Citation
Spectral Distribution of Light in Florida Spring Ecosystems

Material Information

Title:
Spectral Distribution of Light in Florida Spring Ecosystems Factors Affecting the Quantity and Quality of Light Available for Primary Producers
Creator:
Szafraniec, Mary Lucy
Place of Publication:
[Gainesville, Fla.]
Florida
Publisher:
University of Florida
Publication Date:
Language:
english
Physical Description:
1 online resource (269 p.)

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Environmental Engineering Sciences
Committee Chair:
DELFINO,JOSEPH J
Committee Co-Chair:
BROWN,MARK T
Committee Members:
FRAZER,TOM K
COHEN,MATTHEW J
KNIGHT,ROBERT LEE
Graduation Date:
5/3/2014

Subjects

Subjects / Keywords:
Absorptivity ( jstor )
Attenuation coefficients ( jstor )
Biomass ( jstor )
Light water ( jstor )
Modeling ( jstor )
Rainbows ( jstor )
River water ( jstor )
Rivers ( jstor )
Visible spectrum ( jstor )
Water quality ( jstor )
Environmental Engineering Sciences -- Dissertations, Academic -- UF
attenuation -- light
Rainbow River ( local )
Genre:
bibliography ( marcgt )
theses ( marcgt )
government publication (state, provincial, terriorial, dependent) ( marcgt )
born-digital ( sobekcm )
Electronic Thesis or Dissertation
Environmental Engineering Sciences thesis, Ph.D.

Notes

Abstract:
Light availability is a major forcing factor for spring ecosystem productivity and sustainability. It appears that water clarity has decreased in many springs in Florida, which could have had an effect on the quantity and quality of light available for primary producers in springs and spring-runs. Factors controlling the loss of water clarity and light availability are poorly defined in spring systems. To understand the causes of increased light attenuation in these optically complex lotic systems, it is necessary to measure the quantity and the quality of light available to primary producers. Optical water quality determines the underwater light field that provides the basis for habitat suitability, in terms of vegetation abundance, distribution, and survival in springs. Studies were conducted to assess the spectral distribution and potential limitation of wavelength-specific photosynthetically active radiation (PAR) by measuring the percent blue, green and red light available to primary producer communities in springs. Optical properties of the underwater light field were assessed to determine the relative magnitude and contribution of key inherent water clarity driving components. Results revealed a spatial optical gradient in both systems studied. Light attenuation of the blue band was dominated by algal pigment and colored dissolved organic matter absorption, and the green and red light bands were most strongly attenuated by scattering and absorption by particulates. Resulting underwater spectral light field characteristics for each spring system were used to develop and calibrate spectrally explicit, site specific and general spring system, empirical optical models to predict historical water clarity conditions. Optical modeling results indicate that both spring systems studied have not significantly changed since the early 2000s, aside from a few trends where particulate scattering and absorption properties have increased in one system and decreased in the other. The lower portions of each river studied had low levels of blue light available in the benthic zone throughout the time series. Finally, linkages were established among light availability, benthic community composition, distribution, and ecosystem primary productivity. The results from a synoptic field survey of spectral light availability and submerged aquatic vegetation biomass identified a minimum blue light requirement threshold range of 38-45%, below which may not be sufficient to support the growth and production of Sagittaria kurziana, an important primary producer in most spring systems. ( en )
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.
Thesis:
Thesis (Ph.D.)--University of Florida, 2014.
Local:
Adviser: DELFINO,JOSEPH J.
Local:
Co-adviser: BROWN,MARK T.
Electronic Access:
RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2015-05-31
Statement of Responsibility:
by Mary Lucy Szafraniec.

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Embargo Date:
5/31/2015
Classification:
LD1780 2014 ( lcc )

Downloads

This item has the following downloads:


Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID ELNF763S7_WP4PI1 INGEST_TIME 2014-10-03T22:29:12Z PACKAGE UFE0046619_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES



PAGE 1

SPECTRAL DISTRIBUTION OF LIGHT IN FLORIDA SPRING ECOSYSTEMS: FACTORS AFFECTING THE QUANTITY AND QUALITY OF LIGHT AVAILABLE FOR PRIMARY PRODUCERS By MARY LUCY SZAFRANIEC A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF T HE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2014

PAGE 2

2014 Mary Lucy Szafraniec

PAGE 3

To m y parents Alicja and Josef for their lo ving support, and to Chelle, who has been my rock and supported me unconditionally

PAGE 4

4 ACKNOWLEDGMENTS I thank my advisor Dr. Joseph Delfino for his continuous support and encouragement and for allowing me to be his last doctoral student at UF. I also than k my committee members Drs. Mark Brown, Matt Cohen, Bob Knight, and Tom Frazer for their insight and guidance I am very appreciative for the technical guidance provided by Drs. Kellie Dixon and Chris Anastasiou. My research would not have been possible wi thout support from the S outhw est F lorida W ater M anagement D istrict I would like to thank Jennette Seachrist, Eric DeHaven, and Veronica Craw for their support of my research and Jaime Swindasz, Joel Durkee, Jason Hust, Tim Crosby, Matt Jablonski, Laura Ca sino and Ying Wu for field and laboratory assistance I would also like to thank staff from the Florida Department Environmental Protection that provided access to Rainbow and Weeki Wachee Rivers, loaned equipment, and assisted with field work and laborat ory analysis I especially thank Jeff Sowards for his continual desire to assist with field work on the Rainbow River, Judy Ashton for loan of my research vessel Toby Brewer for providing access to Weeki Wachee River, Peggy Morgan for assistance in the fi eld and laboratory, and Joy Jackson for laboratory analysis. I also thank r esearchers at the U niversity of S outh F lorida : Dr. Paula Coble, Ana Arrellano and Jennifer Can n izzaro for use of laboratory equipment and technical expertise. Lastly, I am deeply grateful for the unwavering support throughout the whole process from my friends and family that encouraged me to not give up before I achieve d my goals.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 7 LIST OF FIGURES ................................ ................................ ................................ ........ 10 LIST OF ABBREVIATIONS ................................ ................................ ........................... 17 ABSTRACT ................................ ................................ ................................ ................... 19 1 INTRODUCTION ................................ ................................ ................................ .... 21 2 CAUSES OF LIGHT ATTE NUATION IN FLORIDA S PRING ECOSYSTEMS ........ 30 Introduction ................................ ................................ ................................ ............. 30 Methods ................................ ................................ ................................ .................. 36 Study Area ................................ ................................ ................................ ........ 36 Optical Properties and Water Chemistry ................................ .......................... 37 Statistical Data Analysis ................................ ................................ ................... 42 Results ................................ ................................ ................................ .................... 43 Physical Characteristics ................................ ................................ ................... 43 Optical Properties and Water Chemistry ................................ .......................... 44 Discussion ................................ ................................ ................................ .............. 54 3 DEVELOPMENT AND APPL ICATION OF EMPIRICAL LY DERIVED S PECTRALLY EXPLICIT A ND BROADBAND OPTICAL MODELS OF LIGHT ATTENUATION IN FLORI DA SPRINGS ................................ ................................ 91 Introduction ................................ ................................ ................................ ............. 91 Methods ................................ ................................ ................................ .................. 97 Study Area ................................ ................................ ................................ ........ 97 Model Development ................................ ................................ ......................... 98 Spectral Model Calibration ................................ ................................ ............... 99 Model Application ................................ ................................ ........................... 100 Results ................................ ................................ ................................ .................. 101 Model Calibration and Development ................................ .............................. 101 Model Evaluation ................................ ................................ ............................ 103 Model Application: Hind casting Historical Optical Properties and Water Clarity Regime ................................ ................................ ............................. 109 Discussion ................................ ................................ ................................ ............ 113 Model Applicability ................................ ................................ .......................... 113 Modeling Historical Optical Regimes ................................ .............................. 117 Summary ................................ ................................ ................................ ........ 1 21

PAGE 6

6 4 EFFECTS OF THE UNDER WATER SPECTRAL LIGHT ENVIRONMENT ON SUBMERGED AQUATIC VE GETATION BIOMASS, PR IMARY PRODUCTIVITY AND SPA TIAL DISTRIBUTION AL ONG AN OPTICAL WATE R QUALITY GRADIENT ................................ ................................ ............. 168 Introduction ................................ ................................ ................................ ........... 168 Methods ................................ ................................ ................................ ................ 174 Study Area ................................ ................................ ................................ ...... 174 Water Column Optical Properties and Physico chemical Characteristics ....... 175 Benthic Sampling ................................ ................................ ........................... 177 Light Absorption Properties of Primary Producers ................................ .......... 178 Ecosystem Metabolism ................................ ................................ ................... 181 Statistical Data Analysis ................................ ................................ ................. 184 Results ................................ ................................ ................................ .................. 185 Water Column Apparent Optical Properties and Physico chemical Characteristics ................................ ................................ ............................ 185 Species Composition and Distribution ................................ ............................ 187 Species Morphology and Physiology ................................ .............................. 189 Ecosystem Metabolism ................................ ................................ ................... 191 Abiotic and Biotic Relations ................................ ................................ ............ 195 Discussion ................................ ................................ ................................ ............ 197 5 SUMMARY, CONCLUSIONS AND RECOMMENDATIONS ................................ 252 Summary ................................ ................................ ................................ .............. 252 Conclusions ................................ ................................ ................................ .......... 253 Recommendations ................................ ................................ ................................ 255 LIST OF REFERENCES ................................ ................................ ............................. 258 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 269

PAGE 7

7 LIST OF TABLES Table page 2 1 Summary of mean physical characteristics recorded during sampling events at Rainbow (RR) and Weeki Wachee Rivers (WW). ................................ ............... 83 2 2 Mean (SD) of water chemistry parameter c oncentrations collected in January, April, July and October 2011. ................................ ................................ ............. 84 2 3 Mean values (SD) of apparent optical properties (AOPs). ................................ ... 85 2 4 Mean values (SD) of inherent optical properties (IOPs). ................................ ...... 86 2 5 Pearson correlation results between water chemistry parameters (Turbidity, Chlorophyll a and Color), %Open Canopy, Distance f rom Spring and AOPs and IOPs for Rainbow River sites. ................................ ................................ ...... 87 2 6 Pearson correlation results between water chemistry parameters (Turbidity, Chlorophyll a and Color), %Open Canopy, Distance from Spri ng and AOPs and IOPs for Weeki Wachee River sites. ................................ ............................ 88 2 7 Pearson correlation results between AOPs and IOPs for Rainbow River sites. ..... 89 2 8 Pearson correlation results between AOPs and IOPs for Weeki Wachee River sites. ................................ ................................ ................................ ................... 90 3 1 Spectral and broadband empirical models of diffuse light attenuation in Rainbow and Weeki Wachee Riv ers. ................................ ................................ .............. 160 3 2 Partial absorption coefficient models for detrital, (a d ), CDOM (a g ), and algal pigment (a ph ) at index wavelengths (440, 550, and 660 nm) in Rainbow (RR) and Weeki Wachee (WW) Rivers as shown in Models 6a to 11b. .................... 162 3 3 Total absorption coefficient (at three wavelengths 440, 550 and 660 nm) and vertical diffuse light attenuation (PAR) coefficients combined dataset models 1 2 to 15. ................................ ................................ ................................ ........... 163 3 4 Evaluation of model precision metrics for spectral and broadband empirical optical models of diffuse light attenuation (K d ) in Rainbow River. ..................... 164 3 5 Evaluation of model precision metrics for spectral and broadband empirical optical models of K d in Weeki Wachee River. ................................ ................... 165 3 6a Evaluation of statistically sig nificant temporal trends using simple linear regressions for measured optical water quality concentrations and predicted inherent optical properties for Rainbow and Weeki Wachee River stations. ..... 166

PAGE 8

8 3 6b Evaluation of temporal trends using simple linear regressions for predicted apparent optical properties. ................................ ................................ .............. 167 4 1 Results from one way ANOVA for spatial variability in Rainbow River stat ions of mean (standard deviation given in parentheses) water chemistry parameters. 238 4 2 Results from one way ANOVA for spatial variability in Rainbow River stations of mean (standard deviatio n given in parentheses) apparent optical properties. .. 239 4 3 Results from one way ANOVA for spatial variability in Weeki Wachee River stations of mean (standard deviation given in parentheses) w ater chemistry parameters. ................................ ................................ ................................ ...... 240 4 4 Results from one way ANOVA for spatial variability in Weeki Wachee River stations of mean (standard deviation given in parentheses) apparent optical properties .. ................................ ................................ ................................ ........ 241 4 5 Results from one way ANOVA for spatial variability in Rainbow River transects of mean biomass DW (dry weight, g m 2 ) of filamentous algae (FA) and macrophytes. ................................ ................................ ................................ .... 242 4 6 Results from one way ANOVA for spatial variability in Weeki Wachee River transects of mean biomass DW (dry weight, g m 2) of filamentous algae (FA) and and macrophytes. ................................ ................................ ...................... 243 4 7 Summary statistics for Sagittaria kurziana (SAG) shoot system characteristics in Rainbow and Weeki Wachee River transects. ................................ .............. 244 4 8 Mean quarterly per segment result s in Rainbow River for metabolism characteristics: gross primary productivity (GPP), net ecosystem productivity (NEP), ecosystem respiration (ER), and GPP to ER ratio (P:R). ...................... 245 4 9 Mean quarter ly per segment results in Weeki Wachee River for metabolism characteristics: gross primary productivity (GPP), net ecosystem productivity (NEP), ecosystem respiration (ER), and GPP to ER ratio (P:R). ...................... 246 4 10 Mean quarterly per segment values in Rainbow River for daily raw incident PAR R PAR R efficiency benthic PAR B and benthic PAR R efficiency .............. 247 4 11 Mean quarterly per segment va lues in Weeki Wachee River for daily raw incident PAR R PAR R efficiency benthic PAR B and benthic PAR R efficiency 248 4 12 Summary of MLR ranges, non parametric correlation coefficients (rho) and significance values ( p value) for various levels of subsurface irradiance. ........ 249 4 13 Pearson correlation results between abiotic and biotic parameters for combined datasets of Rainbow and Wee ki Wachee Rivers. ............................ 250

PAGE 9

9 4 14 Pearson correlation results between gross primary productivity (GPP), and abiotic and biotic parameters for combined datasets of Rainbow and Weeki Wachee Rivers. ................................ ................................ ................................ 251

PAGE 10

10 LIST OF FIGURES Figure page 2 1 Rainbow River study reaches. ................................ ................................ ............... 62 2 2 Weeki Wachee River s tudy reaches. ................................ ................................ ..... 63 2 3 Mean spectral diffuse attenuation coefficients, K d stations obtained January to November 2011. ................................ ................... 64 2 4 Mean spectral diffuse attenuation coefficients, K d stations obtained January to November 2011. ................................ ................... 65 2 5 Overall river average spectral diffuse attenuat ion coefficients, K d Rainbow and Weeki Wachee Rivers obtained January to November 2011. ....... 66 2 6 Mean relative % contribution of absorption coefficients (PAR) to total absorption (PAR), wi th and without absorption due to wa ter itself in Rainbow River ......... 67 2 7 Mean relative % contribution of absorption coefficients (440 nm) to total absorption (440 nm), with and without absorption due to wa ter itself in Rainbow River ................................ ................................ ................................ .. 68 2 8 Mean relative % contribution of absorption coefficients (550 nm) to total absorption (550 nm), with and without absorption due to wa ter itself in R ainbow River ................................ ................................ ................................ ... 69 2 9 Mean relative % contribution of absorption coefficients (660 nm) to total absorption (660 nm), with and without absorption due to wa ter itself in Rainbow River ................................ ................................ ................................ .. 70 2 10 Mean relative % contribution of absorption coefficients (PAR) to total absorption (PAR), with and without absorption due to water i tself in Weeki Wachee River ................................ ................................ ................................ ... 71 2 11 Mean relative % contribution of absorption coefficients (440) to total absorption (440), with and without absorption due to water i tself in Weeki Wachee River ................................ ................................ ................................ .... 72 2 12 Mean relative % contribution of absorption coefficients (550) to total absorption (550), with and without absorption due to water i tself in Weeki Wachee River ................................ ................................ ................................ ... 73 2 13 Mean relative % c ontribution of absorption coefficients (660) to total absorption (660), with and without absorption due to water i tself in Weeki Wachee River ................................ ................................ ................................ .... 74

PAGE 11

11 2 14 Avera ge scattering and absorption coefficie nts in relation to distance from the headsprings in A) Rainbow and B) Weeki Wachee Rivers. ................................ 75 2 15 Ratio of scattering and absorpt ion coefficients over the PAR spectrum in relation to distance f rom the headspring in Weeki Wachee and Rainbow Rivers. ................................ ................................ ................................ ................ 76 2 16 Absorption profiles of total (a p ), detrital (a d ) and algal particulates (a ph ); CDOM (a g ), water (a w ), and total absorption with (a t ) and without (a t w ) water. ............... 77 2 17 Absorption profiles of total (a p ), detrial (a d ) and algal particulates (a ph ); CDOM (a g ), water (a w ), and total absorption without (a t w ) water. ................................ ... 78 2 18 Correlations among individual absorption coefficients at reference wavelengths and total absorption for A) Rainbow and B) Weeki Wachee Rivers. ................................ ................................ ................................ ................ 79 2 19 Correlations between water chemistry concentrations and beam attenuation (c) coefficients in A) Rainbow and B) Weeki Wachee Rivers, ( p < 0.05 for all). .. 80 2 20 Correlations between a bsorption (a), scatter (b) and beam attenuation (c) coefficients in A) Rainbow and B) Weeki Wachee Rivers. ................................ .. 81 2 21 Correlations between attenuation (K d ), and beam attenuation (c) coefficients in A) Weeki Wachee and B) Rainbow Rivers. ................................ ..................... 82 3 1 Relationships among p artial absorption coefficients and total absorption at 440 nm for A) Ra inbow and B) Weeki Wachee River ................................ ............. 122 3 2 Relationships shown in A) a d 440 as a function of color and Chla and B) a g 340 as a function of color, and a ph 440 as a function of Chla in Rainbow River. ...... 123 3 3 Correlations among partial absorption coefficients and total absorption at 550 nm (a t 550) for A) Rai nbow and B) Weeki Wachee Rivers ............................... 124 3 4 Correlations among partial absorption coefficients and total absorption at 660 nm (a t 660) for A) Rainbow and B) Weeki Wachee Rivers. ............................... 125 3 5 Relationships shown in A) a d 550 as a function of chlorophyll a (Chla), and B) a d 660 as a fu nction of Chla, and a ph 660 as a function of Chla in Rainbow River. ................................ ................................ ................................ ................ 126 3 6 Correlations between optical water quality concentrations and attenuation coefficients in A) Rainbow and B) Weeki Wache e Rivers. ................................ 127 3 7 Relationships shown in A) a d 440 as a function of turbidity and a g 340 as a function of color, and B) a d 550 and a d 660 as a function of turbidity and a d 660 as a function of Chla i n Weeki Wachee River. ................................ ................. 128

PAGE 12

12 3 8 Quadratic relationship used to best fit the combined dataset (Weeki Wachee and Rainbow River data combined) is shown for Sg as a function of color. ..... 129 3 9 Spectral and broadband model outputs showing measured K d compared with modeled K d for all eight model output results for Rainbow River. ..................... 130 3 10 Spectral and broadband model outputs showing measured K d compared with modeled K d for all eight model output results for Weeki Wachee River. ........... 131 3 11 Measured K d compared with modeled K d f or the selected spectral and broadband model output results for Rainbow River shown in A) and Weeki Wachee River shown in B). ................................ ................................ .............. 132 3 12 Me an annual color, Chla and turbidity concentration data sh own for the period of record measured by the SWFWMD in each Rainbow River station. ............. 133 3 13 Me an annual color, Chla and turbidity concentration data shown for the period of record measured by th e SWFWMD in each Weeki Wachee River station. .. 134 3 14 Mean annual predicted total absorption coefficients at 440, 550, and 660 nm wavelengths shown for the period of record in each Rainbow River station. .... 135 3 15 Mean annual predicted total absorption coefficients at 440, 550, and 660 nm wavelengths shown for the period of record in each Weeki Wachee River station. ................................ ................................ ................................ .............. 136 3 16 Mean annual predicted scattering coefficients at 440, 550, and 660 nm wavelengths shown for the period of record in each Rainbow River station. .... 137 3 17 Mean annual predicted scattering coefficients at 440, 550, and 660 nm wavelengths shown for the period of record in each Weeki Wachee River station. ................................ ................................ ................................ .............. 138 3 18 Percent contribution of ea ch predicted partial absorption coefficient, and a w 440 to the total absorption coefficient (a t 440) for th e upper Rainbow River stations ................................ ................................ ................................ ........... 139 3 19 Percent contribution of each predicted partial absorption coefficient, and a w 440 to the total absorption coefficient (a t 440) for the middle Rainbow River stations. ................................ ................................ ................................ ............ 140 3 20 Percent contribution of each predicted partial absorption coefficien t and a w 440 to the total absorption coefficient (a t 440) for the lower Rainbow River stations. ................................ ................................ ................................ ............ 141 3 21 Percent contribution of each predicted partial absorption coefficient, and a w 550 to the to tal absorption coefficient (a t 550) for the upper Rainbow River stations. ................................ ................................ ................................ ............ 142

PAGE 13

13 3 22 Percent contribution of each predicted partial absorption coefficient, and a w 550 to the total absorption coeffic ient (a t 550) for the middle Rainbow River stations. ................................ ................................ ................................ ............ 143 3 23 Percent contribution of each predicted partial absorption coefficient, and a w 550 to the total absorption coefficient (a t 550) for the lower Rainbow River stations. ................................ ................................ ................................ ............ 1 44 3 24 Percent contribution of each predicted partial absorption coefficient, and a w 660 to the total absorption coefficient (a t 660) for the upper Rainbow River st ations. ................................ ................................ ................................ ............ 145 3 25 Percent contribution of each predicted partial absorption coefficient, and a w 660 to the total absorption coefficient (a t 660) for the middle Rainbow River stations. ................................ ................................ ................................ ............ 146 3 26 Percent contribution of each predicted partial absorption coefficient, and a w 660 to the total absorption coefficient (a t 660) for the lower Rainbow River stations. ................................ ................................ ................................ ............ 147 3 27 Percent contribution of each predicted partial absorption coefficient, and a w 440 to the total absorption coefficient (a t 440) for the upper Weeki Wachee River stations. ................................ ................................ ................................ ... 148 3 28 Percent contribution of each predicted partial absorption coefficient, and a w 440 to the total absorption coefficient for the lower Weeki Wachee River stations. ................................ ................................ ................................ ............ 149 3 29 Percen t contribution of each predicted partial absorption coefficient, and a w 550 to the total absorption coefficient (a t 550) for the upper Weeki Wachee River stations. ................................ ................................ ................................ ... 150 3 30 Percent contribution of each predicted partial absorption coefficient, and a w 550 to the total absorption coefficient (a t 550) for the lower Weeki Wachee River stations. ................................ ................................ ................................ ... 151 3 31 Percent contribution of each predicted p artial absorption coefficient, and a w 660 to the total absorption coefficient (a t 660) for the upp er Weeki Wachee River stations ................................ ................................ ................................ ... 152 3 32 Percent contribution of each predicted partial absorption coefficient, and a w 660 to the total absorption coefficient (a t 660) for the lower Weeki Wachee River stations. ................................ ................................ ................................ ... 153 3 33 Mean annual predicted diffuse attenuation coefficients (K d ) at 440, 550, a nd 660 nm wavelengths shown for the period of record n each Rainbow River station. ................................ ................................ ................................ .............. 154

PAGE 14

14 3 34 Mean annual predicted diffuse attenuation coefficients (K d ) at 440, 550, and 660 nm wavelengths shown for the period of record in each Weeki Wachee River station. ................................ ................................ ................................ ..... 155 3 35 Mean annual predicted percent subsurface spectral irradiance (%PAR) at various wavelengths shown for the period of record i n each Rainbow River station. ................................ ................................ ................................ .............. 156 3 36 Mean annual predicted diffuse subsurface spectral irradiance (%PAR) at various wavelengths shown for the period of record in each Weeki Wachee River statio n. ................................ ................................ ................................ ..... 157 3 37 Mean annual predicted broadband diffuse attenuation coefficients (K d ) for PAR and %PAR shown for the period of record in each Rainbow River station. ................................ ................................ ................................ .............. 158 3 38 Mean annual predicted broadband diffuse attenuation coefficients (K d ) for PAR and %PAR shown for the period of record in each Weeki Wachee River station. ................................ ................................ ................................ .............. 159 4 1 Linear relationships between mean percent broadband subsurface irradiance and %PAR in light bands and distance from the headsprings in A) Rainbow and B) Weeki Wachee Rivers. ................................ ................................ .......... 210 4 2 Weeki Wachee River mean daily flow (m 3 s 1 ) for synoptic water quality sampling dates shown per quarter. ................................ ................................ ... 211 4 3 Relationships between measured biomass DW of FA and SAG and relative abundance (percen t coverage) using data from both spring systems. ............. 212 4 4 Spatial distribution of mean annual relative abundance of filamentous algae (FA) and macrophytes in Rainbow River. ................................ ......................... 213 4 5 Spatial distribution of mean annual relative abundance of FA native and non native macrophyte species at each station in Rainbow River. .......................... 214 4 6 S patial distribution of mean annual relative abundance FA and macrophytes in Weeki Wachee River. ................................ ................................ ....................... 215 4 7 Spatial distribution of mean annual relative abundance FA native and non native macrophyt e species at each station in Weeki Wachee River. ................ 216 4 8 Spatial distribution of mean annual biomass DW of filamentous algae (FA) and macrophytes in Rainbow River. ................................ ................................ ........ 217 4 9 Spatial distribution of mean annual biomass DW of FA and macrophytes in Weeki Wachee River. ................................ ................................ ....................... 218

PAGE 15

15 4 10 Relationship between measured dry weight per shoot of epiphytes and Sagittaria kurziana shoot dry weight using data from both spring systems. ..... 219 4 11 Comparisons of relative percent absorbance scans n ormalized to the maximum OD of FA and nine mac rophyte species found in various locations in Rainbow River. ................................ ................................ ............................. 220 4 12 Spatial variation of percent relative absorptance SAG in Rainbow River.. ......... 221 4 13 Spatial variation of percent relative absorptance of epiphytic algae scraped from SAG samples in Rainbow River.. ................................ ............................. 222 4 14 Pigment content in macrophyte or filamentous algae f ound in Rainbow (RR1 RR8) and Weeki Wachee (WW0 WW5) River transects. ................................ 223 4 15 Relationships between percent relative absorptance of Sagittaria kurziana Chla FW content and epiphytes scraped fro m SAG samples in Rainbow River. ................................ ................................ ................................ ................ 224 4 16 Logarithmic relationship bet ween epiphyte absorptance per reference wavelength as a function of epiphyte biomass per shoot (g) for both systems combined. ................................ ................................ ................................ ......... 225 4 17 Relative absorptance of plankton entrained in the water column at various locations in Rainbow River across the PAR spectrum. ................................ ..... 226 4 18 Boxplots showing temporal variability for benthic PAR in relation to quarter for A) Rainbow, and B) Weeki Wachee Rivers. ................................ ..................... 227 4 19 Boxplots showing temporal variability for benthic e cological efficiency (EE B ) in relation to quarter for A) Rainbow, and B) Weeki Wachee Rivers. ................... 228 4 20 Boxplots showing spatial and temporal variabilit y for GPP in relation to distance of the mi dpoint of each segment from the headsprings for A) Rainbow, and B) Weeki Wachee Rivers. ................................ .......................... 229 4 21 Relationships between average daily GPP in and PAR above the water surface and in the benthic zone. ................................ .... 230 4 22 Relationships between hourly GPP and PAR in the uppermost segment in Rainbow River. ................................ ................................ ................................ 231 4 23 Samp le photosynthesis vs. irradiance curve (P I curve) in the uppermost segment in Rainbow River. ................................ ................................ ............... 232 4 24 Relationships between Sagittaria kurziana (SAG) biomass DW (dry weight) and percent subsurf ace irradiance for different bands ................................ ...... 233

PAGE 16

16 4 25 Relationships between Sagittaria kurziana (SAG) biomass DW (dry weight) and percent subsurface irradiance in the blue band ................................ ........ 234 4 26 Non linear relationships betw een water column chla concentrations and A) NO 3 B) NH 4 + C) TP and D) flow discharge in Rainbow River. ....................... 235 4 27 Relat ionships between water column concen trations and A) FA biomass, relative abundance epiphyte load and type in Rainbow River. ........................ 236 4 28 Relationships between filamentous algae biomass dry weight ( FA BM) and A) velocity, and B) nitrate (NO 3 ) for the combined dataset. ................................ .. 237

PAGE 17

17 LIST OF ABBREVIATIONS % PAR Percent subsurface irradiance (%) % PAR ( Percent s pect ral subsurface irradiance (%) AOP Apparent optical property A ph Spectral absorption coefficient of algal pigments (m 1 ) a d Spectral absorption coefficient of detritus (m 1 ) a g Spectral absorption coefficient of CDOM (m 1 ) a w Spectral absorption coefficient of water ( m 1 ) a t Total spectral absorption coefficient (m 1 ) Total scattering coefficient (m 1 ) BOM Broadband optical model BB Broadband Beam attenuation coefficient (m 1 ) CDOM Colored Dissolved Organic Matter Chl a Chlorophyll a concentration (g L 1 ) EE Ecological efficiency (%, g O 2 mol 1 ) ER Ecosystem respiration (g O 2 m 2 d 1 ) GPP Gross primary production (g O 2 m 2 d 1 ) HSD Horizontal secchi depth (m) IOP Inherent optical property K d Downwelling diffuse attenuation coefficient (m 1 ) K d Spectral downwelling diffuse attenuation coefficient (m 1 ) NEP Net ecosystem production (g O 2 m 2 d 1 ) OWQ Optical water quality PAR Photosynthetically active radiation (mol m 2 s 1 )

PAGE 18

18 Spectral photosynthetically active radiation (mol m 2 s 1 ) POR Period of record PUR Photosynthetically useable radiation (mol m 2 s 1 ) Spectral p hotosynthetically useable radiation (mol m 2 s 1 ) SAV Submerged aquatic vegetation SOM Spectral optical model SWFWMD Southwest Florida Water M anagement District

PAGE 19

19 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 SPECTRAL DISTRIBUTION OF LIGHT IN FLORIDA SPRING ECOSYSTEMS: FACTORS AFFECTING THE QUANTITY AND QUALITY OF LIGHT AVAILABLE FOR PRIMARY PRODUCERS By Mary Lucy Szafraniec May 2014 Chair: Joseph J. Delfino Cochair: Mark T. Brown Major: Environmental Engineering Sciences Light availability is a major forcing factor for spring ecosystem productivity and sustainability. It appears that water clarity has decreased in many springs in Florida which could have had an effect on the quantity and quality of light available for primary producers in springs and spring runs Factors controlling the loss of water clarity and light availability are poorly defined in spring systems. To understand the causes of increased light attenuation in these optically complex lotic systems it i s necessary to measure the quantit y and the quality of light available to primary producers. Optical water quality determines the underwater light field that provides the basis for habitat suitability, in terms of vegetation abundance, distribution, and survival in springs. Studies were co nducted to assess the spectral distribution and potential limitation of wavelength measuring the percent blue, green and red light available to primary producer communities in springs. Optical properties of the unde rwater light field were assessed to determine the relative magnitude and contribution of key inherent water clarity driving components. Results revealed a spatial optical gradient in both systems studied Light attenuation of the blue

PAGE 20

20 band was dominated b y algal pigment and colored dissolved organic matter absorption and the green and red light bands were most strongly attenuated by scattering and absorption by particulates. R esulting underwater spectral light field characteristics for each spring system w ere used to develop and calibrate spectrally explicit site specific and general spring system empirical optical models to predict historical water clarity conditions Optical modeling results indicate that both spring systems studied have not significan tly changed since the early 2000 s, aside from a few trends where particulate scattering and abs orption properties have increased in one system and decreased in the other. T he lower portions of each river studied had low levels of blue light available in th e benthic zone throughout the time series Finally, linkages were established among light availability, benthic community composition distribution and ecosystem primary productivity. The results from a synoptic field survey of spectral light availability and submerged aquatic vegetation biomass identified a minimum blue light requirement threshold range of 38 45%, below which may not be sufficient to support the growth and production of Sagittaria kurziana a n important primary producer i n most spring sys tems.

PAGE 21

21 CHAPTER 1 INTRODUCTION S pring ecosystems in Florida are well known for their exceptional water quality and water clarity and are an important aquatic resource as they provide a unique habitat for aquatic flora and fauna while also providing extensi ve recreational and economic opportunities for human use. North central Florida has an extensive area with a highly permeable karst geology that provides numerous locations and multiple pathways by which nutrient (specifically nitrate) enriched groundwater can be transported to surface water discharged through spring vents from the aquifer systems (Notestein et al. 2003; De Brabandere et al. 2007; Heffernan et al. 2010a; Heffernan et al. 2010b). Unfortunately many spring ecosystems are experiencing a grad ual decline in both w ater quality and water clarity due to anthropogenic activities within their karst springsheds (Jones et al. 1996; Scott et al. 2004). The main water source for Florida springs is groundwater that contains relatively high levels of nitr ate and other constituents which determine the water quality available to the spring ecosystem (Knight and Notestein in Brown et al. 2008; WSI 2010). Nitrate concentrations have nearly doubled in Rainbow Springs within the last 30 years, which is coincide nt with a 20% decline in the percent coverage of the dominant native submerged aquatic vegetation (SAV specifically rooted macrophytes ) species Sagittaria kurziana (Atkins 2012). In Weeki Wachee Springs, nitrate concentrations have also increased from 0.0 7 to 0. 52 mg/L (Frazer et al. 2006) from 1970 to 2000, and have doubled in the last 14 years to above 1 mg/L. The increase in nutrients paralleled a major decline in SAV biomass (75%) from 2000 to 2005 (Frazer et al. 2006). It is

PAGE 22

22 unknown to what extent nut rients or other factors have contributed to the ecological decline in Rainbow and Weeki Wachee Rivers. Efforts have been made to increase protection of spring ecosystems; however land use changes within the watershed and springshed have inevitably led to altered surface and groundwater water chemistry, which considerably increased nutrient loading to the springs (Munch et al. 2006). The decline of aquifer water quality from land use activities in an unconfined karst geological setting has been documented b y many (Notestein et al. 2003; Cohen 2007; Quinlan et al. 2008). Some of the land uses of concern that are putting pressure on spring water quality are residential (septic use), commercial, and agriculture, especially horse farms and cattle operations in the western and central springs of the state (Jones et al. 1996) Other issues that most springs have experienced are reduced wetlands in the watershed, hardened shorelines, introduction of non native invasive flora and fauna, increased disturbances relate d to recreational activities, and natural sediment regimes have been disturbed in spring ecosystems such as Rainbow River from historic phosphate mining (SWFWMD 2004) Other apparent anthropogenic stressors that could potentially a lter Florida springs are related to spring discharge. Spring discharge has been reduced by rainfall deficits due to prolonged droughts and groundwater withdrawal (Knight and Notestein in Brown et al. 2008) All of the factors mentioned could potentially have negative consequences on spring ecosystem health and sustainability. Abiotic and biotic factors and their interactions need to be investigated to provide a bett er understanding of how

PAGE 23

23 to restore these spring systems so that they function efficiently and continue to provide the ecological services that are valued by society To maintain or restore a viable spring ecosystem, it is necessary to understand the forcing functions that drive healthy ecosystem form and function (Odum 1957b; Knight and Notestein in Brown et al. 2008). There are many biotic and abiotic factors that impact the health and ecological integrity of a spring. These factors include sunlight, groundwater flow (with associated nutrients and other chemical constituents), flora and fauna (productivity, herbivory an d other trophic interactions) atmospheric inputs (wind, rain, storms with associated water and nutrients), economic drivers (tourism, markets), and exports from the spring ecosystem (Odum 1957b; Knight and Notestein in Brown et al. 2008). The effects of t he interaction among these factors and processes may be more critical than their singular effects (Rosemond 1993), and each of these factors interact in a complex fashion to influence ecology The various factors that control primary production in spring and other aquatic ecosytems have been studied and described by many (Leconte 1861; Yount 1956; Canfield and Hoyer 1988; Frazer et al. 2001 b Notestein et al. 2003; Kemp and Boynton 2004; De Brabandere et al. 2007; Hauxwell et al. 2007; Quinlan e t al. 2008; Brown et al. 2008; Heffernan et al. 2010a; Heffernan and Cohen 2010 ; King 2012 ). However, perhaps the most comprehensive scientific works concerning productivity in springs and the factors controlling primary productivity were those by H.T. Odu m. Odum studied the productivity, trophic structure, and energy flow in many springs in Florida, with an intensive investigation in Silver Springs (Odum 1957a, 1957b; Quinlan et al. 2008). As a result of his studies, Odum determined that springs are unique chemostatic and

PAGE 24

24 temporally stable environments (Odum 1956, 1957a, 1957b; Knight and Notestein in Brown et al. 2008), and are considered steady state systems (Whitford 1956) with constant temperature, gas and mineral content, and discharge (Knight and Not estein in Brown et al. 2008). The concept of springs in a steady state provides for an ideal opportunity for the identification of environmental factors that control the biological structure and productivity of spring ecosystems (Duarte and Canfield 1990; Knight and Notestein in Brown et al. 2008). The SAV epiphytes (a ttached algae) and benthic algal communities are the key (Odum 1957a). I n Silver Springs, researchers determined that there has been a significant shift in primary producers with a rise in prominence of epiphytic and benthic algal mats in the system y of the vegetation in the 1950 s (Quinlan et al. 2008). Quinlan et al. (2008) attributed high variability of th e spatial distribution of epiphytes to levels of flow discharge in Silver River, with the highest epiphyte biomass occurring in regions of low flow. In another Florida spring, King (2012) found that reduced flow velocities can increase filamentous algae gr owth and production rates. Further, flow velocity c ontrols the drag force and resultant export rates of algae As filamentous algal mats proliferate due to more suitable habitat and hydrologic conditions light availability for SAV is reduced with negativ e consequences for their growth and production (Hauxwell et al. 2007). Previous literature offers insight into the interaction between primary producers such as algal and SAV communities in aquatic ecosystems and the role that forcing factors play in syste m metabolism. As mentioned previously, nitrate concentrations

PAGE 25

25 have increased in many springs over the last 6 0 to 100 years This increase in nitrate has been associated with an increase in nuisance algal communities. However, results from recent studies do not support a quantitative relationship with nitrate concentrations and algal biomass or distribution in springs (Hoyer et al. 2004; Stevenson et al. 2007; Albertin 2009; Pinowska et al. 2009). In a study on the Rainbow River in Florida, Cowell and Dawes (2005) determined that the source of chlorophyll measured in the Rainbow River water column was phytoplankton. Chlorophyll producing organisms were found to be more abundant two kilometers downstream of the headsprings. This trend corresponded to the decre ase in nitrate along the downstream gradient suggesti ve of an inverse correlation between chlorophyll and nitrate ( Cowell and Dawes 2005) however recognizing that this correlation is different than causation Conversely, in a follow up lab experiment by C owell and Dawes (2007), the effect of increased nutrients on phytoplankton growth was investigated, which showed little to no response. An increase in phosphorus and nitrate showed no significant ( p >0.05) increase in phytoplankton biovolumes, and interesti ngly the addition of trace metals to the nutrients caused significant ( p <0.001) phytoplankton growth ( Cowell and Dawes 2007). Therefore, the lack of a direct relationship of algae and nutrients alone suggests that other factors such as light, disturbance, grazing, sediment characteristics, flow velocity, and the presence of SAV can influence the accumulation of phytoplankton and benthic algal biomass (Hoyer et al. 2004; Doyle and Smart 1998; Albertin 2009; Pinowska et al. 2009; Heffernen et al. 2010b ; King 2012; Liebowitz 2013 ).

PAGE 26

26 In several spring studies, researchers have found that light primarily controls (Odum 1957) abundance and distribution while nutrient availability (and other factors) secondari ly affects primary producers (Canfield and Hoyer 1988; Hoyer et al. 2004; Knight and Notestei n in Brown et al. 2008 ). Based on their study Canfield and Hoyer (1988) suggested that light availability is the chief factor limiting growth of SAV in Florida streams, although substrate ty pe, water depth, and current velocity can also be limiting. Nutrient enrichment can result in an increase in algal biomass in some sunlit (not shaded) streams, which suggests the potential of co limitation of primary production by nutrients and light (Rose mond 1993). In a study characterizing influential factors of SAV distribution and abundance in three coastal spring fed rivers in Florida, it was found that the percent of incident light reaching the substrate and salinity concentrations had the greatest i mpact on SAV biomass, while nutrient loads and concentrations did not have a direct influence on the abundance and distribution of SAV (Hoyer et al. 2004). I nteractions among these and other abiotic and ity and distribution of primary producer biomass. As demonstrated by previous investigations into the factors that influence ecosystem function and structure the overall productivity and diversity of floral and faunal assemblages in springs are highly de pendent on light availability and water clarity, which are both controlled by the underwater light environment (Odum 1957a; Duarte and Canfield 1990). In several Florida spring runs declines in water clarity have been observed in overall s ystems, and on s patial scales where water transparency (surrogate measure of clarity) rapidly declines with distance from the headsprings. The

PAGE 27

27 basis for the decline in transparency is not yet well understood, although it is hypothesized that longitudinal decline of water clarity ha s a lways been a n aturally occurring phenomenon. However, changes in water qu a lity as a result of anthropogenic influences may have accelerated the loss of water clarity T he lack of robust water clarity or light attenuation data in springs make i t difficult to determine if overall water clarity and light availability has significantly declined over time The underwater light environment (optical environment) characterizes light availability that provides the basis for community metabolism. Charac teristics that contribute to the underwater light environment directly influence water clarity (Duarte and Canfield 1990; Gallegos 1993; Kirk 1994; Anastasiou 2009). Light availability is often a limiting factor for vegetation communities in many aquatic e cosystems (Canfield and Hoyer 1988; Kirk 1994 ; Biber et al. 2008 ). In clear waters such as spring systems, light attenuation is strongly influenced by background attenuation due to water itself and scatter ing by particles in the water column while less cl ear waters may have other factors attenuating light such as colored dissolved organic matter (CDOM), phytoplankton, and inorganic particles or detritus (Effler 1985; Kirk 1994; Krause Jensen and Sand Jensen 1998). The relative fraction and magnitude of the se factors contributing to light attenuation can significantly impact the amount of light available for photosynthesis by SAV and other primary producers (Kirk 1994; Krause Jensen and Sand Jensen 1998 ; Gallegos 1994 ). With regard to spring ecosystems, fe w studies have considered the optical properties of the underwater environment (Leconte 1861; Duarte and Canfield 1990; Frazer et al. 2001 a ). And to date, no studies have investigated th e effects of a variable

PAGE 28

28 underwater light environment on photosynthesis in springs. However, a great deal of work has been conducted to characterize the underwater light environment in marine and lentic and to a much lesser extent, lotic ecosystems. It has been recognized that the light environments of clear blue waters of m arine systems are similar to the headsprings of clear spring fed rivers in Fl orida (Duarte and Canfield 1990 ). The similarity between the underwater light environments of marine and spring systems provides a basis for the application of methods used to cha racterize the underwater light environments in marine, estuarine and freshwater lakes. Similar approaches and methods can also be applied to derive optical models relating inherent optical properties (IOPs) with factors contributing to light attenuation in springs, and extending that relationship to determine habitat suitability for SAV as it has been done for seagrasses (Gallegos 1993, 1994; Gallegos and Kenworthy 1996; Biber et al. 2008 ) My research aimed at increasing the base of knowledge of spring eco logy and ecosystem function by looking at whole biological communities and selected assemblages of the biological community. The overall objective of my research wa s to assess the abiotic and biotic factors contributing to light attenuation and the effect that light availability has on primary producer community distribution and resultant primary production in Florida springs and spring runs. My research wa s compose d of three separate, but interrelated studies. The first study (Chapter 2) determine d the re lative contributions and magnitude of environmental factors that contribute to light attenuation on a spatial and temporal scale in two Florida spring systems As part of th e first study, I measure d the spectral quantity and quality of light reaching the b enthic surface and characterize d the relative contribution and

PAGE 29

29 magnitude of light attenuating constituent s. In my second study (Chapter 3) I develop ed spectrally explicit empirically derived optical models that relate d optical water quality (OWQ) paramet ers and IOPs to the light attenuation coefficient ( K d ) in spring ecosystems, and I used regression models to predict historical light attenuation from historically available water quality data for Florida springs. Finally in the third study (Chapter 4) I evaluate d the effects of the underwater optical environment and other environmental factors on primary producer biomass distribution and metabolism

PAGE 30

30 CHAPTER 2 CAUSES OF LIGHT ATTE NUATION IN FLORIDA S PRING ECOSYSTEMS Introduction Submerged aquatic veget ation (SAV specifically rooted macrophytes ) requires light for photosynthesis in all aquatic environments. In Florida spring and other clear lotic ecosystems, SAV assemblages are a vital component of the overall pri mary producer community and depend on ab iotic factors such as a suitable light environment i .e. water clarity and other habitat suitability factors such as flow velocity and sediment quality (Biggs 1996) Water resource managers have aimed at preserving and restoring SAV by improving water qua lity and wate r clarity in springs and spring fed rivers Kirk (1988) biosphere or the In estuarine ecosystems, light is generally considered the most important factor controlling the distribution and biomass of seagrasses (Dennison et al. 1993; Gallegos et al. 2009) Seagrass comm unities have declined worldwide, likely caused by a decrease in water clarity brought on by a reduction in water quality. Seagrasses, like many SAV, have a high light requirement (Dennison et al. 1993; Duarte 1995; Gallegos et al. 2009), and freshwater SAV like seagrasses, have also experienced a decline in man y spring ecosystems ( Quinlan et al. 2008) Therefore, it is necessary to characterize light in aquatic environments to determine if habitats are suitable for SAV growth prior to restoration efforts Optical properties not only control the amount of light available for photosynthesis by primary producers they are also important for aesthetic appeal and recreational value of aquatic resources (Kirk 1988). For these reasons, a consideration

PAGE 31

31 must be made to better understand which optical properties have the greatest influence on penetration of solar radiation through the water column. Photosynthetically active radiation (PAR) is a broadband quantity of light that is measured in units of mol photons m 2 s 1 and is integrated across the visible spectrum of 400 700 nm wavelengths. PAR is also referred to as the visible light spectrum and is used in photosynthesis by primary producers (Kirk 1994 ; Kelble et al. 2005 ). PAR is used as an indicator of light quan tity through the water column; however, photons of different PAR wavelengths are not equally absorbed by light absorbing pigments such as chlorophyll or other accessory pigments (Kirk 1994; Anastasiou 2009). Photosynthesis is a function of PAR intensity in the underwater light field ( Duarte 1991 ; Kenworthy and Fonseca 1996 ) More importantly, it is a function of how well the spectral distribution of PAR corresponds to the absorption spectrum of a given species of SAV or algae ( Enriquez et al. 1994; Kirk 199 4 ; Kenworthy and Fonseca 1996; Zimmerman 2003; Drake et al. 2003 ; Gallegos 2009 ). Certain wavelengths of light, such as the blue (400 500 nm) and red (600 700 nm) bands are more readily absorbed by plants, which increases the efficiency of photosynthesis ( Kirk 1994 ; Gallegos et al. 2009 ). The spectral composition of the underwater light field is an important determinant of SAV community composition and distribution ( Zimmerman 2003; Anastasiou 2009; Gallegos et al. 2009; Dixon et al. 2010 ) However, the spe ctral properties of the light field are not taken into account with only broadband PAR measurements, which provides a basis to express PAR in spectral terms and is give ) to effectively, it is imperative to recognize the importance of knowing not only the quantity

PAGE 32

32 e benthic zone of aquatic systems. A need to understand the causes of light loss, or loss of water clarity, is also crucial for resource management, which should include measurements of the spectral properties of the aquatic medium ( Zimmerman 2003; Anastas iou 2009; Gallegos et al. 2009; Dixon et al. 2010 ) As light passes through water it is subject to both absorption and scattering processes, resulting in the attenuation of light. The attenuation coefficient K d and PAR are considered apparent optical prop erties (AOPs) and are very difficult to measure properly (Kirk 1984; Dixon et al. 2010) AOPs can be affected by the ambient light field, including solar angle, cloud cover, shade from water vessels and canopy, water depth, as well as constituents in the w ater column and accurate measures can be difficult to obtain (Kirk 1984). The attenuation of light in aquatic ecosystems is the direct result of i nherent optical properties (IOPs) that include the absorption (a) scattering (b) and beam attenuation (c) co efficients in the water column, and are affected only by the dissolved and suspended particles in the water that cause absorption and scatter of light (Gallegos et al. 1990 ; Durako et al. 2010) The third IOP, the beam attenuation coefficient ( c) can be ca lculated by summing a + b = c (Kirk 1988) Light scattering is a relevant parameter that should be measured as part of the inherent optical properties necessary to characterize the u nderwater light environment. Scattering, unlike absorption does not remov e light Scattering simply causes a photon of light to follow a zig increase in light attenuation (Kirk 1994). However, g iven the complex local hydraulic

PAGE 33

33 dynamics in spring fe d rivers, where flow velocities are relatively rapid and variable throughout the water column and in the benthic zone (King 2012), scattering could play a greater role on controlling total attenuation in the water column than absorption Essentially, all of the light absorption in aquatic systems is carried out by four main components and are described by absorption coefficients (a), with units m 1 : water itself (a w ), colored dissolved organic matter (CDOM) (a g ), algal pigment (a ph ), and detrital material (a d ) suspended in th e water column (Kirk 1980, 1994; Gallegos 2005; Gallegos et al. 2009; Durako et al. 2010) It is important to determine the individual and total contribution of spectral IOPs on light attenuation of the aquatic medium (Gallegos et al. 2 005 2009 ). Each of the different spectral absorption coefficients are measured, and then the total absorption (a t ) spectra ( per wavelength or integrated for each blue, green and red color b and ) can be calculated by summing the individual coefficients at p articular wavelengths (Kirk 1994) as in Equation 2 1 : a t w g ph d Equation ( 2 1 ) The contribution of each of these components is different for each wavelength ; the absorption of water, a w properties of pure water, and it absorbs primarily in the red band (above 550 nm); the absorption of CDOM, a g d blue band; while algal pigments a ph to both the blue and red bands (Kirk 1 994). The absorption spectrum of the aquatic medium determines the relative rates of light attenuation through the water column in the different wavelengths and wavebands of PAR ( Kirk 1994). The attenuation of light in the photosynthetic waveband occurs at

PAGE 34

34 different rates in different parts of the visible spectrum, and as a result, the spectral composition of the downwelling flux of photons changes appreciably with increasin g depth (Kirk 1994). There are two commonly used methods to measure water clarity an d light penetration in water. Secchi disks (m) are used to measure transparency, and the diffuse attenuation coefficient (K d m 1 ) is calculated for downwelling irrad iance from measurements of PAR with quantum irradiance sensors ( Kirk 1994). Light is atten uated in an exponential fashion with depth, and can be calculated as the light attenuation (also called extinction) coefficient K d derived from the Beer Lambert attenuation function (Kirk 1994): K d = ln (I O /I Z ) / z Equation ( 2 2 ) where: I O = Irradiance of light just below the surface of the water I Z = Irradiance of light at depth z Z = depth in water In many spring sampling programs in Florid a, optical quality monitoring has been limited to using Secchi disks to measure transparency alt hough a few studies have incorporated PAR (broadband) measurements (Canfield and Hoyer 1988). In spring systems, good visual clarity causes the water column to exceed the limits of the vertical Secchi disk technique Therefore, for many of the long term mo nitoring programs, the conventional method of measuring transparency with a vertical Secchi disk was modified to approximate Secchi depths beyond the total depth of the sampling site in the spring or spring fed river To account for Secchi distances greate r than th e total depth, water clarity can be measured using a horizontal Secchi disk (Davies Colley 1988 ; Davies Colley and Smith 1995 ; Montes Hugo et al. 2003 ) to measure horizontal Secchi distance ( HSD )

PAGE 35

35 There are many sources of variability that reduce the reliability of general Secchi measurements. Errors that produce great variability include measurement errors, which consist of several factors such as inconsistency in data recording (observer dependent), a high sensitivity of Secchi distance to ambie nt light conditions where cloud cover, angle of the sun, roughness of the water surface or strong currents could skew Secchi readings (Effler 1985; Preisendorfer 1986). In contrast, the measurement of K d is less sensitive to these issues when appropriate m ethods are used to account for changes in ambient light conditions, such as adjustments for on deck (above water surface) air readings to limit inconsistency of cloud cover and for solar elevation (zenith angle) which requires additional data such as dat e and time along with the latit ude /longitude of sample location (Kirk 1994) It has been found that the processes of absorption and scattering affect Secchi and K d measurements differently, which does not support the concept that K d is proportional to 1/SD a commonly used relationship to calculate K d when only Secchi values are available (Effler 1985). Furthermore, light is attenuated and reduced (with depth) through the water column in a vertical direct ion in an exponential manner that can be predicted i f the absorption and scattering of light by water and its contents are known (Kirk 1994). Accordingly, horizontal Secchi measurements cannot account for the vertical attenuation of light, but can be used as a simple approximation of water transparency or visibility, that provides some useful information when more robust methods are unavailable. Effler (1985) demonstrated that there is uncertainty of the product of K d SD when the relative contributions of scattering and absorption are not constant and u ni form This issue is particularly pertinent because the scatter to absorption ratio (b:a)

PAGE 36

36 varies within and among spring rivers and other lotic systems For these reasons, Secchi disk transparency is not considered to be a reliable predictor of light attenu ation for ecological modeling efforts and actual measurements of K d (or spectral K d equipment is available) should be cond ucted ( Effler 1985). However, a s mentioned previously, accurate measurements of K d (or spectral K d ) are difficult to conduct an d OWQ and IOP data could be used to conduct spectrally specific light attenuation modeling to reduce inherent error introduced from changing atmospheric conditions Modeling can also provide a greater spatial distribution of light attenuation data that can then be used to determine if sufficient light is available for SAV ( Gallegos 2005; Gallegos et al. 2009). The purpose of this study wa s to advance the understanding of the underwater light environment in selected spring ecosystems in Florida One objecti ve was to evaluate the spectral downwelling atte nuation ( K d , and to determine t he relative i mportance of detritus, CDOM and algal pigments to light attenuation Information obtained from the data analysis of this study was used in the developm ent of empirical optical models with results presented in Chapter 3. Methods Study Area Two spring fed river systems Rainbow River and Weeki Wachee River, were assessed for the quantity and quality of light and causes of lig ht attenuation. Both of the spring fed river s flow is dominated by groundwater with minimal contributions from surface runoff. The rivers were chosen to cover a range of optical type, riparian shading, and hydrogeomorphic characteristics

PAGE 37

37 Rainbow River ( RR) located in Marion County in northern Florida USA is fed by a complex of several large spring vents and numerous smaller springs that occur within the first mile of the river. It is a relatively large, minimally sinuous (Figure 2 1) relatively clear river with minimal riparian cover that extends over the wide river channel. Weeki Wachee River (WW) is relatively small and is located in Hernando County Florida, USA. It is primarily fed by one large spring vent, with a few smaller springs that contribut e minimally to river flow within the first river mile. Weeki Wachee is a moderately sinuous (Figure 2 2) relatively clear river with a moderate to heavy riparian corridor composed of mixed hardwood wetland forest shading large sections of the river channe l and provid ing considerable tannins to the river during and after heavy storm events Station locations were based on existing wat er quality monitoring stations [ Southwest Florida Water Management District (SWFWMD) Monitoring Program Brooksville, FL, USA ] Station locations for RR and WW are shown in Figures 2 1 and 2 2, respectively Optical Properties and Water Chemistry On a quarterly basis in 2011 (January, April, July and October) surface water samples (0.2 to 0.3 m depth) were collected for measur ements of IOPs at the twelve spring river stations concurrently with ongoing routine water chemistry (turbidity, chlorophyll a, color) and horizontal Secchi distance (HSD) samples The IOP samples were analyzed at the SWFWMD L aboratory on a bench top dual b eam UV/VIS spectrophotometer ( Perkin Elmer Lamda 40 ) for algal pigment (a ph ) and detrital (a d ) absorption coefficients CDOM (a g ) absorption was analyzed at the University of South Florida Spectroscopy Laboratory (St. Petersburg, FL, USA) on a bench top d ual beam UV/VIS spectrophotometer (Hitachi U3900U, Hitachi High Technologies Corporation). S pectral absorption coefficients for the IOPs w ere determined at 1 nm intervals from

PAGE 38

38 2 00 to 8 00 nm following methods as described in the Ocean Optics Field Measureme nts and Data Analysis P rotocols for inherent optical properties ( NASA 2003) The total particulate absorption coefficient (a p ) is the sum of a d and a ph Spectral absorbance of the p articulate fraction (a p ) was determined by the quantitative filter pad tec hnique (Kishino et al. 1985). Absorbance of the river sample was measured relative to a blank wetted with Milli Q water glass fiber filter placed in front of the sensor end of the instument beam to minimize light scattering The volume of river water filte red was dependent on particle load on the filter. Absorbances were conver ted to absorption coefficients (m 1 ) by multiplying the measured absorbance by 2.303 [i.e., ln(10)] and by the geometric pathlength which is found by dividing the area of the filter by the volume of water filtered and then correct ing for the path length amplification factor (Tzortziou et al. 2006) The absorption at 750 nm was assumed to be the residual scattering in the filter caused by either non uniformity in filter wetness or str ay light and was subtracted from the absorbance values at all other wavelengths to calculate a p (Bricaud and Stramski 1990) The spectral absorption of detritus (a d ), or the unpigmented fraction, was determined by cold methanol extraction of pigments from the filters (Kishino et. al1984). The absorption coefficient of algal pigments (a ph ) was obtained by subtracting a d from a p The absorption coefficient for CDOM (a g ) was measured spectrophotometrically at 1 nm intervals on samples that were filtered in th e field immediately after sample collection through a 0.22 um Nucleopore filter into pre combusted, acid cleaned amber glass bottle s and stored at 4C until laboratory analysis. The spectral absorption of

PAGE 39

39 CDOM was measured in 10 cm pathlength quartz cuvett es against a Milli Q water blank. Very low CDOM absorption was encountered at most of the spring fed river sample stations therefore a 1 cm cuvette was not sufficient for accurate measurement of a g CDOM absorption spectra were also corrected by subtracti ng the absorbance at 750 nm from each wavelength and then absorbance values were converted similarly as for particulate absorption coefficients by multiplying by 2.303 and dividing by the 0.1 m pathlength. The method of employing an exponential function of absorption at 440 nm (a g440 ) and spectral slope (S g ) (Bricaud et al.1981) was examined to estimate the CDOM absorption spectra but was not used because a g 440 (m 1 ) for m stations w ere below equipment detection levels hence measured a g values were used. The spectral slope (S g ) of the regression between CDOM and wavelengths in the 200 500 nm region (S 200 500 ), and the ratio of slopes calculated between a g and wavelengths in the 275 to 295 nm range and 350 to 450 nm range ( S R275 295:350 45 0 ) were used to evaluate the source and quality of CDOM in both systems. The spectral slope for a g describes the exponential decrease in CDOM absorption with increasing wavelengths and was determined by applying a regression fit to log transformed a g for w avelengths in the 200 to 500 nm spectral range. K nowledge of the spectral slope and slope ratios are useful metrics that can be used to further characterize CDOM into ranges of molecular weights and differentiating between fulvic and humic acids ( Laurion e t al. 2000 ; Helms et al. 2008 ) Scattering is a direct function of turbidity rather than suspended solids (Gallegos 1994 ) therefore scattering was not measured directly and was calculated from turbid ity

PAGE 40

40 concentrations obtained from water chemistry sample s. Scattering due to water itself is negligible and was not included Scattering (b) due to particulates (alg a l pigment and detrital matter) is dependent on particle size, refractive index, algal species composition, particle mineralogy, and detrital compo sition ( Coble 2007; Anastasiou 2009 ). Scattering was calculated by Equation 2 3 Equation ( 2 3 ) Where, is s cattering at a specific wavelength m 1 ; [Turbidity] is the t urbidity concentration measured in the water column, NTU The beam attenuation coefficient ( c, at a specific wavelengt or integrated for color band or PAR ) was calculated as the sum of total absorption (a) and scattering (b) c = a + b Horizontal Secchi distance (HSD) was measured in situ usually within 10 minutes of PAR measurements. The black and whi te quartered Secchi disk (19.5 cm diameter ) was placed approximately 0.5 m under water in a horizontal direction and a snorkeler in the water swam away from the disk until it was no longer visible. The distance where the disk disappeared was recorded alo ng with the distance where the disk reappeared and those two distances were averaged for the HSD. There is a high potential to introduce measurement error in light measurements, which could be due to several issues such as boat hull reflection, bottom ref lectance from sediments, insufficient integration time, and insufficient depth (<0.5 m) for stable readings. To reduce the effect of angular distribution of light due to low sun horizons during ea rly morning and late afternoon that may contain greater prop ortions of diffuse light, l ight data were collected between the hours of 10:00 and 14:00 hours Eastern Standard Time Avoiding these issues minimized the effect of measurement error on

PAGE 41

41 estimation of K d Broadband PAR (integrated over 400 700 nm) data was c ollected above the water su rface (deck PAR readings) and underwater PAR was measured at 0.5 m incremental depth s down to 0.2 m above the river bed or top of vegetation with dual LI wi th a LI COR (model L I 1400, Lincoln, NE) datalogger Profiles of spectral scans of downwelling spectral irradiance were collected with a planar irradiance cosine collector (Ho bi L abs, Inc., Bellevue, WA) which wa s mounted onto a PVC measuring rod The co sine collector wa s connected via a fiber optic cable to a portable spectrometer ( model HR2000, Ocean Optics, Dunedin, FL). A field laptop PC (Panasonic Toughbook, Panasonic C orporation, New York, NY), was used to store collected data and run the Ocean Opti cs program OOI Base32. For each discrete depth, three consecutive scans were recorded one second apart and averaged to create a composite scan to reduce noise in the light profiles resulting from light focusing from water turbulence (Pfannkuche 2002) Init ial scans were taken in the air just above the water surface followed by surface water reading approximately 0.05 m below the water surface Following the surface water reading, scans were taken at 0.25 m intervals until 0.2 m above the bottom was reached Corrections to light attenuation calculations were made to account for cloud cover and sun elevation angle with each corresponding deck PAR reading. The light attenuation coefficient, K d was calculated in two ways, by Equation 2 2, the transmission meth od, which only takes the top (Io) and bottom (Iz) depths into account The sec ond method, the slope method, was found to be more robust since the entire depth profile of PAR wa s utilized in the calculation of K d where PAR wa s air

PAGE 42

42 normalized and wa s plotte d on a logarithmic scale as a function of depth below the surface, and wa s inclusi ve of all depths of the profile. The slope method wa s the best estimation of K d when concurrent air PAR measurements we re made to correct for changes in incident light from c hangi ng weather or shade conditions, and was the method chosen to report K d for this study. So that diffuse light attenuation values were based on vertical incidence of light, K d values w ere adjusted to correct for differences in solar elevation (zenith angle of the sun) that we re due to day of year, solar declination, latitude, longitude and time of day For light measurements that were made while solar zenith angles were other than 0 (sun directly overhead), t he potential for scattering and absorption was increased due to the lengthened path length that is traveled by a direct beam of photons through the water. K d values were adjusted to a zenith angle of 0 (vertical zenith angle) to rem ove the effect of the longer path lengt h, thereby allowing for com parisons of measurements across seasons and stations and reduc ing diurnal and regional variation in K d (Kirk 1994) The zenith angle adjusted K d was selected to evaluat e functional relationships between water column attenuators such as IOPs and water chemi stry parameters, and wa s calculated by multiplying measured K d by the cos ine of the zenith angle of water ( Kirk 1994). Statistical Data Analysis All statistical analyses of the data w ere performed using MINITAB statistical software Normality tests were c onducted prior to statistical analysis to meet the assumptions of normality and homogeneity of variance was evaluated by reviewing residual plots. D ata transformations or nonparametric alternatives w ere used if data distributions were non normal or varianc e was h eterogeneous R elationships between

PAGE 43

43 inherent and apparent optical properties and physical and chemical characteristics of each spring system were assessed at the disaggregated individual sample level with Pearson correlation analyses and simple lin ear regressions. All correlation and regression coefficients were considered significant when p values we re less than 0.05. Results Physical Characteristics I n q uiescent aquatic systems such as coastal waterbodies and lakes light availability is control led by water column attenuators, mainly dissolved and particulate absorption and scattering in the water column However, in rivers and streams, in addition to water column attenuators, physical characteristics such as hydrology canopy cover, sediment typ e, and channel geometry can also contribute to increased variability in the light regime Over the one year study period, Rainbow River (RR) s pring and resultant river discharge did not vary seasonally, although discharge varied spatially and increased alm ost four fold with distance from the headsprings area (RR1) to the most downstream station RR8 along with a steady spatial increas e in standard deviation (Table 2 1). Weeki Wachee River (WW) discharge remained relatively constant temporally and had minim al variation along the longitudinal gradient. Flow velocity was highly variable in both RR and WW without a clear spatial trend in either system, but mean overall WW flow velocity was almost twice as high as in RR even though RR overall mean discharge wa s 2.5 fold greater than in WW Rainbow River had a n overall water surface width of 44.4 14.2 m (mean standard deviation) and possessed a longitudinal pattern typical of many lotic system s (Allen and Castillo 2007) Weeki Wachee River was much narrower and less variable than RR with a n average width of 15 2.1 m. Water depths ranged from 1.3 to 2.6 m

PAGE 44

44 (mean of 1.9 m) at RR stations and from 0.9 to 1.3 m (mean of 1.2 m) at WW stations with no consistent spatial trend in either system (Table 2 1). Shading due to riparian c anopy cover was determined at each station on two occasions (July and October) with a forest canopy densitometer and was found to be considerably lower in RR, ranging from 0 to 13% closed canopy (mean of 5%) as compared to a range of 8 t o 85% canopy cover (mean of 25%) in WW. The difference in canopy cover wa s likely due to a lesser channel width and a more variable channel orientation in WW as a result of its greater sinuosity, unlike RR that was wider and had a mostly linear channel wit h minimal sinuosity along the entire river reach. Sediment type, size and composition can introduce variability into absorption and scattering in the water column. Qualitative analysis of sediment type resulted in relatively consistent patterns in both sy stems, where sand dominated in the upper part of the rivers and different variations of sand/silty sand were more prevalent in the lower river reaches It should be noted that limestone rock was prevalent near a few stations in RR where velocity was rapid, but those areas were outside of the sampling regions used for this study. Only one station, RR8 was considered to be siltier than the other stations due to a greater amount of fine grain sizes (silt) in the sediment (Table 2 1). Optical Properties and Wat er Chemistry Water chemistry parameters relevant to light attenuation, i.e. color, turbidity, and chlorophyll a increased with distance from the headspri ngs in both RR and WW (Table 2 2) A significant temporal trend was not found for all parameters, howe ver July had the highest color values in the 5 most downstream stations in Rainbow River (RR3 to RR8). The highest chlorophyll a values occurred in April for the upstream stations RR1 to RR3, and in July for the downstream stations RR4 to RR8. In Weeki Wac hee, color

PAGE 45

45 was highest in October for all stations, and turbidity was highest in July for all but the most downstream station (WW5). No discernible temporal trend was found in WW for Chlorophyll a. All three OWQ parameters: color, turbidity, and chlorophyl l a overall means were greater in WW than in RR by 57%, 36%, and 42%, respectively. Mean (and standard deviation) apparent optical property (AOP) data are summarized in Table 2 3 for both RR and WW Rivers. In both systems, horizontal Secchi distance (HSD), and benthic PAR (normalized for near high noon PAR and corrected for shade) both decreased significantly with distance from the headspring (Tables 2 3, 2 5 and 2 6 ). Benthic PAR and HSD values were temporally variable with no significant pattern for eithe r system. Light attenuation coefficient ( K d ) data measured at three reference wavelengths for the blue, green, and red bands, respectively, and for the PAR spectra are shown in Table 2 3. In Rainbow River, all reported K d coefficients ( K dPAR, K d440, K d550, and K d660 ) we re positively correlated with distance from the headspring, which corresponded to the negative correlation of distance with each of the reference benthic PAR values (data not shown; PAR : R 2 = 0.30, p = 0.003; PAR 440 : R 2 = 0.37 p = 0.001 ; PAR 550 : R 2 = 0.28 p = 0.004 ; PAR 660 : R 2 = 0.15 p = 0.044 ). Negative correlations of distance with benthic PAR were considerably stronger in Weeki Wachee River (data not shown; PAR : R 2 = 0.86, p = 0.000; PAR 440 : R 2 = 0.81, p = 0.000; PAR 550 : R 2 = 0.83, p = 0 .000; PAR 660: R 2 = 0.82, p = 0.000) than in Rainbow River. A ll light attenuation coefficients except for K d 660 (red band) we re significant ly, however more weakly correlated with distance in Weeki Wachee than in Rainbow River (Tables 2 5 and 2 6 ) Mean K d 6 60 values were higher in magnitude when compared to K d 440, and K d 550 for all stations except for the most downstream sta tion in Weeki Wachee

PAGE 46

46 River (WW5; Table 2 3) The higher K d 660 values for all but WW5 were likely due to great er depths and little to no color inputs that would affect K d 440 On the other hand the highest light attenuation occurred at the 440 nm wavelength for WW5, due to adjacent riverine swamps that contributed relatively large amounts of CDOM to the river during and after storm events. Mean spectral diffuse attenuation coefficients at each wavelength over the PAR spectrum K d for Rainbow and Weeki Wachee Rivers at each station are shown in Figures 2 3 and 2 4, respectively, which also include interpolated literature values for pure water attenuation coefficients (Pope and Fry 1997) for comparison. It appears that there are distinct optical regimes in both rivers due to the separation in the data across PAR. There are four breaks in the RR data, which represent an attenuation regime for each break. The most upstream stati ons (RR1, RR2 and RR3) fall into the most optically clear category, followed by stations RR4 and RR5 in the middle part of the river as the second clearest category The third break in the data occurs at RR8, which is actually the most downstream station and has less attenuation than station RR7, which is approximately 1 km upstream of RR8. This is likely due to the contribution of phytoplankton from adjacent phosphate mine borrow pits that are hydraulically connected to the lower part of the river just above station RR7. Weeki Wachee River stations seemed to only have two distinct optical regimes although it could be argued that three regimes may exist The four most upstream stations ( WW0 WW0.5, WW1, and WW3 ) were i n the most optically clear regime, with WW0.5 having the lowest attenuation across the PAR spectra and could be considered its own category. T he

PAGE 47

47 second distinct regime ha d the highest light attenuation especially in the blue band (400 500 nm), and occurre d at the most downstream station WW5 (Figure 2 4). Figure 2 5 shows a comparison of overall average spectral light attenuation coefficients in RR and WW and in pure water. Mean K d values we re over an order of magnitude higher in both RR and WW t han in pure water within the blue and blue green regions Maximum light attenuation occurred in the red band (600 700 nm) in the rivers and in pure water, but K d values in pure water wer e approximately 50% of those in RR and WW Weeki Wachee River had high er mean overall light attenuation than Rainbow River across the PAR spectra, with the difference being most pronounced in the blue range The c alculated mean and standard deviation for i nherent optical properties (PAR spec tra) obtained at each station are given in Table 2 4. Total absorption coefficient (a t PAR) values included the contribution of absorption due to water itself (a w PAR) Similar to how overall K d values were lower in RR th an in WW, all IOP parameters measured in this study were greater in WW than in RR This was evident from the absorption coefficient profiles shown in Figures 2 16 and 2 17 where overall system mean absorption coefficients, total and partial, we re plotted f or the full spect rum of PAR. Figure 2 16 included the contribution of water and showed abs orption due to water itself. Figure 2 17 omitted a w so that the contribution of the other partial absorbing components can be compared on an appropriate scale. Overal l mean RR a t PAR was only 21% lower than in WW even though overall mean WW a g PAR, a d PAR, and a ph PAR were 67%, 74%, and 60% higher than in RR Scattering (bPAR) and beam attenuation

PAGE 48

48 (cPAR) coefficients were also higher in WW than in RR by 37% and 31%, respec tively (Table 2 4) Temporal patterns were evident for a g PAR a d PAR a ph PAR bPAR, and cPAR in Weeki Wachee River for the upstream four stations where each of these parameters were highest in July and April in order of magnitude Of significance, a d PAR was two orders of magnitude greater in July (20.9 m 1 ) than in January (0.2 m 1 ) or October (0.5 m 1 ) in the most upstream station (WW0), which may be due to very high recreational activity at a public water park coupled with high filamentous benthic alga e production and export during the peak growing season in the headsprings area just upstream of the WW0 station contributing to detrital and algal pigment absorption and scattering. Unlike the other four upstream stations, WW5 had maximum detrital and alg al absorption, scattering and beam attenuation coefficients in January Rainbow River was more variable and no significant temporal pattern for any of the IOP coefficients was found, although a d PAR was consistently higher in July for all stations, similar to Weeki Wachee River. Absorption b y detrital and algal pigments increased significantly along the river continuum for all PAR and specific reference wavelengths with the except ion of a ph 550 in Rainbow River and a d660 a ph 44 0 and a ph 550 in Weeki Wachee River (T able 2 5 and Table 2 6 ). S cattering and CDOM absorption coefficients were not significantly related to distance from the headsprings in Rainbow River, but they were significantly correlated in Weeki Wachee River (Table 2 6) Water chemistry param eters were correlated to AOPs and IOPs (Tables 2 5 and 2 6) In Rainbow River, of the beam attenuation, light attenuation and absorption coefficients (not including scattering since bPAR is calculated from turbidity

PAGE 49

49 co n centrations ) turbidity was most stro ngly correlated to beam attenuation at 550 nm ( c550 Figure 2 19), K d 440 and a g34 0 (Tables 2 5 and 2 6), respectively As expected, c hlorophyll a concentrations were most strongly correlated with algal pigment absorption for PAR and at 440 and 660 nm in RR and less strongly correlated to c660 (Figure 2 19) In contrast, none of the light attenuation coefficients were significantly correlated with turbidity in WW and a d 440 was most strongly correlated amongst the IOPs with turbidity (Table 2 6), although c550 was highly correlated (Figure 2 19) Chlorophyll a and color were weakly correlated with c440 and c660, respectively (Figure 2 19) and turbidity and chlorophyll a concentrations we re only weakly significantly correlated with a ph66 0 but not a ph44 0 wh ich suggest s that the algal species composition in the water column and their optical properties in Weeki Wachee River differ ed from that in Rainbow River This could be attributed to absorption by accessory pigments in addition to chlorophyll in the blue band a g 44 0 was not significantly correlated to any water chemistry parameter because of the low magnitude of CDOM absorption at that wavelength in RR, but was correlated with chlorophyll a and color in WW However, CDOM absorption at 340 nm was positivel y correlated w ith turbidity, chlorophyll a, and as expected, most strongly correlated with color in RR (Table 2 5) Color was also significantly, but moderately related to c440 (Figure 2 19). I n WW a g 34 0 was also most strongly correlated with color, and w as weakly correlated with chlorophyll a (Table 2 6) Out of the light attenuation coefficients, color was most strongly (although still weakly) correlated with K d PAR in WW and K d 440 in RR

PAGE 50

50 Horizontal Secchi distance was negatively correlated with turbidit y and chlorophyll a in WW and with chlorophyll a and color in RR When comparing correlation strength between HSD and K d across systems, HSD was only slightly less correlated than the various K d values against the water chemistry parameters in RR However, in WW HSD was more strongly correlated with turbidity and chlorophyll a than K d but not with color (Tables 2 5 and 2 6). Percent open canopy was also evaluated for corre lations with light attenuation, absorption and scattering coeffici ents in both syste ms, data were not shown since none wer e significant in RR However, percent open canopy was significantly negatively related in WW with all K d values and all absorption coefficients except for a ph PAR a ph 44 0 and a ph 550 (Table 2 6). Correlation results between AOPs and IOPs measured at reference wavelengths and for the PAR spectra are shown in Table s 2 7 and 2 8 for Rainbow and Weeki Wachee Rivers, respectively. Correlation results only included bPAR for the scattering parameter because the correlation s tatistics were the same for all reference wavelengt hs and bPAR for all correlation result s shown in Tables 2 5, 2 6, 2 7 and 2 8. I t was surprising how many non signific an t correlation s were found for both systems, especially for WW. In RR, correlation co efficients between AOPs and IOPs were stronger than in WW overall There were only a handful of significant albeit weak correlations in WW, which occurred between K d PAR, K d 440 and a t 440, a t 550, a g 340, a d PAR, a d 440, and a d 660. K d 550 wa s correlated with a t 44 0, a d PAR, a d 440, a t 550, and a g 340. Beam attenuation ( c PAR) and light attenuation ( K d PAR) were not significantly correlated in WW (Figure 2 21), which was due to the strong influence of high color and low turbidity at WW5 in the third quarter sample. O f not e is the significant but weak

PAGE 51

51 correlation between K d 660 and a g 340. This suggests that light attenuation could largely be attributed to CDOM and detrital absorption, but not due to direct algal pigment absorption or particulate scattering in WW (Table 2 7) K d PAR was weakly related to cPAR (Figure 2 21), and most strongly correlated with a t 660 in Rainbow River Out of the partial absorption coefficients, the K d values listed in Table 2 7 were not significantly related with only two absorption coefficients, a g 440, and a ph 550. K d 550 and K d 660 were also not correlated with scattering, which was an unexpected result as scattering is usually most strongly correlated with activity associated with detrital and algal material that absorb and scatter in the green an d red region The strongest correlation s between partial absorption coefficients and K d 440, K d 550, and K d 660 were with a d 660 a ph PAR, and a ph PAR, respectively This is atypical of most light attenuation relationships where the absorption coefficient of a s pecific wavelength is usually most strongly related to the corresponding light attenuation coefficient and suggests a more complex optical environment than in clear lotic systems. Similar to the correlation with water chemistry parameters, HSD was correla ted with many absorption coefficients at nearly the same strength as the light attenuation coefficients for both systems, with it being slightly stronger in certain correlations and weaker in others. However, HSD was not correlated with CDOM absorption or scattering in RR, and it was also not correlated to a d 660, a ph 440 or a ph 550 in WW (Tables 2 7 and 2 8 respectively ). Regardless of the few non signif i cant relationships, HSD may have utility as a valid AOP in clear uncolored systems.

PAGE 52

52 Figures 2 6 throug h 2 13 present the mean relative percent contribution of partial absorption coefficients for PAR and three reference wavelengths (440, 550 and 660 nm) to total absorption comparing each with (a t ) and without (a t w ) the contribution of absorption due to wat er itself at each station in both systems When absorption due to water wa s removed (a t w ) from the calculation of a t a g a d and a ph remain ed In Rainbow River, when water wa s included in the calculation of total absorption, a w contribute d the largest fr action to a t PAR a t 550 and a t 660 (Figures 2 6, 2 8 and 2 9) This was in contrast to a t 440, where a w 440 contributed the least among the four coefficients with a g 340 contribut ing the largest percentage ( ca. 40 60%) across all sites in RR except for station RR5 (Figure 2 7 bottom panel ) Detrital absorption dominated a t w for PAR and at 550 nm, but algal pigment s contributed more to total absorption at 660 nm The statistically significant correlations between individual reference absorption coefficients and total absorption (minus water) are shown in Figure 2 18 with the strongest relationship with a d 550, and then in order of decreasing strength a ph 440, a ph 660 and a g 340. Weeki Wachee River had a similar trend as Rainbow River where absorption due to water d ominated most of the upstream sites in all absorption coefficients except for at 440 nm (Figures 2 10, 2 12 and 2 13). However, at 440 nm, total absorption (with and without water) was d u e to a more equal contribution from CDOM and detrital absorption (Fig ure 2 11). Again, detrital absorption is responsible for the majority of a t w for PAR and at 550 nm bu t detrital and algal pigment absorption were relatively equal at 660 nm (Figure 2 13). In Weeki Wachee, the strongest predictor of total absorption (min us water contribution) was with a d 550, followed by a ph 660 and finally with a g 340, a ph 440

PAGE 53

53 was not found to be significantly related to total absorption over the PAR spectra ( F igure 2 18). The relationship between total absorption (a t PAR) and scattering (bP AR) was examined along the river with distance downstream f or both systems (Figure 2 14). a t PAR increased steadily in Rainbow River, but the more variable bPAR peaked approximately midway down the river and was considerably higher in magnitude than absorpt ion until the lowest reach of the river where absorption exceeded scattering T he peak mid river could be due to maximal measured velocities in this area due to shallower depths (Table 2 1) or could also be due to disturbance of bottom sediments from recre ational motor boats or tubers causing resuspension of particulates into the water column (Mumma et al 1996) Weeki Wachee River had increasing trends in both a t PAR and bPAR with distance from the headspring but the magnitude of absorption was larger than scattering overall (Figure 2 14). The scattering to absorption ratio (b:aPAR) was calculated to assess how these optical properties change d together in relation to distance from the headspring and to compare the two systems (Figure 2 15). For the mo st part, Weeki Wachee and Rainbow Rivers both had greater scattering than absorption (i.e., b:aPAR>1) along the river continuum In Weeki Wachee, b:aPAR general ly increase d with distance suggesting that scattering was a larger contributor to overall beam attenuation (c) In Rainbow River, b:aPAR began balanced, and then increased and decreased at different points in the river Correlations of absorption and scattering with beam attenuation are shown in Figure 20. Scattering explains 57% and 39% more beam a ttenuation variability than

PAGE 54

54 absorption in Rainbow and Weeki Wachee Rivers, respectively. The results support that scattering overpower s beam attenuation overall Discussion Because apparent (HSD and K d ) and inherent (a,b, and c) optical properties chara cterize optical water quality (Kirk 1988), it is important to establish optical water quality sampling programs to further the understanding of underwater light environment in Florida springs. This was the first study to assess optical water quality and to evaluate the effect of the magnitude and relative contribution of spectrally specific optical properties and other physical factors on light availability in Florida springs Spatial and temporal patterns were examined to determine what controlled the vari ability of the underwater light environment in springs. Temporal variability (on the inter annual scale) of inherent optical properties due to storm events leading to high flow discharge was not evident in this study as was suggested by others (Duarte and Canfield 1990 ; Davies Colley and N agels 2008, Julian et al. 2008 a 2008b) but was consistent with the notion that Florida springs do not vary considerably on the seasonal scale (Odum 195 7a ) and are functioning in a stead y state (Duarte et al. 2010) How ever, t he majority of optical properties that were examined spatially in the two spring fed rivers in this study displayed a spatial optical gradient within each system Longitudinal patterns are typical for many river processes and sometimes partition the longitudinal profile of a river into three zones that include the erosion zone, the transfer zone and the sediment deposition zone ( Allen and Castillo 2007). than Weeki Wachee and pr ovides a basis for understanding the optical gradient in these

PAGE 55

55 two spring fed rivers. T he strong covariation with distance across many other IOPs provides compelling evidence for longitudinal accrual processes. CDOM absorption and scattering coefficients did not have a similar longitudinal pattern as the other IOPs in Rainbow River, however these two variables were positively significantly related ( R 2 = 0.52, p = 0.000 ). Along the se same lines, chlorophyll a ( R 2 = 0.24, p = 0.008 ) and turbidity (R 2 = 0.52 p = 0.000 ) concentrations were also significantly related to CDOM absorption (340 nm) and by extension, suggests that scattering due to turbidity and chlorophyll a could be the result of biodegrading autotrophic material, likely from senescence and sloug hing of epiphy tic algal diatoms in rapid ly flow ing areas and also from SAV. T he non significant relationship of CDOM with distance downstream in Rainbow River is inconsistent with results from another study that found DOC (proxy for CDOM) rapidly increasin g longitudinally from the headsprings down due to autochthonous sources attributed mainly to SAV (Duarte et al. 2010), although the finding did correspond to results from Weeki Wachee River from this study The consistent accrual /depletion pattern of a g 340 that was found in the upper reaches of both spring systems could be due to several biogeochemical factors. In situ a utotrophic biomass production driven by abundant nutrient and irradiance conditions (Matheson et al. 2012) and allochthonous inputs determine the magnitude and quality of CDOM accrual ( Laurion et al. 2000; Yamashita et al. 2013) The supply of CDOM is then acted upon by biogeochemical processes that biodegrade CDOM or photochemically induce shifts in molecular weig ht of CDOM during p hotobleaching ( Laurion et al. 2000; Margager and Vincent, 2000; Helms et al. 2008 ; Yamashita et al.

PAGE 56

56 2013 ) that occurs under high light (i.e. open canopy areas low light attenuation conditions) and also when CDOM is used in the denitrifi cation process by denitrifying bacteria. Photoautotrophs such as algae and SAV provide an important labile source of carbon and abundant growth of these autotrophs at high irradiances can encourage the development of enhanced heterotrophic biomass and met abolism ( Heffernan and Cohen 2010; Heffernan et al. 2010b; Matheson et al. 2012). In addition to these processes, accrual of CDOM is also due to contributions from allochthonous sources such as riverine floodplain wetlands that surround the Weeki Wachee Ri ver basin. The inference that Weeki Wachee River is more CDOM rich than the CDOM poor Rainbow River can be ex plained by greater canopy cover, thereby reducing the potential for photobleaching and direct influence from c onnectivity with adjacent wetlands. The spectral slope (S) of the regression between CDOM and wavelengths in the 200 500 nm region (S 200 500 ) and the slope ratios ( S R275 295:350 450 ) were used to evaluate the source and quality of CDOM in both systems. The inference that lower molecular weig ht CDOM is due to autotrophic material such as algae and SAV in Rainbow River is supported by the significant relationships between S 200 500 and chlorophyll a ( R 2 = 0.21, p = 0. 0 15,) and turbidity and (R 2 = 0.19 p = 0.019 ) which is highly correlated wit h chlorophyll a Albeit weak, but otherwise important, these relationships are slightly stronger than the relationship with color ( R 2 = 0.17, p = 0.03 ) and the relationship is non significant with a g 440 ( R 2 = 0.007, p =0.67 ), which provides more eviden ce t hat the CDOM source consists of lower molecular weight CDOM in Rainbow River

PAGE 57

57 On the contrary to Rainbow River Weeki Wachee River S 200 500 was most strongly related to color ( R 2 = 0.38, p = 0.003 ) and a g 440 (R 2 = 0.40 p = 0.003 ) and not significantly re lated to chlorophyll a ( R 2 = 0.16, p = 0.08 ) or turbidity ( R 2 = 0.05, p = 0.35,) supporting the assertion that Weeki Wachee River CDOM sources are dominated mostly by terrestrial inputs and not entirely derived from within the spring system as was the case in Rainbow River. The spectral slope ratio S R275 295:350 450 was not significantly related to any of the before mentioned variables (data not shown) although the values we re similar in magnitude that have been reported by other studies in very clear oce an waters and mountain lakes that do not receive terrestrially sourced CDOM rich inputs ( Laurion et al. 2000; Helms et al. 2008). There is a great deal of research that has focused on absorption and scattering characteristics in marine, estuarine and lake systems ( Kirk 1980, 1981, 1984, 1991, 1994; Gallegos 1993, 1994, 2001, 2005; Gallegos and Kenworthy 1996; Biber et al. 2008, Gallegos et al. 1990; Gallegos et al. 2005; Gallegos et al. 2009 ; Perez et al., 2010 ). Riverine optical water quality constituents (IOPs and AOPs ) were quantified in two Midwest ern USA rivers (Julian et al. 2008a) and in over a hundred optically diverse rivers in New Zealand ( Davies Colley and Close 1990; Davies Colley and Nagels 2008). In the Midwestern USA rivers, large fluctuation s in water chemistry was attributed to major tributary inputs and impoundments along the river continuum, along with highly irregular scattering to absorption ratios (b:a) caused by changes in composition of suspended particulate matter (Julian et al. 2008 a) The variable b:a values in the Midwestern USA rivers correspond ed to the variable b:a pattern in Rainbow River in this study. H owever, the source and transformation of the suspended particulate matter was

PAGE 58

58 not contributed by external inputs into Rainbow River rather from within the river itself The global trend of particulates dominating beam attenuation found in the Midwestern USA and New Zealand river studies were corroborated by results from th e spring fed rivers in this study However, the temporal trends due to fluctuation in the hydrologic regime in the other studies was not observed in this study, since the hydrologic regime did not vary temporally in the spring fed rivers ( Davies Colley and Nagels 2008; Julian et al. 2008a) I n springs, only one other study measured total light absorption coefficients at the spring boils of 17 Florida spring systems that is available for comparison to results of this study (Duarte and Canfield 1990). Duarte and Canfield light absorption coefficients ranged f rom 0.038 (Weeki Wachee Springs) to 1.144 m 1 (Old Town Springs), and Rainbow Springs averaged 0.066 m 1 for two sites, but most of the springs had absorption coefficients less than 0.3 m 1 (1990). Duarte and Canfield (1990) did not determine the relativ e importance of each of the absorbing components (a g a ph a d ) on light attenuation in their springs study. Unfortunately they did not provide the details regarding the scale of absorption coefficient, meaning that it is not entirely clear if they were rep orting total absorption (minus the contribution due to water) integrated for PAR, average absorption over PAR, or at one particular wavelength. However, based on the magnitude of the data values provided in Table 1 of their study, it was assumed that they were reporting total absorption averaged for PAR (Duarte and Canfield 1990). Using the assumption that they were reporting PAR averaged total absorption results, Weeki Wachee and Rainbow River s total absorption may have inc reased by

PAGE 59

59 83% and 62% respectively. However, if they were reporting total absorption at 440 nm, may have increased by 77% and 16%, respectively. Either way, at least part of the increase in absorption may be attributed to increased recreation and other human disturbances in both systems. Depending on the method used to calculate total absorption, the previous study provided a rough baseline estimate of total absorption and shows that both spring system s have likely degraded in terms of absorption coefficients As a point of reference, the absorption coefficient s of pure water range from 0.007 to 0.02 0 m 1 within the 400 to 500 nm (blue) region of the absorption spectra (Pope and Fry 1997 ; Kirk 1994). I n contrast, the total absorption of a blackwater river, Lower St. Johns River, on the east coast of Florida was estimated at 30 m 1 with a g 440 being the dominant contributor to the total absorption coefficient, and among the highest values reported for n atural waters (Gallegos 2005). L ight attenuation coefficients K d were correlated with in situ Horizontal Secchi Distance (HSD) measurements and laboratory measured optical properties of the water samples. Light attenuation coefficients are typically more strongly correlated to IOPs in the literature, and results from this study suggest that spring s systems must contain other factors that affect the light environment in addition to the IOPs that control the optical environment. The other factors that could be causing a reduction in predictive power between AOPs and IOPs are likely due to scattering of light caused by turbulence in the upper part of the water column and backscatter due to reflectance from sand, SAV or some other type of difference s in sedimen t albedo that influence optical properties (Boss and Zaneveld 2003) and are an artifact of the method to

PAGE 60

60 measure a light attenuation profile and independent of the method to collect IOPs by collecting surface water sample s well above the bottom substrate In contrast to the trend where visual clarity (measured by HSD) was greatest in the most upstream station near the headsprings area in both systems, light attenuation was actually lowest in the second most upstream s tations (RR2 and WW0.5) and more closel y approximated the pure water attenuation spectra. For a point of reference, each of these stations was approximately 800 m downstream of the headspring areas. This disagreement in AOP results could be attributed to the complexity of the absorption and sca ttering properties of suspended particles such as backscattering and bottom reflectance effects and the limitation of the HSD method to detect changes in light regions outside of the peak sensitivity of the human eye, which is around 550 nm in the green ba nd of the PAR spectrum (Davies Colley 1995). Furthermore, mean turbidity was lower at RR2 than at RR1, and could explain the slight disagreement in AOP results, but this was not the case in WW. ermined an average 63 m horizontal black body visibility distance, which translates to 72 m for a vertical black and white visibility distance, for an ave rage of 67 m for both methods (Davies Colley and Smith 1995). They argued that the visual cla rity ranks among the very clearest in the world and is vir tually indistinguishable from pure water, which has a theoretical visibility between 74 and 80 m for the horizontal black body and vertical black and white body distances, respectively, for an avera ge of 77 m ( Davies Colley and Smith 1995).

PAGE 61

61 Even thoug h v isual clarity methodologies differed slightly between methods used in Pupu Springs and in this study average visual clarity in Pupu Springs wa s not different from visual clarity measured in Rain bow Springs (Station RR1 this study ) where the average measured HSD for RR1 was also 67 m. Th is would suggest that visual clarity in Rainbow S prings rivals the clearest waters of the world, with the caveat that visual clarity can best estimate only the g reen band (i.e. 550 nm) of the PAR spectra In contrast, when average beam attenuation coefficients (c ) at 493 and 528 nm wavelengths were compared for Pupu Springs (0.022 and 0.075 m 1 respectively) and Rainbow Springs ( this study: 0.207 and 0.065 m 1 respective ly), the c(493) value in Rainbow Springs wa s an order of magnitude higher than in Pupu Springs, however, the c(528) value in Rainbow Springs w as only 13% higher than in Pupu Springs These inconsistencies are likely due to differences in sus pended particulates that contribute var iability to absorption and scattering for the selected wavelengths (1995). These results highlight the necessity of evaluating light regimes and how they have changed in spring systems over a period of time longer tha n one year. Also, since light availability is a major control on primary productivity in springs, an evaluation of functional relationships between primary productivity and the light available to primary producers is also needed. Results from this study wi ll be used later to determine the effect of light availability on the composition, distribution and productivity of the primary producer communities in springs (Chapter 4).

PAGE 62

62 Figure 2 1. Rainbow River study reaches. Note: D irection of flow i s indicated on the figure. Sampling stations are in bold and are de noted by circles with an X inside Sampling segments are in italics and denoted with a bi directional arrow. The confluence with the Withlacoochee River is denoted by a star symbol. Rainbow River, FL, USA RRS1 RRS6 RRS5 RRS4 RRS3 RRS2

PAGE 63

63 Figure 2 2. Weeki Wachee River study reaches. Note: The river flows in a westerly fashion and ultimately empties into the Gulf of Mexico. Direction of flow is indicated on the figure. Sampling stations are in bold and are denoted by circles with an X inside. Sampling segments are in italics and denoted with a bi directional arrow. Weeki Wachee River, FL, USA WW S4 WW S 3 WW S 1 WW S 2

PAGE 64

64 Figure 2 3. Mean spectral diffuse attenuation coefficients, K d of Rainbow River (RR) stations obtained January to November 2011. Note: The K d values for pur e water derived by Pope and Fry (1997) are shown for comparison with river stations.

PAGE 65

65 Figure 2 4. Mean spectral diffuse attenuation coefficients, K d of Weeki Wachee River (WW) stations obtained January to November 2011. Note: The K d values for pure water derived by Pope and Fry (1997) are shown for comparison with river stations.

PAGE 66

66 Figure 2 5. Overall river average spectral diffuse attenuation coefficients, K d of Rainbow (RR) and Weeki Wachee (WW) Rivers obtained January to November 2011. Note: The K d values for pure water derived by Pope and Fry (1997) are shown for comparison with river stations.

PAGE 67

67 Figure 2 6. Mean relative % contribution of absorption coefficients (PAR) to total absorption (PAR), with (top panel) and without (bottom panel) absorption due to water itself in Rainbow River (RR).

PAGE 68

68 Figure 2 7. Mean relative % contribution of absorption coefficients (440 nm) to total absorption (440 nm), with (top panel) and without (bottom panel) absorption due to water itsel f in Rainbow River (RR).

PAGE 69

69 Figure 2 8. Mean relative % contribution of absorption coefficients (550 nm) to total absorption (550 nm), with (top panel) and without (bottom panel) absorption due to water itself in Rainbow River (RR).

PAGE 70

70 Figure 2 9. M ean relative % contribution of absorption coefficients (660 nm) to total absorption (660 nm), with (top panel) and without (bottom panel) absorption due to water itself in Rainbow River (RR).

PAGE 71

71 Figure 2 10. Mean relative % contribution of absorption co efficients (PAR) to total absorption (PAR), with (top panel) and without (bottom panel) absorption due to water itself in Weeki Wachee River (WW).

PAGE 72

72 Figure 2 11. Mean relative % contribution of absorption coefficients (440) to total absorption (440), w ith (top panel) and without (bottom panel) absorption due to water itself in Weeki Wachee River (WW).

PAGE 73

73 Figure 2 12. Mean relative % contribution of absorption coefficients (550) to total absorption (550), with (top panel) and without (bottom panel) a bsorption due to water itself in Weeki Wachee River (WW).

PAGE 74

74 Figure 2 13. Mean relative % contribution of absorption coefficients (660) to total absorption (660), with (top panel) and without (bottom panel) absorption due to water itself in Weeki Wache e River (WW).

PAGE 75

75 Figure 2 14. Average scattering (bPAR) and absorption (a t PAR) coefficients in relation to distance from the headsprings in A) Rainbow and B) Weeki Wachee Rivers. A B

PAGE 76

76 Figure 2 15. Ratio of scattering and absorption coefficients (b:a P AR) over the PAR spectrum in relation to distance from the headspring in Weeki Wachee and Rainbow Rivers.

PAGE 77

77 Figure 2 16. Absorption profiles of total (a p ), detri t al (a d ) and algal particulates (a ph ) ; CDOM (a g ) water (a w ) and total absorption with (a t ) and without (a t w ) water. Rainbow River Weeki Wachee River

PAGE 78

78 Figure 2 17. Absorption profiles of total (a p ), detrial (a d ) and algal particulates (a ph ); CDOM (a g ), water (a w ), and total absorption without (a t w ) water. Rainbow River Weeki Wachee River

PAGE 79

79 Figure 2 18. Correlations among individual absorpt ion coefficients at reference wavelengths and total absorption for A) Rainbow and B) Weeki Wachee Rivers. A B

PAGE 80

80 Figure 2 19. Correlations between water chemistry concentrations and beam attenuation (c) coefficients in A) Rainbow and B) Weeki Wach ee Rivers, ( p < 0.05 for all). R 2 = 0.48 R 2 = 0.99 R 2 = 0.44 R 2 = 0.25 R 2 = 0.99 R 2 = 0.31 A B

PAGE 81

81 Figure 2 20. Correlations between absorption (a), scatter (b) and beam attenuation (c) coefficients in A) Rainbow and B) Weeki Wachee Rivers. R 2 = 0.60 p =0.000 R 2 = 0.99 p =0.000 R 2 = 0.42 p =0.000 R 2 = 0.99 p =0.000 B A

PAGE 82

82 Figure 2 21. Correlations between attenuation (K d ), and beam att enuatio n (c) coefficients in A) Weeki Wachee and B) Rainbow Rivers. y = 0.345 + 0.00056x R 2 = 0.043, p = 0.378 y = 0.238 + 0.00109x R 2 = 0.154, p = 0.022 A B

PAGE 83

83 Table 2 1. Summary of mean physical characteristics recorded during sampling events at Rainbow (RR) and Weeki Wachee Rivers (WW). Station Distance From Headspring (m) Depth (m) Chann el width (m) Canopy % Cover Sediment Type Mean Annual Discharge (m 3 /s)* Flow Velocity (cm/s) RR1 85 2.6 51.43 6 Sand/ Si l t y Sand 3.59 0.19 12.10 0.00 RR2 785 1.5 31.55 1 Sand 6.80 0.36 16.27 0.00 RR3 1535 1.9 23.50 3 Sand 8.67 0.46 18.33 0.00 RR4 4030 1.3 44.02 13 Sand 12.67 0.67 26.80 0.00 RR5 6840 2.1 45.51 6 S ilt y Sand 14.08 0.75 18.17 0.00 RR7 7720 1.8 46.90 7 Silt y Sand 14.46 0.77 19.91 0.00 RR8 8570 2.1 67.80 0 Silt 13.57 0.72 14.30 0.00 WW0 136 0.9 18.51 9 Sand 4.13 0.47 38.02 10.51 WW0.5 8 50 1.2 13.63 8 Sand 4.13 0.47 31.15 2.46 WW1 1720 1.2 13.42 13 Sand/ Silt y Sand 4.13 0.47 35.83 8.61 WW3 4885 1.3 14.11 11 Sand/ Silt y Sand 4.19 0.55 34.19 8.61 WW5 8160 1.2 15.14 85 Silt y Sand 4.24 0.65 23.74 7.41 Note: *Mean daily flow discharge data for the water quality sampling dates was acquired from the United States Geological Survey (USGS) website for one station on each river. A fraction coefficient was calculated based on actual flow measurements at each station and at the USGS gaging station. The fraction coefficient was applied to the flow data to estimate flow and velocity at each station as related to the flow at the USGS gaging station. These data were averaged for the year for an annual average.

PAGE 84

84 Table 2 2 Mean (SD) of water chemi stry parameter concentrations collected in January, April, July and October 2011. Station Color (PCU) Turbidity (NTU) Chlorophyll a (g L 1 ) RR1 0.13 0.05 0.16 0.11 0.10 0.03 RR2 0.28 0.13 0.13 0.02 0.18 0.03 RR3 0.38 0.15 0.17 0.04 0.2 6 0.04 RR4 0.90 0.87 0.35 0.15 0.40 0.13 RR5 0.65 0.17 0.17 0.12 0.39 0.10 RR7 0.80 0.27 0.26 0.20 0.93 0.57 RR8 0.90 0.27 0.30 0.19 0.86 0.41 WW0 0.28 0.05 0.21 0.28 0.22 0.18 WW0.5 0.95 1.11 0.31 0.18 0.97 0.50 WW1 0.70 0.41 0.30 0.21 0.73 0.21 WW3 1.23 0.83 0.29 0.20 0.61 0.33 WW5 3.65 2.66 0.62 0.30 1.31 0.17

PAGE 85

85 Table 2 3 Mean values (SD) of app arent optical properties (AOPs). Station HSD (m) Benthic PAR (mol m 2 s 1 ) K d PAR (m 1 ) K d 440 (m 1 ) K d 550 (m 1 ) K d 660 (m 1 ) RR1 66.83 10.12 1021.45 301.09 0.28 0.08 0.16 0.12 0.19 0.08 0.49 0.10 RR2 34.64 9.25 1001.56 121.02 0.26 0.03 0.15 0.01 0.16 0.03 0.50 0.05 RR3 24.27 3.60 1038.64 215.97 0.30 0.11 0.20 0.12 0.19 0.10 0.51 0.12 RR4 20.39 4.17 1074.61 210.03 0.37 0.10 0.30 0.14 0.25 0.09 0.57 0.09 RR5 17.85 5.05 750.49 172.27 0.37 0.06 0.29 0.10 0.25 0.06 0.57 0.06 RR7 13.59 6.21 634.00 201.12 0.54 0.08 0.47 0.13 0.42 0.07 0.75 0.07 RR8 11.21 5.10 734.13 368.96 0.46 0.02 0.37 0.03 0.35 0.02 0.68 0.03 WW0 39.97 16.72 1503.62 117.29 0.34 0.09 0.24 0.10 0.24 0.07 0.56 0.11 WW0.5 14.09 4.01 1453.66 101.08 0.29 0.05 0.16 0.12 0.19 0.03 0.52 0.09 WW1 13.49 2.86 1343.32 266.18 0.41 0.13 0.30 0.17 0.30 0.10 0.64 0.13 WW3 11.65 2.81 1015.95 264.97 0.41 0.09 0.39 0.15 0.28 0.09 0.58 0.07 WW5 7.90 2.64 187.99 69.82 0.76 0.38 1. 04 0.75 0.56 0.27 0.74 0.25 Note: HSD, h o rizontal secchi distance; benthic PAR was normalized for high noon PAR and corrected shade to allow for comparison among stations and varying ambient light conditions ; K d downwelling diffuse attenuation coe fficients at different wavelengths (PAR, 440, 550, 660 nm); d ata was recorded within 10 minutes of water qu ality samples shown in Table 2 2

PAGE 86

86 Table 2 4. Mean values (SD) of inherent optical properties (IOPs) Station a t PAR (m 1 ) a g PAR (m 1 ) a d PAR (m 1 ) a ph PAR (m 1 ) b PAR (m 1 ) c PAR (m 1 ) RR1 47.46 0.38 0.37 0.45 0.89 0.70 0.74 0.30 49.65 33.88 97.19 33.79 RR2 47.99 0.99 0.13 0.00 1.36 0.83 0.73 0.24 39.15 5.27 87.14 4.97 RR3 49.95 0.96 0.68 0.52 2.77 0.26 1.00 0.51 51.81 13.67 101.76 13.20 RR4 51.23 2.32 0.54 0.70 2.79 1.44 2.41 1.84 106.87 46.04 158.10 47.05 RR5 54.17 2.27 0.60 0.43 6.42 1.96 1.65 0.95 51.66 38.03 105.83 38.52 RR7 55.67 5.58 0.81 0.80 4.79 3.33 4 .58 1.81 80.77 61.10 136.44 66.64 RR8 55.13 4.64 0.74 0.51 4.85 2.62 4.04 2.08 92.66 57.62 147.79 61.80 WW0 55.28 9.99 0.22 0.45 7.62 9.72 1.90 0.25 65.63 87.31 120.91 96.12 WW0.5 61.49 6.22 2.42 4.21 9.54 5.00 3.99 0.92 95.13 54.73 156.62 55.05 WW1 59.22 0.96 0.59 0.56 8.36 2.24 4.73 1.83 93.90 65.76 153.11 66.11 WW3 68.21 8.36 1.27 1.07 13.31 7.74 8.11 2.27 90.65 61.35 158.87 65.43 WW5 83.28 8.82 3.88 2.59 25.97 5.70 7.90 6.56 189.95 93.03 273.24 101.22 Note: a t PAR, total ab sorption coefficient ; a g PAR, colored d issolved organic matter (CDOM) absorption coefficient ; a d PAR, detrital non alg al particulate absorption coefficient; a ph PAR algal pigment absorption coeffici ent ; b PAR scattering coefficient; c PAR beam attenuation coefficient ; IOP samples were collected concurrently with water chemistry samples shown in Table 2 2.

PAGE 87

87 Table 2 5. Pearson correlation results between water chemistry parameters (Turbid ity, Chlorophyll a and Color), %Open Canopy, Distance from Spring and AOPs and IOPs for Rainbow River sites. Turbidity (NTU) Chl a ( g L 1 ) Color (PCU) Distance (m) R p value R p value R p value R p value AOPs K d PAR 0.397 0.036 0.699 0.000 0.645 0.000 0.737 0.000 K d 440 0.485 0.009 0.703 0.000 0.720 0.000 0.699 0.000 K d 550 0.344 0.073 0.675 0.000 0.597 0.001 0.733 0.000 K d 660 0.299 0.122 0.631 0.000 0.533 0.004 0.706 0.000 HSD 0.354 0.065 0.633 0.000 0.555 0.002 0.757 0.000 IOPs a t PAR 0.612 0.001 0.897 0.000 0.578 0.001 0.753 0.000 a t 440 0.715 0.000 0.730 0.000 0.798 0.000 0.613 0.001 a t 550 0.492 0.008 0.763 0.000 0.589 0.001 0.800 0.000 a t 660 0.638 0.000 0.867 0.000 0.681 0.000 0.783 0.000 a g 340 0.718 0.000 0.491 0 .008 0.836 0.000 0.304 0.116 a g 440 0.045 0.822 0.074 0.709 0.001 0.997 0.174 0.377 a d PAR 0.490 0.008 0.711 0.000 0.587 0.001 0.676 0.000 a d 440 0.433 0.021 0.630 0.000 0.576 0.001 0.574 0.001 a d 550 0.427 0.024 0.676 0.000 0.548 0.003 0.688 0.0 00 a d 660 0.531 0.004 0.619 0.000 0.583 0.001 0.463 0.013 a ph PAR 0.539 0.003 0.881 0.000 0.347 0.070 0.687 0.000 a ph 440 0.511 0.005 0.793 0.000 0.393 0.038 0.730 0.000 a ph 550 0.206 0.292 0.296 0.126 0.179 0.361 0.348 0.069 a ph 660 0.506 0.00 6 0.771 0.000 0.548 0.003 0.770 0.000 b PAR 1.000 0.645 0.000 0.677 0.000 0.327 0.090

PAGE 88

88 Table 2 6. Pearson correlation results between water chemistry parameters (Turbidity, Chlorophyll a and Color), %Open Canopy, Distance from Spring and AOPs and IO Ps for Weeki Wachee River sites. Turbidity (NTU) Chl a ( g L 1 ) Color (PCU) %Open Canopy Distance (m) R p value R p value R p value R p value R p value AOPs K d PAR ( m 1 ) 0.168 0.480 0.416 0.068 0.537 0.015 0.685 0.001 0.640 0.002 K d 440 ( m 1 ) 0.134 0.574 0.394 0.085 0.495 0.027 0.700 0.001 0.662 0.001 K d 550 ( m 1 ) 0.212 0.370 0.389 0.085 0.519 0.019 0.698 0.001 0.637 0.003 K d 660 ( m 1 ) 0.167 0.482 0.391 0.088 0.510 0.022 0.459 0.042 0.418 0.067 HSD (m) 0.540 0.014 0.604 0.005 0.382 0.096 0.372 0.106 0.562 0.010 IOPs a t PAR ( m 1 ) 0.725 0.000 0.671 0.001 0.510 0.012 0.755 0.000 0.809 0.000 a t 440 ( m 1 ) 0.645 0.002 0.667 0.001 0.869 0.000 0.836 0.001 0.827 0.000 a t 550 ( m 1 ) 0. 6 05 0.005 0.717 0.000 0.531 0.016 0.773 0.001 0.760 0.000 a t 660 ( m 1 ) 0.628 0.003 0.690 0.001 0.331 0.153 0.459 0.042 0.634 0.003 a g 340 ( m 1 ) 0.331 0.153 0.558 0.011 0.995 0.000 0.653 0.002 0.619 0.004 a g 440 ( m 1 ) 0.266 0.258 0.473 0.035 0.915 0.000 0.464 0.039 0.467 0.038 a d PAR ( m 1 ) 0.729 0.000 0.610 0.004 0.390 0.089 0.734 0.000 0.728 0.038 a d 440 ( m 1 ) 0.771 0.000 0.596 0.006 0.435 0.055 0.724 0.000 0.726 0.000 a d 550 ( m 1 ) 0.690 0.001 0.564 0.010 0.345 0.136 0.693 0.001 0.714 0.000 a d 660 ( m 1 ) 0.563 0.010 0.564 0.015 0.277 0.238 0.540 0.014 0.431 0.058 a ph PAR ( m 1 ) 0.482 0.032 0.367 0.112 0.294 0.208 0.364 0.114 0.583 0.007 a ph 440 ( m 1 ) 0.225 0.340 0.072 0.764 0.021 0.931 0.216 0.361 0.255 0.277 a ph 550 ( m 1 ) 0.105 0.659 0.196 0.408 0.152 0.522 0.087 0.716 0.23 8 0.312 a ph 660 ( m 1 ) 0.480 0.032 0.542 0.014 0.199 0.401 0.504 0.023 0.655 0.002 b PAR ( m 1 ) 1 0.559 0.010 0.352 0.128 0.540 0.014 0.485 0.030

PAGE 89

89 Table 2 7. Pearson correlation results between AOPs and IOPs for Rainbow River sites. Apparent O ptical Property (AOP) K d PAR K d 440 K d 550 K d 660 HSD IOP R P value R P value R P value R P value R P value a t PAR 0.627 0.000 0.666 0.000 0.590 0.001 0.533 0.004 0.695 0.000 a t 440 0.589 0.001 0.699 0.000 0.534 0.003 0.438 0.020 0.564 0.002 a t 550 0.616 0.000 0.644 0.000 0.579 0.001 0.540 0.003 0.742 0.000 a t 660 0.663 0.000 0.698 0.000 0.625 0.000 0.575 0.001 0.682 0.000 a g 340 0.461 0.014 0.577 0.001 0.405 0.033 0.320 0.097 0.323 0.094 a g 440 0.032 0.871 0.011 0.957 0.048 0.807 0.047 0.811 0.116 0.555 a d PAR 0.534 0.003 0.602 0.001 0.484 0.009 0.432 0.022 0.655 0.000 a d 440 0.467 0.012 0.555 0.002 0.419 0.026 0.351 0.067 0.572 0.001 a d 550 0.550 0.002 0.596 0.001 0.501 0.007 0.472 0.011 0.638 0.000 a d 660 0.558 0.002 0.638 0 .000 0.499 0.007 0.440 0.019 0.470 0.012 a ph PAR 0.578 0.001 0.538 0.003 0.576 0.001 0.552 0.002 0.555 0.002 a ph 440 0.540 0.003 0.585 0.001 0.517 0.005 0.445 0.018 0.570 0.002 a ph 550 0.229 0.240 0.203 0.300 0.247 0.205 0.221 0.259 0.323 0.094 a ph 660 0.563 0.002 0.550 0.002 0.550 0.002 0.525 0.004 0.624 0.000 b PAR 0.397 0.036 0.485 0.009 0.344 0.073 0.299 0.122 0.354 0.065 Note: AOP units: K d m 1 ; Horizontal Secchi Distance (HSD), m. IOP units: m 1 Non significant correlations ( p >0.05) are shown in italics.

PAGE 90

90 Table 2 8. Pearson correlation results between AOPs and IOPs for Weeki Wachee River sites. Apparent Optical Properties K d PAR K d 440 K d 550 K d 660 HSD IOP R p value R p value R p value R p value R p value a t PAR 0.422 0. 064 0.439 0.053 0.419 0.066 0.257 0.274 0.565 0.009 a t 440 0.489 0.029 0.479 0.033 0.481 0.032 0.387 0.092 0.507 0.022 a t 550 0.486 0.030 0.518 0.019 0.454 0.044 0.290 0.215 0.533 0.016 a t 660 0.287 0.219 0.342 0.140 0.248 0.291 0.117 0.622 0.463 0.040 a g 340 0.489 0.029 0.440 0.052 0.483 0.031 0.484 0.031 0.361 0.118 a g 440 0.373 0.105 0.332 0.153 0.345 0.136 0.401 0.080 0.321 0.167 a d PAR 0.532 0.016 0.539 0.014 0.529 0.017 0.365 0.114 0.480 0.032 a d 440 0.554 0.011 0.539 0.014 0.574 0.0 08 0.422 0.064 0.457 0.043 a d 550 0.419 0.066 0.440 0.052 0.411 0.072 0.258 0.272 0.452 0.045 a d 660 0.447 0.048 0.459 0.042 0.387 0.092 0.355 0.124 0.213 0.367 a ph PAR 0.165 0.486 0.129 0.588 0.143 0.549 0.256 0.277 0.481 0.032 a ph 440 0.2 82 0.228 0.193 0.415 0.323 0.165 0.404 0.078 0.181 0.446 a ph 550 0.242 0.303 0.224 0.342 0.219 0.354 0.287 0.221 0.369 0.109 a ph 660 0.075 0.754 0.125 0.598 0.095 0.690 0.091 0.702 0.484 0.030 b PAR 0.168 0.480 0.134 0.574 0.212 0.370 0.167 0.482 0.537 0.015 Note: AOP units: K d m 1 ; Horizontal Secchi Distance (HSD), m. IOP units: m 1 Non significant correlations ( p >0.05) are shown in italics.

PAGE 91

91 CHAPTER 3 DEVELOPMENT AND APPL ICATION OF EMPIRICALLY DERIVED SPECTRAL LY EXPLICIT AND BROADBAND OPTICAL MODEL S OF LIGHT ATTENUATION IN FL ORIDA SPRINGS Introduction Light penetration is of fundamental importance to aquatic ecosystems and it greatly influences the capacity of primary production, biomass, and spatial distribution of submerged aquatic vegetation (SAV) and algal communities (Canfield and Hoyer 1988; Biber et al. 2005) The primary factors which influence light availability in aquatic systems include chlorophyll biomass, detrital suspended solids, and color d issolved organic matter (CDOM, Kirk 1994). These factors influence optical property magnitude and distribution in the water column, which can significantly impact penetration of solar radiation to the substrate. Optical properties can be described by two types, inherent optical proper ties (IOPs), and appa rent optical properties (AOPs, Kirk 1988, 1994). The IOPs are properties that belong to the water, and they include constituents that are both dissolved and suspended particles (Kirk 1988). The nature of the ambient light field does no t impact the IOPs magnitude or distribution. Absorption ( a units m 1 ) and scattering ( b units m 1 ) coefficients are measured across the visible light spectrum (Kirk 1988). The third IOP, the beam attenuation coefficient ( c) can be calculated by summing a + b = c (Kirk 1988) The four main components that a bsorb light are described by these absorption coefficients, with units m 1 : water itself (a w ), CDOM ( a g ), algal pigments (a ph ), and detrital material (a d ) suspended in th e water column (Kirk 1980, 1994) The four components are summed for a total absorption coefficient, a t

PAGE 92

92 The AOPs are the properties of the light field that can be acted upon by factors other than just constituents in the water, and the most commonly measured properties are Secchi depth (SD) vertical diffuse attenuation coefficient for downward irradiance, (K d units m 1 ), and broadband (integrated over the 400 700 nm portion of the spectrum) photosynthetically active radiation (PAR, units mol m 2 s 1 Kirk 1988, 1994). Advancemen ts in technology allowed for the progression of light measurements over time. Secchi depth measurements, which are a simple visual index of water clarity (Preisendorfer 1986) have given rise to broadband PAR measurements (measure of photons, quantity of li ght), in turn have allowed for the quality light, spectral distribution Cole 2001). The latter two AOPs are more meaningful in terms of quality of light availability when spectral composition is considered. The wavelength specificity of light attenuation has more implications for SAV absorption, in respect to photosynthesis (Gallegos et al. 2009). Therefore, it is more useful to determine the amount of each AOP in resp ect to wavelength, thereby measuring spectral vertical attenuation coefficient for downward irradiance, K d ; Gallegos et al. 2009 ). Although PAR is used as an indicator of light qu antity through the water column, photons of different PAR wavelengths are not equally absorbed by light absorbing pigments such as chlorophy ll or other accessory pigments Chlorophyll absorption and photosynthetic output are more efficient in the blue (400 to 500 nm) and red (600 to 700 nm) regions of the PAR spectrum and much less efficient in the green wavelengths (500 to 600 nm, Anastasiou 2009; Gallegos et al. 2009). Th e concept that autotrophs absorb

PAGE 93

93 light more efficiently from specific regions of the PAR spectrum has been defined as photosynthetically usable radiation (PUR), which is the integral of the quantum spectrum (400 to 700 nm) we ighted by the relative absorption spectrum (A L ) of the SAV of interest (Morel 1978; Gallegos 1994 ; Gallegos et al. 2009 ). PAR adequately quantifies the amount of light available f or photosynthesis, however it is less adequate for developing optical models that are used to determine the dominating factors contributing to light attenuation and loss of clarity (Gallegos 1994). H ence an understanding of the spectral distribution of PAR should be central to optical modeling, ecological studies and SAV restorat ion efforts (Kirk 1994) In addition, developing a linkage between AOPs and IOPs (Mobley 1994) and relating water quality concentrations to SAV restoration targets are important goals in the field of hydrologic optics (Gallegos 2001). Neither Secchi depth nor the attenuation coefficient disclose any information about the components of water quality that cause light attenuation. It is nearly impossible to use these measurements to set management goals for specific substances to achieve a desired water qualit y to protect SAV, and to provide a suitable habitat for restoration of SAV ( Gallegos 1994; Biber et al. 2008). Accurate measurements of K d (or spectral K d ) are difficult to obtain and optical water quality ( OWQ ) and IOP data have been collected synopticall y by several researchers to develop empirical light attenuation models or to calibrate previously developed spectrally specific light attenuation models to reduce inherent error that is introduced by changing field conditions (Gallegos et al. 2009 ; Dixon e t al. 2010 ). This brings to light the reason for measu ring IOPs. Optical models have been developed based on the

PAGE 94

94 absorption and scattering coefficients for many coastal waters and lakes and are used to predict habitat suitability in terms of OWQ and light availability for seagrass and other SAV (Gallegos 1994, 2001; Gallegos and Kenworthy 1996; Biber et al. 2008). Radiative transfer modeling provides the linkage between IOPs and AOPs, as well as optical properties (Biber et al. 2008; Anastasiou 2009). Turbidity is a useful predictor of the absorption coefficient for detrital material (a d ), while color is used to predict the absorption coefficient for CDOM (a g ), and chlorophyll a conc entration is used as an estimate for the absorption coefficient for algal pigments (a ph Gallegos and Kenworthy 1996; Biber et al. 2008). Several recent studies have used similar modeling methods to determine the relative importance of each light absorbing component. A study by Kelble et al. (2005) illustrates the importance of understanding the light environment in Florida Bay and found that tripton (detritus) was the dominant factor controlling light attenuation They used a statistical model to estimate the light attenuation coefficient across the entire bay. Chlorophyll a and CDOM fluorescence and beam transmission (c) data were each regressed against measured concentrations of chlorophyll a CDOM and tripton (detritus) respectively, to develop a relat ionship with water quality concentrations and light attenuation based on fluoresce nce and beam transmission A study by McPherson and Miller (1994) determined partial attenuation coefficients for color and chlorophyll a by derivation using stepwis e multipl e regression analysis. Coefficients derived from the regressions were then multiplied by concentrations to yield the contributions of each component to light attenuation. The

PAGE 95

95 natural logarithm of the ratio of simultaneous in air and in water PAR was plotte d versus depth, followed by a least squares fit to determine the attenuation coefficient. A recent, more technologically progressive study was conducted using a continuous flow through deck mounted system, which was modified to characterize the optical li ght environment (Anastasiou 2009). The researcher modeled absorption coefficients using chlorophyll and CDOM fluorescence relationships, which are site specific and must be calibrated to a specific area. The relative magnitude and contributions of each of the components were determined for Tampa Bay Florida a cross each wavelength in the visible spectrum M inimum light targets for seagrass were determined by looking at percent subsurface irradiance ( % PAR) at the deepest location that seagrass were found, wh ich is also known as the seagrass deep edge (Anastasiou 2009 ). Riverine OWQ is highly variable and difficult to characterize (Davies Colley et al. 2003, Davies Colley and Nagels 2008, Julian et al. 2008a, 2008b) and particulates typically dominate light a ttenuation in rivers, which is understandable that most optically related studies have previously focused on turbidity to characterize riverine OWQ and optical regime (Davies Colley et al. 2003, Davies Colley and Nagels 2008). Given that Julian et al. ( 20 08 b ) characterized the light regime in rivers a s optically complex (Davies Colley and Nagels 2008) and suggested that modeling light in rivers should consider five hydrogeomorphic controls: 1) topography, 2) riparian vegetation, 3) channel geometry, 4) OWQ 5) hydrologic regime. They developed an empirical model to predict spatial and temporal variability of PAR reaching the riverbed (benthic PAR) T he model was used to quantify and generalize hydrogeomorphic controls in optically

PAGE 96

96 diverse rivers, with the f inding that d ischarge was a temporal control and water depth was the spatial control in the large r river T affected by above canopy shading temporally and spatially (Julian et al. 2008b) Kirk (1981, 1984, 1991, a nd 1994 ) presented an empirically derived spectrally explicit optical model that established the relationship between the diffuse attenuation coefficient, and the total absorption and particulate scattering coefficient, which after several iterations of th e model eventually took the form of Equation 3 1: K d 0 [a t 2 + (g 1 0 g 2 ) a t b] 1/2 Equation ( 3 1 ) where: K d ( m 1 ) = vertical diffuse light attenuation coefficient at a specific wavelength a t ( m 1 ) = the total absorption coefficient at a specific wavelength b (m 1 ) = the scatteri ng c oeffici ent at a specific wavelength 0 = the cosine of the zenith angle of the direct solar beam refracted at the air water interface g 1 and g 2 = coefficients that depend on the scattering phase function of the water column, on the optical depth of i nterest and were empirically determined for the midpoint of the euphotic zone This model has proven to be a useful resource management tool in mostly tidal river and estuarine applications where it is difficult and labor intensive to collect accurate lig ht attenuation measurements (Gallegos 1994, 2001, 2005; Gallegos and Kenworthy 1996 ; Biber et al. 2008 ). The model has been validated against Monte Carlo procedures modeling radiative transfer equations to determine the sources of error in predictions of K d by simulating random errors in model coefficients and water quality parameters. The model is reliable in predicting the relevant amounts of light received by SAV over varying seasons, depths, water quality conditions and geographic regions from OWQ data, which is much less intensive than measuring spectral K d over large areas. W applicability and usefulness for predicting

PAGE 97

97 light attenuation and determining the main constituents causing the loss of water clarity can be opt imized for optically complex and unique systems such as spring fed rivers. The objective of this study wa s to develop new and calibrate existing spectral and broadband optical models using functional relationships between IOPs and AOPs and routinely measur ed OWQ variables on a site specific and on a generalized scale The rationale for developing and using optical models to predict spectral light attenuation is based on the practical significance of reducing time in the field and to reduce potential error t hat can be introduced from varying conditions in the field. Historical light regimes were modeled and the model predictions were used to test the hypothesis that light attenuation has significantly increased and benthic light availability, especially in t he blue band (i.e. 4 40 nm) has decreased over time in two spring fed rivers in question, Rainbow and Weeki Wachee Rivers in Florida, USA Site specific r elations hips were developed between historical optical water quality data IOPs and light attenuation d ata to determine benthic habitat suitab ility based on the sufficiency of wavelength specific light available for SAV growth and survival The modeling approach used the combined datasets from Rainbow and Weeki Wachee Rivers, which p rovided a framework and resource management tool to predict spectrally specific light at depth (measure of water clarity) as a function of OWQ parameters and to evaluate the quantity and quality of light available to S AV and other primary producers in other spring systems which had not existed until now. Methods Study Area An optical study was conducted that measured o ptical water quality (OWQ) concentrations (turbidity, chlorophyll a and color), inherent (absorption and scattering

PAGE 98

98 coefficients) and apparent (diffuse vertical l ight attenuation coefficient) optical properties in Rainbow (7 stations) and Weeki Wachee (5 stations) Rivers, both spring fed rivers near the west coast of Florida, USA Figures of the study areas are shown in Figures 2 1 and 2 2 and descriptions of the two systems can be found in the Methods section in Chapter 2. Measurements were conducted quarterly (January, April, July, and October ) in 2011 Model Development Chapter 2 details methods used to collect and analyze OWQ, IOP and AOP samples. D ata obtaine d from the optical study ( color, chlorophyll a, turbidity, a t 440, a t 550, a t 660, a d 440, a g 340, a ph 440, a d 550, a ph 550, a d 660, a ph 660, K d 440, K d 550, and K d 660 ) in Rainbow (n=28), Weeki Wachee (n=20) and combined systems (n=48 the datasets from both systems w ere combined to develop optical model equations that can be used generally in other systems ) were used to develop new and calibrate existing empirical optical model s for the two spring systems Measured K d values were adjusted to a zenith angle of 0 (vert ical zenith angle) to remove the effect of the longer path length, thereby allowing for comparisons of modeled K d values calculated using a direct solar beam. The zenith angle adjusted K d was calculated by multiplying measured K d by the cosine of the zenit h angle of water (Kirk 1994). Four p reviously developed spectral and broadband optical m odels were obtained from the literature ( Models 1, 2a, 2b, 3, Table 3 1) datasets to compare to the optical models modified or develop ed as part of this study (Models 2c, 2d, 4, 5, Table 3 1) To develop the site specific broadband optical models (Models 4 and 5, Table 3 1), and the combined springs total absorption and light attenuation models (Models 12 to 15, Table 3 3), multiple line ar regression analyses

PAGE 99

99 were conducted using OWQ concentrations to achieve the most accurate model s with the greatest predictive power. Model evaluation and comparison involved calculation of a vast set of model precision metrics (Tables 3 4 and 3 5) f o r se lect ion of the most robust and accurate spectral and broadband models. Models were selected based on the lowest level of error, and t he selected site specific models were used to hind cast a historical light environment using historical OWQ data for both s ystems Spectral Model Calibration To obtain input for the spectral models (Models 1, 2a, 2b, 2c, 2d, Table 3 1), scattering coefficients at reference wavelengths (440, 550, 660 nm) were calculated from turbidity by Equation 2 3, and total absorption (a t ) was partitioned as shown in Equation 2 1. Model calibration was dependent on the linear relationships between each measured partial absorption coefficient and its respective measured OWQ concentration. Absorption coefficient values due to water (a w ) itsel f were obtained from the literature (Pope and Fry 1997). O ptical water quality concentrations were used to predict the partial absorption coefficients at reference wavelengths (a d 440, a g 340, a ph 440, a d 550, a ph 550, a d 660, a ph 660 ) The best subsets model bui lder tool in the MINITAB statistical software package was used to determine which individual or group of OWQ variables would best predict each partial absorption coefficient Mallows Cp fitting parameter and the adjusted coefficient of determination (adj R 2 ) were evaluated together to avoid over fitting the model s Only statistically significant models ( p <0.05) were included in the total absorption summation of par tial absorption coefficients Models 2a and 2b contained scattering phase coefficients (g 1 an d g 2 ) that were drawn directly from the literature (Kirk 1984, 1994, Table 3 1). These coefficients were

PAGE 100

100 empirically derived for aquatic systems that greatly differed from freshwater flu vial systems, especially spring fed rivers. Therefore, to further cali brate Model 2b and develop Model 2c adjustments were made to the g 1 and g 2 coefficients. The Excel Solver Analysis Tool was used to find the optimal (minimal) value of the root mean square error (RMSE) by adjusting the g 1 and g 2 coefficients in the mode l. This procedure was conducted for each system individually, and was also for the combined datasets. One final adjustment was made to the algorithm of Model 2c to develop Model 2d. In Chapter 2 Results canopy cover (OC) was found to be significantly c orr elated to light attenuation. A s a result OC was included as an additional explanatory variable to Model 2d in an attempt to improve the model. Model Application Historical OWQ data (color, chlorophyll a, turbidity concentrations) obtained from SWFWMD for the years 2002 to 2011 for Rainbow River and from 2005 to 2011 for Weeki Wachee River were used for both modeling approaches The selected spectral and broadband optical model s (Model 2c and 4, respectively, Table 3 1) were applied to predict historical li ght attenuation coefficients from historical OWQ concentration data. Using Model s 6a to 11b ( Table 3 2), historical OWQ data were used to predict each individual partial absorption coefficient for input into spectral optical Model 2c to hind cast historica l light attenuation coefficients (K d To allow for a comparison of predicted K d for the period of record, spectral K d values were predicted for solar noon on Julian day 152 of each year to assume a vertical beam of radiation with maximum light penetrat ion at each site, irrespective of canopy cover. The location (latitude and longitude) of the headspring area in each sprin g system (Rainbow and Weeki Rivers) was used to calculate the solar noon

PAGE 101

101 (12:00PM EST) zenith angle of water ( overall mean solar eleva tion of 83 ) for all hind cast predictions of spectral light attenuation following methods described by Kirk ( 1994) Benthic light availability was estimated by the p ercent s pect ral and broadband subsurface irradiance [% PAR ( calculated by Equation 3 2 : % PAR ( K d D Equation ( 3 2 ) where K d 1 ) and D is the total depth (m). Temporal trend analyses were conducted on measured OWQ concentrations, predicted IOPs, and predicted AOPs using simple linear regression with time (year) as the explanatory variable to determine if the OWQ, AOPs, and IOPs have increased or decreased significantly over time. A statistically significant trend was evident if p < 0.05. Results M odel Calibration and Development Rainbow River IOP, AOP and OWQ Relationships D etrital, algal pigment and CDOM absorption coefficients were all significant predictors and were used to predict total absorption at 440 nm wavelength ( a t 440 ) as shown in Figure 3 1 Table 3 2 shows the shows the probability of statistical significance ( p value) adjusted coefficient of determination (adj R 2 ), standard error (s) and models developed for partial absorption coefficient regression models using OWQ concentrations as predictors. When modeling the partial absorption coefficients using OWQ concentrations d etr ital absorption (440 nm) was best predicted when both color and chlorophyll a were used to model a d 440 without any indication of multi collineari ty, rather than using just one of the two water quality parameters alone ( Figure 3 2 ) CDOM absorption ( a g 34 0 ) wa s predicted best by

PAGE 102

10 2 color alone. A lgal pigment absorption at 440 nm was significantly related to chlorophyll a and was used to solely predict a ph 440 ( Figure 3 2 ) D etrital and algal pigment absorption ( 550 nm) were both significantly related to a t 550 and a t 660 (Figures 3 3 and 3 4) Chlorophyll a was the only significant predictor for a d 550, a d 660 and a ph 660 (Figure 3 5). Figure 3 6 shows the i nitial univariate correlations that were used to develop the broadband optical multivariate regression model that predict ed K d PAR directly from all three OWQ conc entrations color, turbidity and chlorophyll a (Table 3 1) Weeki Wachee River IOP, AOP and OWQ Relationships For a t 440, algal pigment absorption was not a significant predictor but bo th d etrital and CDOM absorption (440 nm) were significantly related and were used to in the model to predict total absorption at 440 nm (Figure 3 1) Total absorp tion at 550 was only correlated with a d 550, and a t 660 was correlated with both a d 660 and a ph 660 (Figures 3 3 and 3 4). Table 3 2 provides the partial absorption coefficient regression models and fitting parameters using OWQ concentrations as predictors. Fi gure 3 7 shows the OWQ concentrations that were significantly correlated with the partial absorption coefficients. Similar to Rainbow River, only color correlated with a g 340 and chlorophyll a with a ph 660. Turbidity was significantly related to a d 550 and wa s solely used to predict a t 550 since CDOM and algal pigment absorption were not significant predictors of total absorption at 550 nm. Turbidity and chlorophyll a were significantly r elated to a d 660 and a ph 660 respectively (Figure 3 7 ) Figure 3 6 shows th e individual correlations used to develop the multiple linear regression K d PAR model (Table 3 1)

PAGE 103

103 Combined Dataset IOP, AOP and OWQ Relationships Table 3 3 shows the p values, adj R 2 s, spectral total absorption (a t and broadband light attenuation (K d PAR) multi linear regression models that were developed using OWQ concentrations from the combined dataset s (Weeki Wachee and Rainbow Rivers) Total absorption at 440 nm was best predicted by color and turbidity. Color chlorophyll a and turbidity provided the bes t fit model for a t 550, and chlorophyll a and turbidity explained the greatest amount of variation for a t 660. K d PAR was only significantly related to color and chlorophyll a ( Table 3 3 ) The combined dataset al lowed for the empirical determination of spectral slope for the partial absorption CDOM coefficients (a g ) as a function of colo r that was best described by a quadratic equation that is shown in Figure 3 8 The spectral slope (S g ) for color above 2 PCU appe ared to reach an asymptotic value near 0.018 The majority of data points for the combined dataset clustered below color concentration of 2 PCU, with only 5 data points above that value, which may have led to the low coefficient of determination ( R 2 = 0.3 7). Model Evaluation In many cases during spectral light attenuation and absorption and broadband light attenuation model development, best subset model analysis indicated that a model with more variables had higher explanatory power (R 2 ) than a model with less variable s, which occurred for Models 6c, 7, 9a, 10, 11a, 12, and 15. To prevent over fitting models were first select ed based on the lowest Mallows C p scores (determines degree of potential multi collinearity) and second, the highest R 2 values (data not shown). However when conducting t he multi linear regression s, variables were removed from the models that did not substantially improve the explanatory power of the overall

PAGE 104

104 model, which usually meant that the predictor variable did not have a si gnificant effect on the response variable ( p > 0.05 ) In general, only significant variables were used in the multi linear regression modeling efforts to avoid issues such as multi collinearity from over fitting the model that contained too many variables. This was a common occurrence and the resultant models used to predict partial and total absorp tion and diffuse light attenuation coefficients are shown in T ables 3 1, 3 2 and 3 3. All system specific spectral and broadband vertical diffuse light attenuat ion regression model equations, descriptions of model variables, scattering phase coefficients (Models 2a to 2d only) and source of model (i.e. obtained and/or modified from a literature source or developed as part of this study) for Rainbow and Weeki Wach ee Rivers are given in Table 3 1. Spectrally specific and broadband K d s were calculated as described in model equations 1 to 2d (spectral) and 3 to 5 (broadband) and graphical comparisons of modeled output against measured K d values for all models are show n in Figures 3 9 and 3 10 for Rainbow and Weeki Wachee Rivers, respectively. Spectral Optical Models (SOM) Table 3 1 shows the five variations of the spectral optical model : Model 1 included only absorption (a t ) and scattering (b) coefficients; Model 2a incorporate d the scattering phase function coefficients (g 1 and g 2 ) ; Model 2b extended Model 1 to include the cosine of the zenith angle ( 0 ) using literature g 1 and g 2 values; Model 2c and 2d used modified g 1 and g 2 coefficients derived by model optimi zat evaluated the inclusion of an addi tional variable, %open canopy The spectral optical models were compared using seven model precision metrics to determine which model was the most precise

PAGE 105

105 and robust and then that model was selected for use in historical hind casting and future modeling of spectral light attenuation. Rainbow River SOM T he best model was Model 2c and the agreement between modeled versus measured K d values are shown in Figure s 3 9 (in respect to al l other models) and 3 11 Model precision metrics are shown in Table 3 4 for Rainbow River spectral models of diffuse light attenuation. Model 2c and had the highest accuracy overall, with the lowest error values for almost all precision metrics : sum squa re error (1.33 m 1 ) mean square error (0.016 m 1 ), root mean square error (RMSE, 0.126 m 1 ), and %RMSE of the mean (33 .8%) and max (15.2%) K d However, all the models had very similar R 2 values, ranging from 0.6 70 to 0.701 which did not give any indicati on as to which was the most robust model. Model 2a performed the worst out of all spectral models, with very poor accuracy that greatly underestimated K d s. Although Model 2c measured versus modeled K d values had generally good agreement (Figure 3 11), the model underestimated the lower measured K d values (K d < 0.4 m 1 ) and tended to overestimate higher measured K d values (K d >0.4 m 1 ) Weeki Wachee River SOM Similar to the case of Rainbow River, Model 2c ** was also the most robust and precise in terms of m odel error when compared to the other spectral models with a RMSE of 0.157 m 1 which is 36.7% and 12.3% of the mean and max K d respectively (Table 3 4, Figures 3 10 and 3 11). However Model 2d ** also performed well in comparison and only had slightly hi gher error values for all model precision metrics except for the coefficient of determination (R 2 = 0.678) which was slightly better than Model 2c ** R 2 ( 0.604). Nevertheless, Model 2c ** was selected for further modeling efforts due to overall lower error and to simplify the number of

PAGE 106

106 variables contained in the model Even though the added variable in Model 2d ** %Open canopy (OC) was significantly related to K d (at 440, 550 and 660 nm) as was shown in Chapter 2 (Table 2 6 ) modifying the spectral equatio n (Model 2c ** ) to include OC did not improve model perform ance as expected. Model 2a performed the worst out of all spectral models, with poor accuracy that generally underestimated spectral K d s. Th e model agreement seen previously in Rainbow River corresp onded to the agreement in Weeki Wachee River betw een measured and modeled values where Model 2c ** predictions were slightly biased low for the lower measured K d values (K d <0.4 m 1 ) and were slightly biased high for the higher measured K d values (K d >0.4 m 1 Figure 3 11) Combined General Springs SOM R ainbow and Weeki Wachee River datasets were combined to calibrate the scatterin g phase coefficients (g 1 and g 2 ) for spring systems as seen in Model 2c*** (Table 3 1) However, if IOP and AOP da ta we re made av ailable for another spring system calibrating Model 2b using the same approach used to develop the site specific Model 2c by adjust ing the scattering phase coefficients (g 1 and g 2 ) to maximize model precision on a system specific basis would be ideal. Mor eover, since actual partial absorption coefficient data we re unavailable for spring systems other than Rainbow and Weeki Wachee Rivers, a simplified approach to determine total absorption at the reference wavelengths (440, 550, 660 nm) was used by developi ng multivariate prediction models to predict total absorption directly from OWQ concentrations with equations (Model 12 to 14) shown in Table 3 3 This approach saved an extra procedur al step that did not appear to reduce model accuracy with relatively lo w standard error (s) ranging from 0.010 to 0.038 and R 2 values that

PAGE 107

107 were higher on average (mean R 2 = 0.67 for Models 12 to 14 ; Table 3 3) than the overall mean R 2 for the partial absorption coefficient models ( mean of R 2 = 0.5 2 for Models 6a to 11b; Table 3 2) for both systems The spectrally specific total absorption model (Mod els 12 to 14) output data are to be used as input data into the combine springs spectral optical Mode l 2c*** to predict spectral light attenuation for other spring systems Broadba nd Optical Models (BOM) Three variations of broadband optical models were evaluated to predict vertical diffuse light attenuation for PAR (K d PAR) Table 3 1 shows t he broadband models, Model 3, 4 and 5 for Rainbow (denoted with in the table) and Weeki W achee (denoted with ** in the table) Rivers. Model 3, empirically derived by McPherson and Miller ( 1994), was applied to the data collected for this study Model 5 followed the same approach of McPherson and Miller (1994) ; however site specific model coeff icients were obtained by developing simple regressions with each IOP and its corresponding OWQ concent ration to develop partial attenuation coefficients and then summing the partial attenuation coefficients to estimate K d PAR. Development of Model 4 began with a n initial correlation analysis for each individual system and for the combined dataset s that resulted in several significant correlations with K d PAR and each of the OWQ concentrations that were considered for inclusion into each BOM. For each system and com b ined dataset the following p arameters: color, chlorophyll a, turbidity and %canopy cover were considered during mu ltiple linear regression analyse s that determined the minimum number of variables necessary to provide the greatest explanatory power without over fitting the model.

PAGE 108

108 Rainbow River BOM Out of the three broadband optical models evaluated (Model 3, 4, and 5), Model 4 was by far the most robust in terms of model predictive power and uncertainty when compared to the other broadband models w ith an R 2 of 0.697, RMSE of 0.064 m 1 which is only 9.9% and 17.3 % of the mean and max K d respectively (Table 3 4, Figures 3 9 and 3 11). All three OWQ variables, color, chlorophyll a and turbidity were determined to be significant ly related with K d PAR a nd did not over fit the multi linear regression model ( Figure 3 6, Table 3 1). Model 3 was the least accurate with orders of magnitude greater error than Model 4. Model 4 provided good agreement between measured versus modeled value s (Figure 3 11), therefo re Model 4 was selected for further modeling efforts Weeki Wachee River BOM. The most robust model was Model 4 and was in similar form as the BOM for Rainbow River. When conducting univariate regressions, only color and turbidity were significantly relat ed to K d PAR, and chlorophyll a was not ( p = 0.07, Figure 3 6). However the multi variate regression m odel (Model 4 ) was stati stically significant overall ( p = 0.048) when all three OWQ variables were included in the model. Furthermore, with the addition of overall predictive power was improved considerably R 2 value from 0.381 (without OC in the model) to 0.633, nearly doubling the explanatory power of the model The strength of the overa ll significance of the model was also greatly improved with the inclusion of OC, where the p value decreased from 0.048 (without OC) to 0.003 (with OC) Based on the model precision metric evaluation of the three broadband optical models (Table 3 4), Model 4 was the most accurate and was chosen for further modeling efforts.

PAGE 109

109 Combined General Springs BOM The resultant broadband light attenuation (K d PAR) model developed using the combined datasets is shown in Table 3 3. Only color and chlorophyll a concentr ations were statistically significant predictors of K d P AR. Turbidity and OC were not significant predictors in either univariate or multivariate regressions and were not included in the model. Although highly significant ( p = 0.000) the model wa s only abl e to explain 36% of the variation in K d PAR, which may be due to the spatial variability of particulate absorption and scattering and discrete differences in CDOM absorption across the two systems used to develop the regression model. Model Application: Hi nd cast ing Historical Optical Properties and Water Clarity Regime Measured OWQ concentrations (color, chlorophyll a, and turbidity) and predicted inherent (spectrally specific partial and total absorption and scattering coefficients predicted by Models 6a to 11b, Table 3 1 ) and apparent (spectrally specific and broadband vertical light attenuation and percent subsurface irradiance predicted by Models 2c and 4, Table 3 1 ) optical properties were examined to determine if detectable temporal trends were evide nt in the two spring systems Rainbow and Weeki Wachee Rivers for the measured OWQ period of record (POR) A total of 204 simple linear regressio ns were conducted to analyze for statistically significant increasing or decreasing trends for 18 variables for all stations in Rainbow (POR 2002 to 2011) and Weeki Wachee (POR 2005 to 2011) Rivers. Overall, very few significant ( p <0.05) relationsh ips indicating a temporal trend were found only 6 out of 204 or 3% of all regressions, with 4 occurring in Rainbow and 2 in Weeki Wachee River as shown in Table 3 6.

PAGE 110

110 Temporal Trends of Measured OWQ Concentrations Graphical representations of temporally variable measured annual average color, turbidity and chlorophyll a (chla) concentrations are shown in Figure s 3 12 and 3 13 for Rainbow and Weeki Wachee Rivers, respectively. Only 2 significant temporal trends were found where color in WW5 increased and chl a decreased in RR8 POR ( Table 3 6 ) Spatial variation was relatively high in Rainbow Rive r stations for color and chla ( Figure 3 12 ) but seemed to have a consistent sp atial pattern for turbidity, with RR7 or RR8 having higher values for all three parameters than the other stations A ll three parameters were spatially variable in Weeki Wachee River with WW5 having higher color and chla concentrations for all years and WW3 having higher turbidity values for most years ( Figure 3 13 ) Temporal Trends of Predicted Inherent Optical Propertie s Only two predicted IOPs (absorption and scattering coe fficients), a t 660 in RR8 (decreasing trend) and a t 440 in WW5 (increasing trend) were found to have a statistically significant temporal trend for both spatially variable systems ( Figures 3 14, 3 15 and Table 3 6). The most downstream stations in both sys tems (RR7, RR8, WW3 and WW5) had the highest IOP values in comparison to the other stations in each system (Figures 3 14 to 3 17 ) The scattering coefficient patterns (Figures 3 16 and 3 17) resembled patterns shown for turbidity in Figures 3 12 and 3 13 o ver the POR for both systems. Predicted partial absorption coefficients (a d a g a ph and a w ) at reference wavelengths (440, 550 and 660 nm) were evaluated to determine which absorption coefficient s contributed the greatest to total absorption at each res pective wavelength over the POR Figures 3 18 through 3 32 show the relative percent contribution of each

PAGE 111

111 predicted partial absorption coefficient to the total absorption at each station in both systems over the POR In Rainbow River, a w 440 was a minor abs orbing component for most stations and it did not vary much over the POR. a g 340 was variable and contributed the most to total absorption with the highest values in 2002, 2006 and 2007 in the upper river stations ( Figure 3 18 ) In the middle and lower Rain bow River stations, (Figure s 3 19 and 3 20 ) a ph 440 and a g 340 both contributed approximately equally although a slight pattern was seen where as a ph 440 decreased, a g 340 increased over time. In all Rainbow River stations, a w 550 dominated a t 550, wit h no disc ernable temporal trends found for either a w 550 or the only other absorbing component a d 550 ( Figure 3 21, 3 22 and 3 23 ). Total absorption at 66 0 nm was entirely dominated by water itself fo r all stations in Rainbow River. Relative percent contribution was nearly negligible (<5% for upper and middle stations, Figures 3 24 and 3 25, and <10% for lower stations, Figure 3 26) for a ph 660 and even less so for a d 660 which did not change substantially with time over the POR. In all stations in Weeki Wachee River, a w 440 contributed minimally to a t 440, and a ph 44 0 did not contribute at all to total absorption ( Figures 3 27, 3 28 and 3 29 ). In the upper river stations a g 340 contributed more than a d 440 in 2006, 2007, and 2009, but a d 440 contribution was higher than a g 340 in 2005, 2008 and 2011. Though a g 340 and a d 440 appear ed to have opposite temporal trends in WW3 and WW5 lower river stations where detrital absorption dominated total absorption for most years in WW3, and CDOM absorption seemed to increase over the POR and dominated a t 440 during most years (except 2005) in WW5 (Figures 3 28 ). The upper Weeki Wachee River

PAGE 112

112 stations and WW5 had similar spatial trends as Rainbow River where a t 550 was dominated by a w 550 with minor contributions from a d 550 and no apparent te mporal trend over the POR ( Figures 3 29 and 3 30 ). However, WW3 in the lower river tended to have higher detrital absorption for all years except 2011 ( Figure 3 30 ). The same pattern of total absorption at 660 nm was seen in all Weeki Wachee River stations similar to Rainbow River where absorption due to water itself entirely dominated total absorption, with very minor contributions of a d 660 and a ph 660, also with no distinct temporal trend (Figure 3 31 and 3 32 ). Temporal Trends of Predicted Apparent Optica l Propertie s Spectral (K d and broadband (K d PAR) vertical diffuse light attenuation coefficients and percent s pect ral (% PAR the selected SOM and BOMs (Models 2c and 4, Table 3 1 ). For the AOPs, only %PAR 550 had s tatistically significant (increasing) trends in RR7 and RR8 while none of the light a ttenuation coefficients at reference wavelengths (K d 440, K d 550 and K d 660) and for PAR at any station in either system had statistically significant temporal trends (Table 3 6). However, K d 440 had apparent decreasing trends in RR7 and RR8 and increasing trends in WW1 and WW5 over the POR ( Figures 3 33 and 3 34 ) which was mirrored in the opposite direction with %PAR at each reference wavelength and for PAR ( Figures 3 35 and 3 36 ) mea ning that when K d appeared to be decreasing then %PAR corresponded with an apparent increasing trend T he decreasing apparent trend for K d PAR was mirrored by the increasing trend for %PAR in WW0, WW0.5 and WW1 (Figures 3 34 and 3 36)

PAGE 113

113 Over th e predicted period of record for %PAR 440 %PAR 550, %PAR 660 and %PAR, minimum values of ca. 40% ( in 2003), 50% ( in 2003), 20% ( in 2002 and 2009), and 20% ( in 2002 and 2009) respectively, were f ound in the most downstream Rainbow River station RR8 ( Figures 3 35 and 3 37 ). In 2003 Weeki Wachee River station WW3 reached minimal values of 20%, 40%, and 20% for %PAR 440 %PAR 550, and % PAR 660, respectively ( Figure 3 36 ). However, for %PAR (broadband) WW5 had the lowest values with ca. 40% PAR ( Figure 3 38 ). Discussion Model Applicability Lotic systems are known to have high levels of spatial and temporal variability that have present ed difficulties for others while assessing optically complex light regimes in rivers ( Gallegos 2005; Davies Colley and Nagels 20 08; Julian et al. 2008b). This complexity justified the case for expanding upon previous spectral optical modeling efforts (Model 2b, Table 3 1) to predict vertical diffuse light attenuation from inherent optical properties and optical water quality variab les and adapts the approach to the two spring fed systems, Rainbow and Weeki Wachee Rivers. Using the data to calibrate the most robust spectral model (Model 2c, Table 3 1) available at this time provided site specificity that reduce d uncertainty, increase d accuracy by almost 30% and showed generally good agreement between observed and predicted light attenuation values The calibration technique s employed here adjustment of the scattering phase coefficients (g 1 and g 2 ) and calibration of the model using relationships derived from measured optical water quality and absorptions coefficients, were a viable approach to optimize the spectral model that has been developed and used by others in a variety of aquatic systems (Kirk 1984, 1994; Gallegos 1993, 1994, 2001, 2005)

PAGE 114

114 Broadband empirical modeling using multi linear regression analyses provided a simplified and accurate approach to predict vertical diffuse light attenuation in the spring fed rivers evaluated for this study (Model 4, Table 3 1) However, appl ying an empirical model (Model 3, Table 3 1) that was developed for a distinctly different water body type (e.g. estuaries ) to rivers did not provide successful results. Also, using the same approach taken by McPherson and Miller (1994) to partition the l ight attenuation coefficient into a set of partial attenuation coefficients in a pseudo mechanistic manner using data from my study did not prove to be useful The partial attenuation coefficients include scatter and are not strictly linear functions of w ater properties or constituent concentrations, which provide only a rough approximation of tota l light attenuation (Kirk 1983; McPherson and Miller 1994) Davies Colley and Nagels (2008) developed three multiple linear regression equations (see Table 3 in Davies Colley and Nagels 2008) for predicting light attenuation using beam attenuation coefficient at 550 nm (c550), turbidity concentration, and CDOM absorption coefficient at 340 nm (a g 340) data measured in 17 optically diverse rivers in New Zealand Th eir models performed satisfactorily when tested in their rivers However, their models performed poorly when the models were applied to my two spring systems, with R 2 values averaging 0.1 and 0.17 for Rainbow and Weeki Wachee, respectively when comparing measured and modeled values. They did caveat that they did not recommend extrapolating their models to predict light attenuation into waters clearer than their clearest river the Mouteka River ( Davies Colley and Nagels 2008), which in fact, was the case f or most sites in Rainbow and Weeki Wachee Rivers.

PAGE 115

115 Again, site specific optical model development is ideal and quite often a necessity to reduce uncertainty, especially when the system in question contains constituent levels outside of typical ly measured values i.e. extremely clear water or the converse. Even when site specific optical models are developed, there is an inherent measurement error that could be lessened through collection of o ptical water quality samples as a vertical profile composite s o t he data would reflect the vertical profile of light attenuation measurements. Especially in rivers, the potential to introduce measurement error from other aspects of light and optical water quality sampling as was discussed in Chapter 2, is relatively hig h and likely dominates model uncertainty. H owever when light attenuation data are not available to develop site specific models, and when accuracy must be relinquish ed for the sake of simplicity then the combined broadband optical model (Model 15) that was developed as part of this study can be used for other systems displaying similar optical characteristics to Rainbow and Weeki Wachee Rivers The datasets used to develop the regression models we re somewhat diverse in terms of dominant controls on river dynamics such as river size, hydrology, geomorphology, canopy cover, channel orientation and aquatic macrophyte coverage (Julian et al. 2008b) Therefore, applicability of the models developed here to other similar spring systems is reasonable. The broa dband combined optical model (Model 15 Table 3 3 ) developed for this study with the combined dataset is one of the simplest light attenuation models available The model only require s two easily measured variables, color and chlorophyll a concentration da ta to predic t light attenuation. However, t he broadband multi linear

PAGE 116

116 regr ession combined springs model is considerably less accurate than using the combined springs spectrally specific optical model (Model 2c*** Table 3 1) Model 2c*** was calibrated usi ng both datasets and incorporates estimation of total absorption using the regression equations (Models 12 to 14, Table 3 3) developed here to obtain an accurate input for the spectral mode l. The reduction in accuracy in the broadband model may be due to s patial variability of particulate absorption and scattering and discrete differences in CDOM absorption across the two systems used to develop the combined regression model. Furthermore accuracy for both modeling approaches is affected by the presence of vertical gradients due to varying depths on attenuating constituents that are not represented by the OWQ concentration sample data. As often seen in other aquatic systems, scattering (b) had a stronger effect on measured and modeled light attenuation than absorption (a, Kirk 1994; Swift et al. 2006) in the blue, green and red bands of the PAR spectrum (440, 550, 660 nm wavelengths) C orrelation s between modeled b d were more than twice as strong ( mean of b550 and b660 nm, R 2 = 0.98 p = 0.000 ) as the correlations between a t K d ( mean of a t 550 and a t 660, R 2 = 0.45 p = 0.000) in Rainbow River sugges ting that scattering wa s the most dominant attenuator of light in Rainbow River. In contrast scattering was not found to be the domina nt attenuator in Weeki Wachee River due to similar correlation strength (mean of b440, b550 and b660 nm, R 2 = 0. 90 p = 0.000) with light attenuation a s absorption for all three waveband s (mean of a t 440, a t 550 and a t 660, R 2 = 0. 88 p = 0.000)

PAGE 117

117 The u til ity of the spectral modeling approach for the management of SAV in spring systems is the ability to determine which optical water quality constituent is driving light attenuation and potentially limiting SAV growth and distribution. Further insight into th e factors contributing to absorption and scattering properties would aid in the understanding of how light attenuating constituents contribute to the reduction in water clarity and light availability K nowledge of the p article size distribution of particul ates and partitioning of the inorganic and organic fractions of the detrital particulate materi al w ould provide more information on the character of turbidity and would enhance the ability to quantify the relative effect of organic and suspended mineral s ediments on light attenuating constituents ( Swift et al. 2006 ; Julian et al. 2008a) A lgal c ommunity structure and cell physiology dictate the variability of algal pigment absorption and scattering that lea d to contribution s of organic materials in the d etrital and dissolved organic fraction s Chlorophyll a concentration data cannot provide an indication to the cause of variability from changing algal community composition and their associated optical properties that can uniquely affect light absorption a nd scattering properties. The complex relationship between scattering and absorp tion as influenced by differing water column entrained algal communit ies and varying contributions from mineral inorganic and organic particles warrants further investigation i n spring systems Modeling Historic al Optical Regimes As with most empirical optical modeling efforts, there is an intrinsic degree of variability that can be influenced by functions of season, location, local recent meteorological events, and ephemeral l ighting conditions (Preisendorfer 1986)

PAGE 118

118 However, the empirically derived spectral and broadband optical models developed and applied here were meant to provide an approximation of the historical underwater light field, with the caveat that equations deter mined from statistical regression should not be used widely on other aquatic systems, but are meant to be used as a site specific application for the spring systems they were deve loped for and for other spring systems that produce similar optical character istics. The apparent decline over time (<20 years) in the native submerged aquatic vegetation (SAV) communities and visual water clarity in Rainbow (Atkins 2012) and Weeki Wachee and other spring systems provided the basis for hypothesis that empirical optical modeling results would provide evidence that light attenuation has significantly increased leading to a decline in benthic light availability, especially in the blue band (i.e. 440 nm). The predicted results from the historical optical mo deling aff orded by the development of site specific optical models offered additional insight as to the estimated historical light attenuation and subsequent light field conditions in the two spring systems studied. Contrary to expectations, light attenua tion did not increase in either system. In fact, benthic light availa bility in the green band (%PAR 550) in creased significantly in the lower Rainbow River and is assumed to be increasing in the blue and red wavelengths (440 and 660 nm) as well. These resu lts corroborate d Frazer et al. (2006) results in regard to K d PAR in Weeki Wachee River where K d PAR did not change significantly between two sampling periods (1998 2000 vs. 2003 2005). In effect, the upper portion of Rainbow River is behaving at least opt ically, as H. T. Odum ( 1957a and 1957b ) had described spring systems to be have in a steady state,

PAGE 119

119 with temporally stable physico chemical characteristics. However based on results from this study, the lower Rainbow appears to be changing on a positive not e where it is improving in respect to optical water quality and water clarity. On the other hand, Weeki Wachee is more dynamic spatially and temporally and appears to be m oving in the other direction with degrading optical water quality and spectral light attenuation. In spite of the spectrally specific decline of historical optical conditions overall broadband benthic light availability (%PAR) appears to be slightly increasing in response to the apparent decrease in K d PAR in the upper portions of Weeki Wa chee River. The increase in spectral %PAR in Rainbow River corresponds to the downward trend in chlorophyll a, and turbidity, and consequently lower absorption in the green and red bands that are most affected by those optical water quality parameters. Alt hough spectral %PAR is increasing over time in some instances minimum values for predicted %PAR 440, %PAR 660 and %PAR reached 20% on a few occasions during the POR in the lower Rainbow (at ca. 2 m depth) and Weeki Wachee (at ca. 1.5 m depth) River s Valu es nearing 20% PAR ( requirements established for SAV that are necessary for plant growth and survival (Dennison et al. 1993; Gallegos and Kenworthy 1996; Kenworthy and Fonseca 1996 ; Gallegos et al. 2009), suggesting potentia l optical water quality and water clarity impair ments in these two spring systems The significant increase in color concentrations in the lower Weeki Wachee River explains the significant increase in total absorp tion in the blue band (440 nm). However, t his relationship was not extended to result in a significant increase in lig ht attenuation in the blue band. A lthough an apparent increase in light attenuation at 440

PAGE 120

120 nm was seen, and this was matched with an assumed decrease in benthic light availability at 440 nm. The increase in color and CDOM absorption may be due to these two factors or a combination of both, 1) increased algal growth and autotrophic productivity upstream and resulting biomass decomposition that contributed to overall dissolved organ ic matter (DOM) in the water column (Duarte et al. 2010, Heffernan et al. 2010 b ) and/ or 2) increased connectivity to riparian wetlands that may be providing further allochthonous input s of CDOM into the river downstream. The second scenario may be attribu ted to canals that flow through wetland areas that are connected to downstream Weeki Wachee R iver. M easurements of optical proper ties such as CDOM absorption can be used to examine changes in DOM composition, quality and lability which have been shown in previous studies ( Laurion et al. 2000; Helms et al. 2008; Spencer et al. 2010) In this study, I developed a regression model to predict CDOM absorption slop es (S g ) from color concentration data so that the historical sources of DOM may be estimate d in th e two spring systems Based on the range of predicted S g values ( 0.031 to 0 .015 nm 1 data not shown) over the period of record for both systems, it can be speculated that DOM consists of younger, more labile and more photoreactive low molecular weight m aterials that were created within the system H owever further information regarding DOM quality such as lignin biomarkers are required to make a valid assumption (Spencer et al. 2010) It is important to determine how t emporal and spatial variations in DO M quality

PAGE 121

121 (i.e. estuary) ability to process and cycle nutrients. There is potential for future work regarding biogeochemical and nutrient cycling to evaluate DOM sourc es, quality, and fate as it DOM is processed microbially throughout the river continuum. Fluvial ecologists should be compelled to include investigations to understand how spatial and t emporal variability can lead to changes in allochthonous inputs from so urces such as riparian wetlands and how light availability play s a role in DOM molecular weight by influencing photodegrad ation and transformations. Summary This is the first study that has developed empirical regression relationships to model optical r egimes in Florida spring fed rivers. This study calibrated empirical ly derived spectrally specific optical models to provide greater precision and reduce uncertainty of model predictability of light attenuation in spring ecosystems The predictive understa nding derived from the empirical optical modeling approach employed here improves the ability to evaluate how future changes in optical water quality will affect light availability for primary producers in springs The ultimate goal of this study was to p rovide a resource management framework for predicting light attenuation and a means of estimating the primary factors that are impacting the quality and quantity of light available for benthic biota that greatly depend on it. The modeling results and produ cts from this work can be used to develop site specific light requirements for spring systems particularly relating historical light regimes to a minimum water clarity threshold linking SAV to light availability. Modeling resul ts can be used to determine direct causes of water clarity impairment s on a spectral scale which could aid and provide momentum in optical water quality and ecological restoration efforts.

PAGE 122

122 Figure 3 1. Relationships among partial absorption coefficients (ad440, ag440, and aph440) and total absorption at 440 nm (at440) for A) Rai nbow and B) Weeki Wachee River; aph440 was not significantly related to at440 (p>0.05) and was not shown in B). R 2 = 0.70 p = 0.000 R 2 = 0.68 p = 0.000 R 2 = 0.62 p = 0.000 A R 2 = 0.52 p = 0.000 R 2 = 0.72 p = 0.000 B

PAGE 123

123 Figure 3 2. Relationships shown in A) a d 440 as a function of color and chlor ophyll a (Chla), and B) a g 340 as a function of color, and a ph 440 as a function of Chla in Rainbow River. R 2 = 0.33, p = 0.001 R 2 = 0.40, p = 0.000 R 2 = 0.69 p = 0.000 R 2 = 0.62 p = 0.000 A B

PAGE 124

124 Figure 3 3 Correlations among partial absorption coefficients (a d 550 and/or a ph 550) and total absorption at 550 nm (a t 550) for A) Rainbow an d B) Weeki Wachee Rivers ; a ph 660 was not significantly related ( p >0.05) to a t 550, therefore was not shown in B). R 2 = 0.76 p = 0.000 R 2 = 0.17 p = 0. 029 A R 2 = 0.87 p = 0.000 B

PAGE 125

125 Figure 3 4 Correlations among partial absorption coefficients (a d 660 and a ph 660) and total absorption at 660 nm (a t 660) for A) Rai nbow and B) Weeki Wachee Rivers R 2 = 0.57 p = 0.000 R 2 = 0.66 p = 0.000 A R 2 = 0.64, p = 0.000 R 2 = 0.56, p = 0.000 B

PAGE 126

126 Figure 3 5. Relationships shown in A) a d 550 as a function of chlorophyll a (Chla), and B) a d 660 as a function of Chla, and a ph 660 as a function of Chla in Rainbow River. A B R 2 = 0.44 p = 0.000 R 2 = 0.36 p = 0.000 R 2 = 0.58 p = 0.000

PAGE 127

127 Figure 3 6. Correlations betwe en optical water quality concentrations and attenuation (K d PAR) coefficients in A) Rainbow and B) Weeki Wachee Rivers. R 2 = 0.25, p = 0.015 R 2 = 0.25, p = 0.015 R 2 = 0.13, p = 0.07 B A R 2 = 0.40, p = 0.036 R 2 = 0.13, p = 0.000 R 2 = 0.47, p = 0.000

PAGE 128

128 Figure 3 7. Relationships shown in A) a d 440 as a function of turbidity and a g 340 as a function of color, and B) a d 550 and a d 660 as a function of turbidity and a d 660 as a function of Chla in Weeki Wachee River. R 2 = 0.58, p = 0.000 R 2 = 0.36, p = 0.000 R 2 = 0.48, p = 0.000 R 2 = 0.28, p = 0.000 R 2 = 0.26, p = 0.000 A B

PAGE 129

129 Figure 3 8. Quadratic relationship used to best fit the combined dataset (Weeki Wachee and Rainbow River data combined) is shown for Sg (spectral slope of CDOM ab sorption) as a function of color. R 2 = 0.37 p = 0.000 Sg = 0.0007*[ Color] 2 + 0.0067*[Color] 0.0305

PAGE 130

130 Figure 3 9. Spectral and broadband model outputs showing measured K d compared with modeled K d for all eight model output results for Rainbow River. Note: Measured and modeled spectral K d s include data at three wavelen gths, 440, 550, and 660 nm.

PAGE 131

131 Figure 3 10. Spectral and broadband model outputs showing measured K d compared with modeled K d for all eight model output results for Weeki Wachee River. Note: Measured and modeled spectral K d s include data at three wavelen gths, 440, 550, and 660 nm.

PAGE 132

132 Figure 3 11. Measured K d compared with modeled K d for the selected (most robust) spectral (2c) and broadband (4) model output results for Rainbow River shown in A) and Weeki Wachee River shown in B ). Note: Measured and mo deled spectral K d s include data at three wavelengths, 440, 550, and 660 nm. B A

PAGE 133

133 Figure 3 12. Mean annual color, chlorophyll a (Chla) and turbidity concentration data shown for the period of record (2002 2011) measured by the SWFWMD in each Rainbow River st ation. Note: These data were used as inputs to predict annual average partial absorption, scattering and attenuation coefficients in the spectral optical and broadband models during the same period of record.

PAGE 134

134 Figure 3 13. Mean annual color, chlorophyll a (Chla) and turbidity concentration data shown for the period of record (2005 2011) measured by the SWFWMD in each Weeki Wachee River station. Note: These data were used as inputs to predict annual average partial absorption, scattering and attenuation coefficients in the spectral optical and broadband models during the same period of record.

PAGE 135

135 Figure 3 14. Mean annual predicted total absorption coefficients at 440, 550, and 660 nm wavelengths shown for the period of record (2002 2011) in each Rainbow River station.

PAGE 136

136 Figure 3 15. Mean annual predicted total absorption coefficients at 440, 550, and 660 nm wavelengths shown for the period of record (2005 2011) in each Weeki Wachee River station.

PAGE 137

137 Figure 3 16. Mean annual predicted scattering coefficie nts at 440, 550, and 660 nm wavelengths shown for the period of record (2002 2011) in each Rainbow River station.

PAGE 138

138 Figure 3 17. Mean annual predicted scattering coefficients at 440, 550, and 660 nm wavelengths shown for the period of record (2005 2011) in each Weeki Wachee River station.

PAGE 139

139 Figure 3 18. Percent contribution of each predicted partial absorption coefficient (a d 440, a g 340 and a ph 440), and a w 440 to the total absorption coefficient (a t 440) for the upper Rainbow River stations (RR1, RR2, and R R3).

PAGE 140

140 Figure 3 19. Percent contribution of each predicted partial absorption coefficient (a d 440, a g 340 and a ph 440), and a w 440 to the total absorption coefficient (a t 440) for the middle Rainbow River stations (RR4 and RR5).

PAGE 141

141 Figure 3 20. Percent contrib ution of each predicted partial absorption coefficient (a d 440, a g 340 and a ph 440), and a w 440 to the total absorption coefficient (a t 440) for the lower Rainbow River stations (RR7 and RR8).

PAGE 142

142 Figure 3 21. Percent contribution of each predicted partial absor ption coefficient (a d 550, a g 550 and a ph 550), and a w 550 to the total absorption coefficient (a t 550) for the upper Rainbow River stations (RR1, RR2, and RR3).

PAGE 143

143 Figure 3 22. Percent contribution of each predicted partial absorption coefficient (a d 550, a g 550 and a ph 550), and a w 550 to the total absorption coefficient (a t 550) for the middle Rainbow River stations (RR4 and RR5).

PAGE 144

144 Figure 3 23. Percent contribution of each predicted partial absorption coefficient (a d 550, a g 550 and a ph 550), and a w 550 to the total absorption coefficient (a t 550) for the lower Rainbow River stations (RR7 and RR8).

PAGE 145

145 Figure 3 24. Percent contribution of each predicted partial absorption coefficient (a d 660, a g 660 and a ph 660), and a w 660 to the total absorption coefficient (a t 660) for t he upper Rainbow River stations (RR1, RR2, and RR3).

PAGE 146

146 Figure 3 25. Percent contribution of each predicted partial absorption coefficient (a d 660, a g 660 and a ph 660), and a w 660 to the total absorption coefficient (a t 660) for the middle Rainbow River station s (RR4 and RR5).

PAGE 147

147 Figure 3 26. Percent contribution of each predicted partial absorption coefficient (a d 660, a g 660 and a ph 660), and a w 660 to the total absorption coefficient (a t 660) for the lower Rainbow River stations (RR7 and RR8).

PAGE 148

148 Figure 3 27. Perc ent contribution of each predicted partial absorption coefficient (a d 440, a g 340 and a ph 440), and a w 440 to the total absorption coefficient (a t 440) for the upper Weeki Wachee River stations (WW0, WW0.5, and WW1).

PAGE 149

149 Figure 3 28. Percent contribution of each predicted partial absorption coefficient (a d 440, a g 340 and a ph 440), and a w 440 to the total absorption coefficient (a t 440) for the lower Weeki Wachee River stations (WW3 and WW5).

PAGE 150

150 Figure 3 29. Percent contribution of each predicted partial absorption co efficient (a d 550, a g 550 and a ph 550), and a w 550 to the total absorption coefficient (a t 550) for the upper Weeki Wachee River stations (WW0, WW0.5, and WW1).

PAGE 151

151 Figure 3 30. Percent contribution of each predicted partial absorption coefficient (a d 550, a g 550 and a ph 550), and a w 550 to the total absorption coefficient (a t 550) for the lower Weeki Wachee River stations (WW3 and WW5).

PAGE 152

152 Figure 3 31. Percent contribution of each predicted partial absorption coefficient (a d 660, a g 660 and a ph 660), and a w 660 to the to tal absorption coefficient (a t 660) for the upper Weeki Wachee River stations (WW0, WW0.5, and WW1).

PAGE 153

153 Figure 3 32. Percent contribution of each predicted partial absorption coefficient (a d 660, a g 660 and a ph 660), and a w 660 to the total absorption coefficie nt (a t 660) for the lower Weeki Wachee River stations (WW3 and WW5).

PAGE 154

154 Figure 3 33. Mean annual predicted diffuse attenuation coefficients (K d ) at 440, 550, and 660 nm wavelengths shown for the period of record (2002 2011) in each Rainbow River station.

PAGE 155

155 Fi gure 3 34. Mean annual predicted diffuse attenuation coefficients (K d ) at 440, 550, and 660 nm wavelengths shown for the period of record (2005 2011) in each Weeki Wachee River station.

PAGE 156

156 Figure 3 35. Mean annual predicted percent subsurface spectral irr adiance (%PAR) at 440, 550, and 660 nm wavelengths shown for the period of record (2002 2011) in each Rainbow River station.

PAGE 157

157 Figure 3 36. Mean annual predicted diffuse subsurface spectral irradiance (%PAR) at 440, 550, and 660 nm wavelengths shown for the period of record (2005 2011) in each Weeki Wachee River station.

PAGE 158

158 Figure 3 37. Mean annual predicted broadband diffuse attenuation coefficients (K d ) for PAR and %PAR shown for the period of record (2002 2011) in each Rainbow River station.

PAGE 159

159 Figure 3 38. Mean annual predicted broadband diffuse attenuation coefficients (K d ) for PAR and %PAR shown for the period of record (2005 2011) in each Weeki Wachee River station.

PAGE 160

160 Table 3 1. Spectral and broadband empirical models of diffuse light attenuation in Rainbow and Weeki Wachee Rivers Model Equation n* n** g 1 g 2 Source of Model 1 K d t 2 + 0.2743 a t b) 1/2 84 60 NA NA Kirk 1981 2a K d t 2 + (g 1 g 2 ) a t b) 1/2 84 60 0.425 0.190 Kirk 1984 2b K d 0 [a t 2 + (g 1 0 g 2 ) a t b] 1/2 84 60 0.473 0.218 Kirk 1994 2c K d 0 [at 2 + (g 1 0 g 2 ) a t b] 1/2 84 60 6.114* 9.126** 3.921* 6.718** Modified from Kirk 1994 2d K d 0 [at 2 + (g 1 0 g 2 ) at b] 1/2 )*1/OC 84 60 6.114* 9.126** 3.921* 6.718** Modified from Kirk 1994 3 K d PAR = 0.014 [Color] + 0.062 [Turbidity] + 0.049 [Chla] 28 20 NA NA McPherson and Miller 1994 4* K d PAR = 0.2652 + 0.1636 [Color] 0.3885 [Turbidity] + 0.2097 [Chla] 28 NA NA NA This study 5* K d PAR = 0.0.168 + 0.0417 [Color] + 1.0242 [Turbidity] + 0.0144 [Chla] 28 NA NA NA Modified from McPherson and Miller 1994 4** K d PAR = 0.92 58 + 0.0109 [Color] 0.2946 [Turbidity] + 0.0491 [Chla] 0.5819 OC NA 20 NA NA This study 5** K d PAR = 0.0211 + 0.049 [Color] + 0.071 [Turbidity] + 0.015 [Chla] NA 20 NA NA Modified from McPherson and Miller 1994 denotes the number of samp les used to develop or calibrate the model indicates whether the model was obtained from the cited literature or modified using data from this study; or if the model was derived solely from data collected as part of this study denot es that the item only applies to the Rainbow River denotes that the item only applies to the Weeki Wachee River denotes that the item only applies NA indi cates that it is not applicable for that item

PAGE 161

161 Table 3 1. Not es Continued. Model variable definitions and units: K d (m 1 ) is the vertical diffuse light attenuation t (m 1 1 ) is the scattering c 0 is the cosine of the zenith angle of the direct solar beam refracted at the air water interface, g 1 and g 2 are coefficients that depend on the scattering phase function of the water column and on the optical dep th of interest and were empirically determined for the midpoint of the euphotic zone (Kirk 1991 and 1994), OC is %Open Canopy, [Color] is color concentration (PCU), [Turbidity] is turbidity concentration (NTU), [Chla] is chlorophyll a concentration (mg m 3 )

PAGE 162

162 Table 3 2. Partial absorption coefficient models for detrital, ( a d ), CDOM ( a g ), and algal pigment ( a ph ) at index wavelengths (440, 550, and 660 nm) in Rainbow (RR) and Weeki Wachee (WW) Rivers as shown in Models 6a to 11b. Sys tem Model Equation n p adj R 2 s RR 6a a d 440 = 0.0025 + 0.0127 [Color] + 0.0190 [Chla] 28 0.000 0.45 0.012 RR 6b a g 340 = 0.0097 + 0.0417 [Color] 28 0.000 0.59 0.012 RR 6c a ph 440 = 0.0024 + 0.0378 [Chla] 28 0.000 0.62 0.110 RR 7 a d 550 = 0.0039 + 0 .0139 [Chla] 28 0.000 0.44 0.006 RR 8a a d 660 = 0.0008 + 0.0070 [Chla] 28 0.000 0.36 0.004 RR 8b a ph 660 = 0.0022 + 0.0144 [Chla] 28 0.000 0.58 0.005 WW 9a a d 440 = 0.0086 + 0.1734 [Turbidity] 20 0.000 0.57 0.038 WW 9b a g 340 = 0.0048 + 0.0489 [Color ] 20 0.000 0.99 0.009 WW 10 a d 550 = 0.0141 + 0.0710 [Turbidity] 20 0.000 0.45 0.019 WW 11a a d 660 = 0.0065 + 0.0274 [Turbidity] 20 0.001 0.38 0.011 WW 11b a ph 660 = 0.0022 + 0.015 [Chla] 20 0.014 0.26 0.011 Notes: See Table 3 1 Notes for eq uation definitions and table notations. p denotes the significance level ( p 2 adjusted coefficient of determination,

PAGE 163

163 Tab le 3 3. Total absorptio n coefficient (at three wavelengths 440, 550 and 660 nm) and vertical diffuse light attenuation (PAR) coefficients combined dataset models 12 to 15. System Model Equation n p adj R 2 s Combined 12 a t 440 = 0.0149 + 0.0574 [Color] + 0.1856 [Turbidity] 48 0.000 0.86 0.038 Combined 13 a t 550 = 0.0574 + 0.0060 [Color] + 0.0329 [Turbidity] +0.0243 [Chla] 48 0.000 0.55 0.023 Combined 14 a t 660 = 0.4224 + 0.0185 [Chla] + 0.0239 [Turbidity] 48 0.000 0.60 0.010 Combined 15 K d PAR = 0.3353 [Color] + 0.160 6 [Chla] 0.1361 48 0.000 0.36 0.145 Note: Model e quations (12 14) for the combined springs dataset to predict total absorption at reference wavelengths (440, 550, and 660 nm) directly from optical water quality concentration data to be used in the spec tral optical models are shown Model equation 15 shows the equation for the combined springs dataset to predict vertical diffuse attenuation coefficients for broadband PAR directly from optical water quality concentration data. See Tables 3 1 and 3 2 Notes for equation definitions and table notations.

PAGE 164

164 Table 3 4. Evaluation of model precision metrics for spectral and broadband empirical optical models of diffuse light attenuation (K d ) in Rainbow River. Rainbow River Model Model Precisi on Metric 1 2a 2b 2c 2d 3 4 5 Mean Difference (Measured Modeled) 0.150 0.153 0.129 0.030 0.008 0.324 0.000 0.097 Sum Square Error (SSE) 2.851 10.866 2.567 1.330 2.859 3.209 0.113 0.877 Mean Square Error (MSE) 0.034 0.129 0.031 0.016 0.034 0.115 0.004 0 .031 Root Mean Square Error (RMSE) 0.184 0.360 0.175 0.126 0.183 0.339 0.064 0.177 % RMSE of Mean K d 49.5% 96.6% 47.0% 33.8% 49.3% 92.1% 17.3% 48.1% % RMSE of Max K d 22.3% 43.6% 21.2% 15.2% 22.2% 52.9% 9.9% 27.7% R 2 0.694 0.692 0.670 0.701 0.701 0.489 0.697 0.206 Note: Model precision metrics were calculated based on measured versus modeled K d data. The unit for mean difference, SSE, MSE and RMSE is m 1

PAGE 165

165 Table 3 5 Evaluation of model precision metrics for spectral and broadband empiric al optical models of K d in Weeki Wachee River. Weeki Wachee River Model Model Precision Metrics 1 2a 2b 2c 2d 3 4 5 Mean Difference (Measured Modeled) 0.145 0.151 0.117 0.004 0.077 0.364 0.000 0.028 Sum Square Error (SSE) 3.012 3.103 2.806 1.476 1.889 3.572 0.505 2.178 Mean Square Error (MSE) 0.050 0.052 0.047 0.025 0.031 0.179 0.025 0.109 Root Mean Square Error (RMSE) 0.224 0.227 0.216 0.157 0.177 0.423 0.159 0.330 % RMSE of Mean Kd 52.4% 53.2% 50.6% 36.7% 41.5% 95.5% 35.9% 74.6% % RMSE of Ma x K d 17.6% 17.9% 17.0% 12.3% 13.9% 36.0% 13.5% 28.1% R 2 0.494 0.496 0.448 0.604 0.678 0.233 0.539 0.087 Note: Model precision metrics were calculated based on measured versus modeled K d data. The unit for mean difference, SSE, MSE and RMSE is m 1

PAGE 166

166 Table 3 6a. Evaluation of statistically significant temporal trends using simple linear regressions for measured optical Rainbow (RR, 2002 2011 period of record) and Weeki Wachee (WW, 2005 2011 period of record) River stations. Station Color (PCU) Chla (mg m 3 ) Turbidity (NTU) a t 440 (m 1 ) a t 550 (m 1 ) a t 660 (m 1 ) b440 (m 1 ) b550 (m 1 ) b 660 (m 1 ) RR1 # # # # # # # # # RR2 # # # # # # # # # RR3 # # # # # # # # # RR4 # # # # # # # # # RR5 # # # # # # # # # RR7 # # # # # # # # # RR8 # # # * # # # WW0 # # # # # # # # # WW0.5 # # # # # # # # # WW1 # # # # # # # # # WW3 # # # # # # # # # WW5 + # # + # # # # # Note: Notations for statistical significance fo r Tables 3 6a and 3 6b : (#) indicates no statistically significant or apparent temporal trend ( p >0.05) (* +) indicates an i ncreasing significant temporal trend, ( p < 0.05 ) (* ) indicates a d ecreasing significant temporal trend, ( p < 0.05 )

PAGE 167

167 Table 3 Station K d 440 (m 1 ) K d 550 (m 1 ) K d 660 (m 1 ) K d PAR (m 1 ) %PAR 440 %PAR 550 %PAR 660 %PAR RR1 # # # # # # # # RR2 # # # # # # # # RR3 # # # # # # # # RR4 # # # # # # # # RR5 # # # # # # # # RR7 # # # # # + # # RR8 # # # # # + # # WW0 # # # # # # # # WW0.5 # # # # # # # # WW1 # # # # # # # # WW3 # # # # # # # # W W5 # # # # # # # # Note: Notations for statistical significance for Tables 3 6a and 3 6b : (#) indicates no statistically significant or apparent temporal trend ( p >0.05) (* +) indicates an i ncreasing significant temporal trend, ( p < 0.05 ) (* ) indica tes a d ecreasing significant temporal trend, ( p < 0.05 )

PAGE 168

168 CHAPTER 4 EFFECTS OF THE UNDER WATER SPECTRAL LIGHT ENVIRONMENT ON SUBMERGED AQUATIC VEGETATION B IOMASS, PRIMARY PROD UCTIVITY AND SPATIAL DISTRIBUTION ALONG AN OPTICAL WAT ER QUALITY GRADIENT Introd uction Springs and spring runs are highly productive, societally and ecologically valuable ecosystems. In many springs, submerged aquatic vegetation (SAV) dominate the ben thic community and are a keystone species that play a role in the trophodynamics, bio geochemical cycling and stability of substrates, as well as many other functions (Dennison et al. 1993 ; Knight and Notestein in Brown et al. 2008 ). Light availability is an important abiotic variable for influencing SAV growth and productivity and uptake o f nutrients (Odum 1957a; Canfield and Hoyer 1988; Kirk 1994; Hoyer et al. 2004; Hauxwell et al. 2007; Julian et al. 2008 a b, c ; Knight and Notestein in Brown et al. 2008; WSI 2010 ; Matheson et al. 2012 ). To understand photosynthesis in aquatic environment s, it is necessary to define the underwater light environment as the quantity and wavelength of available photons, their distribution and the direction of the radiation (Falkowski and Raven 2007). In a study done on Kings Bay, a spring fed, tidally influe nced freshwater oligohaline system in Florida, researchers found only one statistically significant relationship (out of 64 regressions) where the relative growth rate of Vallinsneria americana (wild celery) shoots increased linearly with surface irradianc e (Odum 1957a; Hauxwell et al. 2007). This relationship substantiates the conjecture that irradiance is essential for the growth for SAV; moreover the amount of irradiance reaching the vegetation is likely a limiting factor for SAV in aquatic ecosystems.

PAGE 169

169 The percent subsurface irradiance (percent of incident light reaching the substrate %PAR ) has been determined to have a major impact on distribution and abundance of SAV biomass in spring ecosystems (Hoyer et a l. 2004). The % PAR is an important variable that is used extensively as a minimum light requirement or target for maintaining or restoring seagrasses in coastal ecosystems (Dennison et al 1993; Gallegos 1994, 2001; Gallegos and Kenworthy 1996; Gallegos et al. 2009; Dixon and Kirkpatrick 2010). Al on g with the % PAR reaching the bottom and the rate at which it is attenuated, depth limits of seagrass are also strongly linked to the l quality. S pect ral PAR [ PAR ( provides better information on the quality and shape of the incoming light field than simply measuring the quantity of light or PAR (Anastasiou 2009; Gallegos et al. 2009 ) Even though the photons of light (PAR or spectral PAR) are available, it is not certain that the available light is actually being absorbed by photosynthesis harvesting pigments in the plants (Kirk 1994) The distribution of PAR in aquatic systems is of fundamental importance in determining the growth and survival of SAV (McPherson and Miller 1994; Miller and McPherson 1995; Gallegos et al. 2005). The wavelengths of light used when expressing PAR are 400 700 nm, and are measured as photon flux or quantum irradiance in units of mol m 2 s 1 (Kirk 1994; Knight and Notestein in Brown e t al. 2008; Anastasiou 2009). The underwater light environment has a spectral composition that is controlled by absorption and scatter by water and dissolved and particulate constituents (Kirk 1994). The quantity and spectral quality of light affect the po tential for photosynthesis and spatial distribution of primary producers in aquatic systems (McPherson and Miller 1994).

PAGE 170

170 Kirk (1994) remarks on photosynthesis as a function of light for plants and algae in aquatic systems. He expressed that the relations hip between incident light characteristics and the rate of photosynthesis is not simple and is dependent on the rate of capture and efficiency to make use of the absorbed energy to fix carbon dioxide (CO 2 ). For that reason, it is necessary to study the eff ects of light intensity and spectral quality of photosynthesis via quantitative methods to determine the photosynthetic rate per unit biomass (Falkowski and Raven 2007) The spectral distribution of PAR through a water column must correspond to the absorp tion spectrum of a given species of SAV or algae for photosynthesis to occur ( Kirk 1994 ). Light harvesting pigments such as carotenoids and chlorophylls in higher plants typically absorb light at the blue and red wavelengths, and only red algae, blue green algae, and the cryptophytes contain billiproteins that allow them to harvest green light ( Kirk 1994 ; Falkowski and Raven 2007; Anastasiou 2009), which indicate that higher plants such as SAV absorb the blue and red bands and not the green wavelengths. The refore, it can be deduced that wavelengths of light that are poorly absorbed by plants, s uch as the green wavelengths (50 0 to 600 nm), are relatively inefficient at driving photosynthesis. ty (Dennison et al. 1993), where flourishing seagrass communities are used as a signal for a productive and diverse coastal ecosystem, and conversely, an absence of seagrasses may be an indicator of degradation of estuarine water quality and light availabi lity (Biber et al. in Bortone 2005). Seagrass es have very high light requirements, which make them sensitive indicators of water quality degradation. Seagrasses (11 37% PAR

PAGE 171

171 subsurface irradiance ) have much higher light requirements than macroalgae o r bent hic microalgae (0.05 3%, Dennison et al. 1993; Biber et al. in Bortone 2005; Gallegos et al. 2009). Rooted macrophytes require more light to supply energy to non photosynthetic tissues to promote re generation, whereas algal species grow vegetatively most of the time and have fewer cells to maintain than more advanced SAV (Sand Jensen and Borum 1991). A study in spring fed streams in Florida found that SAV was restricted to locations in the rivers where at least 10% subsurface light reached the substrate (H oyer et al. 2004). This minimum light requirement value falls within the range of literature values between 4% and 29% PAR for freshwater SAV (Dennison et al. 1993). Due to the high minimal light requirement of both freshwater and marine SAV, good water cl arity is extremely important for SAV. This fact provides the basis for focusing on light attenuation processes to explain the effect of the underwater light field on the distribution and productivity of SAV. An ecological shift has been documented in sev eral spring ecosystems, where macrophyte dominated communities are declining and algal communities are proliferating (Hoyer et al. 2004; Hauxwell et al. 2007; Quinlan et al. 2008 ; Stevenson et al. 2008; H effernan et al. 2010; King 2012 ). In Rainbow Springs and River in northern Florida, SAV have been important primary producers of the ecosystem A bundance of the dominant native keystone SAV species S agittaria kurziana (SAG) has declined over time while fi lamentous benthic and epiphytic algal abundance ha s increased ( Atkins 2012 ) which may have led to a shift in community metabolism rates. However

PAGE 172

172 phototrophs such as epiphytes and benthic algal mats are also important members of the primary producer community (Odum 1957a). This ecological shift could b e partly attributed to recreational effects, which have increased over time and have caused noticeable damage to vegetation communities in spring systems ( Dutoit 1979; Mumma et al. 1996; Cichra et al. 2013) Other potential factors that have been suggested as causes of changes in primary producer abundance and distribution include nutrient enrichment (Odum 1956; Mullholland et al. 2001) reduced flow discharge and velocities (McCutchan et al. 2002 ; King 2012 ) declines of fish and other consumer biomass (Lie bowitz 2013) and aquatic plan t management (Brown et al. 2008 ; WSI 2010 ) In a few studies on spring systems, nutrient concentrations were not found to be related to benthic filamentous algal or SAV abundance or primary productivity rates (Canfield and Hoy er 1988; Duarte and Canfield 1990; Hoyer et al. 2004) although the combined effect of dissolved oxygen concentration and algal grazer abundance were found to be good predictors of algal abundance (Heffern an et al. 2010a ; Liebowitz 2013 ). Also, investigati ons into the effect of hydrology on algal abundance demonstrated t hat flow velocities above 35 cm s 1 were able to constrain certain benthic filamentous algae (King 2012). Benthic algal mats have very high specific rates of photosynthetic and respiratory processing, due to the dense packing of chlorophyll, and differen t spectral absorption characteristics (Sand Jensen and Borum 1991), and t hese photosynthetic rates can exceed SAV ra tes by more than 100 fold In Florida, researchers determined that there h as been a significant shift in primary producers in the Silver Springs (considered to be

PAGE 173

173 system over the last 50 years with a rise in prominence of epiphytic and bent hic algal mats (Quinlan et al. 2008). As epiphytic material and benthic algal mats proliferate and spread, light availability is directly limited to the SAV from shading by algae and can reduce habitat availability for SAV (Hauxwell et al. 2007) hence ind irectly affecting primary production Light, and correspondingly photosynthesis, varies in intensity and photoperiod by several factors and is a function of depth, riparian canopy shading, time of day, and season (Odum 1957b). All of these factors interrel ate to provide the basis for the level of primary production rate that the community is capable of achieving. Even though light is the major driver of photosynthesis (Odum 1956; Kirk 1994; Mullholland et al. 2001; Julian et al. 2008c ) other factors such a s nutrients ( Odum 1956; Mullholland et al. 2001) flow, substrates organic matter, and temperature ( Odum 1956; Biggs et al. 1999 ; Julian et al. 2008c ) can also control primary productivity Calculating ecological efficiencies (EE) is one such method of l inearly relating the efficiency of photosynthesis relative to the incoming light in usable wavelengths that is reaching the primary producers ( Odum 1957b). Therefore, it is possible to compare streams spatially or temporally with this method, in which a hi gher EE in a particular the energy of available sunlight for gross primary production (GPP WSI 2010 ). Odum 1956, 1957a, 1957b) documented that only a small fraction, on ly 4% of inc ident radiation was used in GPP which may be artificially low efficiency estimate because he used incident light that did not account for light attenuation by the water column and was an overestimate of actual benthic light availability The c onversion of incident

PAGE 174

174 radiation for primary production is also called photosynthetic efficiency (Knight and Notestein in Brown et al. 2008 ; WSI 2010 ). In some Florida springs and their associated spring runs the distribution and abundance of SAV has been related to light availability in terms of light quantity (Frazer et al. 2001 a ; Hoyer et al. 2004). However, the quality of light (spectral distribution of PAR) had not yet been investigated as a potential determinant for growth, productivity and spatial di stribution. This study quantitatively evaluate d the spectral and spatial dist ribution of light and primary producer physiolog y and morpholog y of two Florida spring ecosystems Interactions between these characteristics were explored to elucidate the causes of ecological shifts and to assess the potential for light limitation in these spring systems Methods Study Area Rainbow and Weeki Wachee River s are two spring fed rivers that are located near the west central coast of Florida, USA S tudy segments, stat ions and l ocation map s of the study area s are shown in Figures 2 1 and 2 2, and more detailed descriptions of the two systems can be found in the Methods section in Chapter 2. Six study segments were located on Rainbow ( RRS1 to RRS6 n=24 ) and four segmen ts on Weeki Wachee Rivers ( WWS1 to WWS4 n=16 ) for a total of 10 segments (n=40) Some analyses required segment averages (i.e. metabolism analyses ), which were obtained by calculating the mean of two consecutive data points starting with the most upstrea m sampling station to the most downstream station (Figures 2 1 and 2 2) F or example, station RR1 and RR2 data were averaged to obtain RRS1 data, station WW0 and WW0.5 data were averaged for WWS1 data, and so forth. Four quarterly

PAGE 175

175 sampling events for all p arameters were carried out during a one year study period ( January to December 2011) which was separated into four seasonal quarters as follows, Quarter 1: January to March, Quarter 2: April to June, Quarter 3: July to September, and Quarter 4 : October to December. The extensive sampling program did not allow for concomitant sampling of all parameters. O ptical, physico chemical, benthic and primary product ivity characteristic s w ere all collected within the same seasonal quarter but were not always collect ed within the same month. Water Column O ptical Properties and Physico chemical Characteristics Th e Methods section s in Chapter s 2 and 3 provide d the methods that were used to collect and analyze the in stream optical properties ( benthic PAR d d PAR g S R ) and canopy coverage (OC) Water chemistry concentration data (turbidity, color, chlorophyll a, nitrate ammonium, and total phosphorus) were W ater Management Information System (WMIS) database. D i scharge and velocity data were obtained from one or two United States Geological Survey (USGS) flow gages ( Lower Rainbow River: USGS # 02313100, upper and lower Wee ki Wachee River: USGS #02310545 and #02310525 respectively ) on each river Flow discharges a re not regularly measured at any of the SWFWMD water quality monitoring station s on either system However, manual flow discharge (m 3 s 1 ) and velocity (m s 1 ) measurements were available from SWFWMD for each water quality monitoring station (Figure 2 1) i n Rainbow River for one occasion. Weeki Wachee River is primarily fed by one large spring boil in the headspring area. Minor (<10%) additional inputs occur along the river, primarily during high flow events, which were assumed to contribute a minimal amou nt of water flux as compared to the total flow emerging from the spring boil. This permitted the direct use of the

PAGE 176

176 USGS gage flow data to be applied to each station. USGS #02310545 data was assumed to be representative of the u pper (WW0, WW0.5, WW1) statio ns, and USGS #02310525 characterized the lower (WW3, WW5) Weeki Wachee River station s adequately due to proximity and nominal additional inputs. However, s everal relatively large spring s and dozens of smaller inputs (~ 13 27% of total river flow for each g roup of springs per segment) exist at varying points along the upper Rainbow River (RR1 to RR4, Figure 2 1) that contribute an unknown water flux to the overall river flow discharge as represented by the USGS gage in the l ower river This system was more h ydrologically complex than Weeki Wachee River and the simpl e assumption that the USGS gage flow discharge estimates in the lower river would be representative of each station was not valid. Additionally, to obtain accurate p rimary production estimates (see below) and to account for groundwater inputs within each segment, uniformly accurate and more finely resolved flow discharge estimates for each station was required. Thus, the manually measured flow discharge data available from SWFWMD for each station w ere used to determine the fraction of total flow discharge at each station with respect to the USGS gage flow discharge estimates A fraction coefficient was calculated for each station and appl ied to the USGS gage fl ow data to estimate flow discharge at e ach station during the study period S tream segment width s (m) and length s (m) were determined using ArcMap GIS software (Environmental Systems Research Institute, Inc., 2011). Bottom d epth (m) was estimated with a double graduated Keson measuring tape ( Model OTR10M300) with a vertical Secchi disk attached to the end of the tape to act as a weighting anchor Shading due to riparian canopy cover was determined at each station on two occasions

PAGE 177

177 (July and October) with a spherical densiometer (Model C, Forest Densiometers, Rapid City, SD; Lemmon 1957). Benthic Sampling Benthic sampling included qualitative and quantitative measurements of bottom substrates, submerged aquatic vegetation (SAV) and ass ociated epiphytes and benthic filamentous algae (FA). For bo th systems, benthic sampling occurred in March (quarter 1), May (quarter 2), August (quarter 3). Quarter 4 sampling was conducted in October and December for Weeki Wachee and Rainbow Rivers, respectively. All benthic samples were collected from seven and f ive transects on Rainbow and Weeki Wachee Rivers, respectively. Transects were co located with water quality and optical stations (Figures 2 1 and 2 2) Sediment type, e p iphyte type and load, SAV and algal a bundance (percent cover age ) were assessed by div ers at five points along each transect and results were averaged The five sampling points were equally spaced with one in the center of the river channel and two to either side of the center point with the transect running from bank to bank. The t ransect was positioned perpendicularly to river flow. Sediment type was visually evaluated by assigning sediment type to the following categories, sand, silty sand, silt, mud, or rock. The type (e.g. diatom or filamentous) and extent of epiphytic cover (load) on SAV was qualitatively (e. g. low, medium, heavy cover) describe d. SAV and FA percent cover age was estimated at each point on the transect using at a 1 m 2 quadrat Percent cover age was estimated visually on a percent coverage scale (PCS) in 5% increments up to 100% per species observed within each quadrat

PAGE 178

178 At each of the transects, an a bove sediment biomass sample w as collected by randomly tossing a 0.25 m 2 quadrat and all plant and algal biomass within the quadrat was carefully removed and transported to t he laboratory for processing (Canfield and Hoyer 1988; Hoyer et al. 2004). Each biomass s ample underwent several steps of laboratory processing that consisted of: 1) separation of vegetation by species ( Sagittaria kurziana SAG; Vallisneria americana VAL; Potamogeton illinioensis, POT ; Najas guadalupensis NAJ; Ceratophyllum demersum, CER; Hydrilla verticillata HYD ; Utricularia spp ., UTR; Ludwigia repens LUD; Chara spp., CHA; Myriophyllum heterophyllum MYR; filamentous algae, FA were not identified to s pecies and remained at the categorical level ) ; 2) careful removal of epiphytic material ; 3) blot dry ing the samples for fresh wet weight (FW) determin ation and drying the samples in the oven for ~24 h at 100 C to determine a dry weight ( DW ) and mass per a rea Additional analyses were conducted for the gras s like SAV species SAG and VAL Shoot density was determined by counting shoots f or all SAG and VAL plants in each sample L eaf blade lengths and widths were measured and counted per shoot and averaged for 3 to 4 typical shoots Light Absorption Properties of Primary Producers A bsorptan ce (relative absorption curves) values and pigment contents (chlorophylls and carotenoids) were determined for all SAV species and algal assemblages encountered at each st ation in both spring systems including: FA (dominated by Lyngbya wolleii and/or Vaucheria spp. ), 9 SAV species ( SAG, VAL, POT, NAJ, CER, HYD, UTR, LUD, AND CHA) and for the epiphytic communities associated with SAG and V AL Light absorption properties were determined on one occasion for

PAGE 179

179 freshly collected samples collected in July 2011 in the laboratory on a bench top dual beam UV/VIS spectrophotometer ( Perkin Elmer Lamda 40 ) for the species listed above. Samples were processed and analyzed in the laborator y following Su et al. 2010 by the following steps: 1) healthy plants were selected and necrotic material removed; 2) epiphytic material was carefully removed and retained for individual analysis; 3) plant samples were cut into pieces between 9 16 mm 2 ; 4) s amples were measured for total area (to the nearest cm 2 ) and mass ( to the nearest mg); 5) plant and algae cells underwent physical rupture by the grinding settling method with 8 0% acetone to extract pigments into solution; 6) ground sample s in acetone solu tion settled in the dark for 24h at 4 C; 7) volume was measured pre and post centrifuging to obtain a supernatant sample; 8) supernatant was transferred to a 1 cm cuvette; 9) pigment solution absorbance values from 350 to 800 nm wavelengths were measured s pectrophotometrically Absorbance values were converted to photosynthetic absorptances for SAV [A L and algal assemblages [A E according to Kirk (1994) by Equation 4 1, followed by a correction for non photosynthetic absorptance at 750 nm (Drake et al. 2003; Durako 2008). A L or A E 10 Equation ( 4 1 ) Equations reported by Lichtenthaler ( Acetone, 80% v/v; 1987) were used to quantitatively determine c hlorophyll a, b and total carotenoid s ( x+c xanthophylls and carotene ) concentration (g g 1 fresh weight) To compare results with the rigorous laboratory measurements, a simpler f ield method was conducted to measure absorptance on a portable spectrometer ( m odel

PAGE 180

180 HR2000, Ocean Optics, Dunedin, FL) connected to a fiber optic cable and black mounting bracket on a quarterly basis. Incoming solar irradiance (near solar noon) was scanned first ( mean of 3 scans) and between each set of leaf scans Immediately following each solar radiation scan intact SAG and /or VAL leaves ( mean o f 3 to 4 leaves) were scanned for absorbance by put ting the leaf into contact with the bitter end of the fiber optic cable while mounted onto the bracket (Anastasiou 2009) Absorbance values were converted to absorptance using Equation 4 1. Photosynthetica lly useable radiation is defined as the fraction of radiant energy that can be absorbed by a given plant and wa A L Morel 1978 ) B roadband and s pectral ly specific light attenuation due to epiphytic materials that included attached epiphytic algae, epifauna, and associated sediment or detritus on SAG leaf blades was determine d quarterly at each transect in both systems. The procedure used here was a slightly modified form of the epip hyte attenuation method used by Dixon and Kirkpatrick (1995 and 1999). Shoots that were collected for the biomass measurements were used to collec t epiphytic samples. All epiphytic material was scraped from one shoot from each transect and placed into its respective petri dish with a minute amount of deionized (DI) water to evenly suspend the epiphytic material. The intent was to match the dish area to a typical shoot area when possible (Dixon and Kirkpatrick 1995 and 1999). However some of the shoots from deeper transects occupied more area than the dishes that were available in the laboratory, which required an area normalizing calculation. The t ransparent petri dish with the scraped epiphytic sample slurry was placed on top of a clear plexi glass sensor tube that was lined with reflective mylar. The sensor

PAGE 181

181 tube housed the cosine spectral PAR sensor that was connected to a portable spectrometer by a fiber optic cord (described previously in Chapter 2, Methods). Epiphytic light attenuation measurements were conducted under ambient sunlight as near to solar noon as possible under a white acrylic diffuser plate. Prior to and following each epiphyte at tenuation measurement, a spectral blank reading of the irradiance passing through clean DI water was recorded (mean of three spectral scans) that was immediately followed by spectral readings of the irradiance transmitted through the epiphyte sample (mean of three spectral scans). The percent reduction of transmitted irradiance (% epiphyte attenuation) per wavelength of the epiphytic material per SAG shoot was calculated by dividing the light transmitted through the sample by the light transmitted through the blank. The attenuation due to epiphytes was represented as spectral light attenuation per shoot and was also determined as an area normalized condition. This approach allowed a simple and quick method to approximate field a bsorptance (A E ) values witho ut requiring the rigor of algae processing for pigment analysis. Absorptance by epiphytic material was calculated by converting the raw absorbance values in to absorptance using Equation 4 1. Ecosystem Metabolism Whole stream and per segment e cosystem meta bolism was estimated using a modified open channel two s tation method (Odum 1956) to determine g ross primary production ( GPP, g O 2 m 2 d 1 ) and ecosystem respiration (ER, g O 2 m 2 d 1 ). The two station rather than the one station method was chosen because some of the upper river segments were relatively heterogeneous with various spring vent inputs, which could introduce errors in metabolic rate estimates (McCutchan et al. 2002). All 10 segments

PAGE 182

182 (mean = 1650 m and min = 700 m) far exceed the minimum reach l ength required (ca. 20 m in shallow streams ) to detect a significant change in metabolic rates per segment (Riley and Dodds 2013). However, the two systems are relatively deep, thus the length scale had to be predicated on specific discharge rates instead Net ecosystem production (NEP, g O 2 m 2 d 1 ) was calculated by subtracting ER from GPP P: R ratios (GPP/ER) were calculated to determine the relative magnitude of primary production to respiration (Fellow et al. 2006) and to classify each segment as autot rophic (P:R>1), heterotrophic (P:R<1) or balanced (P:R ~1, Odum 1956) Dissolved oxygen (D O) concentration and percent saturation were measured at an upstream and downstream station of each segment with either a n YSI 6600V2 or YSI 600XLM in situ multipara meter sonde (Yellow Springs Instruments, Yellow Springs, OH) every hour over a 3 to 4 day deployment period GPP and ER were calculated for each day (24 h) using the DO rate of change technique ( downstream DO up stream DO, accounting for diffusion and accr ual ; Odum 1956 ) and then averaged for each segment over each quarterly deployment period. Changes in DO are due to several factors including metabolism (photosynthesis and respiration), reaeration, and lateral inflows from groundwater accrual These factor s need to be accurately accounted for and several researchers have made attempts to correct metabolism estimates for DO changes that are not due to photosynthesis or respiration (Odum 1956; Marzolf et al. 1994 ; Young and Huryn 1998; McCutchan et al. 2002; Hall and Tank 2005; Riley and Dodds 2013 ) The difficult y and cost of directly measuring reaeration by using dissolved tracer gases (e.g. propane) has prompted some researchers to estimate reaeration rates by using empirical models based on

PAGE 183

183 diurnal oxygen curves or hydraulics which have worked in some streams, but have resulted in i na ccurate estimates in other streams (sensu Marzolf et. al. 1994 ; Riley and Dodds 2013 ). To calculate a reasonably accurate metabolic rate of DO change, s ome assumptions were n ecessary to minimize effects on metabolism estimates from reaeration (i.e. diffusion) and groundwater accrual ( Hall and Tank 2005; Riley and Dodds 2013) Oxygen diffusion rates (K, g O 2 m 2 hr 1 ) were estimated by using a linear relationship developed using 21 Florida spring systems with similar physical (e.g. low relief topography and stream slope ) and hydraulic characteristics. The regression equation (K = 0.0604 v + 0.0929 R 2 = 0.84 ; Munch et al. 2006 ) was developed with measured (N 2 filled floating dome technique ; Copeland and Duffer 1964 ) diffusion rates as a function of m easured surface velocity rates (v, cm s 1 ) The rate of diffusion is also dependent on the degree of saturation, where under saturation causes oxygen to diffuse in to, and super saturat ion causes oxygen to diffuse out of the water (Odum 1956) To limit the introduction of bias to metabolism estimates from t he influx of oxygen due to accrual of groundwater along each segment accrual was estimated by subtracting downstream from upstream flow discharge to obtain groundwater flux into the reach ( McCutchan et al. 2002 ; Hall and Tank 2005; Fellows et al. 2006) Groundwater flux was coupled with m easured available spring vent DO concentration data within each reach that was obtained from the S WFWMD WMIS database This was done on a limited basis, and was only necessary for the segments that had obvious accrual from spring vents that occurred within segments.

PAGE 184

184 Mean hourly broadband solar radiation data (W m 2 ) were obtained from the Florida Auto mated Weather Network (FAWN, http://fawn.ifas.usf.edu) Oklawaha (#280) station located in Marion County, FL. Solar radiation data were converted to surface irradiance (PAR, mol photons m 2 s 1 ) by using a conversion factor of 4.6 (Duncan 1990) Raw e colog ical (photosynthetic) efficiency (EE R % and g O 2 mol 1 ) was calculated by dividing daily GPP (mol O 2 m 2 ) by daily raw PAR ( PAR R mol photons m 2 d 1 ) The raw EE R uses t he light energy that reaches an un shaded surface on the ground Depending on canopy coverage and light attenuation within the water column at each station, the raw PAR data from the FAWN station may not be entirely representative of the light reaching the benthic zone in a stream. Therefore, it was useful to correct raw EE R based on the l ight reaching the benthic zone (PAR B ), which was determined by accounting for light attenuation by the water column (%PAR, Chapter 2) and shading by canopy cover ( OC, Chapter 2 ) EE B was calculated by dividing GPP by the corrected PAR B Statistical Data A nalysis Statistical analyse s of the data were performed using MINITAB statistical package. Prior to statistical analysis, data were evaluated for assumptions of normality and homogeneity of varian ce. If normality or homogeneity of variance tests failed, th en data were transformed to fit a normal distribution or non parametric analyses were used. cor relation linear and non linear regression analyses were conducted to determine statistically significant ( p < 0.05) relationships among optical, physico chemical, benthic and primary productivity characteristics. Statistical differences among various biotic and abiotic variables at different stations and in different seasons were analyzed individually using one way ANOVA The Tukey

PAGE 185

185 method was used to de termine which groups of quarter or station means were significantly different. Statistical s ignificance for all analyses were assessed at p < 0.05. Results Water Column Apparent Optical Properties and Physico chemical Characteristics A pparent optical properties ( AOPs; benthic PAR, PAR B in blue, green a nd red bands were all normalized for full solar noon s unlight assuming no cloud cover; %PAR in blue, green and red bands ), f low, nutrient concentrations (nitrate, NO 3 ; ammonium, NH 4 + ; tot al Kjeldahl nitrogen, TKN; total phosphorus, TP ) and chlor ophyll a (Chla) were evaluated for differences in respect to season and station location in both Rainbow and Weeki Wachee Rivers Rainbow River Table 4 1 shows statistical differences in means of chemical parameters among stations in the river. Chla was significantly lower in the upper river, intermediate in the middle river and highest in the two most downstream stations. NO 3 and TP had significant differences among stations where NO 3 decreased and TP increased with distance from the headsprings. TKN and NH 4 + did not differ among sites, but NH 4 + was the only chemical parameter that was significantly different among seasons where it was higher in the third and fourth quarter than the first and sec ond quarters (ANOVA, p = 0.000) A pparent optical properties (AOPs) were not found to vary seasonally (ANOVA, p > 0.05). Mean %PAR and %PAR Blue were significantly different among stations, with stations RR1 and RR4 having higher %PAR than RR8 and RR1 to RR4 having significantly higher %PAR Blue than RR7 and RR8 ( Table 4 2) The dec line in %PAR for broadband, blue and green bands with distance from the headsprings was shown in

PAGE 186

186 Figure 4 1. Only PAR B in the blue band was statistically d ifferent across station s, with RR1 to RR4 having higher PAR B than RR7 (Table 4 2 ). Weeki Wachee River Table 4 3 shows statistical differences in means of chemical parameters in relation to station location in the river. Chla varied significantly across Weeki Wachee River. Chla was significantly higher in the most downstream station (WW5) than the furthest upstream station ( WW0 ) and midway down the river ( WW3 ) but similar to the other upper two river stations ( WW0.5 and WW1). NO 3 decreased significantly with distance downstrea m from the headsprings. No significant spatial differences were found for mean TKN or NH 4 + Mean TP was only statistically different in WW3, which had the lowest mean TP measured in the river. The only chemical parameter that was found to be statistically significant for seasonal variation was NH 4 + ( ANOVA, p = 0.001 all other chemical parameters: p > 0.05 ). The first and second quarters were significantly lower than quarters 3 and 4 which was the same trend as in Rainbow River F low was significantly dif ferent across seasons. The fourth quarter had the highest mean daily flow and quarters 1 and 3 had the lowest flow (Figure 4 2 ). Flow did not vary spatially in Weeki Wachee River (ANOVA, p > 0.05) No AOPs were found to vary seasonally (ANOVA, p > 0.05) Mean %PAR, % PAR Blue and %PAR Green declined significantly with distance from the headsprings (Figure 4 1), with %PAR Blue declining at the fas test rate per meter ( 0.004 %PAR Blue m 1 ). Although according to one way ANOVA resul ts shown in Table 4 4 o nly %PA R Blue means were significantly different, with station WW0 and WW0.5 higher than WW5 Mean broadband PAR B and PAR B in the blue, green and red bands were statistically

PAGE 187

187 different across stations and all decreased along the longitudinal gradient (Table 4 4 ). The m ost downstream station (WW5) had significantly lower broadband and spectral benthic PAR which is likely due to heavier canopy shading than at the other stations. Species Composition and Distribution Relative abundance (% coverage) and biomass of SAV and FA were measured along a longitudinal gradient to evaluate the spatial and temporal variability of species composition and distribution in Rainbow and Weeki Wachee Rivers. Biomass and % coverage were significantly correlated for SAG and FA as shown in Figure 4 3. Taxa richness (number of taxa) was not significantly variable across stations or seasons for either system (ANOVA, p > 0.05). Rainbow River Relative Abundance. R elative abundance was measured by estimating the percent (%) coverage of SAV spec ies and FA at each transect ( Figure 4 4 ) Spectral %PAR minima were also shown for each station in Figure 4 4 Nine native species, one non native, and filamentous algae were recorded and are shown in these categories in Figure 4 5 POT was dominant in RR1 SAG was dominant in RR2 and RR3, and FA was co dominant in the middle river and dominant in the lower river. Weeki Wachee Relative Abundance Figure 4 6 shows the spatial distribution of mean annual % coverage of SAV species. Native t axa richness was 3 fold lower in Weeki Wachee than in Rainbow River. Only WW0 had considerable SAG coverage and FA seemed to dominate WW0.5 and WW1 and WW5 Non native taxa (HYD) dominated WW3 (Figure 4 6). Rainbow River Biomass. Within Rainbow River s patial variation of dry weight (DW) biomass was evident for one SAV species and FA (Table 4 5 and Figure 4 8 ) However, statistically significant temporal variation across seasons (quarters) was not

PAGE 188

188 observed for any of the species studied (ANOVA; p >0.05 for all SAV species and FA) Similar findings to % coverage were found for biomass where SAG dominated the upper river and FA dominated the lower river. Differences among transects were found for m ean SAG biomass (ranged from 0.0 to 161.1 g m 2 ) SAG was highest in transect R R3, lowest in RR5, and was not found in RR7 or RR8 Mean VAL biomass was only found in three transects RR2, RR4, and RR5. Mean VAL biomass ranged from 0.0 to 53.6 g m 2 and was not significantly different across sites (Table 4 5 and Figure 4 8 ) Three spe cies were only collected f or biomass in one transect: POT was only found in the headsprin gs areas in the RR1 NAJ was only found in RR4 in the middle portion of the river and CER was only found in RR5. T herefore a statistical assessment was not appropriat e to determine if these species were spatially variable ( Table 4 5 ) Mean HYD biomass ranged from 0.0 to 5.4 g m 2 with the highest mean biomass in RR7 but was not found to vary significantly between transects. Mean FA biomass was found to vary significa ntly across sites, with RR8 having the highest DW biomass of 4 35.8 g m 2 Weeki Wachee River Biomass. Table 4 6 and Figure 4 9 show the spatial variation of SAV and FA. SAG co dominated with FA in WW0, and FA dominated WW0.5 and WW1. HYD was only found in WW3, similar to the % coverage results. SAG, NAJ and HYD were significantly different among transects (Table 4 6). Overall primary producer biomass values were five fold lower in Weeki Wachee than in Rainbow River. Figure 4 10 shows significant relationshi ps between biomass and % coverage for SAG and FA for combined system datasets.

PAGE 189

189 Species Morphology and Physiology Morphological Variation of Sagittaria kurziana The vegetative structure, specifically the above ground portion of the shoot system of SAG was examined for spatial variation. Morphological characteristics such as leaf blade count (per shoot), length, width, shoot density (number per m 2 ), and shoot weight were measured at each transect in both systems and summary statistics are given in Table 4 7 Mean shoot weight and blade length s were 6 fold and 1.6 fold larger respectively, in Rainbow than in Weeki Wachee River which is likely due to overall greater water depths and siltier sediments in Rainbow River The mean number of leaves per shoot was similar for both systems. Mean blade widths and shoot density w ere both two times higher in Weeki than in Rainbow River which is likely due to lower mean water depth in Weeki Wachee River Shoot weight and associated epiphyte weight were significantly pos itively related for the combi ned datasets for both systems ( Figure 4 1 0 ). Light Absorption Properties of Primary Producers Physiological characteristics such as relative absorptance and pigment content were investigated for SAV and FA species in both sys tems to determine which environmental factors influence physiological differences of SAV and algae Figures 4 1 1 4 1 2 and 4 1 3 show relative absorbance of each species found, spatial variability of relative absorptance of SAG (%AL) and absorptance of SAG associa t ed epiphyte s (%AE) respectively, in Rainbow River. Certain species had similar absorbance scans and were grouped together to compa re absorbance peaks (Figure 4 1 1 ). Distinct groupings were f ound for 1) POT, UTR, and SAG, 2) SAG, CER, and VAL, 3 ) NAJ a nd LUD, and 4) CHA, HYD and FA based on the magnitude and extent or width of peaks in the blue and red

PAGE 190

190 regions. The SAG absorbance scan was similar to SAV in both groups 1) and 2), hence its inclusion in both groups. In terms of absorptance, SAG ha d a significant broad peak across the blue and a second sharper peak in the red ban d for all stations with RR3 having the highest absorptance (Figure 4 1 2 ) A lthough inter s ite variability was rather low, aside from the spread between sites in the gree n red region (500 to 650 nm) Each of those peaks correspond ed to regions associated with chlorophyll and carotenoid pigment light absorption. The green region had considerably lower %AL for all absorption spectra Epiph yte absorptance was highly variable, indicating greater inter site variatio n among epiphyte as semblages (Figure 4 1 3 ) High levels of absorptance were seen across the PAR spectrum for mo st of the sites except for RR5. RR5 did not have high absorptance in the green band and clearly consisted of a different epiphyte community than the other four sites. Pigment content, specifically chlorophyll a (Chla g g 1 ) and total carotenoids (x+c g g 1 ) inter species and inter site var iability is shown in Figure 4 1 4 Chla content ranged from 157.9 to 1790.1 g g 1 and total caroteoids ranged from 8.1 to 371.1 g g 1 No distinct spatial pattern of pigment content could be deduced based on the data. However, NAJ, HYD and LYN seemed to contain higher levels of carotenoids across transects and rivers. Significant ( p <0.05) positive relationships are shown in Figure 4 1 5 between %AL and %AE ( in the blue, green and red bands ) with chlorophyll a content (Chla fresh weight ) measured in SAG and associated epiphytic assemblages, respectively. The strongest rel ationships were found for epiphytes (i.e. %AE vs Chla of epiphytes)

PAGE 191

191 specifically in the green and red bands and the weakest correlation was in the blue band. %AL for SAG relationships were lower than epiphytes, likely due to SAG also having considerable c arotenoid pigments that harvest light in similar wavebands as chlorophylls. A significant positive logarithmic relationship was found for epiphyte weight per shoot and epiphyte absorptance at reference wavelengths 440, 550, 6 60 and for PAR (Figure 4 1 6 ). A E%440 (blue band) was most strongly correlated with epiphyte weight per shoot. Absorptance by algae entrained in the water column (plankton) was also evaluated in Rainbow River (Figure 4 1 7 note different magnitude on y axes ). RR1 (uppermost station) had the lowest absorptance. The mid river station (RR4) had relatively low absorptance as well (mean ca. 1 0%) RR8 (lowermost station) had the highest absorptance (mean ca. 20%) within the main river channel, which was likely influenced by the phytoplankton that were being exported from a relic phosphate mine pit (Blue Cove) that is adjacent and hydraulically connected to the main channel in the lower river. Ecosystem Metabolism Table s 4 8 (Rainbow River) and 4 9 (Weeki Wachee River) provide mean quarter p er segment results for the following me t abolism characteristics: gross primary productivity (GPP), net ecosystem productivity (NEP), ecosystem respiration ( E R), and GPP to ER ratio (P : R). Rainbow River GPP, NEP, ER, and P:R ranged from 8.75 to 26.17 g O 2 m 2 d 1 0 to 22.54 g O 2 m 2 d 1 0.21 to 25.41 g O 2 m 2 d 1 and 1.0 to 115.7, respectively. Weeki Wachee River GPP, NEP, ER, and P:R ranged from 2.39 to 15.62 g O 2 m 2 d 1 0.51 to 12.11 g O 2 m 2 d 1 0 to 7.05 g O 2 m 2 d 1 and 1.1 to 62.8, respectively. Overall means of GPP and ER were 2 fold and 3.5 fold higher,

PAGE 192

192 respectively, in Rainbow than in Weeki Wachee River. No statistically significant differences were found between seasons (quarters) in either system for GPP, NEP, ER, or P:R (ANOVA, p >0.05 for a ll). Tables 4 10 (Rainbow River) and 4 11 (Weeki Wachee River) give mean quarterly values for daily raw incident PAR R PAR R efficiency (EE R ), benthic PAR B and benthic PAR R efficiency (EE B ) Overall mean values for Rainbow River PAR R EE R (g O 2 mol 1 and % ), PAR B and EE B (g O2 mol 1 and %) values were 42.5 6 mol m 2 d 1 0.41 g O 2 mol 1 7.87%, 20.86 mol m 2 d 1 0.92 g O 2 mol 1 17.51%, respectively (Table 4 10). Overall mean values for Weeki Wachee River PAR R EE R (g O 2 mol 1 and %), PAR B and EE B (g O2 mol 1 and %) values were 33.94 mol m 2 d 1 0.31 g O 2 mol 1 5.84%, 18.4 mol m 2 d 1 0.54 g O 2 mol 1 10.4 %, respectively (Table 4 11 ). Raw ecological efficiency values were between 3 and 27% (mean of 10%) lower than benthic ecological efficiency values in Rainbow River (Table 4 10). Differences between raw and benthic ecological efficiency were less in Weeki Wachee River, with EE B values between 1 and 10% (mean of 5%) greater than EE R values (Table 4 11). S ignificant temporal differences were found for benthic PAR where the 2 nd quarter ( April ) was significantly high er than the other quarters in Rainbow River (Figure 4 18 and Table 4 10 ). In Weeki Wachee River benthic PAR was highe r in the 2 nd quarter (April ) than the 1 st and 4 th quarter s (January and O ctober, Figure 4 18 and Table 4 11). Statistically significant differences between quarter means were also found for EE B where the 3 rd quarter mean was higher than quarter 2 in Rainbow River and mean values for quarters 1 and 4 we re significantly higher t han quarter 2 in Weeki Wachee River (Figure 4 19).

PAGE 193

193 The lower segments in Rainbow and upper segments in Weeki Wachee Rivers had more variable GPP values than the upper and lower segments, respectively. Significant spatial differences of GPP means were not found in Rainbow River, but were found between segments for Weeki Wachee ( Figure 4 20 ), with WWS1 (uppermost segment) being significantly higher than WWS3 and WWS4 (lowermost segments) Light and GPP relationships were evaluated in a variety of ways. D ail y GPP was found to be significantly, but weakly positively related to daily PAR above the surface of the water and in the benthic zone for the combined Weeki and Rai nbow River datasets (Figure 4 21 ). Benthic PAR was slightly better correlated with GPP than irradiance above the water surface. Relationships between hourly GPP and instantaneous PAR (s 1 ) above the water surface, and in the benthic zone in the upp ermost segment in Rainbo w River are shown in Figure 4 2 2 The daily course of GPP on an hourly bas is, specifically photosynthesis and respiration is depicted by changing dissolved oxygen (DO) concentrations i n relation to PAR The periods of increasing and decreasing DO (shown by the arrows in Figure 4 2 2 ) are likely due to effects from longitudinal d ispersion reaeration, and advection rates that caused a distinct time lag (ca. 1 2 h) between peak photosynthesis and peak PAR. The lags seemed to lengthen with distance from the headsprings, with the lowermost segment experiencing a 5 6 hour lag between GPP and PAR peaks (data not shown). The effect of dispersion on light utilization for photosynthesis was similar to the effect seen in Heffernan and Cohen (2010) work where they found lags between photosynthesis and nitrogen assimilation in the spring f ed Ichetucknee River in Florida

PAGE 194

194 Sample ecosystem level photosynthesis vs. irradiance curve s (P I curve) in the uppermost segment in Rainbow River are shown in Figur e 4 2 3 for PAR above the water (Io) and PAR remaining in the benthic zone (Iz) after ligh t was attenuated by the water column GPP values were lagged by two hours behind peak irradiance to match peaks within the day to develop t he P I curve s shown in Figur e 4 2 3 The P I curve typically has three distinct regions: a light limited region shown by the regression line and denoted 1 (pertains to Io) and 2 (pertains to Iz), a light saturated region, which is nominal in Figure 4 23 and a photoinhibited region that is denoted by 1 (pertains to Io) and 2 (pertains to Iz; Falkowski and Raven 20 07). The light saturated region of the curve is usually m ore pronounced extending for a period before the curve reaches (Falkowski and Raven 2007) However this deviation could be due to an effect of scale, i.e. ecosystem level vs. organism level. The P I curve (Figure 4 2 3 ) l ight compensation point for Io (C o 1) wa s 700 and Iz (C o 2) wa s 350 mol m 2 s 1 which equated to 3 4 % and 1 8 % PAR respectively, for a full sun day that p eaked at 2 05 0 mol m 2 s 1 The s aturation irradiance (E k mol m 2 s 1 ) wa s the value at the top end of the regression line. E k wa s 1150 mol m 2 s 1 for Io and 850 for Iz mol m 2 s 1 Pmax that occurred at E k for both Io and Iz wa s 0.71 g m 2 hr 1 1 (pertains to Io) wa s 2043 mol m 2 s 1 and 2 (pertains to Iz) wa s 1002 mol m 2 s 1 Half saturation wa s 1250 mol m 2 s 1 for Io and 600 mol m 2 s 1 for Iz, which equated to 63% and 30% PAR, when using 205 0 mol m 2 s 1 as max PAR. The 1 and 2 were 0.0007 and 0.0014 g O 2 mol 1 m 2 and represent the quantum yield of photosynthesis at a specific irradiance ( E g O 2 mol 1 m 2 ).

PAGE 195

195 Abiotic and Biotic Relations Results from n on correlations between Sagittaria kurziana ( SAG) dry weight biomass and percent subsu rface irradianc e in the blue, green and red bands for both systems are shown in Figure 4 2 4 The results suggest that potential blue light limitation exists between 38 and 45%PAR Blue A linear regression relationshi p also determined a statistically significant positive relationship between all data and mean annual data points for %PAR Blue and SAG biomass (Figure 4 2 5 ). The annually aggregated data provided a stronger relationship than the full dataset. The circled po int on the graph shows the estimated point of light limitation which is ca. 38%, the low value in the previously described blue light threshold range (Figure 4 25) Additionally, broadband %PUR L (%PAR weighted for %AL) and the blue band were significant ly related with SAG biomass (rho = 0.47, p = 0.011; rho = 0.56, p = 0.002, respectively). When epiphytes were considered to determine the effect of light available for SAG growth (i.e. %PUR LE vs. SAG biomass), the strength of the significance value of the correlation increased by an order of mag nitude for both broadband and the blue band ( p = 0.003, p = 0.0003 respectively ). Table 4 12 summarizes the minimum light requirement threshold ranges for broadband and blue band %PAR, and %PUR (weighted for SAG an d epiphytes). Figures 4 2 6 and 4 2 7 show significant relationships between water column chlorophyll a (chla) concentrations and NO 3 NH 4 + TP, flow discharge FA biomass, FA relative abundance, epiphyte load observed on macrophytes (categorical variable ranks of load: 1= clean, 2= light, 3= moderate, 4= heavy), and epiphyte type observed on macrophytes (categorical variable ranks of type: 1= diatoms, 2= diatoms and FA mixed,

PAGE 196

196 3= FA) Positive relationships were found for all except for NO 3 However, s inc e flow and NO 3 we re also highly correlated ( p < 0.05) residuals from the regression between flow and NO 3 were regressed against Chla (data not shown) to determine if a relationship remained after the effect of flow was removed from NO 3 A significant rel ationship was not found between the residuals and Chla, therefore t he apparent correlation between NO 3 and Chla is likely an artifact of the effect of flow on NO 3 driving the negative relationship with Chla. Table 4 13 provides statistically significant Pearson correlation results between abiotic and biotic parameters. AL% and SAG Chla content decreased and SAG shoot weight increased with distance from the headsprings PAR, PAR Red %PAR, %PAR Red velocity sh oot density, blade width, SAG Chla content an d epiphyte Chla content all decreased with increasing water depth and leaf lengths increased with greater water depths Positive relationships were found between velocity and SAG Chla content shoot density, and negative correlations between velocity and %AE and epiphyte:shoot weight ratio. A p ositive and negative correlation was seen between NO 3 and %AE and with SAG Chla content respectively. TP was p ositive ly correlated with SAG biomass, shoot weight, %AE, and epiphyte Chla content. A negative relat ionship w as found between TP and SAG Chla content, and %AL A negative correlation was found between PAR and %PAR (blue and red bands), and shoot weight. There was a p ositive correlation between PAR and %PAR (blue and red bands) and SAG Chla content, epi p hyte :shoot ratio and blade width. Negative relationship s were found between %AL and %AE and epiphyte Chla content SAG Chla

PAGE 197

197 content was also p ositive ly correlated with blade width, %AL, and negatively correlated with shoot weight, %AE and epiphyte Chla content SAG biomass was positively correlate d with leaves per shoot, and blade length. Shoot density was also negatively correlated with shoot weight. FA biomass was not significantly correlated with nitrate (Figure 4 2 8 ) or %PAR (data not shown). The r e lationship shown in Figure 4 2 8 between FA biomass and velocity for both systems combined datasets did seem to suggest that higher velocities (>35 cm s 1 ) constrained FA biomass, which supports recent work by King (2012). FA percent coverage and CDOM slope ( S g 0.33, p = 0.044). SAG and CHA percent coverage were also correlated with S g (r = 0.58, p = 0.008; r = 0.47, p = 0.04, respectively). No significant relationships were found for S g a nd any SAV species or algal biomass values. Table 4 14 provides statistically significant Pearson correlation results between GPP and abiotic and biotic parameters. GPP was negatively correlated with color and positively correlated with S g percent open canopy, community respiration, flow daily PAR B native biomass, native % coverage, SAG biomass, and SAG % coverage. Discussion This ecosystem scale study assess ed the effects of the sp ectral underwater light climate and other environmental factors on th e relative abundance biomass, optical characteristics, productivity and spatial distribution of primary producers along the downstream gradient. Correlatio ns were found that suggest spectral light availability influences morphological characteristics such as shoot weight, blade width and chlorophyll content production in Sagittaria kurziana (SAG), an dominant species in many spring systems. In lower blue and red light conditions, p lants direct ed more

PAGE 198

198 energy into growing larger shoots and wider blades to all ow for efficient li ght harvesting to drive their photosynthetic processes, which is consistent with the results of other studies (Abal et al. 1994) However, contrary to results from a study where Chla content in marine SAV was found to be higher in low l ight conditions (Abal et al. 1994), Chla content in SAG in this study was higher in high light conditions. An explanation of the difference in results could be attributed to the other study being carried out as a controlled laboratory growth experiment T h is study was conducted in the field, which could have be e n influenced by other environmental factors (e.g. sediment nutrients, gra zing pressures, flow velocity ) and their interactions with light. Pigment composition in SAV and algae is dependent on speci es composition, where diatoms and SAV have chlorophyll a, b, c, fucoxanthins, and ziadinoxanthins as their major pigments and cyanophytes have chlorophyll a, zeaxanthins, phycobiliproteins as their major pigments (Le et al. 2013) N umerous species of algae coexist in the same water mass and attached to SAV that can lead to variability in pigment content and composition (Falkowski and Raven 2007) Pigment composition can also lead to an indication of photoadaptation due to high light conditions where cells t ypically produce more photoprotective carotenoid pigments, whereas in low light conditions, cells produce more photosynthetic carotenoids ( Abal et al. 1994; Falkowski and Raven 2007; Le et al. 2013). Carotenoids were evaluated as part of this study, but we re not found to be significantly correlated with light availability. The negative relationship found between SAG absorptance and associated epiphy te absorptance and Chla content implies that epiphytes can impress upon the

PAGE 199

199 physiological response of their host plant by controlling the amount of light absorbed for photosynthesis and other functions When the range of epiphyte attenuation (20 to 75%) found in this study is factored in, the quantity and quality of l ight remaining after attenuation by the water column and by epiphytes may not be sufficient for SAV growth and survival (Sand Jensen and Borum 1991) Sleszynski (2009) studied environmental influences from ch anges in flow velocity and nutri ents on the morphology and physiology of Sagittaria kurziana (SAG) and epiphytes in the Rainbow River H is estimates of epiphytic algae attenuation in areas of good velocities ranged from 27% to 68% PAR broadband epiphyte attenuation estimates in both Weeki Wachee and Rainbow Riv er. greater epiphyte cover and thus greater light attenuation t he results from this study also found that l ower velocities may have provided better environmental conditions for epiph yte growth and light absorption (%AE). The absorption spectra determined for all of the SAV, FA and epiphyte communities found in these spring systems as part of this study provided a detailed look into the physiological workings of these photoautotrophs, which evidently utilize light differently based on the quality of incoming irradiance. When increases in depth and/or light attenuation are factored in, the relationship between light availability and plant survival becomes dependent on the ability of SAV to morphologically acclimate or adapt by using stored carbohydrate reserves to elongate above ground cells to reach more efficient levels of light in the water column, likely resulting in lower below ground biomass ( Middleboe and Markager 1997). The incre ase in above ground biomass and reduction of below ground biomass from increased

PAGE 200

200 depths could potentially lead to the plant uprooting in unstable sediments. The elongation mechanism was evident in this study with the positive correlation between mean depth and mean leaf lengths, but leaf lengths were not directly correlated with light attenuation. Diversity measures such as taxa richness (number of taxa) showed that Rainbow River overall had greater diversity than Weeki Wachee River by 3 fold. The upper hal f (ca. 0.2 km 2 (mean 88%), where the lower half (ca. 0.2 km 2 ) of the river was dominated by filamentous algae. SAG dominated the native SAV for the upper half of the river, where CER was the dominant native SAV in the lower river. SAG, and a similar species, VAL, are meadow forming species. Other species present in the river, such as POT, CER, and HYD are considered canopy forming. CER and HYD were more prominent in the lower portions of the systems studied here. The different growth strategies are responsible for different light requirements and the ability of the plant to harvest light (Middleboe and Markager 1997) due to self shading by the macrophyte itself in the upper canopy (canopy forming) and from epiphytes growing on the lower (older) leaves that are sloughed off with the leaves approximately every two months (Sand Jensen and Borum 1983). In contrast, epiphytes accumulate on the apical (top, older) portion of the leaves of meadow forming SAV that are closer to the greatly influencing the quality of light reaching the leaf. T he growth strategies of the canopy forming macrophytes may be more suitable for lower light conditions since epiphytes seem to have a lesser e ffect on light attenuation.

PAGE 201

201 Light was not the only factor evaluated to determine effects on SAV abundance distribution and productivity Other factors such as nutrients and flow were also considered, so that the study could account for interactions betwe en environmental factors and provide more understanding of their role in shaping the spring ecosystem communities. Nutrients such as nitrate and phosphorus were not found to be correlated with SAV or filamentous algae biomass, abundance or GPP Hoyer et al (2004) assessed factors controlling the abundance and distribution of SAV in three Florida spring fed coastal rivers and s imilar to this and other studies, and nutrients were not found to be a significant driver of SAV biomass (Canfield and Hoyer 19 88; D uarte and Canfield 1990). H owever light availability, salinity and substrate type, as governed by flow rates, were found to be the key factors constraining SAV abundance and di stribution (Hoyer et al. 2004). A comprehensive ecosystem level study measured metabolism characteristics in twelve springs in Florida including Rainbow and Weeki Wachee Rivers (WSI 2010), The WSI (2010) study reported metabolism values for two stations in both systems, one statio n in the spring pool and the other station in the spring run. D ata from the July sampling event from segment s RRS1 and RRS2 from this study w ere used to compare metabolism values against Rainbow Springs pool and run stations (collected in June 2009), respectively. Likewise, metabolism data from the April sampling event from segments WWS1 and WWS2 from this study w ere Wachee Springs pool and run stations (collected in March 2009), respectively.

PAGE 202

202 In Rain bow River, pool (RRS1, this study) and run (RRS2, this study) values for GPP, NEP, E R, P:R and EE R were relatively similar but slightly lower to values reported by WSI (2010) I n Weeki Wachee River pool (WWS1, this study) and run (WWS2, this study), values for metabolism characteristics did not agree as well as in Rainbow River. Values for GPP, NEP, ER, P:R and EE R in this study (WWS1) were all considerably higher (nearly 2 fold) than the values reported for the pool segment in the WSI report ( 2010) I n con trast, most of these values were markedly lower (also 2 fold) in this study (WWS2) as compared to WSI (2010) run segment. Duarte and Canfield (1990) evaluated the primary abiotic and biotic factors that affected SAV biomass and daily maximum primary pro duction (DMP) in 31 spring runs in Florida. Out of a wide array of explanatory variables that included in stream total nitrogen and total phosphorus concentrations, only shading by canopy cover was found to be significantly negatively correlated with DMP a nd with SAV biomass after censoring the data by removing sites that had dissolved oxygen concentrations less than 1 mg L 1 (Duarte and Canfield 1990; Canfield and Hoyer 1988). For that reason, nutrients were not considered to be limiting nor the primary dr iver controlling the abundance of SAV in a few dozen Florida streams (Duarte and Canfield 1990; Canfield and Hoyer 1988). They attributed the lack of nutrient limitation of SAV biomass to the continuous supply of nutrient rich water, reduction of boundary layer due to in stream turbulence and that SAV can acquire nutrients from the sediment (Duarte and Canfield 1990). Similar to results to this study, D uarte and Canfield (1990) also found that DMP was positively correlated with SAV biomass and their associa ted epiphytic biomass. However, they did not find any statistical correlations between stream discharge,

PAGE 203

203 velocity or depth with SAV biomass or DMP results where discharge was positively correlated with GPP Duarte an d Canfield (1990) stated that chlorophyll a in streams was generally due to pseudoplankton and not phytoplankton, which occur due to several reasons: 1) absence of algae in the aquifer, 2) high hydraulic flushing rates and 3) spring runs emerge from a sing le localized point instead of the runs arising from an array of low order streams which enhance the development of stream algae. This conclusion was corroborated results of chlorophyll a concentrations in the water column where light absorp tion spectra were considerably higher in areas of the river that had higher residenc e times and lower velocities. However, chlorophyll a concentrations in the water column were positively correlated with ammonia and phosphorus. In studies on other spring s ystems, nutrient concentrations were not found to be related to benthic filamentous algal abundance, although the combined effect of dissolved oxygen concentration and algal grazer abundance were found to be good predictors of algal abundance (Heffern an et al. 2010a ). Also, investigations into the effect of hydrology on filamentous algal abundance supported the result of another study that suggested flow velocities above 35 cm /s were able to constrain certain benthic filamentous algae (King 2012). The pho tosynthetic rate for an organism or ecosystem is controlled by the efficiency of light utilization (Falkowski and Raven 2007). The instantaneous photosynthetic rate (P E ) can be estimated by multiplying E by PUR (obtained by multiplying the spectrally inte grated percent absorptance by PAR), which is the light absorbed in the PAR range by the primary producer in question (Falkowski and Raven

PAGE 204

204 2007), assuming that the primary producer was dominant in the reach and was the primary contributor to GPP. The spectr ally integrated absorbed irradiance (PUR) includes absorption by pigments such as chlorophylls, carotenoids, phycobilipigments (cyanophytes), and accessory chlorophylls (Falkowski and Raven 2007). The photosynthetic rate can be calculated for total irradia nce or can be adjusted for specific wavebands or wavelengths in the PAR spectrum to determine the photosynthetic rate at the desired wav eband or wavelength. When %PUR wa s used to adjust the benthic light ) became steeper (data not shown). The ecosystem level P I curves for the uppermost Rainbow River segment showed mid day suppression of GPP for incident PAR above the water (Io) and PAR remaining in the benthic zone (Iz) after light was attenuated by t he water column. Mid day suppression of photosynthesis may be due to photoinhibition in this high light environment (Falkowski and Raven 2007) In contrast, a study in a New Zealand river did not find ecosystem level photoinhibition or light saturation ( Yo ung and Huryn 1996). However, t he upper river segments of this study had phototrophic communities that we re a dapted to higher light conditions, which mean t that mor e light wa s required to reach maximum photosynthesis levels Correlations between GPP and pe rcent open canopy and daily benthic PAR provided support for the relationships developed from the P I curves to predict the rate of photosynthesis, provided that the dominant known to calculate PUR. Other important correlations that were found between GPP

PAGE 205

205 and native SAV and SAG abundance and biomass reinforce the notion that the quality of light is critical to support native SAV for efficient GPP. The change in composition and decline in SAG were found to be signifi cantly correlated to the decline of blue light in the lower portions of these systems Based on the distribution of SAG in relation to percent subsurface irradiance and available benthic PAR in the blue band, it appears that the lower Rainbow River is blue light limited. This inference pr ovides sufficient evidence to reject the null hypothesis that blue light does not control the abundance or distribution of SAG within spring systems Although SAG seems to be controlled by the quality of light available to the benthic zone, SAG was not found to be limited by the green or red bands. Moreover, the reduction of light by epiphytic algae can further reduce benthic light a vailability to very low levels and needs to be taken into account when considering light lim itation of SAV in general This conclusion explicitly incorporates the light attenuation due to epiphytes because varying levels of epiphyte coverage were observed at each site where SAG biomass was measured. Therefore, it seems that a dichotomy may exist within these systems w here photoinhibition may be occurring in the upper river and light limitation may be taking place in the lower river. Several researchers have proposed minimum light requirements (ranging from 4 to 29% PAR ) for certain freshwater SAV species based on controlled laboratory light response experiments (Sand Jensen and Madsen 1991; Dennison et al. 1993 ). Another more recent light response laboratory study determined that SAG became light saturated at levels between 6 to 15% ( Sleszynski 20 09 ). Sleszynski results were considerably lower than estimates from this study ( ca. 40%). In contract to this field

PAGE 206

206 laboratory organism scale study results suggest that SAG was tolerant to low light levels A n important distinction to make is that epiphytes were removed from the leaves prior to conducting Sleszynski (2009) light response study, which may have alleviated the potential for light limitation since the effects of epiphyte attenuation on light absor ption by the SAG leaves were not incorporated in the s tudy Another reason for the difference in results could be from the level of scale of the studies This study was conducted a t the ecosystem community level, which incorporated various species of SAV, benthic microalgae and macroalgae, water column plankton, and consumers, which all could affect the overall community level GPP. S leszynski (2009) study was conducted at the organism level which may have decreased the light saturation level due to the d ifference in scale This brings to light an important management implication of the need to establish a blue light requirement for springs The results from this study suggest that a minimum b lue light requirement between 38 to 45% in the benthic zone wo uld provide sufficient light, while considering epiphyte attenuation, to support establishment and growth of the SAG, the sentinel springs species When spectral subsurface irradiance in the benthic zone cannot be determined and only broadband %PAR can be measured, a minimum broadband PAR requirement of 36 to 40% can be used as a surrogate that would encompass the minim blue light requirement. Accordingly, the relationship between SAG biomass and the amount of PUR and PUR in the blue band that is weighted for the absorption spectra of SAG ( %PUR L ) wa s stronger (considerably lower p values) and explained 11% more of the variability than

PAGE 207

207 %PAR. Likewise, the amount of useable light can also be used as a light requirement (%PUR L of 10 to 25% and %PUR L Blue of 15 to 25%). However, measurements like these are more demanding and laborious than simply measuring irradiance in the water column to calculate %PAR. The relationship between light and photosynthesis has been studied by many in different types of aquatic ecosystems, with the majority of the research focusing on broadband PAR and spectrally explicit PAR in estuaries, oceans, and lakes. The few river ecological studies that have considered the quantity of light did not progress beyond the broadband spectrum to assess the quality of light. Knowledge of the quantity of light available (PAR) to primary producers can be greatly enhanced by also SAV community to carry out photosynthesis efficiently. Alterations to the underwater light field from changes in optical water quality (OWQ) can produce changes in the spectral distribution of light, thereby leading to a change in the composition, distribution and photosynthetic ef ficiency of the primary producer community. The lack of spectrally defined ecological studies in flowing waters can be attributed to the complexity of conducting spectral light measurements in rivers due to the high spatial variation and difficulty of meas uring light in turbulence. However, high resolution spectrally specific empirical optical modeling methods are now available (Chapter 3) for spring fed rivers that can be applied to lotic systems that exhibit similar optical characteristics as the rivers u sed for model development. The spectrally specific optical modeling methods, coupled with the GPP PAR B relationship developed as part of this study advances the currently available approaches to predict GPP and to determine

PAGE 208

208 if spectral light availability i s sufficient for SAV growth and survival. Future work abounds that could include investigations into the effects of light intensities and phototroph type and biomass on components of the P I Curve and overall GPP rates. The original perception that sprin g systems are not light limited needs to be revisited and more thoroughly examined to support or reject the conclusions suggested in this study. The %PAR requirements recommended here for SAG are somewhat higher than light requirements suggested by others for other SAV such as seagrasses or other freshwater plants mentioned earlier However, lotic systems function very differently in terms of potential hydrologic impacts to sediments, nutrient transport and delivery, and other environmental factors that can increase the potential for light limitation and the minimum amount of light of more sensitive sentinel species such as SAG. To ensure that a suitable benthic habitat is available for healthy SAG growth and distribution, other factors such as sediment qual ity, grazing pressure, flow velocity, residence time, and epiphytic cover that all interact with light requirements also need to be evaluated when any resource management restoration efforts are considered. The expected linkage between the quality of light reaching SAV, in terms of more efficient wavelengths such as the blue band of PAR, should invoke a sense of urgency in resource managers to determine ways to limit the constituents in the water column that are causing attenuation of the limiting wavelengt hs. A considerable change in light availability with respect to wavelengths in the water column can have detrimental effects on spring ecosystem health and productivity. It is a fact that there is competition for light between nuisance (undesirable) vegeta tion such as benthic algae, epiphytes and with

PAGE 209

209 desirable SAV (Sand Jensen and Borum 1991). This competition for high quality light is a critical factor in the balance among these phototrophic communities. In spite of this, the balance may be tipping towar ds an optical water quality habitat that favors phototrophs that can use low quality light wavebands that desired SAV cannot use for efficient photosynthesis. If and when aquatic ecosystems experience a change in the spectral distribution of the light clim ate, then a shift in vegetation community distribution will likely occur. In most springs, there is still a substantial quantity of light available, however the quality of light is changing and that alteration needs to be addressed to provide good optical water quality for a suitable habitat for SAV communities to flourish.

PAGE 210

210 Figure 4 1. Linear relationships between mean percent broadband subsurface irradiance (%PAR) and %PAR in the blue, green, and red bands (%PAR Blue %PAR Green, and %PAR R ed, respectively) and distance from the headsprings in A) Rainbow and B) Weeki Wachee Rivers.

PAGE 211

211 Figure 4 2 Weeki Wachee River mean daily flow (m 3 s 1 ) for synoptic water quality sampling dates shown per quarter. Note: Data were obtained from USGS webs ite. ANOVA results shown. The Tukey method was used to determine which groups of means were significantly different. Means that do not share a letter were significantly different if p<0.05.

PAGE 212

212 Figure 4 3 Relationship s between measured biomass DW (dry we ight, g m 2 ) of filamentous algae (FA) and Sagittaria kurziana (SAG) and relative abundance (percent coverage) using data from both spring systems.

PAGE 213

213 Figure 4 4 Spatial distribution of mean annual relative abundance (percent coverage) of filamentous algae (FA) and macrophytes in Rainbow River Note: M acrophyte species: Sagittaria kurziana (SAG), Vallisneria americana (VAL), Potamogeton illinioensis (POT), Najas guadalupensis (NAJ), Utricularia spp. (UTR), Ludwigia repens (LUD), Ceratophyllum demersu m (CER), Myriophyllum heterophyllum (MYR), Chara spp. (CHA), Hydrilla verticillata (HYD). Rainbow River stations are shown in brackets on the x axis. The table above the figure shows %PAR (A), %PAR Blue (B), %PAR Green (C), and %PAR Red (D) annual minima values above each station.

PAGE 214

214 Figure 4 5 Spatial distribution of mean annual relative abundance (percent coverage) of filamentous algae (FA), native and non native macrophyte species at each station in Rainbow River. Note: Native species include all sp ecies listed in Figure 4 4, except for HYD, which is the only non native species. Rainbow River stations are shown in brackets on the x axis.

PAGE 215

215 Figure 4 6 Spatial distribution of mean annual relative abundance (percent coverage) of filamentous algae (FA) and macrophyte s in Weeki Wachee River Note: Macrophyte species: Sagittaria kurziana (SAG), Vallisneria americana (VAL), Najas guadalupensis (NAJ), Chara spp (CHA), Hydrilla verticillata (HYD). Weeki Wachee River stations are shown in brackets on t he x axis. The table above the figure shows %PAR (A), %PAR Blue (B), %PAR Green (C), and %PAR Red (D) annual minima values above each station.

PAGE 216

216 Figure 4 7 Spatial distribution of mean annual relative abundance (percent coverage) of filamentous algae ( FA), native and non native macrophyte species at each station in Weeki Wachee River. Note: Native species include all species listed in Figure 4 6 except for HYD, which is the only non native species. Weeki Wachee River stations are shown in brackets on the x axis.

PAGE 217

217 Figure 4 8 Spatial distribution of mean annual biomass DW (dry weight) of filamentous algae (FA) and macrophytes in Rainbow River Note: M acrophyte species: Sagittaria kurziana (SAG), Vallisneria americana (VAL), Potamogeton illinioens is (POT), Najas guadalupensis (NAJ), Ceratophyllum demersum (CER), and Hydrilla verticillata (HYD). Rainbow River stations are shown in brackets on the x axis.

PAGE 218

218 Figure 4 9 Spatial distribution of mean annual biomass DW (dry weight) of filamentous alg ae (FA) and macrophytes in Weeki Wachee River Note: M acrophyte species: Sagittaria kurziana (SAG), Najas guadalupensis (NAJ), Chara spp (CHA), and Hydrilla verticillata (HYD). Weeki Wachee River stations are shown in brackets on the x axis.

PAGE 219

219 Figure 4 10. Relationship between measured dry weight per shoot (wt shoot 1 g ) of epiphytes and Sagittaria kurziana shoot dry weight using data from both spring systems.

PAGE 220

220 Figure 4 11 Comparisons of relative percent absorbance scans or optical density (OD) normalized to the maximum OD of filamentous algae (F A) and nine macrophyte species found in various locations in Rainbow River. Note: M acrophyte species: Sagittaria kurziana (SAG), Vallisneria americana (VAL), Potamogeton illinioensis (POT) Najas guada lupensis (NAJ), Utricularia spp (UTR), Ludwigia repens (LUD), Ceratophyllum demersum (CER), Chara spp (CHA), Hydrilla verticillata (HYD) Wavelngth (nm)

PAGE 221

221 Figure 4 12 S patial variation of percent relative absorptance (%AL) of Sagittaria kurziana (SAG) in Rainbow Riv er. The mean %AL is shown for all five stations.

PAGE 222

222 Figure 4 13 S patial variation of percent relative absorptance (%AE) of epiphytic algae scraped from Sagittaria kurziana (SAG) samples in Rainbow River. The mean %AE is shown for all five stations.

PAGE 223

223 Figure 4 14 Pigment content in macrophyte or filamentous algae found in Rainbow (RR1 RR8) and Weeki Wachee (WW0 WW5) River transects. Note: The x axis labels indicate the transect location and the macrophyte (SAV) or filamentous algae (FA) measured: Sa gittaria kurziana (SAG), Vallisneria americana (VAL), Potamogeton illinioensis (POT) Najas guadalupensis (NAJ), Utricularia spp (UTR), Ludwigia repens (LUD), Ceratophyllum demersum (CER), Myriophyllum heterophyllum (MYR), Chara spp (CHA), Hydrilla vert icillata (HYD), Lyngbya wollei (LYN), and Vaucheria spp (VAU). The red bars are above the median (indicated by black lines) and the grey bars are below the median. Chlorophyll a (Chla, g g 1 ) content is shown in A ), and total carotenoid (x+c, g g 1 ) co ntent is shown in B). A

PAGE 224

224 Figure 4 15 Relationships between A) percent relative absorptance of Sagittaria kurziana (SAG, %AL) and chlorophyll a (Chla) FW (fresh weight) content of SAG, and B) epiphytic algal absorptance (%AE) and chlorophyll a content of epiphytes scraped from SAG samples in Rainbow River. A B

PAGE 225

225 Figure 4 16. Logarithmic relationship between epiphyte absorptance (% AE) per reference wavelength as a function of epiphyte biomass per shoot (g) for both systems combined .

PAGE 226

226 Figure 4 17 Relative a bsorptance of plankton entrained in the water column at various locations in Rainbow River across the PAR spectrum. Note: Left axis is for RR1 (uppermost station near the headsprings) and right axis is for RR4 (mid river, upstream of Blue Cove) Blue Cove (relic phosphate mine pit adjacent to main lower river channel) and RR8 (lowermost station downstream of Blue Cove) samples.

PAGE 227

227 Figure 4 18 Boxplots showing temporal variability for benthic PAR in relation to quarter for A) Ra inbow, and B) Weeki Wachee Rivers. Note: For statistically significant one way ANOVA, the Tukey method was used to determine which groups of means were significantly different. Means that do not share a letter were significantly different if p <0.05. p = 0. 0 0 0 A B B B A p = 0. 0 0 3 B A B B AB

PAGE 228

228 Figure 4 19. Boxplots showing temporal variability for benthic ecological efficiency (EE B ) in relation to quarter for A) Rainbow, and B) Weeki Wachee Rivers. Note: For statistically significant one way ANOVA, the Tukey method was used to det ermine which groups of means were significantly different. Means that do not share a letter were significantly different if p <0.05. B AB p = 0. 017 p = 0. 0 0 4 A AB B A B A AB A

PAGE 229

229 Figure 4 20 Boxplots showing spatial and temporal variability for gross primary production (GPP) in relation t o distance of the midpoint of each segment from the headsprings for A) Rainbow, and B) Weeki Wachee Rivers. Note: For statistically significant one way ANOVA, the Tukey method was used to determine which groups of means were significantly different. Means that do not share a letter were significantly different if p <0.05. A AB B B A B p = 0. 0 05 p > 0.05

PAGE 230

230 Figure 4 21 Relationships between average daily GPP in segments (n = 39) and PAR above the water surface and in the benthic zone. Note: Gross primary productivit y, GPP; A) average daily photosynthetically active radiatio n (PAR) above the water surface, and B) average daily PAR in the benthic zone (river bottom). PAR and GPP were both natural log transformed. A B

PAGE 231

231 Figure 4 22 Relationships between hourly GPP and PAR in the uppermost segment in Rainbow River. Note: Gross primary productivity, GPP; photosynt hetically active radiation, PAR; A) above the water surf ace, and B) in the benthic zone Arrows indicate direction of change during periods of increasing an d decreasing dissolved oxygen during photosynthesis and respiration. A B

PAGE 232

232 Figure 4 23 Sample photosynthesis vs. irradiance curve (P I curve) in the uppermost segment in Rainbow River Note: The P I curve shows the mid day suppression of gross primary pr oductivity (GPP) for incident photosynthetically active radiation (PAR) above the water (Io) and PAR remaining in the benthic zone (Iz) after light was attenuated by the water column. The P I curve usually has three distinct regions: a light limited region shown by the regression line and denoted 1 (pertains to Io) and 2 (pertains to Iz), a light saturate d region, and a photoinhibited region that is denoted by 1 (pertains to Io) and 2 (pertains to Iz ). C o 1 and C o 2 are the l ight compensation point s for Io and Iz, respectively. 1

PAGE 233

233 Figure 4 24 Relationships between Sagittaria kurziana (SAG) biomass DW (dry weight) and percent subsurface irradiance for different bands Note: A) the whole PAR spectrum (%PAR) ; different portions of the PAR spectrum reaching the benthic zone: B) blue (400 500 nm, %PARBlue), C) green (500 600 nm, %PARGreen) and D) red (600 700 nm, %PARRed). The circles on the data points in B) indicate the low and high ranges of potential blue light limitation, 38 and 45%, respectively. Non ho) and significance p values A

PAGE 234

234 Figure 4 2 5 Relationships between Sagittaria kurziana (SAG) biomass DW (dry weight) and percent subsurface irradiance in the blue band (400 500 n m, %PAR Blue ) Note: A) all data points combined for both systems and B ) mean annual data for each variable. Data was natural log transformed. The circled point shows the estimated point of light limitation. A B

PAGE 235

235 Figure 4 26 Non linear relationships b etween water column chlorophyll a (chla) concentrations and A) nitrate (NO 3 ), B) ammonium (NH 4 + ), C) total phosphorus (TP), and D) flow discharge in Rainbow River. Note: None of these relationships were significant ( p >0.05) in Weeki Wachee River, theref ore data were not shown. A B C D

PAGE 236

236 Figure 4 27 R elationships between water column concentrations and A) FA biomass, B) FA relative abundance (%coverage), C) epiphyte load observed on macrophytes D) epiphyte type observed on macrophytes in Rainbow River Not e: C hlorophyll a, chla; filamentous algae, FA; epiphyte load ca tegorical variable ranks : 1= clean, 2= light, 3= moderate, 4= heavy; epiphyte type categorical variabl e ranks : 1= diatoms, 2= diatoms and FA mixed, 3= FA None of these relationships were signi ficant ( p >0.05) in Weeki Wachee River, therefore data were not shown. D C A B

PAGE 237

237 Figure 4 28 Relationships between filamentous algae biomass dry weight (FA BM) and A) velocity and B) nitrate (NO 3 ) for the combined dataset. A B

PAGE 238

238 Table 4 1. Res ults from one w ay ANOVA for spatial variability in Rainbow River stations of mean (standard deviation given in parenthes es) water chemistry parameters. RR1 RR2 RR3 RR4 RR5 RR7 RR8 ANOVA ( p value) Chla 0.10 0.18 0.26 0.40 0.39 0.93 0.86 mg m 3 (0.03) (0.03) (0 .04) (0.13) (0.10) (0.57) (0.41) D CD BC B B A A 0.000 NO 3 2.07 1.79 1.68 1.55 1.52 1.47 1.46 mg L 1 (0.145) (0.03) (0.03) (0.03) (0.03) (0.05) (0.04) A B BC CD D D D 0.000 TKN 0.080 0.023 0.034 0.063 0.026 0.071 0.051 mg L 1 (0.05) (0.02) (0. 03) (0.02) (0.02) (0.04) (0.03) A A A A A A A 0.116 NH4+ 0.009 0.009 0.009 0.009 0.009 0.011 0.011 mg L 1 (0.002) (0.002) (0.002) (0.002) (0.002) (0.004) (0.003) A A A A A A A 0.649 TP 0.029 0.031 0.032 0.031 0.035 0.036 0.036 mg L 1 (0.0006) ( 0.001) (0.0008) (0.002) (0.003) (0.005) (0.003) B AB AB AB A A A 0.002 Note: Water chemistry parameters: Chlorophyll a, Chla; nitrate, NO 3 ; ammonium, NH 4 + ; total Kjeldahl nitrogen, TKN; and total phosphorus, TP. The Tukey method was used to determi ne which groups of means were significantly different. Means that do not share a letter were significantly different if p <0.05.

PAGE 239

239 Table 4 2. Res ults from one way ANOVA for spatial variability in Rainbow River stations of mean (standard deviation given in parenthese s) apparent optical properties. RR1 RR2 RR3 RR4 RR5 RR7 RR8 ANOVA ( p value) PAR B 1021.4 1001.6 1038.6 1074.6 750.5 634.0 734.1 mol m 2 s 1 (301.1) (121.0) (216.0) (210.0) (172.3) (201.1) (367.0) 0.077 PAR B (Blue) 390.1 386.0 384. 5 370.2 279.8 228.8 281.3 mol m 2 s 1 (49.1) (32.4) (48.1) (78.0) (42.9) (87.2) (153.6) A A A A AB B AB 0.039 PAR B (Green) 406.9 395.9 429.3 426.5 315.8 260.7 295.6 mol m 2 s 1 (115.8) (40.2) (102.1) (85.3) (82.8) (72.1) (142.9) 0.103 PAR B (Red) 224.4 219.7 224.8 277.9 155.0 144.5 157.2 mol m 2 s 1 (150.8) (60.0) (68.0) (59.0) (55.0) (55.0) (92.0) 0.263 %PAR 57.1 59.9 55.0 61.0 46.5 37.6 29.7 (10.3) (10.0) (12.5) (9.4) (7.2) (14.1) (20.6) AB A AB A AB AB B 0.013 %PAR (Blue) 69.7 72.0 65.8 66.8 54.0 42.9 43.3 (7.7) (10.2) (17.4) (11.9) (10.4) (17.1) (17.8) A A A A AB B AB 0.039 %PAR (Green) 64.4 67.7 64.0 68.2 55.6 43.5 43.9 (11.7) (9.5) (13.6) (9.5) (7.6) (14.8) (20.2) 0.103 %PAR (Red) 37.4 40.3 35.5 48.2 29.9 26.5 26.0 (18.9) (11.1) (7.4) (8.3) (5.2) (11.7) (14.6) 0.141 Note: Benthic PAR, PAR B in blue, green and red bands; %PAR in blue, green and red bands The Tukey method was used to determine which groups of means were significantly different. Means that do not sha re a letter were significantly different if p <0.05.

PAGE 240

240 Table 4 3. Res ults from one way ANOVA for spatial variability in Weeki Wachee River stations of mean (standard deviation given in parentheses) water chemistry parameters WW0 WW0.5 WW1 WW3 WW5 ANOVA ( p value) Chla 0.22 0.97 0.73 0.61 1.31 mg m 3 (0.18) (0.50) (0.21) (0.33) (0.17) A AB ABC BC C 0.002 NO 3 0.93 0.89 0.87 0.85 0.81 mg L 1 (0.05) (0.04) (0.04) (0.04) (0.03) A A AB AB B 0.006 TKN 0.020 0.025 0.025 0.034 0.056 mg L 1 (0.02 ) (0.03) (0.01) (0.02) (0.03) A A A A A 0.196 NH4+ 0.009 0.009 0.009 0.009 0.010 mg L 1 (0.002) (0.002) (0.002) (0.002) (0.008) A A A A A 0.597 TP 0.016 0.010 0.010 0.009 0.010 mg L 1 (0.007) (0.001) (0.001) (0.000) (0.001) A AB AB B AB 0.0 48 Note: Water chemistry parameters: Chlorophyll a, Chla; nitrate, NO 3 ; ammonium, NH 4 + ; total Kjeldahl nitrogen, TKN; and total phosphorus, TP. The Tukey method was used to determine which groups of means were significantly different. Means that do not share a letter were significantly different if p <0.05.

PAGE 241

241 Table 4 4. Res ults from one way ANOVA for spatial variability in Weeki Wachee River stations of mean (standard deviation given in parentheses) apparent optical properties See Table 4 2 notes for t able notations. WW0 WW0.5 WW1 WW3 WW5 ANOVA ( p value) PAR B 1503.6 1453.7 1343.3 1016.0 188.0 mol m 2 s 1 (117.3) (101.1) (266.2) (265.0) (69.8) A A AB B C 0.000 PAR B (Blue) 530.2 491.1 466.2 358.6 58.1 mol m 2 s 1 (93.0) (66.3) (81.4) (104.8) (27.6) A A A A B 0.000 PAR B (Green) 568.6 572.8 520.6 402.5 71.6 mol m 2 s 1 (67.4) (33.5) (120.7) (99.4) (22.9) A A A A B 0.000 PAR B (Red) 404.8 389.8 356.4 255.0 58.4 mol m 2 s 1 (62.5) (27.9) (97.7) (76.2) (22.0) A AB AB B C 0. 000 %PAR 74.6 71.8 70.5 58.2 57.5 (3.6) (8.1) (5.5) (5.2) (18.3) 0.067 %PAR (Blue) 81.2 80.7 77.9 61.0 51.8 (4.9) (11.7) (3.0) (9.4) (24.2) A A AB AB B 0.017 %PAR (Green) 79.6 77.9 76.2 66.5 63.6 (3.4) (7.8) (6.2) (6.0) (19.3) 0.145 %PAR ( Red) 63.3 57.2 57.4 47.3 57.3 (4.6) (5.9) (10.1) (1.8) (13.9) 0.171

PAGE 242

242 Table 4 5. Re sults from one way ANOVA for spatial variability in Rainbow River transects of mean biomass DW ( dry weight, g m 2 ) of filamentous algae (FA) and macrophytes. RR1 RR2 RR3 RR4 RR5 RR7 RR8 ANOVA ( p value) SAG 83.6 113.0 161.1 123.0 10.6 0.0 0.0 (77.4) (49.8) (64.4) (102.9) (12.5) (0.0) (0.0) AB A A A B B B 0.002 VAL 0.0 11.3 0.0 1.0 53.6 0.0 0.0 (0.0) (22.5) (0.0) (2.1) (75.8) (0.0) (0.0) A A A A A A A 0.154 POT 129.6 0.0 0.0 0.0 0.0 0.0 0.0 (259.1) (0.0) (0.0) (0.0) (0.0) (0.0) (0.0) * * * * NAJ 0.0 0.0 0.0 14.6 0.0 0.0 0.0 (0.0) (0.0) (0.0) (29.22) (0.0) (0.0) (0.0) * * * * CER 0.0 0.0 0.0 0.0 204.4 0.0 0.0 (0.0) (0.0 ) (0.0) (0.0) (369.3) (0.0) (0.0) * * * * HYD 0.0 0.0 2.9 0.0 0.0 5.4 0.2 (0.0) (0.0) (5.6) (0.0) (0.0) (6.6) (0.3) B B AB B B A B 0.178 FA 11.5 7.5 0.2 65.2 53.6 138.7 435.8 (14.2) (10.3) (0.3) (60.6) (73) (160.6) (205.9) A A A A A A B 0.000 Note: Sagittaria kurziana (SAG), Vallisneria americana (VAL), Potamogeton illinioensis (POT), Najas guadalupensis (NAJ), Ceratophyllum demersum (CER), Hydrilla verticillata (HYD). Standard deviation is given in parentheses. The Tukey metho d was used to determine which groups of means were significantly different. Means that do not share a letter were significantly different if p<0.05. *denotes this species was only found in one transect.

PAGE 243

243 Table 4 6. Results from one way ANOVA for spatial va riability in Weeki Wachee River transects of mean biomass DW (dry weight, g m 2) of filamentous algae (FA) and and macrophytes. WW0 WW0.5 WW1 WW3 WW5 ANOVA ( p value) SAG 29.8 0.0 0.0 0.0 0.0 (18.8) (0.0) (0.0) (0.0) (0.0) A B B B B 0.000 NA J 0.0 0.0 0.1 11.0 2.8 (0.0) (0.0) (0.02) (8.4) (2.5) B B B A AB 0.005 CHA 8.62 14.76 1.90 0.00 0.04 (16.5) (24.46) (2.67) (0.00) (0.07) A A A A A 0.460 HYD 0.0 0.0 0.0 18.0 0.0 (0.0) (0.0) (0.0) (0.0) B B B A B 0.010 FA 33.0 56.5 25. 0 13.9 0.4 (60.5) (85.6) (42.1) (24.5) (0.8) A A A A A 0.625 Note: Sagittaria kurziana (SAG), Najas guadalupensis (NAJ), Chara spp. (CHA), Hydrilla verticillata (HYD). Standard deviation is given in parentheses. The Tukey method was used to deter mine which groups of means were significantly different. Means that do not share a letter were significantly different if p <0.05.

PAGE 244

244 Table 4 7. Summary statistics for Sagittaria kurziana (SAG) shoot system characteristics in Rainbow and Weeki Wachee R iver transects. Rainbow River Dry per shoot wt (g) Density (shoots m 2 ) Leaves shoot 1 Blade Length (cm) Blade Width (cm) Mean 1.45 106.70 14.22 41.83 0.57 Standard Error 0.36 22.01 0.78 4.74 0.04 Standard Deviation 1.54 98.45 3.30 20.09 0. 17 Minimum 0.16 0.00 9.00 14.05 0.38 Maximum 6.47 360.00 19.83 84.54 0.89 Number of Samples 18 20 18 18 18 Weeki Wachee River Mean 0.24 238.00 13.00 25.51 1.01 Standard Error 0.13 56.91 1.15 4.44 0.03 Standard Deviation 0.26 113. 82 2.31 8.89 0.07 Minimum 0.10 76.00 11.00 13.88 0.95 Maximum 0.63 336.00 15.00 34.50 1.10 Number of Samples 4 4 4 4 4

PAGE 245

245 Table 4 8. Mean quarterly per segment results in Rainbow River for metabolism characteristics: gross primary producti vity (GPP), net ecosystem productivity (NEP), ecosystem respiration (ER), and GPP to ER ratio (P:R) Segment Quarter GPP (gO 2 m 2 d 1 ) NEP (gO 2 m 2 d 1 ) ER (gO 2 m 2 d 1 ) P:R RRS1 1 25.38 0.00 25.38 1.00 RRS1 2 17.16 0.41 16.75 1.02 RRS1 3 16.07 0.00 16.07 1.00 RRS1 4 19.76 0.00 19.76 1.00 RRS2 1 18.17 0.09 18.09 1.01 RRS2 2 19.54 0.26 19.28 1.01 RRS2 3 15.38 1.60 13.79 1.12 RRS2 4 21.41 2.58 18.83 1.14 RRS3 1 15.04 0.87 14.17 1.06 RRS3 2 16.20 0.00 16.20 1.00 RRS3 3 13.14 0.00 13.14 1.00 RRS3 4 16 .28 0.00 16.28 1.00 RRS4 1 18.82 9.69 9.12 2.07 RRS4 2 15.62 10.49 5.13 3.07 RRS4 3 11.80 0.25 11.55 1.02 RRS4 4 10.45 5.54 4.91 2.13 RRS5 1 14.78 10.98 3.81 3.95 RRS5 2 22.75 22.54 0.21 115.71 RRS5 3 18.22 17.09 1.14 16.02 RRS5 4 8.75 4.92 3.82 2. 29 RRS6 1 11.65 8.61 3.05 3.95 RRS6 2 16.70 3.53 13.17 1.28 RRS6 3 26.17 0.76 25.41 1.04

PAGE 246

246 Table 4 9. Mean quarterly per segment results in Weeki Wachee River for metabolism characteristics: gross primary productivity (GPP), net ecosystem productivit y (NEP), ecosystem respiration (ER), and GPP to ER ratio (P:R) Segment Quarter GPP (gO 2 m 2 d 1 ) NEP (gO 2 m 2 d 1 ) ER (gO 2 m 2 d 1 ) P:R WWS1 1 6.83 3.61 3.22 2.46 WWS1 2 11.88 5.86 6.02 1.99 WWS1 3 15.62 9.44 6.18 2.70 WWS1 4 12.11 12.11 0.00 WWS2 1 5.58 1.11 4.48 1.40 WWS2 2 8.37 2.17 6.20 1.35 WWS2 3 6.51 5.38 1.13 WWS2 4 13.61 6.57 7.05 3.74 WWS3 1 4.33 0.65 3.68 1.19 WWS3 2 3.99 0.79 3.20 1.27 WWS3 3 5.91 1.82 4.09 1.46 WWS3 4 5.28 2.73 2.55 2.14 WWS4 1 2.39 0.72 1.68 1.56 WWS4 2 2.92 2 .87 0.05 62.77 WWS4 3 4.82 0.71 4.11 1.18 WWS4 4 4.88 0.51 4.36 1.11

PAGE 247

247 Table 4 10 Mean quarterly per segment values in Rainbow River for daily raw incident PAR R PAR R efficiency (EE R ), benthic PAR B and benthic PAR R efficiency (EE B ). Segmen t Quarter PAR R (mol m 2 d 1 ) EE R (%) EE R (gO 2 mol 1 ) PAR B (mol m 2 d 1 ) EE B (%) EE B (gO 2 mol 1 ) RRS1 1 41.04 11.91 0.62 25.38 19.25 1.01 RRS1 2 55.08 5.96 0.31 27.01 12.16 0.64 RRS1 3 30.73 10.04 0.52 17.68 17.45 0.91 RRS1 4 42.96 8.81 0.46 24.67 15.33 0.80 RRS2 1 42.82 8.17 0.43 28.38 12.33 0.64 RRS2 2 55.08 6.78 0.35 32.62 11.46 0.60 RRS2 3 30.73 9.65 0.50 16.08 18.43 0.96 RRS2 4 42.96 9.53 0.50 19.34 21.18 1.11 RRS3 1 41.04 7.07 0.37 22.48 12.90 0.67 RRS3 2 55.08 5.63 0.29 35.64 8.70 0.45 RRS3 3 30.73 8.17 0.43 13.72 18.31 0.96 RRS3 4 42.96 7.25 0.38 21.28 14.65 0.77 RRS4 1 41.04 8.84 0.46 18.46 19.65 1.03 RRS4 2 55.08 5.43 0.28 31.92 9.37 0.49 RRS4 3 30.73 7.36 0.38 12.95 17.45 0.91 RRS4 4 42.96 4.65 0.24 19.27 10.37 0.54 RRS5 1 39.98 7.1 1 0.37 14.11 20.14 1.05 RRS5 2 55.08 7.90 0.41 27.20 16.00 0.84 RRS5 3 35.48 10.04 0.52 9.71 36.68 1.92 RRS5 4 36.97 4.96 0.26 16.62 11.03 0.58 RRS6 1 42.82 5.25 0.27 8.60 26.13 1.37 RRS6 2 52.01 6.26 0.33 24.17 13.46 0.70 RRS6 3 35.48 14.28 0.75 12. 58 40.27 2.11

PAGE 248

248 Table 4 11 Mean quarterly per segment values in Weeki Wachee River for daily raw incident PAR R PAR R efficiency (EE R ), benthic PAR B and benthic PAR R efficiency (EE B ). Segment Quarter PAR R (mol m 2 d 1 ) EE R (%) EE R (gO 2 mol 1 ) PAR B (mol m 2 d 1 ) EE B (%) EE B (gO 2 mol 1 ) WWS1 1 15.41 10.61 0.55 10.17 16.09 0.84 WWS1 2 58.86 3.88 0.20 38.61 5.92 0.31 WWS1 3 38.58 9.36 0.49 27.25 13.25 0.69 WWS1 4 22.90 11.96 0.63 14.93 18.34 0.96 WWS2 1 15.41 8.98 0.47 9.19 15.07 0.79 WWS2 2 58.86 2.73 0.14 36.38 4.42 0.23 WWS2 3 38.58 3.97 0.21 25.84 5.93 0.31 WWS2 4 22.90 13.24 0.69 17.57 17.26 0.90 WWS3 1 15.41 6.92 0.36 8.74 12.20 0.64 WWS3 2 58.86 1.30 0.07 33.59 2.28 0.12 WWS3 3 38.58 3.52 0.18 20.13 6.74 0.35 WWS3 4 22.90 4.82 0.25 13.7 2 8.05 0.42 WWS4 1 15.41 3.65 0.19 5.70 9.85 0.51 WWS4 2 58.86 0.95 0.05 16.58 3.39 0.18 WWS4 3 38.58 2.94 0.15 8.72 12.99 0.68 WWS4 4 22.90 4.69 0.25 7.33 14.65 0.77

PAGE 249

249 Table 4 12 Summary of MLR ranges, non parametric correlation coefficient s (rho) and significance values ( p value) for various levels of subsurface irradiance. SAG Biomass (g m 2 ) SAG % Coverage Broadband MLR Range (%) Rho p value MLR Range (%) Rho p value %PAR 36 40 0.33 0.067 36 40 0.45 0.01 %PUR L 10 25 0.47 0.011 10 14 0.4 0.022 %PUR LE 9 12 0.54 0.003 9 12 0.44 0.011 Spectral %PAR Blue 38 45 0.37 0.035 39 45 0.51 0.003 %PUR L Blue 15 25 0.56 0.002 15 24 0.43 0.015 %PUR LE Blue 12 20 0.64 0.0003 12 20 0.47 0.006 Note: Minimum light requirement (MLR %, percent subsurface irradiance at benthic zone) Subsurface irradiance included: broadband and spectral (blue band of PAR spectrum) percent subsurface irradian ce (%PAR), %PAR weighted for light absorption by Sagittaria kurziana (SAG, %PUR L), and %PUR L weighted for epiphyte light absorption (%PUR LE). Both Rainbow and Weeki Wachee River datasets were used to develop MLR ranges.

PAGE 250

250 Table 4 13. Pearson corre lation results between abiotic and biotic parameters for combined datasets of Rainbow and Weeki Wachee Rivers. Distance Depth Velocity Nitrate TP %PAR Blue %PAR Red PAR %AL SAG Chla content SAG Biomass SAG Biomass 0.46* SAG Chla Content 0 .51* 0.61** 0.64** 0.76*** 0.83*** 0.63** 0.63** 0.63** Epiphyte Chla Content 0.81*** 0.53** 0.81*** 0.71*** SAG Shoot Weight 0.61** 0.55** 0.75*** 0.62** 0.62** 0.62** 0.58** SAG Shoot Density 0.51* SAG Blade Width 0.71*** 0.51** 0.51** 0.51** 0.57** SAG Blade Length 0.81*** 0.44* SAG Leaves per Shoot 0.53* Epiphyte:Shoot Weight 0.57** 0.52** 0.52** 0.52** %AL 0.44* 0.51* 0.78*** %AE 0.58** 0.71*** 0.50* 0.61** 0.69*** PAR 0.47* PAR Red 0.52* %PAR 0.49* %PAR Red 0.58** Velocity 0.78*** (*) indicates (0.05> p >0.01) (**) indicates (0.01> p >0 .001) (***) indicates ( p <0.001)

PAGE 251

251 Table 4 1 4 Pearson correlation results between gross primary productivity (GPP), and abiotic and biotic parameters for combined datasets of Rainbow and Weeki Wachee Rivers. GPP (g O 2 m 2 d 1) Color (PCU) 0.39* S g (m 1 ) 0.38* % Open Canopy 0.64*** Community respiration (g O 2 m 2 d 1 ) 0.72*** Flow (m 3 d 1 ) 0.49** Daily PAR B (umol m 2 s 1 ) 0.41** Native Biomass (g m 2 ) 0.43** Native % Coverage 0.59*** SAG Biomass (g m 2 ) 0.39** SAG % Coverage 0.57*** No (*) indicates (0.05> p >0.01) (**) indicates (0.01> p >0.001) (***) indicates ( p <0.001)

PAGE 252

252 CHAPTER 5 SUMMARY, CONCLUSION S AND RECOMMENDATIONS Summary The objectives of my research were to gain an understand ing of the factors contributing to loss in water clarity and light availability and how these factors influence primary producer distribution and abundance in two Florida springs Optical water quality data was coupled with light attenuation and inherent optical propert y data to det ermine the factors governing light availability in springs. Several research que stions were asked regarding spatial and temporal specificity and the relative con tribution and magnitude of c ontributing light attenuators such as water column constituents and epiphytes A fundamental question with regard to the sufficiency of the q uantity and quality of light available for submerged aquatic vegetation (SAV) w as asked to evaluate the potential for light limitation in these springs. Underwater light re gimes were characterized for Rainbow and Weeki Wachee spring systems. As expected, p articulate matter w as found to be the over arching water column attenuating factor that controlled light availability via absorption and scattering processes in the two spring systems. Epiphytic attenuation was also assessed to determine the extent of direct light attenuation caused b y the attached algal assemblage s associated with beneficial SAV. Spectrally explicit empirical optical models were develop ed from synoptic inherent and apparent optical properties, and optical water quality data to assess historical light attenuation conditions in springs Primary producer ( SAV and algae) were assessed in both Rainbow and We eki Wachee Rivers that included a comprehensive assessment of species c ompositio n (biomass and abundance) dis tribution morphology (shoot system characteristics) and

PAGE 253

253 physiology (relative absorptance and pigment content) A high resolution e cosyste m metabolism evaluation was conducted for both systems that evalua ted gross primary productivity (GPP) community respiration (CR) net ecosystem productivity (NEP) and the effects of l ight and other factors on metabolism estimates on varying spatial and tempo ral scale s Environmental effects of a bioti c and biotic factors on SAV and algae were evaluated to determine if functional relationships exist ed that could provide an explanation of ecological decline in the two s ystems Conclusions My research demonstrated that Florida spring fed rivers are complex optical environments that varied significantly w ithin and among spring systems. A significant longitudinal decli ne in spectr al light availability was found in both systems. Inter annual seasonal and predicted long term temporal variability in optical water quality and light attenuation was not found in these systems. S patially variant wate r column light attenuators were mostly dominated by particulate absorption and scattering ( dominated by non algal detrit al particulate s ) and less so by abs orption by color dissolved organic matter (CDOM) Water column l ight atte nuation ( broadband PAR) ranged from 2 0 to 90% w hich translate d to a range of 10 to 80%PAR ( i.e. percent sub surface irradi ance r emaining following water column attenuation ) for both s prings systems Epiphyt ic attenuation further reduced light ava ilability for SAV by 20 to 75% As a result, when high levels of epiphytic and water colu mn attenuation we re encountered insufficient light was found to reach the le aves of SAV in s everal of the downstream cases. The b roadband optical model s developed from optical water quality and light attenuation data w ere simplistic, but useful. C alibrated p seudo mechanistic spectral

PAGE 254

254 model s were 30% more accurate than un calibrated spectral models. Site specific m ulti linear regression models outperformed and provided nearly 2 fold greater accuracy in optical model output i n comparison to calibrated pseudo mechanistic optical models. CDOM slope analysis found that CDOM consisted primarily of low molecular weight (LMW) mater ial that was likely sourced from aut ochthonous material or reduced to a lower form by photobleaching in the Rainbow River Weeki Wachee River had lower CDOM slopes, which indicated a more recalcitrant allochthonous source of CDOM. The more labile LMW material wa s attributed to higher nutrient removal rates in Rainbow with lower rates in Weeki Wachee River The direct relationships that were found b etween SAV abundance, biomass and primary productivity with CDOM quality and light availability influence s the degree of nutrient removal which presen ts management implications for the receiving waters such as nutrient limited estuaries. Mean GP P estimates did not vary spatially in Rainbow, but declined s ignificantly in Weeki Wachee River. Light w as directly significantly related to GPP for both systems. Other positive correlations were found between GPP and native SAV and Sagittaria kurziana abundance and biomass, flow, and CDO M slope. Color wa s found to be negatively correlated with GPP. Nutrient concentrations (total phosphorus and nitrate) were not found to be correlated with GPP or with filamentous algae. However, total phosphorus (TP) was posi t iv ely correlated with S. kurziana biomass and shoot weigh t leading to greater bio volumes of SAV due to increased TP concentrations B oth TP and nitrate concentrations were negatively correlated with S. kurziana chlorophy ll a pigment content T his suggests that higher nutrient concentrations may be causing a reduction in

PAGE 255

255 chlorophyll a production in S. kurziana which may affect overall GPP and ecosystem metabolism T he p ositive relationsh ips that were found between nutrient concentrations and epiphytic absorptance suggested that nutrients may be inducing greater absorption of light by attached algae, therefore reducing the amount of light available for beneficial SAV. The d ominant macrophyte species, S kurziana was found to be adapted to hig h light conditions from p hotosynthesis i rradiance (P I) curve s suggesting ph otoinhibition in the clear upper portions of the two spring systems. O n the contrary S. kurziana was found to be light l imited in the poor water clarity portions of both lower Rainbow and lower Weeki Wachee Rivers due to considerably reduced blue light availability A minimum blu e light requirement (MBLR) of 38 to 45% was concluded from field synoptic surveys of spectral light availability and biomass measurements A cco rdingly m anagement implications regarding the e ffects of nutrients on beneficial SAV morphology and epi phyt ic physiology must also consider the p ote ntial impacts of light limitation due to increased light attenuation in spring systems. This conclusion supports the contention that spectral knowledge of light availability is imperative to meet management goals for future restoration. Recommendations As mentioned previously, o ptical water quality coupled with inherent and apparent optical propert y d ata can greatly inform which factors are driving water clarity decline s in aqu atic resource s. Only two first magnitude springs were comprehensively evaluated for spectral distribution and optical water quality as part of this study. Opportunity ab ounds for synoptic investigations into th e spectral quality and quantity of light in more than 30 other first magnitude springs in Florid a H undreds of other smaller

PAGE 256

256 springs i n the sta te would also greatly benefit from this type of work that would lead to t he develo pment and calibrat ion of site specific spectral models. Additional d ata collection in Rainbow and Weeki Wachee Rivers could also i mprove the modeling component of this study for re calibration and validatio n of the spectral models to enhance robustness and reduce uncertainty of model output results. Sit e specificity is key in obtaining high ly accurate model output, and should be a major consideration when developing and utilizing optical models. Additionally, f urther information regarding the character of turbidity such as the particle size distribution and composition of mineral and organic frac tions of particulate matter, a long with knowledge of the epiphyt ic community composition c ould strengthen opt ical model input and interpretation of light attenuation assessment results. Future studies regarding biogeochemical and nutrient cycling should include a focused evaluation of CDOM source and quality This can be carried out through measurement of lignin fluxes, CDOM m olecular weight, and how these factors may be affected by hydrologic dynamics that lead to changes in inputs from allochthonous sources such as riparian weltands. An in depth evaluation of CDOM quality in relation to in situ nutrient removal would provide resource managers the ability to plan in system nutrient removal action plans to reduce nutrient s in downstream receiving water bodies. The spectrally specific optical modeling methods, coupled with the GPP PAR B relationship d eveloped as part of this study advances the currently available approaches to predict GPP and to determine if spectral light availability is sufficient for SAV growth and survival. Future work abounds that could include investigations into the

PAGE 257

257 effects of light intensities and phototroph type and biomass on components of the P I c urve and overa ll GPP rates. The MBLR was recommended to allow sufficient useable light in the blue range of the PAR spectrum that should support the growth and survival of S. kurziana This light requirement accounts for variable levels of environmental conditions or stressors such as varying epiphyte coverage, sediment quality, velocity rates, nutrient fluxes, grazers or other disturbances that wer e encountered by S. kurziana during this study. T he results from this field scale synoptic study warrant an experiment to c onfirm the suggested MBLR However, th e results from this study provide sufficient ev idence that should compel resource managers to consider the spectr al quality of light, specifically t he MBLR in future springs restoration activities where S. kurziana is the target species for restoration.

PAGE 258

258 LIST OF REFERENCES ABAL, E.G., N. LONERAGAN, P. BOWEN, C. J. PERRY, J. W. UDY, W. C. DENNISON. Physiological and morphological responses of the seagrass Zostera capricorni Asc hers. to light intensity. J. Exp. Mar. Biol. Ecol. 178 : 113 129. ALBERTIN, A. R. 2009. Nutrient dynamics in Florida springs and relationships to algal blooms. Ph D Thesis. Univ. of Florida. ALLEN, J. AND M. M. CASTILLO. 2007. Structure and function of running waters, 2 nd ed. Princeton University Press. ANASTASIOU, C. J. 2009. Characterization of the underwater light environment and its relevance to seagrass recovery and sustainability in Tampa Bay, Florida. Ph D Thesis. Univ. of South Florida. AT KINS. 20 12 Rainbow River vegetation evaluation Southwest Florida Water Management District. BIBER, P. D C. L. GALLEGOS, AND W. KENWORTHY. 2008. Calibration of a b io optical m odel in the North River, North Carolin a (Albemarle Pamlico Sound): A t ool to evaluate water quality impacts on s eagrasses. Estuaries and Coasts 31: 177 191. ______ H. W. PAERL, C. L. GALLEGOS, W. J. KENWORTHY. 2005. Evaluating indicators of seagrass stress to light. Chp 13 In S. A. Bortone [ed.], Estuarine Indicators: Proce edings of Estua rine Indicators Workshop. Sanibel Captiva Cons ervation Foundation. CRC Press. ______ W. J. KENWORTHY, AND H. W. PAERL. 2009. Experimental analysis of the response and recovery of Zostera marina (l.) and Halodule wrightii (Ascher.) to re peated light limitation stres s. J. Exp. Mar. Bio l Eco l 369 : 110 117 BIGGS, B. J. 1996. Hydraulic habitat of plants in streams. Regulated Rivers Res. Ma nag. 12 : 1 31 144. ______, R. A. SMITH, AND M. J. DUNCAN 1999. Velocity and sediment disturbance of periphyton in headwater streams: biomass and metabolism. J. North Am. Benthol. Soc. 18 : 222 241. BOSS, E., AND J. R. ZANEVELD 2003 The effect of bottom substrate on inherent optical properties: Evidence of biogeochemical processes Limnol. Oceanogr 48 : 346 354. BRAUN BLANQUET, J. 1965. Plant sociology: the study of plant c ommunities. Haffner.

PAGE 259

259 BRICAUD, A., AND D. STRAMSKI 1990 Spectral absorption of living phytoplankton and nonalgal biogenus matter. A comparison between the Peru upwelling are a and the Sargasso Sea. Limnol. Oceanogr. 35 : 562 582. ______, A. MOREL, AND L. PRIEUR. 1981 Absorption by dissolved organic matter of the sea (yellow substance) in the UV and visible domains. Limnol. Oceanogr. 26 : 43 53. BROWN, M. T., K. C REISS, M J. COHEN, J. M. EVANS, T. K. FRAZER, C. A. JACOBY, E. J. PHLIPS, R. L. KNIGHT, S. K. NOTESTEIN, AND K. A. MCKEE. 2008. Summary and synthesis of the available literature on the effects of nutrients on spring organism and systems. Florida Department of Environmental Protection. CANFIELD, D. E., AND M. V. HOYER. 1988. Influence of nutrient enrichment and light availability on the abundance of aquatic macrophytes in Florida streams. Can. J. Fish. Aquat. Sci. 45 : 1467 1472. CARRUTHERS, T. J., B. J. L ONGSTAFF, W. C. DENNISON, E. G. ABAL, AND K. AIOI. 2001. Measurement of light penetration in relation to seagrass, p. 369 392. In F T. SHORT, AND R. G. COLE [eds.], Global Seagrass Research Methods, Amsterdam, Netherlands: Elsevier Science B. V. CICHR A, C., S. HOLLAND J. STEPHENS, AND A. CROTEAU 2013. Recreational impacts on the Rainbow River. Aquatics: A Publication of the Florida Aquatic Plant Management Society 35 : 9 14. COBLE, P. G. 2007 Marine optical biogeochemistry: The chemistry of ocean color. Chemical Reviews 107 : 17. COHEN, M. J. 2007. Sources, transport and transformations of nitrate N in the Florida environment. Special Publication SJ2007 SP10. St. Johns River Water Management District. COPELAND, B. J., AND W. R. DUFFER. 1964. U se of clear plastic dome to measure gaseous diffusion rates in natural waters. Limnol. Oceanogr. 9 : 494 499. COWELL, B. C. AND C. J. DAWES. 2005. Sources of chlorophyll a in Rainbow River, Florida compared to three spring fed rivers with varying dissol ved inorganic nitrogen concentrations (DIN). Southwest Flor ida Water Management District. ______, AND C. J. DAWES. 2007. The influence of phosphate on phytoplankton in the Rainbow River, Florida. Southwest Florida Water Management District DAVIES COLLEY R. J. 1988. Measuring water clarity with a black disk. Limnol. Oceanogr. 33: 616 623.

PAGE 260

260 ______, AND M. E. CLOSE 1990 Water colour and clarity of New Zealand rivers under baseflow conditions. N. Z. J. Mar. Freshwater Res. 24 : 357 365. ______, AND D. G. SMITH. 1995. Optically pure waters in Waikoropupu ('Pupu' ) Springs, Nelson, New Zealand. N. Z. J. Mar. Freshwater Res 29 : 251 256. ______, AND J. W. Nagels. 2008 Predicting light penetration into river waters J. Geophys. Res. 113 : G03028 DE BRABAN DERE, L., T. K. FRAZER, AND J. P. MONTOYA. 2007. Stable nitrogen isotope ratios of macrophytes and associated periphyton along a nitrate gradient in two subtropical, spring fed streams. Freshw. Biol. 52 : 1564 1575. DENNISON, W. C., R. J. ORTH, K. A. MOORE J. C. STEVENSON, V. CARTER S. KOLLAR, P. W. BERGST ROM, AND R. A. BATIUK 1993. Assessing water quality with submersed aquatic vegetation. BioScience 43: 86 94. DIXON, K. L., AND G. J. KIRKPATRICK. 1995. Light attenuation with respect to seagrasses in Sarasota Bay, Florida. Sarasota Bay National Estuary Program. ______, AND G. J. KIRKPATRICK. 1999. Causes of light attenuation with respect to seagrasses in upper and lower Charlotte Harbor. Southwest Florida Water Management District. ______, E R. HALL, AND G. J. KIRKPATRICK. 2010. A spectrally explicit optical model of attenuation for Charlotte Harbor seagrasses. Southwest Florida Water Management District. DOYLE R. D. AND R. M. SMART. 1998 Competitive reduction of noxious Lyngbya wollei mats by rooted aquatic plants. Aquat. Bot. 61 : 17 32. DRAKE, L. A., F. C. DOBBS, AND R. C. ZIMMERMAN 2003 Effects of epiphyte load on optical properties and photosynthetic potential of the seagrasses Thallasia testudinum Bank ex Konig and Zostera mari na L. Limnol. Oceanogr. 48 : 456 463. DUARTE, C. M. AND D. E. CANFIELD. 1990. Macrophyte standing crop and primary production in some Flo rida spring runs. Water Resour. B ull. 26 : 927 934. DUARTE, C. M. 1991 Seagrass depth limits. Aquat. Bot 40 : 363 3 77. ______. 1995. Submerged aquatic vegetation in relation to different nutrient regimes Ophelia 41 : 87 112.

PAGE 261

261 ______, R. MARTINEZ, Y. T. PRAIRIE, T. K. FRAZER, M. V. HOYER, S. K. NOTESTEIN, AND D. E. CANFIELD. 2010 Rapid accretion of dissolved organic carbon in the Springs of Florida: the most organic poor natural waters Biogeosciences Discuss., 7 : 5253 5267 DUNCAN, M. J. 1990. Measurement of light energy, p. 147 151. In R. C. Phillips and C. P. McRoy [eds.], Seagrass Research Methods. UNESCO. D URAKO, M. J. 2007. Leaf optical properties and photosynthetic leaf absorptances in several Australian seagrasses. Aquat. Bot. 87 : 83 89. ______ P. KOWALCZUK, M. A. MALLIN, W. J. COOPER J. J. SOUZA, AND D. H. WELLS 2010 Interannual variation in photo synthetically significant optical properties and water quality in a coastal blackwater river plume Estuaries and Coasts 33 : 1430 1441. EFFLER, S. 1985. Attenuation versus transparency. J. Environ. Eng. 8 : 448 459. ENRIQUEZ, S., S. AGUSTI, AND C. M. DU ARTE, C. M 19 94. Light absorption by marine macrophytes Oecologia 98 : 121 129. FALKOWSKI, P. G. AND J. A. RAVEN. 2007. Aquatic Photosynthesis, 2 nd ed. Princeton. FELLOWS, C. S., H. M VALETT, C. N. DAHM, P. J. MULHOLLAND, AND S. A. THOMAS. 2006. C oupling nutrient uptake and energy flow in headwater streams. Ecosystems 9 : 788 804. FRAZER, T. K., M. V. HOYER, S. K. NOTESTEIN, J. A. HALE AND D. E. CANFIELD. 2001 a Physical, chemical and vegetative characteristics of five gulf coast rivers. Southwe st Florida Water Management District. ______, E. J. PHLIPS, S. K. NOTESTEIN AND C. JETT. 2001 b Nutrient limiting status of phytoplankton in five gulf coast rivers and their associated estuaries. Southwest Florida Water Management District ______, S.K NOTESTEIN, AND W.E PINE, JR 2006. Changes in the physical, chemical and vegetative characteristics of the Homosassa, Chassahowitzka,and Weeki Wachee Rivers. Southwest Florida Water Management District. GALLEGOS, C. L. 1993. Determination of optical water quality requirements for growth of seagrasses in the Indian River near Ft. Pierce, Florida, with emphasis on the impact of colored water discharges. South Flo rida Water Management District. ______ 1994. Refining habitat requirements of submersed aquatic vegetation: Role of optical models. Estuaries 17 : 187 199.

PAGE 262

262 ______ 2001. Calculating optical water quality targets to restore and protect submerged aquatic vegetation: overcoming problems in portioning the diffuse attenuation coefficient for photo synthetically active radiation. Estuaries 24 : 381 397. ______. 2005. Optical water quality of a blackwater river estuary: the Lower St. Johns River, Florida, USA. Estuarine Coastal and Shelf Science 63 : 57 72 ______, D. L. CORRELL, AND J. W. PIERCE. 199 0. Modelling spectral diffuse attenuation, absorption, and scattering coefficient s in a turbid estuary. Limnol. Oceanogr. 35 : 1486 1502. ______, AND W. J. KENWORTHY. 1996. Seagrass depth limits in the Indian River Lagoon: Application of an optical water q uality model. Estuarine Coastal and Shelf Sciences 42 :267 288. ______, T. E. JORDAN, A. H. HINES, AND D. E. WELLER. 2005. Temporal variability of optical properties in a shallow, eutrophic estuary: Seasonal and interannual variability. Estuarine, Coastal and Shelf Science 64 : 156 170. ______, W. J. KENWORTHY, P. D. BIBER, AND B. S. WOLFE. 2009. Underwater spectral energy distribution and seagrass depth limits along an optical water quality gradient. Smithsonian Contributions to the Marine Sciences 38 : 359 367. HALL, R. O., AND J. L. TANK 2005. Correcting whole stream estimates of metabolism for groundwat er input. Limnol. Oceanogr.: Methods 3 : 222 229. HAUXWELL, J. A., T. K. FRAZER, AND C. W. OSENBERG. 2007. An annual cycle of biomass and productivity of Vallisneria americana in a subtropical sp ring fed estuary. Aquat. Bot. 87 : 61 68. HEFFERNAN, J. B., A ND M. J. COHEN. 2010. Direct and indirect coupling of productivity and diel nitrate dynamics in a subtropica l spring fed river. Limnol. Oceanogr. 55 : 677 688. ______, D. M. LIEBOWITZ, T. K. FRAZER, J. M. EVANS, AND M. J. COHEN. 2010 a Algal blooms and the nitrogen enrichment hypothesis in Florida springs: Evidence, alternatives, and adaptive management. Ecol. Appl. 20 : 816 829. ______, M. J. COHEN, T. K. FRAZER, R. G. THOMAS, T. J. RAYFIELD, J. GULLEY, J. B. MARTIN, J. J. DELFINO, AND W. D. GRAHAM. 2010 b Hydrologic and biotic influences on nitrate removal in a sub tropical spring fed river. Limnol. Oceanogr. 55 : 249 263.

PAGE 263

263 HELMS, J. R., A. STUBBINS, J. D. R ITCHIE, E. C. MINOR D. J. KIEBER, AND K. MOPPER. 2008 Absorption spectral slopes and slope ra tios as indicators of molecular weight, source, and photobleaching of chromophoric dissolved organic matter. Limnol. Oceanogr. 53 : 955 969. HOYER, M. T., T. K. FRAZER, S. K. NOTESTEIN, AND D. E. CANFIELD. 2004. Vegetative characteristics of three low lyi ng Florida coastal rivers in relation to flow, light, salinity and nutrients. Hydrobiologia 528 : 31 43. JONES, G. W., S. B. UPCHURCH AND K. M. CHAMPION. 1996. Origin of nutrients in ground water discharging from Rainbow Springs, Marion County, Florida. Southwest Florida Water Management District JULIAN, J. P., M. W. DOYLE, S. M. POWERS, E. H. STANLEY, AND J A. RIGGSBEE. 2008b a Optical water quality in rivers. Water Resour. Res. 44 : W10411 ______, M. W DOYLE, AND E. H. STANLEY 2008 b Empirical mo deling of light availability in rivers. J. of Geophys. Res. 113 : G03022 ______, E. H. STANLE Y, AND M. W. DOYLE 2008 c Basin scale consequences of agricultural land use on benthic light availability and primary production along a sixth order temperate river. Ecosystems 11 : 1091 1105. ______, DAVIES COLLEY, R. J., C. L. GALLEGOS, AND T.V. TRAN. 2013. Optical Water Quality of Inland Waters: A Landscape Perspective Annals of the Association of American Geographers 103 : 309 318. KELBLE, C. R., P. B. ORTINER, G. L. MITCHCOCK, AND J. N. BOYER. 2005. Attenuation of photosynthetically availabl e radiation (PAR) in Florida Bay: potential for light limitation of primary producers. Estuaries. 28 : 560 571. KEMP, W. M. AND W. R. BOYNTON. 2004. Productivity, trophic structure, and energy flow in the steady state ecosystems of Silver Springs, Florid a. Ecological Modelling 178 : 43 49. KE NWORTHY, W. J., AND M. S. FONSECA. 1996 Light requirements of seagrasses Halodule wrightii and Syringodium filiforme derived from the relationship between diffuse light attenuation and maximum depth distribution Estuaries 19 : 740 750. KING, S. A. 2012. Effects of flow on filamentous algae and nutrient limitation in lotic systems. Ph D Thesis. Univ. of Florida. ______, J. B. HEFFERNAN, AND M. J. COHEN 2014 Nutrient flux, uptake, and autotrophic limitation in streams and rivers. Freshw Sci. 33 : 1 16.

PAGE 264

264 KIRK, J. T. O. 1980. Spectral absorption properties of natural waters: contribution of the soluble and particulate fractions to light absorption in some inland waters of South eastern Australia. Aust. J. Ma r. Freshwater Res. 31 : 287 296. ______. 1984. Dependence of relationship between inherent and apparent optical properties of water on solar altitude. Limnol. Oceanogr. 29 : 350 356 ______. 1988. Optical water quality what does it mean and how should w e measure it? Journal of Water Pollution Control Federation 60 : 194 197. ______. 1991 Volume scattering function, average cosines, and the underwater light field. Limnol. Oceanogr. 36 : 455 467. ______. 1994. Light and photosynthesis in aquatic ecosys tems, 2nd ed. Cambridge. KISHINO, M., C. R. BOOTH, AND N. OKAMI 1984. Underwater radiant energy absorbed by phytoplankton, detritus, dissolved organic matter, and pure water. Limnol. Oceanogr. 22 : 340 349. ______, TAKAHASHI, N., N. OKAMI, S. ICHIMUR A 1985 Estimation of the spectral absorption coefficients of phytoplankton in the sea Bulletin of Marine Sciences 37 : 634 642. KNIGHT, R. L. AND S. K. what do we know? The Fl orida Watershed Journal 11 15 KRAUSE JENSEN, D., AND K. SAND JENSEN. 1998. Light Attenuation and photosynthesis of aquatic plant communities. Limnol. Oceanogr. 43 : 396 407. LAURION, I., M. VENTURA J. CATALAN, R. PSENN ER, AND R. SOMMARUGA 2000 Attenuation of ultraviolet radiation in mou ntain lakes: Factors controlling the amongh and within lake variability. Limnol. Oceanogr 45 : 1274 1288. LECONTE, J. 1861. On the optical phenomena presented by the Silver Spring in Marion County, Florida. American Journal of Science and Arts 31: 1 1 2. LE C. E H. CHUANMIN, D. A. ENGLISH, J. A. CANNI ZZARO, Z. A. CHEN, C. KOVACH, C. J. ANASTA SIOU, J. ZHAO, K. L. CARDER. 2013. Inherent and apparent optical properties of the complex estuarine waters of Tampa Bay: What controls light? Estuarine, Coas t al and Shelf Science 117 : 54 69 LICHTENTHALER, H. K. 1987. Chlorophylls and carotenoids: pigments of photosynthetic biomembranes. Methods in Enzymology 148 : 350 382. LIEBOWITZ, D. M. 2013. Environmentally mediated consumer control of algae proliferatio n in Florida springs. Ph D Thesis. Univ. of Florida.

PAGE 265

265 MARGAGER, S., AND W. F. VINCENT 2000 Spectral light attenuation and the absorption of UV and blue light in natural waters. Limnol. Oceanogr. 45 : 642 650. MARZOLF, E R. P. J. MULLHOLAND, AND A. D. S TEINMAN. 1994. Improvements to the diurnal upstream downstream dissolved oxygen change technique for determining whole stream metabol ism in small streams. Can. J. Fish. Aquat. Sci. 51 : 1591 1599. MATHESON, F. E., J. M. QUINN, AND M. L. MARTIN. 2012 Ef fects of irradiance on diel and seasonal patterns of nutrient uptake by stream periphyton. Freshw. Biol. 57 : 1617 1630. MCCUTCHAN, J. H., J. F. SAUNDERS, W. M. L EWIS, AND M. G. HAYD EN 2002. Effects of groundwater flux on open channel estimates of strea m metabolism. Limnol. Oceanogr. 47 :321 324. MCPHERSON, B. F AND R. L. MILLER. 1994. Causes of light attenuation in Tampa Bay and Charlotte Harbor, Southwestern Florida. Water Res our Bull. 30 : 43 53. MIDDELBOE A. L. AND S MARKAGER 1997 Depth limit s and minimum light requirements of freshwater macrophytes. Freshw. Biol. 37 : 553 568 MILLER, R. L., AND B. F. MCPHERSON. 1995. Modeling Light Available to Seagrasses in Tampa Bay, Florida. Florida S cientist 58: 116 125 MOBLEY, C. D 1994 Light and w ater. Radiative transfer in natural waters. Academic Press. MONTES HUGO, M. A., S. ALVAREZ BORREGO, AND A. D. G ILES GUZMAN 2003 Horizontal sighting range and secchi depth as estimators of underwater PAR attenuation in a coastal lagoon. Estuaries 26 : 1 302 1309 MOREL, A. 1978 Available, usable, and stored radiant energy in relation to marine photosynthesis. Deep Sea Research 25 : 673 688 MUMMA, M. T., C. E CICHRA., AND J. SOWARDS 199 6. Effects of recreation on the submersed aquatic plant community of Rainbow River, Flor ida. J. Aq. Plant Manag. 34 : 53 56. MULHOLLAND, P. J., C S. FELLOWS, J. L. TANK, N. B. GRIMM, J R. WEBSTER, S. K. HAMILTON, E. MART I, L. ASHKENAS, W. B BOWDEN, W. K. DODD S, W. H. MCDOWELL, M. J. P AUL, AND B. J. PETER SON 2001. Inter biome comparison of factors controll ing stream metabolism. Freshw. Biol. 46 :1503 1517.

PAGE 266

266 MUNCH, D. A., D. J. TOTH, C. HUANG, J. B. DAVIS, C. M. FORTICH, W. L. OSBURN, E. J. PHLIPS, E. L. QUINLAN, M. S. ALLEN, M. J. WOODS, P. COONEY, R. L. KNIGHT, R. A. CLARKE, AND S. L. KNIGHT. 2006. Fifty year retrospective study of the ecology of Silver Springs, Florida. Florida Department of Environmental Protection MITCHELL, B. G., KAHRU, M. WIELAND, J. AND M. STRAMSKA. 2003. In J. L. Mueller, G. S. Fargion, and C. R. McClain [eds], Ocean optics protocols for satellite ocean color sensor validation, revision 4, volume IV: Inherent optical properties: instruments, characterizations, field measurements and data analysis protocols. National Aeronautics and Space Administration (NASA). NOTESTEIN, S. K., T. K. FRAZER, M. V. HOYER, AND D. E. CANFIELD. 2003. Nutrient limitation of periphyton in a spring fed, coastal stream in Florida,USA. J. Aq. Plant Manag. 41 : 57 60. ODUM, H. T. 1956. Primary Production in Flow ing Waters. Limnol. Oceanogr. 1 :102 117. ______. 1957 a Primary production measurements in eleven Florida springs and a marine turtle grass community. Limnol. Oceanogr. 2 :85 97. ______. 1957 b Trophic structure and productivity of Silver Springs, Flor ida. Ecolog. Monogr 27 :55 112. PEREZ, G. L., A. TORREMORELL, J. B USTINGORRY, R. ESCAR AY, P. PEREZ, M. DIEGUEZ 2010 Optical characteristics of shallow lakes from the Pampa and Patagonia regions of Argentina. Limnologica Ecology and Management of Inl and Waters 40 : 30 39. PFANNKUCHE, J. 2002 Optical properties of Otago Shelf waters: South Island New Zealand Estuarine, Coastal and Shelf Science 55 : 613 627. POPE, R. M ., AND E. S. FRY. 1997 Absorption spectrum (380 700 nm) of pure water. II. Integ rating cavity measurements. Applied Optics 36 : 8710 8723. PINOWSKA, A., R. J. STEVENSON, J. O. SICKMAN, A. ALBERTIN AND M. ANDERSON. 2007. Integrated interpretation of survey for determining nutrient thresholds for macroalgae in Florida springs, Florida Department of Environmental Protection PREISENDORFER, R. W. 1986. Secchi disk science: Visual optics of natural waters. Limnol. Oceanogr. 31 : 909 926.

PAGE 267

267 QUINLAN, E. L., E. J. PHLIPS, K.A. DONNELLY, C.H. JETT, P. SLESZYNSKI AND S. KELLER. 2008. Pr imary producers and nutrient loading in Silver Springs, FL, USA. Aquat. Bot. 88 : 247 255. RILEY, A. J., AND W. K. DODDS. 2013. Whole stream metabolism: strategies for measuring and modeling diel trends of dissolved oxygen. Freshw Sci. 32 : 56 69. ROSE MOND, A. D. 1993. Interactions among irradiance, nutrients, and herbivores constrain a stream algal community. Oecologia 94 : 585 594. ROSENAU, J. C., G. L. FAULKNER, C. W. HENDRY JR. AND R. W. HULL. 1977. Bulletin 31: Springs of Florida. Florida D ep artment of Natural Resources. SAND JENSEN, K. AND J. BORUM. 1983. Regulation of growth of eelgrass ( Zostera marina L.) in Danish coastal waters. Mar. Techn. Soc. J. 17 : 15 21. ______, AND J. BORUM. 1991. Interactions among phytoplankton, periphyton, and macrophytes in temperate fr eshwaters and estuaries. Aquat. Bot. 41 : 137 175. ______, AND T. V. MADSEN. 1991. Minimum light requirements of submerged freshwater macrophytes in laborator y growth experiments. J. Ecology. 79 : 749 764. SCOTT, T., AND OTHERS. 2004. Bulletin 66 : Springs of Florida. Florida Geological Survey. SLESZYNSKI, P. A. 2009. Ecology of Sagittaria kurziana and distribution of periphyton in the spring fed Rainbow River. Ph D Thesis. Univ. of Florida. SMITH, R. C. AND K. S. BAK ER. 1981. Optical properties of the clearest natural waters (200 800 nm). Applied Optics 20 : 177 184. SOUTHWEST FLORIDA WATER MANAGEMENT DISTRICT (SWFWMD) 2004. Surface Water Improvement and Management Plan for the Rain bow River. SPENCER, G. M., P. J. HERNES, R. RUF A. BAKER R. Y. DYDA, A. STUBBINS 2010 Temporal controls on dissolved organic matter and lignin biogeochemistry in a pristine tropical river, Democratic Republic of Congo. J. Geophys. Res. 115 : G03013 STEVENSON R.J., A. PINOWSKA A. ALBERTIN A., AND J. O. SICKMAN. 2007. Ecological condition of algae and nutrients in Florida springs: The Synthesis Report. WM858, Florida Department of Environmental P rotection.

PAGE 268

268 SWIFT, T. J., J. PEREZ LOSADA, S. G. SCHLAD OW, J. E. REUTER, AN D C. R. G OLDMAN. 2006 Water clarity modeling in Lake Tahoe: Linking suspended matter characteristics to Secchi depth Aquat. Sci. 68 : 1 15. TOMASKO, D. A., AND M. O. HALL. 1999. Productivity and biomass of the seagrass Thalassia testudinum along a gradient of f reshwater influence in Charlott e Harbor, Florida. Estuaries 22 : 592 602. TZORTZIOU, M., J. R HERMAN, C. L. GALLEG OS, P. J. NEALE, A. SU BRAMANIAM, AND L. W. HARDING 2006 Bio optics of the Chesapeake Bay from measurements and radiataive transfer closu re. Esturarine, Coastal and Shelf Science 68 : 348 362. WETLAND SOLUTIONS (WSI). 2007. Pollutant load reduction goal analysis for the Wekiva River and Rock Springs Run, Florida. Special Publication SJ2008 SP5. St. Johns River Water Management District. ______ 2010. An ecosystem Conservation Commission. WHITFORD L. A. 1956. The communities of algae in the springs and spring streams of Florida. Ecology 37 : 433 442. YAMASHITA, Y., Y. NOSAKA, K. SUZUKI, H. OGAWA, K. TAKAHASHI, AND H. SAITO 2013 Photobleaching as a factor controlling spectral characteristics of chromophoric dissolved organic matter in open ocean. Biogeosciences 10 : 7207 7217. YOUNG, R. G, AND A. D. HURYN. 1999. Effects of land use on stream metabolism and organic matt er turnover. Ecol. Appl. 9 : 1359 1376. ZIMMERMAN, R C. 2003 A biooptical model of irradiance distribution and photosynthesis in seagrass canopies. Limnol. Oceanogr. 48 : 568 585

PAGE 269

269 BIOGRAPHICAL SKETCH Mary S zafraniec grew up in California where she spent most of her free time swimming and SCUBA diving in the Pacific Ocean. After she moved to Florida in high school, she immediately began exploring the beauty of the Atlantic Ocean and the Gulf of Mexico waters. To expand h er knowledge of the environment, she received her Bachelor of Science degree in Biology at the University of South Florida. She worked as an environmental scientist at the Florida Department of Environmental Protection in the Watershed and Reso urce Management Program for four years before being hired by the Southwest Florida Water Management District and worked in the Surface Water Improvement and Management Program to restore water quality and habitat in lakes, rivers, and estuaries. To strengt hen her knowledge and understanding of aquatic ecology and ecological engineering principles, she received h er Master of Science degree in Environmental Engineering Sciences at the University of Florida. Her love of aquatic resources and passion for spring ecosystems drove her to continue on and receive her Doctor of Philosophy degree in Environmental Engineering Sciences at the University of Florida in 2014