Environmentally Mediated Consumer Control of Algae Proliferation in Florida Springs

Material Information

Environmentally Mediated Consumer Control of Algae Proliferation in Florida Springs
Liebowitz, Dina M
Place of Publication:
[Gainesville, Fla.]
University of Florida
Publication Date:
Physical Description:
1 online resource (122 p.)

Thesis/Dissertation Information

Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Interdisciplinary Ecology
Committee Chair:
Cohen, Matthew J
Committee Members:
Phlips, Edward J
Frazer, Tom K
Brenner, Mark
Heffernan, James B
Graduation Date:


Subjects / Keywords:
Algae ( jstor )
Biomass ( jstor )
Ecology ( jstor )
Ecosystems ( jstor )
Flumes ( jstor )
Grazing ( jstor )
Herbivores ( jstor )
Hypoxia ( jstor )
Oxygen ( jstor )
Snails ( jstor )
Interdisciplinary Ecology -- Dissertations, Academic -- UF
algae -- grazers -- hypoxia -- hysteresis
City of Tallahassee ( local )
bibliography ( marcgt )
theses ( marcgt )
government publication (state, provincial, terriorial, dependent) ( marcgt )
born-digital ( sobekcm )
Electronic Thesis or Dissertation
Interdisciplinary Ecology thesis, Ph.D.


Herbivores control autotroph biomass across many aquatic ecosystems, therefore the loss of grazing pressure can induce dramatic changes in autotroph abundance and composition. Symptoms of consumer loss in aquatic systems (e.g., nuisance algae accumulation) may be similar to those of the alleviation of nutrient limitation, necessitating discrimination between these mechanisms. We tested the hypothesis that the loss of top-down control by aquatic gastropods explains benthic filamentous algae proliferation in Florida’s iconic springs and spring-fed rivers. We further hypothesized that the dominant grazer control is  dissolved oxygen (DO) concentration, which varies substantially among, and longitudinally within, springs. Finally, we hypothesized that algae-domination may persist if algae biomass exceeds a critical herbivore escape level, beyond which grazers can no longer constrain algae accumulation. We used three scales of investigation to test these hypotheses: state-wide observational surveys, in situ grazer experiments in Ichetucknee Springs, and artificial stream mesocosms with controlled dissolved oxygen levels. We observed a strong and temporally consistent negative association between algae and gastropod biomass in the surveys (gastropods predicting  up to 40% of the algal variation), and gastropods were by far the most important explanatory variable in multivariate prediction models as well. DO concentrations were significantly, but only modestly, positively associated with gastropod density in field surveys, but hypoxic conditions strongly reduced gastropod survival and grazing rates in controlled experiments. The in situ grazing experiments showed that high, but ecologically relevant, densities of grazers (>100 g m-2wet weight Elimia spp.) could inhibit algal bloom formation, but even > 330 g m-2 of Elimia spp. could not reduce high algal biomass consistently, suggesting that algal biomass may escape herbivore control and entrain spring systems in algal-dominated states. Recognition that the fauna are not merely passive recipients of the system effects, but direct drivers themselves, may suggest new approaches to restoration and help focus future research. ( en )
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
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 (Ph.D.)--University of Florida, 2013.
Adviser: Cohen, Matthew J.
Statement of Responsibility:
by Dina M Liebowitz.

Record Information

Source Institution:
Rights Management:
Copyright Liebowitz, Dina M. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
887737141 ( OCLC )
LD1780 2013 ( lcc )


This item has the following downloads:

Full Text




2 2013 Dina M. Liebowitz


3 To Harold and Judy Liebowitz


4 ACKNOWLEDGMENTS I would like to express my deepest gratitude to my advisor, Matt Cohen, for his endless encouragement scientific curiosity, brilliance, kindness, and good humor. He has been an inspiration to me throughout this process, and I cannot thank him enough. I also sincerely thank my wonderful committee, Tom Frazer, Jim Heffernan, Ed Phlips, and Mark Brenner for their insights patience, and support I could not have accomplished all the gator defying field work without the help the mild mannered man of steel, Larry Korhnak. I am also extremely grateful to many people f or field and lab assistance, including Crystal Hartman, Ashley Aarenson Jen n Dragunchuk, Josh Bellamy, Loren Matthews Chad Foster, David Kaplan, Jonathan Field, and Jennifer Hoyer M any so urces of financial support all owed me to focus on this work. T he NSF IGERT program on Adaptive Man agement: Wetlands, Water, and Watersheds under provided financial support and a stimulating intellectual community of colleagues. Assistantships through t he School of Natural Resources and the Environment, teaching experience with the biology department, and a grant from the Three Rivers Fnpc, Inc. provided generous additional financial support and training experience s Last but far from least, I send warm thanks to my amazing family and friends for all their love humor, creati vity, and general fabulousness My parents, Harold and Judy Liebowitz, my sisters, Debbie Kantor Naomi Maron, and Esther Glahn, all my beloved nieces and nephews, and my aunt Sharon Green and uncle Jay Friedman were all bastion s of support and encourageme nt throughout this send warm thanks to my friends who were particularly involved in motivating me th rough the


5 endgame: Margaret Tolbert Gaby stocks, Jennifer Hoyer, Kelly Biedenweg, Bill Bryson, Daphna Davidson, Jesse Rodin, Jonat han Field, Sarah Norman, Sonora and Kevin Thomas Kathleen McKee, and Danielle and Adam Watts. Thank you all!


6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF FIGURES ................................ ................................ ................................ .......... 9 ABSTRACT ................................ ................................ ................................ ................... 11 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 13 Top Down Versus Bottom Up Contr ol of Algae Proliferation ................................ .. 13 Controls of Aquatic Grazer Populations ................................ ................................ .. 14 ................................ ............ 15 2 ECOSYSTEM LEVEL PATTERNS OF CONSUMER CONTROL OF ALGAL ................................ .......................... 17 Introduction ................................ ................................ ................................ ............. 17 ................................ ............. 18 Hypotheses ................................ ................................ ................................ ...... 20 Methods ................................ ................................ ................................ .................. 21 Study Site Selection ................................ ................................ ......................... 21 Field Sampling Design ................................ ................................ ...................... 22 Field Sampling Protocols ................................ ................................ .................. 22 Biomass Sample Processing ................................ ................................ ............ 24 Data Analysis ................................ ................................ ................................ ... 24 Results ................................ ................................ ................................ .................... 27 Biomass Composition and Predictors of Distribution Patterns .......................... 27 Algae Distributions ................................ ................................ ........................... 28 Gastropod Distributions ................................ ................................ .................... 29 Longitudinal Patterns ................................ ................................ ........................ 30 Bimodality ................................ ................................ ................................ ......... 31 Discussion ................................ ................................ ................................ .............. 32 Gastropod Control of Filamentous Algae Blooms ................................ ............. 32 Dissolved Oxygen and Additional Controls on Gastropods .............................. 34 Gastropods, Algae, and Hysteretic Ecosystem Change ................................ ... 38 Management Implications ................................ ................................ ................. 39 3 EXPERIMENTAL EVIDENCE OF GRAZER CONTROL OF FILAMENTOUS ALGAE IN A SPRING FED RIVER ................................ ................................ ......... 53 Introduction ................................ ................................ ................................ ............. 53


7 Case Study: Flori ................................ ................................ .......... 55 Hypotheses ................................ ................................ ................................ ...... 56 Methods ................................ ................................ ................................ .................. 57 Study Site and Biota ................................ ................................ ......................... 57 Gastropod Surveys ................................ ................................ ........................... 58 Experimental Design and Field Sampling ................................ ......................... 58 Labora tory Processing ................................ ................................ ...................... 61 Data Analyses ................................ ................................ ................................ .. 62 Results ................................ ................................ ................................ .................... 63 Spatial Patterns of Gastropod Grazers ................................ ............................. 63 Experiment 1: Low Initial Algae Conditions ................................ ...................... 64 Experiment 2: High Initial Algae Conditions ................................ ...................... 65 Discussion ................................ ................................ ................................ .............. 66 Grazer Impacts in Low Initial Algae Conditions ................................ ................ 66 Environmental Media tion of Grazer Impact ................................ ...................... 68 Grazer Impacts in Algae Bloom Conditions ................................ ...................... 69 4 DISSOLVED OXYGEN IMPACTS ON ALGAL GRAZING BY ELIMIA FLORIDE NSIS IN STREAM MESOCOSMS ................................ ........................... 83 Introduction ................................ ................................ ................................ ............. 83 Hypoxic Stress ................................ ................................ ................................ 83 Flori ................................ ................................ ............................... 85 Methods ................................ ................................ ................................ .................. 87 Stream Mesocosms ................................ ................................ .......................... 87 Grazing Impacts ................................ ................................ ............................... 88 Gastropod Behavior and Survival ................................ ................................ ..... 89 Sample Collection and Processing ................................ ................................ ... 90 Data Analysis ................................ ................................ ................................ ... 90 Results ................................ ................................ ................................ .................... 91 Discussion ................................ ................................ ................................ .............. 93 Impacts of DO on Grazi ng Potential ................................ ................................ 93 Lethal and Sublethal Mechanisms of Hypoxia ................................ .................. 94 Effects of Grazer:Algae Biomass Ratios ................................ ........................... 95 ................................ ............... 96 5 CONCLUSION ................................ ................................ ................................ ...... 106 Grazer Biomass Thresholds and Alternative States ................................ .............. 106 Gastropod Conservation ................................ ................................ ....................... 107 LIST OF REFERENCES ................................ ................................ ............................. 110 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 122


8 LIST OF TABLES Table page 2 1 Categorization of sprin gs study sites: t he 32 springs with sufficient prior data t o be considered for the study were categorized by NO 3 DO and Algae % cover as described in the methods. ................................ ................................ .... 42 2 2 Pearson r correlations for primary variables of interest included in the regres aggregation. ................................ ................................ ................................ ...... 43 2 3 Three general linear regression models for algae biomass, with a suite of potential explanatory variables, N = 74.. ................................ ............................. 44 2 4 Three general linear regression models for gastropod biomass with a suite of potential explanatory variables, N = 77. ................................ .............................. 45 2 5 grazer biomass thresholds.. ................................ ................................ ................ 46 3 1 Factorial ANOVAs for the four response variables for experiment 1 (left col umn) and experiment 2 (right column). ................................ .......................... 74 3 2 Daily accumulation rates of filamentous algae (AFDM g m 2 day 1 ) per experimental treatment (g m 2 wet weight snail biomass) at each site, calculate d as the slope of the regression line of algae AFDM by date,. ............. 75 4 1 ANOVA table of four response variables (DM, %R, RR, and Export) by experimental treatments (DO, Date, Snail presence) for ea ch Biomass Treatment.. ................................ ................................ ................................ ......... 98 4 2 ANOVA comparisons of Percent Rem oval (%R) by biomass treatments, categorized by date and DO ................................ ................................ ............... 99


9 LIST OF FIGURES Figure page 2 1 Photographs of the same location at the Ichetucknee Headspring (A) in 1976 (photo: C. DuToit) and (B) 2010 (photo: D. Liebowitz) ................................ ........ 47 2 2 Bivariate regressions of gastropods and algae at three scales of data aggregation.. ................................ ................................ ................................ ....... 48 2 3 Bivariate regression of gastropods and DO at three scales of data aggregation.. ................................ ................................ ................................ ....... 49 2 4 Probability of gastropod biomass being above the proposed threshold of 20.1 g m 2 (above red line), as a function of DO ................................ ........................ 50 2 5 L ongitudinal behavior of gastropod biomass (A), and algal biomass (B) by longitudinal site for high vs. low DO springs. ................................ ...................... 51 2 6 Histograms of residuals of the regression of algal biomass with gas tropod biomass, fit for unimodal vs. bimodal distributions. ................................ ............. 52 3 1 A) Location of the Ichetucknee springshed B) map of the springshed C) Ichetucknee Springs State Park, with the four experime ntal sites marked and D) photograph of the experimental apparatus. ................................ ................... 76 3 2 Five day dissolved oxygen profiles for the four sites with experimental installations, measured between 21 February 20 11 and 30 March 2011. .......... 77 3 3 Gastropod distribution and composition along the length of the Ic hetucknee River on 13 June 2011 ................................ ................................ ...................... 78 3 4 Photograph of t hree flumes in experiment 1, at HS (Headspring) on 3 March 2011. ................................ ................................ ................................ .................. 79 3 5 Experiment 1 (left column, low initial algae) and Experiment 2 (right column, high initial algae) mean s and standard error bars for AFDM, percent removal (%R), and removal rate (RR) ................................ ................................ .............. 80 3 6 Plot of algal AFDM by gastropod wet weight for Experiment 1, categorized by location (HS (Headspring), GF (Grassy Flats), MP (Mill Pond), and ST (South Takeout)). ................................ ................................ ................................ ........... 81 3 7 Algae AFDM g m 2 accumulation over time, plotted by collection date and categorized by snail treatment (g m 2 wet weight snails) ................................ .... 82 4 1 Model of the artificial stream mesocosm and experimental design. .................. 100


10 4 2 Photo of flumes for A) three high DO treatment rep licates, and B) three low DO treatment replicates on 15 May 2011. ................................ ........................ 101 4 3 Mean algae dry mass ( standard err or) categorized by DO treatment and presence or absence of snails for A) BT 1, B) BT 2, and C) BT 3 .................... 102 4 4 Mean algae dry mass ( standard error) categorized by DO and presence (333 g m 2 ) or absence of snails, for biomass export collected downstream of BT 1 on the final coll ection day, 21 June 2012. ................................ ................ 103 4 5 Mean Percent Removal and Removal Rate (%R standard error), separated by Biomass Treatment (BT1, BT 2, BT 3) and categorized by collection date (15 June 2012 and 21 June 12) and DO level ................................ .................. 104 4 6 Means standard error for behavioral data averaged by day of the experiment (13 21 June): % Grazing, % Breaking the surface, % Inert, and % Living. ................................ ................................ ................................ ............... 105


11 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 ENVIRONMENTALLY MEDI ATED CONSUMER CON TROL OF ALGAE PROLIFERATION IN FLO RIDA SPRINGS By Dina M. Liebowitz May 2013 Chair: Matthew J. Cohen Major: Interdisciplinary Ecology Herbivores control autotroph biomass across many aquatic ecosystems therefore the loss of grazing pressure can induc e dramatic changes in autotroph abundanc e and composition. S ymptoms of consumer loss in aquatic systems (e.g., nuisance algae accumulation) may be similar to those of the alleviation of nutrient limitation, necessitating di scrimination between these mechan isms. W e test ed the hypothesis that the loss of top down control by aquatic gastropods explains benthic filamentous alga e proliferation spring fed rivers. We further hypothesized that the dominant grazer control is dissolved oxygen (DO) concentration which varies substantially among, and longitudinally within springs. Finally, we hypothesized that alga e domination may persist if algae biomass exceeds a critical herbivore escape level beyond which grazers can no longer cons train algae accumulation. We used three scales of investigation to test th ese hypotheses: state wide observational surveys, in situ grazer experiments in Ichetucknee Springs and artificial stream mesocosms with controlled dissolved oxygen levels. We obser ved a strong and temporally


12 consistent negative association between algae and gastropod biomass in the surveys ( gastropods predicting up to 40% of the algal variation) and gastropods were by far the most important explanatory variable in multivariate prediction mode ls as well DO concentrations were significantly but only modestly positively associated with gastropod density in field surveys but hypoxic conditions strongly reduced gastropod survival and grazing rates in controlled experiments. The in situ grazing experiments showed that high but ecologically relevant, densities of grazers (>100 g m 2 wet weight Elimia spp .) could inhibit a lgal bloom formation, but even > 330 g m 2 of Elimia spp could no t reduc e high algal biomass consistently, suggesting that algal biomass may escape herbivore control and entrain spring sys tems in algal dominated states. Recognition that the fauna are not merely passive recipients of the system effects but direct d r ivers themselves may suggest new approaches to restoration and help focus future research.


13 CHAPTER 1 INTRODUCTION Top Down Vers us Bottom Up Control of A lga e P roliferation Aquatic ecosystems are changing on a global scale (Diaz 2001, Steffen and Tyson 2001, Lotze et al. 2006) ; o ne of the symptoms of t hese changes is the proliferation of nuisance algal blooms which can have impacts ranging from aesthetic changes to disruption of ecological function and large scale d egradation of aquatic resources (Estes et al. 2011, Barnosky et al. 2012) A long r unning debate has q uestioned the causes of these changes in primary producer biomass and composition, with early ecological enquiry asking whether factors control pr imary producers (Hairston et al. 1960, P ower 1992, Hillebrand 2002) Models aimed at understanding the controls on primary producer diversity and productivity became progressively more sophisticated in the attempt to explain highly variable experimental outcomes (Leibold et al. 1997) incorporating the concurrent and interactive effects of both top down and bottom up factors, with a multitude of potential mediating factors such as plant edibility and species compositional change (Leibold et al. 1997) ecosystem type and producer evenness (Hillebrand et al. 2007) and positive feedbacks leading to hysteretic interactions and alternative stable states (Dent et al. 2002, Scheffer et al. 2008) Despite the complexity and context dependence of the controls on the ecosystems aquatic ecosystem management efforts have generally focused on nutrient control as the most tractable driver amenable to inter vention and remediation However, emerging evidence suggests that system declines often attributed to nutrient enrichment may instead primarily be responses to loss of consumer control and trophic cascades


14 (Heck and Valentine 2007, Gruner et al. 2008, Baum and Worm 2009, Estes et al. 2011, Poore et al. 2012) Once consumer control is lost and algal blooms proliferate, t hese ecosystem changes can be maintained as a lternative stab le states as a consequence of biomass amasses to an extent that it escapes herbivore control through loss of palatability (Scheffer et al. 2008). Therefore, the evaluation of th e role of grazers in ecosystems of interest, as well as the factors controlling those grazer populations, may be an important step in successfully managing and restoring ecosystems Controls of A quatic Grazer P opulations There are a large number of anthro pogenic changes that can lead to the decline of aquatic fauna, including habitat destruction through dams, dredging, or severe flow reductions W ater quality changes are equally as important, particularly factors such as increasing temperature heavy silta tion, n utrient enrichment, pesticides pharmaceuticals, and a host of emerging contaminants. In relatively unmodified ecosystems, such as those located on protected lands, low dissolved oxygen (DO) concentration can be one of the strong con trols on aquatic faunal health. The highly visible effects of m arine and estuarine hypoxic (low DO) or anoxic ( no DO) waters, so (Rabalais et al. 2002) have prompted significant investigation, leading to an early definition of hypoxia at 2.9 mg O 2 L 1 (Diaz and Rosenberg 1995) later refined to hypoxia at < 2.0 mg O 2 L 1 and severe hypoxia at < 0.5 mg O 2 L 1 (Diaz 2001) However, those levels may underestimate the DO concentrations necessary for healthy ecosystem s due to subtle sublethal effects (Vaquer Sunyer and Duarte 20 08) Fresh water systems encompass similar ranges (Malard and Hervant 1999) but have received less systematic atte ntion so far


15 Exploring the Role of C onsumers in prings North ern Florida has one of the highest densit ies of large freshwater springs in the world, with more than 700 named springs that provid e both aesthetic and economic benefits to the S tate. Seminal work in aquatic ecology was made possible by the short term chemostatic and hydrostatic properties of these systems (Odum 1957b) but recently there have been dramatic increases in nitrate leve ls in many spring s. N uisance algal blooms have developed and submerged aquatic vegetation (SAV) has declined in many of these springs as well leading to detrimental impacts on aquatic habitat s hu man health and ecosystem aesthetics. N utrient increases have been implicated a s the cause of nuisance algal blooms and manage ment attention has overwhelmingly focused on nutrient tracking and control (Stevenson e t al. 2004, 2007, Brown et al. 2008b) Several lines of evidence, h owever, suggest that nitrate enrichment may not be the primary driver of algal proliferation in th e s e system s and that alternative mechanisms such as consumer effects should be considered (Heffernan et al. 2010 b ). The following chapters use three scales of inquiry to test t hree main hypothese s : 1) top the loss of this c onsumer control helps explain the observed benthi c filamentous algal proliferation ; 2) hypoxia controls the presence and grazing rate of dominant invertebrate grazer in the springs, Elimia spp. ; and 3) the algal dominated state can become persistent if grazer populations remain low or the algae rea ch an herbivore escape density (a large biomass of algae that grazers can no longer diminish) We conducted observational surveys across the state to detect an ecosystem level signal of the grazer effect on algae, the potential for alternative states ( the persistence of high or low algae states ), and environmental controls o n grazers. We then used in situ experiments to refine our


16 understanding of grazer impacts by demonstrating causal effects of gastropods on algae, and enumera ting grazer thresholds and herbivore escape densities leading to hysteretic processes. Lastly, we used artificial stream mesocosms to control DO levels and test the effects of varying degrees of hypoxia on grazer survival and grazing rates at multiple grazer and algal biomass levels.


17 CHAPTER 2 ECOSYSTEM LEVEL PATTERNS OF CONSUMER CONTROL OF ALGAL P Introduction Broad scale shifts in the composition and abundance of aquatic plant assemblages, in particular the proliferation of nuisance algae, have focused attention on causal mechanisms of aquatic ecosystem degradatio n A large and well studied suite of potential drivers of algal blooms ha s been invoked, including increased nutrient delivery and changes in stoichiometry (Dodds 2006, Hillebrand et al. 2007, Elser et al. 2007) reductions in consumer biomass (Gruner et al. 2008, Hillebrand 2009, Poore et al. 2012) changes in light availability (Odum 1956, Hillebrand 2005) and changes in water velocity and hydrology (Hoyer et al., 2004 Riseng et al., 2004 ) along with myriad interactions. Nutrients are often considered the most influential drivers of algal blooms and thus are the focus of management interventions and there are a plethora of well documented instances of nutrient en richment leading to algal dominance particul arly in coas tal and estuarine regions, lakes, and large rivers (Carpenter et al. 1995, Jrgensen and Richardson 1996, Worm and Lotze 2006, Conley et al. 2009) However, e vidence suggests that herbivore control, specifically consumer removal of primary producer biomass through consumption or mechanical disturbance, can be an equally strong force in structuring ecosystems (Hillebrand 200 2, Heck and Valentine 2007, Gruner et al. 2008, Baum and Worm 2009, Estes et al. 2011) removing an average of 59% of algal standing stock across aquatic ecosystems (Hillebrand 2009) A decline in graze r populations can lead to algal dominated systems which can persist if grazer populations remain low or if algae reach a critical herbivore escape density Scheffer et al. (2008) modeled the equilibria of primary producer dominated


18 vers u s herbivore dominated states, finding that both press and pulse declines in the herbivore population can induce a switch to a stable algal dominated state, if high algal biomass or older growth forms are sufficiently resistant to grazing. Grazers have a par ticularly strong influence on early algal successional stages and reduce colonization potential for many filamentous forms, with much weaker effects on adult algal forms (Alstyne et al. 1999 ; Korpinen et al. 2008). Once algae accumulate beyond an herbivore escape density, these positive feedback mechanisms allow the persistence of unpalatable algal growth forms or species; systems c an thus stabilize into an algal dominated state, insensitive to even intense grazing pressure (Gliwicz 1990, van de koppel et al. 1996, Gragnani et al. 1999, Lotze and Worm 2000) Alternatively if algal growth rate is higher than the grazer consumption rate but al gae do not reach an herbivore escape density alga l biomass could be reduced by additional influx of grazers or increases in overall consumption rates (Lamberti et al. 1987, Suding and Hobbs 2009) C a uses of gastropod declines and extinctions worldwide include hydrological modification, habitat loss, declining water quality and quantity, invasive species, and their many potential interactions (B rown et al. 2008a, Lysne et al. 2008) Low dissolved oxygen (DO), in particular, can directly cause mortality through su ffocation or indirectly through slowing of feeding rates and resulting starvation reduced fecundity and behavior al changes that incr ease predation vulnerability (Diaz 2001, Wu 2002, Cheung et al. 2008) Grazers, O xygen, and A lgae in prings North ern Florida has the highest densit y of large freshwater springs in the world, with more tha n 700 named springs that provid e both aesthetic and economic value


19 (Bonn and Bell 2003, Scott et al. 2004) N uisance algal blooms have developed and submerged vascular plants (or submerged aquatic vegetation SAV) have declined in many of these springs over the last few decades leading to detrim ental impacts on habitat, human health and aesthetics (Fig. 2 1 ). Increases in nitrate (NO 3 ) concentration have been implicated as the cause of nuisance algal blooms and manage ment attention has overwhelmingly focused on nutrient reduction and related re mediation (Stevenson et al. 2004 Brown et al 2008) However, several lines of evidence suggest that nitrate enrichment may not be the primary driver of algal proliferation in th e s e system s and that alternative mechanisms, including loss of top down control, should be considered (Heffernan et al. 2010). Given the well established effects of grazers on primary producers in streams in the southeastern US (Mulholland et al. 1991, Hill 1992, Rosemond et al. 2000) explicit consideration of the potential for top springs is critically needed. Studies in east coast temperate streams found gastropods from the family Pleuroceridae (mostly of the genera Juga and Elimia ) to be particularly important consumers, with natural populations reaching 1 000 individuals m 2 and accounting for up to 95% of invertebrate biomass (Newbold et al. 1983, Rosemond et al. 2000, Stewart and Garcia 2002) and eating a wide array of algae from diatoms to green algae and phaeophytes (Hill, Boston & Steinman 1992). Although population den sities of gastropods are poorly chara pleurocerids can (Walsh 2001) documented up to 579 g m 2 ( Dutoit 1979), suggesting a potentially strong role of these herbivores in controlling algal biomass. The potential for c onsumer regulation has only recently been explo


20 (Dormsjo 2008), and there have been no broad scale investigati ons of top down control s Despite evidence of DO declines and increasing incidence of hypoxia (DO < 2 mg L 1 ; Diaz 2001) in springs (Heffernan et al. 2010), the effect s of DO on freshwater gastropods are poorly understood. Previous laboratory studies found slight association s between oxygen stress and metabolic patterns in the lab and weak relationships between DO and gastropod distributions in the field (Berg and Ockelmann 1959, Hanley and Ultsch 1999) More recently, Dormsjo (2008) and Liebowitz (Chapter 3 ) found that Elimia spp. suffered high mortality and potential sub l ethal effects under hypoxic co nd itions in Ichetucknee springs as well as in laboratory streams (Liebowitz, Chapter 4) Pleurocerids are heavy shelled, particularly slow snails with low dispersal rates (Brown et al. 2008a) making them slow to repopulate which suggests potentially persistent effects of even pulse h ypoxic events that cause mortality. Hypotheses W e sought to test the hypothesis that gastropod grazers control algal accumulation algal blooms. We further hypothesized that dissolved oxygen influences grazer abundance, and that feedbacks between algal accumulation and decreased grazing ability create the po tenti al for bifurcation in to high or low algal states Th ese hypotheses led to three linked predictions : (1) gastropods and algae biomass will exhibit a negative association in the field surveys ; (2) DO c oncentrations will be positively associated with gastropod densitie s; and (3) that below a certain gastropod biomass threshold, residuals of algal biomass will be uniformly high, while above that gastropod threshold,


21 the algal r esiduals will be bimodal ly distributed (either high or low) consistent with positive feedbacks that create alternative states. Methods Study Site Selection We selected eight spring runs from the more than 700 named springs in Fl orida to represent gradients in environmental variables of interest and to satisfy t he following several criteria. First and foremost, we used existing data on biological, chemical and physical attributes (Scott et al. 2004, Stevenson et al. 2004) to group springs based on three binary classifications: (1) high vs low nitrate concentration (breakpoint at 0.35 mg L 1 which corresponds to the recently adopted nu meric nutrient criteri on; Obreza et al. 2011) ; (2) high vs low vent DO concentration ( using the widely cited threshold of hypoxia at 2 mg L 1 ; Diaz 2001) ; and (3) high vs low alga l abundance ( breakpoint at 50% cover using Stevenson et al. ( 2004 ) for most springs, and qualitative high vs low using Scott et al. (2004) if quantitative data were lacking) This created eight groupings, from which o ne spring per group was selected (Table 2 1 ) based on discharge and spring length criteria Mean discharge of each spring was between 0.3 and 2.8 m 3 s 1 (2 nd magnitude springs; Meinzer 1928) Additionally, springs required a spring run (the lotic ecosystem downstream of t he vent) at least 200 m long before confluence with another water body, and with no tidal or salinity influence. This criterion was applied to permit investigation of within system longitudinal DO variation while minimizing the number of confounding facto rs In addition to the eight springs selected, we also obtained measurements in t hree additional springs from the Ichetucknee S prings complex to further populate the sample of low DO s ystems. We note that this study design necessarily preclude d considerati on of larger, iconic 1 st magnitude springs (e.g.


22 Silver Springs, Rainbow Springs). Additionally, we treat the results pertaining to nitrate and DO associations with algae with caution, as the sampling design specifically crossed those variables with algal abundance therefore bivariate associations are not random. Field Sampling Design A hierarchical sampling design was employed to document variation among springs as well as wi thin springs along a longitudinal gradient (e.g., downstream changes in DO or n itrate) Observational surveys were conducted three times during 2009 in each of the eleven springs to examine temporal effects At each spring, we selected three sites spanning from the spring vent (Site 1) to locations 200 m or more downstream. Downstre am sites in high DO springs were 100 m from the vent (Site 2) and 200 m from the vent (Site 3). In low DO springs, we varied the location of Site 3 to ensure that DO levels at all Site 3 locations were above 2 mg L 1 at the trough of diel DO variation, mea sured using a YSI 6920 multi parameter sonde (Yellow Springs Instruments, Yellow Springs, OH) deployed for hourly measurements over an extended period (4 6 days) prior to sampling. Field Sampling Protocols At each site, we installed two line transects fro m bank to bank ( 10 m apart from each other along the bank and perpendicular to spring flow ) with three sampling locations evenly spaced along each transect. At each of these sample locations, w e estimated benthic cover of filamentous algae, diatoms, and s ubmerged aquatic vegetation (SAV) using a 5 step Braun Blanquette scale (Braun Blanquet 1932) (binned at 1=0 5%, 2=6 25%, 3=26 50%, 4=51 75%, 5=76 100%). At e ach sampling location, we also collected biomass and estimated c anopy cover using a densitometer, surface


23 flow velocity using the float method, and DO, water temperature, specific conduct ance and pH using a YSI 6920 multi parameter sonde At each site we also collected a 500 mL water sample into an acid washed polyethylene bottle that was acidified to pH 2.0 using hydrochloric acid, and stored below 4C until analysis. Water samples were filtered using a 0.45 m glass fiber filter and analyze d within 28 days for nitrate (cadmium reduction method; EPA 353.2) and soluble reactive phosphorus (ascorbic acid method; EPA 365.1). We sampled vegetation and large invertebrates us ing a Hess type invertebrate sampler with a circular footprint of 0.086 m 2 modified with a mesh cap to keep samples from escaping in deeper water and with a larger mesh size (ca. 1 m m) that facilitated sampling in flocculent and diverse substrates Because p revious studies reported, and site reconnaissance confirmed, that gas tropods are often dominant grazers we used a method well suited to gastropod assessment across a variety of substrates but less accurate for smaller invertebrates. The sampler was placed on the substrate and pushed down 1 2 cm to create a bottom seal. A t deep water sites where the sampler had to be completely submerged (> 0.75 m), a mesh cover with a single access slit was attached to the top of the sampling device to ensure none of the sample was lost. Vegetation was clipped at the roots and guided into the mesh bag attached to the downstream side of the sampler. The sediment was then agitated to collect remain ing invertebrates and algae Sampling was considered complete when three consecutive sweeps with a hand net yielded no additional biota Samples w ere t ransferred into labeled containe rs, and placed on ice for transport to the lab.


24 Biomass Sample P rocessing Samples were separated into four biomass categories and rinsed with DI water to remove sediment and foreign materials. Plant species were catego rized into coarse groupings of: 1) submerged aquatic vegetation (SAV), primarily S a gi t taria kurziana and Vallisne ria americana ; 2) filamentous algae; 3) aquatic plants of other species (e.g., Ceratophyl lum sp. Hydrocotyl e sp. Chara sp. ); and 4) invasive species ( Hydrilla verticillata and Elodea densa ). Each biomass component was blotted dry, weighed for wet mass, oven dried at 70 C for 24 hours and reweighed for dry mass. Aquatic vegetation other than Sagittaria and Vallisneria was not included in subs equent statistical analyses, as it accounted for a trivial fraction of total biomass. Invertebrates were sorted from the samples by hand, taking 1 6 hour processing time per sample depending on sample biomass and complexity. Invertebrates were separated by major taxa (family level for gastropods, order level for all others), counted, blotted dry, weighed dried at 70 C for 24 hours and then reweighed for dry mass. Because of sampling and sorting constraints, i nsect and decapod abundances and biomass measur ements were considered informative, but incomplete representations of prevailing conditions, particularly for smaller taxa (e.g., chironomids). Data A nalysis W et and dry weight were strongly correlated for all biomass categories (SAV: y = 0.12 + 0.08*x, r = 0.97, p < 0.001 ; Algae: y = 2.90 + 0.33*x, r = 0.95, p < 0.001 ; Gastropods: y = 0.11 + 0.64*x, r = 0.99, p < 0.001). Because each metric of algal presence has biases, the correlation between filamentous algal biomass and estimated % cover was evaluated (y = 3.074 + 1.869*x; r = 0.844, p < 0.0001) The f ive samples (out of 480) that showed a large discrepancy between % cover and biomass were


25 omitted from further analysis Additionally, cover and biomass showed discrepancies at S ite 1 in Mill Pond as a r esult of the dominant algal growth form, whose structural disaggregation during sampling rendered mass recovery extremely low A lga l data from this site were excluded from all statistical analyses unless otherwise noted We examined the data in a hierarchi cal manner. At the highest level, data were averaged for each spring, and segregated by sampling period; analysis of these data provided the lowest statistical power and therefore the most conservative assessment of the inferred relationships. Data were an alyzed subsequently at the site level (i.e., three sites per spring), which is likely more appropriate, because the distance between sites is sufficient to preclude gastropod traversal (Hu ryn et al. 1997) Additionally, because key environmental variables (e.g., NO 3 SRP, light, water velocity) showed non systematic but significant differences among sites within springs, we considered the site level independent for these analyses. Bivaria te analy ses were examined at multiple data aggregation. Finally, we assessed bivariate relationships between algae, gastropods, and DO at the individual sample level, but did not analyze these data using multivariate approaches. Biomass of algae and gastropods, as well as environmental variables, exhibited skew and were natural log transformed prior to statistical analysis. All regressions were initially run for each sampling period separately, yielding nearly identical results. D ata were also checked for temporal differences using the Kruskal Wall is test with no significant effect of time of sampling period 2 = 0.39, p = 0.82) or gastropod biomass ( 2 = 0.64, p = 0.73) so data were pooled across sampling periods.


26 We use d bivariate and multivariate general regression models t o test relationships between environmental metrics, gastropod biom ass, and algal biomass ; all analyses were performed i n STATISTICA v10 (Statsoft, Inc.). To test prediction 1 (a negative relationship between algae and grazers) we first modeled the pairwise association between algal and gastropod biomass at each level of data aggregation. In a second model, we used multiple regression to assess the conditional effects of grazers and other drivers (i.e., nutrients, water velocity and light); a third model included the entire suite of environmental variables. The corrected A kaike Information Criteria (AICc = 2k + n(Ln(RSS/n)) + (2k(k+1))/(n k 1) where k is the number of parameters and n is the number of samples ) was used to compare models We also explored the role of grazer composition on algal abu ndance by analyzing the e ffect of fractional biomass of the four primary snail families (Pleuroceridae [ Elimia spp .], Hydrobiidae, Planorbidae, and Viviparidae ) on algal biomass. To test prediction 2 (a positive relationship between DO and grazers) we used the same basic approach as for prediction 1 We first evaluated the bivariate association between DO and grazers, and then considered the suite of environmental controls by building multivariate models at two levels of complexity, the first using the most likely controls on graz ers, and the second including all measured environmental variables To further explore whether high DO springs (> 2.0 mg L 1 DO at the vent) exhibited different longitudinal behavior than low DO springs (< 2.0 mg L 1 DO at the vent) we compared patter n s o f algal and gastropod biomass with distance downstream for springs grouped by DO concentration


27 To test prediction 3 (bi modality of algal biomass above a grazer density threshold, suggesting positive feedbacks that create alternative ecosystem states) w e examined the residuals of the relationship between gastropod and algal biomass Because bimodality was expected to be manifest at or above a critical gastropod density, w e evaluated bimodality above and below five gastropod biomass thresholds : 12, 16, 20, 26, and 33 g m 2 gastropod dry weight ; this analysis was performed using natural log transformed data, resulting in residual values that are likewise transformed All bi modality analyses were performed in R v.2.15.0 ( R Foundation for Statistical Computi ng ). was employed to test the null hypothesis that residual distribution s above and below each threshold are unimodal, with the alternative hypothesis of at least bimodality (Hartigan and Hartigan 1985) Using the mclust package in R, we compared the fit of the residuals to both a normal distribution ( N is a normal distribution wi th mean and standard deviation ), and a bimodal distribution given by P= q N(x, 1 1 ) + (1 q) N(x, 2 2 ), where q is a mixing coefficient with values from 0 to 1 that describes the relative contribution to the density function of mode 1 and x is the residual algal biomass not explained by gastropods. We selected between unimodal and bimodal distributions using both the Bayesian Information Criteria ( BIC) and the diptest to present the most consistent threshold. Results Biomass Composition and P redic tors of Distribution P atterns Twenty faunal taxa including 9 mollusks, 7 insects and 4 decapods, were i dentified across the samples from the springs. T he dominant taxa in terms of biomass were the gastropods account ing for 60 100% of the biomas s in each spring. Algal biomass averaged 153 g m 2 (SD = 426, max = 3,419); SAV biomass averaged 49 g m 2


28 (SD = 86 max = 580), gastropod biomass averaged 67 g m 2 (SD = 107 max = 589), decapod biomass averaged 0.6 g m 2 (SD = 1.8 max = 18.3), and insect biomass averaged 0.04 g m 2 (SD = 0.13 max = 1.4). The large mesh size needed to sample flocculent s ubstrates led to low capture efficiency of insect and small decapo d biomass, therefore they are not included in subsequent analyses as a consequence of likely unde rrepresentation in samples. We note, however, that for decapods, we found no significant relationship with algal biomass (p = 0.12), and a significant but weak positive relationship with DO (p = 0.001 r 2 = 0.13); similarly, insect biomass was uncorrelat ed with algal biomass (p = 0.54), and was weakly, positively associated with DO (p = 0.013, r 2 = 0.07). Algae D istribution s Gastropod biomass was a strong predictor of algal biomass in the bivariate correlations at all three data aggregation levels (Fig. 2 2 A C). In the most conservative and algae association was significant and consistent across all sampling periods (period 1, n=10, p = 0.01, r = 0.79, slope = 2.3; pe riod 2, n = 11, p = 0.003, r = 0.83, slope = 2.9; period 3, n = 11, p = 0.01, r = 0.72, slope = 2.4). This relationship remained strong and consistent when the data were pooled across seasons (n = 32, p < 0.0001, r = 0.76). At the most ecologically re same held true: the negative algae and grazer relationship was consistent across sample periods, as well as for pooling all samples (combined data: n = 88, p < 0.00 01, r = 0.61, slope = 1.29). T he sig n and magnitude of the gastropod effect indicated that the gastropods were the strongest and most consistent predictor of algal biomass, but other bivariate relationships w ere also significant (Table 2 2). At the individual sample


29 level, we still observed strongly significant relationships, but lower explanatory power for individual and pooled sampling periods (combined data: n = 451, p<0.0001, r = 0.35, slope = 0.56). Among the multivariate algal biomass prediction models of varying complexity (Table 2 3 ), model 2 includes only those factors most often associated with algal blooms (nutrients, canopy, water velocity, and consumers), and explains ~45% of the variation in algal biomass. However, only gastropods (negative relationship) and light (positive rel ationship) are statist ically significant predictors. Inclusion of the full suite of measured environmental factors improved overall explanatory power (adj. R 2 = 0.53 ), but not AIC c suggesting that model 3 may be overfit, with additional variables not addin g sufficient additional predictive power. Despite the improved overall fit and explanatory power, only one variable (gastropod biomass) in model 3 was a statistically significant predictor. The fraction of biomass as Elimia spp was negatively correlated with algal biomass (r = 0.458, p < 0.0001), whereas % h ydrobiids (r = 0.348, p = 0.001) and % p lanorbids (r = 0.278, p = 0.009) were positively correlated with algal biomass. However, % Elimia spp. was also positively correlated with overall gastropod bio mass (r = 0.401, p = 0.0001), while h ydrobiids and p lanorbids were both negatively correlated. Gastropod populations dominated by Elimia spp. had much higher biomass overall (up to 575 g m 2 ) than those dominated by other species ( p lanorbids up to 113 g m 2 h ydrobiids up to 106 g m 2 ). Gastropod D istributions Dissolved oxygen concentration was a significant predictor of gastropod biomass across two of the three data aggregation levels. A t the highest


30 aggregation, where sample size limit s power and an y longitudinal changes in DO are lost to averaging the DO gastropod relationship was marginally significant and weakly positive, (p = 0.08, r = 0.31 ; Fig. 2 3A) was highly significant (p < 0.001, r = 0.42), but heavily influenced by three points from one spring site (i.e. MP Site 1) the exclusion of which reduced the strength of the DO grazer association (Fig. 2 3B) gastropod regression w as significant (p = 0.0003, r = 0.32; Fig. 2 3C). Only two springs, both low DO sites, had significant within spring relationships between D O and gastropod biomass (MP, p = 0.01; r = 0. 91; RS, p = 0.0006; r = 0.91). Notably, Williford springs had DO levels of 0.8 mg L 1 at the vent, yet as much as 27 g m 2 of ga stropods. I nclusion of additional predictive factors in Model 2 (Table 2 4) indicated that all but temperature exerted a significant univariate effect, and together explain ed 50% of gastropod variati on. Model 3 included the full suite of environmental variables, with all but temperature and SRP exerting a significant effect. DO, water velocity, and light were all positive predictors, wh ereas NO 3 conductivity, pH, and SAV were negative predictors, to gether explaining over 60% of gastropod variation. However, Model 2 showed a slightly more favorable AICc score due to its parsimony, suggesting that Model 3 may be overfit. Longitudinal P atterns L ongitudinal gastropod biomass changes were different acro ss spring s Over 70% of low DO springs ( 5 of 7 ) showed significant (p < 0.05) positive associations between gastropod biomass and distance downstream; one more was marginally significant (p = 0.09) and the last had no significant relationship. Gastropods i ncreased


31 significantly downstream in only one o f the four high DO springs ; gastropods also significantly decreased in one and showed no relationships in the other two Longitudinal gradients of gastropod and algal biomass differed between high and low D O springs (Fig. 2 5). In low DO springs, g astropod biomass increased from upstream to downstream; gastropod densities decreased downstream in high DO springs, though they had a hig her overall population than low DO springs throughout (F (1,85) = 9.1, p = 0. 003) Longitudinal patterns of algal biomass were slightly different, decreasing with downstrea m distance in both high and low DO springs. T his trend was weaker in the low DO springs than in high DO springs, and low DO springs had hi gher overall algae biom ass (F(1,82) = 7.3, p = 0.008). Bimodality were unimodal below all potential threshold levels but that above the se threshold s unimodality could be rejected for all e xcept at a threshold of 26 g m 2 (T able 2 5). BIC model comparisons (i.e., bi modal vs. unimodal model fits ) were generally in agreement with this out put. H owever, as the gastropod threshold increase d beyond 26 g m 2 distributions become more ambiguous. Above a threshold of 26 g m 2 good ness of fit bec ame more equivocal (i.e., less than three units separate models) With a threshold of 33 g m 2 residuals both above and below the threshold were better described by bi modal distributions. Because ambiguity in the data distributions first appeared at 26 g m 2 we selected 20 g m 2 as the most defensible threshold for which model residuals are presented (Fig. 2 6). The probability of gastropod biomass exceeding this 20 g m 2 threshold varies dramatically with DO conditions (Fig. 2 4); 10% of observations with


32 DO below 1 mg L 1 exceeded critical snail densities, whereas more than 70% of observations exceeded this snail biomass threshold at sites with DO > 5 mg L 1 Discussion Gastropod Control of Filamentous A lgae B looms T his study lends stro ng suppo rt for the claim that consumers play an important role in control ling algal proliferation We found that g astropod biomass exhibited a strong negative relationship with algal biomass across all levels of data aggregation in thi s study, including at the entire spring level for which our sample size was small. This is consistent with recent meta analyses of grazer experiments illustrating the paramount role of consumers in controlling primary producers across most aquatic ecosyst em s (Hillebrand 2002, 2009, Hughes et al. 2004, Gruner et al. 2008, Poore et al. 2012) Observational surveys such as this study and others (Riseng et al. 2004) compl e ment and contextualize the experimental manipulations that dominate the assembled evidence for top down controls. While experiment al manipulations provide stronger inference about causality and constitute the next step in scale observational survey demonstrate s the plausibility of consumer, and specifically gastropod, control on algal biomass in the context of complex ecosystems. Negative correlations between Elimia spp biomass and algal abundance suggest that this group plays a particularly important role in top down control in Florida Springs. Gastropod populations domi nated by o ther major gastropod families (Hydrobiidae, Planorbidae, and Viviparidae) were in contrast, weakly but positively associated with algal biomass. Elimia spp are generalist grazers with a demonstrated ability to shift algal composition and select ively remove filamentous cyanobacteria, keeping these


33 species from accumulating (Tuchman and Stevenson 1991, Feminella and Hawkins 1995) They have been documented at exceedingly high biomass levels dominating southeastern st reams (Hill 1992) and have been shown to control algal accumulation in natural stream systems (Brown et al. 2008a) However, this association may also reflect covariation between species composition and density (rather than greater grazing efficacy of Elimia at the individual level) as Elimia dominated populations reach five fold higher biomass levels than any other species in thes e systems Although gastropods consistently exhibited the strongest bivariate relationship with algae, a number of additional variables (phosphate (SRP) conductivity, temperature, and submerged aquatic vegetation ( SAV ), all positive) had significant biv ariate correlations. Moreover, multivariate models including the full suite of environmental factors explained 15% more algal variability than gastropods alone. However, in those more complex multivariate models, all variables except for gastropods were n o longer significant, suggesting complex interactions among variables that cannot be discerned with the low power of this particular study design. Notably, NO 3 the variable most widely cited as the cause of algal proliferation, showed no association wi th algal biomass in this study We caution however that the expectation of a relationship may be confounded by the study design which intentionally selected springs with high and low alga l cover at high and low nitrate levels. down versus bottom acknowledge the complex and nuanced interactions and contingencies that mediate the strengths of multiple drivers. Although studies suggest that nutrients generally increase primary producer biomass i n the absence of sufficient herbivory (Gruner et al. 2008) a


34 host of physical, chemical, physiological, and ecological factors can mediate this patte rn (Borer et al. 2005, Gruner et al. 2008, Poore et al. 2012) T his study addressed only some of these potential mediating factors, yet the fact that grazer biomass was consistently the best predictor of algal biomass while all other variable showed weaker associations, strongly supports the conclusion that grazers exert a dominant effect on algal proliferation in these systems. Dissol ved Oxygen and Additional Controls on G astropods Our results support the hypoth esis that dissolved oxygen influences gastrop od biomass, but also that other stressors need to be considered to understand grazer declines At most levels of data aggregation there was a significant positive relationship between DO and gastropod biomass, but DO only explained 18% of gastropod variat expected positive direction, but was not statistically significant. Whereas hypoxic sites generally had lo w gastropod density t here were exceptions to this pattern. The Williford S pring s ite for example, had DO levels of 0.8 mg L 1 at the vent, but also had relatively high gastropod biomass (27 g m 2 ) at that location. Discerning critical levels of dissolved oxygen for aquatic organism he alth has proven to be a vexing problem. Both chronic hypoxia and pulses of severe hypoxia or anoxia affect gastropod populations, potentially leading to apparent low correspondence between concurrent DO and gastropod l evels for instance, a brief low DO pulse could dec imate a community, but not be in evidence if DO measurements are taken days later. Hypoxia can cause a range of responses, including depressed feeding and reproduction rates as a consequence of shifting metabolic pathways (Kapper and Stickle 1987, Hanley and Ultsch 1999) behavioral modifications that can lead to


35 predator vulnerability, and direct mortality (Sagasti et al. 2001, Wu 2002, Cheung et al. 2008) Gastropods may be able to behaviorally adjust to conditions of hypoxia by taking advantage of the DO concentration gradient at the air water interf ace in shallow waters (Wu 2002) However, multivariate cross system analyses of water depth, DO, and velocity indicated no interactive effects in this study. We observe d that at DO concentrations below 1 mg L 1 g astropod density was uniformly low, while at higher DO concentraions ther e was a wide range of gastropod densities. However, as DO levels can fluctuate through time, as well as vary spatially in microhabitats, our observations may ha ve been insufficient to capture the dynamics of DO effects in space and time (Garvey 2007) Experimental manipulation of dissolved oxygen to discern effects on grazing rates and gastropod mortality is clearly needed. L ongitudinal patterns of DO within springs further illustrate the nuanced effect of DO on gastropods, and the ir commensurate effects on algal density. Low DO springs uniformly showed significant gastropod increases downstream, whereas th e longitudinal patterns in high DO springs were inconsistent. Additionally, low DO springs consistently h ad greater alga l abundance and fewer gastropods at all sites despite gastropod increases downstream Overall, the low gastropod abundances at the vent of low DO springs suggests that gastropods benefit from rearation downstream That they continue to exhib it lower biomass in low DO springs, may suggest both reduced survivorship and lower fecundity from low level physiological stress. One explanation for higher algal biomass at downstream locations in low DO springs may be high rates of longitudinal drift of sloughed algal tissues which subsequently settle downstream; there is strong


36 evidence of this effect in Alexander Springs (Coh en et al. 2012), one of our low DO sites. Whereas our evidence supports the conclusion that DO influences grazer populations in F lorid a springs, the influence of other factors is clearly evident in the data. In the multivariate regressions, most of the environmental variables added significant predictive power, collectively explaining 60% of gastropod variability while DO alone ex plained only 20%. Nitrate concentrations, conductivity, pH, and SAV were all negative ly correlated with gastropod abundances, whereas light, water velocity, and DO were positively correlated These factors generally align with the expected direction of eff ect. The effect of nit rate is of particular interest. Several studies have shown lethal impacts of nitrate on aquatic fauna H owever gastropods generally appear to be less susceptible to t hese effects particularly at the relatively low levels seen in the springs (Camargo et al. 2005, Mattson et al. 2007, Brown et al. 2008b) Conductivity, on the other hand, is generally negatively associated with freshater gastropods (Lodge et al. 1987) with potential implications for coastal springs susceptible to salt water intrusion, and those where connate sea water influences spring wa ter chemistry (e.g., Alexander Springs; Davis et al. 2001 ) The negative relationship with SAV observed in multivariate analyses is not present in the bivariate relationship, and is surprising, as gastropods frequently graze on strapleaf vegetation; however, they do aggregate more densely on complex vegetative forms such as the macro algae Chara spp. ( Liebowitz, pers. obs.), and the highest biomass aggregations were on bare rocky or sandy substrates. Gastropods have been found to be positively associated with light availability (possibly due to positive phototax is), and flow relationships can vary greatly depending on habitat


37 and life stage (Johnson and Brown 1997) Together, these factors suggest that the persistence of low nutrients and conductivity, cooler water, and high flow and DO are characteristics of optimal gastropod habitat. This suite of variables helps set our expectation s of grazer abundances; those variables that are amen able to management may be candidates for restorative action. O ur results suggest that gastropod popu lations respond predictably to environmental drivers, however a large fraction of the variance in snai l biomass remains unexplained. This is particularly salient in light of evidence of gastropod declines in this data versus earlier surveys (DuToit 1979) and obser vations made by local biologists who noted declines in Manatee S prings (R. Mattson, pe rs. comm.), and in Ichetucknee S prings (J. Stevenson, pers. com m.; C. Parenteau, pers. comm.). Although gastropod densities are relatively undocumented in Florida, extrem e declines are seen throughout the southeastern US (Neves et al. 1997, Brown et al. 2008a, Lysne et al. 2008) ; f o r example, 65% o G2 by the Nature Conservancy (Lydeard and Mayden 1995) I t seems likely that the faunal responses to anthropogenic changes in Florida s springs exhibit similar patterns. A wide range of emerging contaminants may be complicit in broadly observed grazer declines, and are potentially important to understanding residual biomass variation in the models. Pesticides and herbicides entering the water, either unintentionally from terrestrial applications or wastewater, or int entionally as part of aggressive aquatic weed management, could negatively impact invertebrate health and remain one of the understudied realms of conservation science (Lawler et al. 2006) In a


38 review of the effects of the management of invasive aquatic flora Brown et al. ( 2008b) outlined a number of herbicides used currently or historically in springs, including copper and diquat, both of which ar e toxic to invertebrates. A recent survey of four springs in the St. Johns River Water Management District found low levels of a variety of pesticides (such as DEET), herbicides, and wastewater compounds all were below levels of concern for DEP; however, r ecent studies have found that pesticides can cause ecological degradation at levels within legal bounds For example, the most com monly used fungicide in the US, Chlorothalonil ( US EPA 2004) has been shown to have strong ne g ative effects on invertebrates (including gastropods ) and and lead to algal blooms at ecologically relevant levels of contamination (McMahon et al. 2012). Additionally, few studies have looked at the effects of cocktails of various compounds, which could p otentially act as multiple sources of stress and therefore have large negative impacts in combination (Phelps et al. 2006). Gastropods, Algae and Hysteretic Ecosystem C hange Feedbacks between organisms and environmental factors can create multiple persis tent ecosystem configurations under t he same environmental drivers. The suite of examples in which such alternative stable states are observed is growing ( e.g., v an de Koppel et al. 2001; Beisner et al. 2003; Schroder et al. 2005; Heffernan 2008) as is recognition of the importance of these behaviors for environmental management (Suding and Hobbs 2009). Our analysis of mode l residuals suggests the presence of a critical gastropod density below which algal biomass is uniformly high and above which two ecosystem configurations might persist In theoretical models of algal escape from herbivore control ( Scheffer et al. 2008 ) this condition arises in part in response to the historical trajectory of the system, depending on whether the system has experienced a


39 stressor that allowed algae to grow beyond herbivore control If Elimia spp. populations decline d because of a lethal pu lse of low DO or some other stressors their slow recolonization rates could release herbivore control long enough to allow algal blooms, even if contemporary gastropod densities are, in other systems, sufficient to keep algal biomass at low levels This escape dens ity may persist long after bloom conditions initiate and gastropod populations recover, due to less palatable growth forms or alter ed species compositions after succession. B i modality is only one line of evidence in support of the presence of alternat ive stable states within spring ecosystems, and this study was only able to consider variation across sites rather than detecting change w ithin a given system over time. However, these results offer a plausible scenario to guide future research The gastrop od threshold level yielding the clearest division between unimodal and bimodal distributions was 20 g m 2 dry weight of gastropods (equivalent to ~ 30 g m 2 wet weight of gastropods) suggesting that systems with l ower gastropod biomass will generally have algal b iomass at nuisance levels but system s above that threshold may or may not, depending on Management I mplications Our results indicate that loss of top down control i s a plaus i ble factor leading to other environmental variables, helps explain patterns of gastr opod abundance and distribution Photosynthetic o xygen production and atmo spheric reaeration are negligible in aquifers, suggesting that DO levels in groundwater are primarily a function of water age and microbial respiration. In short, older water and DOC enriched water (which allows respiratory consumption of DO) generally hav e the lowest DO levels in


40 the Floridan Aquifer (Malard and Hervant 1999, Martin and Dean 2001) Therefore, on a pract ical level, DO may be changed by (1) droughts or anthropogenic use of younger (shallower) aquifer water leading to lower flows and relative dominance of older water, and (2) groundwater organic matter enrichment leading to higher microbial respiration and DO depletion. Given that DO plays a role in grazer abundance, and that both pulse and press hy po xia could harm populations, ensuring adequate flows and reducing organic pollutants merit attention in management plans. Further work to isolate and experiment ally verify the validity of DO as a restoration objective are needed, but our results support both ongoing investigation and management attention. R estoration plans for springs with nuisance filamentous alga e should consider potential alternative states and herbivore escape densities Our conclusion from these surveys that high algae or low algae states exist and can persist is supported by the in situ experiments (Chapter 3) which f ound that low algae biomass can be maintained by > 100 g m 2 of gastropod wet weight, but that even > 330 g m 2 of gastropod wet weight m ay not be enough to decrease an algal bloom Additionally, preliminary evidence suggests that Elimia can inhibit further growth of Lyngbya but not decrease the existing standing crop (Liebowitz, unpublished data) These findings suggest that once a system manner. While further study is n ecessary to explore the critical transitions between states, it is likely that mature algal growth forms would have to be cleared to allow grazers to access early algal successional stages that they are capable of controlling. Alternative stable states and simple time lags are difficult to distinguish without


41 systematic long term monitoring of spring biology, which is a critical gap in the State of Where gastropod populations fall below the 20 g m 2 dry weight (~ 30 g m 2 wet weight) threshold, reintro ducing native gastropods may be a tool for restoration, though the cause of grazer declines would first have to be ameliorated. Sixty percent of the springs in this study currently have gastropod populations above the threshold, indicating that these bioma ss levels are a viable target for most springs. Low DO conditions may require large scale changes in water consumption to mitigate further declines, and DO is also impacted by climate induced variation s in discharge that are beyond local management control Other factors, such as organic matter inflows (e.g., from septic tanks, wastewater sprayfields), are slightly more tractable to management oversight. In cases where low populations are a result of a n historic pulse event from which populations were slow to recolonize, breeding and restocking native grazers to their historic ranges and densities could be an effective intervention (Lysne et al. 2008, Whelan et al. 2012)


42 Table 2 1 Categorization of springs study sites The 32 springs with sufficient prior data to be considered for the study were categorized by NO 3 DO and Algae % number of these 32 springs that fit in each category. The cell to the right of each displays the spring chosen for the survey and its basic chemical composition. Three additional low DO springs were added: Mission (MS), Mill High Algae (> 50% cover) Low Algae (< 50% cover) # Springs Sampled spring # Springs Sampled spring High N High DO 6 Blue Hole (BH) 6 Gilchrist Blue (GB) (>0.35 mg L 1 ) ( >2mg L 1 ) N O 3 = 0.66 mg L 1 NO 3 = 1.7 mg L 1 DO = 2.1 mg L 1 DO = 4.0 mg L 1 F 03 Algae = 87% Qualitative low Low DO 5 Manatee (MN) 5 Rock Springs (RS) (<2 mg L 1 ) NO 3 = 1.8 mg L 1 NO 3 = 1.3 mg L 1 DO = 1.6 mg L 1 DO = 1.0 mg L 1 F 03 Algae = 94% Qualitative low Low N High DO 3 Fern Hammock (FH) 5 Cypress (CP) (<0.35 mg L 1 ) (>2 mg L 1 ) NO 3 = 0.12 mg L 1 NO 3 = 0.3 mg L 1 DO = 6.0 mg L 1 DO = 5.2 mg L 1 F 03 Algae = 54% F 03 Algae = 0% Low DO 1 Alexander (AX) 1 Williford (WL) (<2 mg L 1 ) NO 3 = 0.07 mg L 1 NO 3 = 0.07mg L 1 DO = 1. 6 mg L 1 DO = 1.4 mg L 1 F 03 Algae=74% F 03 Algae = 13% Additional Springs Mill Pond (MP) NO 3 = 0.23 mg L 1 DO = 0.39 mg L 1 F 03 Algae=52% Mission (M S) NO 3 = 0.51 mg L 1 DO = 0.83 mg L 1 F 03 Algae=86.4% NO 3 = 0.39 mg L 1 DO = 0.42 mg L 1 F 03 Algae = 67%


43 Table 2 2. Pearson r correlations for primary variables of interest included in the regression models, average aggregation. Numbers marked with asterisks are significant at p < 0.05 SRP NO 3 DO Cond. Temp. pH Vel. Light Algae SAV SRP NO 3 0.16* DO 0.12 0.11 Conductivity 0.36* 0.07 0. 01 Temperature 0.41* 0.03 0.02 0.56* pH 0.04 0.47* 0.27* 0.02 0.22 Velocity 0.34* 0.16 0.16 0.07 0.19 0.14 Light 0.17 0.24* 0.28* 0.42* 0.17 0.15 0.05 Algae 0.24* 0.12 0.12 0.56* 0.42* 0.00 0.06 0.17 SAV 0.63* 0.31* 0.07 0.22* 0.26* 0.17 0.47* 0.15 0.20 Gastropods 0.13 0.06 0.42* 0.33* 0.13 0.14 0.12 0.32* 0.61* 0.16


44 Table 2 3. Three general linear regression models for algae biomass, with a suite of potential explanatory vari ables, N = 74. Model 1 uses the hypothesized predictor alone, model 2 includes the suite of common drivers of algal abundance, and model 3 includes all sampled environmental variables (data aggregation at site level; n=74 for all model ). Variable SS Df MS F P Beta () Adjusted R 2 Whole Model F Whole model p AICc Model 1 Intercept 600.7 1 600.7 95.95 <0.0001 0.37 44.54 <0.0001 135.76 Gastropods 278.8 1 278.8 44.54 <0.0001 0.62 Error 450.7 72 6.26 Model 2 Interc ept 9.8 1 9.8 1.89 0.202 0.45 14.47 <0.0001 126.58 Gastropods 302.6 1 302.6 58.23 <0.0001 0.67 SRP 2.3 1 2.3 0.44 0.511 0.06 NO 3 0.03 1 0.03 0.01 0.938 0.01 Velocity 17.5 1 17.5 3.37 0.071 0.18 Canopy 58.2 1 58.2 11.19 0.001 0 .31 Error 353.4 68 5.4 Model 3 Intercept 10.4 1 10.4 2.22 0.141 0.53 9.25 <0.0001 126.31 Gastropods 101.1 1 101.2 21.55 <0.0001 0.54 SRP 3.9 1 3.9 0.83 0.366 0.11 NO 3 6.2 1 6.2 1.31 0.256 0.12 DO 3 1 3 0.6 5 0.425 0.08 Conductivity 14.6 1 14.6 3.12 0.082 0.24 Temperature 10.6 1 10.6 2.26 0.138 0.18 pH 2.3 1 2.3 0.49 0.488 0.08 Velocity 9.4 1 9.4 2.00 0.162 0.15 Canopy 6.8 1 6.8 1.46 0.232 0.14 SAV 2.7 1 2.7 0.57 0.454 0.09 Error 295.6 63 4.7


45 Table 2 4. Three general linear regression models for gastropod biomass with a suite of potential explanatory variables, N = 77. Model 1 uses the hypothesized predictor alone, model 2 includes the suite of common predictors of gastropod abundance, and model 3 includes all sampled environmental variables. Variable SS df MS F p Beta () Adjusted R 2 Whole Model F Whole model p AICc Model 1 Intercept 113.2 1 113.2 46.9 <0.0001 0.228 23.402 <0.0001 67.9 2 DO 56.5 1 56.5 23.4 <0.0001 0.49 Error 181.1 75 2.4 Model 2 Intercept 0.05 1 0.1 0.0 0.846 0.555 16.793 <0.0001 24.56 DO 22.1 1 22.1 15.9 <0.0001 0.39 Temperature 1.8 1 1.8 1.3 0.263 0.11 Velocity 10.7 1 10.7 7 .7 0.007 0.25 Canopy 41.2 1 41.2 29.6 <0.0001 0.50 SAV 8.9 1 8.9 6.4 0.014 0.23 Conductivity 35.4 1 35.4 25.4 <0.0001 0.54 Error 97.4 70 1.4 Model 3 Intercept 7.5 1 7.5 6.1 0.016 0.609 14.129 <0.0001 25.53 SRP 2.9 1 2.9 2.4 0.128 0.17 NO 3 7.2 1 7.2 5.9 0.018 0.22 DO 29.8 1 29.8 24.3 <0.0001 0.43 Conductivity 28.7 1 28.7 23.4 <0.0001 0.51 Temperature 0.2 1 0.2 0.2 0.688 0.04 pH 14.2 1 14.2 11.6 0.001 0.33 Velocity 12.2 1 12.1 9.9 0.002 0.27 Canopy 29.3 1 29.3 24.0 <0.0001 0.50 SAV 8.9 1 8.9 7.2 0.009 0.28 Error 82 67 1.2


46 Table 2 5. BIC m grazer biomass thresholds. Each gastropod threshold was tested for fit for curves with 1 vs. 2 modes. Shown is the best of the two models for each threshold, and the BIC difference between it and the less fit model. statistic tests the null hypothesis that the distribution is unimo dal, with significant p values indicating multimodality Where BIC and the diptest Gastropod threshold (g m 2 ) # Modes BIC diff. Mean 1 Mean 2 Variance Hartigan's Diptest: D p value <12.1 1 5.072 0.263 2.525 0.072 0.459 >12.1 2 8.663 2.827 2.006 1.659 0.072 0.009 <15.6 1 3.859 0.085 2.75 0.051 0.795 >15.6 2 7.77 2.793 2.208 1.77 0.074 0.015 <20.1 1 3.436 0.201 3.182 0.047 0.825 >20.1 2 8.01 2.769 2.284 1.729 0.077 0.014 <25.7 1 3 .252 0.117 3.358 0.044 0.928 >25.7 1 or 2 8.76 2.73 2.508 1.709 0.071 0.07 <33.1 1 or 2 1.216 3.543 0.612 1.818 0.029 0.993 >33.1 2 16.723 2.74 2.599 0.488 0.085 0.018


47 B Figure 2 1 Photographs of the same location at the Ich etucknee Headspring (A) in 1976 (photo: C. DuToit) and (B) 2010 ( photo: D. Liebowitz), exhibiting a system shift from primary producer dominance by a diverse community of submerged aquatic macrophytes to dominance by the filamentous cyanobacteri um Lyngbya sp. A


48 Figure 2 2. Bivariate regressions of gastropods and algae at three scales of data aggregation. (A) Spring level data averaging (11,609.4x 1.68 R = 0.37, p < 0.001); (B) site level data averaging coded by spring (fit on figure); and (C) non averaged, sample level data (y = 5.0x 0.40 R = 0.08, p < 0.001). Note the different axis scales on each graph. A) B ) C ) Gastropod biomass (g m 2 ) Algae biomass (g m 2 )


49 Figure 2 3. Bivariate regression of gastropods and DO at three scales of data aggregation. (A) S pring leve l data averaging (y = 19.9x 0.59 R = 0.10, p = 0.08); (B) site lev el coded by spring with symbols sites represented by circles are high DO springs, all others are low DO springs; solid line is the fit including the ci rcled MP outliers (y = 6.5x 1 .21 R = 0.19, p < 0.001), dashed line is th e fit without MP (y = 14.3x 0.62 R = 0.07, p = 0.01); (C) sample level, no n aggregated data (y = 2.4x 1.28 R = 0.10, P < 0.001). Dissolved Oxygen (mg L 1 ) A) C) B) Gastropod biomass (g m 2 )


50 Figure 2 4. Probability of gastropod biomass being above the proposed threshold of 20.1 g m 2 (above red line), as a function of DO, pre sented for breaks at DO = 0 1 mg L 1 (severe hypoxia), 1 5 mg L 1 (mild hypox ia to normoxia), and above 5 mg L 1 (above DEP threshold, high oxygen).


51 Figure 2 5. Longitud inal be havior of gastropod biomass (A), and algal biomass (B ) by longitudinal sit e for high vs. low DO springs. Bars represent means standard error. ANOVA F and p values for the log transformed data for: A) G astropods vs. DO level (9.11, 0.002), Site (4. 39, 0.02), and DO*Site ( 8.69, < 0.001); and B) A lgae vs. DO level ( 7.33, 0.01 ), Site ( 0.06, 0.94 ), and DO*Site ( 0.33, 0.72 ) in lower case letters. 0 50 100 150 200 250 300 1 2 3 Average Algae Biomass (g m 2 ) Site Low DO High DO 0 20 40 60 80 100 120 140 1 2 3 Average Gastropod biomass (g m 2 ) Site Low DO High DO A ) B) a b b b b b


52 Figure 2 6. H istograms of residuals of the regression of algal biomass with gastropod biomass, fit for unimodal vs. bimodal distributions. Graphs contain the values and fit curves for gastropod biomass < 20.1 g m 2 (A ), and >20.1 g m 2 (B ). BIC fitting data and Hartiga are presented in table 2 5. Proportional Frequency Gastropod biomass < 20 g m 2 Gastropod biomass > 20 g m 2 6 4 2 0 2 4 6 6 4 2 0 2 4 6 Residual algae biomass 0.00 0.05 0.10 0.15 0.20 0.25 0.00 0.05 0.10 0.15 0.20 0.25 A) B) Residual algae biomass


53 CHAPTER 3 EXPERIMENTAL EVIDENCE OF GRAZER CONTROL OF FILAMENTOUS ALGAE IN A SPRING FED RIVER Introduction Consumer control of primary producers is increasingly viewed as a key driver of ecosystems states with evidence of significant top down control in terrestrial ecosystems (Estes et al. 2011) and even stronger effects in aquatic systems (Cyr and Pace 1993, Borer et al. 2005, Heck and Valentine 2007, Gruner et al. 2008, Estes et al. 2011) Proliferation of nuisance algae has motivated a focus on aquatic ecosystems where bot tom up nutrient and top down grazer impacts are potentially confounded due to the similar ity of the symptoms Experimental studies have provided clear evidence of top down control. A meta analysis of 25 consumer control experiments (Cyr and Pace 1993) fo und grazers remove a median of 79% of biomass and a nother meta analysis of 89 stream experiments found ambient grazer densities reduc ed periphy ton biomass in 70% of studies ( Feminella and Hawkins 1995) A much larger meta analysis of 865 experimental studies across multiple ecosystem types concluded that grazers remove on average 59% of periphyton biomass, with no significant differences between aquatic biomes (Hil lebrand 2009) Despite this convergence of evidence, considerable uncertainties and mediating factors remain regarding the presence and strength of consumer control across systems therefore top down controls s hould be demonst rated on a case by case basi s. Positive relationships between consumers and algae are observed in 10% of the studies (Hillebrand 2009) and many additional factors directly and indirectly affect primary producer bio mass, such as hydrology (Poff et al. 1990, Riseng et al. 2004) light (Odum


5 4 1956, Phlips et al. 2000) nutrients (Biggs 2000, Dodds 2007) and a host of indirect effects (Holomuzki et al. 2010) The stre ngth s top conditional and depe nd on a wide variety of factors, such as consumer composition (Lamberti et al. 1987, Steinman 1996, Shurin et al. 2002) water velocity (Opsahl et al. 2003, Poff et al. 2003) and light (Mallory and Richardson 2005) Dissolved oxygen (DO) influences grazer survival a s well as grazing rates (Diaz 2001, Wu 2002) and contaminants may also reduce grazer populations (McMahon et al. 2012) Population density is an importa nt factor for grazing, as higher density populations will generally have higher cumulative grazing rates until they overcrowd to the point of competition (Feminella and Hawkins 1995) Loss of consumer control can lead to alg al blooms (Lamberti et al. 1989, Wo rm and Lotze 2006) and positive feedback mechanisms can make bloom conditions stable (Scheffer et al. 2008) and thus resistant to restoration. In the simplest conceptualization of the autotroph/grazer relationship (assuming only energetic caps on grazer biomass, and type I functional response) h igher autotroph pr oductivity is linearly related to higher cumulative grazing rates, leading to increasing herbivore biomass and consistent fractional standing crop consumption (Rosenzweig 1971, Feminella and Hawkins 1995, Borer et al. 2005) Under such circumstances healthy grazer populations can potentially keep algal biomass consistently low In many examples, however, non linear threshold eff ects have been observed with varying grazer densities. In these cases, when grazer populations fall below threshold levels (for any exogenous reason), algal biomass proliferates. If grazer populations can rebound and resume consumption at a higher rate th an algal growt h, and the algae is still palatable the system will ultimately


55 return to low biomass equilibrium (Steinman et al. 1987, Suding and Hobbs 2009) However, if algae become less palatable or inaccessible as biomass accumulate s (Gliwicz 1990, Alstyne et al. 1999, Gragnani et al. 1999, Lotze and Worm 2000, Korpinen et al. 2008) an algal dominated state can persist or even become exacerbated by further diminishing the grazer population (Scheffer et al. 2008) prings In north Florida, the karst ic Floridan Aquifer feeds the highest density of large artesian springs in the world, with > 30 first magnitude springs (> 2.8 m 3 s 1 ) and > 700 named spri ngs (Scott et al. 2004) These springs exhibit remarkable stability in flow, te mperature and water chemistry, and th e chemostatic properties have made them foci for ecological research (Odum 1956, 1957a, 1957b, Heffernan et al. 2010a) Alarming ecosystem changes have occurred in m any springs in the past 20 years as the autotrophs switched from dominance by submerged aquatic macrophytes to proliferat ions of dense mats of filamentous algae (Stevenson et al. 200 4 ) T he causes of these transitions however, remain unclear (Heffernan et al 2010b ) Nutrient enrichment has been a leading explanation for these algal bloo ms (Stevenson et al. 2007) but a review of the available data suggests that the role of nutrients is equivocal, while the role of consumers and the factors that c ontrol them merit greater attention (Heffernan et al. 2010b) Research in similar systems in the sou theast US ha s shown that grazers, and gastropods in particular exert strong control algal accumulation (Hill et al. 1992, Rosemond et al. 1993, Brown et al. 2008a) and recent studies suggest that this could be an i ( 2010b ) for example, found that DO concentrations and qualitative measures of grazer presence were


56 significan t predictors of algal biomass in a study spanning 28 springs. Additionally, Chapter 2 demonstrated that algal biomass and gastropod biomass were negati vely cor related across multiple sites in 11 springs, while other potential predictors of algal biomass (nutrients, flow) were not significant predictors of algal biomass In a ddition, these investigations found that in low gastropod environments (< 30 g m 2 wet weight), algal biomass was always high, wh ereas in the presence of higher gastropod biomass (> 30 g m 2 ), algae biomass could be either high or low. Dormsjo ( 2008) provided the first experimental evidence ings Conducting experiments with 10 cm diameter petri dish enclosures she found that grazers could impact algal accumulation under high oxygen conditions but not in severely hypoxic sites Studies that combine experiment s and field observations, althoug h rare, are required to ensure that patterns observed in nature are consistent with hypothesized mechanisms (Feminella and Hawkins 1995) This study uses an experimental approach to test mechanisms of algal cont rol by grazers suggested by ecosystem level surveys ( C hapter 2 ). Hypotheses We sought to test two hypotheses regarding grazer controls of algal abundance in springs. The first is that grazer density controls algal accumulation. This leads to the prediction that experimental manipulations of grazer density will influence algal standing crop with high densities leading to reduced algal biomass. The second hypothesis is that grazers are able to control algal accumulation when initial algae conditions are low, but are unable to do so when the initial algae biomass is high (i.e., algae have a critical escape density from gastropod consumers) The predictions that follow from this hypothesis are that a) at low initial algal biomass, high snail biomass will mainta in low


57 algal levels, but that b) algal biomass will not decline with any level of grazer biomass when initial algal abundance is high. Methods Study S ite and B iota The Ichetucknee River, located in the Ichetucknee Springs State Park in north central Flor ida (Fig. 3 1), originates at the Ichetucknee Head spring and flows for 8 km before discharging into the Santa Fe River. It is fed along its length by nine named springs with distinct chemistry. Median Ichetucknee Head spring discharge is 1.3 m 3 s 1 Addi tional spring flows along the length of the river result in median river discharge of 9 12 m 3 s 1 (Heffernan et al. 2010a) As a state park, t he Ichetuc k nee is relatively well protected, but is subject to heavy recreational pressure during the summer. A carrying capacity study ( DuToit 1979) led to seasonal visitat ion restrictions, which have enabled submerged aquatic vegetation recovery (Kurz et al. 2003, 2004) Nuisance algal levels are ill defined, but EPA Rapid Bioassess ment Protocols suggest biomass thresholds at > 50 g m 2 AFDM, or > 10 0 g chl a m 2 (Barbour et al. 1999) which have been recorded near many of the spring vents (Frydenborg 2006 Liebowitz et al, in prep.). The aquatic fauna of the Ichetucknee River includes a wide variety of invertebrate species, with biomass dominated by the P leurocerid snail Elimia floridensis The genus Elimia can account fo r up to 95% of biomass in some s outheastern US streams (Newbold et al. 1983, Hill et al. 2001) Elimia spp. are known to be generalist grazers, capable of preventing accumulation of filamentous alga e particularly cyanobacteria (Tuchman and Stevenson 1991) Early surveys of three locations in the Ichetucknee River report a mean Elimia density of 315 g m 2 wet weight (range 13 8 579 g m 2 ; DuToit 1979) Recent sampling found that Elimia accounted for 86% of


58 invertebrate abundance in the river, but only 10% in the feeder springs where algal proliferation is most severe (Dormsjo 2008). While high Elimia density is still occasionally observed in the river (surveys in 2010 recorded one site with 517 g m 2 ; D. Liebowitz, unpublished data) most densities were much lower in these surveys, averaging 37 g m 2 at nine spring and river locations. Gastropod S ur veys On 6/13/11, we conducted a survey of the biota in 10 cross sectional transects, each approximately evenly spaced along the length of the Ichetucknee River fro m the Headspring (HS) to the S outh T ake out (ST). Three samples were collected per transect, evenly spaced across the width of the river. A modified Hess type sampler (area = 0.084 m 2 ) was used to collect all biomass within the sampling device and each sample was individually bagged, labeled, and kept on ice until processing within 72 hours. Each sample was washed and sorted to family level taxa, and all gastropod shells were checked for the presence of bodies, to ensure empty shells were not included. Animals were tamped with paper towel to remove excess water then weighed. Wet weights were conv erted to dry weights using an empirically derived relationship from surveys in Chapter 2 (y = 0 .64x 0.11, R = 0.98), and the three samples per transect were averaged to assess the range of variation in consumer density within the Ichetucknee River Ex p erimental Design and Field S ampling E xperiment s were conducted at four sites (Fig. 3 1C) with slightly different water chemistry (DO; nitrate, NO 3 ; orthophosphate, PO 4 3 ), channel morphology, canopy cover, and DO regimes (Fig. 3 2). Diel DO profiles take n for a week at each site between 21 February and 30 March 2011 (Fig. 3 2) show the magnitude of variation in


59 this key environmental variable T wo sites were near spring vents (Ichetucknee Headspring, HS, high DO, 0.8 mg NO 3 N L 1 20 g PO 4 P L 1 ; Mill Po nd, MP, low DO, 0.2 mg NO 3 N L 1 55 g PO 4 P L 1 ) and two were in the main stem of the river ( Grassy Flats GF, m edium fluctua ting DO, 0.38 mg NO 3 N L 1 41 g PO 4 P L 1 ; South Takeout, ST, high fluctuating DO, 0.31 mg NO 3 N L 1 40 g PO 4 P L 1 ). In si tu enclosures (flume s) were created using 15 cm diameter polyvinyl chloride (PVC) tubes, cut to 77 cm length and split laterally, creating a footprint area of 0.12 m 2 Each flume was enclosed with nylon mesh (ca. 1 mm aperture) at the top and both sides, with one side affixed with hook and loop seals to allow repeated access. Strips of plastic lattice raised tiles 1 cm off the base of the flume to minimize sediment accumulation. Flumes were suspended with PVC stakes affixed on each corner to maint ain flume s 10 cm below the water surface, and oriented parallel to the flow (Fig. 3 1D). This study consisted of two experiments, the first initiated with low algae biomass, the second initiated with high algae biomass. Expe riment 1 was conducted at all four prev iously described sites. Each site contained nine flumes, allowing three replicates of three treatments: (1) 0 snail exclosures (0 g m 2 ), (2) ambient snail densities where flumes were open on the ends (mesh was present but unsealed at the upstream end to maintain consistent flow effects with the two sealed treatments, and absent in the downstream end) and (3) high snail enclosures (70 E. f loridensis individuals per enclosure, equivalent to ~306 g m 2 or ~ 600 adult individuals m 2 ). This high snail enclos ure level of 306 g m 2 wet weight is in the top 10% of pleurocerid biomass values from the 480 samples from Chapter 2 (which found a maximum


60 pleurocerid density of 845 g m 2 wet weight in Cypress Spring) ; however it is approximately the mean of the pleuro cerid biomass found in Ichetucknee Springs in 1979 (Dutoit 1979) and a pproximates extant levels at GF therefore an ecologically relevant value Algae was grown on u nglazed ceramic tiles (5 cm 2 ) which were placed in e ach flume and allowed to incubate for five days prior to initiation of Experiment 1 on 4 February 2011. Tiles were collected at the start of the experiment, then at days 4, 12, and 28 of the experimental duration Flumes were inspected every 2 3 days during which the top mesh was brushed clean, excess sediment was removed by gently agitatin g the water column to re suspend and flush out the sediment, and dead snails were replaced, if ne cessary Experiment 2 was conducted only at HS, GF, and ST; MP was not used because of extremely high grazer mortali ty during Experiment 1. In this experiment we used high initial algae biomass and four grazer density treatments E ach site contained 12 flumes, with three replicates of each treatments: (1) snail exclosures (0 g m 2 ), (2) low density enclosure (25 snails, 110 g m 2 ), (3) medium density enclosure (50 snails, 220 g m 2 ), and (4) high density enclosure (75 85 snails, 330 g m 2 ). Unglazed ceramic tiles were cultured for approximately two months prior to initiation of the experiment. However, during this culturing period a wave of chironomid larvae i nfiltrated the flumes, necessitating tile clearing and additional algal accumulation time. This delayed the start and shortened the duration of the experiment, the end of which was constrained by the start of high park visitation. During Experiment 1 we ob served very little algae accumulation at ST even in ungrazed conditions, either on tiles or in the river. Therefore tiles incubated at HS were transported to ST at the initiation of Experiment 2, to explore


61 whether algae did not naturally recruit there, or alternatively could not grow once established. Due to the shortened timeline of Experiment 2, tiles were collected at initiation ( 13 May 2011 ), and on day s 7 and 13. Because GF was not affected by visitor pressure, the experiment was continued for an addi tional 10 days at that site. Environmental variables were recorded for the duration of both experiments. We measured canopy cover using a densiometer, surface flow velocity using the float method, and DO, water temperature, specific conductance, and pH us ing a YSI 6920 multi parameter sonde (Yellow Springs Ins truments, Yellow Springs, OH). At ea ch site we also collected a 500 mL water sample into an acid washed polyethylene bottle that was acidified to pH 2.0 using hydrochloric acid, and stored below 4C u ntil colorimetric analysis for NO 3 and SRP For both experiments, three tiles were collected from the same location in each flume and placed in individual, labeled sealable plastic bags. They were put on ice in a dark cooler for transport to the laborato ry, and kept on ice until processed, within 48 hours of sample collection. Laboratory P rocessing Each experimental tile was scraped with a clean razor blad e, rinsed into a beaker using deionized water to a volume of 200 mL and split into two subsamples of equal volume. Subsamples were individually filtered through a pre combusted and pre weighed W hat man GF/F glass fiber filter One filter was frozen for Chlorophyll a analysis (Standard Methods 10200 H (APHA 2005) with the modificatio n of pigment extraction in solution (Sartory and Grobbelaar 1984) ). The other filter was processe d for ash free dry mass (AFDM) by drying the filters at 60 C for 24 hours, weighing them, combusting them with a muffle furnace at 500 C for 1 hour, and then r eweighing them to calculate AFDM. Water samples were filtered using a 0.45 m glass fiber filte r and


62 analyzed within 28 days for NO 3 ( as N, using EPA 353.2) and PO 4 3 ( as P, using EPA 365.1) Data Analyses Four response variables were examined individually and together to explore algal responses to grazing: (1) Chlorophyll a g m 2 (2) AFDM g m 2 (3) percent removal (%R), and (4) removal rate (RR). The first and second response variables are direct measurements. The third response variable is the percent of algae AFDM removed (Worm and Lotze 2006) comparing grazed biomass (A G ) to ungrazed biomass (A U ) treatments on the fin al collection date: %R = 100*(A U A G )/A U (3 1) The fourth response variable is the daily removal rate of AFDM per treatment: RR = ( A U A G ) / t (3 2) where A U and A G are the same as they are for %R, compared between sampling times, and t i s the number of days of grazing function of both direct consumption and bioturbation or other forms of mechanical disturbance and export (Cattaneo and Mousseau 1995) All response variables were tested for normality u sing the Kolmogorov Smirnov test. When variables deviated significantly from normality, we conducte d the Box Cox analyses to find the mos t appropriate transformation with resulting lambdas consistently near zero suggesting log arithmic transf ormations (Osborne 2010) Therefore, all response values were natural log transformed for statistical analyses, though averages are presented for untransformed values.


63 To test the significance of both site and treatment on each response variable, we used f actorial ANOVA on natural log transformed data to detect significant differences among sample means at p < 0.05 Specifically, we examined site, treatment, and hoc tests were used to detect significant results (p < 0.05) for factors without interaction effects (treatment, site). All analyses were performed using Statistica (v10 STATsoft, Inc.). The relationship of ungrazed samples with environmental variables was tested with ordinary least squares (OLS) regressi on using site means (n=4), providing very low statistical power but preliminary information for future study. To account for varying grazing pressure in ambient treatments between sites, we used linear regression on log transformed data to assess bivariat e relationships between grazer density and algal biomass Results Spatial Patterns of Gastropod Grazers Gastropods from six native families and one invasive species ( family Thiaridae) were distributed along the river in a highly variable manner (Fig. 3 3 ), with the family Pleuroceridae ( Elimia spp .) comprising the dominant biomass presence. Pleuroceridae were 57% of the biomass, and the family Ampullariidae (apple snails, a genus that primarily consumes macrophytes rather than alga e (Dillon 2000) ) was the second most abunda nt, accounted for 14% of the sampled gastropod biomass. The highest snail density was found at GF, with total gastropod biomass approaching levels seen by Dutoit (1979). Additional Elimia density metrics derived from the presence of gastropods in the ambie nt flumes on 3 March 20 11 showed a wide range of densities, with 55 155 g m 2 at HS, 107 363 g m 2 at GF, 0 g m 2 at MP, and 4 128 g m 2 at ST.


64 Experiment 1: Low Initial Algae C onditions We observed strong and significant effects of Elimia grazing on all measures of algal accumulation, with visually apparent (Fig. 3 4) and statistically significant variability between sites (Fig. 3 5). Chlorophyll a and AFDM were significantly different for all treatments (p < 0.001 for both response variables). Post hoc c omparisons ranked 0 grazers > ambient grazers > enclosed high grazers (Table 3 1). Algal AFDM and grazer biomass exhibited a strong negative relationship (Fig. 3 6), but with some variation between sites. However, combining all sites together showed the p resence of a potential threshold at 100 g m 2 of gastropods: when gastropod biomass was below that level, algae began to accumulate beyond 10 g m 2 a level equated with low algae biomass across grazer studies (Feminella and Hawkins 1995). Above 100 g m 2 of gastropods, the algal AFDM (M= 2.6, SD = 1.7 g m 2 ) was not significantly different from the initial conditions (M= 2.1, SD = 0.95 g m 2 ; t(38) = 1.1, P = 0.25), w hereas below 100 g m 2 of gastropods, algal AFDM (M= 16.4, SD = 12.6 g m 2 ) was significan tly higher than the same initial conditions (t(42) = 5.5, p < 0.001). Mean percent removal (%R) differences between ambient (48%) and high g razer enclosure (78%) treatment s was nearly significant (p = 0.057) across all sites. However, there was strong evid ence of a site level effect for this response variable; HS and GF both averaged more than 90% algal removal in high grazer treatments, while snails at MP and ST were much less effective (14% to 64%; Fig. 3 5). There was no effect of treatment on removal ra te (RR), suggesting that grazers consumed cons tant amounts per unit biomass. However, we did observe one significant site effect; HS and GF had average removal rates of 1.6 and 1.0 g AFDM m 2 d 1 respectively, which were both significantly higher than MP and ST with 0.2 and 0.1 g AFDM m 2 d 1 respectively


65 None of the environmental relationships with algal AFDM were statistically significant, which was expected because of low power (n = 4) for detecting those effects However we note the few marginally significant factors: water velocity exhibited a negative relationship in exclosures (r = 0.94, p = 0.06) and DO concentration had a positive relationship in the enclosures (r = 0.92, p = 0.08). Experiment 2: High Initial A lgae C onditions We observed significant gr azer effects on both Chlorophyll a and AFDM, but only for the highest grazing treatment (330 g m 2 snail wet weight); intermediate densities (110 g m 2 and 220 g m 2 ) were not significantly different from snail free exclosures (Fig. 3 5). Grazing efficienc y (%R) and removal rate (RR) were not significant ly different across the three grazer treatments (Table 3 1). The highest snail density treatment (330 g m 2 ) averaged 54% removal (strongly influenced by extreme removal of 82% at HS), while the intermediate (220 g m 2 ) and low (110 g m 2 ) treatments averaged only 30% and 23% removal, respectively. Removal rates varied significant ly among sites and treatments, with a high of 2.5 g m 2 day 1 in GF (at 220 g m 2 ) and lows of 0.09 g m 2 day 1 at ST (220 g m 2 ) (Fig. 3 5). Algae AFDM increased in all treatment s at GF during the extra week of experimental duration (Fig. 3 7). At the end of the experiment, AFDM had reached 49 g m 2 in the snail free exclosures, and the snail enclosure treatments were not significa ntly different (F = 0.4, p = 0.8, Table 3 1). Although there was negligible snail mortality for the first three weeks ( until 26 May ) at GF there was unexplained snail mortality between 26 May and 5 June Final snail biomass averaged 258 g m 2 in the 2 2 treatments.


66 Negligible biomass accumulation on tiles at ST during the 2 month culturing period necessitated transplanted algae for experiment 2 at that site. As such, the experim ent test s whether algae simply did not recruit at that site, or if algal biomass was actively diminished. All grazed treatments at ST los t biomass w hereas the 0 snail treatmen t gain ed a small amount of biomass however none of the treatments w ere significant (F(3, 8) = 1.27, p = 0 35). Discussion Grazer Impacts in Low Initial Algae C onditions This study provides s trong evidence that grazers c an inhibit formation of algal blooms, given high snail density and low initial algae conditions. In the high grazer enclosures, the final AFD M was even lower than the negligible levels present at initiation of the experiment for HS, GF, and ST, while grazer exclosures exhibited positive but varying levels of algal accumulation. Ambient grazer levels were intermediate between grazer enclosure an d exclosure treatments; all three treatments together exhibited a highly significant negative association between grazer and algal biomass, consistent with our central prediction for hypothesis 1, and with behaviors observed broadly across aquatic ecosyste ms (Hillebrand 2009) and specifically in streams (Feminella and Hawkins 1995) That snails inhibited algal accumulati on except at the hypoxic MP sit e, despite site variation in ungrazed algal productivity (Fig. 3 5) suggests that given sufficient snail biomass, algal productivity is transferred to secondary production or sloughed off rather than accumulating at the primary producer level. Thi (Rosenzweig 1971) wherein increases in primary production are rea lized at higher trophic levels. It also reinforces the findings of Feminella and Hawkins ( 1995)


67 whose meta analysis of stream grazing experiments found that grazers generally maintained low algae biomass (< 10 g m 2 ), despite dramatically different production (0 to 80 g A FDM m 2 ) in grazer free areas This effect of higher grazing with higher productivity is also seen in the relatively consistent percentage of algae biomass consumed across productivity gradients (Cyr and Pace 1993, Worm and Lotze 2006) The absence of algal accumulation at the South Take out (ST) could have been a consequence of the high water velocities or low NO 3 at thi s site. H owever we suspect that a large factor was the cage effect, as the water at that point in the river had a relatively large suspended sediment load, which it deposited when the velocity decreased upon entering the flume through the mesh partial b arrier. Therefore, the algal declines at ST may have been a result of sedimentation The ability of grazers at ambient densities to maintain a low algal biomass illustrates t he management relevance of grazing effects. At HS and GF, ambient gastropod levels (94 and 217 g m 2 respectively) were able to maintain the same low algal biomass as enclosures (330 g m 2 ) for about two weeks. By the end of the month, algal biomass in the ambient treatments at HS had accumulated significantly, wh ereas GF had increased only slightly (Fig. 3 7), likely because ambient snail levels at GF were closer to enclosure levels. The regression analysis of gastropod biomass versus algal biomass shows that g astropod density above approximately 100 g m 2 maintained al gae AFDM at the same level as the initial conditions, with most points below 4 g m 2 algae AFDM and MP having two points at 6 g m 2 algae AFDM, perhaps slightly higher due to lower efficiency grazing from hypoxic stress While grazed tiles remained at low biomass ungr aze d tiles reached 51.2 g m 2 exceeding the US EPA nuisance algae


68 threshold of 50 g m 2 (Barbour et al. 1999) in a short period of time. Historical Elimia densities (mean = 315 g m 2 minimum = 138 g m 2 ; DuToit 1979) far exceed this 100 g m 2 wet weight gastropod biomass threshold suggesting that snails would have been capable of ma intaining low algal biomass under historic conditions In contrast, curr ent average grazer levels are mostly below these values except at GF. Env ironmental Mediation of Grazer I mpact It seems likely that DO impacts algal growth in grazed treatments by su ppressing grazing rates. DO was not significantly related to AFDM in grazer exclosures as a direct driver of algae, but marginally significant in the enclosures (r = 0.92, p = 0.08) indicating impacts on grazing potential. The hypoxic MP site was the only one where grazer enclosures were ineffective and allowed algal biomass to increase, likely because of high snail mortality that did not occur in any of the other sites. Snails in the enclosures were sourced from the immediate vicinity of the experimental unit, and therefore presumed to be acclimated to the local conditions. We speculate that these snails may survive in this extremely low DO environment through behavioral modifications (Hanley and Ultsch 1999, Wu 2002, Cheung et al. 2008) specifically positioning themselves close to the air water in terface, particularly at night. The flumes prohibited access to the water surface, and thus precluded the behavioral coping mechanism. Although snails that died were replaced regularly, enclosure tiles were less completely grazed than at other sites suggesting that DO stress m ay lower grazing rates, even where biomass may is high. This is consistent with results from chapter four, which found lower per capita grazing rates in hypoxic treatments. We caution that there was only one site with severe hypoxia, and the mortality coul d also have been caused by some other unmeasured factor. However, extreme low DO is stressful for most


69 aquatic organisms (Diaz 2001) prior research at this location also suggested higher mortality and torpor (Dormsjo 2008) and controlled DO experiments (C hapter 4) found a strong effect of hypoxia on the survival and grazing rate of pleurocerids Algal accumulation may also be mediated by environmental factors in the absence of grazing with water velocity potentially influencing growth and proliferation (Poff et al. 1990 ) In the grazer treatments, only DO showed even marginal associatio ns with AFDM, while i n the grazer free treatments, water velocity was marginally significant (r = 0.96, p = 0.06). Additionally, turbidity increases downriver (Wetland Solutions Inc. 2011) and as noted previously sett ling of sediments at ST may smother algae and inhibit biomass a ccumulations. We discuss the environmental data despite the non significant p val u es in order to avoid Type II errors (given the low power of N = 4 and high correlations), with the caveat that further research is needed Th ese data suggest that it is wort h testing the hypothesis that with high grazer density, environmental factors are secondary, but in the absence of strong grazer control, the environmental factors may affect primary producers more directly. Grazer Impacts in Algae Bloom C onditions Grazi ng effects at high initial algal biomass (Experiment 2) were more variable by site and snail density, and suggested potential herbivore escape dynamics. Algal biomass decreased at the highest level of grazing (330 g m 2 ) at all three sites for the first 2 weeks, indicating that grazing rates can exceed algal growth rates; if the system continued on that trajectory, it would eventually re turn to low algal standing crop One important caveat is that at the one sit e with the extended experiment duration (GF), there was a reversal in the algal biomass control trajectory, leading to algal accumulation (Fig. 3 7). This could be due to the longer duration of the experiment, and


70 the rest of the sites may have eventually lost control of algal biomass as well (after t he initial impacts of bioturbation leading to temporary declines in algal biomass), given a longer time frame. A lternat iv ely this may have been a consequence of high gastropod mortality at GF 2 with approximately 258 g m 2 However, that is still well above the 100 g m 2 gastropod biomass which prevented algal accumulation in low initial algae sites illustrating the presence of hysteretic behavior Different thresholds could also be a function o f different algal assemblages in the blooms, as some may be more palatable than others. We were unable to run full palatability experiments, but preliminary trials suggest that Elimia can limit new growth in Lyngbya mats, but cannot decrease pre existing b iomass (Liebowitz, unpublished data). Control of algae was less effective at i ntermediate gastropod densities At HS, we observed net accumulation of algae at and below the 220 g m 2 treatment, while at GF the intermediate gastropod density treatments d id decrease AFDM by week 2, but AFDM growth was positive by week 3. This again could be due to snail mortality during that last week, but showed that gastropods could not reduce AFDM at the new average level of biomass treatments (330 = 258 g m 2 220 = 18 7 g m 2 and 110 = 100 g m 2 ). Together, these findings suggest that grazer densities of ca. 330 g m 2 are required to reverse algal accumulation. As a restoration target, this gra zer density is extremely high. Densities in excess of 330 g m 2 are present surveys (450 samples across multiple sites and dates in 11 springs between 2008 and 2010) found only ~ 10% of samples were above this threshold ( chapter 2 ) O nly 25% of samples were above


71 110 g m 2 a density at whic h no evidence of biomass reduction w as found in the current study. In short, all snail treatments in this experiment were high in relation to existing (natural) densities, suggesting that most locations have gastropod populations that are too low to revers e t he trajectory of algal growth. We note, however, that this finding and the threshold at which reversal of nuisance algal biomass is possible, may depend on algal productivity at each site. One mechanism for creating an herbivore escape density is the loss of palatability as algae reach mature growth forms, which inhibit s herbivore population growth and allow s autotrophic biomass accumulation that is unaffected by herbivory (Scheffer et al. 2008) This mechanism does not appear to hold in this study, where grazers continued to remove algae at similar rates (per un it snail; RR) in both experiments. Alternatively, simple declines in grazing efficiency at high prey density (i.e., type II functional response; Holling) or consumer satiation could lead to the cap in grazing as well. T he different thresholds of gastropod density needed to maintain low algal biomass from low initial algae conditions ( 100 g m 2 snails) vs. the density needed to reduce algae from high algae initial states ( > 330 g m 2 snails) suggests a hysteretic system. This could arise from three pot ential scenarios for the longer term dynamics of algae grazer interactions, which hinge on algal palatability and grazer population recovery. The first posits that if the alga e removal in Experiment 2 was due to consumption and exogenous stressors are absent th e grazer population could recover and cumulative grazing rates would increase. This would allow the grazers to eventually reduce a bloom and return to an herbivore dominated system. In the second scenario, if


72 the algae are palatable but the exogenous stres sor (s) remains, the grazer population density may remain low or further decline. Thus cumulative grazing rates would remain depressed, effectively maintaining an algal dominated state. In the third scenario, if the algal removal in experiment 2 was primari ly due to mechanical interference, and the bloom formation is unpalatable or otherwise impenetrable, then the algal dominance would be entrenched, whether or not grazer populations ever recovered. As this study was temporally restricted and was not run lon g enough to capture long term population dynamics, it remains unknown whether this system could eventually recover through a clear unpa latable algal species. This stud y highlights the vital role of grazers in controlling aquatic autotroph biomass, and delineates threshold levels of gastropods needed to maintain herbivore dominated states in Ichetucknee Springs. Management attention is generally focused on the abiotic dr ivers of ecosystem change, but this study provides unequivocal evidence that healthy gastropod communities can forestall nuisance algal blooms and even reduce them at sufficiently high grazer densities M anaging nuisance algae in l ikely benefit from research and management efforts that result in maintaining viable grazer populations A variety of stressors could affect gastropod populations with DO only one of many potential factors, suggesting the importance of assessing multiple hypotheses to fully understand the drivers of ecosystem change (Heffernan et al. 2010b) For example industrial and agricultural contaminants are emerging as an area of concern for the health of aquatic fauna (Evans White and Lamberti 2009, McMahon et al. 2012) and


73 many contaminants have been shown to trigger trophic cascades and ove rabundant primary producers (Fleeger et al. 2003) Aquatic gastropods have so far received little conservation or legal attention in the southeastern US, despite the fact th at it is a hotspot for gastropod diversity (Neves et al. 1997, Lydeard et al. 2004, Brown et al. 2008a, Lysne et al. 2008) Because Elimia spp. live as long as 10 11 years, and start breeding only after they reach 1 2 years of age (Hill et al. 1992, Johnson and Brown 1997, Powles 2000) their populations may be p articularly sensitive to disturb ance and slow to recover Recognition of the central role of grazers as regulators of primary producer composition and abundance should suggest broader approaches to restoration and future research.


74 Table 3 1 Factorial AN OVAs for the four response variables for experiment 1 (left column) and experiment 2 (right column). Post interaction terms. The site*treatment effects for AFD M, %R, and RR are displayed in F igure 3 5 Variable F p Among factor difference (Tukey HSD) F P Among factor difference (Tukey HSD) Experiment 1: Low initial algae Exp 2: High initial algae Algae Chl a Intercept 121.5 p < 0.001 1364.5 p < 0.001 Site 19.3 p < 0.001 HS,MP>ST,GF 33.2 p < 0.001 ST>GF>HS Treatment 37.7 p < 0.001 1>2>3 22.6 p < 0.001 1,2,3>4 Site x Treatment 3.9 0.008 8.3 p < 0.001 Algae AFDM Intercept 554.7 p < 0.001 2969.9 p < 0.001 Site 39.0 p < 0.001 HS,MP>GF>ST 51.2 p < 0.001 GF>HS>ST Treatment 49.3 p < 0.001 1>2>3 13.4 p < 0.001 1,2,3>4 Site x Treatment 5.1 0.002 3.8 0.008 % Removed Intercept 1537.3 p < 0.001 539.2 p < 0.001 Site 2.5 0.099 0.5 0.608 Treatment 4.2 0.05 7 3.2 0.065 Site x Treatment 0.7 0.589 1.5 0.252 Removal Rate Intercept 28.5 p < 0.001 37.2 p < 0.001 Site 14.3 p < 0.001 HS,GF,MP>MP,ST 6.5 0.008 GF, HS>HS,ST Treatment 0.4 0.552 2.4 0.12 Site x Treatment 0.1 0.941 0.6 0.661


75 Table 3 2. Daily accumulation rates of filamentous algae (AFDM g m 2 day 1 ) per experimental treatment (g m 2 wet weight snail biomass) at each experimental site calculated as the slope of the regression line of algae AFDM by date, categorized by trea tment. Note that a few treatments have final AFDM < initial AFDM, yet show positive (but negligible) accumulation rates this is due to small fluctuations of AFDM in intermediate dates. Treatment HS (Headspring) GF (Grassy Flats) MP (Mill Pond) ST (Sou th Takeout) Experiment 1. Low initial algae: 4 February 2011 3 March 2011 Initial AFDM g m 2 (average) 3.36 1.46 2.03 1.65 Final AFDM in 306 g m 2 (average) 2.72 1.41 7.53 1.08 0 g m 2 1.42 0.79 0.55 0.09 Ambient 0.67 0.08 0.40 0.02 306 g m 2 0.05 0.03 0.27 0.00 Experiment 2. High initial Algae: 13 May 2011 26 May 2011 (5 June 2011) Initial AFDM g m 2 27.26 31.83 17.34 Final AFDM g m 2 in 330 g m 2 7.06 29.88 (41.07) 7.32 0 g m 2 0.96 1.33 (0.82) 0.16 110 g m 2 0.12 0.53 (0 .49) 0.47 220 g m 2 0.62 0.47 (0.21) 0.72 330 g m 2 1.80 0.01 (0.35) 0.92


76 F igure 3 1. A) Location of the Ichetucknee springshed in north Florida; B) map of the springshed showing confining layers above the karst aquifer, with the Ichet ucknee Springs State Park boundary drawn in white; C) Ichetucknee Springs State Park, with the four experimental sites marked with stars (Head Spring (HS), Grassy Flats (GF), Mill Pond (MP), and South Takeout (ST)) ; D) photograph of the experimental appara tus at M ill P ond (MP) before being lowered under the water surface


77 Figure 3 2. Five day dissolved oxygen profiles for the four sites with experimental installations, measured between 2 1 February 2011 and 30 March 2011. HS (Heads pring) and MP (Mill Pond) are spring vents and exhibit relatively stable oxygen concentrations. In contrast, GF (Grassy Flats) and ST (South Takeout) are in the main stem of the river, subject to higher fluctuations in DO concentration as a consequence of diurnal fluctuations in b enthic primary productivity and respiration. 0 2 4 6 8 10 12 Dissolved Oxygen (mg L 1 ) Time HS GF MP ST


78 Fig ure 3 3. Gastropod distribution and composition along the length of the Ichetucknee River on 13 June 2011 beginning at HS ( Headspring, transect 1) and ending at ST ( South Takeout, transect 10). Each t ransect represents average values from three biomass samples collected at points evenly distributed along the cross section of the river. 0 50 100 150 200 250 1 2 3 4 5 6 7 8 9 10 Biomass (g m 2 ) Transect Planorbids Physids Pleurocerids Hydrobiids Thiarids Viviparids Ampullariids


79 Figure 3 4. Three flumes in experiment 1, at HS (Headspring) on 3 March 20 11. The center flume shows the hig h Elimia floridensis treatment (~306 g m 2 ), the left flume shows the ambient level, and the right shows the dark algae mat in the 0 snail exclosures


80 Figure 3 5 Experiment 1 (left column, low initial algae) and Experiment 2 (right column, high initial a lgae) means and standard error bars for AFDM (top row ), percent removal (%R) (middle row ), and removal rate (RR) (bottom row ), for the listed grazer treatments at the four locations for Experiment 1 (HS (Headspring) GF (Grassy Flats) MP (Mill Pond) and ST (South Takeout) ) and three locations for Experiment 2 (HS, GF, ST)


81 Figure 3 6 Plot of algal AFDM by gastropod wet weight for Experiment 1, categorized by location (HS (Headspring), GF (Grassy Flats), MP (Mill Pond), and ST (South Takeout)) The we ight of snail biomass was measured on 3 March 2011 for enclosures, any snails that infiltrated the exclosures, and snails found within the ambient grazer treatments. All fits are significant at p < 0.05 level. y = 38.13e 0.009x R = 0.93 y = 14.95e 0.008x R = 0.85 y = 12.84e 0.005x R = 0.55 y = 2.46e 0.003x R = 0.64 0 5 10 15 20 25 30 35 40 45 0 50 100 150 200 250 300 350 Algae AFDM (g m 2 ) Gastropod wet weight (g m 2 ) HS GF MP ST Expon. (HS) Expon. (GF) Expon. (MP) Expon. (ST)


82 Figure 3 7 Algae AFDM g m 2 accu mulation over time, plotted by collection date and categorized by snail treatment (g m 2 wet weight snails) for A) Headspring (HS) Experiment 1 ( low initial algae); B) HS Experiment 2 ( high initial algae) ; C) Grassy Flats (GF) Experiment 1; and D) GF Expe riment 2. Note that GF has a supplemental 10 days duration, allowing an additional collection date on 5 June 20 11. A) B) C) D)


83 CHAPTER 4 DISSOLVED OXYGEN IMPACTS ON ALGAL GRAZING BY ELIMIA FLORIDENSIS IN STREAM MESOCOSMS Introduction Herbivory is a dominant contr ol on the biomass and composition of primary producers (Cyr and Pace 199 3, Estes et al. 2011, Poore et al. 2012) Across a wide array of aquatic systems, grazers regulate algal biomass (Cyr and Pace 1993, Heck and Valentine 2007, Gruner et al. 20 08, Estes et al. 2011) even where nutrient enrichment might otherwise lead to nuisance algal accumulation. That is, where grazers are unaffected by environmental or anthropogenic stressors, increased primary production can be transfe r r ed into secondary p roduction, leading to conditions of high algal productivity and high grazer biomass but low standing crops of algae (Rosenzweig 1971, Feminella and Hawkins 19 95, Borer et al. 2005) Indeed, g razer exclusion studies indicate that consumers reduce algal standing crops by 59 79% yearly (Cyr and Pace 1993, Hillebrand 2009) However, this top down control var ies as a function of both consumer abundance and consumer efficiency, both of wh ich are susceptible to external stressors (Breitburg et al. 1997) Thus, external factors that affect grazers including sub lethal effects, can induce dramatic effects on both the biomass and composition of the autotroph s. Hypoxic S tress Hypoxi c conditions ha ve driven widespread degradation of aquatic ecosystems across the globe (Diaz 2001, Rabalais et al. 2002, St eckbauer et al. 2011) T he effects yet enumerating thresholds for mortality and sublethal effects on the biota can be challenging (Diaz and Rosenberg 1995, Diaz 2001, Steckbauer et al. 2011) Grazer


84 abundance an d distribution is controlled by many factors in addition to dissolved oxygen including predat ion parasites, invasive species, food resources, flow regime, and pollutants, confounding consistent predictive associations between DO and biota, particularly i n flowing waters (Hanley and Ultsch 1999, Garvey 2007) Despite these confounding factors, studies have found that m ean lethal oxygen concentrations (LC 50 ) for freshwater lotic invertebrates range from a 2 day LC 50 of 0.27 mg L 1 for an e phe mer opteran to a 30 day LC 50 of 4.8 mg L 1 for a p lecopteran (Landman et al. 2005) Gastropods are generally more tolerant of low DO with 28 day LC 50 values as low as 0.51 mg L 1 (Das and Stickle 1993) C onventional thresholds of hypoxia ( < 2 mg L 1 ) and severe hypoxia ( < 0.5 mg L 1 ) emerge from comprehensive review s of marine hypoxia studies (Diaz and Rosenberg 1995, Diaz 2001) but may still underestimate sublethal effects. Vaquer Sunyer and Duarte ( 2008) suggest that conventional level s underestimate both lethal and sublethal effe cts in marine systems underscoring the considerable uncert ainty that remains in less well studied freshwater systems. S ublethal effects are far less visible than lethal effects but can impose strong limitation s on individual beha vior that scale up to co mmunity level impacts if the animals can not escape the hypoxic zones Highly mobile species such as fish are generally less resistant to hypoxia, and most respond by leaving the hypoxic environment (Dia z and Rosenberg 1995, Connolly et al. 2004) Less mobile species can respond with behavioral changes and small scale movements such as gastropod migration up the water column to access increased partial pressure of O 2 near the air water interface (Hanley and Ultsch 1999, Vaquer Sunyer and Duarte 2008) a behavior that may ameliorate DO stress but increase predation risk (Breitburg et al. 1997, Saloom and


85 Duncan 2005) And w hile m any species can survive transient hypoxic stress through mobility leaving the ecosystem relatively unimpacted l ong term effects of hypoxic stress can chang e ecosystem s via dramatic decreases in secondary production, which lead to the loss of higher trophic levels as well (Baird et al. 2004, Sturdivant 2011) Th is loss of production in the faunal community is likely due to the shifting of energy a llocation Short term metabolic compensation includes anaerobic respiration or reduced metabolic rates, which decreases most activity (Stickle et al. 1989) and leads to the slowing or cessation of feeding (Kapper and Stickle 1987, Forbes and Lopez 19 90, Das and Stickle 1993) and reproduction (Wu 2002, Cheung et al. 2008) H ypoxia can additionally act as an endocrine disruptor, further inhibiting reproduction (Wu 2002) or weakening resistance to other stressors such as pesticides and other contaminants (Evans White and Lamberti 2009) Surveys of sublethal DO thresholds for different species have found effects at DO levels in freshwater as high as 10.9 mg L 1 for cod, but 90% of the surveyed experiments fo und sublethal effects below 5 mg L 1 (Vaquer Sunyer and Duarte 2008) This coincides well with the US EPA freshwater regulations, which accepted a 5 mg L 1 DO th impairment for nonsalmonid fauna (Chapman 1986) prings Florida has more than 700 named karst aquifer springs, many of which have experienced a shift in primary producers fr om submerged aquatic vegetation to nuisance filamentous algae blooms. W hile historic data on gastropod population densities are scarce, data from the Ichetucknee River, an entirely spring fed river in Columbia County FL, indicate that populations of the na tive gastropod Elimia spp a


86 long lived pleurocerid snail, ( Huryn and Denny 1997, Huryn et al. 2013 ) have declined dramatically (315 161 g m 2 in 1979 vs. 63 70 g m 2 in 2012; DuToit 1979, Liebowitz C hapter 3 ) and that contemporary populations across numerous systems are generall y low ( Chapter 2 ) Recent in situ experiments have illustrated th at native gastropod s can inhibit filamentous algal accumulation in well oxygenated ar eas, but not in hypoxic sites due to high mortality (Dormsjo 2008, Liebowitz C hapter 3 ) D O levels are generally thought to be constant in springs over months or even years yet significant declines have been noted on long er t ime scales (mean across 42 springs of 3.03 0.29 mg DO L 1 in 1972 vs. 2.00 0.24 in 2002 with 18 of 42 springs < 1 mg L 1 ; Heffernan et al. 2010) Moreover because of low frequency water quality monitoring, short term low DO pulses in response to spring flow reversals as downstream waters flood (Martin and Dean 2001) could easily go undetected. Because Elimia in particular are long lived (Huryn et al. 1994) the impacts of even short term lethal events could have long term implications for population recovery and algal reduction via grazing. We sought to test four linked hypotheses about DO effects on Elimia grazing First, we hypothesized that overall relationships between grazer effectiveness and DO would be positive with low DO levels diminish ing g astropod control of algal biomass. T his effect could arise via three mechanisms: decreased survival, decreased per capita grazing rate, and decreased export from bioturbation. S econd we hypothe sized that initial algae conditions impact grazer efficiency, with the specific predicti on that grazers will be able to reduce more absolute biomass in the high algae conditions, but will be less efficient and unable to reduce the algal biomass to low levels. Third we


87 hypothesized that the percent of algae removed i s positively associated with gastropod density with gastropods exerting a stronger overall grazing impact at high densities F ourth we hypothes ized that pulse hypoxi a event s can have lasting impacts with the specific prediction that return of grazing ra tes will lag behind return of oxic conditions Together, these hypotheses explore the overarching question of whether chang es in DO can contribute, via a trophic cascade, to the changes in algal biomass observed in many springs. Methods Stream M esocosm s Experiments were conducted using nine recirculating stream mesocosms (Fig. 4 1) installed in a greenhouse on the University of Florida campus The greenhouse received natural, shaded light conditions, with photosynthetically active radiat ion (PAR) rang ing from 0 150 mol photons m 2 second 1 depending on weather and time of day. M esocosms were constructed of 15 cm diameter 100 cm long PVC flumes. Each flume contained approximately 2.5 L of water ; experimental flow of approximately 1 L min 1 resulted in a residence time of 2.5 min. Perforated clear plexiglass partitions were inserted in each flume to separate biomass treatments (longitudinally), maintain snails in the appropriate compartments (grazed vs. ungrazed, lateral) and evenly distribute flow (Fig 4 1). DO levels were manipulated using f our oxygen stripping towers constructed using 15 c m diameter 15 0 cm tall PVC pipes filled with bio filter media and capped on both ends Water was pumped in at the top and N 2 gas fed in the bottom ; this DO strippi ng method avoids potentially confounding effects on biota associated with other methods (Connelly et al. 2004) Water temperature was maintained between 19 and 23.5 C, a range typical of the system from which the snails were collected.


88 Unfiltered spring w ater was collected from the main spring vent at Blue Springs in Gilchrist County (NO 3 = 990 ppb, SRP = 18 ppb), and seeded with an algal slurry sourced from the same location. Three replicates each of three experimental DO treatments were established ( 0 .25 mg L 1 ): 1) low, 0.5 mg L 1 2) medium, 1.5 mg L 1 3) high, 5.5 mg L 1 (YSI 6920 sonde with 6150 ROX optical DO probe, YSI, Inc.). Water was recirculated in a mann er that mixed both high and low DO waters in Basin 1 before redistributing them (Fig. 4 1) to avoid confounding variation in nutrient availability through the experiment Water for the high DO treatment was pumped from Basin 1 to B asin 2, and gravity distributed with flow controlled by valves to the flumes. W ater for the low DO flumes was pu mped from B asin 1 into the top of one DO stripping tower, suspended above, and feeding into, B asin 3 which was sitting at the same elevation as B asin 2. Three additional suspended stripping towers drew water from and returned water to B asin 3, allow ing a greater quantity of water to be rapidly deoxygenated. L ow DO water from B asin 3 was gravity di stributed via valves to the low DO flumes. M edium DO flumes received water from B asins 2 and 3, blended using flow control valves to yield water at the target DO level. During the 2 day hypoxia event all flu mes were covered by clear plexi glass lids to inhibit re aeration ; lids were open at the bottom of the flumes to allow venting of excess heat. Grazing Impacts Three concurrent biomass treatments were evaluated using a fractional factorial design using two initial algae biomass levels and two snail densities which along with ungrazed controls, were subjected to three DO levels (Fig. 4 1). Biomass treatment 1 (BT 1) was located at the head of the flume, using t iles that had been grown in the


89 mesocosms for 1.5 months prior to the experiment (mean = 26.6 g m 2 dry mass). This treatment contained 10 snails per enclosure (approximately 0.5 grams wet weight each) yielding a density of 333 g m 2 snail biomass on a we t weight bases The second treatment (BT 2) contained a lower initial algal biomass ( mean = 6.0 g m 2 dry mass, grown for 2 weeks prior to experi mental initiation) and the same snail biomass. The third treatment (BT 3) contained the same low algae biomass as BT 2 ( mean = 5.8 g m 2 2 weeks of growth) and 10 snails, but double the enclosed area, creating a lower snail density of 167 g m 2 Using an empirical conversion for Elimia wet:dry weight ( y = 0.64x 0.11, R = 0.98 ), we divided gastropod dry mass by algae dry mass to get the following gastropod to algae biomass ratios: BT 1 = 8, BT 2 = 36, BT 3 = 16. At the conclusion of the experiment, total algal export (bioturbation) was collected from the downstream end of each BT 1 section by removing the mesh p artition and using tweezers to collect biomass which was then dried and weighed. Gastropod Behavior and S urvival Grazing activity, emergence, torpor, and mortality were monitored 2 3 times throughout the day for each treatment Data are reported as a prop ortion of individuals exhibiting each behavior (K olar and Rahel 1993) Snails were considered actively grazing if their tentacles were visible and the ir head was observed sweeping back and forth across the substrate. Snails breaking the water surface were considered emerging, and those with their opercu lum closed, or unmoving ( but attached to the surface ) were considered inert. Inert snails suspected of being dead were noted, and if they did not move from their location in 24 hours, were recorded and removed from the flumes.


90 Sample Collection and P roce ssing G astropods were collected from Blue Spring s in Gilchrist County. Elimia were added to the base of the flumes (segregated from algae growth tiles) and acclimated at high DO levels for 24 hours prior to being distributed among the treatments at the sta rt of the experiment on the morning of 13 June 2012 DO was decreased slowly to one of the three experimental levels (0.5, 1.5, and 5.5 mg L 1 ) over a 2 hour period, kept at these levels for 48 hours and then turned off following tile sampling on 15 June 20 12; at that time, DO levels return ed to 5.5 6.5 mg L 1 One week after the 2 day low DO pulse was ended (on 21 June 2012 ) the second set of tiles was collected. Each collection involved obtaining one tile from each compartment in each of the three repl icate flumes. Tiles were placed in re seal able bags, kept in the dark and on ice, and processed within 72 hours. A lgae biomass was scraped off the tile with a razor, diluted with DI to a volume of 200 mL, subsampled by half to allow successful filtration o f high biomass samples and filtered through pre weighed W h atman GF/F 0.45 M filters. Filters were then dried at 60C for 24 hours and weighed subtracting the filter weight to obtain the dry mass (DM) Data A nalysis Three metrics were used to analyze gra zer effects on alga e responses: 1) dry mass (DM), 2) percent removal (%R), and 3) removal rate (RR). DM is a direct measure of dr y algal biomass, and %R is a measure of grazing efficiency in comparison to the ungrazed control: %R = (1 DM G /DM U )*100 (4 1 ) where DM G is algal dry mass in the grazed treatment and DM U is algal dry mass in the ungrazed control. Finally, RR denotes snail specific removal rate computed as:


91 RR = U G ) / ( t N ) (4 2 ) w here DM U and DM G are as in Eq. 1, but ev aluated with respect to initia l and final biomass estimates, t is the number of days between initial and final estimates and N refers to the number of live snails at the beginning of the time step T o assess the overall effect of DO levels on algal accumu lation, we used factorial ANOVA, specifically evaluating the interactive effects of DO and snail density on algal biomass for each initial biomass treatment and for biomass export at the final time step ( 21 June 2012 ). ANOVA was conducted for RR and %R ove r both time steps ( 13 15 June and 15 21 June ) to assess DO treatment effects following the 2 day hypoxia stress, and also after a week of recovery. The %R variable allows comparison of relative grazer impact across initial algal biomass treatments (Hillebrand 2009) ANOVA was used to assess differential impacts of high versus low initial algae biomass (BT 1 vs. BT 2) and high versus low gastropod biomass (BT 2 vs. BT 3) as a function of D O, and analyzed separately for eff ects during and following the 2 day induced hypoxia. Pairwise treatment comparisons were evaluated using differences (HSD) multiple comparison tests Behavioral data were examined using ANOVA comparing DO effects on snail grazin g, emergence, torpor, and death at the end of the hypoxia pulse (15 June) and the experiment ( 21 June ). Results Dissolved oxygen treatments had a statistically significant effect on all algal response variables (Dry Mas s, % Removal, and Removal Rate) within each biomass treatment, with the exception of RR for B T 3 (Table 4 1) Average algal dry mass (on 21 June ) was significantly lower in all snail treatments than in snail free controls H igh DO treatments were signific a ntly lower than medium and low DO treatments, which were


92 not significantly different from each other (Fig. 4 3 ) Th e same pattern was observed for %R, which was greater in the high DO treatment than in the medium and low DO treatments ; the latter were not significantly different from each other. R emoval rate in the medium DO treatment was intermediate between high and low DO but was not significantly different from either (Fig. 4 5 ) While the low DO treatment had the highest mortality, snails in both hy poxic treatments had significantly high er mortality rates than in the high DO treatments (Fig. 4 6 ). Additionally, snails in the hypoxic treatments were more inert, were observed breaking the surface far more frequently, and were observed actively grazing less frequently. DO also significantly affected the per capita grazing rate for BT 1 and BT 2, but not BT 3 (Table 4 1). Mean e xport rates were higher in the higher DO treatments, but these differences were not statistically significant; the presence of sn ails w as the only factor that significantly affected biomass export (Table 4 1, Fig. 4 4 ). I nitial algae biomass and gastropod density did not influence the percent removal of algae (Table 4 2). The comparison of BT 1 vs. BT 2 (high initial algae vs. low initial algae, same snail density) showed that BT 2 had a significantly higher percent removal on 6/15 but the two biomass treatments were no long er significantly different by 21 June The percent of algae removed by high vs. low gastropod density on low initial algal biomass showed no significant differences on either date. During the two days of hypoxic stress, behavioral metrics of grazer health (active grazing, breaking the water surface, torpor, and mortality) were significantly different between the high DO and both of the hypoxic treatments, though low and medium DO were not significantly different from each other (Fig. 4 6 ) D uring the subsequent


93 recovery week in which high DO was restored to all treatments s ome indic ators of grazer health and cap acity to control algae showed improvement d However, while medium DO treatments recovered (showing significantly higher % grazing and lower % inert and lower death rates ), low DO treatments did not. Discussion Impacts of DO on G razing P otential This stu dy supports the hypothesis that DO can strongly influence the accumulation of algal biomass through both lethal and sublethal effects on grazers In the high DO treatments of this study, t he percent removal of algal biomass approximated global averages ho wever removal fell far belo w average in the medium and low DO treatments. The high DO treatments ranged from 60 74% removal across all three biomass treatments by 21 June (no significant differences among biomass treatments). Our observation of a consisten t percent age of algae biomass removed despite large differences in gastropod:algae dry mass ratios ( ranging from 8 to 36), is congruous with multiple meta analyses of herbivore removal experiments which found that h erbivores removed an average of 59 79% of annual algae biomass, with no significant differences among ecosystem types or productivity gradients (Cyr and Pace 1993, Hillebrand 2009) Notably, h owever, the medium and low DO treatments in this study produced much lower removal percentages ranging from 14 47% removal in medium DO (1.5 mg L 1 ), and 1 15% in low DO (0.5 mg L 1 ) by the end of the experiment. Multiple mechanisms may explain reduced grazing effects in low DO treatments, includ ing snail mortality as well as changes in behavior inhibiting active grazing and biot urbation. This suggests that the native gastropod Elimia floridensis exerts typical grazer effects on algal accumulation under high DO circumstances but


94 that these impact s are dramatically reduced to near negligible removal under DO stress Because gastro pods are the dominant invertebrate grazers, and E. floridensis is often the dominant gastropod, this provides strong mechanistic evidence that hypoxia limits the potential for native biological control of algal blooms in Florida spring systems Lethal and Sublethal M echanisms of Hypoxia Multiple concurrent mechanisms appear to explain declining control of algae under hypoxic stress D irect mortality was a main factor with 70% mortality in the 0.5 mg L 1 and 52% mortality in the 1.5 mg L 1 treatment over the 8 day study. This strongly supports the overarching hypothesis that Elimia floridensis grazing pressure is sensitive to hypoxia We note, however, that 12% mortality in the 5.5 mg L 1 treatment indicates other unidentified factors that influence surv ival in these mesocosms. These results are consistent with field experiments that show a strong, but variable effect of DO on mortality (Dormsjo 2008 ; Liebowitz C hapter 3 ) Meta analyses of lethal and sublethal hypoxia levels indicate that gastropods are among the most hypoxia tolerant taxa, with mean LC 50 levels of 0.89 0.11 mg L 1 DO (Vaquer Sunyer and Duarte 2008) with documented 28 day LC 50 values of 0.79 mg L 1 (Kapper and Stickle 1987) and 0.51 mg L 1 (Das and Stickle 1993) While these studies employed different methodo logies and are therefore not directly comparable, the near 50% mortality at 1.5 mg L 1 in this study suggests that Elimia may be more sensitive than other tested gastropods. However, extrapolation of laboratory DO responses to field behavior can be difficu lt due to the high variability of DO in the field, as well as co mpensatory mechanisms for short term hypoxic exposures (Berg and Ockelmann 1959, Hanley and Ultsch 1999, Garvey 2007)


95 Sublethal impacts can be subtle and difficult to detect, bu t have strong effects by reshaping the trophic interactions and energy flows in a system (Breitburg et al. 1997, Evans White and Lamberti 2009, Sturdivant 2011) The behavioral responses that we observed in response to low DO included snails spending more time at the air water surface or in torpor, and less time ac tively grazing Together, these behavioral adjustments help explain lower per capita grazing rate s Elimia exhibited behavioral modification s that reallocate energy from feeding to basic survival strategies, as was seen in the gastropod genus Nassarius (Cheung et al. 2008) These strategies may aid in short term survival, but could eventually lead to lower fecundity and population growth and indirectly to the accumulation of algal biomass However, despite behavioral modifications, 48 % mortality of Elimia in response to a two day pulse at 1.5 mg L 1 DO indicates that the threshold for sub lethal effects must be higher than that. E ffects of Grazer:A lgae Biomass R atios The effects of contrasting initial biomass ratios were smaller than expected suggesting that both algal density and grazer density had little impact on potential biomass removal within the ranges we tested While this result is consistent with other studies that show consistent grazer effects across wide productivity gra dients (Feminella and Hawkins 1995, Hillebrand 2009) it may also be due to low biomass accumulation in the flumes overall which limited differentiation among biomass treatments We did observe that algal percentage removal was higher in BT 2 than BT 1 at the en d of the 2 day hypoxia event ( 15 June ; Table 4 2), which likely follows from the higher snail t o algae biom ass ratio ( 36 versus 8 ). However, by 21 June the difference had disappeared. There was no evidence of herbivore escape densities of algae, which would have been manifest as a persistent low percent age of algal biomass grazed in BT


96 1 where algae biomass st arted with 27 g m 2 dry mass We note, h owever that this biomass is well threshold of 50 g m 2 ash free dry mass As a result this experiment was unable test the effects of what would be considered large bloom formations. Also, t he effect of high versus lower gastropod biomass was not significant within the range of gastropod densities tested (Table 4 2). While in situ experiments generally find positive relationships between snail density and algal consumption lab experim ents often find no effects or negative effects of snail densities on algal consumption Feminella and Hawkins (1995) speculated that t his may be a result of overcrowding in laboratory treatments though the range of grazer densities used in this laboratory experiment corresponded with the ranges in the in situ experiments in Chapter 3 in which higher grazer densities sh owed higher algal consumption Recovery prings The week of recovery period saw modest increases in grazing behavior and less surfacing in the 1.5 mg L 1 DO treatments suggesting slight recovery, however the percent of algae removed from each biomass treatment did not increase significantly in the hypoxia treatments after the recovery week So although snail behavior showed fewer signs of stress their capacity to graze efficiently did not fully return Additionally the 0.5 mg L 1 DO treatments did not exhibit si gns of recovery, and snails continued to perish, suggesting that even a two day pulse of severe hypoxia can have lasting population level effects. Due to time constraints, we were unable to run multiple iterations of the hypoxia trial to identify sublethal DO levels between 1.5 and 5.5 mg L 1 or cross those levels with varying pulse durations, therefore further studies of longer


97 duration will be required to determine environmental DO needs for snail popula tions to thrive and maintain high grazing capacitie s.


98 Table 4 1. ANOVA table of four response variables (DM, %R, RR, and Export) by experimental treatments (DO, Date, Snail presence) for each Biomass Treatment. Significant effects show differences between treatment levels or univariate effects, and significant interaction effects are labeled in Figure 4 2 and 4 4. Response Variable Effect F p Among factor diff. (Tukey HSD) F P Among factor diff. (Tukey HSD) F P Among factor diff. (Tukey HSD) BT 1: High Algae, High Snails BT 2: Low Algae, High Snails BT 3: Low Algae, Low Snails Algae Intercept 378.16 <0.0001 556.68 <0.0001 588.49 <0.0001 Dry Mass Snails 15.1 0.002 0 > 1 43.42 <0.0001 0 > 1 15.61 <0.0001 0 > 1 DO 3.56 0.061 13.59 0.001 1,2 > 3 16 0.002 1,2 > 3 Snails*DO 5.85 0.017 5.55 0.02 3.07 0.084 % Intercept 56.15 <0.0001 94.52 <0.0001 19.13 0.001 Removed Date 3.29 0.095 2.64 0.13 3 0.109 DO 16.72 <0.0001 3 > 2,1 19.83 <0.0001 3 > 2,1 7.48 0.008 3 > 1,2 Date*DO 4.05 0.045 0.84 0.457 0.44 0.656 Removal Intercept 22.45 <0.0001 24.03 <0.0001 4.37 0.059 Rate Date 5.57 0.036 1 > 2 8.91 0.011 1 > 2 0.21 0.655 DO 4.22 0.041 3 > 1 6.68 0.011 3 > 1 1.44 0.275 Date*DO 0.06 0.945 2.9 0. 094 1.24 0.325 Export Intercept 25.22 <0.0001 Snails 7.04 0.021 1 > 0 DO 0.61 0.560 Snails*DO 0.52 0.609


99 Table 4 2. ANOVA comparisons of Percent Removal (%R) by biomass treatments categorized by d ate and DO The column on the left shows high initial algae (BT 1; mean = 26.6 g m 2 algae DM) vs. low initial algae (BT 2; mean = 5.9 g m 2 DM algae) during hypoxic stress ( 1 5 June ) and after a week of reco very ( 21 June ), with high snail biomass (333 g m 2 ) in both treatments. The column on the right compares biomass treatments with high snail biomass (BT 2) versus low snail biomass (BT 3, 167 g m 2 snails), with low algae biomass initial conditions in both treatments. Percent Removal (%R) Effect F p Among factor diff. (Tukey HSD) F P Among factor diff. (Tukey HSD) BT 1 vs. BT 2 BT 2 vs. BT 3 6/15 Intercept 79.9 <0.001 19.5 < 0.001 BT 4.9 0.047 2 > 1 3.4 0.09 DO 17.2 <0.001 3 > 2, 1 7.2 0.009 3 > 2,1 BT*DO 2.9 0.092 0.85 0.45 6/21 Intercept 77.2 <0.001 97.6 <0.001 BT 2.3 0.16 3.5 0.09 DO 20.5 <0.001 3 > 2 > 1 19.4 <0.001 3 > 2,1 BT*DO 0.5 0.63 2.0 0.18

PAGE 100

100 Figure 4 1. Model of the artificial stream mesocosm and experimental design. Each vertica l rectangle represents one of nine flumes, each with a vertical partition separating snail treatments (1) from snail free controls (0). Biomass treatments (BT 1 3) are aligned vertically, separated by mesh barriers indicated by dashed lines. Water flow fol lows the direction of the arrows and feeds the high medium and low DO treatments. The dry biomass of algae was 27 g m 2 2 T he dry bio mass ratios of gastropods (converted from the wet weights via em pirically derived conversion factor of 0.64x 0.11 ) to algae for each biom a ss treatment are: BT 1 = 8, BT 2 = 36, and BT 3 = 18 (see text).

PAGE 101

101 A) B) Figure 4 2. Photo of flumes for A) three high DO treatment replicates, and B) three low DO treat ment replicates on 15 May 2011, showing visual impacts of snails (black dots) versus no snails on algae standing crop. Treatment key is shown in figure 4 1. Note the effects of snails on algal cover (darker brown = more algae) in high DO treatments, but no t in low DO treatments.

PAGE 102

102 F igure 4 3 Me an algae dry mass ( standard err or) categorized by DO treatment and presence (snail biomass in g m 2 ) or absence o f snails for biomass treatments with varying ratios of gastropod:algae biomass. A) BT 1 (biomass ratio = 8) B) BT 2 (biomass ratio = 36) and C) BT 3 (biomass ratio = 18) on the final collection day, 21 June 20 12. Significant interactive effects within each biomass treatment marked on the graph ) and univariate effects per treatment are shown in Table 4 1. 0 10 20 30 40 50 60 0.5 1.5 5.5 BT 1, Ratio = 8 Algae Dry Mass g m 2 0 g snails 333 g Snails 0 2 4 6 8 10 12 14 16 0.5 1.5 5.5 BT 2, Ratio = 36 Algae Dry Mass g m 2 0 g snails 333 g snails 0 2 4 6 8 10 12 14 16 0.5 1.5 5.5 BT 3, Ratio = 18 Algae Dry Mass g m 2 DO treatment (mg/L) 0 g snails 167 g snails a a a a b a,b a a a,b a,b b,c c A) B) C)

PAGE 103

103 Figure 4 4 Me an algae dry mass ( standard error) categorized by DO and presence (333 g m 2 ) or absence o f snails, for biomass export collected downstream of BT 1 on the final collecti on day, 21 June 20 12. Interactive effects were not significant, and univariate effects per treatment are shown in Table 4 1. 0 5 10 15 20 25 30 0.5 1.5 5.5 Export Algae Dry Mass g m 2 DO Treatment (mg/L) 0 g snails 333 g snails

PAGE 104

104 Figure 4 5 Mean Perce nt Removal and Removal Rate ( SE ), separated by B iomass T reatment ( (BT ) and corresponding gastropod: algae biomass ratio (Ratio) ) cat egorized by collection date ( 15 June and 21 June 20 12) and DO level < 0.05) are marked on graph, and univariate effects are shown in Table 4 1. -25 -5 15 35 55 75 95 0.5 1.5 5.5 BT 3, Ratio = 18 Percent Removal Dissolved Oxygen (mg/L) 15-Jun 21-Jun -25 -5 15 35 55 75 95 0.5 1.5 5.5 BT 1, Ratio = 8 Percnt removal 15-Jun 21-Jun -0.3 -0.1 0.2 0.4 0.6 0.8 1.0 1.2 0.5 1.5 5.5 Removal Rate (g m 2 day 1 ) 15-Jun 21-Jun -25 -5 15 35 55 75 95 0.5 1.5 5.5 BT 2, Ratio = 36 Percent Removal 15-Jun 21-Jun -0.1 0 0.1 0.2 0.3 0.4 0.5 1.5 5.5 Removal Rate (g m 2 day 1 ) 15-Jun 21-Jun -0.1 0.0 0.1 0.2 0.3 0.5 1.5 5.5 Removal Rate (g m 2 day 1 ) Dissolved Oxygen (mg/L) 15-Jun 21-Jun a a,b b b b b

PAGE 105

105 Figure 4 6 Means s tandard error for behavioral data averaged by day of the experiment ( 13 21 June ): % Grazing, % Breaking the s urface, % Inert, and % Living. ANOVA for DATE (during hypoxia on 15 June an d after a week of recovery on 21 June ) by DO treatment for each behavio ral variable: A) % Grazing (F(2, 128) = 6.9, p = 0.002), B) % Breaking the surface (F(2, 128)=11.5, p < 0.001), C) % Inert (F(2, 128)=5.8, p=.004), and D) % Live (F(2, 128)=10.2, p < 0.01). Grey bar indicates duration of hypoxia treatments. A) B) C) D)

PAGE 106

106 CHAPTER 5 C ON CLUSION Grazer Biomass T hresholds and Alternative S tates This study provides multiple lines of evidence to support the hypothesis that high densities of grazers have the potential to inhibit algal blooms in Florida spring s, but that they are less able to reduce algal biomass once a bloom has formed. The observational field study found a potential grazer biomass threshold at about 30 g m 2 of wet weight o f gastropods below which algal biomass was uniformly high, and above w hich algae could be high or low ; above 4 2 8 g m 2 gastropod wet weight, algae were uniformly low. This is a wider set of threshold s than were found in the in situ studies, in which ~ 100 g m 2 of Elimia were needed to maintain low algal biomass when starting from initially low algae co nditions, and > 330 g m 2 Elimia were necessary to reduce a pre existing algal bloom. Although there is an apparent discrepancy in the threshold values, it is common for surveys and experimental findings to arrive at different values, potentially due to differenc es in temporal scale historical contingencies, and the myriad interactions that occur in the field but which are reduced in controlled experiments (Leibold et al. 1997) The bimodal residuals from the field surveys along with the different grazer densities needed to d ecrease algae in high versus low algal conditions in the in situ experiments suggest the presence of an herbiv ore escape de nsity and alternative states. While true alternative stable states are difficult to demonstrate definitively (Schroder et al. 2005), these two lines of evidence build a strong case that the s ystem is hysteretic at the least The implication for restoration is that even if gastropod populations were restored to high levels areas with high algal standing stocks may not

PAGE 107

107 revert to lo w algal states unless the syst em is reset in some manner, making the system resistant to switching back to an herbivore dominated state. Additionally, as there are few locations that currently have the high levels of gastropods that demonstrated control on algal accumulation, the causes of grazer declines must be well understood before attempting biological restorations. Gastropod C onservation Although most attention to the global biodiversity extinction cris i s focuses on tropical eco systems or megafauna declines in less charismatic groups such as gastropods can have large impacts ngs are specifically listed as hotspot s of global hydrobiid diversity with 84 species of hydrobiids (43 of which are endemic), as well more than 100 species of the genus Elim ia (Brown et al. 2008a, Strong et al. 2008) a genus which appear s to be particularly important in controlling algal proliferation i n Florida s springs (Chapter 2) G astropod populations can be a dominant invertebrate presence in springs and streams o f the southeastern US, but their population s are declining severely throughout the region (Lydeard et al. 2004, Brown et al. 2008a) T here is little documented historic data about the abundance or declines of gastropods in Florida however declin ing populations in Ichetucknee S prings (DuToit 1979) and anecdotal reports from biologists and park managers suggest that Florida is experiencing similar trends to those noted throughout the region. T herefore causes of gastropod declines should be d iscerned and addressed. As there have been noted changes in DO levels in the springs, with many sites dipping below 1 mg L 1 (Heffernan et al. 2010b) t his study tested the effects of hypoxia on Elimia mortality and grazing rate s to see how such changes may impact the ecosystem. We found that while there was a weak correlation between gastropods and

PAGE 108

108 DO c oncentrations in the field, our experimental studies showed high mortality and reduced grazing under hypoxic condi tions. The weak relationships in the field may have been due in part to the fact that DO is not constant; surveys such as these necessarily mu st rely on snapshots of environme n tal conditions, which may miss pulse events that can restructure ecosystems but leave no trace of the stressor itself. Additionally, sublethal effects can cause asynchrony in the biomass signal in response to DO stress ma king trends difficult to discern G astropods have higher tolerance for hypoxia than most invertebrates with LC 50 values as low as 0.5 mg L 1 (Vaquer Sunyer and D uarte 2008) however we found severe mortality and sublethal effects at and below 1.5 mg L 1 This makes DO an important factor for management attention. H ypoxia explains a portion of gastropod distributions in the f ield, yet additional factors are li kel y at play. The multivariate models found a variety of ex pected factors predicting gastropod distribution, such as light and flow, but there still remained a measure of unexplained variability. Though beyond the realm of this study, we suggest that the e merging body of work in the field of aquatic contaminants such as fungicides, pesticides, pharmaceuticals, and industrial pollutants (Fleeger et al. 2003, Phelps et al. 2006, Brown et al. 2008b, Evans White and Lamberti 2009, McMahon et al. 2012) may help explain reg ional gastropod declines and constitute an important focus for future research. Additionally, t he habitat metrics of flow and sediment regimes have been shown to severely impact g astropod distribution (Stewart and Garcia 2002), as have invasive species (Ri ley et al. 2008) and salinity (James et a l. 2003, Ramakrishnan 2007) All these factors together help define the habitat needs of gastropod populations,

PAGE 109

109 to allow them to thrive and maintain their roles as powerful drivers of healthy ecosystems.

PAGE 110

110 LIST OF REFERENCES Alstyne, K. L. Van, J. M. Ehlig and S. L. Whitman. 1999. Feeding preferences for juvenile and adult algae depend on algal stage and herbivore species. Mar Ecol Prog Ser 180:179 185. APHA. 2005. Standard Methods for the Examination of Water and Wastewater21st Editi. American Public Heal th Association, Washington, D.C. Baird, D., R. R. Christian, C. H. Peterson, and G. A. Johnson. 2004. Consequences of Hypoxia on Estuarine Ecosystem Function: Energy Diversion from Consumers to Microbes. Ecological Society of America 14:805 822. Barbour, M T., J. Gerritsen, B. D. Snyder, and J. B. Stribling. 1999. Rapid bioassessment protocols for use in wadeable streams and rivers: periphyton, benthic macroinvertebrates, and fish. Washington, D.C. Barnosky, A. D., E. a Hadly, J. Bascompte, E. L. Berlow, J. H. Brown, M. Fortelius, W. M. Getz, J. Harte, A. Hastings, P. a Marquet, N. D. Martinez, A. Mooers, P. Roopnarine, G. Vermeij, J. W. Williams, R. Gillespie, J. Kitzes, C. Marshall, N. Matzke, D. P. Mindell, E. Revilla, and A. B. Smith. 2012. Approachin g a state shift 8. Baum, J. K., and B. Worm. 2009. Cascading top down effects of changing oceanic predator abundances. Journal of Animal Ecology:699 714. Beisner, B. E., D. T. Haydon, and K. Cuddington. 2003. Alternative Stable States in Ecology. Frontiers in Ecology and the Environment 1:376 382. Berg, K., and K. Ockelmann. 1959. The respiration of freshwater snails. Journal of Experimental Biology 36:690 708. Biggs, B. J. F. 2000. Eutrophication of Streams and Rivers: D issolved Nutrient Chlorophyll Relationships for Benthic Algae. Journal of the North American Benthological Society 19:17. Bonn, M. A., and F. W. Bell. 2003. Economic Impact of Selected Florida Springsm on Sur rounding Local Areas. Pp 1 99. Tallahassee, FL. Borer, E. T., E. W. Seabloom, J. B. Shurin, K. E. Anderson, C. A. Blanchette, B. Broitman, S. D. Cooper, and B. S. Halpern. 2005. What Determines the Strength of a Trophic Cascade? Ecology 86:528 537.

PAGE 111

111 Braun Blanquet. 1932. Plant sociology. The study of pla nt communities. Page 439 Fuller, G.D., Conard, H.S. (Eds.), Plant Sociology: The Study of Plant Communities. McGraw Hill, New York. Reprint: Lubrecht & Cramer, Ltd., Forestburgh, NY Breitburg, D. L., T. Loher, C. A. Pacey, and A. Gerstein. 1997. Varying E ffects of Low Dissolved Oxygen on Trophic Interactions in an Estuarine Food Web. Ecological Monographs 67:489 507. Brown, K. M., B. Lang, and K. E. Perez. 2008a. The conservation ecology of North American pleurocerid and hydrobiid gastropods. Journal of th e North American Benthological Society 27:484 495. Brown, M. T., K. C. Reiss, M. J. Cohen, J. M. Evans, T. K. Frazer, C. A. Jacoby, and E. J. Phlips. 2008b. Summary and Synthesis of the Available Literature on the Effects of Nutrients on Spring Organisms a nd Systems. Pages 1 376. Gainesville FL Camargo, J. A., A. Alonso, and A. Salamanca. 2005. Nitrate toxcity to aquatic animals: a review with new data for freshwater invertebrates. Chemospere 58:1255 1267. Carpenter, S. R., D. L. Christensen, J. J. Cole, K. L. Cottingham, J. R. H. Xi He, J. F. Kitchell, S. E. Knight, and M. L. Pace. 1995. Biological Control of Eutrophication in Lakes. Environmental Science & Technology 29:784 786. Cattaneo, A., and B. Mousseau. 1995. Empirical analysis of the removal rate of periphyton by grazers. Oecologia 103:249 254. Chapman, G. 1986. Ambient Water Quality Criteria for Dis solved Oxygen. Pp 1 46 US EPA Washington, D.C. Cheung, S. G., H. Y. Chan, C. C. Liu, and P. K. S. Shin. 2008. Effect of prolonged hypoxia on food cons umption, respiration, growth and reproduction in marine scavenging gastropod Nassarius festivus. Marine Pollution Bulletin 57:280 286. Conley, D. J., H. W. Paerl, R. W. Howarth, D. F. Boesch, S. P. Seitzinger, K. E. Havens, C. Lancelot, and G. E. Likens. 2 009. Controlling eutrophication: nitrogen and phosphorus. Science 323:1014 5. Connolly, N., M. Crossland, and R. Pearson. 2004. Effect of low dissolved oxygen on survival, emergence, and drift of tropical stream macroinver tebrates. Journal of the North Ame rican Benthological Society 23:251 270. Cyr, H., and M. Pace. 1993. Magnitude and patterns of herbivory in aquatic and terrestrial ecosystems. Nature 361:148 149.

PAGE 112

112 Das, T., and W. Stickle. 1993. Sensitivity of crabs Callinectes sapidus and C. similis and th e gastropod Stramonita haemastoma to hypoxia and anoxia. Marine Ecology Progress Series 98:263 274. Davis, S. N., L. D. Cecil, M. Zreda, and S. Moysey. 2001. Chlorine 36, bromide, and the origin of spring water. Chemical Geology 179:3 16. Dent, C. L., G. S Cumming, and S. R. Carpenter. 2002. Multiple states in river and lake ecosystems. Philosophical transactions of the Royal Society of London. Series B, Biological sciences 357:635 45. Diaz, R. J. 2001. Overview of hypoxia around the world. Journal of envi ronmental quality 30:275 81. Diaz, R. J., and R. Rosenberg. 1995. Marine Benthic Hypoxia: A Review of its Ecological Effects and the Bahavioural Responses of Benthic Macrofauna. Oceanography and Marine Biolog y: an Annual Review 33:245 303. Dillon, R. T. 20 00. The Ecology of Freshwater Molluscs. Cambridge University Press, Cambridge, MA. 509 pp. Dodds, W. K. 2006. Eutrophication and trophic state in rivers and streams. Limnology and Oceanography 51:671 680. Dodds, W. K. 2007. Trophic state, eutrophication an d nutrient criteria in streams. Trends in ecology & evolution 22:669 76. Dormsjo, K. 2008. Oxygen mediated grazin g impacts in Florida springs. University of Florida. Gainesville, FL. DuToit, C. H. 1979. 1979. The Carrying Capacity of the Ichetucknee Spring s and River. University of Florida. Gainesville, FL. Elser, J. J., M. E. S. Bracken, E. E. Cleland, D. S. Gruner, W. S. Harpole, H. Hillebrand, J. T. Ngai, E. W. Seabloom, J. B. Shurin, and J. E. Smith. 2007. Global analysis of nitrogen and phosphorus limi tation of primary producers in freshwater, marine and terrestrial ecosystems. Ecology letters 10:1135 42. Estes, J. a, J. Terborgh, J. S. Brashares, M. E. Power, J. Berger, W. J. Bond, S. R. Carpenter, T. E. Essington, R. D. Holt, J. B. C. Jackson, R. J. M arquis, L. Oksanen, T. Oksanen, R. T. Paine, E. K. Pikitch, W. J. Ripple, S. a Sandin, M. Scheffer, T. W. Schoener, J. B. Shurin, A. R. E. Sinclair, M. E. Soul, R. Virtanen, and D. a Wardle. 2011. Trophic downgrading of planet Earth. Science 333:301 6.

PAGE 113

113 Ev ans White, M. a, and G. a Lamberti. 2009. Direct and indirect effects of a potential aquatic contaminant on grazer algae interactions. Environmental toxicology and chemistry / SETAC 28:418 26. Feminella, J. W., and C. P. Hawkins. 1995. Interactions between stream herbivores and periphyton: a quantitative analysis of past experiments. Journal of the North American Benthological Society 14:465 509. Fleeger, J. W., K. R. Carman, and R. M. Nisbet. 2003. Indirect effects of contaminants in aquatic ecosystems. Th e Science of the total environment 317:207 33. Forbes, T., and G. Lopez. 1990. The effect of food concentration, body size, and environmental oxygen tension on the growth of the deposit feeding polychaete, Capitella species 1. Limnology and Oceanography 35 Francoeur, S. N. 2001. Meta Analysis of Lotic Nutrient Amendment Experiments: Detecting and Quantifying Subtle Responses. Journal of the North American Benthological Society 20:358 368. Frydenborg, R. 2006. Water Quality Study of the Ichetucknee River. B ureau of Laboratories, Florida Department of Environmental Protection. Tallahassee, FL. Garvey, J. 2007. A hierarchical model for oxygen dynamics in streams. Canadian Journal of Fisheries and Aquatic Science 64:1816 1827. Gliwicz, Z. M. 1990. Why do cladoc erans fail to control algal blooms? Hydrobiologia 200 201:83 97. Gragnani, A., M. Scheffer, and S. Rinaldi. 1999. Top Down Control of Cyanobacteria: A Theoretical Analysis. The American Naturalist 153:59 72. Gruner, D. S., J. E. Smith, E. W. Seabloom, J. T Ngai, S. Sandin, H. Hillebrand, W. S. Harpole, J. J. Elser, E. E. Cleland, M. E. S. Bracken, E. T. Borer, and B. Bolker. 2008. A cross system synthesis of consumer and nutrient resource control on producer biomass. Ecology Letters 11:740 755. Hairston, N ., F. Smith, and L. Slobodkin. 1960. Community structure, population control, and competition. American Naturalist 94:421 425. Hanley, R., and G. Ultsch. 1999. Ambient oxygen tension, metabolic rate, and habitat selection in freshwater snails. Arch. Hydrob iol. 144:195 214. Hartigan, J. A., and P. M. Hartigan. 1985. The Dip Test of Unimodality. Annals of Statistics 13:70 84.

PAGE 114

114 Heck, K. L. J., and J. F. Valentine. 2007. The primacy of top down effects in shallow benthic ecosystems. Estuaries and Coasts 30:371 3 81. Heffernan, J. B. 2008. Wetlands as an Alternative Stable State in Streams. Ecology 89:1261 1271. Heffernan, J. B., M. J. Cohen, T. K. Frazer, R. G. Thomas, T. J. Rayfield, J. Gulley, J. B. Martin, J. J. Delfino, and W. D. Graham. 2010a. Hydrologic and biotic influences on nitrate removal in a subtropical spring fed river. Limnology and Oceanography 55:249 263. Heffernan, J. B., D. M. Liebowitz, T. K. Frazer, J. M. Evans, and M. J. Cohen. 2010b. Algal blooms and the nitrogen enrichment hypothesis in Flor alternatives and adaptive management. Ecological Applications 20:816 829. Hill, W. R. 1992. Food Limitation and Interspecific Competition in Snai Dominated Streams. Can. J. Fish. Aquat. Sci. 49. Hill, W. R., H. L. Boston, and A. D. Steinman. 1992. Grazers and Nutrients Simultaneously Limit Lotic Primary. Can. J. Fish. Aquar. Sci. 49:504 512. Hill, W. R. W., P. J. P. Mulholland, and E. R. Marzolf. 2001. Stream Ecosystem Responses to Forest Leaf Emergence in Spring. Ecology 82:2306. Hillebrand, H. 2002. Top down versus bottom up control of autotrophic biomass a meta analysis on experiments with periphyton. Journal of the North American Benthological Society 21:349 369. Hillebrand, H. 2005. Light regime and consumer control of autotro phic biomass. Journal of Ecology 93:758 769. Hillebrand, H. 2009. Meta analysis of Grazer Control of Periphyton Biomass Across Aquatic Ecosystems. J. Phycol. 45, 45:798 806. Hillebrand, H., D. S. Gruner, E. T. Borer, M. E. S. Bracken, E. E. Cleland, J. J. Elser, W. S. Harpole, J. T. Ngai, E. W. Seabloom, J. B. Shurin, and J. E. Smith. 2007. Consumer versus resource control of producer diversity depends on ecosystem type and producer community structure. Proceedings of the National Academy of Sciences of the United States of America 104:10904 9. Holomuzki, J. R., J. W. Feminella, and M. E. Power. 2010. Biotic Interactions in Freshwater Benthic Habitats. Journal of the North American Benthological Society 29:220 244.

PAGE 115

115 Hoyer, M., T. Frazer, and S. Notestein. 200 4. Vegetative characteristics of three low lying Florida coastal rivers in relation to flow, light, salinity and nutrients. Hydrobiologia 528:31 43. Hughes, A. R., K. J. Bando, L. F. Rodriguez, and S. L. Williams. 2004. Relative effects of grazers and nutr ients on seagrasses: a meta analysis approach. Marine ecology. Progress series 282:87 99. Huryn, A. D., M. W. Denny, and N. Carolina. 1997. A Biomechanical Hypothesis Explaining Upstream Movements by the Freshwater Snail Elimia. Functional Ecology 11:472 4 83. Huryn, A., J. Koebel, and A. Benke. 1994. Life history and longevity of the pleurocerid snail Elimia: a comparative study of eight populations. Journal of the North American Benthological Society 13 :4 James, K. R., B. Cant, and T. Ryan. 2003. Response s of freshwater biota to rising Journal of Botany 51:703 713. Johnson, P. D., and K. M. Brown. 1997. The Role of Current and Light in Explaining the Habitat Distribution of the Lotic Snail Elimia semicarinata (Say). Journal of the North American Benthological Society 16:545. Jrgensen, B. B., and K. Richardson. 1996. Eutrophication in Coastal and Marine Ecosystems. Coastal Estuarine Studies. AGU, Washington, D. C. Kapper, M ., and W. Stickle. 1987. Metabolic responses of the estuarine gastropod Thais haemastoma to hypoxia. Physiological zoology 60:159 173. Kolar, C., and F. Rahel. 1993. Interaction of a biotic factor (predator presence) and an abiotic factor (low oxygen) as a n influence on benthic invertebrate communities. Oecologia 95:210 219. Van de Koppel, J., P. P. M. J. Herman, P. Thoolen, C. C. H. R. Heip, and J. Van De Koppel. 2001. Do alternate stable states occur in natural ecosystems? Evidence from a tidal flat. Ecol ogy 82:3449 3461. Van de koppel, J., J. Huisman, R. Van der Wal, and H. Wolff. 1996. Patterns of herbivory along a gradient of primary productivity: an empirical and theoretical investigation. Ecology 77:736 745. Korpinen, S., V. Jormalainen, and J. Ikonen 2008. Selective consumption and facilitation by mesograzers in adult and colonizing macroalgal assemblages. Marine Biology 154:787 794.

PAGE 116

116 Kurz, R. C., P. Sinphay, W. E. Hershfeld, A. B. Krebs, A. T. Peery, D. C. Woithe, S. K. Notestein, T. K. Frazer, J. A. Hale, and S. R. Keller. 2003. Mapping and Monitoring Submerged Aquatic Vegetation in Ichetucknee Submitted to Kurz, R. C., D. C. Woithe, S. K. Notestein, T. K. Frazer, J. A. Hale, and S. R. Keller. 2004. Mapping and Monitoring Submerged Aquatic Vege tation in Ichetucknee Springs. Live Oak, FL. Lamberti, G. A., L. R. Ashkenas, S. V. Gregory, a nd A. D. Steinman. 1987. Effects of three herbivores on periphyton communities in laboratory streams. J. N. Am. Benthol. Soc. 6:92 104. Lamberti, G. A., S. V Gregory, L. R. Ashkenas, and A. D. Steinman. 1989. Productive Capacity of Periphyton as a Determin ant of Plant Herbivore Interactions in Streams. Ecology 70:1840 1856. Landman, M. J. M., M. V. D. M. R. Van Den Heuvel, and N. Ling. 2005. Relative sensitivities of common freshwater fish and invertebrates to acute hypoxia. New Zealand Journal of Marine an d Freshwater Research 39:1061 1067. Lawler, J. J., J. E. Aukema, J. B. Grant, B. S. Halpern, P. Kareiva, C. R. Nelson, K. Ohleth, J. D. Olden, M. A. Schlaepfer, B. R. Silliman, and P. Zaradic. 2006. year report card In a nutshel 4:473 480. Leibold, M. a., J. M. Chase, and, J. B. Shurin, and A. L. Downing. 1997. Species Turnover and the Regulation of Trophic Structure. Annual Review of Ecology and Systematics 28:467 494. Lodge, D., K. Brown, S. Klosiewski, R. Stein, A. Covich, B. Leathers, and B. Bronmark. 1987. Distribution of freshwater snails: spatial scale and the relative importance of physicochemical and biotic factors. Am Malacol Bull 5:73 84. Lotze, H. K., H. S. Lenihan, B. J. Bourque, R. H. Bradbur y, R. G. Cooke, M. C. Kay, S. M. Kidwell, M. X. Kirby, C. H. Peterson, and J. B. C. Jackson. 2006. Depletion, degradation, and recovery potential of estuaries and coastal seas. Science 312:1806 9. Lotze, H., and B. Worm. 2000. Variable and complementary ef fects of herbivores on different life stages of bloom forming macroalgae. Marine Ecology Progress Series 200:167 175. Lydeard, C., R. H. Cowie, W. F. Ponder, A. E. Bogan, P. Bouchet, S. a. Clark, K. S. Cummings, T. J. Frest, O. Gargominy, D. G. Herbert, R. Hershler, K. E. Perez, B. Roth, M. Seddon, E. E. Strong, and F. G. Thompson. 2004. The Global Decline of Nonmarine Mollusks. BioScience 54:321.

PAGE 117

117 Lysne, S. J., K. E. Perez, K. M. Brown, R. L. Minton, and J. D. Sides. 2008. A review of freshwater gastropod c onservation: challenges and opportunities. Journal of the North American Benthological Society 27:463 470. Malard, F., and F. Hervant. 1999. Oxygen supply and the adaptations of animals in groundwater. Freshwater Biology 41:1 30. Mallory, M. a., and J. S. Richardson. 2005. Complex interactions of light, nutrients and consumer density in a stream periphyton grazer (tailed frog tadpoles) system. Journal of Animal Ecology 74:1020 1028. Martin, J. B., and R. W. Dean. 2001. Exchange of water between conduits and matrix in the Floridan aquifer. Chemical Geology 179:145 165. Mattson, R. A., M. Lehmensiek, and E. F. Lowe. 2007. Nitrate Toxicity in Florida Springs and Spring Run Streams: A Review of the Literature and Its Implications. Pages 1 31. SJRWMD. Palatka, FL McMahon, T. a, N. T. Halstead, S. Johnson, T. R. Raffel, J. M. Romansic, P. W. Crumrine, and J. R. Rohr. 2012. Fungicide induced declines of freshwater biodiversity modify ecosystem functions and services. Ecology letters 15:714 22. Meinzer, O. E. 1928. Compressibility and elasticity of artesian aquifers. Econ. Geol. 23:263 291. Mulholland, P. J., A. D. Steinman, A. V Palumbo, J. W. Elwood, and D. B. Kirschtel. 1991. Role of Nutrient Cycling and Herbivory in Regulating Periphyton Communities in Laboratory Streams. Ecology 72:966 982. Neves, R. J., A. E. Bogan, J. D. Williams, S. A. Ahlstedt, and P. W. W. Hartfield. 1997. Status of Aquatic Mollusks in the Southeastern United States: A Downward Spiral of Diversity. Page 554 in G. W. Benz and D. E. Collins, e ditors. Aquatic Fauna in Peril: The Southeastern PerspectiveSpecial pu. Southeast Aquatic Research Institute, Lenz Design and Communications, Decatur, GA. Dynamics in a Woodla Ecology 64:1249 1265. Odum, H. 1957a. Trophic structure and productivity of Silver Springs, Florida. Ecological monographs 27:55 112. Odum, H. T. 1956. Primary Production in Flowing Waters. Limnology an d Oceanography 1:102 117.

PAGE 118

118 Odum, H. T. 1957b. Primary Production Measurements in Eleven Florida Springs and a Marine Turtle Grass Community. Limnology and Oceanography 2:85 97. Opsahl, R. W., T. Wellnitz, and N. LeRoy Poff. 2003. Current velocity and invert ebrate grazing regulate stream algae: results of an in situ electrical exclusion. Hydrobiologia 499:135 145. Osborne, J. 2010. Improving your data transformations: Applying the Box Cox transformation. Practical Assessment, Research & Evaluation 15. Phelps, G. G., S. J. Walsh, R. M. Gerwig, W. B. Tate, and B. G. G. Phelps. 2006. Characterization of the Hydrology Water Chemistry and Aquatic Communities of Selected Springs in the St Johns River Water Management District Florid a 20 04. Phlips, E. J., M. Cichra, F. J. Aldridge, J. Jembeck, J. Hendrickson, and R. Brody. 2000. Light Availability and Variations in Phytoplankton Standing Crops in a Nutrient Rich Blackwater River. Limnology and Oceanography 45:916 929. Poff, N. L., N. J. Voelz, J. Ward, and R. Lee. 1990. Algal Colonization under Four Experimentally Controlled Current Regimes in High Mountain Stream. Journal of the North American Benthological Society:303 318. Poff, N. L., T. Wellnitz, and J. B. Monroe. 2003. Redundancy among three herbivorous i nsects across an experimental current velocity gradient. Oecologia 134:262 269. Poore, A. G. B., A. H. Campbell, R. a. Coleman, G. J. Edgar, V. Jormalainen, P. L. Reynolds, E. E. Sotka, J. J. Stachowicz, R. B. Taylor, M. a. Vanderklift, and J. Emmett Duffy 2012. Global patterns in the impact of marine herbivores on benthic primary producers. Ecology Letters 15:912 922. Power, M. E. 1992. Top Down and Bottom Primacy. Ecology 73:733 746. Powles, H. 2000. Assessing and protecting endangered marine species. ICES Journal of Marine Science 57:669 676. Rabalais, N. N., R. E. Turner, and W. J. Wiseman. 2002. Gulf of Mexico Hypoxia, a.K.a. 263. Ramakrishnan, V. 2 007. Salinity, ph, temperature, desiccation and hypoxia tolerance in the invasive freshwater a pple snail Pomacea Insularum. University of Texas at Arlington.

PAGE 119

119 Riseng, A. C. M., M. J. Wiley, and R. J. Stevenson. 2004. Hydrologic Disturbance and Nutrient Effe Covariance Structure Analysis. Journal of the North American Benthological Society 23:309 326. Rosemond, A. D., P. J. Mulholland, and S. H. Brawley. 2000. Seasonally shifting limitation of str eam periphyton: response of algal populations and assemblage biomass and productivity to variation in light, nutrients, and herbivores. Can. J. Fish. Aquat. Sci 57:66 75. Rosemond, A. D., P. J. Mulholland, and J. W. Elwood. 1993. Top Down and Bottom Up Con trol of Stream Periphyton: Effects of Nutrients and Herbivores. Ecology 74:1264 1280. Rosenzweig, M. 1971. The paradox of enrichment: destabilization of exploitation ecosystems in ecological time. Science 171:385 387. Sagasti, A., L. C. Schaffner, and J. E Duffy. 2001. Effects of periodic hypoxia on mortality feeding and predation in an estuarine epifaunal community. Journal of Experimental Marine Biology and Ecology 258:257 283. Saloom, M. E., and R. S. Duncan. 2005. Low dissolved oxygen levels reduce a nti predation behaviours of the freshwater clam Corbicula fluminea. Freshwater biology 50:1233 1238. Sartory, D., and J. Grobbelaar. 1984. Extraction of chlorophyll a from freshwater phytoplankton for spectrophotometric analysis. Hydrobiologia 187. Scheffe r, M., E. H. Nes, M. Holmgren, and T. Hughes. 2008. Pulse Driven Loss of Top Down Control: The Critical Rate Hypothesis. Ecosystems 11:226 237. Schroder, A., L. Persson, A. M. D. A. De Roos, and A. Schrder. 2005. Direct experimental evidence for alternati ve stable states: a review. Oikos 110:3 19. Scott, T. M., G. H. Means, R. P. Meegan, R. C. Means, S. B. Upchurch, R. E. Copeland, J. Jones, T. Roberts, and A. Willet. 2004. Springs of Florida: Bulletin No. 66. USGS. Tallahassee FL Shurin, J., E. Borer, E. W. Seabloom, C. A. Blanchette, B. Broitman, S. D. Cooper, and B. Halpern. 2002. A cros s e cosystem comparison of the strength of trophic cascades. Ecology Letters 5:785 791. Steckbauer, a, C. M. Duarte, J. Carstensen, R. Vaquer Sunyer, and D. J. Conley. 2011. Ecosystem impacts of hypoxia: thresholds of hypoxia and pathways to recovery. Environmental Research Letters 6:025003.

PAGE 120

120 Steffen, W. L., and P. Tyson. 2001. Global change and the earth system: a planet under pressure., 4th edition. International Geosph ere Biosphere Programme. Steinman, A. D. 1996. Effects of Grazers on Freshwater Benthic Algae. Pages 341 373 in R. J. Stevenson, M. L. Bothwell, R. L. Lowe, and J. H. Thorp, editors. Algal Ecology. Academic Press, San Diego. Steinman, A. D., C. D. Mcint ire, S. V Gregory, G. A. Lamberti, and R. Ashkenas. 1987. Effects of herbivore type and density on taxonomic structure and physiognomy of algal assemblages in laboratory streams. Journal of the North American Benthological Society 6:175 188. Stevenson, R. J., A. Pinowska, A. Albertin, and J. Sickman. 2007. Ecological condition of algae and nutrients in florida springs: the synthesis report. Tallahassee FL Stevenson, R. J., A. Pinowska, and Y. Wang. 2004. Ecological Condition of Algae and Nut rients in Florida Springs. Tallahassee FL Stewart, T. W., and J. E. Garcia. 2002. Environmental Factors Causing Local Variation Density and Biomass of the Snail Leptoxis carinata,in Fishpond Creek, Virginia. The American Midland Naturalist 148:172 180. S tickle, W., M. Kapper, L. Liu, E. Gnaiger, and S. Wang. 1989. Metabolic adaptations of several species of crustaceans and molluscs to hypoxia: tolerance and microcalorimetric studies. The Biological Bulletin 177:303 312. Strong, E. E., O. Gargominy, W. F. Ponder, and P. Bouchet. 2008. Global diversity of 166. Sturdivant, S. K. 2011. The effects of hypoxia on macrobenthic production and function in the lower Rappahanno ck River, Chesape ake Bay, USA. The College of William & Mary in Virginia. Suding, K. N., and R. J. Hobbs. 2009. Threshold models in restoration and conservation: a developing framework. Trends in ecology & evolution 24:271 9. Tuchman, N. C., and R. J. Stevenson. 1991. Eff ects of Selective Grazing by Snails on Benthic Algal Succession. Journal of the North American Benthological Society 10:430 443. US EPA. 2004. Pesticide Industry Sales and Usage: 2000 and 2001 Market Estimates. Office of Pesticide Programs, Washington, DC. Van de Koppel, J., P. Herman, P. Thoolen, C. Heip. 2001. Do alternate stable states occur in natural ecosystems? Evidence from a tidal flat. Ecology 82 (12), 3449 3461.

PAGE 121

121 Van de Koppel, J., J. Huisman, R. van der Wal, H. Olff. 1996. Patterns of herbivory a long a prouductivity gradient: An empirical and theoretical investigation. Ecology 77(3), 736 745. Vaquer Sunyer, R., and C. M. Duarte. 2008. Thresholds of hypoxia for marine biodiversity. Proceedings of the National Academy of Sciences of the United State s of America 105:15452 7.

PAGE 122

122 BIOGRAPHICAL SKETCH Dina Liebowitz graduated from the University of Pennsylvania with a B.S. in Biology and Environmental Studies and a thesis exploring land use and fire patterns in Bolivia. After spending a year volunteering w ith human rights and environmental NGOs in Israel, she worked as a field biology coordinator for Harvard University's Department of Organismic an d Evolutionary Biology. In 2007 she completed an M.S. in wildlife ecology and c onservation from the University of Florida, with a concentration in tropical conservation and d evelopment, focusing on marine resource use patterns and stakeholder assessments for marine protected areas in the Bahamas. During that time, she had the opportunity to spend five weeks i n a Pa cif ic Islands with a field training course exploring community based natural resource management in the Solomon Islands. She was an NSF IGERT Fellow in the Adaptive Management of Water, Wetlands, and Watersheds program during her doctoral training, and now plans to combine her interdisciplinary experiences to work towards the goals of natural resource management and conservation.