental treatment limitations (e.g. possible insufficient trial duration 24 hrs, and/or insufficient pond water sample volumes 1 L, or predator number). Insufficient replication and trial duration likely occurred during the 24 hour captive feeding trial due to time constraints, only four pond replicates were obtaine d from each of the four pond nutrient treatments (PRO, UNP, CSM and INO) when six replicate ponds were available. Pooling nutrient treatments also was partially justified as no significant differences in densities of the three major large zooplankton ta xa groups (copepods, Filinia rotifers, daphnids), occurred among the Initial treatments of the four pond nutrient treatments ( Figure 6 52 6.55). Additionally, the main objective of the 24hour feeding trial was to determine the relative effects of swordta il feeding on plankton assemblages from four different pond nutrient regimes, and not to evaluate differences in plankton assemblages among pond nutrient treatments per se. Additionally, live food nave fish from a single source were used for the feeding tria l s F ish were not obtained from ponds that had received different applied nutrients. This ruled out previous nutritional history from outdoor pond applied nutrients (e.g., feeds, cottonseed meal), or live prey exposure as potential experimental arti facts or biases among swordtails randomly assigned to large plankton assemblages obtained from different pond nutrient treatments.

PAGE 439

439 24Hour Trial Community %IRI Values Within each of the 24hour feeding trial treatments (Initial, NF, Fish), large plankton assemblage community %IRI values were fairly similar among pond nutrient treatments (Figures 656 661). Interestingly, copepod and daphnid %IRI values appeared to be inversely proportional within the Initial and NF control treatments (treatments lackin g potential fish predator), as copepod and daphnid %IRI values were never simultaneously low or high ( Figures 652 653, 655 656). This was similar to the inversely proportional copepod and daphnid %IRI relationship observed during the 2006 12week outdoor pond trial (Chapter 5; Figures 5 13 520). The Fish treatment showed the most dramatic differences in large plankton taxa %IRI values, compared to their Initial and NF control treatment counterparts ( Figures 654, 657). Relative decreases in daphnid %IRI, percent volume (%V), and percent number (%N), coupled with simultaneously increases in copepod %IRI, %V, and %N indicate that swordtails were preferentially selecting daphnid prey within the Fish treatment during the 24hour feeding trial. S wordtails also were utilizing copepod s and other prey, as overall densities of copepods, daphnids, and total large plankton volumes and densities decreased within the Fish treatment relative to the Initial and NF control treatments (Figures 6 1 624, 638 641, 647 657). Although copepod density was reduced within the Fish treatment, copepod relative importance (%IRI) increased in the presence of swordtail predation. The large antennae of Calanoid copepods may have made them a less attractive foo d source relative to daphnids as the large antennae of copepods may have increased copepod prey handling time, or reduced fish feeding efficiency by clogging gill rakers, or by producing physical irritation (Chapter 5; Figure 537a,b).

PAGE 440

440 24Hour Trial Simplified Morisitas and Percent Similarity Indices Within pond nutrient treatments, large plankton community Morisitas similarity (MSI), and to a lesser degree percent similarity (PSI) indices were higher between the Initial and NF feeding trial control treatments, than they were between the Initial and Fish, and NF and Fish comparisons (Figures 6 62 663). Notable exceptions were the PRO Initial and NF PSI values which were markedly lower than those of the other three pond nutrient treatments (Figure 663). Within nutrient treatments, Initial and Fish, and NF and Fish, large plankton community pairwise MSI and PSI values did not appear to greatly differ from each other, but both values were dramatically lower than those of the Initial and NF pair wise MSI and PSI values ( Figures 6 62 663). These results indicate that the presence of a single fish predator for a 24 hour period fundamentally changed previously similar large plankton communities (NF and Fish). Bootstrap Initial, NF, and Fish MSI and PSI Estimates Captive f eeding trial large plankton community bootstrap MSI values did not differ among the feeding trial treatment comparisons (Initial and NF, Initial and Fish, NF and Fish) within any of the four pond nutrient treatments, but MSI values did differ among pond nutrient treatments within given feeding trial treatmentpair comparisons: Initial and NF MSI PRO and INO, Initial and Fish MSI UNP and INO. The reason that these particular experimental treatment pair MSI values differed among pond nutrient treatments was due to the greater taxa % number similarities of the INO Initial and NF large plankton assemblages (Figures 6 22 623; positive X axis) relative to those of the PRO treatment (Figures 6 13 614), which may have been due to the l arge number of small spball taxa (believed to be fungal spores or unidentified cysts) within the PRO NF large plankton assemblage, which were largely absent within the PRO Initial treatment. The reason the Initial and Fish experimental treatment MSI va lues differed between the UNP and INO treatments was due to the large numbers of Pediastrum spp. green algae

PAGE 441

441 present within the Fish treatment assemblage of the UNP treatment ( Figure 618) that were absent within the UNP Initial plankton assemblage ( Figure 6 16). Generally, bootstrapgenerated MSI values of experimental treatment combinations containing swordtails (Initial and Fish, NF and Fish) were lower than those of the Initial and NF treatment combination, except for those of the PRO pond treat ment. However, variation among replicate large plankton assemblages was fairly high, resulting in the absence of significant bootstrap MSI value differences among the experimental treatment pair combinations. Bootstrap large plankton community PSI values did not differ among experimental treatment group pair combinations within nutrient treatments, or within experimental treatment group pair combinations among the four pond nutrient treatments. Large plankton community bootstrap PSI difference patterns wi thin pond nutrient treatments and among the three experimental treatment group pair combinations followed the same pattern as that of the bootstrap MSI values. With the exception of the PRO nutrient treatment, experimental treatment pairwise combination MSI value rankings were the highest between the Initial and NF treatments, which were then followed by the Initial and Fish, or NF and Fish treatment pairs ( Figures 6 69 671) Initial and Fish, and NF and Fish MSI values were nearly identical. The reason the Initial and NF large plankton assemblages were less similar (lower MSI and PSI values) within the PRO nutrient treatment were likely due to the relatively large numbers of spball spores or cysts within the NF (no fish) experimental treatment, rather than a fundamental difference in the PRO Initial and NF large plankton community assemblages ( Figures 61 62, 613 614, 662 663). Because no significant differences in large plankton community PSI values occurred among feeding treatmentpair combinations within given pond nutrient treatments, pond nutrient

PAGE 442

442 treatments were pooled to increase the experimental treatment sample size from four to sixteen. Again, no significant differences occurred among the three pooled experimental treatment comb inations ( Figure 6 72), which may have been due to the large variation in taxa % numbers within replicate plankton assemblages. Interestingly, only within the UNP and CSM treatments were the Initial and NF MSI values considered similar, and the remaining t wo pairwise treatment combinations (Initial and Fish, NF and Fish) not considered similar, according to the arbitrarily chosen critical threshold value (MSI(critical) feeding trial treatment plankton assemblage MSI comparisons (Initial and NF, Initial and Fish, NF and Fish) were all similar within the INO nutrient treatment (MSI(critical) Similarly, UNP and CSM Initial and NF PSI values were considered similar (PSI(critical) 60%), and the Initial and Fish, and NF and Fish P SI values were not similar ( Figure 6696.70). Again, the INO experimental treatment PSI values were all similar (Figure 6 71).

PAGE 443

443 Figure 61. Initial (0 hrs) processed feed (PRO) treatment large zooplankton community pooled taxa parameters [ % number (%N), % volume (%V), % frequency of occurrence (%FO)] for 24hour feeding trial taxa %N, %V, or %FO %FO cop cop trirot dap dap tlr elm non-lhc bbrot naup 40 10 20 30 40 20 30 10 50 60 70 80 90 50 60 70 80 90 %V %N PROInitial control treatment

PAGE 444

444 Figure 62. N F ( no fish 24 hr incubation) processed feed treatment (PRO) large zooplankton pooled community taxa parameters [ % number (%N), % volume (%V), % frequency of occurrence (%FO)] for 24hour feeding trial taxa %N, %V, or %FO 0.5%; n = 4. %FO cop cop trirot dap dap tlr elm non-lhc bbrot naup vluke spbill 40 20 30 10 10 20 30 40 50 60 70 80 90 50 60 70 80 90 %V %N PRO No Fish control treatment

PAGE 445

445 Figure 63. F ish ( single fish 24 hr incubation) processed feed (PRO) treatment large zooplankton pooled community taxa parameters [ % number (%N), % volume (%V), % frequency of occurrence (%FO)] for 24hour feeding trial taxa %N, %V, or %FO 0.5%; 1L water sample per pond, n = 4 ponds %FO cop cop trirot dap dap tlr non-lhc bbrot naup bin mite lne sclr cst vluke dpp %V %N PRO 24 hour Fish treatment90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90

PAGE 446

446 Figure 64. Initial (0 hrs ) unprocessed feed (UNP) treatment large zooplankton pooled community taxa parameters [ % number (%N), % volume (%V), % frequency of occurrence (%FO)] for 24hour feeding trial taxa %N, %V, or %FO 1L water sample per pond, n = 4 ponds. %FO cop cop trirot dap dap non-lhc cst UNP Initial control treatment90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 447

447 Figure 65. N F ( no fish 24 hr incubation) unprocessed feed (UNP) treatment large zooplankton pooled community taxa parameters [ % number (%N), % volume (%V), % frequency of occurrence (%FO)] for 24hour feeding trial taxa %N, %V, or %FO 0.5%; 1L water sample per pond, n = 4 ponds. %FO cop cop trirot dap dap non-lhc cst lne elm vluke naup bbrot UNP No Fish control treatment 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 448

448 Figure 66. F ish ( single fish 24 hr incubation) processed feed (PRO) treatment large zooplankton pooled community taxa parameters [ % number (%N), % volume (%V), % frequency of occurrence (%FO)] for 24hour feeding trial taxa %N, %V, or %FO 0.5%; 1L water sample per pond, n = 4 ponds. %FO cop cop trirot dap dap non-lhc cst tlr bin mite lne sclr ost elm vluke naup dpp UNP 24 hour Fish treatment90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 449

449 Figure 67. Initial (0 hrs) cottonseed meal (CSM) treatment large zooplankton pooled community taxa parameters [ % number (%N), % volume (%V), % frequency of occurrence (%FO)] for 24hour feeding trial taxa %N, %V, or %FO 1L water sample per pond, n = 4 ponds. %FO cop cop trirot dap dap non-lhc cst naup dpp CSM No Fish control treatment 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 450

450 Figure 68. N F ( no fish 24 hr incubation) cottonseed meal (CSM) treatment large zooplankton pooled community taxa parameters [ % number (%N), % volume (%V), % frequency of occurrence (%FO)] for 24hour feeding trial taxa %N, %V, or %FO 0.5%; 1L water sample per pond, n = 4 ponds. %FO cop cop trirot dap dap non-lhc cst naup elm vluke dpp seed CSM No Fish control treatment90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 451

451 F igure 69. F ish ( single fish 24 hr incubation) cottonseed meal (CSM) treatment large zooplankton pooled community taxa parameters [ % number (%N), % volume (%V), % frequency of occurrence (%FO)] for 24hour feeding trial taxa %N, %V, or %FO 1 L water sample per pond, n = 4 ponds. %FO cop cop trirot dap dap non-lhc cst naup dpp tlr bin lne sclr vluke CSM 24 hour Fish treatment90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 452

452 Figure 610. Initial (0 hrs) inorganic fertilizer (INO) treatment large zooplankton pool ed community taxa parameters [ % number (%N), % volume (%V), % frequency of occurrence (%FO)] for 24hour feeding trial taxa %N, %V, or %FO 1L water sample per pond, n = 4 ponds. %FO cop cop trirot dap dap bbrot naup bin cst elm vluke spbill tlr non-lhc seed INO Initial control treatment 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 453

453 Figure 611. N F ( no fish 24 hr incubation) inorganic fertilizer (INO) treatment large zooplankton pooled community taxa parameters [ % number (%N), % volume (%V), % frequency of occurrence (%FO)] for 24hour feeding trial taxa %N, %V, or %FO 0.5%; 1L water sample per pond, n = 4 ponds. %FO cop cop trirot dap dap bbrot naup bin cst elm lne spbll INO No Fish control treatment90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 454

454 Figure 612. F ish ( no fish 24 hr incubation) inorganic fertilizer treatment large zooplankton pooled community taxa parameters [ % number (%N), % volume (%V), % frequency of occurrence (%FO)] for 24hour feeding trial taxa %N, %V, or %FO 0.5%; 1L water sample per pond, n = 4 ponds. %FO cop cop trirot dap dap naup bin cst elm lne sclr ost non-lhc seed sn INO 24 hour Fish treatment 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 455

455 Figure 613. Initial (0 hrs) processed feed (PRO) treatment large zooplankton unpooled community taxa parameters: % number ( % N 1 SE), % volume (% V 1 SE), and frequency of occurr ence (% FO) for 24hour feeding trial taxa %N, %V, or %FO 0.5%; 1L water sample per pond, n = 4 ponds. %FO cop cop trirot dap dap tlr elm non-lhc bbrot naup bin cst arth PRO Initial control treatment 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 456

456 Figure 614. N F ( no fish 24 hr incubation ) processed feed (PRO) treatment large zooplankton unpooled community taxa parameters : % number (%N 1 SE), % volume (%V 1 SE), and % frequency of occurrence (%FO)] for 24hour feeding trial taxa %N, %V, or %FO 1L water sample per pond, n = 4 ponds. %FO cop cop trirot dap dap tlr elm non-lhc bbrot naup vluke spbill PRO No Fish control treatment 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 457

457 Figure 615. F ish ( single fish 24 hr incubation ) processed feed (PRO) treatment large zoo plankton unpooled community taxa parameters: % number (%N 1 SE), % volume (%V 1 SE), and % frequency of occurrence (%FO) for 24hour feeding trial, taxa %N, %V, or %FO 1L water sample per pond, n = 4 ponds. %FO cop cop dap dap non-lhc bbrot naup bin sclr vluke fls PRO 24 hour Fish treatment 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 458

458 Figure 616. Initial (0 hrs) unpr ocessed feed (UNP) treatment large zooplankton unpooled community taxa parameters: % number (N 1 SE), % volume (%V 1 SE), and frequency of occurrence (%FO) for 24hour feeding trial taxa %N, %V, or %FO 0.5%; 1L water sample per pond, n = 4 ponds. %FO cop cop trirot dap dap non-lhc cst ost dpp spbll UNP Initial control treatment 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 459

459 Figure 617. NF ( no fish 24 hr incubation) unprocessed feed (UNP) treatment large zooplankton unpooled community taxa parameters: % number (%N 1 SE), % volume (%V 1 SE), and % frequency of occurrence (%FO) for 24hour feeding trial, taxa %N, %V, or % FO 1L water sample per pond, n = 4 ponds. %FO cop cop trirot dap dap non-lhc cst lne elm vluke naup bbrot bin ost UNP No Fish control treatment 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 460

460 Figure 618. F ish ( single fish 24 hr incubation) unprocessed feed (UNP) treatment large zooplankton unpooled community taxa parameters: % number (%N 1 SE), % volume (%V 1 SE), and % frequency of occ urrence (%FO) for 24hour feeding trial, taxa %N, %V, or %FO 1L water sample per pond, n = 4 ponds. %FO cop cop dap dap non-lhc cst bin mite lne sclr ost elm vluke naup dpp UNP 24 hour Fish treatment 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 461

461 Figure 619. Initial (0 hrs) cottonseed meal fertilizer (CSM) treatment large zooplankton unpooled community taxa parameters: % number (N 1 SE ), % volume (%V 1 SE), and frequency of occurrence (%FO) for 24hour feeding trial taxa %N, %V, or %FO 1L water sample per pond, n = 4 ponds. %FO cop cop trirot dap dap cst naup dpp mite CSM Initial control treatment 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 462

462 Figure 620. N F ( no fish 24 hr incubation) cottonseed meal (CSM) treatment large zooplankton unpooled community taxa parameters: % number (%N 1 SE), % volume (%V 1 SE), and % frequency of occurrence (%FO) for 24hour feeding trial, taxa %N, %V, or %FO 1L water sample per pond, n = 4 ponds. %FO cop cop trirot dap dap non-lhc cst naup elm vluke dpp seed spbll CSM No Fish control treatment 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 463

463 Figure 621. F ish ( single fish 24 hr incubation) cottonseed meal (CSM) treatment large zooplankton unpooled community taxa parameters: % number (%N 1 SE), % volume (%V 1 SE), and % frequency of occurrence (%FO) for 24hour feeding trial, taxa %N, %V, or %FO 1L water sample per pond, n = 4 ponds. %FO cop cop trirot dap dap non-lhc cst naup dpp tlr bin lne sclr vluke iht CSM 24 hour Fish treatment 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90% N % V

PAGE 464

464 Figure 622. Initial (0 hrs) inorganic fertilizer (INO) treatment large zooplankton unpooled community taxa parameters: % number (N 1 SE), % volume (%V 1 SE), and frequency of occurrence (%FO) for 24hour feeding trial taxa %N, %V, or %FO 0.5%; 1L water sample per pond, n = 4 ponds. %FO cop cop trirot dap dap bbrot naup bin cst elm vluke spbill tlr non-lhc ost seed INO Initial control treatment 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 465

465 Figure 623. N F ( no fish 24 hr incubation) inorganic fertilizer (INO) treatment large zooplankton unpooled community taxa parameters: % number ( % N 1 SE), % volume (%V 1 SE), and f requency of occurrence (%FO) for 24hour feeding trial, taxa %N, %V, or %FO 1L water sample per pond, n = 4 ponds. %FO cop cop trirot dap dap bbrot naup bin cst elm lne spbll INO No Fish control treatment 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 466

466 Figure 624. F ish ( single fish 24 hr incubation) inorganic fertilizer (INO) treatment large zooplankton unpooled community taxa parameters: % number (%N 1 SE), % volume (%V 1 SE), and % frequency of occurrence (%FO) for 24hour feeding trial, taxa %N, %V, or %FO 1L water sample per pond, n = 4 ponds. %FO cop cop trirot dap dap naup bin cst elm sclr ost non-lhc seed sn INO 24 hour Fish treatment 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 467

467 Figure 625. 24hour feeding trial large ( > 200 m) plankton assemblage taxa total volumes (mm3/L) among four pond nutrient treatments [ processed feed (PRO) unprocessed feed (UNP) cottonseed meal fertilizer (CSM) inorganic fertilizer (INO ) ] and three treatments [Initial, NF (no fish), Fish] ; 1 L water sample per pond, n = 4 ponds. 0 2 4 6 8 10 12 14 Initial NF Fishmm3/ L PRO UNP CSM INO

PAGE 468

468 Figure 626. Processed feed (PRO) treatment 24hour feeding trial large (> 200 m) plankton assemblage volumes ( mm3/L 95% CI) within three treatments [Initial, NF (no fish), Fish] four replicate s per treatment; unshared letters denote statistical differences between treatments (P < 0.05, Bonferroni post test ) 0 5 10 15 20 25 30 Initial NF Fishmm3/ Lab b a

PAGE 469

469 Figure 627. Unprocessed feed (UNP) treatment 24hour feeding trial large (> 200 m) plankton assemblage volumes ( mm3 /L 95% CI) within three treatments [Initial, NF (no fish), Fish] four replicate pond water samples per treatment; unshared letters denote statistical differences between treatments (P < 0.05, Bonferroni post test ) 0 5 10 15 20 25 30 Initial NF Fishmm3/ Lb b a

PAGE 470

470 Figure 628 Cottonseed meal fertilizer (CSM) treatment 24hour feedin g trial large (> 200 m) plankton assemblage volume s ( mm3/L 95% CI) within three treatments [Initial, NF (no fish), Fish] four replicate pond water samples per treatment; unshared letters denote statistical differences between treatments (P < 0.05, Bonf erroni post test ) 0 5 10 15 20 25 30 Initial NF Fishmm3/ La a a

PAGE 471

471 Figure 629. Inorganic fertilizer (INO) treatment 24hour feeding trial large (> 200 m) plankton assemblage volumes ( mm3/L 95% CI) within three treatments [Initial, NF (no fish), Fish] four replicat e pond water samples per treatment; unshared letters denote statistical differences between treatments (P < 0.05, Bonferroni post test ) 0 5 10 15 20 25 Initial NF Fishmm3/ La a a

PAGE 472

472 Figure 630. Pooled (among nutrient treatment) 24hour feeding trial large (> 200 m) plankton assemblage volumes ( mm3/L 95% CI) within three treatments [Initial, NF (no fish), Fish] 16 pooled 1 L replicate pond water samples per treatment; unshared letters denote statistical differences between treatments (P < 0.05, Bonferroni post test ) 0 2 4 6 8 10 12 14 16 Initial NF Fish mm3/ La b a

PAGE 473

473 Figure 631. 24hour feeding trial large (> 200 m) plankton assemblage densit ies (plankton number/L) among four pond nutrient treatments [ processed feed (PRO) unprocessed feed (UNP) cottonseed meal fertilizer (CSM) inorganic fertilizer (INO ) ] and three treatments [Initial, NF (no fish), Fish]. 0 50 100 150 200 250 300 350 400 Init Ave N NF Ave N Fish Ave N Number / L

PAGE 474

474 Figure 632. Processed feed (PRO) treatment 24hour f eeding trial large (> 200 m) plankton assemblage densit ies ( plankt er number/L 95% CI) within three treatments [Initial, NF (no fish), Fish] four replicate samples per treatment; unshared letters denote statistical differences between treatments (P < 0.05, Bonferroni post test ) 0 100 200 300 400 500 600 700 800 900 Initial NF Fish Number / Lb a a

PAGE 475

475 Figure 633. Unprocessed feed (UNP) treatment 24hour feeding trial large (> 200 m) plankton assemblage densit i es ( plankt er number/L 95% CI) within three treatments [Initial, NF (no fish), Fish] four replicate samples per treatment; unshared letters denote statistical differences between treatments (P < 0.05, Bonferroni post test ) 0 100 200 300 400 500 600 700 800 900 Initial NF Fish Number / Lb b a

PAGE 476

476 Figure 634. Cottonseed me al fertilizer (CSM) treatment 24hour feeding trial large (> 200 m) plankton a ssemblage densit ies ( plankt er number/L 95% CI) within three treatments [Initial, NF (no fish), Fish] four replicate samples per treatment; unshared letters denote statistical differences between treatments (P < 0.05, Bonferroni post test ) 0 50 100 150 200 250 300 350 Initial NF FishNumber / La a a

PAGE 477

477 Figure 635. Inorganic fertilizer (INO) treatment 24hour feeding trial large (> 200 m) plankton assemblage densit ies ( plankt er number /L 95% CI) within three treatments [Initial, NF (no fish), Fish] four replicate samples per treatment; unshared letters denote statistical differences between treatments (P < 0.05, Bonferroni post test ) 0 50 100 150 200 250 300 350 400 450 500 Initial NF Fish Number / La a a

PAGE 478

478 Figure 636. Pooled (among nutrient treatment) 24hour feeding trial large (> 200 m) plankton a ssemblage densit ies ( plankt er number/L 95% CI) within three treatments [Initial, NF (no fish), Fish] 16 pooled replicates per treatment; unshared letters denote statistical differences between treatments (P < 0.05, Bonferroni post test ) 0 50 100 150 200 250 300 350 400 450 Initial NF Fish Number / Lb b a

PAGE 479

479 Figure 637. Processed feed (PRO) treatment 24hour feeding trial large (> 200 m) plankton assemblage Diaptomous copepod densit ies ( copepods /L 95% CI) within three treatments [Initial, NF (no fish), Fish] four repl icates per treatment; unshared letters denote stat istical differences between treatments (P < 0.05, Bonferroni post test ) 0 50 100 150 200 250 300 350 400 450 Init Ave N NF Ave N Fish Ave N Copepods / Lab b a

PAGE 480

480 Figure 638. Unprocessed feed (UNP) treatment 24hour feeding trial large (> 200 m) plankton assemblage Diaptomous copepod densit ies ( copepods /L 95% CI) within three treatments [Initial, NF (no fish), Fish] four replicates per treatment; unshared letters denote statistical differences between treatments (P < 0.05, Bonferroni post test ) 0 50 100 150 200 250 300 350 400 450 Init Ave N NF Ave N Fish Ave N Copepods / La a a

PAGE 481

481 Figure 639 Processed feed (PRO) treatment 24hour feeding trial large (> 200 m) plankton assemblage Diaptomous copepod densit ies ( copepods /L 95% CI) within three treatments [Initial, NF (no fish), Fish] four replicates per treatment; unshared letters denote statistical differences between treatments (P < 0.05, Bonferroni post test ) 0 50 100 150 200 250 300 350 400 450 Init Ave N NF Ave N Fish Ave N Copepods / La a a

PAGE 482

482 Figure 640. Inorganic fertilizer (INO) treatment 24hour feeding trial large (> 200 m) plankton assemblage Diaptomous copepod densit ies ( copepods /L 95% CI) within three treatments [Initial, NF (no fish), Fish] four r eplicates per trea tment; unshared letters denote statistical differences between treatments (P < 0.05, Bonferroni post test ) 0 50 100 150 200 250 300 350 400 450 Init Ave N NF Ave N Fish Ave N Copepods / La a a

PAGE 483

483 Figure 641. Pooled (among nutrient treatment) 24hour feeding trial large (> 200 m) plankton assemblage Diaptomous copepod densit ies ( copepods /L 95% CI) within three treatments [Initial, NF (no fish), Fish] 16 replicate s per treatment; unshared letters denote statistical differences between treatments (P < 0.05, Bonferroni post test ) 0 20 40 60 80 100 120 140 160 180 200 Initial NF Fish Copepods / La a b

PAGE 484

484 Figure 642. Processed feed (PRO) treatment 24hour fe eding trial large (> 200 m) plankton assemblage Filinia rotifer densit ies ( rotifer s /L 95% CI) within three treatments [Initial, NF (no fish), Fish] four replicate water samples per treatment; unshared letters denote statistical differences between trea tments (P < 0.05, Bonferroni post test ) 0 10 20 30 40 50 60 70 Init Ave N NF Ave N Fish Ave N Rotifers / Lb a a

PAGE 485

485 Figure 643. Unprocessed feed (UNP) treatment 24hour feeding trial large (> 200 m) plankton assemblage Filinia rotifer densit ies ( rotifer s /L 95% CI) within three treatments [Initial, NF (no fish), Fish] four replicate water samples per treatment; unshared letters denote statistical differences between treatments (P < 0.05, Bonferroni post test ) 0 10 20 30 40 50 60 70 Init Ave N NF Ave N Fish Ave NRotifers / La a a

PAGE 486

486 Figure 644. Cottonseed meal fertilizer (CSM) treatment 24hour feeding trial large (> 200 m) plankton a ssemblage Filinia rotifer densit ies ( rotifers /L 95% CI) within three treatments [Initial, NF (no fish), Fish] unshared letters denote statistical differences between treatments (P < 0.05, Bonferroni post test); four replicate s per treatment 0 10 20 30 40 50 60 70 Init Ave N NF Ave N Fish Ave NRotifers / La a a

PAGE 487

487 Figure 645. Inorganic fertilizer (INO) treatment 24hour feeding trial large (> 200 m) plankton assemblage Filinia rotifer densit ies ( rotifer s /L 95% CI) within three treatments [Initial, NF (no fish), Fish] four replicate water samples per treatment; unshared letters denote statistical differences between treatments (P < 0.05, Bonferroni post test ) 0 10 20 30 40 50 60 70 Init Ave N NF Ave N Fish Ave N Rotifers / La a a

PAGE 488

488 Figure 646. Pooled (among nutrient treatment) 24hour feeding trial large (> 200 m) plankton assemblage Filinia rotifer densit i es ( rotifer s /L 95% C I) within three treatments [Initial, NF (no fish), Fish] 16 replicate water samples per treatment; unshared letters denote statistical differences between treatments (P < 0.05, Bonferroni post test ) 0 2 4 6 8 10 12 14 16 18 Initial NF Fish Rotifers / La a a

PAGE 489

489 Figure 647. Processed feed (PRO) treatment 24hour feeding trial large (> 200 m) plankton assemblage Moina macrocopa (Daphnid) d ensit i es ( daphnids /L 95% CI) within three treatments [Initial, NF (no fish), Fish] four replicates per treatment; unshared letters denote statistical differences between treatments (P < 0.05, Bonferroni post test ) 0 100 200 300 400 500 600 Init Ave N NF Ave N Fish Ave N Daphnids / La a a

PAGE 490

490 Figure 648. Unprocessed feed (UNP) treatment 24hour feeding trial large (> 200 m) plankton assemblage Moina macrocopa (Daphnid) d ensit i es ( daphnids /L 95% CI) within three treatments [Initial, NF (no fish), Fish] four repli cate s per treatment; unshared letters denote statistical differences between treatments (P < 0.05, Bonferroni post test ) 0 100 200 300 400 500 600 700 Init Ave N NF Ave N Fish Ave N Daphnids / Lb ab a

PAGE 491

491 Figure 649. Cottonseed meal fertilizer (CSM) treatment 24hour feeding trial large (> 200 m) plankton assemblage Moina macrocopa (Daphnid) densit i es ( daphnids/L 95% CI) within three treatments [Initial, NF (no fish), Fish] four replicates per treatment; unshared letters denote statistical differences between treatments (P < 0.05, Bonfe rroni post test ) 0 20 40 60 80 100 120 140 Init Ave N NF Ave N Fish Ave N Daphnids / La a a

PAGE 492

492 Figure 650. Inorganic fertilizer (INO) treatment 24hour feeding trial large (> 200 m) plankton assemblage Moina macrocopa (Daphnid) densit i es ( daphnids /L 95% CI) within three treatments [Initial, NF (no fish), Fish] four replica tes per treatment; unshared letters denote statistical differences between treatments (P < 0.05, Bonferroni post test ) 0 20 40 60 80 100 120 140 160 180 200 Init Ave N NF Ave N Fish Ave N Daphnids / Lab a b

PAGE 493

493 Figure 651. Pooled (among nutrient treatment) 24hour feeding trial large (> 200 m) plankton assemblage Moina macrocopa (Daphnid) densit ie s ( daphnids /L 95% CI) within three experimental treatments [Initial, NF (no fish), Fish] 16 samples per treatment; unshared letters denote statistical differences between treatments (P < 0.05, Bonferroni post test ) 0 50 100 150 200 250 Initial NF Fish Daphnids / La a b

PAGE 494

494 Figure 652. 24hour feed ing trial Initial control treatment major taxa densities ( plankt er number / L ) within four pond nutrient treatments [ processed feed (PRO) unprocessed feed (UNP) cottonseed meal fertilizer (CSM) inorganic fertilizer (INO ) ] ; four 1 L replicate water samples per treatment 0 50 100 150 200 250 PRO UNP CSM INO Plankton / L COP CLD DAP

PAGE 495

495 Figure 653. 24 hr feeding trial Initial treatment Diaptomous copepod densities ( copepods / L 95% CI) among four treatments [ processed feed (PRO) unprocessed feed (UNP) cottonseed meal fertilizer (CSM) inorganic fertilizer (INO ) ] four replicates per treatment; unshared letters denote statistical differences between treatments (P < 0.05, Bonferroni post test). 0 50 100 150 200 250 300 350 400 450 PRO UNP CSM INO Copepods / La a a a

PAGE 496

496 Figure 654. 24 hr feeding trial Initial treatment Filinia spp. rotifer densiti es ( rotifers /L 95% CI) within large ( > 200 um) plankton assemblages [ processed (PRO) and unprocessed (UNP) feed, cottonseed meal (CSM) and inorganic fertilizer (INO ) ], unshared letters denote statistical differences (P < 0.05 Bonferroni posttests) ; four replicates per treatment. 0 10 20 30 40 50 60 70 PRO UNP CSM INO Rotifers / L a ab b ab

PAGE 497

497 Figure 655. 24 hr feeding trial Initial treatment Moina macrocopa (Daphnid) densities ( daphnids / L 95% CI) within four nutrient treatments [ processed (PRO) and unprocessed (UNP) feed, cottonseed meal (CSM) and inorganic (INO) fertilizer ] four replicates per tre atment; unshared letters denote statistical differences between treatments (P < 0.05, Bonferroni post test ) 0 100 200 300 400 500 600 700 PRO UNP CSM INO Daphnids / La a a a

PAGE 498

498 Figure 656. 24hour feeding trial Initial (0 hrs) treatment major (taxa %IRI > 0.5% ; IRI: index of relative importance ) large (> 200 m) plankton assemblage taxa %IRI values pooled among four nutrient treatments [ processed (PRO) and unprocessed (UNP) feed, cottonseed meal (CSM) and inorganic (INO) fertilizer ] 0 10 20 30 40 50 60 70 80 90 100 Initial PRO Initial UNP Initial CSM Initial INO% IRI COP TRIROT DAP TLR BIN CST NAUP NON LHC

PAGE 499

499 Figure 657. 24hour feeding trial NF (no fish + 24 hr s ) treatment ma jor (taxa %IRI > 0.5%; IRI: index of relative importance ) large (> 200 m) plankton assemblage taxa %IRI values pooled among four pond nutrient treatments [ processed (PRO) and unprocessed (PRO) feed, cottonseed meal (CSM) and inorganic (INO) fertilizer ] 0 10 20 30 40 50 60 70 80 90 100 NF PRO NF UNP NF CSM NF INO % IRI COP TRIROT DAP TLR BIN CST ELM VLUKE NAUP NON LHC SEED BBROT SPBLL

PAGE 500

500 Figure 658. 24hour feeding trial Fish (one fish + 24 hr s ) treatment major (taxa %IRI > 0.5% ; IRI: index of relative importance ) large (> 200 m) plankton assemblage taxa %IRI values pooled among four pond nutrient treatments [ processed (PRO) and unprocessed (UNP) feed, cottonseed meal (CSM) and inorganic (INO) fertilizer] 0 10 20 30 40 50 60 70 80 90 100 Fish PRO Fish UNP Fish CSM Fish INO% IRI COP TRIROT DAP TLR BIN LNE SCLR CST ELM VLUKE NAUP NON LHC DPP BBROT

PAGE 501

501 Figure 659. 24hour feeding trial Initial (no fish + 0 hr s ) treatment major (taxa %IRI > 0.5% ; IRI: index of relative importance ) large (> 200 m) plankton assemblage taxa unpooled %IRI values among four pond nutrient treatments [ processed (PRO) and unprocessed (UNP) feed, and cottonseed meal (CSM) and inorganic (INO) fertilizer ] 0 10 20 30 40 50 60 70 80 90 Initial Pro Initial Unpro Initial CSM Initial INO% IRI (unpooled) COP TRIROT DAP TLR BIN MITE CST ELM NAUP NON LHC BBROT

PAGE 502

502 Figure 660. 24hour feeding trial NF (no fish + 24 hr s ) treatment major (taxa %IRI > 0.5%; IRI: index of relative importance ) large (> 200 m) plankton assemblage taxa unpooled %IRI values among four pond nutrient treatments [ processed (PRO) and unprocessed (UNP) feed, and cottonseed meal (CSM) and inorganic (INO) fertilizer] 0 10 20 30 40 50 60 70 80 90 100 NF Pro NF Unpro NF CSM NF INO%IRI (unpooled) COP TRIROT DAP TLR BIN CST ELM VLUKE NAUP NON LHC SEED BBROT SPBLL

PAGE 503

503 Figure 661. 24hour feeding trial Fish (one fish + 24 hr s ) treatment major (taxa %IRI > 0.5% ; IRI: index of relative importance ) large (> 200 m) plankton assemblage taxa unpooled %IRI values among four pond nutrient treatments [ processed (PRO) and unprocessed (UNP) fee d, and cottonseed meal (CSM) and inorganic (INO) fertilizer] 0 10 20 30 40 50 60 70 80 90 100 Fish Pro Fish Unpro Fish CSM Fish INO %IRI (unpooled) COP TRIROT DAP TLR BIN SCLR CST ELM VLUKE NAUP NON LHC DPP

PAGE 504

504 Figure 662. Deterministic large plankton assemblage simplified Morisitas s imilarity i ndex values (using taxa %IRI) between 24hour feeding trial treatments within four pond nutrient treatments [ processed (PRO) and unprocessed (UNP) feed, and cottonseed meal (CSM) and inorganic (INO) fertilizer ] 0 0 0 2 0 4 0 6 0 8 1 0 1 2 Initial v. NF Initial v. Fish NF v. Fish MSI ( 0 1 ) PRO UNPRO CSM INO

PAGE 505

505 Figure 663. Deterministic large plankton assemblage percent s imilarity i ndex values [ using taxa %IRI (index of relative importance ) ] between 24hour feeding trial treatments within four pond nutrient treatments [ processed (PRO) and unprocessed (UNP) feed, and cottonseed meal (CSM) and inorganic (INO) fertilizer]; four replicates per treatm ent 0 10 20 30 40 50 60 70 80 90 100 Initial v. NF Initial v. Fish NF v. Fish PSI (%) PRO UNPRO CSM INO

PAGE 506

506 Figure 664. 24hour feeding trial large (> 200 m) plankton assemblage bootstrap (taxa % number ; 1000 iterations ) simplified Morisitas s imilarity i ndex ( MSI 95% CI ) between treatments within the processed feed (PRO) treatment four replicate water samples; unshared letters denote statistical differences (P < 0.05 Tukeys multiple comparison test ) 0 0 0 2 0 4 0 6 0 8 1 0 1 2 Initial v. NF Initial v. Fish NF v. Fish MSI ( 0 1 )a a a

PAGE 507

507 Figure 665. 24hour feeding trial large (> 200 m) plankton assemblage bootstrap (taxa % number ; 1000 iterations ) simplified Mo risitas s imilarity i ndex ( MSI 95% CI ) between treatments within the unprocessed feed (UNP) treatment, four replicate water samples; unshared letters denote statistical differences (P < 0.05 Tukeys multiple comparison test ) 0 0 0 2 0 4 0 6 0 8 1 0 1 2 Initial v. NF Initial v. Fish NF v. Fish MSI ( 0 1 )a a a

PAGE 508

508 Figure 666. 24hour feeding trial large (> 200 m) plankton assemblage bootstrap (taxa % number ; 1000 iterations ) simplified Morisitas s imilarity i ndex ( MSI 95% CI ) between treatments within the cottonseed meal (CSM) treatment, four replicates per treatment; unshared lett ers denote statistical differences (P < 0.05 Tukeys multiple comparison test ) 0 0 0 2 0 4 0 6 0 8 1 0 1 2 Initial v. NF Initial v. Fish NF v. Fish MSI ( 0 1 )a a a

PAGE 509

509 Figure 667. 24hour feeding trial large (> 200 m) plankton assemblage bootstrap (taxa % number : 1000 iterations ) simplified Morisitas s imilarity i ndex ( MSI 95% CI ) between treatments within the inorganic fertilizer (INO) treatment, four replicates per treatment; unshared letters denote statistical differences (P < 0.05 Tukeys multiple comparison test ) 0 0 0 2 0 4 0 6 0 8 1 0 1 2 Initial v. NF Initial v. Fish NF v. Fish MSI ( 0 1 )a a a

PAGE 510

510 Figure 668. 24hour feeding trial large (> 200 m) plankton assemblage bootstrap (taxa % number ; 1000 iterations ) percent s imilarity i ndex ( PSI 95% CI ) between treatments within the processed feed (PRO) pond nutrient treatment four replicates per treatment; unshared letters denote statistical differences (P < 0.05, Tukeys multiple comparison test ) 0 10 20 30 40 50 60 70 80 90 100 Initial v. NF Initial v. Fish NF v. Fish PSI (%)a a a

PAGE 511

511 Figure 669. 24hour feeding trial large (> 200 m) plankton assemblage bootstrap (taxa % number ; 1000 iterations ) percent s imilarity i ndex ( PSI 95% CI ) between treatments within the unprocessed feed ( UNP) treatment, four replicate s per treatment; unshared letters denote statistical differences (P < 0.05 Tukeys multiple comparison test ) 0 10 20 30 40 50 60 70 80 90 100 Initial v. NF Initial v. Fish NF v. Fish PSI (%)a a a

PAGE 512

512 Figure 670. 24hour feeding trial large (> 200 m) plankton assemblage bootstrap (taxa % number ; 1000 iterations ) percent s imilarity i ndex ( PSI 95% CI ) between treatments within the cottonseed meal fertilizer (CSM) treatment f our replicate s per treatment ; unshared letters denote statistical differences (P < 0.05, Tukeys multiple comparison test ) 0 10 20 30 40 50 60 70 80 90 100 Initial v. NF Initial v. Fish NF v. Fish PSI (%)a a a

PAGE 513

513 Figure 671. 24hour feeding trial large (> 200 m) plankton assemblage bootstrap (taxa % number ; 1000 iterations ) percent s imilarity i ndex ( PSI 95% CI ) between treatments within the inorganic fertilizer (INO) treatment, four replicates per treatment; unshared letters denote statistical differences (P < 0.05, Tukeys multiple comparison test ) 0 10 20 30 40 50 60 70 80 90 100 Initial v. NF Initial v. Fish NF v. Fish PSI (%)a a a

PAGE 514

514 Figure 672. 24hour feeding trial large (> 200 m) plankton assemblage bootstrap (taxa % number ; 1000 iterations ) percent s imilarity i ndex ( PSI 95% CI ) betw een treatments within pooled pond nutrient treatment s, 16 replicates per treatment; unshared letters denote statistical differences (P < 0.05 Tukeys multiple comparison test ) 0 10 20 30 40 50 60 70 80 Initial v. NF Initial v. Fish NF v. Fish PSI (%)a a a

PAGE 515

515 CHAPTER 7 S UMMARY, CONCLUSIONS, AND RECOMMENDATIONS The overall goal of this research was to provide insight into the trophic dynamics occurring with in commercial tropical fish ponds. The use of robust stable isotope measurement techniques, allowed descri ption of the trophic interactions occurring in tropical fish ponds that had been previously based upon a vast, but incomplete knowledge base. Hopefully, this study ha s added scientific and practical aquaculture management information to this knowledge base. Indoor Fr y Growth and Isotopic Tracing Validation Trial The primary objective of the indoor feeding trial was to validate the isotope tracing methodology by feeding five experimental feeds with differing isotopic profiles to swordtail fry, and determine if different fry groups subsequently developed unique isotopic profiles. An additional objective of the indoor trial was to derive carbon and nitrogen enrichment magnitudes for a single trophic level, and apply these values to outdoor pond trials to estimate trophic positions of pond taxa based upon their carbon and nitrogen isotope signatures. Indoor Fry Growth The 100% wheat protein diet fry exhibited less growth than those of the other four feed treatments from day 30 onward, and were significantly smaller tha n the other groups at the end of the trial (Figure 2 12 ). Although it seems likely that lysine deficiency was responsible for reduced fry growth within this treatment lower feed palatability of the 100% wheat protein diet may have reduced feed consumption, resulting in lower growth. Unfortunately, due to ad libitum feeding ( instead of restricted rations), and the impracticality of measuring uneaten food, it is unknown whether feed palatability or suboptimal amino acid profile was responsible for the reduced growth in the 100% wheat protein diet fed fry.

PAGE 516

516 Again, the finding that elemental carbon to nitrogen ratios of fry did not differ among treatment groups at harvest, tends to contradict, but not rule out reduced feed palatability as a possible reason for the lower growth observed for the 100% wheat protein diet fry. Lack of carbon to nitrogen ratio differences suggests that the overall nutritional status of fry did not differ among treatments as nitrogen and carbon ratios are considered direct indicators of protein (nitrogen), and carbohydrate and lipid (carbon) content ( Valiela 1995). Elemental carbon to nitrogen ratio s suggest that the suboptimal amino acid composition of the 100% wheat protein diet may have been the primary reason for reduced fry growth observed within this treatment. The lack of differences in both fry survival rates and fry elemental carbon to nitrogen ratios among experimental feed treatments suggest that the nutritional states of fry did not differ among treatments. An overall pr otein, lipid and carbohydrate deficiency from malnutrition and feed aversion would likely manifest itself in survival rate differences due to a chronic protein and energy defici ency If the 100% wheat protein diet fry were eating less feed due to palatabi lity issues, it would be expected that both elemental carbon to nitrogen ratios (and therefore nitrogen content) and treatment fry survival rates would differ from fry of the other treatments due to lower total nitrogen (protein) intake (Dabrowki et al. 20 07). However, reduced palatability of the 100% wheat protein diet cannot be ruled out C oldwater salmonids have been show to differentiate lysine deficient diets from complete diets and have exhibited less growth and lower consumption on lysine deficient diets (Yamamoto et al. 2001). In this study, lysine was present in the 100% wheat protein diet, but at levels considered insufficient for maximum growth ( NRC 1993) Conversely, i ncreased food consumption to compensate for low lysine feed, may have led to increased SDA (specific dynamic activity) and

PAGE 517

517 greater energy expenditure in the 100% wheat protein diet, possibly to the point growth was reduced (Tandler and Beamish 1981, Jobling 1983). Ultimately, the lower growth rate of the 100% wheat protein diet fry was of minor importance as the primary objective of the study was to validate the ability of the isotope tracing method to differentiate consumer groups that had consumed isotopically different food sources. This was demonstrated by the isotopic signature profiles of the five experimental feeds, which were isotopically mirrored by the isotopic signature s exhibited by the pre harvest and harvest swordtail fry (Figures 2 16 218). Initial Fry Isotope Signature Female broodstock, male broodstock, and initial fry carbon isotope signatures did not differ at the start of the trial, whereas significant nitrogen isotope signature differences were present between the male broodstock and initial fry (Figure 2 15). Thes e findings clearly contrast the differing strengths and purposes of carbon and nitrogen isotope signature determinations, which help ed to validate the isotope tracing methodology. Carbon signatures were highly conserved between fry offspring and broodstoc k females; all three groups had similar carbon isotope signatures due to their direct nutrition from the control feed (TetraMin tropical flake feed) in the case of the female and male broodstock, and more indirect maternal nutrition (lecitrophic) in the c ase of fry. In contrast to carbon signature results, fry nitrogen signatures were significantly higher than male broodstock, and numerically higher, but not significantly higher than female broodstock. This indicates that fry (via maternal egg nutrition ) were a quasi trophic level enriched relative to male broodstock, and that fry were significantly enriched in the heavier nitrogen isotope (15N) illustrating the trophic position identifying capability of nitrogen isotope ratio determination s

PAGE 518

518 Pre harvest fry carbon and nitrogen isotope signatures diverged quickly and in a predictable manner (Figure 215). Changes in isotopic signatures appeared to follow fairly linear trajectories even though pre harvest fry isotopic profile data points were all from single replicates, d emonstrating that isotopically identical fry undergo systematic changes in isotope signature following diet switching to a food, with a different isotopic profile These results support the validity of the carbon and nitrogen isotope tracing methods. At harvest, carbon and nitrogen isotope signatures significantly differed among the majority of treatments except for four pairwise treatment comparisons (10 possible) for carbon and two pairwise treatment comparisons (10 possible) for nitrogen (Figure 218). Relative to carbon, greater nitrogen signature differences occurred among harvest fry, which likely contributed to the greater number of harvest fry nitrogen signature differences. Isotopically similar consumers largely differed after 12 weeks of consuming isotopically contrasting diets; additionally, isotopic differences of fry reflected the isotopic differences of the experimental feeds they consumed in a regular and predictable manner (Figures 218, 221). These findings support the use of the isotopic tracing method as a valid means of determining the primary food sources of a consumer. Additionally, the observed mean carbon and nitrogen isotope signature enrichment magnitudes were well within published values for a simple, s ingle trophic level food chain. This allow ed aquaculture pond food sources t o be identified with a high degree of confidence, and also allow ed aquaculture pond food web organism trophic position to be identified with a high degree of confidence. Fry Gro wth and Survival Rates Paradoxically, the feed treatment that displayed the lowest overall growth (100% plant protein diet), also displayed the highest overall fry survival rate, although overall survival rates did not significantly differ among treatment groups. Increasing aggressive behavior among fry

PAGE 519

519 (pers. obs.) within replicate aquaria (10 subsample fry at trial inception) may have been responsible for some observed mortality. As fry grew and matured, aggression among tank mates may have increased due to the onset of sexu al maturity and/or physical crowding due to fry somatic growth. Lower overall growth of fry within the 100% plant protein diet treatment also may have resulted in lower rates of intra specific aggression due to the delayed onset of sexual maturity and/or less physical crowding due to nutritional differences among the applied experimental feeds. Resulting in a higher fry survival rate at harvest for the 100% plant protein diet treatment group, again, this difference in survival rate was not statistically significant. Indoor Trial Experimental Design Improvements Ten subsample fry were housed within each aquarium replicate at trial inception. Unfortunately, noticeable aggression among fry was observed in the last 2 to 3 weeks of the trial, resulting in nearly a dozen suspected mortalities (pers. obs .). Whether these mortalities resulted from overcrowding due to increased somatic growth and/or increased ontogenetic aggression is not known ( Sohn 1977, Hannes and Franck 1983, Hannes et al. 1984, Dionne 1985, Franck et al. 1985, Campton and Gall 1988, Ribowski and Franck 1993, Jones et al. 1998). Ideally and in possible future studies, it may be advantageous to have single fry housed within aquaria S urvival rates were relatively high (~80 90%) with ten f ry per aquarium, and would likely be higher with a single fish per tank. Additionally, aggressive behavior and social hierarchies may have led to reduced feed intake /stunting and mortality in smaller and less dominant individuals Using a single fish per replicate aquarium may eliminate these confounding factors, and increase the experimental rigor (experimental design orthogonality and statistical power) and statistical strength of any conclusions produced ( Sokal and Rohlf 1981, Zar 1984). Ideally, the s ingle daily feeding would have been equally divided into two or three meals spaced into morning and evening, or morning, midday and evening; multiple small meals

PAGE 520

520 typically result in better feed and protein utilization efficiency, and higher growth rates than a single daily feeding ( Boujard and Corraze 1995, Boujard et al. 1996, Boilliet et al. 2004, Glineau et al. 1996). Unfortunately, due to the large number of manhours required to provide weighed rations (0.001 gram accuracy) for daily feedings (~ 120 minutes), and alternate day tank cleaning (~ 45 hrs), fry were fed only once daily between 11:0013:00 hrs (EST). Fry were fed at a high initial daily ration level of 15% body weight per day (dry feed wt. /wet fry wt. x 100). This level of ration was greater than the 30 minute satiation level, which is the ability of fish within a tank to completely consume a meal within 30 minutes (Lovell 1998, Halver 2003). Lower initial ration levels (below 30 minute satiation level) would have been desirable as m aximal fish growth was not the primary objective of the indoor trial. The primary research objectives of the indoor trial were to produce maximum feed protein utilization in fry from five different feed treatments, and measure the resulting carbon and nit rogen isotope fractionation magnitudes between fry and their respective feed treatments, to validate feed source identification methods and calculate mean isotopic enrichment magnitudes. These objectives would have been more easily achieved and the result ing conclusions less limited without the confounding factor of ad libitum feeding, as equivalent nitrogen budgets may not have been established for individual replicates for all five experimental feed treatments; feed consumption rates may have differed am ong replicates and/or treatments, due to factors such as feed palatability, appearance, visual contrast with tank background, or tactile and olfactory cues. Unfortunately, the five experimental indoor feeds varied greatly in appearance (Figure 25). T he i sotopic nutrient tracing and enrichment magnitude measurement s for a single trophic leve l objectives were met with moderate success (Figure 2 21) A dditional objectives, such as feed quality evaluation of the experimental feeds would have been more easily achieved and the

PAGE 521

521 resulting conclusions less limited without the confounding factor of ad libitum feeding, as equivalent nitrogen budgets may not have been established for fry groups among all five experimental feed treatments. Feed consumption rates may have differed among replicates and/or treatments, due to factors such as feed appearance ( visual contrast with white tank background) feed tactile and olfactory cues, or swordtail behavioral issues (aggression, feeding hierarch ies), etc.; for example the five experimental indoor feeds varied greatly in appearance (Figure 2 5). Feed palatability also may have been a factor that reduced relative feed consumption rate s R educed consumption of feeds with high plant content has be en observed in many salmonid studies (Fowler 1980, Refstie 1998, Arndt et al. 1999, Regost et al. 1999, Yamamoto et al. 2007, Barrows et al. 2008). With subsatiation ration levels, the probability that treatment fish were consuming differing amounts of f eed would likely have been reduced. Outdoor Pond Production Trial Prior to harvest, mean fry growth was periodically monitored during the 12 week outdoor pond trial. At harvest, swordtail biomass was significantly greater within the two commercial feed tr eatment ponds than within the two fertilizer treatment ponds (Figure 3 9). Additionally, marketable fish (> 31 mm SL) numbers were highest within the two feed treatments, and the inorganic fertilizer ponds had significantly fewer marketable swordtails than those of the three other treatments (Figure 3 11) Pre harvest Fry Growth Five to seven weeks into the 12week outdoor pond trial, increased female broodstock recruitment from precocious female fry and higher fecundity due to increased water temperature led to an increase in the numbers of young/ small fry within study ponds, reducing mean pre harvest fry weights. Pre harvest fry mean weights differed among sampling dates within given pond nutrient treatments (Figures 3 5 38), but pre harvest fry mea n weights did not differ

PAGE 522

522 among nutrient treatments for given sampling dates. Unfortunately, the presence of any true differences in fry growth rates among nutrient treatments were obscured by the continuous production of multiple age cohorts; also, the ma thematical function describing fry growth could not be determined due to the presence of multiple cohorts and continuous spawning. A major problem that occurred during pre harvest fry sample collection was the tendency for larger, presumably older swordt ails to congregate near the surface at the center of the pond, which prevented their capture with the sampling method employed to capture pre harvest fry (wading along the pond perimeter with a dip net). Aggregations of larger fish were routinely observed at the surface, associating with the center aeration line that bisected ponds along the longest axis in the deepest section (Figure 31a,c,d). Fish at Harvest Lack of weight uniformity among large harvest fish within the INO ponds (Figures 3 13, 315), may have been a consequence of food limitation resulting in large numbers of stunted fish. Again, this would be an undesirable performance trait for an applied nutrient as uniform livestock size is desired by both wholesalers and retail buyers (Karplus et al. 2003) Under certain situations, large numbers of small unmarketable fish may actually be desirable as future marketable stock during times when fish livestock numbers are low and wholesale prices are high. If making a choice strictly between the two fertilizers evaluated in the outdoor pond trial, cottonseed meal application alone would produce higher marketable fish yields than the inorganic fertilizer, due to the inorganic fertilizers relatively poor fish production performance (Figures 3 9, 311). Among the different metrics used to analyze pond treatments at harvest (total biomass, total fish numbers, marketable fish numbers, unmarketable fish numbers), mean marketable fish numbers at harvest was the best measure of a ponds financial value ( Tamaru et al. 2001).

PAGE 523

523 Among the four treatments, the two feed treatments (PRO and UNP) significantly produced more marketable fish than the two fertilizer (CSM and INO) treatments (Figure 3 11). If commercial feed and cottonseed fertilizer application costs (costs of feed, fertilizer, and labor) are negligibly greater than inorganic fertilizer application, the harvest data tends to support the concerted use of commercial feed and periodic cottonseed meal fertilizer application. C ottonseed meal m ay enhan ce fish production by increasing the production of supplemental live feeds, which may be important as primary feeds for fry and young juveniles (Chapters 46) ; supplemental live feeds also may provide necessary growth factors (omega 3 and 6 fatty acids) th at increase somatic growth rates and broodstock fecundity, immune system boosters (vitamins, minerals, antioxidants), and color enhancing pigments that may increase ornamental fish appearance and marketability (Royes and Chapman 2003, Ezhil et al. 2008). Although the PRO and UNP marketable fish numbers at harvest did not statistically differ, the PRO treatment mean was approximately 500 (~ 20%) marketable fish per pond higher than the UNP treatment (Figure 3 11). As marketable fish numbers are the best measure of livestock value, this makes commercial feeds an obvious choice over fertilizer application alone to maximize production. Additionally, not only did PRO ponds produce the highest numbers of large marketable fish, PRO ponds also produced the highest numbers of small unmarketable fish; large numbers of small unmarketable fish also may be desirable, as livestock that can be harvested later when market conditions are better (supplies low and wholesale prices high) or empty ponds are available for further fry grow out (mover ponds). PRO and UNP treatments also displayed the lowest variations in individual large harvest (SL > 31 mm) fish weight (Figures 3 13, 315), which also is desirable, as tropical fish wholesalers prefer fish of uniform size, as well as appearance.

PAGE 524

524 Water Chemistry and Pond Physical Environment Water temperature increased steadily during the 12week trial (Figure 3 16 ) from late March to mid June. Increased water temperatures likely increased swordtail fecundity, growth rates and zooplankton and phytoplankton biomass, as well as pond respiration rates and r educed oxygen solubility in water ( Hall 1964, Boyd 1979, 1995, Milton and Arthington 1983, Kruger et al. 2001). Experimental Design Considerations and Further Research Swordtails (Family Poeciliidae) were not an ideal fish species choice for this study. The reproductive life history of this specie s made individual fish somatic growth and fry survival rate calculations impossible using standard life table generation methods (Pianka 1970, Pearl and Slobodkin 1976, Cailliet et al. 1996, Hutchings 1993). Reproductive and survival rate calculations wer e not possible, due to the continuous reproduction of swordtail broodstock, and reproductive recruitment of older, precocious fry at approximately five to seven weeks of age, (Figure 3 4). S tocking ponds with known numbers of eggs, or fry from a highly f ecund egg laying fish species with an age of first reproduction greater than the 12week study duration, would have allowed the growth of a single fry cohort to be followed via periodic sampling prior to harvest (single age cohort), as well as allowed the calculat ion of fry survival rates at harvest. Additionally, genetic diversity could have been reduced by using highly inbred broodstock to produce the fry/eggs used in the outdoor pond trials, increasing treatment effects on fish size and production, whil e reducing size variation due to genetic and age differences among fry/eggs. By using sexually immature fish of known age, and prior knowledge that excess bioenergetic resources (beyond maintenance, activity, and specific dynamic activity needs) are largely

PAGE 525

525 channeled into somatic growth, stronger conclusions could be made regarding preferred pond nutrient treatment choice s An important advantage of having a no applied nutrient /extensive aquaculture control treatment would have been the possession of a ct ual baseline fish production information. Baseline extensive management fish production information could have provided a means by which to evaluate absolute fish production performances of the various applied nutrient treatments (Chapter 3, relative fish production) Because the studys experimental design lacked a no applied nutrient control treatment, only relative Regardless of fish production increases resulting from human mediated nutrient addition, the relative contributions of autochthonous (recycled nutrients from previous production cycles, regeneration of nutrients from sediments) and non a nthropogenic allochthonous (groundwater nutrients, windblown dust, leaves, terrestrial insects, etc .) nutrient sources to total pond fish production could still not be measured and evaluated solely by the addition of a no applied nutrient control treatme nt. Further information regarding nonhuman mediated allochthonous nutrient inputs [i.e., magnitudes (mass balance) and isotopic signatures of nitrogen within windblown dust and groundwater DIN inputs] would still need to occur before the relative fish production rate comparisons could be made among the different experimental treatments (Chapter 3) Fish production resulting solely from applied nutrient application (absolute fish production) could have been easily calculated for each treatment by subtracting mean baseline no applied nutrient fish production values from the mean gross fish production value s measured for each of the four pond treatments (Chapter 3). Because baseline extensive aquaculture management scenario fish production values were not available, the actual increase in fish production realized from anthropogenic nutrient addition is some what conjecture.

PAGE 526

526 impacts of autochthonous and nonanthropogenic allochthonous nutrients sources upon pond fish production could be properly evaluated. Isotopic Investigation of Pond Trophic Dynamics Carbon and nitrogen isotope signatures of four applied nutrient treatments were traced to three plankton size assemblages and swordtails within 24 aquaculture ponds. Applied nutrients pri marily moved in parallel to swordtails and plankton assemblages within the two feed treatments, serially within the inorganic fertilizer treatment (fertilizer phytoplankton zooplankton fish), and intermediate between parallel and serial within the organic cottonseed fertilizer ponds. Additionally, swordtail yields at harvest reflected these nutrient movement patterns, with more direct movement of nutrients into swordtails (applied commercial feed treatments), producing greater overall swordtail yield s at harvest (Figure 4 1). Poor fish yields observed within the inorganic fertilizer treatment (Chapter 3), were reflected in the greater number of trophic levels believed to separate swordtails and the applied basal nutrient indicated by isotopic tracing. Pond Nutrient Isotope Signatures As expected, processed and unprocessed feed carbon and nitrogen isotope signatures did not differ due to their manufacture from identical ingredients. Applied cottonseed meal isotope signatures ( C, N) differed from both feeds and from the inorganic fertilizers nitrogen signature. The inorganic fertilizer nitrogen signature differed from the other applied n utrients, and although the inorganic fertilizer did not contain carbon, INO treatment taxa carbon isotope signatures differed noticeably from those of the other three treatments. Fortunately, these differences in nutrient carbon (or INO pond DIC) and nitrogen isotope signatures produced contrasting isotopic signatures among the respective pond taxa, of the different pond nutrient treatments. Isotopic signature differences among pond nutrient treatments for given taxa was

PAGE 527

527 necessary to determine whether applied nutrients had altered pond taxa isotope signatures in the absence of baseline taxa isotopic signatures from a no applied nutrient control treatment. Although isotopic signature differences among treatments within a given taxa were not neces sary to investigate trophic dynamics occurring within nutrient treatment ponds; within nutrient treatment trophic dynamic investigations required that taxa from different trophic levels exhibit relative isotopic signature differences. A llow ing an applied n utrient to be traced as it progressed through its respective food webs/chains. Applied nutrient isotope signature differences and differences among treatments for given taxa, helped reduce the magnitude of the problem created by the lack of baseline ta xa isotope signatures that would have been provided by a no applied nutrient control treatment. Unfortunately, applied nutrient isotopic signature differences did not address a major problem created by this lack of baseline taxa isotopic signature infor mation, which was whether anthropogenic nutrient applications were of sufficient magnitude and/or nutrients were in chemical forms suitable to produce isotopic differences within given taxa, among the four different treatments (via primary producer assimilation). Whether applied nutrient applications were capable of isotopically altering the various pond taxa within the different pond treatments, was uncertain at the onset of the outdoor pond trial portion of the study. A lack of isotopic signature diffe rences within individual food web guilds among the different pond nutrient treatments ( Figure 418; e.g., PRO and INO large zooplankton 15N values ) would not have made it possible to reject the null hypothesis that applied nutrients had no effect upon t he isotopic composition of the food web components of their respective treatment ponds This would i ndicat e that isotopically contrasting applied nutrients had not been assimilated by pond taxa to an appreciable degree, either because applied nutrients w ere not in a

PAGE 528

528 suitable chemical form and/or applied nutrients had not been present in sufficient quantities to produce any noticeable effects, and allowing for the remote possibility that autochthonous or nonanthropogenic allochthonous (e.g., groundwater N O3 -) nutrient sources were primarily responsible for biomass production within the treatment ponds, rather than the applied nutrients. As a result, only relative If taxa did not isotopically differ within a given treatment, than we could not have rejected the null hypothesis applied nutrients do not alter pond trophic dynamics. It would remain uncertain whether applied nutrients were in the proper chemical form and/or amounts necessary to alter pond trophic dynamics or responsible for any differences in fish production among pond nutrient treatments. trophic interaction conclusions regarding taxa from different trophic levels within individual treatments can be made using taxa stable isotope signatures. If no applied nutrient control treatment baseline taxa isotope signature values had been available, experimental treatment taxa isotope signature differences with those of the no applied nutrient control taxa could have been used to determine trophic processes likely occurring within the various experimental treatment ponds, even if given taxa isotopic signatures had not differed among the applied nutrient treatments (e.g., PRO and INO large zooplankton 13C and/or 15N ). If pond taxa isotope signatures had not differed w ithin given treatm ents, then identification of functional trophic processes occurring within the different pond treatments could not have been easily made using the stable isotope signature tracing methodology. This would have prevented answer ing two questions central to t his study: (1) what trophic processes are occurring within the different treatment pond? (e.g., are feeds driving fish production within the feed treatment ponds ? ) and (2) are nutrients moving directly and efficiently to the target species or are nutrient s moving inefficiently through several trophic intermediaries before (or even if)

PAGE 529

529 becoming sequestered within the target species? O nly empirical measurements (e.g., fish production rates) could have been used to evaluate treatment efficacies, and then only if significant differences in harvest fish production were present among treatments. This would still have left unanswered these two questions central to this study As previously mentioned, the problems created by the lack of external (absolute sta ndard) baseline taxa isotope signatures, were largely offset by the presence of contrasting isotopic signatures of individual taxa among treatments. The nitrogen and carbon isotope signature contrast s provided by the INO treatment food web components rela tive to those of the other three nutrient treatments, acted as an internal standard control (relative standard) for the outdoor trial isotope tracing methodology ( Sawyer et al. 2003 ). The often contrasting isotopic signatures of given taxa from differen t pond treatments (e.g., PRO and INO large plankton assemblage 15N), implies that some (if not all) of the applied nutrients altered the isotopic signatures of taxa, and by proxy, the trophic dynamics of the ponds in which they were applied. It remains a possibility that liquid inorganic fertilizer application did not result in increased fish production. It is possible that the applied inorganic liquid fertilizer nitrogen was not in the proper chemical form and/or in sufficient quantities to alter INO treatment pond trophic dynamics L iquid inorganic fertilizer may have immediately become bound within the sediments following application and therefore unavailable to pond primary producers. H aving greater isotopic contrasts among the applied nutrient sources would have been desirable, in that it would have allowed greater confidence to be placed in the food web interpretations and conclusions. However high isotopic contrasts were not necessary to successful ly trace pond treatment nutrients, as naturally contrasting isotopic signatures within the selected applied nutrient sources appeared to have been adequate. Additionally, although

PAGE 530

530 a rtificially enriched or depleted nutrient sources were available, their high cost and the large quantities requir ed for the present study made their use prohibitive. Production biomass and i sotopic measurement of incidental taxa (crayfish, tadpoles, large insects, and snails) and pond sediments also would have been useful in identifying some of the more conspicuous unintended fate s of applied nutrients This also may have allowed nutrient losses resulting from pest species production to be measured, and the identif ication of which species were the largest nutritional drains /sinks upon pond food webs, other than the intended swordtail target species. If given incidental fauna were isotopically identified and present in sufficiently large numbers (e.g., over a selected biomass threshold) then pond management countermeasures might be implemented to reduce the impact of the se more problematic species, and hopefully increas e swordtail fish production. Unfortunately, due to time and budgetary issues, incidental pond fauna were not isotopically examined nor were their production rates. In addition to the desire for bas eline fish production (Chapter 3), fish and plankton assemblage isotopic signature information, having large zooplankton community composition and relative abundance information (Chapter 5) in the absence of fish predators (no fish control treatment), al so might have been useful in providing important insights into pond food web dynamics. Adding these two control treatments (no applied nutrient and no fish ), would have allowed a better evaluation of the effects of typical aquaculture industry pond ma nagement practices on pond food web dynamics and fish production rates. A no fish control treatment would have provided large plankton assemblage biomass and community information in the absence of swordtail production. Differences in large plankton as semblage communities in the presence and absence of swordtail predation may be used to identify preferred macro zooplankton taxa prey Pond management strategies might then be altered to enhance production

PAGE 531

531 of preferred macro zooplankton prey, potentially increasing swordtail production. Regardless, strong inferences regarding pond food web dynamics in the absence of this information were possible in the absence of no nutrient and no fish control treatments Due to anatomical (e.g., pharyngeal teeth), life history (e.g., continuous reproduction mixed size/age cohorts), and ecological (e.g., omnivorous generalist versus zooplanktivore specialist) aspects of the model animal, the determination of functional trophic processes that occurr ed within the various treatment ponds was addressed using pond taxa isotopic signatures. Understanding functional trophic processes occurring within aquaculture ponds allow s the prediction of relative production efficiencies of different pond management strategie s (feeds, fertilizers, or both) Functional process models also can tell why certain strate gies are better at producing fish than others which from a scientific viewpoint is much more interesting and satisfying than what can be learn ed from a simple emp irical model. Dynamical e cosystem trophic functional models were the primary objectives of this study. However, empirical models and measurements are still vital in determin ing what has actually occurred (input and output rates) within an aquaculture pond system and in evaluat ing the validity (ground truthing) of different functional ecosystem models. Pre Harvest Fish Isotopic Signatures Limited isotopic analyses of consecutively sampled (weekly) pre harvest fish followed expected trajectories, initially similar fry isotope signatures diverged among treatments over time, and fry signatures stabilized about mean values for latter sampling dates within treatments. Fry from the two feed treatments did not differ isotopically (C and N) for any of the three sampling dates analyzed, most likely because the two applied feeds were made from the same ingredients. Mean fry weight gains of 650 to 1,150 % initial weight (Chapter 2; indoor trial

PAGE 532

532 initial fry weight value) were necessary before C signatur e differences between fry from different treatments were detectable between the PRO and INO, and UNP and INO treatments, which occurred at approximately three weeks post stocking. At two weeks post stocking, pre harvest fry weight gains were much lower (C hapter 2; 200 to 250 % initial fry wt) and fewer pairwise carbon isotope signature differences between treatment fry were present (i.e., PRO and INO, and UNP and INO C did not differ at two weeks post stocking). This was roughly half of the minimum bio mass gain necessary (~ 4X initial weight) to be isotopically detectable following diet switching ( Fry 2006, T. Frazer pers. comm.). During the initial weeks of the study, feed treatment pre harvest fry may have been able to consume greater quantities of t he finely ground applied feeds due to low predation pressure from the lower density of adult fish (250 broodstock/pond) present at the onset of the study, and lower overall swordtail biomass, which may have increased feed availability for small fry and young juvenile fish due to low er competition for applied feed from within and amongfry and juvenile cohorts The previously hypothesized greater fry and juvenile swordtail dependence upon live foods, may have taken several weeks to develop as large zooplankton standing stocks increased over time (Figure 5 1), along with the increasing fry, juvenile, and adult swordtail numbers and biomass observed at harvest ( Figures 3 9 3 10). Increasing competition over time for feeds from other fry, juveniles, and adults, and increasing potential predation pressure (cannibalism) from adults, may have caused microhabitat shifts for both fry and juvenile swordtail s to the more vegetated areas along pond perimeters, and a trophic shift to epiphytic and planktonic live foods due to reduced availability of applied feed in these areas.

PAGE 533

533 Harvest Fish Isotopic Signatures As expected, large harvest fish isotope signatures of the two feed treatments did not differ for either carbon or nitrogen. This was likely due to their consumption of the chemically and isotopically identical ingredients used to manufacture the processed and unprocessed feeds. Larger fish were regularly observed feeding (pers. obs .) upon the applied feeds and finely ground cottonseed meal at the surface in the deeper, central portions of the nonINO treatment ponds. Within the nonINO treatment ponds, s maller fish were rarely seen at the surface and may not have been able to exploit the applied feeds and cottonseed meal as readily as larger fish. Casual observations of differences in feeding behavior and microhabitat usage between large and small fish, such as feeding at the surface, water depth, and association with plant cover may have combined with morphological limitations such as maximum mouth gape and food particle size differences to produce the nitrogen isotope signature differences observed between co occurring small and large fish in the nonINO treatment ponds. Additionally, during periodic pre harvest fry sampling using a large, fine mesh va t net while wading within the shallows along pond perimeters, few adult fish were captured ( pers. obs.). Only the INO large harvest fish carbon isotope signature differed from those of the other three treatments. This was interesting in that only the INO treatment food webs carbon sources were derived from autochthonous and nonanthropogenic allochthonous s ources that apparently had a composite carbon isotope signature differing greatly from those of the nonINO treatment nutrients. Carbon dioxide in the atmosphere, aqueous solution, and ground water carbonates were the most likely sources of inorganic carb on for photosynthetic organisms within the INO treatment ponds ( Hall 2004). Additionally, carbon released through the decomposition of organic material (leaves, cellulose, dead plants and animals) and carbon dioxide released during the respiration of pond organisms also contributed to the dissolved inorganic carbon pool

PAGE 534

534 utilized by INO treatment pond primary producers. P rimary production resulting from dissolved inorganic carbon (DIC) fixed by pond phytoplankton and submergent and emergent vascular plants, formed the base of the food web within the INO treatment ponds ( Smith and Walker 1980, Rounick and Winterbourn 1986, Peterson and Fry 1987, Zohary et al. 1994, Takahashi et al. 1999, Vander Zanden and Rasmussen 2001). Within the nonINO treatments, a pplied nutrient carbon isotope signature differences were the result of the differing origins of their constituent ingredients: plant/animal, terrestrial/aquatic. Terrestrial cereal crops such as corn and wheat are C4 monocotyledonous plants ( C4 = 13 ; Boutton 1991), while cotton ( Gossypium hirsutum ) and soybean ( Glycine max) are C3 dicotyledonous plant (C3 ~ 27 C ; Boutton 1991). The c ommercial feeds were manufactured using a combination of both C3 ( soybean meal) and C4 plants [ rice bran ( Oryza sativa ) and wheat middlings ( Triticum spp .) ] that resulted in composite carbon isotope signature values of ~ 24 C. The c ottonseed meal carbon isotope signature value was 26.43 ( C), which is close to the value reported for C3 ( 27 C) plants in the literature ( Minagawa and Wada 1984, OLeary 1988, Finlay and Kendall 2007 ). Organic carbon within the INO treatment trophic guilds was most likely derived from INO pond DIC f ixed via the C3 pathway utilized by freshwater green algae ( INO small plankton C = 23.26 ) Large harvest fish nitrogen isotope signatures did not differ among treatments, which was somewhat surprising given the conspicuous nitrogen isotope signature differences among the four applied nutrient treatments. How ever, given that nitrogen isotope signature enrichment rates are much larger and typically more variable than carbon isotope enrichment rates between trophic levels, the lack of nitrogen isotope signature differences at the highest trophic levels are under standable ( Kling and Fry 1992, Fry et al. 1999, Vander Zanden and Rasmussen 1999, Fry

PAGE 535

535 2006.). Additionally, the presence of more diffuse trophic food webs as opposed to simple linear food chains within the two fertilizer treatments likely contributed t o the lack of nitrogen isotope signature differences within large harvest swordtails Nutrients within the ponds of the two fertilizer treatments, moved through more trophic levels than within those of the two feed treatments prior to becoming assimilat ed by swordtails within their respective treatment ponds. The low inorganic fertilizer nitrogen isotope signature (~ 0.00 N) was enriched via the inorganic fertilizer pond food web to such a degree that INO large swordtail signature values did not statistically differ from those of the two feed treatments. An initial nitrogen isotope signature difference of ~ 4.7 N (4.61 N PRO/INO, 4.88 N UNP/INO) existed between the two feed treatments and the inorganic nitrogen fertilizer. The fewer isotopic differences present among higher trophic level taxa than among lower taxa within the four treatments w ere likely due to the attenuation of isotopic differences (tracer signals) present among the applied nutrients. Tracer signal attenuation, likely resulted from zooplankton omnivory, zooplanktivore predation upon multiple trophic levels, predation within (e.g., medium size rotifers consuming medium size phytoplankton) and between plankton assemblages, including can nibalism (e.g., adult copepod predation upon copepod nauplii), as well as omnivory and cannibalism among swordtails ( Dionne 1985, Jones et al. 1998, Tamaru et al. 2001). Cannibalism among swordtails may have been particularly common within the INO treatm ent, as indicated by the large nitrogen isotope signature difference (trophic difference) between the two INO harvest fish size classes compared to those of their non INO treatment counterparts. INO ponds had higher percentages of smaller, possibly stunte d fish (Chapter 3; Figure 312 ). The greater coefficient of variation (Chapter 3; Figure 3 15 ) among individual

PAGE 536

536 INO large harvest fish mean weights relative to those within the other three treatments, also may suggest low prey availability within the INO ponds compared to the other treatments ( Holbrook and Schmitt 1992, 1994, Holbrook and Unwin 1997, Mittlebach et al. 1999). Greater predation on limited planktonic live foods may have increased swordtail cannibalism rates within the INO ponds. Alternately, higher nitrogen isotope signature differences between INO small and large harvest fish (Figure 4 42) may have occurred due to the complete reliance of INO swordtails upon live foods, and fish size stunting and high size variation resulting from food lim itation. Lower trophic level estimates for INO small harvest fish relative to INO large harvest fish, may have been due to ontogenetic dietary differences arising from ontogenetic microhabitat and morphological differences, such as mouth gape limitations. Food limitation may have reduced the trophic niche breadth of INO swordtail size classes, as well as producing a trophic shift among smaller fry and juveniles toward smaller, trophically lower prey items, resulting in greater swordtail size class nitrogen isotope signature (and N derived trophic distance estimates) differences within the INO relative to the CSM treatment. The high population percentage of small fish within the INO ponds relative to the CSM ponds, may have resulted in greater swordt ail food limitation, within the INO treatment ponds due to lower small plankton availability [ Figures 311 312 (Chapter 3), 4 52]. Departures from a strictly linear trophic food chain (omnivory, cannibalism, etc.) may have weakened or even obliterated i sotopic tracer signal differences present at the base and lower trophic levels as they progressed to the higher trophic levels within the ponds. This was evident among the harvest fish size classes, as small harvest fish isotopic signature differences amo ng the four treatments, were no longer present among their large harvest fish counterparts.

PAGE 537

537 Small Harvest Fish Isotope Differences Unlike INO large harvest fish, which only differed from their non INO treatment counterparts for carbon isotope signatures, INO small harvest fish differed from their non INO treatment counterparts for both carbon and nitrogen isotope signatures. Similarly, large, medium, and small plankton assemblages displayed nearly identical carbon and nitrogen isotope signature difference patterns among treatments feed treatment taxa and harvest fish did not isotopically differ, but feed treatment taxa frequently differed from those of the two fertilizer treatments, which frequently differed from each other A gain, this was likely due t o the manufacture of the two feeds from identical materials. This suggests that, in addition to feed treatment swordtails, feed treatment plankton assemblages also were isotopically influenced by applied feed nutrients; possibly due to direct feed consumption by larger macrozooplankton, and remineralization and subsequent assimilation of feed nutrients by feed treatment pond primary producers and any heterotrophs that nutritionally utilized them within the medium and small plankton assemblages. Swordtail Ontogenetic Dietary Differences As previously mentioned, fry and juvenile swordtail microhabitat associations with dense plant cover along pond perimeters may have been due to factors such as lower adult swordtail predation pressure and greater food availability. Appropriatelysized prey associated with plant s, may have been limited due to mouth gape limitations of fry and juvenile swordtails As a result small harvest fish may have obtained a greater proportion of their nutrition from smaller, lower trophic level live foods (e.g., greater omnivory) within the vegetation than large harvest fish that congregated at the surface in the non vegetated deeper areas of the ponds. Small and large harvest fish carbon iso tope signature differences followed similar patterns among treatments: nonINO harvest fish carbon isotope signatures were lower than their INO

PAGE 538

538 counterparts. This indicated that INO harvest fish, and the taxa they either directly or indirectly (e.g., phyt oplankton) consumed, utilized a DIC basal nutrient pool distinct from those utilized within the non INO treatments Additionally, the INO DIC pool acted as a quasi no applied nutrient control for carbon isotope signature values from the nonINO treatmen ts, as this carbon was not anthropogenic in origin. This also provided circumstantial evidence that nonINO nutrient applications were of sufficient magnitudes, and in forms (feeds, seed meal) sufficient to alter carbon isotope signatures among the variou s taxa within the nonINO treatment ponds. Paradoxically, nonINO large harvest fish nitrogen isotope signatures were lower than their small fish conspecifics, s uggesting that non INO large harvest fish were trophically lower than nonINO small harvest f ish which was a non sequitur conclusion based upon what is know n of fish biology and aquatic ecosystems. This could have occurred if larger fish within these treatments were primarily utilizing isotopically lower (15N) nutritional sources (applied feeds and cottonseed meal), while fry and juvenile swordtails were heavily utilizing live pond prey (plankton assemblages) that also were utilizing these same applied nutritional sources. Potential live prey may have been directly consuming applied nutrients or have been isotopically influenced by applied nutrients via primary producer assimilation of DIC and DIN derived from remineralized applied nutrients. This was supported by the finding that plankton assemblage isotope signatures (medium and large plankton) were frequently enriched (15N) relative to the applied nutrient treatment (Figures 4 39 442). If ontogenetic differences in resource utilization/partitioning were occurring for non INO swordtails for applied pond nutrients (feeds and cottonseed meal ) and live pond foods T his could have resulted in small ( swordtails within the non INO ponds being more isotopically enriched than larger swordtails (>

PAGE 539

539 31 mm SL), resulting in smaller fish being perceived as being at a higher trophic level than larger individuals. In contrast, due to the chemical form of the inorganic nitrogen fertilizer used within the INO ponds, INO small harvest fish were obtaining their nutrition from live foods Because primary producers were assimilating the inorganic fertilizer nitrogen within the INO ponds, heterotrophs that utilized this primary production and those occurring further up the food chain (including swordtails), became further enriched with each increasing trophic level. As a result of the INO pond food web structure the INO large harvest fish (apex predator) displayed the greatest nitrogen isotope signature difference (trophic distance) relative to the applied inorganic nitrogen fertilizer, followed by INO small harvest fish. This was the expected pa ttern of nitrogen isotope signature increase from lower to higher trophic levels within the INO ponds (monotonous increase in 15N signature). In summary, higher small harvest fish isotope signatures relative to large harvest fish within the non INO treatments were primarily due to the parallel movements of applied nutrients into the various pond taxa. Greater dependence of small fish upon live foods within the nonINO treatments may explain their h igher isotope signatures relative to their large harvest fish counterparts. This gave the false impression that non INO small harvest fish were at higher trophic levels than nonINO large harvest fish, which was not possible given what is know n of fish fe eding behavior and ecosystem dynamics. In contrast, the serial movement of applied nutrients through the various pond plankton taxa and possible swordtail cannibalism likely explain INO small harvest fish having lower isotope signatures than INO large har vest fish. This was an interesting finding in that it clearly illustrated a major flaw in relying solely upon

PAGE 540

540 isotopic signature results when interpreting trophic dynamics within an aquatic ecosystem and the need for additional, corroborating trophic ana lyses methods (Chapters 5 6). Plankton Isotope Signatures Non INO plankton assemblages had lower C signatures than their INO counterparts. This again indicated that INO organisms were utilizing a different carbon source (DIC) than those within the nonINO treatments applied nutrients and/or DIC derived primarily from applied nutrients. The finding that the INO plankton assemblages ( small medium large ) had much lower nitrogen isotope signatures than their counterparts in the other three pond nutrient treatments was reasonable if the applied inorganic nitrogen was first assimilated by pond primary producers, which were then utilized by heterotrophic organisms. Inorganic fertilizer was the presumed nitrogen source for the INO food webs, and had a much lower nitrogen isotope signature than the applied nutrients of the other three treatment groups ( INO 0.08 N ) due to its manufacture from atmospheric nitrogen feedstock ( Haber 1920 ) The lower nitrogen isotope signatures observed in the INO pond trophic guilds ( plankton and harvest fish ) were likely a reflection of the lower nitrogen isotope signature of the INO food web basal nutrient (inorganic fertilizer). This provid ed in direct evidence that INO pond organisms utilized the applied inorganic f ertilizer. There is a remote possibility that the primary nitrogen source driving production within the INO treatment ponds was actually derived from the fixation of atmospheric nitrogen via cyanobacteria (blue green algae). If the primary nitrogen sour ce within the INO food webs was derived from blue green algae production, the nitrogen isotope signature of the resulting taxa would have values similar to those observed using the inorganic liquid fertilizer; again, due to the manufacture of the applied i norganic liquid fertilizer from atmospheric nitrogen. However,

PAGE 541

541 this does not appear likely as conspicuous mats of blue green algae were not observed during the 12week pond trial. A heavy infestation of filamentous green algae occurred in one pond (pers. obs.), but was quickly controlled (~2 days) by mechanical removal. Herbicide was then applied to all ponds as a preventative measure (prevent reinfestation) early in the study (diuron during week 4). I t was still possible that the applied inorganic nitr ogen fertilizer did not alter pond primary production rates and was not utilized by pond taxa, due to the chemical form and/or amount of fertilizer applied to ponds. Applied nutrients may have become bound to bottom sediments and become unavailable to pond heterotrophs. In the absence of a no applied nutrient control, it remains uncertain whether INO treatment pond swordtail production rates are higher than those that would be realized without human mediated nutrient additions. Similarly, due to potent ial lack of inorganic fertilizer assimilation by INO pond autotrophs, it remains unknown whether INO treatment taxa isotopic signatures would differ from those of no applied nutrient control treatment ponds. It was assumed that for each trophic level oc curring within the INO ponds from primary producer (small plankton assemblage) to swordtails, that the nitrogen isotope signature of the subsequent trophic level would become further enriched in 15N (~ + 3.03 N per trophic level; Chapter 2). INO tr ophic guild nitrogen isotope signatures did not become enriched to the point where they differed from those of their non INO counterparts, until applied inorganic fertilizer nitrogen had reached the top predator that we isotopically examined, and had becom e sequestered within the INO large harvest fish (Figure 4 11). These observations (pond taxa isotope signatures/lack of cyanobacteria growth) support the hypothesis that the primary inorganic nitrogen pool utilized within the INO ponds originated

PAGE 542

542 from th e applied inorganic fertilizer. They also suggest that the somewhat arbitrarily selected plankton size assemblages (via mesh sizes chosen by the author) partly coincided with actual trophic assemblages. To some degree within all ponds, applied nutrients were consumed or assimilated by trophic groups (plankton), which in turn may have been consumed by higher trophic level heterotrophs (fish, zooplankton). In general, applied nutrients and lower trophic groups were assimilated or consumed by successive tr ophic groups higher up the food web. This was indicated by the carbon and nitrogen isotope signatures ( Figures 4 38 4 42) of these various trophic groups (applied nutrients/plankton assemblages), and trophic distances estimates derived from isotope signature difference values between trophic groups (Table 4 4; e.g., feeds and feed treatment swordtails). Interesting further research would have been to continue monitoring the isotopic signatures of plankton size assemblages after the cessation of applied p ond nutrients T o observe if and when isotopic signatures again converged to a common set point value or remained on different deterministic trajectories (D Canfield pers. comm.). Harvest Fish and Applied Nutrient Isotope Differences Carbon isotope signature differences between large harvest fish and their applied nutrients were lower within the two feed treatments relative to the cottonseed meal fertilizer treatment (applied inorganic fertilizer did not contain carbon). However, carbon isotope signature differences between small harvest fish and their applied nutrients did not differ among treatments. This may have been due to the high variation in carbon isotope signature results, relative to the low carbon isotope signature enrichm ent magnitudes typically observed between successive trophic levels (~ 0.5 13C per trophic level) within an ecosystem. Unfortunately,

PAGE 543

543 these factors combine to make trophic relationship and distance estimates, based upon carbon isotope signature differences, broad and open to interpretation. This is cited as a major limitation inherent to using carbon isotope signature results alone, as an analytical tool in feeding habit studies, stressing the need for trophic relationship conclusions to be based upon both carbon and nitrogen isotope signature results. C arbon isotope signature results are better suited for tracing food web linkages and identifying potential nutrient sources rather than taxa trophic positions ( DeNiro and Epstein 1978, Minigawa and Wada 1984, Vander Zanden and Rasmussen 1999) Greater nitrogen isotope signature differences between trophic guilds relative to differences measured for carbon isotope signature are expected given that nitrogen isotope signature enrichment is typically on the order of six to seven times that observed for carbon isotope enrichment for a single trophic level increase ( DeNiro and Epstein 1981, Peterson and Fry 1981, Fry 2006). The nitrogen isotope signature differences between large harvest fish and the two applied fertilizer treatments were much greater than those occurring within the two feed treatments, and more so within the INO than within the CSM ponds. These nitrogen isotope signature r esults were the basis for the conclusion that applied feeds were directly consumed by fish and possibly by larger zooplankton, and a basic pattern of parallel nutrient movement occurred within the feed treatment ponds. Applied inorganic fertilizer nutrien t w as moving serially (becoming further enriched with each trophic level) within the INO pond food web. Additionally, INO treatment nutrients were moving serially [via a trophic cascade: inorganic N2 phytoplankton small zooplankton (rotifers, nauplii) large macro zooplankton (copepods, daphnids) swordtails] among pond taxa to a greater degree than nutrients within the CSM ponds. This was i ndicated by the larger nitrogen isotope signature differences between INO treatment taxa and the applied

PAGE 544

544 inorganic fertilizer, as well as the higher overall variation in nitrogen isotope signatures among the INO treatment taxa ( Figures 429 430, 44 3). H igh protein cottonseed meal may have been acting as both a directly ingestible (and assimilated) feed for sw ordtails (pers. obs.), and via its decay, a s a green manure which was enhancing swordtail live food production (via CSM food web). Due to the intermediate nitrogen isotope signature differences between small and large CSM swordtails and their applied organic basal nutrient relative to those of the other three trea tments, a nutrient movement pattern intermediate to parallel and serial appears to be occurring within the CSM pond food webs. Based upon applied nutrient and pond taxa isotope signature values, functional trophic processes occurring within treatment ponds were dependent upon the form of the applied pond nutrient. These patterns appeared to apply more for large fish than for smaller fish and fry, which may have been more reliant upon live planktonic and epiphytic foods occurring within the ponds, regardle ss of treatment (e.g., processed feed treatment small fish). Static isotope signature food web models assume isotopic equilibrium between predator and prey groups, and do not take into account isotopic signature changes occurring within both prey and predators ( Ben David and Schell 2001 ). Predator and prey isotopic signature changes can result from such factors as seasonal plankton community changes ( e.g. plankton community succession), predator and prey reproductive status, prey switching due to changes in prey abundance and nutritional quality, and ontogenetic changes in predator feeding behavior (e.g., predator mouth gape), among others (Holbrook and Schmitt 1992, 1994, Cailliet et al. 1986 ). Unfortunately, dynamic isotope signature food web models, w hich are more robust in accurately predicting and explaining stable isotope signatures and their changes over time, are also more complex and require additional analyses that were not available in this study.

PAGE 545

545 Isotopically Derived Fish Diet Scenario Issues T he CSM medium plankton assemblage carbon and nitrogen isotope signatures placed it directly upon a line (collinear) midway between the carbon and nitrogen isotope signature coordinates of the CSM large zooplankton assemblage and applied cottonseed meal ( F igure 441). This made it difficult to determine whether CSM harvest fish were primarily consuming prey from the medium size plankton assemblage or roughly equal proportions of large zooplankton and the applied cottonseed meal, as both of these dietary st rategies would result in similar large harvest fish isotopic signatures ( Fry and Sherr 1984, Phillips 2001, Phillips and Gregg 2001, 2003, Phillips and Koch 2002). Direct observation of CSM large swordtails feeding directly upon applied cottonseed meal and that large swordtails and large macro zooplankton aggregations cooccur at the pond surface within the central portions of the CSM ponds, tends to support the conclusion that large swordtail diet within the CSM ponds consist primarily of roughly equal pr oportions of large macrozooplankton and directly ingested (and partially assimilated) cottonseed meal. INO swordtail diet estimations derived from INO taxa isotope signature results, also m ight have encountered the isotopic collinearity problem (Figure 442), due to the isotopic resemblance of the INO medium plankton assemblage to that of a roughly equal mixture of INO large zooplankton and small plankton. An INO swordtail diet consisting primarily of medium plankton or one consisting of roughly equal parts large zooplankton and small plankton would be isotopically similar. However, the presence of isotopic collinearity among the small, medium, and large plankton assemblages within the INO treatment appeared to be a moot point as the isotopic plot positions of small and large swordtails were not in close proximity to the INO small and medium plankton assemblage plot positions (Figure 4 42).

PAGE 546

546 When given a choice between equally valid (isotopically) alternative diets, the determination of which diet is m ore likely must rely upon supplemental, nonisotopic information, e.g., direct observation of feeding behavior, gut content analysis, lipid analysis, etc. Unfortunately, not all of these methods were feasible given the biology and anatomical characteristi cs of the model animal, as well as a limited research budget. Plankton Assemblage Trophic Identity Issues The plankton size assemblages analyzed for the study, were chosen due to a combination of commercially available filter mesh sizes, information gathered from researchers in the scientific field (M. Cichra pers. comm.), and from the published literature ( Pen nak 1989, APHA/AWWA 1999 ). Mesh sizes were not chosen according to known size thresholds for phytoplankton versus zooplankton assemblages, which did not actually exist due to the large overlap in phytoplankton and zooplankton cell sizes ( pers. obs.; Chapter 5). The plankton assemblage derived from the largest mesh size consisted primarily of large macro zooplankton ( with some phytoplankton present; Chapter 5), while the plankton size assemblage resulting from the smallest mesh sizes consisted primarily of phytoplankton, and the intermediate medium plankton size assemblage consisted largely of a mixture of larger phytoplankton, rotifers, zooplankton eggs, and copepod nauplii. This may have been the primary reason that clear cut isotopic thresholds/functional trophic guilds (e.g., 1, 2, 3 consumer) were not observed between the different plankton size class assemblages, as there were no di scernable size class cut off thresholds for trophic guilds and the fact that assemblages changed over time (Chapter 5). Additionally, cursory microscopic examination of samples from the medium plankton assemblage, revealed that many of the individuals present in the medium assemblage were smaller, younger zooplankton of the same taxa as those found in the large assemblage [e.g., copepod nauplii (medium plankton assemblage), and adult copepods (large plankton

PAGE 547

547 assemblage)], further blurring the taxonomic, trophic and isotopic divisions among the three plankton size assemblages. In contrast, isotopic signature values of fish, pond nutrients and to a lesser degree the large zooplankton assemblage, appeared to be more clearly defined and assumed values expected for trophic groups from simple predator prey systems. This was probably due to the more compartmentalized feeding ecology of a single taxon (fish) versus the multiple taxa (and likely trophic positions) within the different plankton size assemblages. The gross definitions of large zooplankton, medium zoo plankton, and small phytoplankton used in this study were actually a gradation of organism sizes from varying trophic levels. Additionally, many taxa of the large zooplankton assemblage were likely consuming organisms from multiple trophic levels (i.e,. primary producers, primary consumers, and secondary consumers). As a result, actual trophic positions (1 producer, 1 and 2consumers, etc ) were not as clearly defined and closely associated with pl ankton size assemblages (ecologically relevant trophic levels) as desired. Potential omnivory, cannibalism and predation from multiple trophic levels within the three pond plankton size assemblages were systematically ignored to simplify pond food web str ucture and hypothesis testing. Realized swordtail trophic positions (i.e., actual food web position) were almost certainly lower than their potential trophic positions (i.e., strict food chain with swordtail as top predators), possibly due to the same trophic processes previously mentioned and believed to be occurring within the pond plankton assemblages: omnivory and zooplanktivory from multiple trophic levels, and possible cannibalism. Although only qualitatively assessed, it was interesting to note tha t small size plankton assemblage filtrates were usually a light to dark green color In contrast, medium size plankton

PAGE 548

548 assemblage filtrates were usually a dark brown color, and large plankton filtrates were composed of conspicuous red macroscopic organism s, usually copepods and daphnids, when filtrates were viewed on a bright white filter disk background (Figure 43). A primary reason isotopically derived swordtail trophic distance estimates were non integer values, may have been due to their omnivorous dietary habits (realized trophic position; Vander Zanden and Rasmussen 1999, Kling and Fry 1992). If isotopic signatures upon which trophic distance estimates were based, resulted largely from swordtail omnivory, then the simplified trophic food chain mod el (small phytoplankton medium phytoplankton/zooplankton (rotifers/larger algae) large zooplankton (copepods/daphnids) small and large harvest fish ) was largely invalid (potential trophic position; Vander Zanden and Rasmussen 2001, Vander Zanden and Vadeboncoeur 2002). The more complex, but still simplistic food web model was much more likely to have occurred within the INO ponds (Figure 445), and to a lesser degree within the CSM ponds ( Figure 4 44). Additionally, the presence of zooplankton and swordtail cannibalism would cause a further departure from the simple linear food chain, requiring a more complex food web model to describe the trophic interactions occurring within the INO and possibly nonINO ponds. Unfortunately, the degree of canniba lism occurring within the INO ponds was unknown, and therefore its effects upon the isotopic signatures of INO large harvest fish (as well as non INO large fish) are unknown. Juvenile Swordtail Live Food Dependence It is my belief that swordtails within all four nutrient ponds obtained a significant portion of their juvenile nutrition from live foods. This was supported by the casual visual observation that fry and juvenile fish were generally absent from surface waters, especially in the deeper sections of the ponds where floating applied feeds and cottonseed meal tended to concentrate. O ntogenetic swordtail micro habitat differences may have required fry and juvenile fish to be

PAGE 549

549 more dependent upon smaller live pre y (due to mouth gape limitations and predation pressures ) associated with plant cover that occurred only along the pond perimeter. Large aggregations of macroscopic zooplankton (e.g., daphnids) w ere often observed at the surface in deeper sections of the ponds some distance from the pond perimeter s ( pers. obs.) Unfortunately, a thorough survey of pond plankton and epiphytes, and their occurrence with vegetative cover distance from shore, and depth was not performed. The hypothesis of greater juvenile de pendence upon live foods also was supported by the relative isotopic signature values of small and large harvest fish (i.e., non INO small fish N > large fish N) that may be explained as residual nitrogen isotope signals resulting from greater fry an d juvenile fish dependence upon live foods. Within all pond treatments, this juvenile (ontogenetic) dietary isotopic signal, produced from smaller planktonic and epiphytic prey, may have been present within small harvest fish, and present as a latent weak er isotopic signal within all large harvest fish groups. If trophic guilds within the PRO, UNP, and CSM ponds were largely dependent upon applied nutrients and not the interdependence of living pond organisms ( trophic cascade: trophic guilds/predation ,), two expected outcomes were: (1) some pond trophic guilds would have roughly equal nitrogen isotope signatures that would have been expected to have isotopic signatures that reflected trophic dependence ( ~ 3. 4 N per trophic level) of one guild upon another (e.g. PRO large fish and large zooplankton N values were nearly equal), and (2) because fry and juvenile fish were believed to have a greater dependence upon live feeds relative to larger fish (mouthgape/predation pressure) smaller fish would be isotopically enriched relative to larger fish within the non INO treatments If small non INO fish were isotopically enriched in comparison to large nonINO fish, then the estimated trophic positions of small fish

PAGE 550

550 also would be higher than large fish w ithin the non INO treatments, due to the derivation of taxa trophic position estimates from taxa nitrogen isotope signature differences relative to their applied basal nutrient. This was the general pattern observed for nonINO small and large harvest fish, and more so within the two feed treatments than within the CSM pond treatment. Again, parallel applied nutrient movement among phytoplankton, zooplankton, and omnivor ous swordtail s and greater juvenile swordtail live food dependence may have been resp onsible for producing trophic level estimates (N based) greater than unity for the two feed and cottonseed meal small harvest fish groups ( Table 4 4). As nonINO small fish grew and entered the large fish size class, estimated small fish trophic distan ces to their applied nutrients ( > 1.0), became less pronounced (near unity). Serial nutrient movement within the INO treatment ponds produced a monotonous increase in nitrogen isotope signature and trophic distance estimate ( N based) as lower to higher (e.g., INO large zooplankton N < small swordtail N < large swordtail N) trophic level organisms were isotopically measured. Applied Nutrient Isotopic Signature Influences upon Pond Taxa Based upon the initial alternate hypothesis, it was expected that applied feeds would be directly consumed by and promote fish production, and that applied fertilizers would need to be assimilated by various pond taxa before becoming available in forms (live prey) capable of promoting fish production. If correct, this alternate hypothesis predicts that nitrogen isotope signature differences between harvest fish and their applied nutrient treatments w ould be greater for the CSM and INO treatments than for the two feed treatments as applied nutrients were utilized (and consumers became isotopically enriched) and moved serially within the food webs of the two fertilizer treatments. This is what was actually observed for the fish and isotopically analyzed pond plankton taxa within the various pond nutrient treatments (Figures 430 431).

PAGE 551

551 In fact, nitrogen isotope signature difference patterns were identical for both small and large harvest fish size classes within all four treatments s upport ing the contention that harve st fish within the four treatments were obtaining their nutrition from the applied nutrients and that the food webs of the two fertilizer treatments were longer than those of the two feed treatments. Although do fish assimilate feeds?, and do fertil izers promote biomass production, which fish then utilize? were simple questions, which had obvious expected outcomes for the pond trial and from many decades of previous aquaculture pond practice and research actually answering these questions with any degree of certainty was quite difficult (Boyd and Bowman 1997, Coman et al. 2003, Lochmann and Phillips 1996). Biases Using the ED and IS Dietary Estimate Methods Based upon CSM taxa isotopic signature profiles (Figure 4 41) and the two diet estimation me thods (Table 41; ED and IS), there existed two isotopically plausible CSM large harvest fish diet scenarios: (1) primarily CSM medium plankton, or (2) roughly equal proportions of applied cottonseed meal and CSM large zooplankton; a ssuming equal assimilat ion efficiencies of the medium plankton and mixed large plankton and cottonseed meal diets From direct observations of CSM large harvest fish feeding upon applied cottonseed meal at the pond surface, it appears much more likely that CSM large harvest fis h consume (diet scenario 2) roughly equal proportions of large zooplankton and cottonseed meal. CSM taxa isotope signatures appear to have produced erroneous CSM large harvest fish ED and IS diet estimates Within the CSM treatment, the collinear isotop ic signature plot positions of cottonseed meal fertilizer, and the medium and large plankton assemblages, may have incorrectly indicated that the CSM medium plankton assemblage was the primary prey of CSM large harvest fish ( Phillips 200 1, Phillips and Gre gg 2001, 2003).

PAGE 552

552 Without other (nonisotopic) fish dietary information, such as direct observations of fish feeding behavior, the diet consisting primarily of CSM medium plankton could not have been eliminated as a plausible CSM large harvest fish diet Again, illustrating a major limitation of relying solely upon isotope signature based diet estimation methods (Euclidean distance/IsoSource), and the need for corroborating direct observation and/or other methods to determine fish diet and feeding behavi or [e.g., gut content analysis, potential prey availability information, controlled prey choice/selection experiments, controlled feeding experiments (Chapter 6), etc.]. Additionally, a large visual predator such as the swordtail would be expected to pref erentially predate upon larger macro zooplankton, rather than smaller prey, when larger prey are present. Although cottonseed meal was indicated to be an important CSM large harvest fish dietary component using the ED and IS diet estimates ( Table 4 1 ), wi thout actual feed assimilation data for cottonseed meal fed to swordtails in a controlled setting (e.g. cottonseed meal feed efficiency estimates), it is not entirely certain that ingested cottonseed meal was assimilated by swordtails to an appreciable degree (Elliot and Persson 1978 Diana 1981). Regardless of potential CSM treatment diet estimation biases due to nutrient and plankton assemblage isotopic signatures, both the IS and ED CSM large harvest fish diet estimates produced similar results (Table 4 1; ~ 55 65% medium plankton, ~1520% large zooplankton, ~15 20% cottonseed meal). The finding that the IsoSource and Euclidean distance diet estimation methods produced similar results, even though they used different calculation methods (linear algebra versus coordinate distances), tends to indicate that the ED and IS diet estimation methods had high precision, although the accuracy of both methods appears poor when taxa have collinear isotopic signature values ( Phillips 2001, BenDavid and Schell 2001 )

PAGE 553

553 Supplemental information such as direct observation of CSM treatment large fish actively consuming applied cottonseed meal and their microhabitat preference for surface waters in deeper areas of the ponds proved invaluable in differentiating between two isotopically plausible dietary scenarios for CSM treatment large harvest fish. INO treatment ED and IS diet estimations determined that large harvest fish primarily utilized large zooplankton, which was expected as the applied inorganic fertilizer was not directly usable by pond heterotrophs (Table 41). It also was not unexpected that visual predators like swordtails would be concentrating on larger macrozooplankton relative to smaller zooplankton and phytoplankton (Jones et al. 2007). Small Harvest F ish ED and IS Diet Estimation Issues Small harvest fish ED and IS diet estimations were much more varied than their large harvest fish counterparts within all pond treatments (Table 4 1). Additionally, s mall harvest fish diet composition estimates differed greatly between the ED and IS estimation methods, which may indicate an inherent limitation of using stable isotopes to estimate diet s among smaller / younger individuals following diet switching within a consumer population ( Fry 2006). A small difference in diet among individual fry or among fry from different replicate pond s from the same treatment will have a disproportionately larger isotopic signature effect upon smaller individual s This is due to the lower predator to prey mass ratio of small er fish relative to large r fish and a greater prey isotopic tracer signal within a smaller predator. Additionally, larger/older fish are more likely to be in isotopic equilibrium with their prey sources due to temporal variations in diet being averaged over a longer time (greater fish age) and less isotopic dilution of a larger fishs greater elemental pool/biomass due to diet changes ( T Frazer pers. comm. ). Rather than providing reasonable estimates of subpopulation diets within given treatments, small harvest fish isotopic analyses may have only produced confounding diet estimations results.

PAGE 554

55 4 Alternately or in conjunction with lower small harvest fish weights being responsible for higher variation among small harvest fish ED and IS diet estimates pl ankton and epiphytes consumed by smaller fish within the experimental ponds may have been much more varied, both taxonomically and trophically (1 producers, 1 and 2 consumers, etc .) than food items consumed by larger fish within these ponds. Within the shallower areas where smaller fish were known to occur, greater trophic variation among the planktonic and epiphytic community was highly plausible considering the large amount s of vegetative structure and surface area available, in contrast to the more exposed and homogenous deep water areas of the pond frequented by larger fish. F or small fish a more ecologically diverse and speciose live food base would tend to consist of organisms of greater trophic diversity ( Paine 1980). Greater trophic and taxon omic variation in prey items (e.g., omnivory) would likely be correlated with greater isotopic variation in prey, leading to greater isotopic variation in any potential predators and isotopically based (i.e., ED and IS) diet estimates (Fry and Peterson 1984, Phillips and Gregg 2003). Within the present study, t he ED and IS diet estimation methods only utilize d two elements, which can produce dissimilar and inaccurate diet estimations when potential prey isotopic signatures are similar or prey signatures are collinear when plotted on a Cartesian plane ( Phillips 2001, Ben David and Schell 2001). When potential prey isotopic plots are collinear along the side of a predators isotopic resource polygon ( e.g., Figures 439 442, 453), it can be difficult to isotopically discern a mixed diet composed of prey groups that are isotopically collinear to or between vertices, and a diet composed primarily of a prey group that is collinear to, and between these two prey groups (Figures 446, 453). Unfortunately, due to isotopically similar mixtures of plausible prey organisms /nutrient sources, multiple plausible dietary scenarios can occur, which can lead to large diet estimation errors ( Phillips and Gregg 2003)

PAGE 555

555 Small Harvest Fish Dietary Estimate Issues In summa ry, small harvest fish ED and IS diet estima te prey/nutrient importance ranking differences appeared to be the result of three potential factors: (1) isotopic similarities of prey/nutrient groups, (2) prey/nutrient group(s) isotope signature s laying along a n isotopic gradient between other potential prey /nutrient groups ( isotopically collinear prey /nutrient groups), and (3) low individual small harvest fish weights The first of these factors did not appear to be an issue as the four applied nutrients and their respective plankton assemblages appeared to be isotopically separated (Figures 4 39 442) to varying degrees T he second factor was likely an issue within both the CSM and INO treatment s. Both the CSM and INO treatments had applied nutrient and/or plankton taxa that had collinear isotopic plot positions, which can produce isotopically similar composite diets from differing proportions of collinear potential nutrient sources. However, the collinear plankton assemblages we re not in close proximity to the isotopic plot position of either swordtail size class. The third factor may have been occurring within the INO treatment and to varying degrees within the three nonINO treatments S mall harvest fish may not have reached isotopic equilibrium (lower fish biomass/prey ratios) with their diets, resulting in higher levels of isotopic variation among individual fish within a pond and among fish from different ponds, producing ambiguous small harvest fish dietary estimates using the ED and IS methods. Plankton Size Assemblage and Trophic Position/Isotopic Signature Issues Collinear isotopic plot positions for different plankton assemblages were likely an experimental artifact resulting from plankton sieve selections P lankton s ize assemblages appeared to be roughly correlated with functional trophic groups (1 producer, 1, 2 and 3 consumers, etc.) along a trophic (and isotopic) continuum from primary producer to herbivor e to zooplanktivor e (excluding omnivor es). Microscopic examination of the three plankton size

PAGE 556

556 assemblages indicated that these assemblages were not discrete trophic groups, but e xist ed as mixed (especially medium) trophic groups (phytoplankton and zooplankton), except for the possible exception of the large pl ankton assemblage, which consisted primarily of large macro zooplankton (Chapter 5). Plankton size and trophic gradients likely produced the gradual isotopic gradient s (collinear isotopic signature plots) observed among the CSM (Figure 4.41; small and medium plankton assemblages), and INO (Figures 442; small, medium, and large plankton assemblages ) fertilizer treatments As previously mentioned, three or more isotopically collinear plankton assemblage and/or applied nutrient plot positions occurring along the side of a potential resource polygon ( Figure 4394.42) made diet estimations difficult using the ED and IS methods Although, the use of stable carbon and nitrogen isotope signatures to answer biological and geological research questions is now commonplace and has been widely applied within many different scientific fields (geology, ecology, anthropology, oceanography, climatology, etc.), the utilization of additional elements and their isotopes (i.e., 32S/34S and 16O/18O), has only recently become routine and affordable, allowing greater precision and accuracy in isotopic diet estimations ( Fry 2006). By utilizing additional elements such as sulfur, hydrogen and oxygen, potential food source candidates can be plotted in three and even four dim ensions, allowing implausible diets to be more easily eliminated than by the use of two dimensional (e.g., C, N) isotopic signature plots (Figures 4 39 442). An additional isotopic signature plot position issue was that INO small and large harvest fish carbon and nitrogen isotope plot positions were outside the food resource polygon connecting the large zooplankton, medium plankton, and small plankton (Figure 442), indicating that an important food source(s) may have been omitted ( Phillips 2001, Saito 2001,

PAGE 557

557 Fry 2006). Due to time and budget restrictions, emergent plant material and associated epiphytes and macroinvertebrates were not isotopically analyzed. These may have been important nutrient sources for fry and juvenile swordtails living in t he INO treatment ponds (possibly within all treatments), and may warrant further study to determine if emergent plants and associated epiphytes and macroinvertebrates are important nutritional sources in the early life history of swordtails. Another concer n in estimating fish and plankton assemblage trophic positions and trophic differences using isotopic signature data within outdoor ponds, was the major assumption that indoor feed assimilation efficiencies and isotopic enrichment rates for carbon and nitr ogen (Chapter 2; C 0.64 N 3.03 per trophic level) were comparable to the assimilation efficiencies and isotopic enrichment rates occurring within outdoor ponds for swordtails, applied nutrients, and various live foods. Many factors including temperature, photoperiod, predator characteristics (activity level, growth rates, dietary history, size/age, specific dynamic activity rates), food availability, diet proximate composition, diet protein to energy ratios, and prey handling time may have aff ected assimilation, and therefore isotopic composition. Further research investigating swordtail live feed assimilation rates and isotopic enrichment rates using cultured live foods fed defined or partially defined nutrients (including the nutrients used i n this study) in controlled feedings trials in laboratory or outdoor mesocosms, would be extremely useful in further characterizing and modeling intensive aquaculture pond trophic dynamics. This would give a more realistic approximation of aquaculture pond culture conditions, and a better understanding of tropical aquaculture pond trophic dynamics.

PAGE 558

558 Small Plankton and Nutrient Isotope Signature Differences Although it was hypothesized that feed nutrients were generally moving in parallel within the feed t reatment pond taxa, carbon isotope signature differences between the PRO and UNP small plankton assemblages and their respective applied feeds did not support this conclusion. The large C signature differences between feed treatment small plankton assemblages and their respective applied feeds indicated that phytoplankton within these treatments were not directly utilizing the applied feeds ( via ingestion ) as a carbon source. Feed treatment small plankton assemblages consisted primarily of autotrophic phytoplankton that utilized dissolved inorganic carbon compounds (DIC: CO2, HCO3 -, CO3 2), some of which may have been derived from remineralized and highly recycled applied feed nutrients (designed for heterotrophic consumption) released during respiratio n, excretion, egestion, and decay. This may have been responsible for the carbon isotope signature similarity of the two feed treatment small plankton assemblages, and their isotopic dissimilarity with the INO treatment small plankton assemblage. Photosynthetic fixation of available DIC causes autotroph carbon to be highly fractionated in favor of the lighter 12C isotope as physical (e.g., diffusive fractionation) and enzymatic fractionation (i.e., ribulose bisphosphate carboxylase kinetics RuBisCo) proc esses are strongly biased toward isotopically lighter carbon compounds ( OLeary 1988). Nevertheless, the DIC source that is utilized by autotrophs, still influences the carbon isotope signature of the resulting primary production (~ 20 C freshwater algae and inorganic carbon source; Peterson and Fry 1987). In addition to autochthonous, nonanthropogenic allochthonous DIC inputs, and remineralized allochthonous applied nutrient DIC, these carbon fractionation processes were likely respon sible for the large carbon isotope signature differences observed between the nonINO small plankton assemblages and their respective applied nutrients.

PAGE 559

559 Due to the presence of small plankton assemblage carbon isotope signature differences among treatments (Figure 4 28), feed and cottonseed meal application magnitudes were apparently sufficient to isotopically influence phytoplankton within the treatment ponds. Although DIC pool carbon isotope signature values were not measured for the four pond treatments, the fact that small plankton assemblage carbon isotope signatures differed among treatments implied that the dissolved inorganic carbon pools used for phytoplankton primary production also differed isotopically among treatments. Although the outdoor pond trial lacked a no nutrient control treatment, INO treatment taxa carbon isotope signature values were likely close to actual baseline taxa carbon isotope signature values that would have been observed within a no applied nutrient control treatment, because the dissolved inorganic carbon fixed and driving production within the INO pond food web was not intentionally human derived. Further research in vestigat ing dissolved inorganic carbon pool sources, input rates and their carbon isotope signatures would be extremely useful Including such information as the amounts and isotopic signatures of aqueous carbonates and carbon dioxide entering the pond a s groundwater and well water used to maintain pond volume losses due to seepage and evaporation; as well as autochthonous DIC derived from pond substrates, and decaying organic matter (detritus, humus). Again, c arbon isotope signature information for the various plankton assemblages and fish in the absence of anthropogenic nutrient applications would be extremely important and helpful for making further inferences and conclusions about tropical pond aquaculture trophic systems and their responses to diffe rent aquaculture management strategies. Fortunately, INO treatment taxa carbon isotope signature values served as de facto no nutrient control treatment baseline taxa carbon isotope signatures.

PAGE 560

560 Similar to non INO small plankton assemblage carbon isotop e signatures, nonINO small plankton nitrogen isotope signatures also appear to have been influenced by remineralized and recycled allochthonous applied nutrients. With the same general pattern of nitrogen isotope signature differences among treatment tax a, i.e., nonINO treatment small plankton assemblage nitrogen isotope signature values did not differ, but all differ ed from th at of the INO small plankton assemblage (Figure 4 28) T he nitrogen isotope signature influence s of applied nutrients were much more direct than that of their carbon isotope signature s with only a minimal nitrogen isotope signature difference occurring between small plankton assemblages and their respective applied nutrient treatments ( Figure 4 38 ) relative to the often large carbon isotope signature difference between small plankton assemblages and their respective applied nutrients (again, excluding the INO treatment which did not contain carbon) Dissolved inorganic nitrogen originally derived f rom applied nutrients designed for animal consumption (feeds) or as a combination animal feed and slow nutrient release fertilizer (cottonseed meal), appear to have influenced the nitrogen isotope signature of the small plankton assemblages within the nonINO treatments (Figure 4 28). Alternately, applied inorganic nitrogen fertilizer may have lowered the INO small plankton assemblage nitrogen isotope signature to a value significantly lower than those within the nonINO treatments. Additionally, both processes may have been occurring simultaneously, inorganic fertilizer application lowered INO primary producer N signature, and non INO treatment applied nutrients were remineralized and present in concentrations sufficient to influence nonINO primary producer N signatures. Regardless of the actual mechanisms by which food web taxa were isotopically influenced by applied nutrients, applied nutrients

PAGE 561

561 isotopically influenced the autotrophic phytoplankton communities within the outdoor treatment ponds. U nfortunately, because all ponds received nitrogen from their applied nutrient treatments, it could not be determine d if all applied nutrients isotopically ( N) influenced their respective taxa, or if only certain applied nutrients affected the isotopic s ignatures of their pond taxa. Autochthonous or unidentified allochthonous (e.g., groundwater DIN) nitrogen also may have made a large, but unmeasured contribution to all pond DIN pools. These issues again stressed the need for a no nutrient applied control treatment to obtain baseline taxa nitrogen isotope signature values as well as baseline extensive aquaculture fish production rates. Baseline extensive aquaculture pond DIN pool concentration data would have been extremely valuable in determining the overall contribution to pond DIN pools made by the addition of applied nutrients (Figure 4 46; via subtraction) and those contributions derived from autochthonous and nonanthropogenic allochthonous sources within the various treatments. Additionally absolute contributions to plankton and fish production, resulting from nonanthropogenic sources and from applied nutrients (via subtraction) could have been determined with the presence of no nutrient control treatment ponds. The INO small plankton as semblag e nitrogen isotope signature (1.19 N) was only slightly enriched relative to its applied fertilizer (INO 0.08 N), although this nitrogen isotope enrichment rate was approximately twice that observed within the non INO treatments it was close to the isotopic enrichment rate (depending on phytoplankton growth rates) measured for marine phytoplankton assimilating inorganic nitrogen in the western North Pacific Ocean (Wada 1980). Again, this was not surprising given that most autotrophs undergo little nitrogen isotope signature fractionation relative to their nitrogen source unless nitrogen is greatly in

PAGE 562

562 excess of metabolic need and/or autotroph growth rates are extrememly low (Fogel and Cifuentes 1993, Evans et al. 1996, Finlay and Kendall 2007 ). Other investigators have found larger phytoplankton nitrogen isotope enrichment rates, but these studies were done under laboratory settings where high nutrient concentrations were employed, and high inorganic nutrient concentrations (luxury uptake) have been known to strongly influence (generally increase) phytoplankton isotopic signature enrichment rates (Waser et al 1998, 1999, Needoba et al. 2003). In particular, high nitrogen availability often results in a greater nitrogen isotope enri chment magnitude between algae and the exploited inorganic nitrogen pool, due to higher nitrogen turnover rates within algal cells (Pennock et al. 1996, Cole et al. 2004, Needoba 2004, Finlay and Kendall 2007). As mentioned earlier, fertilizer treatment po nds (CSM and INO) had higher total ammonia nitrogen (TAN) levels than those of the two feed treatments ( Figure 4 4 6). The greater nitrogen isotope signature difference between the INO small plankton assemblage and the inorganic fertilizer may have been due to the greater availability of dissolved inorganic nitrogen (TAN) within the INO ponds relative to TAN concentrations within ponds of the two feed treatments ( Figure 4 46 ). Interestingly, although CSM pond DIN concentrations did not differ from those observed within the INO treatment ponds (Figure 4 46), greater nitrogen isotope fractionation was not observed between the CSM small plankton assemblage and applied cottonseed meal fertilizer. This lack of a large nitrogen isotope signature enrichment magn itude occurring between CSM small plankton and cottonseed meal may have been due to higher phytoplankton concentrations within the CSM ponds (Figure 449 ), which may have reduced the per capita DIN availability for individual phytoplankton cells. Due to nitrogen limitation, CSM phytoplankton may have had to

PAGE 563

563 assimilate all available DIN, thereby undergoing little nitrogen turnover and therefore little isotope fractionation relative to their nitrogen sources occurred. CSM treatment pond DIN was believed to be primarily derived from the metabolism and decay of applied cottonseed meal protein. Many factors can affect the rate at which nitrogen is taken up by algae and can affect the nitrogen balance and overall isotopic signatures of biological materials, including phytoplankton surface area to volume ratio s nitrogen availability, nitrogen chemical form (NH4 +, NO3 -, NO2 -, urea) within the aqueous environment, current and past history of nutrient availability, algal characteristics (species, cell size and age, growth rates, genetic differences), photoperiod, and temperature (Fogg and Thake 1987, Wada and Hattori 1978, Mont oya et al. 1991, Montoya and McCarthy 1995, Valisneria 199X, Hein et al. 1995). Interestingly, freshwater lentic phytoplankton nitrogen isotope signature values have been reported as varying from 3 to 8 N, influenced by factors including phytoplankton population growth rates, algal species, and season ( spring and summer months ) in a Japanese eutrophic lake (Wada and Hattori 1976, Yoshioka et al. 1994), from 4.2 to 6.0 N in an oligotrophic lake located along the Sierra Nevada foothills along the C alifornia Nevada border (Estep and Vigg 1985), and ~ 3 to 7.8 N in carp polyculture ponds in Southern China (est. Fig. 1; Gu et al. 1996). All reported values agreed well with the small plankton assemblage nitrogen isotope results for the PRO, UNP a nd CSM treatments (Figure 4 28; range 4.29 to 4.38 N), which was surprising given the variety of climates/seasons, water body nutrient levels (oligotrophic to eutrophic), phytoplankton species and investigator plankton sieve size choices. Again, bas eline nitrogen isotope signatures of the various plankton assemblages and fish, in the absence of allochthonous nutrient application, would have been useful to better determine

PAGE 564

564 actual food web processes occurring at lower trophic levels (e.g., whether domi nant pond DIN source(s) were anthropogenic or non anthropogenic). Additionally, how fish production (Chapter 3) and large plankton assemblage compositions and production rates (Chapter 5) were altered by nutrient additions, and how these factors would have been affected by the absence of anthropogenic nutrient addition, would have been extremely valuable information. Baseline taxa nitrogen isotope signatures also m ay have provide d greater insights into pond trophic dynamics and led to more meaningful and robust conclusions. Due to the lack of nitrogen isotopic signatures for autochthonous and nonanthropogenic allochthonous DIN pools as well as the DIN pool sizes (concentration) being utilized by phytoplankton (small plankton assemblage) within the four pond treatments not resulting from human invention. C onclusions regarding what trophic processes [e.g., identifying autochthonous and nonanthropogenic allochthonous DIN input source(s) and their relative contributions to total DIN] were occurring at the base of the food web within the four pond treatments were somewhat conjecture. Additionally, because baseline fish production values were lacking, the impact of applied nutrients within the different treatments must be primarily ass essed from relative empirical measurements of fish production, rather than absolute fish production measurements compared against baseline fish production in the absence of applied nutrients Unfortunately, due to limited pond replicate availability, and the typically high variation in fish production that occurs among pond replicates (even within the same treatment), ponds that could be assigned to a no applied nutrient control treatment were not available for the outdoor pond study. Plankton Assemblage Experimental Artifact and Model Species Issues Several experimental design modifications and additions could have been made to the outdoor pond study, which likely w ould have facilitated interpretation of the experimental data, giving a better underst anding of aquaculture pond trophic dynamics and an increased ability to

PAGE 565

565 evaluate aquaculture pond management practices. Plankton size assemblages, resulting from sieve selection certainly produced experimental artifacts that divided pond plankton assemb lages into convenient, but artificial assemblages that were not grouped into actual functional trophic groups but rather by similar gross particle size. Physically separating large zooplankton into species or genus groupings prior to isotopic analyses wou ld have been extremely useful in producing true trophic guilds, although this would have greatly increase d the time and expense of analyses. Poeciliid swordtails were not the ideal experimental model animal choice, due to their iteroparus reproductive lif e history (~ 2430 days per brood) relatively small brood size (~ 30 fry per brood) and early age of first reproduction, which has been reported as early as six to eight weeks of age (Wischnath 1993, Tamaru et al. 2001). Continuous reproduction of pond broodstock and sexually mature older fry created numerous fry age/size cohorts with mixed genetic histories. Unfortunately, mixed age cohorts and continuous reproduction made individual fish growth rate and fry survival rate calculations impossible. Gener alist dietary habits also made trophic estimates problematic, as did potential cannibalism due to large size disparities between broodstock (and older pond spawned fish) and smaller juveniles and natal fry ( Jones et al. 1998, Tamaru et al. 2001). A more id eal experimental model fish species would have had life history characteristics such as: highly fecund egg layer (high recruitment), first age of reproduction occurring at greater than three months, and highly specialized dietary habits (e.g., zooplanktivore specialist). Common ornamental fishes such as the zebra danio ( Danio rario) or redtail shark ( Epalzeorhynchos bicolor ) would have been better model animals to characterize trophic dynamics occurring within tropical aquaculture ponds within the experim ental design of the

PAGE 566

566 present study. A large number of eggs/fry resulting from a limited number of female broodstock would have been desirable to limit the genetic and size/age variation among the fry/eggs used to randomly stock the individual treatment ponds, and any potential maternal dietary history effects (i.e., maternal nutrient isotopic signature effects). Another advantage of having only a single age cohort would have been greatly reduced potential cannibalism rates due to lower fish size disparitie s ( Jones et al. 1998, 2007). A single age/size cohort also would have greatly aided in the determination of individual fish (somatic) growth rate calculations. Short pond trial duration, lack of reproduction within the ponds, and a single fry cohort also would have greatly aided fry survival rate estimates for obvious reasons. Additionally, the presence of ontogenetic changes in fish trophic position estimates would likely have either occurred in a predictable pattern, or may have been entirely absent due to a lack of intraspecific predation pressures and the necessity for microhabitat differences among fish of different size/age classes due to cannibalism and/or food availability issues. Large Zooplankton Assemblage Taxa and Abundances Weekly large ( > 200 m) plankton assemblage taxa were enumerated, and identified to order, and farther when possible. Large zooplankton volumes and densities were calculated based upon taxa numbers and taxa geometries derived from photomicrographs. Differences in larg e zooplankton densities, volumes and major taxa (highest %IRI) did not greatly differ among pond treatments but frequently differed among sampling dates within treatments Additionally, trends in total zooplankton density and volume, major taxa %IRI and major taxa density did not greatly differ among treatments, but were similar among treatments. Large Zooplankton Densities As previously mentioned l arge zooplankton assemblage density and volume differences were generally absent among the four nutrient treatments for the ten weekly sampl es. Large

PAGE 567

567 zooplankton density and volume differences among samples were more a function of time within pond treatments rather than a function of nutrient treatment differences. P lankton densit i es and volumes early in t he study were significantly lower than later densities and volumes due to the time required for large plankton communities to develop following pond draining and sterilization T ime integrated large zooplankton density and volume averages did not differ among the four treatments i ndicat ing that large plankton standing stocks and th us live food availabilities probably did not differ for swordtails among the four treatments. However, this does not indicate whether swordtails within the different treatme nts were utilizing large zooplankton to differing degrees since actual plankton production rate ( i. e. mg large zooplankton C/day) and swordtail predation rate ( i.e., mg large zooplankton/fish/day) values were lacking Large zooplankton production rate may have been higher within a given treatment, but if predation upon the zooplankton was higher within this treatment, large zooplankton standing stock measurements alone would not indicate whether swordtail dependence upon large zooplankton communities differed among treatments. The presence of a no fish control treatment would have been useful in providing baseline large zooplankton standing crop information in the absence of fish predation. A t a minimum this would have allowed a gross estimate of swordtail predation rates upon large zooplankton via subtraction and then integrat ion over the 10 weeks. Also, the addition of a no fish and no nutrient control treatment would have been useful in determining the bas eline large zooplankton production rates (again via subtraction) in the absence of both fish predation and anthropogenic nutrient addition.

PAGE 568

568 Unexpectedly, pond nutrient treatment was not a significant factor in determining large zooplankton volume in contr ast to sampling date which was the primary factor determining large zooplankton standing stock volume (Figure 5 7). It was initially expected that ponds within the two fertilizer treatments would have had greater live food productivity relative to their feed treatment pond counterparts, due to fertilizer applications being intended to stimulate phytoplankton production and subsequent zooplankton production. However, greater swordtail dependence upon live foods within the fertilizer treatments may have r educed CSM and INO large zooplankton standing stock/ biomass levels below that of feed ponds. Not surprisingly, large zooplankton densities increased as a function of time, zooplankton standing stocks increased as water temperatures and photoperiods increased, and as nutrients were continuously applied on a daily (feeds and cottonseed meal), weekly or bi weekly basis (inorganic liquid fertilizer). Large plankton assemblage maximum density means occurred roughly midtrial for the PRO, UNP and CSM treatments, although PRO mean densities did not statistically differ among the ten weekly sampl es. INO large plankton assemblage density means were bimodal with a second larger peak occurring near the end of the study (Figure 511). Again, decreasing lar ge zooplankton standing stocks during the latter half of the study may have been due to a combination of factors including increased zooplanktivory, and lower phytoplankton prey availability. Lower phytoplankton prey availability also may have resulted f rom photoinhibition and/or self shading by the phytoplankton community. Phytoplankton primary production also may have been reduced by higher water temperatures or higher night time pond respiration rates (hypoxia) that were suboptimal for primary product ion. However, phytoplankton photoinhibition does not appear to be likely due to the generally stable or increasing chlorophyll

PAGE 569

569 [a] during the latter half of the twelve week trial (Figures 4 45 448). Phytoplankton self shading may have occurred, as prominent algal blooms were present within the majority of the treatment ponds during the study (pers. obs.). Although slight decreases in chlorophyll [a] occurred during the study this may have been due to physical (density independent) factors such as war mer water temperatures, and increased solar insolation (slight photoinhibition) as well as biological factors (density dependent) such as shading and increased nighttime respiration and possible hypoxia stress that may have resulted in smaller phytoplankt on standing stocks. Zooplankton herbivory did not appear to be the likely cause for the slight decrease in phytoplankton during the last weeks of the study, due to lower large zooplankton densities starting in the last third of the 12week trial (Figure 5 25.5). Interestingly, large zooplankton assemblages did not dramatically differ among the four pond nutrient treatments over the 12 week study. Of the large zooplankton taxa that regularly occurred within the weekly samples (%IRI > 0.5%), the pattern th at taxa colonized newly established ponds was highly consistent regardless of treatment. The three major large zooplankton taxa (highest %IRI rank) displayed predictable %IRI and abundance patterns (density: number/liter): copepods were the dominant group by %IRI and often by number (density) followed closely by Filinia rotifers, and daphnid %IRI and densities consistently ranked third. Similar to large zooplankton assemblage standing stocks, large zooplankton assemblage taxa compositions did not appear to be correlated with any particular nutrient treatment. Time had a much greater influence upon large zooplankton assemblage taxa compos itions. Not surprisingly, large zooplankton abundances and assemblage species diversity increased as the 12week outdoor pond study progressed.

PAGE 570

570 Ponds typically had large numbers of copepods and Filinia rotifers which closely mirrored (inversely proportional) each other in %IRI (especially pooled %IRI; Figures 514 5 17), possibly due to predation of copepods upon Filinia rotifers. Daphnids first appeared within ponds about midstudy and apparently increased at the expense of both copepods and Filinia rot ifers, probably due to competitive exclusion of copepods and direct predation upon Filinia rotifers based on the relative sizes of these taxa. Density counts of the three highest %IRI taxa indicated the same pattern of importance as %IRI within the large zooplankton assemblage community. Large numbers of copepods and daphnids usually resulted in subsequent decreases in Filinia rotifer numbers, and copepod numbers decreased as daphnid numbers increased ( Figures 5 21 536). Due to the extremely small amo unts of biological material needed for isotopic analysis (Chapter 4; 600800 g), pooled samples separated into individual taxa could easily be performed for future studies. Combined with time series isotopic analysis of fish predators, this would allow the determination of whether fish predators switch dietary taxa as prey abundances change, which would be expected for a generalist omnivore like the swordtail. Additionally, stable isotope signature analysis of individual zooplankton taxa could resolve t he suspicion that copepods and daphnids are directly consuming Filinia rotifers to a significant degree. Deterministic Large Plankton Assemblage Similarity Indices As expected, similar large plankton assemblage taxon %IRI values resulted in high large plankton assemblage Morisitas (MSI ) and percent ( PSI ) similarity indices values between treatments and sampling dates within treatments (Figures 538, 545). Pairwise treatment MSI trajectories displayed much greater variation among sampling dates than their PSI counterparts, and noticeably oscillated during the first six to nine weeks of the study, before plateauing to fairly constant and high values during the last few weeks (Figure 538).

PAGE 571

571 PSI value trajectories displayed much lower variation among sampling dates than weekly MSI values, and all but one treatment pair combination w ere similar (PSI 0%). This made the PSI virtually useless, in that it did not provide any new information regarding large zooplankton assemblage communities and any differences that may have existed between them, nor how these communities may have changed over time. Additionally, the higher PSI critical value (PSI 80%) used by Silver (Silver 1975), did not greatly reduce the utility of the percent similarity index. The simplified Morisitas similarity index was much more sensitive to changes in large zooplankton assemblage composition differences between treatment groups or sampling dates within treatments, changing both temporally as the study progressed and possibly in direct response to large zooplankton community perturbations (i.e., diuron herbicide application on 19 April 2006). PSI did not appear to noticeably change due to large zooplankton community changes and development over time, nor in response to diuron he rbicide application, with the possible exception of the PRO and CSM PSI time series values (Figure 5 47). These results indicate that MSI was superior to PSI in determining plankton assemblage similarities within the current study (Krebs 19 98). Bootstrap Large Plankton Assemblage Similarity Indices As expected, bootstrap similarity indices typically displayed lower variation among estimates as draw pool size and iteration numbers increased ( Figures 552 555, 568 573). To observe the effects of incr eased draw pool size on bootstrap similarity index estimates, draw size pools were increased to 20 resampled replicates per iteration, which resulted in moderate curve smoothing as variation among resampled PSI mean values decreased ( Figures 5 68 573). However, draw pool size should not be increased greatly beyond the actual number of

PAGE 572

572 replicates within the original dataset, as bootstrap generated statistical parameters (e.g. mean) may become suspect when draw size is greatly increased ( Simon 1997). Int erestingly, bootstrap MSI and PSI time series values, generated using taxon %N closely resembled each other, unlike the deterministic MSI and PSI time series values generated using %IRI values. Also, unlike their deterministic %IRI generated PSI counterp arts, bootstrap generated PSI time series values appeared to be sensitive to large plankton community disturbances (e.g., diuron application) as were both the deterministic %IRI and bootstrap %N generated MSI estimates. This may have been due to the use o f %N for bootstrap derived estimates and the more complex %IRI for deterministic estimates. Unfortunately, bootstrap MSI and PSI estimates using %IRI values could not be generated due to the use of percent frequency of occurrence (%FO) within the %IRI, an d the resulting lack of multiple datasets (replicates) from which to randomly draw (with replacement) for the generation of resampled population parameters (i.e. means, standard errors, and 95 % CI ). The primary rationale for using resampling techniques to generate bootstrap similarity index estimates and error measures ( Appendix 5B ; resampling program) was an attempt to more objectively identify within treatment differences in similarity values among sampling dates other than by the use of arbitrarily c hosen critical threshold values, 0.65 for the simplified Morisitas similarity index and 60% for the percent similarity index (Zaret and Rand 1971, Cailliet and Barry 1978, Krebs 1998, Lindquist 1998). Large zooplankton assemblage communities did not c orrelate with pond nutrient treatments but did change in predictable patterns over time. Additionally, large zooplankton assemblages did not display greater similarities with similar pond nutrient types (i.e. feeds versus fertilizers). Over the course o f the 12week study, large zooplankton communities largely

PAGE 573

573 stabilize d among the different treatments. Both MSI and PSI pairwise treatment large zooplankton assemblage similarity values leveled off during the last three to four weeks of the study. L arge plankton assemblage biomass (density and volume) parameters and community analyses indices (%IRI, MSI, PSI) did not appear to correlate with differences in swordtail biomass production (Chapter 3) due to the lack of any clear differences in large zoop lankton biomass or large zooplankton community composition with pond nutrient treatment type. It was initially believed that large zooplankton assemblage biomass and/or community composition indices would be correlated with applied pond nutrient treatment s and swordtail harvest biomass differences among treatments. However, interesting and predictable trends in large zooplankton biomass and community composition did occur among weekly sampling intervals within treatments, allowing some limited insight into ecosystem functioning within the large zooplankton communities occurring within the different pond treatments. The experimental design of the outdoor pond study was not intended to quantitatively measure the degree to which swordtails exploited availabl e live foods (ration requirements) within the different pond treatments (e.g., serial capture and gut content analysis over diel periods ; Elliott and Persson 1978, Diana 1979). Swordtail daily ration estimates from future studies might be combined with i nformation from the current study (Chapter 4, isotopically derived diet estimates; Chapter 5, large zooplankton taxa %IRI) to provide better quantitative estimates of applied nutrient and live food resource usage. Additionally, information regarding large zooplankton and swordtail spatial distributions and microhabitat usage (depth, distance from shore, association with vegetation, etc ) within individual ponds and pond nutrient treatments would be useful in further determining food availability differenc es for swordtails

PAGE 574

574 due to microhabitat associations (including ontogenetic microhabitat changes) of swordtails and their potential large zooplankton prey. The integrated water column plankton sampling method employed in the current study (Chapter 4; Figure 42, PVC pipe sampler) attempted to account for typically patchy distributions of plankton by pooling subsamples collected from each side of the rectangular ponds. It is unlikely that this adequately accounted for actual plankton availability within the ponds, as large zooplankton aggregations were often seen at the surface midway between the pond perimeter/shallows and the floating airline bisecting individual ponds A dult fish were often seen congregating in association with the floating airline and ai r stone ganglia. Unfortunately, swordtail anatomy (pharyngeal jaws) prevented a thorough and direct survey of prey taxa exploited by swordtails within the experimental treatments. In future studies, selection of a species that lacks pharyngeal teeth may allow improved identification of gut content prey items (Stone et al. 2006 Lochmann and Phillips 1996). Suitable fish predator species selection allowing direct gut content analysis combined with potential prey isotopic analyses might provide a clearer an d more robust interpretation of aquaculture pond trophic dynamics among differing applied nutrients, nutrient types (i.e., feeds, fertilizers) and pond management scenarios (e.g. twice daily feeding versus weekly and bi weekly pond fertilization). Future research using large zooplankton assemblage community taxa analysis and detailed stable isotope signature information of individual major large zooplankton community taxa and fishes might indicate which live prey taxa (if any) are preferred and/or provide the highest nutritional quality to the target aquaculture species. Possibly allowing aquaculture pond management plans to be designed that differentially enhance the production of pond live food taxa/communities that best contribute to commercial fish sp ecies production ( Schroeder 1978,

PAGE 575

575 1983, 1987, Lochmann and Phillips 1996, Kruger et al. 2001 ). Also, stable isotope and community analyses of the benthos may be necessary to properly describe and model tropical pond aquaculture systems, as pond benthos ma y make a sizable contribution to pond food webs and fish production ( Fry and Parker 1979, Cifuentes et al. 1988, Yusoff and McNab 1989, Vander Zanden and Vadeboncoeur 2002, Melville and Connolly 2003, Tyler et al. 2003). 24Hour Large Zooplankton Assemblage Captive Feeding Trials Due to the pharyngeal jaw anatomy within swordtail buccal cavities, swordtail gut content analysis was not feasible as ingested prey is typically reduced to a homogenous bolus. It was therefore necessary to employ more indirect methods, by comparing large zooplankton biomass and community compositions in the presence and absence of a single swordtail predator after 24 hours, and with a sample that was immediately fixed with preservative following collection. Captive feeding trials were conducted within 1liter samples from the four nutrient treatments over a 24hour period. Three treatments were employed for this portion of the study: one liter no fish for 24 hours, one liter single fish for 24 hours, one liter immediately pre served upon collection. Large Plankton Assemblage Biomass Large plankton assemblage biomass ( density and volume) did not differ among pond nutrient treatments within individual feeding trial treatments (e.g. PRO Initial and UNP Initial), but did differ among feeding trial treatments (Initial, NF, Fish) within certain pond nutrient treatments (i.e. UNP and INO). When pond nutrient treatments were pooled, biomass within the Fish treatment was significantly lo wer than those of the two control treatments. Two feeding trial control treatments were necessary to account for the effects of time/trial duration which likely included such factors as intra plankton p redation and possible physiological stressors (tempe rature, DO, starvation, etc ) effects upon large plankton assemblage biomass (NF and

PAGE 576

576 Fish ). Time and fish presence both reduced large plankton biomass lower biomass resulting from the 24hour duration of the trial alone (NF) was likely due to both intr a plankton predation and physiological stressors, but only fish presence (Fish) resulted in significantly lower biomass measurements ( unpooled data: PRO and UNP treatments only). This finding indicated that the presence of a single swordtail fish predator over a 24 hour period was sufficient to lower large plankton biomass, most likely as a direct result of fish planktivory. When densities of the three major large plankton taxa were examined, taxa densit ies were lower within the Fish experimental treatment than within those of the two other treatments. Major taxa densities within the Fish were not significantly lower across all pond nutrient treatments, but copepods were significantly lower for the INO and daphnid densities were significantly lower for th e UNP and INO treatments. When pond nutrient treatments were pooled, copepod and daphnid densities decreased among the three feeding trial treatments in the following order: Initial, NF, and Fish. The Fish treatment had significantly fewer copepods and da phnids than the two control treatments, which did not differ from each other. This indicates that swordtails were an effective predator, given the experimental conditions for the captive feeding trial (24 hour trial duration, 1L pond water sample, one sw ordtail predator, no aeration, etc .). Among all captive feeding trial replicate large plankton assemblages, Filinia numbers were much lower than those of copepods and daphnids, which may have been why rotifer density analysis provided inconsistent resul ts among pond nutrient treatments and among the three feeding trial treatments. Significant differences in rotifer density only occurred among the captive feeding treatments within the PRO pond treatment ( Figure 642). Not surprisingly, pooled Filinia de nsity averages also did not differ among feeding trial treatments. Again, this

PAGE 577

577 was likely due to the small rotifer densities present among the different experimental treatments. Low Filinia numbers may have been due to seasonal effects, as the feeding tr ials occurred in mid September 2007, whereas the outdoor pond trials occurred from late March to mid June 2006. Interestingly, daphnids were preferred prey for swordtails as indicated by changes in relative copepod and daphnid %N and %IRI values among the Initial, and both 24hour incubation treatment groups [NF (no fish) and Fish]. Copepod % number and % volume were markedly higher than daphnid % number and % volume within the Fish treatment relative to these parameters within the Initial and NF treatmen ts. Additionally, copepod %IRI was much higher following fish predation than prior to (Initial) or in the absence of fish predation (NF), although absolute copepod numbers were still reduced within the Fish treatment large plankton assemblages, likely as a result of fish predation. These taxa differences were much more apparent within pooled pond nutrient treatment comparisons of feeding trial data. P ooling data is not an ideal strategy for data analyses and several limitations and problems associated with data pooling must be addressed. Simply increasing sample size can produce a statistically significant difference even when the difference has no biological meaning ( Zar 1984, Krebs 1998). However, this usually entails increasing sample sizes to extr emely high numbers (>100), while pooling data for these experiments only increased sample size from four to sixteen. Because the 24 hour feeding tria l experimental design was constructed without the consideration that data might be pooled among pond nutrient treatments, pooling treatment groups that were not biologically equivalent was a possibility; to partially compensate, significance level ( ) can be lowered when pooling data in an attempt to increase the rigor of conclusions based upon analyses of pooled data, however, this increases the probability of a Type II (not rejecting a false

PAGE 578

578 null Ho) statistical error ( Ott et al. 2000 ). Adjusting the sign ificance level was not deemed necessary as Initial large plankton assemblages derived from different pond nutrient treatments, qualitatively resembled each other (community composition, taxa %IRI, % numbers and % volumes), indicating that fundamental dif ferences in large plankton communities did not appear to exist among pond nutrient treatments. Additionally, the objective of the 24hour captive feeding trial was not to compare large plankton assemblage biomass and communities among pond nutrient treatm ents per se, but to compare these parameters in the presence and absence of swordtail predation. Pooled copepod densities displayed the same significance pattern observed when pooled large zooplankton volumes were compared among the three feeding trial tre atments. Copepod densities, in replicates incubated with fish for 24 hours decreased to the point that this treatment had significantly fewer copepods than either the Initial or NF (no fish) 24 hour controls. Although single fish were able to significantly reduce copepod densities (assumed via grazing) in 1L incubation jars after 24 hours, the actual density of fish within the ponds almost certainly differed from that used for the 24hour captive feeding trial (i.e., one fish per liter). Additionally, fish and zooplankton are not evenly distributed within ponds; larger fish were routinely observed aggregated in schools near the surface with smaller fry and juveniles among the emergent vegetation along the perimeter of the pond. M acrozooplankton were observed in large aggregations at the pond surface midway between the pond center and shore. Again, t he primary purpose of the 24hour captive feeding trial was to evaluate and validate the potential for swordtail s to produce a detectable grazing/predation pressure upon captive zooplankton assemblages.

PAGE 579

579 Bootstrap estimates of large plankton community MSI and PSI using large plankton taxa % number values were more erratic than their deterministic %IRI MSI and PSI counterparts. For example, large number s of extremely small Pediastrum spp within the UNP Fish experimental treatment and low Pediastrum spp. density in its INO Fish treatment counterpart (Figures 6 6, 6 12) were likely responsible for the observed significant difference between Initial and F ish MSI values of the UNP and INO treatments that otherwise did not appear to greatly differ in taxa composition or relative taxa % numbers or % volumes. Interestingly, MSI values differed among pond nutrient treatments within a given captive feeding trial treatmentpair comparison for some pond nutrient treatments (e.g., Initial and NF UNP, and Initial and NF INO), whereas no differences among pond nutrient treatments within feeding trial treatments were found for PSI. Again, PSI may be less sensitive to slight taxa % number differences among large plankton assemblage comparisons relative to MSI (Chapter 5 Discussion/Conclusion). This suggests that the use of % number was inferior to % IRI for many of the reasons described by Cortes ( Cortes 1997, E. Corte s pers. Comm.) in that large numbers of small taxa can skew MSI and PSI values disproportionally to their importance within the population being analyzed (i.e., large plankton community). The captive feeding trial experiments were designed to qualitativel y evaluate potential swordtail predation effects upon large plankton community taxa under controlled conditions. It was shown that a single swordtail can reduce the overall taxa densities, bio volume s, and alter community composition (i.e., feeding prefer entially upon daphnids) within a small volume of pond water within a 24 hour period. T o better quantify swordtail feeding behavior, several simple experimental trials also could be performed. Different potential prey taxa handling time comparisons, and diet preference/dietary choice trials could be performed in simple aquaria, or

PAGE 580

580 feeding behavior choice trials could be performed within Y shaped aquaria in which predators are allowed to choose from a variety of potential prey ( Mills et al. 1984, DeVries et al. 1998). Additionally, potential large zooplankton palatability issues may arise from the anthropogenic application of different pond nutrients, which may influence swordtail prey/food source choice. L arger pond water samples (e.g., 2 L) and/or longer trial duration (> 24 hrs.) may have more clearly demonstrated potential swordtail predation upon the large phytoplankton and zooplankton community. To adequately investigate potential swordtail predation upon pond large plankton communities, a fully orthogonal trial employing different potential predation magnitudes ( numbers of fish in a container, e.g., 1, 2, 4) and trial durations (e.g., 12, 24, 36, 48 hrs.) would better describe (and quantify) the potential impact of swordtail predation upon the large p lankton community and the magnitude of predation per fish. This combined with more detailed baseline investigations of large zooplankton abundance and distribution within ponds, and swordtail population size structure, abundance and distribution within the ponds would give a clearer picture regarding the potential trophic effects and limitations of swordtail predation upon the large zooplankton community. In floating pond mesocosms, or pond side aquaria with continuous pumping of filtered pond water int o aquaria, may have produced environmental conditions (temperature, dissolved oxygen) more similar to those actually occurring within the outdoor ponds. The reduction or elimination of physiological stressors such as low dissolved oxygen, ammonia, nitrite and high temperatures etc., may have increased the similarity of the large plankton communities within the two control treatments. Additionally, reduction of large community taxa differences between the two control treatments, and increasing the number of replicate ponds sample d ( from four to six), would likely have increased the statistical power of the various analyses. I ncreasing the

PAGE 581

581 likelihood that pooling of replicates among pond nutrient treatments would not have been necessary. Daily ration es timates combined with gastric evacuation rate measurement of swordtails (and various swordtail size classes) would also allow rough estimates of swordtail carrying capacity and the proportion of potential swordtail production attributable to live foods and the proportion derived from direct anthropogenic nutrient additions (intended for direct consumption by fish), given that bioenergetic needs of swordtails, and potential swordtail food calorimetri c s were known ( Winberg 1956, Elliott and Persson 1978, Eggers 19 79).

PAGE 582

582 APPENDIX A LARGE PLANKTON ASSEM BLAGE %N, %V, AND %F O Large Zooplankton Assemblage Community Taxa Parameters Legend Key to Graphs: COP CYCLOPIDAE copepod CLD Filinia rotifer DAP Moina macrocopa (CLADOCERA)TLR rotifer (Brachionus havanaensis) OST ostracod CST bryozoan statoblast ELM copepod egg mass (unattached) VLUKE large single egg VSUKE small single egg NAUP copepod nauplii BIN big insect CLR rotifer (Keratella spp.) MITE water mite (HYDRACARINA) LNE nematode SCLR spiked rotifer (Keratella spp.) EPH Ephemerata insect NON-LHC 'rare' copepod FLS terrestrial floral stamen DPP Pediastrum spp. algae SEED terrestrial grass seed BBROT brown ball rotifer SN snail SPBLL spore ball NLR Synchaeta spp. rotifer ARTH arthropod (misc) IHT insect head RR round rotifer GSR grape seed rotifer PHR -" Papal Tiara" rotifer SBT Spikey Ball Thing

PAGE 583

583 Figure 5A 1 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment 13 April 2006; taxa data pooled among six replicate ponds (0.015 hectares). %FO cop cop bin elm vsuke lne tlr vlukePRO 4/13/2006(%N or %V => 0.5%) clr mite naup seed spbll Figure 5A.1: PRO pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 584

584 Figure 5A 2 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP) treatment 13 April 2006; taxa data pooled among six replicate pond s (0.015 hectares). %FO cop cop bin vsuke lne tlr vlukeUNPRO 4/13/2006(%N or %V => 0.5%) seed spbll cst Figure 5A.2: UNP pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %Vbbrot

PAGE 585

585 Figure 5A 3 La rge zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment 13 April 2006; taxa data pooled among six replicate pond s (0.015 hectares). %FO cop cop lne tlr non-lhc bin elm vsuke vluke seedCSM 4/13/2006(%N or %V => 0.5%) Figure 5A.3: CSM pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 586

586 Fig ure 5A 4 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment 13 April 2006; taxa data pooled among six replicate pond s (0.015 hectares) %FO cop cop bin ost elm vsuke lne tlr vluke dppINO 4/13/2006(%N or %V => 0.5%) Figure 5A.4: INO pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 587

587 Figure 5A 5 Large zooplankton (size > 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment 20 April 2006; taxa data pooled among six replicate pond (0.015 hect ares). %FO cop cop bin vsuke lne vlukePRO 4/20/2006(%N or %V => 0.5%) elm eph ost dpp clr spbll Figure 5A.5: PRO pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 588

588 Figure 5A 6 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP) treatment 20 April 2006; taxa data pooled among six replicate pond s (0.015 hectares). %FO cop cop bin vsuke lne vlukeUNPRO 4/20/2006(%N or %V => 0.5%) elm cst clr spbll seed bbrot Figure 5A.6: UNP pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 589

589 Figure 5A 7 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment 20 April 2006; taxa data pooled among six re plicate ponds (0.015 hectares). %FO cop cop bin vsuke lne vlukeCSM 4/20/2006(%N or %V => 0.5%) spbll cst clr mite elm eph arth non-lhc sn Figure 5A.7: CSM pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 590

590 Figure 5A 8 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment 20 April 2006; taxa data pooled among six replicate pond s (0.015 hectares). %FO cop cop bin vsuke lne vlukeINO 4/20/2006(%N or %V => 0.5%) cst elm eph tlr mite ost dpp Figure 5A.8: INO pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 591

591 Figure 5A 9 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment 27 April 2006; taxa data pooled among six replicate ponds (0.015 hectares). %FO cop cop tlr lne vlukePRO 4/27/2006(%N or %V => 0.5%) cst cld seed bin gsr arth mite elm naup Figure 5A.9: PRO pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 592

592 Figure 5A 10 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP) treatment 27 April 2006; taxa data pooled among six replicate pond s (0.015 hectares). %FO cop cop lne vlukeUNPRO 4/27/2006(%N or %V => 0.5%) cst cld bin mite elm naup iht fls sbt Figure 5A.10: UNP pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 593

593 Figure 5A 11 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment 27 April 2006; tax a data pooled among six replicate pond s (0.015 hectares). %FO cop cop tlr lne vlukeCSM 4/27/2006(%N or %V => 0.5%) elm cst cld bbrot seed bin mite naup Figure 5A.11: CSM pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 594

594 Figure 5A 12 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment 27 April 2 006; taxa data pooled among six replicate pond s (0.015 hectares). %FO cop cop tlr lne vlukeINO 4/27/2006(%N or %V => 0.5%) cst cld seed bin clr ost arth fls dpp bbrot Figure 5A.12: INO pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 595

595 Figure 5A 13 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment 4 May 2006; taxa data pooled among six replicate ponds (0.015 hectares). %FO cop copPRO 5/4/2006(%N or %V => 0.5%) cld bin naup vsuke tlr bbrot clr lne sn nlr cst Figure 5A.13: PRO pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 596

596 Figure 5A 14 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP) treatment 4 May 2006; taxa data pooled among six replicate ponds (0.015 hectares). %FO cop cop vlukeUNPRO 5/4/2006(%N or %V => 0.5%) cld bin naup vsuke tlr bbrot clr lne cst elm spbll Figure 5A.14: UNP pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 597

597 Figure 5A 15 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal ferti lizer (CSM) treatment 4 May 2006; taxa data pooled among six replicate pond s (0.015 hectares). %FO cop cop lne vlukeCSM 5/4/2006(%N or %V => 0.5%) cld bin mite elm naup vsuke tlr clr ost bbrot Figure 5A.15: CSM pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 598

598 Figure 5A 16 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment 4 May 2006; taxa data pooled among six replicate ponds (0.015 hectares). %FO cop cop vlukeINO 5/4/2006(%N or %V => 0.5%) cld bin mite elm naup vsuke tlr clr ost bbrot cst dpp Figure 5A.16: INO pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 599

599 Figure 5A 17 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment 11 May 2006; taxa data pooled among six replicate ponds (0.015 hectares). %FO cop cop vlukePRO 5/11/2006(%N or %V => 0.5%) cld bin vsuke tlr bbrot elm nlr naup gsr seed dpp phr Figure 5A.17: PRO pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 600

600 Figure 5A 18 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP) treatment 11 May 2006; taxa data pooled among six replicate ponds (0.015 hectares). %FO cop cop vlukeUNPRO 5/11/2006(%N or %V => 0.5%) cld bin vsuke tlr bbrot nlr naup seed ost Figure 5A.18: UNP pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 601

601 Figure 5A 19 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO )] for cottonseed meal fertilizer (CSM) treatment 11 May 2006; taxa data pooled among six replicate pond s (0.015 hectares). %FO cop cop vlukeCSM 5/11/2006(%N or %V => 0.5%) cld bin vsuke tlr bbrot clr elm phr seed nlr Figure 5A.19: CSM pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 602

602 Figure 5A 20 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment 11 May 2006; taxa data pooled among six replicate pond s (0.015 hectares). %FO cop cop vlukeINO 5/11/2006(%N or %V => 0.5%) cld bin vsuke tlr bbrot elm nlr mite naup sn Figure 5A.20: INO pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 603

603 Figure 5A 21 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment 18 May 2006; taxa data pooled among six replicate ponds (0.015 hectares). %FO cop copPRO 5/18/2006(%N or %V => 0.5%) cld nlr tlr dap elm phr vluke naup cst Figure 5A.21: PRO pond treatment large plankton90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 604

604 Figure 5A 22 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP) treatment 18 May 2006; taxa data pooled among six replicate ponds (0.015 hectares). %FO cop copUNPRO 5/18/2006(%N or %V => 0.5%) cld elm vluke naup tlr ost Figure 5A.22: UNP pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 605

605 Figure 5A 23 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequ ency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment 18 May 2006; taxa data pooled among six replicate pond s (0.015 hectares). %FO cop cop vlukeCSM 5/18/2006(%N or %V => 0.5%) cld vsuke tlr bbrot nlr ost dap cst elm gsr phr non-lhc spbll Figure 5A.23: CSM pond treatment large plankton90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 606

606 Figure 5A 24 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment 18 May 2006; taxa data pooled among six replicate pond s (0.015 hectares). %FO cop copINO 5/18/2006(%N or %V => 0.5%) cld vsuke tlr nlr bin dap elm phr ost iht dpp Figure 5A.24: INO pond treatment large plankton90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 607

607 Figure 5A 25 Large zooplankton (> 200 m) assemblage community taxa param eters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment 25 May 2006; taxa data pooled among six replicate ponds (0.015 hectares). %FO cop copPRO 5/25/2006(%N or %V => 0.5%) cld elm tlr dap cst vluke seed Figure 5A.25: PRO pond treatment large plankton90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 608

608 Figure 5A 26 Large zooplankton (> 200 m) assemblage community taxa param eters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP) treatment 25 May 2006; taxa data pooled among six replicate ponds (0.015 hectares). %FO cop copUNPRO 5/25/2006(%N or %V => 0.5%) cld elm tlr cst vluke seed ost Figure 5A.26: UNP pond treatment large plankton90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 609

609 Figure 5A 27 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment 25 May 2006; taxa data pooled among six replicate pond s (0.015 hectares). %FO cop copCSM 5/25/2006(%N or %V => 0.5%) cld vsuke elm vluke tlr cst bin naup Figure 5A.27: CSM pond treatment large plankton90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 610

610 Figure 5A 28 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment 25 May 2006; taxa data pooled among six replicate pond s (0.015 hectares). %FO cop copINO 5/25/2006(%N or %V => 0.5%) cld vsuke elm tlr ost dap bin cst vluke naup Figure 5A.28: INO pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 611

611 Figure 5A 29 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment 1 June 2006; taxa data pooled among six replicate ponds (0.015 hectares). %FO cop copPRO 6/1/2006(%N or %V => 0.5%) cld elm cst vluke dap bin mite tlr ost naup fls dpp Figure 5A.29: PRO pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 612

612 Figure 5A 30 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP) treatment 1 June 2006; taxa data pooled among six replicate ponds (0.015 hectares). %FO cop copUNPRO 6/1/2006(%N or %V => 0.5%) cld elm cst vluke dap bin mite ost Figure 5A.30: UNP pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 613

613 Figure 5A 31 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment 1 June 2006; taxa data pooled among six replicate pond s (0.015 hectares). %FO cop copCSM 6/1/2006(%N or %V => 0.5%) cld elm cst vluke dap bin mite naup non-lhc fls Figure 5A.31: CSM pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 614

614 Figure 5A 32 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment 1 June 2006; taxa data pooled among six replicate ponds (0.015 hectare s). %FO cop copINO 6/1/2006(%N or %V => 0.5%) cld elm cst vluke dap bin mite tlr ost Figure 5A.32: INO pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 615

615 Figure 5A 33 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment 8 June 2006; taxa data pooled among six replicate ponds (0.015 hectare s). %FO cop copPRO 6/8/2006(%N or %V => 0.5%) cld elm cst vluke dap tlr sclr vsuke Figure 5A.33: PRO pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 616

616 Figure 5A 34 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP) treatment 8 June 2006; taxa data pooled among six replicate ponds (0.015 hectares). %FO cop copUNPRO 6/8/2006(%N or %V => 0.5%) cld elm cst vluke dap tlr sclr bin lne ost naup fls dpp seed Figure 5A.34: UNP pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 617

617 Figure 5A 35 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment 8 June 2006; taxa data pooled among six rep licate pond s (0.015 hectares). %FO cop copCSM 6/8/2006(%N or %V => 0.5%) cld elm cst vluke dap bin ost tlr sclr vsuke naup Figure 5A.35: CSM pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 618

618 Figure 5A 36 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment 8 June 2006; taxa data pooled among s ix replicate ponds (0.015 hectares). %FO cop copINO 6/8/2006(%N or %V => 0.5%) cld elm cst vluke dap bin ost tlr Figure 5A.36: INO pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 619

619 Figure 5A 37 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment 15 June 2006; taxa data pooled among six replicate ponds (0.015 hectares). %FO cop copPRO 6/15/2006(%N or %V => 0.5%) cld vluke dap tlr elm lne naup Figure 5A.37: PRO pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 620

620 Figure 5A 38 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP) treatment 15 June 2006; taxa data pooled among six replicate ponds (0.015 hectares). %FO cop copUNPRO 6/15/2006(%N or %V => 0.5%) cld vluke dap tlr elm naup ost cst Figure 5A.38: UNP pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 621

621 Figure 5A 39 Large zooplankton (> 200 m) assemblage community taxa parameters [% number ( %N ) % volume ( %V ) % frequency of occurrence ( %FO) ] for cottonseed meal fertilizer (CSM) treatment 15 June 2006; taxa data pooled among six replicate pond s (0.015 hectares). %FO cop copCSM 6/15/2006(%N or %V => 0.5%) cld vluke dap tlr naup cst elm vsuke Figure 5A.39: CSM pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 622

622 Figure 5A 40 Large zooplankton (> 200 m) assemblage community taxa parameters [% number ( %N ) % volume ( %V ) % frequency of occurrence ( %FO) ] for inorganic fertilizer (INO) treatment 15 June 2006; taxa data pooled among six replicate pond s (0.015 hectares). %FO cop copINO 6/15/2006(%N or %V => 0.5%) cld vluke dap tlr cst elm Figure 5A.40: INO pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 623

623 Figure 5A 41 Large zooplankton (> 200 m) assemblage community taxa parameters [% number ( %N ) % volume ( %V ) % frequency of occurrence ( %FO) ] for processed feed (PRO) treatment 13 April 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate ponds (0.015 hectares). %FO cop cop bin elm vsuke lne tlr vlukePRO 4/13/2006 not pooled(%N or %V => 0.5%; +/1 SE) clr mite naup seed spbll bbrot Figure 5A.41: PRO pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 624

624 Figure 5A 42 Large zooplankton (> 200 m) assemblage community taxa parameters [% number ( %N ) % volume ( %V ) % frequency of occurrence ( %FO) ] for unprocessed feed (UNP) treatment 13 April 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate ponds (0.015 hectares). %FO cop cop bin vsuke lne tlr vlukeUNPRO 4/13/2006 not pooled(%N or %V => 0.5%; +/1 SE) seed spbll cst elm Figure 5A.42: UNP pond treatment large planktonbbrot 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 625

625 Figure 5A 43 Large zooplankton (> 200 m) assemblage community taxa parameters [% number ( %N ) % volume ( %V ) % frequency of occurrence ( %FO) ] for cott onseed meal fertilizer (CSM) treatment 13 April 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate pond s (0.015 hectares). %FO cop cop lne tlr non-lhc bin elm vsuke vluke seedCSM 4/13/2006 not pooled(%N or %V => 0.5%; +/1 SE) cst iht mite Figure 5A.43: CSM pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 626

626 Figure 5A 44 Large zooplankton ( > 200 m) assemblage community taxa parameters [% number ( %N ) % volume ( %V ) % frequency of occurrence ( %FO) ] for inorganic fertilizer (INO) treatment 13 April 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate pond s (0.015 hectares). %FO cop cop bin ost elm vsuke lne tlr vluke dppINO 4/13/2006 not pooled(%N or %V => 0.5%; +/1 SE) spbll mite cst Figure 5A.44: INO pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 627

627 Figure 5A 45 Large zooplankton (> 200 m) assemblage community taxa parameters [% number ( %N ) % vo lume ( %V ) % frequency of occurrence ( %FO) ] for processed feed (PRO) treatment 20 April 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate ponds (0.015 hectares). %FO cop cop bin vsuke lne vlukePRO 4/20/2006 not pooled(%N or %V => 0.5%; +/1 SE) elm eph ost dpp clr spbll cst Figure 5A.45: PRO pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 628

628 Figure 5A 46 Large zooplankton (> 200 m) assemblage community taxa parameters [% number ( %N ) % volume ( %V ) % frequency of occurrence ( %FO) ] for unprocessed feed (UNP) treatment 20 April 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate ponds (0.015 hectares). %FO cop cop bin vsuke lne vlukeUNPRO 4/20/2006 not pooled(%N or %V => 0.5%; +/1 SE) elm cst clr spbll seed bbrot ost eph Figure 5A.46: UNP pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 629

629 Figure 5A 47 Large zooplankton (> 200 m) assemblage community taxa parameters [% number ( %N ) % volume ( %V ) % frequency of occurrence ( %FO) ] for cottonseed meal fertilizer (CSM) treatment 20 April 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate pond s (0.015 hectares). %FO cop cop bin vsuke lne vlukeUNPRO 4/20/2006 not pooled(%N or %V => 0.5%; +/1 SE) elm cst clr spbll seed bbrot ost eph Figure 5A.46: UNP pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 630

630 Figure 5A 48 Large zooplankton (> 200 m) assemblage community taxa parameters [% number ( %N ) % volume ( %V ) % frequency of occurrence ( %FO) ] for inorganic fertilizer (INO) treatment 20 April 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate pond s (0.015 hectares). %FO cop cop bin vsuke lne vlukeINO 4/20/2006 not pooled(%N or %V => 0.5%; +/1 SE) cst elm eph tlr mite ost dpp Figure 5A.48: INO pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 631

631 Figure 5A 49 Large zooplankton (> 200 m) assemblage community taxa parameters [% number ( %N ) % volume ( %V ) % frequency of occurrence ( %FO) ] for processed feed (PRO) treatment 27 April 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate ponds (0.015 hectares) %FO cop cop tlr lne vlukePRO 4/27/2006 not pooled(%N or %V => 0.5%; +/1 SE) cst cld seed bin gsr arth mite elm naup dpp Figure 5A.49: PRO pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 632

632 Figure 5A 50 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP) treatment 27 April 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate ponds (0.015 hectares). %FO cop cop lne vlukeUNPRO 4/27/2006 not pooled(%N or %V => 0.5%; +/1 SE) cst cld seed bin mite elm naup vsuke iht sbt dap tlr dpp Figure 5A.50: UNP pond treatment large plankton90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 633

633 Figure 5A 51 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) t reatment 27 April 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate pond s (0.015 hectares). %FO cop cop tlr lne vlukeCSM 4/27/2006 not pooled +/1 se(%N or %V => 0.5%; +/1 SE) elm cst cld bbrot seed bin mite naup dpp Figure 5A.51: CSM pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 634

634 Figure 5A 52 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment 27 April 2006; ta xa %N ( 1 SE) and % V ( 1 SE) for six replicate pond s (0.015 hectares). %FO cop cop tlr lne vlukeINO 4/27/2006 not pooled(%N or %V => 0.5%; +/1 SE) cst cld seed bin clr ost arth fls dpp bbrot cst Figure 5A.52: INO pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 635

635 Figure 5A 53 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment 4 May 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate pond s (0.015 hectares). %FO cop copPRO 5/4/2006 not pooled(%N or %V => 0.5%; +/1 SE) cld bin naup vsuke tlr bbrot dpp clr lne sn nlr ost cst Figure 5A.53: PRO pond treatment large plankton90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 636

636 Figure 5A 54 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP) treatment 4 May 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate ponds (0.015 hectares). %FO cop cop vlukeUNPRO 5/4/2006 not pooled(%N or %V => 0.5%; +/1 SE) cld bin naup vsuke tlr bbrot clr lne cst elm spbll iht Figure 5A.54: UNP pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 637

637 Figure 5A 55 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment 4 May 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate pond s (0.015 hectares). %FO cop cop lne vlukeCSM 5/4/2006 not pooled(%N or %V => 0.5%; +/1 SE) cld bin mite elm naup vsuke tlr clr ost bbrot iht dpp Figure 5A.55: CSM pond treatment large plankton90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 638

638 Figure 5A 56 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % freq uency of occurrence (%FO)] for inorganic fertilizer (INO) treatment 4 May 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate ponds (0.015 hectares). %FO cop cop vlukeINO 5/4/2006 not pooled(%N or %V => 0.5%; +/1 SE) cld bin mite elm naup vsuke tlr ost bbrot cst dpp lne iht phr rr Figure 5A.56: INO pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 639

639 Figure 5A 57 Large zooplankton ( > 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment 11 May 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate ponds (0.015 hectares). %FO cop cop vlukePRO 5/11/2006 not pooled(%N or %V => 0.5%; +/1 SE) cld bin vsuke tlr bbrot elm nlr naup gsr seed dpp phr ost cst Figure 5A.57: PRO pond treatment large plankton90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 640

640 Figure 5A 58 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP) treatment 11 May 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate ponds (0.015 hectares). %FO cop cop vlukeUNPRO 5/11/2006 not pooled(%N or %V => 0.5%; +/1 SE) cld bin vsuke tlr bbrot nlr naup seed ost cst elm dpp Figure 5A.58: UNP pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 641

641 Figure 5A 59 Large zooplankton (size > 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment 11 May 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate pond s (0.015 hectares). %FO cop cop vlukeCSM 5/11/2006 not pooled(%N or %V => 0.5%; +/1 SE) cld bin vsuke tlr bbrot clr elm phr nlr eph naup dpp mite seed Figure 5A.59: CSM pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 642

642 Figure 5A 60 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment 11 May 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate pond s (0.015 hectares). %FO cop cop vlukeINO 5/11/2006 not pooled(%N or %V => 0.5%; +/1 SE) cld bin vsuke tlr bbrot elm nlr mite naup sn cst Figure 5A.60: INO pond treatment large plankton90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 643

643 Figure 5A 61 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment 18 May 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate ponds (0.015 hectares). %FO cop copPRO 5/18/2006 not pooled(%N or %V => 0.5%; +/1 SE) cld nlr tlr dap elm phr vluke naup cst lne bin Figure 5A.61: PRO pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 644

644 Figure 5A 62 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP) treatment 18 May 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate ponds (0.015 hectares). %FO cop copUNPRO 5/18/2006 not pooled(%N or %V => 0.5%; +/1 SE) cld elm vluke nlr tlr ost vsuke gsr phr Figure 5A.62: UNP pond treatment large plankton90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 645

645 Figure 5A 63 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment 18 May 2006; taxa % N ( 1 SE) and % V ( 1 SE) for six replicate pond s (0.015 hectares). %FO cop cop vlukeCSM 5/18/2006 not pooled(%N or %V => 0.5%; +/1 SE) cld vsuke tlr nlr elm gsr cst dap bin clr lne sclr naup phr Figure 5A.63: CSM pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 646

646 Figure 5A 64 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment 18 May 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate pond s (0.015 hectares). %FO cop copINO 5/18/2006 not pooled(%N or %V => 0.5%; +/1 SE) cld vsuke tlr nlr bin dap elm phr ost iht dpp lne cst naup Figure 5A.64: INO pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 647

647 Figure 5A 65 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for proces sed feed (PRO) treatment 25 May 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate ponds (0.015 hectares). %FO cop copPRO 5/25/2006 not pooled(%N or %V => 0.5%; +/1 SE) cld elm tlr dap cst vluke seed Figure 5A.65: PRO pond treatment large plankton90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 648

648 Figure 5A 66 Large zooplankton (size > 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed treatment 25 May 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate pond (0.015 hectares). %FO cop copUNPRO 5/25/2006 not pooled(%N or %V => 0.5%; +/1 SE) cld elm tlr cst vluke ost Figure 5A.66: UNP pond treatment large plankton90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 649

649 Figure 5A 67 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % fre quency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment 25 May 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate pond s (0.015 hectares). %FO cop copCSM 5/25/2006 not pooled(%N or %V => 0.5%; +/1 SE) cld vsuke elm vluke tlr cst bin naup Figure 5A.67: CSM pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 650

650 Figure 5A 68 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment 25 May 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate pond s (0.015 hectares). %FO cop copINO 5/25/2006 not pooled(%N or %V => 0.5%; +/1 SE) cld vsuke elm tlr ost dap bin cst vluke naup bbrot Figure 5A.68: INO pond treatment large plankton90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 651

651 Figure 5A 69 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment 1 June 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate pond s (0.015 hectares). %FO cop copPRO 6/1/2006 not pooled(%N or %V => 0.5%; +/1 SE) cld elm cst vluke dap bin mite tlr ost naup fls lne Figure 5A.69: PRO pond treatment large plankton90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 652

652 Figure 5A 70 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (UNP) treatment 1 June 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate ponds (0.015 hectares). %FO cop copUNPRO 6/1/2006 not pooled(%N or %V => 0.5%; +/1 SE) cld elm cst vluke dap bin mite ost tlr Figure 5A.70: UNP pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 653

653 Figure 5A 71 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment 1 June 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate pond s (0.015 hectares). %FO cop copCSM 6/1/2006 not pooled(%N or %V => 0.5%; +/1 SE) cld elm cst vluke dap bin mite naup non-lhc fls ost Figure 5A.71: CSM pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 654

654 Figure 5A 72 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment 1 June 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate ponds (0.015 he ctares). %FO cop copINO 6/1/2006 not pooled(%N or %V => 0.5%; +/1 SE) cld elm cst vluke dap bin mite tlr ost Figure 5A.72: INO pond treatment large plankton90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 655

655 Figure 5A 7 3. Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment 8 June 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate ponds (0.015 hectares). %FO cop copPRO 6/8/2006 not pooled(%N or %V => 0.5%; +/1 SE) cld elm cst vluke dap tlr sclr vsuke ost naup dpp cst Figure 5A.73: PRO pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 656

656 Figure 5A.74. Large zooplankton (size > 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed treatment 8 June 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate pond (0.015 hectares). %FO cop copUNPRO 6/8/2006 not pooled(%N or %V => 0.5%; +/1 SE) cld elm cst vluke dap tlr sclr lne ost fls dpp seed Figure 5A.74: UNP pond treatment large plankton90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 657

657 Figure 5A 75 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment 8 June 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate pond s (0.015 hectares). %FO cop copCSM 6/8/2006 not pooled(%N or %V => 0.5%; +/1 SE) cld elm cst vluke dap bin ost tlr sclr vsuke naup bbrot Figure 5A.75: CSM pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 658

658 Figure 5A 76 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment 8 June 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate ponds (0.015 hectares). %FO cop copINO 6/8/2006 not pooled(%N or %V => 0.5%; +/1 SE) elm cstvluke dap bin ost tlr cld Figure 5A.76: INO pond treament large plankton90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 659

659 Figure 5A 77 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for processed feed (PRO) treatment 15 June 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate ponds (0.015 hectares). %FO cop copPRO 6/15/2006 not pooled(%N or %V => 0.5%; +/1 SE) cld vluke dap tlr elm lne naup bin cst Figure 5A.77: PRO pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 660

660 Figure 5A 78 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for unprocessed feed (PRO) treatment 15 June 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate ponds (0.015 hectares). %FO cop copUNPRO 6/15/2006 not pooled(%N (number) or %V (volume) => 0.5%; +/1 SE) cld vluke dap tlr elm naup ost cst Figure 5A.78: UNP pond treatment large plankton90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 661

661 Figure 5A 79 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for cottonseed meal fertilizer (CSM) treatment 15 June 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate pond s (0.015 hectares). %FO cop copCSM 6/15/2006 not pooled(%N or %V => 0.5%; +/1 SE) cld vluke dap tlr naup cst elm vsuke mite dpp Figure 5A.79: CSM pond treatment large plankton 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 662

662 Figure 5A 80 Large zooplankton (> 200 m) assemblage community taxa parameters [% number (%N), % volume (%V), % frequency of occurrence (%FO)] for inorganic fertilizer (INO) treatment 15 June 2006; taxa %N ( 1 SE) and % V ( 1 SE) for six replicate pond s (0.015 hectares). %FO cop copINO 6/15/2006 not pooled(%N or %V => 0.5%; +/1 SE) cld vluke dap tlr cst elm ost Figure 5A.80: INO pond treatment large plankton90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90%N %V

PAGE 663

663 APPENDIX B BOOTSTRAP P ING SIMILARITY INDEX PROGRAM Bootstrap Simpl ified Morisitas Similarity and Percent Similarity Index Estimator with Error Measures BASIC Computer Program (written by J.S. Kao). 10 CLS Similarity Index Bootstrap Program, Written by Jon S. Kao (12/12/2008). 20 RANDOMIZE TIME$("milliseconds") seed random # generator w/ # milliseconds past midnight DefaultDir$ = "C: \ gtvar = 0 DIM psidrawdev(50) DIM pnum(50) DIM catnum(120) DIM zoo1(200,200) DIM zoo2(200,200) DIM zoo1$(200,200) DIM zoo2$(200,200) DIM preypct1(150) DIM preypct2(150) DIM Pdraw(45) DIM psitotal(6000) DIM psiiter(6000) DIM psivariter(6000) DIM psistdeviter(6000) DIM psisterriter(6000) DIM psiarraytrials(6000) DIM Mpreypct1(150) DIM Mpreypct2(150) DIM Mdraw(45) DIM MORiter(6000) DIM MORvariter(6000) DIM MORstdeviter(6000) DIM MORsterriter(6000) DIM MORarraytrials(6000) DIM Mcrossprod(150) DIM Pdraw(50) DIM tval(1000) 30 PRINT Treatment Comparison Module 40 PRINT "Welcome to the Similarity Index Bootstrap Program 50 PRINT:PRINT "for c a lculating Simplified Morisita's Index and Percent Similarity Index" 60 PRINT:PRINT Written by Jon S. Kao" 70 PRINT:PRINT:PRINT "95% C onfidence I ntervals c alculated for a t wo t ailed Student's t c ritical v alue, alpha = 0.0 5" 80 PRINT:PRINT:PRINT "enter pond treatment comparison:" 90 PRINT 100 T$(1)= "(1) PRO vs. UNP":T$(2)="(2) PRO vs. CSM":T$(3)="(3) PRO vs. INO":T$(4)="(4) UNP vs. CSM":T$(5)="(5) UNP vs. INO":T$(6)="(6) CSM vs. INO" 120 PRINT T$(1):PRIN T T$(2):PRINT T$(3):PRINT T$(4):PRINT T$(5):PRINT T$(6) 130 PRINT 140 INPUT "enter a number 16, followed by "; comp 150 IF comp <> int(comp) THEN 140 160 IF comp < 1 OR comp > 6 THEN 140

PAGE 664

664 170 PRINT 200 GOSUB [date_check_module] 210 GOSUB [read_t_values] read t values into array for correct 95% CI calcu l ation 220 input "true replicate" pond data: pond numbers, # prey categories and zooplankton counts for 230 each category 2 40 FOR treats = 1 TO 2 treatment groups one and two 260 PRINT 300 GOSUB [treatment_array] tells user what treatments are being compared, does nothing else 320 PRINT 380 INPUT "enter number of ponds: "; pnum(treats) number of ponds = pnum 400 IF pnum(treats) <> int(pnum(treats)) THEN 380 420 IF pnum(treats)<= 0 THEN 380 460 PRINT 500 PRINT "you entered ";pnum(treats); ponds" 520 PRINT 540 cn = 0 600 INPUT "enter total # of prey categori es: "; catnum(treats) 640 IF catnum(treats) <> int(catnum(treats)) THEN 600 680 IF catnum(treats) <= 0 THEN 600 700 PRINT 720 PRINT "you entered ";catnum(treats); categories" 740 GOSUB [data_file_check] 760 PRINT 770 IF dfile$ = "y" THEN 1120 manual zooplankton number input bypass 780 FOR x = 1 TO pnum(treats) create dimensional arrays for # of ponds 800 FOR y = 1 TO catnum(treats) create dimensional array within pond # x for entire zooplankto n suite 820 PRINT how many prey category # ";y 840 PRINT are present for : pond # ";x 860 PRINT 880 IF treats = 2 THEN 980 900 INPUT "enter number of zooplankt er taxa found: "; zoo1(x,y) 920 PRINT zooplankton numbers inser ted into arrays zoo1(x,y) and zoo2(x,y) 940 IF zoo1(x,y) <> int(zoo1(x,y)) OR zoo1(x,y) < 0 THEN 900 960 GOTO 1060 980 INPUT "enter number of zooplankt e r taxa found: "; zoo2(x,y) 1000 PRINT 1040 IF zoo2(x,y) <> int(zoo2(x,y)) OR zoo2(x, y) < 0 THEN 980 1060 PRINT 1080 NEXT y 1100 NEXT x 1120 NEXT treats 1140 df = min(pnum(1),pnum(2)) 1 calculate degrees of freedom for t value calculation: (smaller number of ponds 1) 1160 GOSUB [calc_95CI] choose Student's t score for 95% CI calculation

PAGE 665

665 1190 line 1200 is starting point after iteration run 1200 PRINT manual selection of # of draws and iterations, or pre selected draw and iteration values 1300 PRINT "would you like to select the number of draws and iterations manually?" 1400 PRINT 1500 PRINT or would you like to use pre selected values: 1600 PRINT: PRINT 4, 6, 8, 15 and 20 draws; 10, 50, 100, 200, 500, and 1000 iterations" 1700 PRINT 1800 INPUT "please press or

, followed by : ";manpre$ 1900 IF manpre$ = "m" THEN 2200 2000 IF manpre$ <> "p" THEN 1800 2100 GOTO 43000 2200 module to perform Simplified Morisita's Index or Percent Similarity Index 2220 PRINT 2260 PRINT "you chose manual bootstrap program execution." 2300 PRINT "Are we performing" 2320 PRINT "(1) Simplified Morisita's Index" 2360 PRINT "(2) Percent Similarity Index": PRINT 2400 INPUT "enter 1 or 2 followed by :";index 2440 IF ind ex <> int(index) or index < 0 THEN 2400 2480 if index <> 1 and index <> 2 then 2400 2520 PRINT:PRINT "Your Choice: 2560 IF index = 1 then 2640 2600 Print: Print "Percent Similarity Index" 2620 GOTO 3000 2640 PRINT: PRINT "Morisita's Index" 3000 PRINT select the number of draws module 3100 INPUT "enter the number of draws (typically 3 20) to be taken per iteration: ";draws 3200 IF draws <> int(draws) THEN 3100 3240 IF draws <= 2 THEN 3100 3280 IF draws < 45 THEN 5000 3300 PRINT 3320 PRINT too many 'replicates' drawn!, choose a smaller number!" 3350 PRINT 3380 GOTO 3100 5000 PRINT select the number of iterations to perform module 5100 INPUT "enter the number o f iterations to be performed: "; iter iter a.k.a. trialnum 5200 IF iter <> int(iter) THEN 5100 5300 IF iter <= 1 then 5100 9000 IF index = 1 then 30000 goto Morisita's Index module if index = 1 10000 Percent Similarity Index Module 10 050 PRINT start trial (iteration loop), typically 500 or more iterations 10100 FOR trials = 1 TO iter 10200 set initial 'draw' variables to zero 10300 PSItot3 = 0 PSItot = 0 psidrawsum = 0 10400 FOR pull = 1 to draws run single draw of treatment comparisions 10500 select one "true" pond replicate from each of 2 treatment groups in a two way comparison

PAGE 666

666 10600 get n1 (random pond # among treatment 1) and n2 (random pond # among treatment 2) 10700 gosub [random_number_selection_II] 10800 converts numbers of zooplankton (from randomly selected replicate) into percent of total zooplankton 10820 and inserts these values into two arrays: preypct1(intn1) < -treatment 1 and preypct2 (intn2) < -treatment 2 10850 gosub [generate_percentages] 10880 sets prey category number to either treatment one or two, which ever is less 10900 IF catnum(1) < catnum(2) THEN 10950 ELSE 11000 10950 cn = catnum(1) 10970 GOTO 11100 11000 cn = catnum(2) 11100 run through every category for treatment one and two, which ever is the smaller is added into variable PSItotal 11200 PSI calculation module 11300 FOR y = 1 TO cn run through each zooplankton category 11400 IF preypct1(y) < preypct2(y) THEN PSItot = PSItot + preypct1(y) zooplankton category percent of treatment one is placed into PSItot 11500 IF preypct1(y) > preypct2(y) THEN PSItot = PSItot + preypct2(y) zooplankton cateogry percent of treatmen t two is placed into PSItot 11600 PSItot is sum of minimum percentages for two compared treatments, aka PSI value treat 1 v. treat 2 11700 NEXT y 12000 PSI value from single treatment comparison, a draw of 'x' of these comparisons are average d into draw average 12200 PRINT "single comparison PSI: ";:print using("##.##",PSItot) display single, truncated PSI value 12700 Pdraw(pull) = PSItot 'place individual PSI values of each draw into array Pdraw(x) 12800 PSItot = 0 set sing le comparison value to zero after storing value in array, next loop generates new PSItot 12900 next pull 13000 subroutine [draw_deviation] calculates and prints the draw PSI average, variance, std dev and std error. 13500 gosub [PSI_draw_deviation] 13700 subroutine [grand_total_stats] 13900 gosub [grand_total_stats_PSI] 14400 next trials # of iterations loop 14600 subroutine [grand_total_stats_II] 14700 gosub [grand_total_stats_II] 14800 print 1490 0 bsPSImean = psiitermean/iter 15000 print "the overall PSI (bootstrap) mean is : ";:PRINT USING("##.##",bsPSImean) 15100 PRINT 15200 bsPSIvar = psiitervariance/iter 15300 print "the overall PSI (bootstrap) variance is : ";:PRINT USING("####.####",bsPSIvar)

PAGE 667

667 15400 PRINT 15500 bsPSIstdev = psiiterstddeviation/iter 15600 PRINT "the overall PSI (bootstrap) standard deviation is : ";:PRINT USING("####.####",bsPSIstdev) 15700 PRINT 15800 bsPSIsterror = psiiterstderror/iter 15900 PRINT "the o verall PSI (bootstrap) standard error is : ";:PRINT USING("####.####",bsPSIsterror) 16000 bs95PSILL = bsPSImean tscore2*bsPSIsterror 16100 PRINT 16200 PRINT "the overall PSI (bootstrap) 95% LL is ";: PRINT USING("##.##",bs95PSILL) 16300 bs95PSIUL = bsPSImean + tscore2*bsPSIsterror 16400 PRINT 16500 PRINT "the overall PSI (bootstrap) 95% UL is ";: PRINT USING("##.##",bs95PSIUL) 16600 PRINT 16700 PRINT "the Student's t score for ";df;" degrees of freedom is : ";tscore2 21700 fname$ = G$+"Manual_SimIndex"+sdate$+".txt" 21800 PRINT: PRINT your output file will be : ";fname$ 21900 PRINT 22000 OPEN fname$ FOR APPEND AS #1 23700 PRINT #1, "PSI: "; 23800 PRINT #1, draws: "; 23840 PRINT #1, draws; 23860 PRINT #1, ", iter: "; 23900 PRINT #1, iter; 24000 PRINT #1, ", mean: "; 24100 PRINT #1, bsPSImean; 24200 PRINT #1, ", std err: "; 24300 PRINT #1, bsPSIsterror 24400 PRINT #1, "" 24500 PRINT #1, CI LL: "; 24600 PRINT #1, bs95PSILL; 24700 PRINT #1, ", CI UL: "; 24800 PRINT #1, bs95PSIUL; 24900 PRINT #1, ", tscore: "; 25000 PRINT #1, tscore2; 25100 PRINT #1, alpha = 0.05, twotailed. 25200 CLOSE #1 25300 GOTO 39000 30000 Morisita's Index module 30100 PRINT 30200 FOR trials = 1 TO iter 'start iteration loop, typically 500 or more iterations 30300 set initial 'draw' variables to zero 30400 MORtot3 = 0 MORtot = 0 MORdrawsum = 0 30500 FOR pull = 1 to draws set nu mber of replicate comparisons to be made for each iteration 30600 select one "true" pond replicate from each of 2 treatment groups in a two way comparison 30700 gosub [random_number_selection_II] get n1 (random pond # among treatment 1) and n2 (ra ndom pond # among treatment 2) 30800 converts numbers of zooplankton (from randomly selected replicate) into percent of total zooplankton 30900 and inserts these values into two arrays: preypct1(intn1) < -treatment 1 and preypct2(intn2) < -t reatment 2

PAGE 668

668 31000 gosub [generate_percentages] 31600 Gosub [Morisita_Draw_Module] 31700 PRINT "single comparison Morisita's Index: ";:print using("##.##",MorisitaDraw) display single, truncated Morisita's value 31900 Mdraw(pull) = MorisitaDraw 'place individual Morisita's Index values of each draw into array Mdraw(n) array 32000 MorisitaDraw = 0 reset MorisitaDraw value to zero after value is stored in array Mdraw(n), loop to next value 32100 next pull 32200 subroutine [dr aw_deviation] calculates and prints the draw MOR average, variance, std dev and std error. 32300 gosub [MOR_draw_deviation] 32400 subroutine [grand_total_stats] 32500 gosub [grand_total_stats_MOR] 32600 next trials loop to next iteration of size iter 32700 subroutine [grand_total_stats_II], generate bootstrap statistical values 32800 gosub [grand_total_stats_II_MOR] 32900 PRINT 33000 bsMORmean = MORitermean/iter 33100 print "the overall Morisita's Index (bootstra p) mean is : ";:PRINT USING("##.##",bsMORmean) 33200 PRINT 33300 bsMORvar = MORitervariance/iter 33400 print "the overall Morisita's Index (bootstrap) variance is : ";:PRINT USING("####.####",bsMORvar) 33500 PRINT 33600 bsMORstdev = MORiterstddeviation/iter 33700 PRINT "the overall Morisita's Index (bootstrap) standard deviation is : ";:PRINT USING("####.####",bsMORstdev) 33800 PRINT 33900 bsMORsterror = MORiterstderror/iter 34000 PRINT "the overall Morisita's Index (bootstrap) standard er ror is : ";:PRINT USING("####.####",bsMORsterror) 34100 bs95MORLL = bsMORmean tscore2*bsMORsterror 34200 PRINT 34300 PRINT "the overall Morisita's (bootstrap) 95% LL is ";: PRINT USING("##.##",bs95MORLL) 34400 bs95MORUL = bsMORmean + tscore2*bsMORsterror 34500 PRINT 34600 PRINT "the overall Morisita's (bootstrap) 95% UL is ";: PRINT USING("##.##",bs95MORUL) 34650 PRINT 34700 PRINT:PRINT your output file will be : ";fname$ 34750 PRINT 34800 OPEN fname$ FOR APPEND AS #1 34850 PRINT #1, "" 34900 PRINT #1, "Morisita: "; 34920 PRINT #1, draws: "; 34940 PRINT #1, draws; 35000 PRINT #1, ", iter: "; 35100 PRINT #1, iter;

PAGE 669

669 35200 PRINT #1, ", mean: "; 35300 PRINT #1, bsMORmean; 35400 PRINT #1, ", std err: "; 35500 PRINT #1 bsMORsterror 35550 PRINT #1, "" 35600 PRINT #1, CI LL: "; 35700 PRINT #1, bs95MORLL; 35800 PRINT #1, ", CI UL: "; 35900 PRINT #1, bs95MORUL; 36000 PRINT #1, ", tscore: "; 36100 PRINT #1, tscore2; 36200 PRINT #1, ", alpha = 0.05, twotailed. 36300 PRINT #1, 36400 CLOSE #1 37000 GOTO 58950 Finished MANUAL MODULE go to "Do you wish to continue?" Branch >>>>>>>>>>>>>>>>>>>>> PRE SELECTED VALUES ROUTINE <<<<<<<<<<<<<<<<<<< 43000 read data into arrays for # draws and # iterations, and create loops 43100 PRINT 43200 FOR bigruns = 1 to 26 43300 READ iter, draws 43400 IF iter = 0 and draws = 0 THEN 59000 end of data, go to "do you wish to continue?" module 44000 PRINT Percent Similarity Index Module (start trial iteration loop, typically 500 or more iterations) 44100 FOR trials = 1 TO iter 44200 set initial 'draw' variables to zero 44300 PSItot3 = 0 PSItot = 0 psidrawsum = 0 44700 FOR pull = 1 to draws run single draw of treatment comparisions 44800 select one "true" pond replicate from each of 2 treatment groups in a two way comparison 44900 get n1 (random pond # among treatment 1) and n2 ( random pond # among treatment 2) 45000 gosub [random_number_selection_II] 45100 converts numbers of zooplankton (from randomly selected replicate) into percent of total zooplankton 45200 and inserts these values into two arrays: preypct1(i ntn1) < -treatment 1 and preypct2(intn2) < -treatment 2 45300 gosub [generate_percentages] 45400 sets prey category number to either treatment one or two, which ever is less 45500 IF catnum(1) < catnum(2) THEN 45600 ELSE 45800 45600 cn = ca tnum(1) 45700 GOTO 46000 45800 cn = catnum(2) 46000 run through every category for treatment one and two, which ever is the smaller is added into variable PSItotal 46100 PSI calculation module 46200 FOR y = 1 TO cn run through each zooplankton category 46300 IF preypct1(y) < preypct2(y) THEN PSItot = PSItot + preypct1(y) zooplankton category percent of treatment one is placed into PSItot

PAGE 670

670 46400 IF preypct1(y) > preypct2(y) THEN PSItot = PSItot + preypct2(y) zoopla nkton cateogry percent of treatment two is placed into PSItot 46500 PSItot is sum of minimum percentages for two compared treatments, aka PSI value treat 1 v. treat 2 46800 NEXT y 47000 PSI value from single treatment comparison, a draw of x' of these comparisons are averaged into draw average 47100 PRINT "single comparison PSI: ";:print using("##.##",PSItot) display single, truncated PSI value 47200 Pdraw(pull) = PSItot 'place individual PSI values of each draw into array Pdraw(x) 47300 PSItot = 0 set single comparison value to zero after storing value in array, next loop generates new PSItot 47400 next pull 47500 subroutine [draw_deviation] calculates and prints the draw PSI average, variance, std dev and std err or. 47700 gosub [PSI_draw_deviation] 47800 subroutine [grand_total_stats] 47900 gosub [grand_total_stats_PSI] 48000 next trials # of iterations loop 48100 subroutine [grand_total_stats_II] 48200 gosub [grand_total_stats_II] 48300 print 48500 bsPSImean = psiitermean/iter 48600 print "the overall PSI (bootstrap) mean is : ";:PRINT USING("##.##",bsPSImean) 48700 PRINT 48800 bsPSIvar = psiitervariance/iter 48900 print "the overall PSI (bootstrap) variance is : ";:PRINT USING("####.####",bsPSIvar) 49000 PRINT 49100 bsPSIstdev = psiiterstddeviation/iter 49200 PRINT "the overall PSI (bootstrap) standard deviation is : ";:PRINT USING("####.####",bsPSIstdev) 49300 PRINT 49400 bsPSIsterror = psiiterstderror /iter 49500 PRINT "the overall PSI (bootstrap) standard error is : ";:PRINT USING("####.####",bsPSIsterror) 49600 bs95PSILL = bsPSImean tscore2*bsPSIsterror 49700 PRINT 49800 PRINT "the overall PSI (bootstrap) 95% LL is ";: PRINT USING("##.##",bs 95PSILL) 49900 bs95PSIUL = bsPSImean + tscore2*bsPSIsterror 50000 PRINT 50100 PRINT "the overall PSI (bootstrap) 95% UL is ";: PRINT USING("##.##",bs95PSIUL) 50110 PRINT 50115 PRINT "the Student's t score is ";tscore2;" for ";df;" degrees of free dom" 50120 fname$ = G$+"SimIndex"+sdate$+".txt" 50140 PRINT: PRINT your output file will be : ";fname$ 50160 PRINT 50180 OPEN fname$ FOR APPEND AS #1 50200 PRINT #1, T$(comp); 50210 PRINT #1, "; 50220 PRINT #1, sdate$ 50230 PRINT #1,

PAGE 671

671 50240 PRINT #1, "Draws: "; 50260 PRINT #1, draws 50280 PRINT #1, "" 50300 PRINT #1, "PSI: "; 50400 PRINT #1, iter: "; 50500 PRINT #1, iter; 50600 PRINT #1, ", mean: "; 50700 PRINT #1, bsPSImean; 50800 PRINT #1, ", std err: "; 50900 PRINT #1, bsPSIsterror 51000 PRINT #1, CI LL: "; 51100 PRINT #1, bs95PSILL; 51200 PRINT #1, ", CI UL: "; 51300 PRINT #1, bs95PSIUL; 51400 PRINT #1, ", tscore: "; 51500 PRINT #1, tscore2; 51600 PR INT #1, ", alpha = 0.05, two tailed. 51700 PRINT #1, 51800 CLOSE #1 53000 Morisita's Index module 53100 PRINT 53200 FOR trials = 1 TO iter 'start iteration loop, typically 500 or more iterations 53300 se t initial 'draw' variables to zero 53400 MORtot3 = 0 MORtot = 0 MORdrawsum = 0 53500 FOR pull = 1 to draws set number of replicate comparisons to be made for each iteration 53600 select one "true" pond replicate from each of 2 treatment groups in a two way comparison 53700 gosub [random_number_selection_II] get n1 (random pond # among treatment 1) and n2 (random pond # among treatment 2) 53800 converts numbers of zooplankton (from randomly selected replicate) into percent of total zooplankton 53900 and inserts these values into two arrays: preypct1(intn1) < -treatment 1 and preypct2(intn2) < -treatment 2 54000 gosub [generate_percentages] 54100 Gosub [Morisita_Draw_Module] 54200 PRI NT "single comparison Morisita's Index: ";:print using("##.##",MorisitaDraw) display single, truncated Morisita's value 54300 Mdraw(pull) = MorisitaDraw 'place individual Morisita's Index values of each draw into array Mdraw(n) array 54400 MorisitaDraw = 0 reset MorisitaDraw value to zero after value is stored in array Mdraw(n), loop to next value 54500 next pull 54600 subroutine [draw_deviation] calculates and prints the draw MOR average, variance, std dev and std error. 54700 gosub [MOR_draw_deviation] 54800 subroutine [grand_total_stats] 54900 gosub [grand_total_stats_MOR]

PAGE 672

672 55000 next trials loop to next iteration of size iter 55100 subroutine [grand_total_stats_II], generate bootstrap statistical values 55200 gosub [grand_total_stats_II_MOR] 55300 print 55400 bsMORmean = MORitermean/iter 55500 print "the overall Morisita's Index (bootstrap) mean is : ";:PRINT USING("##.##",bsMORmean) 55600 PRINT 55700 bsMORvar = MORitervariance/iter 55800 print "the overall Morisita's Index (bootstrap) variance is : ";:PRINT USING("####.####",bsMORvar) 55900 PRINT 56000 bsMORstdev = MORiterstddeviation/iter 56100 PRINT "the overall Morisita's Index (bootstrap) standard deviation is : ";:PRINT USING("####.####",bsMORstdev) 56200 PRINT 56300 bsMORsterror = MORiterstderror/iter 56400 PRINT "the overall Morisita's Index (bootstrap) standard error is : ";:PRINT USING("####.####",bsMORsterror) 56500 bs95MORLL = bsMORmean tscore2*bsMORsterro r 56600 PRINT 56700 PRINT "the overall Morisita's (bootstrap) 95% LL is ";: PRINT USING("##.##",bs95MORLL) 56800 bs95MORUL = bsMORmean + tscore2*bsMORsterror 56900 PRINT 57000 PRINT "the overall Morisita's (bootstrap) 95% UL is ";: PRINT USING("# #.##",bs95MORUL) 57020 PRINT 57040 PRINT "the Student's t score is ";tscore2;" for ";df;" degrees of freedom" 57060 PRINT 57100 OPEN fname$ FOR APPEND AS #1 add analyses to text file 57200 PRINT #1, "Morisita: "; 57300 PRINT #1, iter: ; 57400 PRINT #1, iter; 57500 PRINT #1, ", mean: "; 57600 PRINT #1, bsMORmean; 57700 PRINT #1, ", std err: "; 57800 PRINT #1, bsMORsterror 57900 PRINT #1, CI LL: "; 58000 PRINT #1, bs95MORLL; 58100 PRINT #1, ", CI UL: "; 58200 PRINT #1, bs95MORUL; 58300 PRINT #1, ", tscore: "; 58400 PRINT #1, tscore2; 58500 PRINT #1, ", alpha = 0.05, twotailed. 58700 PRINT #1, 58800 CLOSE #1 58900 NEXT bigruns 58950 RESTORE reset data pointer to beginning data value 59000 PRINT do you wish to continue module 59100 INPUT "do you wish to continue? press or , followed by : ";z$ 59200 IF z$ = "y" THEN 59500 59300 PRINT 59400 PRINT "goodbye." : END

PAGE 673

673 59500 PRINT 59600 PRINT "Do you wish to continue with the same data set?" 59700 INPUT "press or , followed by : ";dset$ 59800 RESTORE 59900 IF dset$ = "y" THEN 1200 60000 CLS 62000 GOTO 60 various subroutines [treatment_array] 100000 if comp <> 1 or treats <> 1 then goto 100100 100050 print "please enter data for PRO treatment" 100080 return 100100 if comp <> 1 or treats <> 2 then goto 100200 100150 print "please enter data for UNP treatment" 100180 return 100200 if comp <> 2 or treats <> 1 then goto 100300 100250 print "please enter data for PRO treatment" 100280 return 100300 if comp <> 2 or treats <> 2 then goto 100400 100350 print "please enter data for CSM treatment" 100380 return 100400 if comp <> 3 or treats <> 1 then goto 100500 100450 print "please enter data for PRO treatment" 100480 return 100500 if comp <> 3 or treats <> 2 then goto 100600 100550 print "please enter data for INO treatment" 100580 return 100600 if comp <> 4 or treats <> 1 then goto 100700 100650 print "please enter data f or UNP treatment" 100680 return 100700 if comp <> 4 or treats <> 2 then goto 100800 100750 print "please enter data for CSM treatment" 100780 return 100800 if comp <> 5 or treats <> 1 then goto 100900 100850 print "please enter data for UNP treatment" 100880 return 100900 if comp <> 5 or treats <> 2 then goto 101000 100950 print "please enter data for INO treatment" 100980 return 101000 if comp <> 6 or treats <> 1 then goto 101100 101050 print "please enter data for CSM treatment" 101080 return 101100 if comp <> 6 or treats <> 2 then goto 101180

PAGE 674

674 101150 print "please enter data for INO treatment" 101300 return [random_number_selection_II] 110000 module to generate random numbers to select which replicate is chosen in treatment comparison 110200 seed1 = time$("milliseconds") 110300 seed1$ = str$(seed1) 110400 seeda = val(right$(seed1$,2)) 110500 if seeda <=0 then 110200 110600 randomize seeda 110700 n1 = rnd(seeda/100)+1 replicate for treatment one chosen 110800 n1 = n1 int(n1) 110900 n1int = int(n1 1000) 111000 if n1int > pnum(1) or n1int = 0 then 110700 random number generated up to 'true replicate' sample size 111100 for t = 1 to 5000: next t 111200 seed2 = time$("milliseconds") 111300 seed2$ = str $(seed2) 111400 seedb = val(right$(seed2$,2)) 111500 if seedb <= 0 then 111200 111600 randomize seedb 111700 n2 = rnd(seedb/100)+1 replicated for treatment two chosen 111800 n2 = n2 int(n2) 111900 n2int = int(n2 1000) 112000 IF n2int > pnum(2) OR n2int = 0 THEN 111700 random number generated up to 'true replicate' sample size 112100 RETURN [generate_percentages] 117100 totprey1 = 0 totprey2 = 0 117200 FOR y = 1 TO catnum(1) 117300 totprey1 = totprey1 + zoo1(n1int,y) counts all prey of category one; aka cat sum treatment one 117400 NEXT y 117500 FOR y = 1 TO catnum(2) 117600 totprey2 = totprey2 + zoo2(n2int,y) counts all prey of category two; aka cat sum treatment two 117700 NEXT y 118250 fo r y = 1 to catnum(1) 118280 preypct1(y) = ((zoo1(n1int,y))/totprey1)*100 generate and store percentage value of prey category y, treat 1 118300 next y 118600 for y = 1 to catnum(2) 118700 preypct2(y) = ((zoo2(n2int,y))/totprey2)*100 generate and store percentage value of prey category y, treat 2 118900 next y 119000 return [PSI_draw_deviation] 130900 psidrawsum = 0 131000 for pulls = 1 to draws 131100 psidrawsum = psidrawsum + Pdraw(pulls) 131200 next pulls 131250 psidrawavg = psidrawsum /draws average PSI for pull of size 'draw' replicates 132000 for pulls = 1 to draws

PAGE 675

675 132100 psidrawdevsum = psidrawdevsum + (Pdraw(pulls) psidrawavg)^2 132200 next pulls 132400 for pulls = 1 to draws 133500 psidrawdevtot = psidra wdevtot + psidrawdev(pulls) sum of squares of deviaitons of ind. PSI from mean PSI of draw 133600 next pulls 133800 PRINT "PSI draw mean = ";:PRINT USING("##.##",psidrawavg) 133900 psidrawvar = psidrawdevsum/(draws 1) 134000 print "PSI draw variance = ";:PRINT USING("####.####",psidrawvar) 134100 psidrawdev = sqr(psidrawvar) 134200 print "PSI draw standard deviation = ";:PRINT USING("#####.####",psidrawdev) 134300 psidrawsterr = psidrawdev/sqr(draws) 134400 print "PSI draw standard error = ";:PR INT USING("####.####",psidrawsterr) 134500 psidrawsum = 0 134700 psidrawdevsum = 0 140000 return [grand_total_stats_PSI] 150100 module to place individual draw stats into overall PSI 'iterations' array and stats 150200 psiiter(trials) = psidrawavg 150300 psivariter(trials) = psidrawvar 150400 psistdeviter(trials) = psidrawdev 150500 psisterriter(trials) = psidrawsterr 160000 return [grand_total_stats_II] 170000 module to run through all iteration arrays and generate overall 'bootstrap' stats 170050 psiitermean = 0 psiitervariance = 0 psiiterstddeviation = 0 psiiterstderror = 0 170100 for trials = 1 to iter 170200 psiitermean = psiitermean + psiiter(trials) 170300 psiitervariance = psiitervariance + psivariter(trials) 170400 psiiterstddeviation = psiiterstddeviation + psistdeviter(trials) 170500 psiiterstderror = psiiterstderror + psisterriter(trials) 170800 next trials 180000 return [MOR_draw_deviation] 200000 MORdrawsum = 0 MORdrawdevsum = 0 Mprey1sum = 0 Mprey2sum = 0 Mcrosssum = 0 200100 for pulls2 = 1 to draws 200200 MORdrawsum = MORdrawsum + Mdraw(pulls2) Mdraw(x) array contains individual draw comparisons... 4, 6, 8, etc. 200300 next pulls2 200400 MORdrawavg = MORdrawsum /draws average PSI for pull of size 'draw' replicates 200500 for pulls2 = 1 to draws 200600 MORdrawdevsum = MORdrawdevsum + (Mdraw(pulls2) MORdrawavg)^2 200700 next pulls2

PAGE 676

676 210200 PRINT "Simplified Morisita's Index draw mean = ";:PRINT USING("##.##",MORdrawavg) 210300 MORdrawvar = MORdrawdevsum/(draws 1) 210400 print "Morisita's Index draw variance = ";:P RINT USING("####.####",MORdrawvar) 210500 MORdrawdev = sqr(MORdrawvar) 210600 print "Morisita's Index draw standard deviation = ";:PRINT USING("#####.####",MORdrawdev) 210700 MORdrawsterr = MORdrawdev/sqr(draws) 210800 print "Morisita's Index draw standard error = ";:PRINT USING("####.####",MORdrawsterr) 220000 return [grand_total_stats_MOR] 225100 module to place individual draw stats into overall PSI 'iterations' array and stats 225200 MORiter(trials) = MORdrawavg 225300 MORvariter(trials) = MORdrawvar 225400 MORstdeviter(trials) = MORdrawdev 225500 MORsterriter(trials) = MORdrawsterr 300000 return [grand_total_stats_II_MOR] 310000 module to run through all iteration arrays and generate overall 'bootstrap' stats 310100 MORitermean = 0 MORitervariance = 0 MORiterstddeviation = 0 MORiterstderror = 0 310300 for trials2 = 1 to iter 310400 MORitermean = MORitermean + MORiter(trials2) 310500 MORitervariance = MORitervariance + MORvariter(trials2) 310600 MORiterstddeviation = MORiterstddeviation + MORstdeviter(trials2) 310700 MORiterstderror = MORiterstderror + MORsterriter(trials2) 310800 next trials2 311000 return [Morisita_Draw_Module] 318810 Mprey1sum = 0 Mprey2sum = 0 Mcrosssum = 0 Mnumsum1 = 0 Mnumsum2 = 0 Mnumsumsum1 = 0 Mnumsumsum2 = 0 318820 for y = 1 to catnum(1) 318840 Mpreypct1(y) = preypct1(y)*(preypct1(y) 1) 318880 Mprey1sum = Mprey1sum + Mpreypct1(y) Mprey1sum aka Morisita's numerator 318900 next y 319000 for y = 1 to catnum(2) 319020 Mpreypct2(y) = preypct2(y)*(preypct2(y) 1) Mprey2sum aka Morisita's numerator 319040 Mprey2sum = Mprey2sum + Mpreypct2(y) 319060 next y 319080 for y = 1 to catnum(1) 319100 Mnumsum1 = Mnumsum1 + preypct1(y) cat sum treatment 1 319200 next y 319300 for y = 1 to catnum(2) 319400 Mnumsum2 = Mnumsum2 + preypct2(y) cat su m treatment 2 319500 next y 319600 Mnumsumsum1 = Mnumsum1*(Mnumsum11) 319700 Mnumsumsum2 = Mnumsum2*(Mnumsum21)

PAGE 677

677 320000 Lambda1 = Mprey1sum/Mnumsumsum1 aka Morisita's denominator treatment 1 320100 Lambda2 = Mprey2sum/Mnumsumsum2 aka Morisita's denominator treatment 2 320200 if catnum(1) < catnum(2) then 320800 320300 for y = 1 to catnum(2) 320400 Mcrossprod(y) = preypct1(y) preypct2(y) 320500 Mcrosssum = Mcrosssum + Mcrossprod(y) 320600 ne xt y 320700 goto 321200 320800 for y = 1 to catnum(1) 320900 Mcrossprod(y) = preypct1(y) preypct2(y) 321000 Mcrosssum = Mcrosssum + Mcrossprod(y) 321100 next y 321200 MorisitaDraw = (Mcrosssum*2)/((Lambda1+Lambda2)*(Mnumsum1*Mnumsum2)) 323000 re turn [date_check_module] 330000 INPUT "enter sample date month: , followed by : ";smonth 330100 IF smonth < 1 or smonth > 12 THEN 330000 330200 IF smonth <> int(smonth) THEN 330000 330300 INPUT "enter sample date day:

, followed by : ";sday 330400 IF sday < 1 or sday > 31 THEN 330300 330500 IF sday <> int(sday) THEN 330300 330600 INPUT "enter sample date year: , followed by : ";syear 330800 IF syear <> int(syear) THEN 330600 331000 smonth$ = str$(smonth) 331200 sday$ = str$(sday) 331400 syear$ = str$(syear) 331500 IF len(syear$) <> 4 THEN 330600 331700 sdate$ = smonth$ + sday$ + syear$ 331800 IF comp = 1 THEN G$ = "PU" 331900 IF comp = 2 THEN G$ = "PC" 332000 IF comp = 3 THEN G$ = "PI" 332100 IF comp = 4 THEN G$ = "UC" 332200 IF comp = 5 THEN G$ = "UI" 332300 IF comp = 6 THEN G$ = "CI" 333000 RETURN [data_file_check] 335000 PRINT 335100 PRINT "do you want to read data from a text file?" 335200 PRINT 335300 INPUT "please press or , fol lowed by : ";dfile$ 335400 IF dfile$ = "n" THEN 340000 335500 IF dfile$ = "y" THEN 336000 335700 GOTO 335300 336000 PRINT 336050 IF treats = 2 THEN 337750 336100 PRINT "please load file with data from first treatment of treatment pair : 3 36150 PRINT T$(comp) 336200 FILEDIALOG, "Open", "c: \ *.txt", fname$ 336300 IF fname$ = "" then 336200 336400 PRINT

PAGE 678

678 336500 OPEN fname$ FOR INPUT AS #2 336600 FOR x = 1 to pnum(treats) 336700 FOR y = 1 to catnum(treats) 337000 zoo1$(x,y) = inputto$(#2,",") 337050 FOR t = 1 TO 400: next t 337100 zoo1(x,y) = val(zoo1$(x,y)) 337300 NEXT y 337400 NEXT x 337500 CLOSE #2 337600 FOR t = 1 to 5000: next t 337700 IF treats = 1 THEN 340000 337750 PRINT 337800 PRINT "please load file with data from second treatment of treatment pair : 337900 PRINT T$(comp) 338000 FILEDIALOG, "Open", "c: \ *.txt", fname2$ 338100 IF fname2$ = "" then 336200 338300 OPEN fname2$ FOR INPUT AS #3 338500 FOR x = 1 to pnum(treats) 338700 FOR y = 1 to catnum(tr eats) 338900 zoo2$(x,y) = inputto$(#3,",") 339100 for t = 1 to 100: next t 339300 zoo2(x,y) = val(zoo2$(x,y)) 339500 NEXT y 339700 NEXT x 339900 CLOSE #3 340000 RETURN [read_t_values] 360000 FOR RT = 1 to 35 360500 READ tscore 360550 IF tscore = 0 THEN 360700 360600 tval(RT) = tscore 360700 NEXT RT 360900 RETURN [calc_95CI] calculate t score based upon degrees of freedom (i.e. smaller pond treatment number of replicates 1) 370500 IF df <= 30 THEN hdf = df 380000 IF df > 30 AN D df < 40 THEN hdf = 31 380100 IF df >= 40 AND df < 60 THEN hdf = 32 380200 IF df >= 60 AND df < 100 THEN hdf = 33 380400 IF df >= 100 AND df < 130 THEN hdf = 34 380500 IF df >= 130 THEN hdf = 35 390000 tscore2 = tval(hdf) set t score based upon array holding t scores and the degrees of freedom 39500 RETURN t score array: alpha = 0.05, twotailed DATA 12.706, 4.303, 3.182, 2.776, 2.571, 2.447, 2.365, 2.306, 2.262, 2.228 DATA 2.201, 2.179, 2.160, 2.145, 2.131, 2.120, 2.110, 2.101, 2.093, 2.086 DATA 2.080, 2.074, 2.069, 2.064, 2.060, 2.056, 2.052, 2.048, 2.045, 2.042 DATA 2.021, 2.000, 1.980, 1.960, 0

PAGE 679

679 automatic preset number of iterations and draws (not manual) DATA 10, 4, 10, 6, 10, 8, 10, 15, 10, 20 DATA 50, 4, 50, 6, 50, 8, 50, 15, 50, 20 DATA 100, 4, 100, 6, 100, 8, 100, 15, 100, 20 DATA 500, 4, 500, 6, 500, 8, 500, 15, 500, 20 DATA 1000, 4, 1000, 6, 1000, 8, 1000, 15, 1000, 20 DATA 0, 0 end of data

PAGE 680

680 Data input file format and exa mple output file for bootstrap simplified Morisitas and Percent Similarity Index program: Input file format (Microsoft Notepad text file *.txt): each line one replicate, 30 taxon categories (# of each taxon within given replicate) Typical data input file (actual data PRO ponds 1 June 2006): 24,6,31,0,0,0,4,1,0,0,14,0,0,0,0,0,1,0,0,0,0,0,1,0,0,0,0,0,0,0 30,0,44,0,1,0,0,0,0,2,1,1,0,0,0,3,0,0,0,0,0,0,1,0,0,0,0,0,0,0 181,14,50,5,0,0,2,1,0,0,3,6,0,2,0,0,0,0,0,0,0,1,1,0,0,0,0,0,0,0 37,0,20,0,0,0,14,0,0,0,3,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 6,48,8,5,2,1,6,1,0,4,1,1,0,2,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0 86,54,37,3,1,0,6,1,0,0,4,4,0,4,0,3,0,0,0,0,0,1,0,6,0,0,0,0,0,0 Actual partial output file (consisting of 4, 6, 8, 15, 20 draws per iteration) as Microsoft Note pad file (*.txt): (Output format: # iterations, mean index value, standard error, 95% CI LL, 95% CI UL) (5) UNP vs. INO, PSI 10, 56.4819258, 10.2436675, 36.4043376, 76.5595141 Morisit a, 10, 0.83698221, 0.09955040, 0.64186342, 1.032101 (5) UNP vs. INO, PSI 10, 61.5602319, 10.1860904, 41.5954948, 81.524969 Morisita, 10, 0.7322554, 0.97426733e 1, 0.541299, 0.9232118 (5) UNP vs. INO, PSI 10, 53.7052246, 7.67315261, 38.6658455, 68.7446037 Morisita, 10, 0.71025277, 0.83782942e 1, 0.54 603821, 0.87446734 (5) UNP vs. INO, PSI 10, 56.9714794, 6.48779756, 44.2553962, 69.6875626 Morisita, 10, 0.6490744, 0.77994348e 1, 0.49620547, 0.80194332 (5) UNP vs. INO, PSI 50, 60.2229343, 13.893479, 32.9917153, 87.4541532 Morisita, 50, 0.74084471, 0.12845958, 0.48906392, 0.99262549 (5) UNP vs. INO, PSI 50, 62.0175474, 11.9668357, 38.5625496, 85.4725453 Morisita, 50, 0.7168991, 0.12916443, 0.46373683, 0.97006138 (5) UNP vs. INO, PSI 50, 58.6950058, 9.73918113, 39.6062108, 77.7838008

PAGE 681

681 Morisita, 50, 0.68623326, 0.11613755, 0.45860366, 0.91386285 (5) UNP vs. INO, PSI 50, 57.2556306, 7.4335738, 42.685826, 71.8254353 Morisita, 50, 0.68424005, 0.88926879e 1, 0.50994337, 0.85853674 (5) UNP vs. INO, PSI 50, 56.5456242, 6.32544941, 44.1477434, 68.9435051 Morisita, 50, 0.68632794, 0.76004842e 1, 0.53735845, 0.83529743 (5) UNP vs. INO, P SI 100, 60.1864029, 14.4598765, 31.845045, 88.5277609 Morisita, 100, 0.79337088, 0.12171307, 0.55481327, 1.03192849 (5) UNP vs. INO, PSI 100, 58.60070 76, 11.018936, 37.0035929, 80.1978222 Morisita, 100, 0.73957487, 0.12044588, 0.50350094, 0.97564879 (5) UNP vs. INO, PSI 100, 58.5079084, 9.89758171, 39.1086482, 77.9071685 Morisita, 100, 0.74588278, 0.10739447, 0.53538962, 0.95637594 (5) UNP vs. INO, PSI 100, 58.361429, 7.65176116, 43.3639772, 73.3588809 Morisita, 100, 0.69685134, 0.85044149e 1, 0.53016481, 0.86353788 (5) UNP vs. INO, PSI 100, 59.4303341, 6.57935342, 46.5348014, 72.3258668 Morisita, 100, 0.68755745, 0.74509769e 1, 0.54151831, 0.8335966 (5) UNP vs. INO, PSI 500, 61.4148541, 12.3721985, 37.165345, 85.6643633 Morisita, 500, 0.77519512, 0.13241533, 0.51566108, 1.03472916 (5) UNP vs. INO, PSI 500, 59.1547502, 11.6211914, 36.377215, 81.9322853 Morisita, 500, 0.73502015, 0.12419932, 0.49158948, 0.97845082

PAGE 682

682 LIST OF REFERENCES Adams, C.M., S.L. Larkin, D.J. Lee, R.L. Degner, and J.W. Milon. 2001. International trade in live, ornamental fish in the U.S. and Florida. Sea Grant Technical Paper 113. Florida Sea Grant / University of Florida, Gainesville, Florida. U.S.A. Ali, A.B. 1990. Seasonal dynamics of microcrustacean and rotifer communit ies in Malaysian rice fields used for rice fish farming. Hydrobiologia 206:139148. APHA/AWWA. 1999. Standard methods for the examination of water and wastewater, 20th edition. American Public Health Association and American Water Works Association publis hers, Washington, DC, U.S.A. Arndt, R.E., Hardy, R.W., Sugiura, S.H., and F.M. Dong. 1999. Effects of heat treatment and substitution level on palatability and nutritional value of soy deffated flour in feeds for Coho Salmon. Aquaculture 180:129 145. Art hington, A.H. 1989. Diet of Gambusia affinis holbrooki, Xiphophorus helleri, X. maculates and Poecilia reticulate (Pisces: Poeciliidae) in streams of Southeastern Queensland, Australia. Asian Fisheries Science 2:193 212. Axelrod, H.R. and L. Wischnath. 1991. Swordtails and platies. TFH Publications Inc. Neptune City, New Jersey, U.S.A. Barkoh, A. 1996. Effects of three fertilization treatments on water quality, zooplankton, and striped bass fingerling production in plastic lined ponds. The Progressive Fish Culturist 58:237247. Barrows, F.T., Bellis, D., Krogdahl, A., Silverstein, J.T., Herman, E.M., Sealey, W.M., Rust, M.B., and D.M. Gatlin III. 2008. Report of the plant products in aquafeed strategic planning workshop: an integrated, interdisciplinar y research roadmap for increasing utilization of plant feedstuffs in diets for carnivorous fish. Reviews in Fisheries Science 16:449 455. Beaudoin, C.P., Tonn, W.M., Prepas, E.E., and L.I. Wassenaar. 1999. Individual specialization and trophic adaptabilit y of northern pike ( Esox lucius ): an isotope and dietary analysis. Oecologia 120:386 396. Ben David, M., Hanley, T.A., Klein, D.R., and D.M. Schell. 1997. Seasonal changes in diets of coastal and riverine mink: the role of spawning Pacific salmon. Canadia n Journal of Zoology 75:803811. Ben David, M., and D.M. Schell. 2001. Mixing models in analyses of diet using multiple stable isotopes: a response. Oecologia 127:180184.

PAGE 683

683 Benstead, J.P., March, J.G., Fry, B., Ewel, K.C., and C.M. Pringle. 2006. Testing isosource: stable isotope analysis of a tropical fishery with diverse organic matter sources. Ecology 87:326333. Boutton, T.W. 1991. Stable carbon isotope ratios of natural materials: II. atmospheric, terrestrial, marine and freshwater environments. Pages 173186 in D.C. Coleman and B. Fry editors. Carbon isotope techniques. Academic Press. San Diego, California, U.S.A. Boyd, C.E. 1974. Lime requirements of Alabama fish ponds. Bulletin 459. Agricultural Experiment Station, Auburn University, Auburn, Alabama, U.S.A Boyd, C.E. 1979. Water quality in warmwater fish ponds. Auburn University, Auburn, Alabama, U.S.A. Boyd, C.E. 1995. Bottom soils, sediment and pond aquaculture. Chapman and Hall, New York, N.Y. U.S.A. Boyd, C.E. 1997. Practical aspects of chemistry in pond aquaculture. The Progressive FishCulturist 59:8593. Boyd, C.E. and J.R. Bowman. 1997. Pond bottom soils. Pages 135162 in H.S. Egna and C.E. Boyd, editors. Dynamics of pond aquaculture. CRC Press L.L.C., New York, N.Y., U.S.A. Brett, M.T., Wiackowski, K., Lubnow, F.S, Mueller Solger, A, Elser, J.J., and C.R. Goldman. 1994. Species dependent effects of zooplankton on planktonic ecosystem processes in Castle Lake, California. Ecology 75:22432254. Briand, F. and J.E. Cohen. 1987. Environmental correlates of food chain length. Science 238:956960. Brooks, J.L. 1968. The effects of prey size selection by lake planktivores. Systematic Zoology 17:273291. Brown, P.B., Twibell, R., Jonker, Y., and K.A. Wilson. 1997. Evaluation of thr ee soybean products in diets fed to juvenile hybrid striped bass Morone saxitilis M. chrysops. Journal of the World Aquaculture Society 28: 215233. Brown, C.D., Hoyer, M.V., Bachmann, R.W., and D.E. Canfield, Jr. 2000. Nutrient chlorophyll relationship s: and evaluation of empirical nutrient chlorophyll models using Florida and northtemperate lake data. Canadian Journal of Fisheries and Aquatic Sciences 57:15741583. Burns, C.W. and M. Schallenberg. 2001a. Calanoid copepods versus cladocerans: consumer effects on protozoa in lakes of different trophic status. Limnology and Oceanography 46:15581565.

PAGE 684

684 Burns, C.W. and M. Schallenberg. 2001b. Short term impacts of nutrients, Daphnia, and copepods on microbial foodwebs of an oligotrophic and eutrophic lake New Zealand Journal of Marine and Freshwater Research 35:695 710. Cabana, G., and J.B. Rasmussen. 1994. Modeling food chain structure and contaminant bioaccumulation using stable nitrogen isotopes. Nature 372:255257. Cailliet and Barry 1978. Comparison of food array overlap measures useful in fish feeding habit analysis Pages 67 79 in S.J. Lipovsky and C.A. Simenstad, editors. Fish food habit studies, proceedings of the second Pacific Northwest technical workshop University of Washington, Was hington Sea Grant Program, Seattle, Washington, U.S.A. Cailliet, G.M., M.S. Love, and A.E Ebeling. 1986. Fishes: a f ield and l aboratory m anual on t heir s tructure, i dentification and natural h istory. Wadsworth Publishing Company, Belmont, CA U.S.A. Campton, D.E., and G.A.E. Gall. 1988. Effect of individual and group rearing on age and size at maturity of male mosquitofish, Gambusia affinis Journal of Fish Biology 33:203212. Carpenter, S.R., Kitchell, J.F., and J.R. Hodgson. 1985. Cascading trophic interactions and lake productivity. BioScience 35:634 639. Carpenter, S.R., Kitchell, J.F., Hodgson, J.R., Cochran, P.A., Elser, J.J., Elser, M.M., Lodge, D.M., Kretchmer, D., and X. He. 1987. Regulation of lake primary productivity by food web structure Ecology 68:18631876. Cho, C.Y. and D.P. Bureau. 1997. Reduction of waste output from salmonid aquaculture through feeds and feeding. The Progressive FishCulturist 59:155160. Chong, A.S.C., Ishak, S.D., Osman, Z. and R. Hashim. 2004. Effect of dietar y protein level on the reproductive performance of female swordtails Xiphophorus helleri (Poeciliidae). Aquaculture 234:381392. Cifuentes, L.A., Sharp, J.H., and M.L. Fogel. 1988. Stable carbon and nitrogen isotope biogeochemistry in the Delaware estuary Limnology and Oceanography 33:11021115. Cohen, J.E., Pimm, S.L., Yodzis, P., and J. Saldana. 1993. Body sizes of animal predators and animal prey in food webs. Journal of Animal Ecology 62:6778. Cohen, J.E., Jonsson, T., and S.R. Carpenter. 2002. Ecol ogical community description using the food web, species abundance and body size. Proceedings of the National Academy of Sciences 100:17811786. Cole, M.L., Valiela, I., Kroeger, K.D., Tomasky, G.L., Cebrian, J., Wigand, C., McKinney, R.A., Grady, S.P., and M.H.C. da Silva. 2004. Assessment of a d15N isotopic method

PAGE 685

685 to indicate anthropogenic eutrophication in aquatic ecosystems. Journal of Environmental Quality 33:124132. Coman, F.E., Connolly, R.M., and N.P. Preston. 2003. Zooplankton and epibenthic fauna in shrimp ponds: factors influencing assemblage dynamics. Aquaculture Research 34:35371. Corliss, J.B., Dymond, J., Gordon, L.I., Edmond, J.M., Von Herzen, R.P., Green, K., Williams, D., Bainbridge, A., Crane, K., and T.H. Van Andel. 1979. Submarine t hermal springs on the Galapagos Rift. Science 203:10731083. Cortes, E. 1997. A critical review of methods of studying fish feeding based on analysis of stomach contents: application to elasmobranch fishes. Canadian Journal of Fisheries and Aquatic Sciences 54:726738. Cowey, C.B. and M.J. Walt on. 1989. Intermediate metabolism. Pages 259329 in J.E. Halver, editor. Fish Nutrition 2 nd edition, Academic Press, New York, N.Y., U.S.A. Dabrowski, K., Arslan, M, Terjesen, B.F., and Y. Zhang. 2007. The efficiency of dietary indispensible amino acid imbalances on feed intake: is there a sensing of deficiency and neural signaling present in fish? Aquaculture 268:136142. Dawes, J. A. 1991. Livebearing f ishes. a guide to t heir a quarium c are, biology and c lassification. Blandforn Press, London, England. Dawson, T.E., S. Mambelli, A.H. Plamboeck, P.H. Templer, and K.P. Tu. 2002. Stable isotopes in plant ecology. Annual Review of Ecological Systems 33:507559. Day, O.J., and H.G. PlascenciaGonz alez. 2000. Soybean protein concentrate as a protein source for turbot Scophthalmus maximus L. Aquaculture Nutrition 6:221 228. De Brabandere, L., Frazer, T.K., and J.P. Montoya. 2007. Stable nitrogen isotope ratios of macrophytes and associated periphyto n along a nitrate gradient in two subtropical, springfed streams. Freshwater Biology 52:15641575. DeNiro and Epstein. 1978. Influence of the diet on the distribution of carbon isotopes in animals. Geochimica Cosmochimica Acta 42:495 506. DeNiro and Eps tein. 1981a. Influence of the diet on the distribution of nitrogen isotopes in animals. Geochimica Cosmochimica Acta 45:341 351. DeVries, D.R., Bremigan, M.T., and R.A. Stein. 1998. Prey selection by larval fishes as influenced by available zooplankton and gape limitation. Transactions of the American Fishereies Society 127:10401050. Diana, J.S. 1979. The feeding pattern and daily ration of a top carnivore, the northern pike ( Esox lucius ). Canadian Journal of Zoology 57:21212127.

PAGE 686

686 Diana, J.S. 1981. An e xperimental analysis of the metabolic rate and food utilization of northern pike. Comparitive Biochemistry and Physiology 71A:395399. Diana, J.S., Lin, C.K., and P.J. Schneeberger. 1991. Relationships among nutrient inputs, water nutrient concentration s, primary production, and yield of Oreochromis niloticus in ponds. Aquaculture 92:323241. Dionne, M. 1985. Cannibalism, food availability, and reproduction in the mosquito fish ( GAMBUSIA AFFINIS ): a laboratory experiment. American Naturalist 126:16 23. Dussault, G.V., and D.L. Kramer. 1981. Food and feeding behavior of the guppy, Poecilia reticulata (Pisces: Poeciliidae). Canadian Journal of Zoology 59:684701. Eggers, D.M. 1979. Comments on some recent methods for estimating food consumption by fish. Journal of the Fisheries Research Board of Canada 36:10181019. Egna, H.S., Boyd, C.E., and D.A. Burke. 1997. Introduction: dynamics of pond aquaculture. Pages 1 18 in H.S. Egna and C.E. Boyd, editors. Dynamics of pond aquaculture. CRC Press LLC, Boca Raton, Florida, U.S.A. Elliott, J.M. and L. Persson. 1978. The estimation of daily rates of food consumption for fish. Journal of Animal Ecology 47:977991. Estep, M.L.F., and S. Vigg. 1985. Stable carbon and nitrogen isotope tracers of trophic dynamics in natural populations and fisheries of the Lahontan lake system, Nevada. Canadian Journal of Fisheries and Aquatic Sciences 42:17121719. Evans, R.D., Bloom, A.J., Sukrapanna, S.S., and J.R. Ehleringer. 1996. Nitrogen isotope composition of tomato ( Lycopersicon esculentum Mill. cv. T 5) grown under ammonium or nitrate nutrition. Plant, Cell and Environment 19:13171323. Ezhil, J., Jeyanthi, C., and M. Narayanan. 2008. Marigold as a carotenoid source on pigmentation and growth of red swordtail, Xiphophorus helleri. Turkish Journal of Fisheries and Aquatic Sciences. 8:99 102. FAO. 1994. Handbook on s mall s cale f reshwater f ish f arming (publication number 24). F ood and A griculture O rganization of the United Nations, Rome, Italy. FDACS (Florida Department of A griculture and Consumer Services). 2000. Aquaculture Best Management Practices. Rule 5L 3, FDACS Division of Aquaculture. 77p. FASS/FDACS. 2001, 2002. Florida Agriculture Statistics Service/ Florida Department of Agriculture and Consumer Services online article: http://www.nass.usda.gov/Statistics_by_State/Florida/Publications/Aquaculture/02aqua4. pdf.

PAGE 687

687 FASS/FDACS. 2005. Florida Agriculture Statistics Service/Florida Department of Agriculture and Consumer Services online article: http://www.nass.usda .gov/Statistics_by_State/Florida/Publications/Aquaculture/06aqua4. pdf Finlay, J.C. and C. Kendall. 2007. Stable isotope tracing of temporal and spatial variability in organic matter sources to freshwater ecosystems. Pages 283 333 in R. Michener and K. Laj tha, editors. Stable isotopes in ecology and environmental science. Blackwell Publishing Ltd., Malden, Massachussetts, U.S.A. Florida Lakewatch. 2001. A beginners guide to water management water clarity. 3rd ed. Circular 103. Institute of Food and Agr icultural Sciences, University of Florida; Gainesville, Florida. Fogel., M.L., Cifuentes, L.A., Velinsky, D.J., and J.H. Sharp. 1992. Relationship of carbon availability in estuarine phytoplankton to isotopic composition. Marine Ecology Progress Series V8 2:291300. Fogg, G.E. and B. Thake. 1987. Algal cultures and phytoplankton ecology. 3rd edition. University of Wisconsin Press publisher, Madison, Wisconsin, U.S.A Fowler, L.G. 1980. Substitution of soybean and cottonseed products for fish meal in diets fed to Chinook and Coho salmon. The Progressive FishCulturist 42:87 91. Franck, D., Hannes, R.P., Lanffermann, H., and A. Ribowski. 1985. Effects of social isolation on aggressivenes s in fish with special reference to the swordtail (Xiphophorus helleri). Behavioural Processes 10:415 427. Frazer, T.K. 1315N) composition of Antarctic krill, Euphausia superba, and two of its potential food sources in winter Journal of Plankton Research 18:14131426. Fry, B., and P.L. Parker. 1979. Animal diet in Texas seagrass meadows: 13C evidence for the evidence for the importance of benthic plants. Estuarine and Coastal Marine Science 8:499509. Fry, B. and E.B. Sher r. 1984. d13C measurements as indicators of carbon flow in marine and freshwater ecosystems. Contributions in Marine Science 27:13 47. Fry, B. 1991. Stable isotope diagrams of freshwater food webs. Ecology 72:22932297. Fry, B. 1999. Using stable isotope s to monitor watershed influences on aquatic trophodynamics. Canadian Journal of Fisheries and Aquatic Sciences 56:21672171.

PAGE 688

688 Fry, B., Mumford, P.L., Tam, F., Fox, D.D., Warren, G.L., Havens, K.E., and A.D. Steinman. 1999. Trophic position and individual feeding histories of fish from Lake Okeechobee, Florida. Canadian Journal of Fisheries and Aquatic Sciences 56:590600. Fry, B. 2006. Stable isotope ecology. Springer Science and Business Media LLC New York, N.Y., U.S.A. Gilbert, J.J. 1988. Suppression of rotifer populations by Daphnia: a review of the evidence, the mechanisms, and the effects on zooplankton community structure. Limnology and Oceanography 33:12861303. Gilbert, J.J. 1989. The effect of Daphnia interference on a natural rotifer and cilia te community: short term bottle experiments. Limnology and Oceanography 34:606617. GraphPad Software Inc. 2005. GraphPad Prism 4.0 statistical software package. LaJolla, California, U.S.A. Grassle, J.F. 1985. Hydrothermal vent animals: distribution and biology. Science 229:713717. Evans, H.E. 1992. Anatomy of tropical fishes. Pages 7193 in J.A. Gratzek and J.R. Mathews, editors. Aquariology: a natomy, genetics, and breeding. Tetra Press, Morris Plains, New Jersey, U.S.A. Gregory Eaves, I., Demers, M.J., Kempe, L., Krummel, E.M., MacDonald, R.W., Finney, B.P., and J.M. Blais. 2007. Tracing salmonderived nutrients and contaminants in freshwater food webs across a pronounced spawner density gradient. Environmental Toxicology and Chemistry 26:11001108. Grey, J. Waldron, S. and R. Hutchinson. 2004. The utility of carbon and nitrogen isotope analyses to trace contributions from fish farms to the receiving communities of freshwater lakes: a pilot study in Esthwaite Water, UK. Hydrobiologia 524:253262. Gu, B., Schelske, C.L., and M.V. Hoyer. 1996. Stable isotopes of carbon and nitrogen as indicators of diet and trophic structure of the fish community in a shallow hypereutrophic lake. Journal of Fish Biology 49:12331243. Gu, B., Schell, D.M., Huang, X., and F. Yie. 1996. Stable isotope evidence for dietary overlap between two planktivorous fishes in aquaculture ponds. Canadian Journal of Fisheries and Aquatic Sciences 53:28142818. Gu, B., D.M. Schell, T. Frazer, M. Hoyer, and F.A. Chapman. 2001. Stabl e carbon isotope evidence for reduced feeding of Gulf of Mexico sturgeon during their prolonged river residence period. Estuarine, Coastal and Shelf Science 53:275 280.

PAGE 689

689 Hall, D.J. 1964. An experimental approach to the dynamics of a natural population of D aphnia Galeata Mendotae Ecology 45:91112. Hall, E.R. 2004. Transformation and f ate of dissolved organic m atter o riginating in the Suwanee River w atershed: a s table i sotope a pproach. Ph.D. dissertation, Department of Fisheries and Aquatic Sciences, Unive rsity of Florida, Gainesville, FL U.S.A. 160 pp. Halver, J.E. 1989. Fish Nutrition 2nd edition. Academic Press Inc., New York, N.Y. U.S.A. Haines, E.B. and C.L. Montague. 1979. Food sources of estuarine invertebrates analyzed using 13C/12C ratios. Ecolo gy 60:4856. Hannes, R.P. and D. Franck. 1983. The effect of social isolation on androgen and corticosteroid levels in a cichlid fish ( Haplochromis burtonis ) and in swordtails ( Xiphophorus helleri ). Hormones and Behavior 17:292301. Hannes, R.P., Franck, D., and F. Liemann. 1984. Effects of rankorder fights on whole body and blood concentrations of androgens and corticosteroids in the male swordtail ( Xiphophorus helleri ). Zeitschrift fr Tierpsychologie 65:53 65. Hansson, L.A. and L.J. Tranvik 1997. Algal species composition and phosphorus recycling at contrasting grazing pressure: an experimental study in subAntarctic lakes with two trophic levels. Freshwater Biology 37:4553. Hansson, S., J.E. Hobbie, R. Elmgren, U. Larsson, B. Fry, and S. Johansson. 1997. The stable nitrogen isotope ratio as a marker of foodweb interaction and fish migration. Ecology 78:22492257. Harrigan, P., Zieman, J.C., and S.A. Macko. 1989. The base of nutritional support for the gray snapper ( Lutjanus griseus ): an evaluation based on a combined stomach content and stable isotope analysis. Bulletin of Marine Science 44:6577. Hardy, R.W. 2003. Soybeans: the time is now. Aquaculture Magazine September/October 2003:5862. Havens, K.E. 1994. An experimental comparison of the effects of two chemical stressors on a freshwater zooplankton assemblage. Environmental Pollution 84:245251. Hideshige, T., and E. Wada. 1990. Use of 15N/14N rations to evaluate the food source of the mysid, Neomysis intermedia Czerniawky, in a eutrophic lake in Japan. Hydrobiologia 194:8590. Hoeinghaus, D.J., Winemiller, K.O., and A.A. Agostinho. 2008. Hydrogeomorphology and river impoundment affect foodchain length of diverse neotropical food webs. Oikos 117:984995.

PAGE 690

690 Holbrook, S.J., and R. J. Schmitt. 1992. Causes and consequences of dietary specialization in surfperches: patch choice and intraspecific competition. Ecology 73(2):402412. Holbrook, S.J., and R.J. Schmitt. 1994. Compensation in resource use by foragers released from interspec ific competition. Journal of Experimental Marine Biology and Ecology 185:219233. Hopkins, C.L., and M.J. Unwin. 1997. The effect of restricted springtime feeding on growth and maturation of freshwater reared Chinook salmon, Oncorhynchus tshawytscha (Walb aum). Aquaculture Research 28:545549. Horn, H.S. 1966. Measurement of overlap in comparative ecological studies. The American Naturalist 100:419424. Hoyer, M.V., Gu, B., and C.L. Schelske. 1998. Sources of organic carbon in the food webs of two Florida lakes indicated by stable isotopes. Pages 381392 in E. J eppessen, M. Sondergaard, and K. Christoffersen, ed itors. The s tructuring r ole of s ubmerged m acrophytes in l akes Springer Verlag, New York, N.Y. U.S.A. Hutchings, J.A. 1993. Adaptive life histories effected by age specific survival and growth rate. Ecology 74:673684. Ingram, B.A. 2009. Culture of juvenile Murray cod, trout cod and Macquarie perch (Percichthyidae) in fertilized earthen ponds. Aquaculture 287:98106. James, R. and K. Sampath 1998. Effects of feed and water quality (hardness) on growth in red swordtail Xiphophorus helleri fry. Indian Journal of Fisheries 45:307313. Jannasch, H.W., and M.J. Mottl. 1985. Geomicrobiology of deepsea hydrothermal vents. Science 229:717725. Jobling, M. 1983. Towards an explanation of specific dynamic action (SDA). Journal of Fish Biology 23:549555. Jobling, M. 2001. Feed composition and analysis. Pages 1 24 in D. Houlihan, T. Boujard, and M. Jobling, editors. Food i ntake in f ish. Blackwell Science. Oxford, England Jones, C.L.W., Kaiser, H., Webb, G.A., and T. Hecht. 1998. Filial cannibalism in the swordtail Xiphophorus helleri (Poeciliidae). Aquarium Sciences and C onservation 2:7988. Jones, C.L.W., Kaiser, H., and T. Hecht. 1998. Effect of shelter, broodstock number, and sexration on juvenile production in the swordtail Xiphophorus helleri under intensive culture conditions. Journal of the World Aquaculture Soci ety 29:9296.

PAGE 691

691 Jones, C.L.W., Kaiser, H., and T. Hecht. 2007. Intercohort cannibalism and parturitionassociated behavior of adult swordtail, Xiphophorus hellerii Heckel (pisces: Poeciliidae). Aquaculture Research 38:718727. Jones, J.I. and S. Waldron. 2003. Combined stable isotope and gut contents analysis of food webs in plant dominated, shallow lakes. Freshwater Biology 48:13961407. Keeley, J.E., Sternberg, L.O., and M.J. DeNiro. 1986. The use of stable isotopes in the study of photosynthesis in fres hwater plants. Aquatic Botany 26:213223. Karplus, I., Gottdiener, M., and B. Zion. 2003. Guidance of single guppies (Poecilia reticulate) to allow sorting by computer vision. Aquacultural Engineering 27:177190. Kastner, R.J. and C.E. Boyd. 1996. Produc tion of sunfish Lepomis spp. in ponds treated with controlled release fertilizers. Journal of the World Aquaculture Society 27:228234. Kelly, A.M. and C.C. Kohler. 1997. Clima te, site, and pond design. Pages 109133 in H.S. Egna and C.E. Boyd. Dynamics of pond aquaculture, CRC Press LLC, Boca Raton, Florida, U.S.A. Kline, T.C. Jr., Goering, J.J., Mathisen, O.A., and P.H. Poe. 1993. Recycling of elements transported upstream by runs of Pacific salmon: II. 15N and 13C evidence in the Kvichak river watershed, Bristol Bay, southwestern Alaska. Canadian Journal of Fisheries and Aquatic Sciences 50:2350 2365. Kling, G.W., and B. Fry. 1992. Stable isotopes and planktonic trophic structure in arctic lakes. Ecology 73:561566. KnudHansen, C.F., Batterson, T.R., and C.D. McNabb. 1993. The role of chicken manure in the production of Nile tilapia, Oreochromis niloticus (L.). Aquaculture and Fisheries Management 24:483493. KnudHansen, C.F. 1998. Pond fertilization: ecological approach and practical application. Aquaculture Collaborative Research Support Group, Oregon State University, Corvallis, Oregon, U.S.A Krebs, C.J. 1999. Ecological Methodology, 2nd ed ition AddisonWelsey Educational Men lo Park, California, U.S.A. Kruger, D.P., Britz, P.J. and J. Sales. 2001. The influence of livefeed supplementation on growth and reproductive performance of swordtail ( Xiphophorus helleri Heckel 1848) broodstock. Aquarium Sciences and Conservation 3:265273. Kurten, G., Hall, L., and N. Thompson. 1999. Evaluation of cottonseed meal supplementation of inorganically fertilized Florida largemouth bass spawning ponds. North American Journal of Aquaculture 61:115125.

PAGE 692

692 Landau, M. 1992. Introduction to aquaculture John Wiley and Sons, New York, N.Y., U.S.A. Larson, G.L., McIntire, C.D., Buktenica, M.W., Girdner, S.F., and R.E. Truit t. 2007. Distribution and abundance of zooplankton populations in Crater Lake, Oregon. Developments in Hydrobiology 191:217233. Lemly, A.D. and J.F. Dimmick. 1982. Structure and dynamics of zooplankton communities in the littoral zone of some North Carolina lakes. Hydrobiologia 88:299307. Lim, S.R., S.M. Choi, X.J. Wang, K.W. Kim, I.S. Shin, T.S. Min, and S.C. Bai. 2004. Effec ts of dehulled soybean meal replacer in diets for fingerling and growing Korean rockfish ( Sebastes schlegeli ). Aquaculture 231:457468. Lin, C.W., Teichert Coddington, D.R., Green, B.W., and K.L. Veverica. 1997. Fertilization regimes. Pages 73 107 in H.S. Egna and C.E. Boyd editors. Dynamics of pond aquaculture. CRC Press L.L.C., New York, N.Y., U.S.A. Lindquist, D.C. 1998. The effects of erosion on the trophic ecology of fishes in Elkhorn Slouch, CA. MS Thesis, California State University, Hayward, Cali fornia, U.S.A. Lochmann, R., and H. Phillips. 1996. Stable isotopic evaluation of the relative assimilation of natural and artificial foods by golden shiners Notemigonus crysoleucas in ponds. Journal of the World Aquaculture Society 27:168177. Lochmann, R., H. Phillips, S. Dasgupta, D. Gatlin, S. Rawles. 2001. Stable carbon isotope ratios and standard production data as indices of golden shiner, Notemigonus crysoleucas, performance in pond feeding trials. Journal of Applied Aquaculture 11:21 34. Lovell, T. 1998. Nutrition and f eeding of f ish, 2 nd edition. Van Norstrand Rheinhold, New York, N.Y., U.S.A. Ludwig, G.M., Stone, N.M., and C. Collins. 1998. Fertilization of fish fry ponds. Publication 469, Southern Regional Aquaculture Center, Cooper ative States Research, Education and Extension Service, United States Department of Agriculture. Mallin, M.A. and H.W. Paerl. 1994. Planktonic Trophic Transfer in and Estuary: Seasonal, Diel and Community Structure Effects. Ecology 75:21682184. McCutchan, J.H. Jr., W.M. Lewis Jr., C. Kendall, and C.C. McGrath. 2003. Variation in trophic shift for stable isotope ratios of carbon, nitrogen and sufur. Oikos 102:378390. McKenzie Jr., W.D., Crews, D., Kallman, K.D., Policansky, D., J.J. Sohn. 1983. Age, weight and the genetics of sexual maturation in the platyfish, Xiphophorus maculates Copeia 1983:770774.

PAGE 693

693 Melville, A.J., and R.M. Connolly. 2003. Spatial analysis of stable isotope data to determine primary sources of nutrition for fish. Oecologia 136:499507. Michener, R.H., and L. Kaufman. 2007. Stable isotope ratios as tracers in marine food webs: an update. Pages 238282 in R. Michener and K. Lajtha, editors. Stable isotopes in ecology environmental science, Blackwell Publishing, Malden, Massa chusetts, U.S.A. Mills, E.L., Confer, J.L., and R.C. Ready. 1984. Prey selection by young yellow perch: the influence of capture success, visual acuity, and prey choice. Transactions of the American Fisheries Society 113:579 587. Milton, D.A., and A.H. Arthington. 1983. Reproductive biology of Gambusia affinis hobrooki Baird and Girard, Xiphophorus helleri (Gnther) and X. maculates (Heckel) (Pisces; Poeciliidae) in Queensland, Australia. Journal of Fish Biology 23:2341. Mittelbach, G.C., and C.W. Osen berg. 1993. Stage structured interactions in bluegill: consequences of adult resource variation. Ecology 74:23812394. Mittelbach, G.C., Osenberg, C.W., and P.C. Wainwright. 1999. Variation if feeding morphology between pumpkinseed populations: phenotypic plasticity or evolution? Evolutionary Ecology Research 1:111128. Minagawa, M. and E. Wada. 1984. Stepwise enrichment of 15N along food chains: further evidence and the relation between 15N and animal age. Geochimica et Cosmochimica Acta 48:11351140. Montoya, J.P. and J.J. McCarthy. 1995. Isotopic fractionation during nitrate uptake by phytoplankton grown in continuous culture. Journal of Plankton Research V17:439464. Motulsky, H. 1995. Intuitive biostatistics Oxford University Press New York, N.Y., U.S.A. Montgomery, D.C. 2001. Design and analysis of experiments, 5th edition. John Wiley and Sons, Inc., New York, N.Y., U.S.A. NRC. 1993. Nutritional requirements of fish. National Research Council. National Academy Press. Washington, D.C., U.S.A. OLeary, M.H. 1988. Carbon isotopes in photosynthesis. BioScience 38:328336. Olive, P.J.W., J.K. Pinnegar, N.V.C. Polunin, G. Richards, and R. Welch. 2003. Isotope trophic step fractionation: a dynamic equilibrium model. Journal of Animal Ecology 72:608617.

PAGE 694

694 Osenberg, C.W., G.G. Mittelbach, and P.C. Wainwright. 1992. Twostaged life histories in fish: the interaction between juvenile competition and adult performance. Ecology 73:255267. Ott, L., Ott., R.L., and M.T. Longenecker. 2000. An introduction to statistical methods and data analysis, 5 th edition. Brooks Cole, Belmont, California, U.S.A. Paine, R.T. 1980. Food webs: linkage, interaction strength and community infrastructure. Journal of Animal Ecology 49:667685. Parker, R.E. 1979. Introductory statistics for biology, 2nd edition. Press Syndicate of the University of Cambridge, Cambridge, United Kingdom. Pennak, R.W. 1989. Freshwater invertebrates of North America, third edition. John Wiley and Sons, Toronto, Ontario, Canada. Pennock, J.R., V elinsky, D.J., Ludlam, J.M., and J.H. Sharp. 1996. Isotopic fractionation of ammonium and nitrate during uptake by Skeletonema costatum : implications for 15N dynamics under bloom conditions. Limnology and Oceanography 41:451459. Perschbacher, P.W., Ludwig, G.M., and N. Slaton. 2002. Effects of common aerially applied rice herbicides on the plankton communities of aquaculture ponds. Aquaculture 214:241246. Perschbacher, P.W. and G.M. Ludwig. 2004. Effects of diuron and other aerially applied cotton herbicides and defoliants on the plankton communities of aquaculture ponds. Aquaculture 233:197203. Peterson, B.J., B. Fry. 1987. Stable isotopes in ecosystem studies. Annual Review of Ecology and Systematics 18:293320. Phillips, D.L. 2001. Mixing models in analyses of diet using multiple stable isotopes: a critique. Oecolgia 127:166170. Phillips, D.L. and J.W. Gregg. 2001. Uncertainty in source partitioning using stable isotopes. Oecologia 127:171179. Phillips, D.L. and P.L. Koch. 2002. Incorporating concentration dependence in stable isotope mixin g models. Oecologia 130:114 125. Phillips, D.L. and J.W. Gregg. 2003. Source partitioning using stable isotopes: coping with too many sources. Oecologia 136:261269. Pianka, E.R. 1970. On r and K selection. American Naturalist 104:592596.

PAGE 695

695 Pinkas, L., Oliphant, M.S., and I.L.K. Iverson. 1971. Food habits of albacore, bluefin tuna, and bonito in California waters. California Fish and Game Fish Bulletin 152. Post, J.R., and D.J. McQueen 1987. The impact of planktivorous fish on the structure of a plankton community. Freshwater Biology V17:7989. Post, D.M. 2002a Using stable isotopes to estimate trophic position: models, methods, and assumptions. Ecology 83:703718. Post, D.M. 2002b. The long and short of foodchain length. Trends in Ecology and Evolution V17:269277. Refstie, S., Storebakken, T., and A.J. Roem. 1998. Feed consumption and conversion in Atlantic salmon ( Salmo salar ) fed diets with fish meal, extracted soybean meal o r soybean meal with reduced content of oligosaccharides, trypsin inhibitors, lectins and soya antigens. Aquaculture 162:301312. Regost, C., Arzel, J., and S.J. Kaushik. 1999. Partial or total replacement of fish meal by corn gluten meal in diet for turbo t (Psetta maxima ). Aquaculture 180:99117. Ribowski, A., and D. Franck. 1993. Subordinate swordtail males escalate faster than dominants: a failure of the social conditioning principle. Aggressive Behavior 19: 223229. Riley, J. and P. Edwards. 1998. Statistical aspects of aquaculture research: pond variability and pseudoreplication. Aquaculture Research 29:281288. Rodelli, M.R., Gearing, J.N., Gearing, P.J., Marshall, N., and A. Sasekumar. 1984. Stable isotope ratios as a tracer of mangrove carbon in Malaysian ecosystems. Oecologia 61:326333. Rounick, J.S., and M.J. Winterbourn. 1986 Stable carbon isotopes and carbon flow in ecosystems. Bioscience 36:171 177. Royes, J.A. and F.A. Chapman. 2003. Preparing your own feeds. University of Florida Extension Circular #97, University of Florida, Gainesville, F lorida, U.S.A. Saito, L., Johnson, B.M., Bartholow, J., and R.B. Hanna. 2000. Assessing Ecosystem Effects of Reservoir Operations Using Food Web Energy Transfer and Water Quality Models. Ecosystems 2001:105125. Sawyer, C.N., McCarty, P.L., and G.F. Parkin. 2003. Chemistry for environmental engineering and science, 5 th edition. McGraw Hill, New York, N.Y., U.S.A.

PAGE 696

696 Schmitt, R.J. and S.J. Holb rook. 1984a. Ontogeny of prey selection by black surfperch Embiotoca jacksoni (Pisces: Embiotocidae): the roles of fish morphology, foraging behavior, and patch selection. Marine Ecology Progress Series 18:225239. Schmitt, R.J. and S.J. Holbrook. 1984b. Gape limitation, foraging tactics and prey size selectivity of two microcarnivorous species of fish. Oecologica 63:612. Schoeninger, M.J. and M.J. DeNiro. 1984. Nitrogen and carbon isotopic composition of bone collogen from marine and terrestrial animals Geochimica et Cosmochimica Acta 48:625 639. Schroeder, G.L. 1978. Autotrophic and heterotrophic production of microorganisms in intensely manured fish ponds, and related fish yields. Aquaculture 14:303325. Schroeder, G.L. 1983. Stable isotope ratios as naturally occurring tracer in the aquaculture food web. Aquaculture 30:203210. Schroeder, G.L. 1987. Carbon and nitrogen budgets in manured fish ponds on Israels coastal plain. Aquaculture 62:259279. Schwarcz, H.P. 1991. Some theoretical aspects of isotope paleodiet studies. Journal of Archaeological Science 18:261 275. Silver, M.W. 1975. The habitat of Sal pa fusiformis in the California Current as defined by indicator assemblages. Limnology and Oceanography 20:230:237. Smart, T.S., Riley, J., an d P. Edwards. 1998. Statistical aspects of aquaculture research: sample sizes for pond experiments. Aquaculture Research 29:373379. Smart, T.S., J. Riley, and D.C. Little. 2001. Statistical aspects of aquaculture research: optimum block size in pond expe riments. Aquaculture Research 32:6576. Smith, F.A., and N.A. Walker. 1980. Photosynthesis by aquatic plants: effects of unstirred layers in relation to assimilation of CO2 and HCO3 and to carbon isotopic discrimination. New Phytologist 86:245 259. Sohn, J.J. 1977. Socially induced inhibition of genetically determined maturation in the platyfish, Xiphophorus maculates Science 195:199201. Sokal, R.R., and F.J. Rohl. 1981. Biometry, 2nd edition. W.H. Freeman and Company, New York, N.Y., U.S.A. Takahashi, K., Yoshioka, T., Wada, E., and M. Sakamoto. 1990. Temporal variations in carbon isotope ratio of phytoplankton in a eutrophic lake. Journal of Plankton Research 12:799808.

PAGE 697

697 Tranvik, L.J. and L.A. Hansson. 1997. Predator regulation of aquatic m icrobial abundance in simple food w e bs of subAntarctic lakes. Oikos 79:347356. Tamaru, C.S., B. Cole, R. Bailey, C. Brown and H. Ako. 2001. A manual for the commercial production of the swordtail, Xiphophorus helleri CTSA Publication #128, University of Hawaii Sea Grant Extension Service School of Ocean Earth Science and Technology, Honolulu, Hawaii U.S.A. Tandler, A., and F.W.H. Beamish. 1981. Apparent specific dynamic action (SDA), fish weight and level of caloric intake in largemouth bass, ( MICRO PTERUS SALMOIDES (Lacepede)). Aquaculture 23:231242. Teichert Coddington, D.R., Popma, T.J., and L.L Lovshin. 1997. Attributes of tropical pond cultured fish. Pages 183198 in H.S. Egna and C.E. Boyd editors. Dynamics of pond aquaculture. CRC Press L.L.C ., New York, N.Y., U.S.A. Tepe Y. and C.E. Boyd. 2003. A reassessment of nitrogen fertilization for sunfish ponds. Journal of the World Aquaculture Society 34:505 511. Tyler, A.C., McGlathery, K.J., and I.C. Andersen. 2003. Benthic algae control sediment water column fluxes of organic and inorganic nitrogen compounds in a temperate lagoon. Limnology and Oceanography 48:21252137. Valiela, I. 1995. Marine ecological processes, 2nd edition. Springer Verlag, New York, N.Y., U.S.A. Vander Zanden, M.J., Cab ana, G., and J.B. Rasmussen. 1997. Comparing trophic position of freshwater fish calculated using stable nitrogen isotope ratios ( 15N) and literate dietary data. Canadian Journal of Fisheries and Aquatic Sciences 54:11421158. Vander Zanden, M.J., and J.B. Rasmussen. 1999. Primary consumer 13C and 15N and the trophic position of aquatic consumers. Ecology 80:13951404. Vander Zanden, M.J., Shuter, B.J., Lester, N., and J.B. Rasmussen. 1999. Patterns of food chain length in lakes: a stable isotope stu dy. American Naturalist 154:406416. Vander Zanden, M.J., and J.B. Rasmussen. 1513C trophic fractionation: Implications for aquatic food web studies. Limnology and Oceanography 46:20612066. Vander Zanden, M.J., and Y. Vadeboncoeur. 2002. Fishes as i ntegrators of benthic and pelagic f ood w ebs in l akes. Ecology 83:21522161. Velinsky, D.J. and M.L. Fogel. 1999. Cycling of dissolved and particulate nitrogen and carbon in the Framvaren Fjord, Norway: stable isotope variations. Marine Chemistry 67:161 180.

PAGE 698

698 Wada, E. 1980. Nitrogen isotope fractionation and its significance in biogeochemical processes occurring in marine environments. Pages 375398 in E.D. Goldberg, Y. Horibe, and K. Saruhashi editors. Isotope m arine c hemistry. Uchida Rokakuho, Tokyo, Japan. Wada, E., and A. Hattori. 1978. Nitrogen isotope effects in the assimilation of inorganic nitrogenous compounds by marine diatoms. Geomicrobiology Journal 1:85101. Wainwright, P.C., and B.A. Richard. 1995. Predicting patterns of prey use from morphology of fishes. Environmental Biology of Fishes 44:97 113. Warren, P.H., and M. Spencer. 1996. Community a nd foodweb responses to the manipulation of energy input and disturbance in small ponds. Oikos 75:407418. Waser, N.A., Yin, K., Yu, Z., Tada, K., Harrison, P.J., Turpin, D.H., and S.E. Calvert. 1998. Nitrogen isotope fractionation during nitrate, ammoni um and urea uptake by marine diatoms and coccolithophores under various conditions of N availability. Marine Ecology Progress Series 169:2941. Waser, N.A., Yu, Z., Yin, K., Nielsen, B., Harrison, P.J., Turpin, D.H., and S.E. Calvert. 1999. Nitrogen isotopic fractionation during a simulated diatom spring bloom: importance of N starvation in controlling fractionation. Marine Ecology Progress Series 179:291296. Werner, E.E., and D.J. Hall. 1979. Foraging efficiency and habitat switching in competing sunfis hes. Ecology 60:256264. Winberg. G.G. 1956. Rate of metabolism and food requirements of fishes. Byelorussian University, Minsk. Translated from Russian, 1960. Fisheries Research Board of Canada Translation Series 194. Yamamoto, T., Shima, T., Furuita, H ., Suzuki, N., Sanchez Vazquez, F.J., and M. Tabata. 2001. Self selection and feed consumption of diets with a complete amino acid composition deficient in either methionine or lysine by rainbow trout, Oncorhynchus mykiss (Walbaum). Aquaculture Research 32 (s1):8391. Yamamoto, T ., Shima, T., Furuita, H., Sugita, T., and N. Suzuki 2007. Effects of feeding time, water temperature, feeding frequency and dietary composition on apparent nutrient digestibility in rainbow trout Oncorhynchus mykiss and common carp Cyprinus carpio Fisheries Science 73:161 170. Yoshii, K., Melnik, N.G., Timoshkin, O.A., Bondarenko, N.A., Anoshko, P.N., Yoshioka, T., and E. Wada. 1999. Stable isotope analyses of the pelagic food web in Lake Baikal. Limnology and Oceanography 44:502511.

PAGE 699

699 Yusoff, F.M., and C.D. McNabb. 1989. Effects of nutrient availability on primary production and fish production in fertilized tropical ponds. Aquaculture 78:303319. Zar, J.H. 1984. Biostatistical Analysis, 2nd edition. Prentice Hall, Englewood Cliffs, New Jersey, U.S.A. Zaret, T.M. and A.S. Rand. 1971. Competition in tropical stream fishes: support for the competitive ex c lusion principle. Ecology 52:336342. Zohary, T., Erez, J, Gophen, M., Berman Frank, I., and M. Stiller. 1994. S easonality of stable carbon isotopes within the pelagic food web of Lake Kinneret. Limnology and Oceanography 39:10301043.

PAGE 700

700 BIOGRAPHICAL SKETCH Jon S. Kao was born in Waterloo, Iowa. The youngest of three children, he spent his childhood in C edar Falls, Iowa, and summers with his mother in Milwaukee, Wisconsin Jon graduat ed from Cedar Falls High School in 1983 and then spent two months with his mother, sister and brother, traveling through China before heading off to college He attended th e University of Iowa, and in 1990 earned his B .A. majoring in biology and botany. Pursuing his childhood dream to be a marine biolog ist Jon was faced with a major dilemma a s the state of Iowa was surprisingly devoid of opportunities to study marine biology. H e soon packed up most of his worldly belongings and moved to California to attend graduate school studying marine biology. As a budding ichthyologist, he exhaustively Upon completi on of his Ph.D. program, Jon, his veterinarian significant other Shari, and their large brood of dogs and cats, will again make a continent wide trek back to Americas west coast to start a commercial fish and livestock farm of modest, but hopefully comfortable means. studied the diet and feeding physiology of the leopard shark ( Triakis semifasciata ) at Moss Landing Marine Laboratories, and earned his M.S. in Marine Science in 2000. He then again gathered all of his worldly possessions to drive 2000+ miles to study aquaculture as a Ph.D. student at the University of Florida, at the former Depar tment of Fisheries and Aquatic Sciences ( currently the program in Fisheries and Aquatic Sciences in the School of Forest Resources and Conservation).