<%BANNER%>

Habitat Selection and Foraging Success of Wading Birds in Impounded Wetlands in Florida


PAGE 1

HABITAT SELECTION AND FORAGI NG SUCCESS OF WADING BIRDS IN IMPOUNDED WETLANDS IN FLORIDA By ERIC DOUGLAS STOLEN A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2006

PAGE 2

Copyright 2006 by Eric Douglas Stolen

PAGE 3

This document is dedicated to my late father W. Douglas Stolen who believed in me and my late father-in-law Leslie Kinney who inspired me.

PAGE 4

iv ACKNOWLEDGMENTS First and foremost, I want to thank Dr. Ja ime Collazo for serving as my co-advisor, mentor, colleague, and friend. Jaime made th e process of working on this dissertation meaningful and helped me develop professi onally in many, many ways. Likewise, as my other co-advisor Dr. Franklin Percival was a constant source of help and encouragement. He also taught me a great deal about the philosophy part of the degree. I also had the pleasure of getting to know the other members of my comm ittee during various stages of my studies. Dr. Wiley Kitchens was an insp iring teacher and a solid source of advice, especially in the initial stages of planni ng and implementing my field work. Dr. Clay Montague was also a great teacher and truly made me think about how to solve problems in science. Dr. Peter Frederick was an inva luable source of knowledge about wading bird ecology and also a very practical guide to na vigating the process of earning my degree. His insightful comments on earlier draf ts greatly improved this manuscript. I was fortunate to have the help of ma ny very able field technicians who spent countless hours knee-deep in mud under the ba king sun. Special thanks go to Francisco Collazo, Alison Pevler, Elizabeth Labunsky, Ja mes Borgmeyer, Heather Eijanga and Jap Eijanga. Others also helped with fiel d work on many days, including Zoe Donaldson, Ashley Below, Ryan Woods, Geoff Carter, Donna Oddy, and Ron Brockmeyer, to name a few. Geoff Carter gave me valuable editori al advice and spent hours reading various drafts of this dissertation; I am extremely fo rtunate to share an office with such a great

PAGE 5

v naturalist and writer. Brean Duncan spent many hours helpin g with sticky GIS problems and was always ready to help no matter how busy he was. Dave Breininger got me started on wading birds and treated me as a co lleague even before I was one. Dave also has been very patient in allowing me to pursu e my research interests at the expense of publications documenting our monitoring program (a situation I will at last have the time to remedy). My bosses Carlton Hall, Ross Hinkle and Doug Britt were very supportive and encouraging throughout. My NASA manage rs, Kelly Gorman and Burt Summerfield were also very supportive of my work on wading bird foraging ecology and have the vision to see the value of knowledge about our natural resources. I am also grateful for the support and en couragement by the staff at the Merritt Island National Wildlife Refuge, especially Marc Epstein, Ron Height, and Ralph Loyd. My mother, Sandra Morgan, and my mo ther-in-law, Regina Kinney, helped innumerable times taking care of my children wh ile I was occupied at the computer. If it were not for that help I never could have fini shed writing. Finally, I want to deeply thank my wife Megan and my children Ethan St olen and Erin Stolen, whose support was irreplaceable and whose patience was in exhaustible. I did it all for them.

PAGE 6

vi TABLE OF CONTENTS Page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES...........................................................................................................viii LIST OF FIGURES.........................................................................................................xiii ABSTRACT.....................................................................................................................xvi CHAPTER 1 INTRODUCTION........................................................................................................1 2 FORAGING HABITAT SELECTION BY NESTING GREA T EGRETS AND SNOWY EGRETS......................................................................................................11 Methods......................................................................................................................13 Study Site.............................................................................................................13 Foraging Flight Observations..............................................................................14 Habitat Use Analysis...........................................................................................15 Results........................................................................................................................ .19 Foraging Flight Characteristics...........................................................................19 Habitat Use Analysis...........................................................................................21 Discussion...................................................................................................................29 3 THE DISTRIBUTION OF WADING BIRD PREY IN IMPOUNDED WETLANDS IN THE NORTHERN INDIAN RIVER LAGOON ESTUARY........33 Introduction.................................................................................................................33 Methods......................................................................................................................35 Study Site.............................................................................................................35 Prey Sampling.....................................................................................................38 Analysis...............................................................................................................40 Results........................................................................................................................ .43 Fixed-Station Sampling.......................................................................................43 Random-Site Sampling........................................................................................43 Discussion...................................................................................................................64

PAGE 7

vii 4 EFFECTS OF HABITAT STRUCTUR E AND PREY DISTRIBUTION ON WADING BIRD FORAGING HABITAT USE........................................................68 Introduction.................................................................................................................68 Methods......................................................................................................................72 Results........................................................................................................................ .83 Discussion.................................................................................................................103 Foraging Habitat Selection................................................................................103 Mixed-Species Foraging Aggregations.............................................................108 5 BENEFITS TO INDIVIDUALS IN PI SCIVOROUS WADING BIRD MIXEDSPECIES FORAGING AGGREGATIONS.............................................................110 Introduction...............................................................................................................110 Methods....................................................................................................................113 Study Site...........................................................................................................113 Foraging Behavior Observations.......................................................................114 Results.......................................................................................................................121 Prey Density at Used and Paired Random Sites................................................121 Behavior-Prey Sampling...................................................................................121 Discussion.................................................................................................................133 6 CONSERVATION IMPLICATIONS AND MANAGEMENT RECOMMENDATIONS..........................................................................................138 APPENDIX A WATER LEVELS WITHIN IMPOUNDMENTS AND INDIAN RIVER LAGOON ESTUARY MEASURED DU RING WETLANDS INITIATIVE.........145 B MODEL SELECTION RESULTS FO R INDIVIDUAL WADING BIRD SPECIES...................................................................................................................148 LIST OF REFERENCES.................................................................................................160 BIOGRAPHICAL SKETCH...........................................................................................174

PAGE 8

viii LIST OF TABLES Table page 2.1. Summary of the sizes of foraging aggrega tions joined by Great Egrets and Snowy Egrets followed from three colonies........................................................................23 2.2. Summary of flight distance, duration and speed for Great Egrets and Snowy Egrets followed from three colonies........................................................................23 2.3. Areas of Great Egret and Snowy Egret fora ging habitats within three flight-radius distances from three nesting colonies.......................................................................24 2.4. Comparison of the proportion of bird use versus the proportion of habitat availability, for three types of fora ging habitat by nesting Snowy Egrets...............25 2.5. Comparison of the proportion of bird use versus the proportion of habitat availability, for three types of fora ging habitat by nesting Great Egrets.................26 3.1. Number of points sampled during random -station fish sampling in unvegetated (unveg.) and vegetated (veg.) habita t during 5 seasons in 7 impoundments...........47 3.2. Occurrence of fish by species in throwtrap samples in 6 fixed-station sampling impoundments on KSC/MINWR.............................................................................47 3.3. Fixed-station fish sampling model sel ection results for GLM analysis with ln[1+fish density] as respons e variable and season..................................................48 3.4. ANOVA table for GLM with fixed-station ln(1+fish density) as response variable and explanatory variables season and location........................................................48 3.5. Parameter estimates for GLM with fixe d-station ln(1+fish density) as response variable and explanatory va riables season and location...........................................49 3.6. Occurrence of non-fish nekton in ra ndom-site throw-trap samples in 7 impoundments on KSC/MINWR.............................................................................49 3.7. Occurrence of fish by species in throwtrap samples in wetland habitat with and without emergent vegetation in 7 random-site impoundments................................50 3.8. Density of fish (individuals/m2) in wetland habitat without emergent vegetation in 7 random-site impoundments on KSC/MINWR......................................................51

PAGE 9

ix 3.9. Density of fish (individuals/m2) in wetland habitat with emergent vegetation in 7 random-site impoundments on KSC/MINWR.........................................................52 3.10. Biomass (g/m2) of fish in wetland habitat w ithout emergent vegetation in 7 random-site impoundments on KSC/MINWR.........................................................53 3.11. Biomass (g/m2) of fish in wetland habitat with emergent vegetation in 7 randomsite impoundments on KSC/MINWR......................................................................54 3.12. Mean length and mean biomass of indi vidual fish captured within random-site throw-trap samples by species..................................................................................55 3.13. Mean length of individual fish in 2 ha bitat types within random-site throw-trap samples, listed by species.........................................................................................55 3.14. Random-site model selection results for GL M analysis with ln[fish density+1] as response variable and habitat type...........................................................................56 3.15. ANOVA table for GLM with random-site ln (fish density+1) as the explanatory variable and habitat*season and habita t*impoundment interac tions included........56 3.16. Parameter estimates for GLM with random-site ln(fish density+1) as the explanatory variable and habitat*season and habitat*impoundment interactions...57 3.17. ANOVA table for GLM with random-site ln (fish density+1) as the explanatory variable and habitat*season, habita t*impoundment and season*impoundment......58 3.18. Parameter estimates for GLM with random-site ln(fish density+1) as the explanatory variable and habita t*season, habitat*impoundment and......................59 4.1. Models set for analysis of factors that effect wading bird density in foraging habitat on KSC/MINWR..........................................................................................78 4.2 Occurrence of wading birds during ground surveys of 9 impoundments on the KSC/MINWR, January July 2001.........................................................................86 4.3. Water depth at locations where fo raging wading birds were observed in impoundments during ground surveys of impoundments on KSC/MINWR...........87 4.4. All species of wading birds observed show ed selection for edge habitat over both vegetated and unvege tated habitats..........................................................................88 4.5. Information-theoretic model selection resu lts for GLM analysis with (wading bird density)1/2 as response variable, fish de nsity (P), water depth (D)...........................89 4.6. ANOVA table for GLM with (wading bird density)1/2 as response variable and including the fish dens ity*habitat interaction..........................................................90

PAGE 10

x 4.7. Parameter estimates for GLM with (wading bird density)1/2 as response variable and including the fish de nsity*habitat interaction....................................................90 5.1. Mean fish density and mean total biomass (with 95% confidence intervals) at sites used by foraging wading bird s and paired random sites........................................124 5.2. Mean fish density and mean total biomass (with 95% confidence intervals) at sites used by foraging individuals a nd groups of wading birds.....................................124 5.3. Mean length and mean mass (with 95% confid ence intervals) of fish in samples at sites used by wading birds and paired-unused sites...............................................124 5.4. Correlations between mean le ngth and mean mass of fish in samples at sites used by wading birds. The sample size was 120 for all comparisons...........................124 5.5. Information-theoretic model selection re sults for GLM analysis of Great Egret (captures / min)1/2 as response variable, prey density (PD) as covariate................125 5.6 ANOVA table for GLM of Great Egret (captures / min)1/2 as response variable with the binary group variab le (GB) as a factor.....................................................125 5.7 Parameter estimates for GLM of Great Egret (captures / min)1/2 with the binary group variable (GB) as a factor..............................................................................125 5.8 ANOVA table for GLM of Great Egret (captures / min)1/2 as response variable with prey density (PD) as covariat e and the binary group variable (GB)..............126 5.9. Parameter estimates for GLM of Great Egret (captures / min)1/2 as response variable with prey density (PD) as covariate and the binary group variable.........126 5.10 ANOVA table for GLM of Gr eat Egret (captures / min)1/2 as response variable with group category (GC) as a factor.....................................................................126 5.11 Parameter estimates for GLM of Great Egret (captures / min)1/2 as response variable with group category (GC) as a factor.......................................................127 5.12. Information-theoretic model selection re sults for GLM analysis of Snowy Egret (captures / min)1/2 as response variable, prey density (PD) as covariate................127 5.13 ANOVA table for GLM of Sn owy Egret (captures / min)1/2 as response variable with the binary group variab le (GB) as a factor.....................................................127 5.14 Parameter estimates for GLM of Snowy Egret (captures / min)1/2 with the binary group variable (GB) as a factor..............................................................................127 5.15. ANOVA table for GLM of Snowy Egret (captures / min)1/2 as response variable with prey density (PD) as covariat e and the binary group variable (GB)..............128

PAGE 11

xi 5.16. Parameter estimates for GLM of Snowy Egret (captures / min)1/2 as response variable, prey density (PD) as covari ate and the binary group variable (GB).......128 5.17. ANOVA table for GLM of Snowy Egret (captures / min)1/2 as response variable with the interaction between prey density (PD).....................................................128 5.18. Parameter estimates for GLM of Snowy Egret (captures / min)1/2 as response variable with the interacti on between prey density (PD).......................................129 5.19. Information-theoretic model selection re sults for GLM analysis of Tricolored Heron (captures / min)1/2 as response variable, prey density (PD).........................129 5.20. Summary of foraging rates of Great Egrets in various sized groups......................129 5.21. Summary of foraging rates of Snowy Egrets in various sized groups....................129 5.22. Summary of foraging rates of Tricolor ed Herons in various sized groups.............130 5.23. Summary of studies investigating the be nefits to wading birds of foraging in aggregations compared to foraging solitarily.........................................................137 B.1. Model selection results for GLM an alysis with (Gre at Egret density)1/2.................148 B.2. Model selection results for GLM analysis with (Snowy Egret density)1/2...............149 B.3. Model selection results for GLM anal ysis with (Tricolored Heron density)1/2........150 B.4. Model selection results for GLM an alysis with (Wh ite Ibis density)1/2...................151 B.5. ANOVA table for GLM of (Great Egret density)1/2 with management type as a factor and fish density as a covariate.....................................................................151 B.6. Parameter estimates for GL M of (Great Egret density)1/2 with management type as a factor and fish de nsity as a covariate..............................................................152 B.7. ANOVA table for GLM of (Great Egret density)1/2 with fish density as a covariate.................................................................................................................152 B.8. Parameter estimates for GL M of (Great Egret density)1/2 fish density as a covariate.................................................................................................................152 B.9. ANOVA table for GLM of (Great Egret density)1/2 with the interaction between prey density and habitat type..................................................................................153 B.10. Parameter estimates for GLM of (Great Egret density)1/2 with the interaction between prey density and habitat type...................................................................153 B.11. ANOVA table for GLM of (Great Egret density)1/2 with the interaction between prey density and management type........................................................................154

PAGE 12

xii B.12. Parameter estimates for GLM of (Great Egret density)1/2 with the interaction between prey density and management type..........................................................154 B.13. ANOVA table for GLM of (Great Egret density)1/2 with management type as a factor.......................................................................................................................155 B.14. Parameter estimates for GLM of (Great Egret density)1/2 with management type as a factor...............................................................................................................155 B.15. ANOVA table for GLM of (Great Egret density)1/2 with habitat type as a factor and fish density as a covariate................................................................................155 B.16. Parameter estimates for GLM of (Great Egret density)1/2 with habitat type as a factor and fish density as a covariate.....................................................................156 B.17. ANOVA table for GLM of (Great Egret density)1/2 with fish density and water depth as covariates.................................................................................................156 B.18. Parameter estimates for GLM of (Great Egret density)-2 with fish density and water depth as covariates........................................................................................156 B.19. ANOVA table for GLM of (Snowy Egret density)1/2 with the interaction between prey density and habitat type...................................................................157 B.20. Parameter estimates for GLM of (Snowy Egret density)1/2 with the interaction between prey density and habitat type...................................................................157 B.21. ANOVA table for GLM of (Tricolored Heron density)1/2 with the interaction between prey density and habitat type...................................................................158 B.22. Parameter estimates for GLM of (Tricolored Heron density)1/2 with the interaction between prey de nsity and habitat type.................................................158 B.23. ANOVA table for GLM of (White Ibis density)1/2 with the interaction between prey density and habitat type..................................................................................159 B.24. Parameter estimates for GL M of (White Ibis density)1/2 with the interaction between prey density and habitat type...................................................................159

PAGE 13

xiii LIST OF FIGURES Figure page 1.1. A conceptual model of the factors influe ncing the foraging success of individual wading birds in shallow wetlands..............................................................................9 1.2. Diagram depicting the relationship betw een the functional value of the northern Indian River Lagoon system for wading bird populations.......................................10 2.1. Foraging locations of Great Egrets (circl es) and Snowy Egrets (triangles) followed from three colonies in the north ern Indian River Lagoon, Florida..........................27 2.2. Great Egret and Snowy Egret foraging f light habitat resour ce selection ratios..........28 3.1. Map of study site showing location of study impoundments.....................................37 3.2. Frequency of 1-m2 throw trap samples (tosses) within 7 density categories..............60 3.3. Mean fish densities measured dur ing fixed-station fish sampling..............................60 3.4. Mean fish density was always greater for vegetated versus paired nearby unvegetated sites based on random-site sampling from 5 quarterly sampling.........61 3.5. Mean fish biomass was usually greater for vegetated versus paired nearby unvegetated sites......................................................................................................61 3.6. Frequency of 1-m2 throw trap samples (tosses) within 7 density categories in unvegetated and vegetated flooded habitat..............................................................62 3.7. Estimated marginal means of fish dens ity (back-transformed) in vegetated and unvegetated habitats within 4 im poundments on MINWR/KSC by season............62 3.8. Estimated marginal means of fish dens ity (back-transformed) in vegetated and unvegetated habitats within 4 impoundments on MINWR/KSC.............................63 4.1. Map of study site showing lo cation of 9 study impoundments..................................79 4.2. Aerial photographs of study impoundments showing sections used to record locations of foraging wading birds...........................................................................80 4.3. Aerial photographs of study impoundments showing sections used to record locations of foraging wading birds...........................................................................81

PAGE 14

xiv 4.4. Aerial photographs of study impoundments showing sections used to record locations of foraging wading birds...........................................................................82 4.5. Mean density of wading birds during ground surveys of 9 impoundments on KSC/MINWR, January July 2001.........................................................................91 4.6. Mean density of wading birds by im poundment section (s ee figures 4.2-4.4) during ground surveys of 9 impoundments on KSC/MINWR.................................92 4.7. A linear regression showed that as the proportion of open habitat increased the wading bird density decreased.................................................................................93 4.8. A linear regression showed that as the proportion of edge habitat increased the wading bird density increased..................................................................................93 4.9. Habitat use by species of wading birds observed during ground surveys of foraging habitat use of 9 impoundments on KSC/MINWR.....................................94 4.10. Proportion of unvegetated habitat contained within 9 impoundments on KSC/MINWR, circa 2001........................................................................................94 4.11. Distance to edge for observations of wading birds of 10 species during ground surveys of 9 impoundments on KSC/MINWR, January July 2001......................95 4.12. Resource selection ratios (wi) for wading birds foraging in impoundments on MINWR/KSC...........................................................................................................96 4.13. Histogram of sizes of wading bird fo raging aggregations observed during ground surveys of impoundments on KSC/MINWR, January July 2001.........................97 4.14. Composition of wading bird foraging aggregations observed during ground surveys of 9 impoundments on KSC/MINWR, January July 2001......................97 4.15. The number of birds observed in each of 4 group sizes categories by species.........98 4.16. Spatial distribution of wadi ng bird foraging aggregations.......................................99 4.17. Temporal distribution of wadi ng bird foraging aggregations.................................100 4.18. Wading bird density in 7 impoundme nts measured during aerial surveys concurrent with fish sampling periods...................................................................101 4.19. Mean wading bird density in 2 types of foraging habitat.......................................101 4.20. Mean wading bird density in 2 seas ons measured in 6 impoundments during aerial surveys concurrent w ith fish sampling periods............................................102 4.21. Results of model predicting wading bird density as a function of prey density and habitat type for Wetlands Initiative impoundments........................................107

PAGE 15

xv 5.1 Map of study site showing locatio n of several study impoundments on the Kennedy Space Center / Merritt Isla nd National Wildlife Refuge........................120 5.2. Great Egrets had a higher capture rate when foraging in groups than when foraging alone, while Snowy Egrets had hi gher capture rate foraging solitarily...131 5.3. Great Egrets increased their cap ture rate as group size increased............................132 A.1. Mean monthly water levels w ithin 6 impoundments on KSC/MINWR between February 2000 and February 2003.........................................................................146 A.2. Mean monthly water levels at 3 ga uge stations near study impoundments on KSC/MINWR between February 2000 and February 2003...................................147

PAGE 16

xvi Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy HABITAT SELECTION AND FORAGING SUCCESS OF WADING BIRDS IN IMPOUNDED WETLANDS IN FLORIDA By Eric Douglas Stolen May 2006 Chair: Franklin Percival Major Department: Wildlife Ecology and Conservation Most wading birds (Ciconiiformes) in Flor ida are tied to wetla nd habitats, and thus, to wetland protection and management. I inves tigated factors that influence piscivorous wading birds foraging habitat selection and us e at three spatial scales, and reasons why individuals choose to forage with others. Field work was conducted within the northern Indian River Lagoon, a sub-tropical es tuary in east central Florida. Based on follows from colonies, I found that nesting Great Egrets ( Ardea alba ) and Snowy Egrets ( Egretta thula ) preferred estuarine wetland ha bitat and avoided freshwater wetlands. Interpretation of selection pattern s depended on scale and habitats included. All birds followed foraged within 13 km of breeding colonies, well within the maxima reported for other colonies, s uggesting that resources with in impounded habitat met their energetic requirements. To facilitate testing hypotheses regarding wading bird ha bitat selection, I measured wading bird prey density in paired vegetated and unvegeta ted sites within impounded salt

PAGE 17

xvii marsh. Over 88% of the fish captured belonged to three species ( Cyprinodon variegatus Gambusia. holbrooki and Poecilia latipinna ). Fish were highly clumped spatially, and density and biomass varied substantially between impoundments and seasons for both habitats. Habitat selection within impoundments matc hed previous studies. Most species preferred unvegetated to vegeta ted habitat and all strongly pref erred the area within 0.5 m of the boundary between types. Prey density had a strong effect on wading bird habitat use in unvegetated but not in vegetated wetland sites. I measured the success rate of wading bi rds foraging alone and in mixed-species aggregations, and also measured the prey de nsity at the foraging sites and nearby unused sites. Foraging sites had higher biomass (but not density) than the average level available throughout the landscape. Foraging groups occurre d at the sites with higher prey density (but not biomass) than solitary foragers. Gr eat Egrets benefited from foraging in groups, Snowy Egrets did not. This study underscored the importance of edge habitat for foraging wading birds in impounded marshes, supported the social facilitation hypothes is regarding group foraging, and provided baseline data to measure respons es from restoration and management of marsh habitats in Florida.

PAGE 18

1 CHAPTER 1 INTRODUCTION Florida is home to 16 regularly occurring species within the order Ciconiiformes, birds commonly referred to as the long-le gged wading birds (Robertson and Woolfenden 1992). Due to a long history of direct harvesti ng, habitat loss, and habitat alteration, most species of wading birds in Florida have expe rienced two distinct reductions in population sizes during the last 2 cent uries (Ogden et al. 1978, Fr ohring et al. 1988, Ogden 1994). Following an initial decline due to plume hun ting in the late nineteenth century, most species showed healthy increases following le gal protection up to the 1930s. However, a second decline associated w ith extensive alteration of wetlands, resulted in great reductions to populations of most species in south Florida (Ogden 1994). Recently, population declines have been documented throughout the state for many species (e.g., Frederick 1996; Ogden 1996a, 1996b). Comp arison of 2 statewide nesting colony surveys conducted in 1976-1978 and 1986-1989 indicated that the numbers of wading birds breeding in the state declined, and populations became more fragmented, between the two periods (Runde 1991). Preliminary resu lts of a third statew ide nesting colony survey conducted in 1998-1999 show that th e decline in numbers of breeding wading birds and fragmentation of wading bird populations continues (J. Swan and J. Dodge, Florida Fish and Wildlife Conservation Commission, personal co mmunication). Thus, many species of wading birds are listed as sp ecies of special concer n by the Florida Fish and Wildlife Conservation Commission (Wood 1996).

PAGE 19

2 The interpretation of population declines in Florida is made more difficult by the tremendous shifts in the distri bution of wading birds during the last several decades in the southeastern United States (Ogden 1994, Fleu ry and Sherry 1995, Frederick et al. 1996, Frederick and Ogden 1997). These changes have been attribut ed to changes in management of wetlands (especially hydrology) in the regions involve d (Bancroft et al. 1994, Frederick and Spalding 1994, Ogden 1994). The ability of wading birds to shift their use of foraging habitat in response to changing conditi ons over smaller temporal and spatial scales also makes it difficult to track wading bird populations. It can be difficult to separate such changing patterns of ha bitat use from changes in the status of populations. Thus, there is a need for a mo re mechanistic unders tanding of factors influencing wading bird hab itat selection and use. Wading birds are intricately ti ed to their wetland habita ts, which provide resources for all aspects of their existence. Unfortunately, wetland loss in Florida has occurred at a staggering rate with over half of all ma rshlands (or 1.57 million ha) destroyed by 1987 (Kautz 1993). Many experts ag ree that the greatest curren t threat to wading bird populations in Florida is the continued loss and alteration of wetland habitat in the state (Frederick and Spalding 1994; Hoffman et al. 1994; Ogden 1994, 1996a; 1996b; Rodgers 1996; Runde 1996; Kushlan 1997). With a c onstantly expanding human population, it does not seem likely that the pressure on natural wetland habitats in Florida will ease soon (DeFreese 1991, Gilmore 1995). It is clear th at the future of wading birds in Florida is tied to protection of wetlands, a nd where possible (e.g., managed wetlands) multispecies or integrated management. It is al so clear that to formulate and implement a

PAGE 20

3 suitable multi-species management strategy, a better understanding of several features of their ecology is needed. My goal in conducting this research was to augment our understanding of wading bird selection of foraging habita t at different scales. In par ticular, I wished to learn more about factors that influence i ndividual piscivorous wading bi rds when they make choices about what type of foraging habitat to use, a nd whether or not to forage with others. I believe that science is essent ially a way of gaining inform ation about nature, and that there are several different methods that we can use to gain such knowledge. Often, controlled, replicated, manipulative experiment s are viewed as the best way to gain reliable scientific information. But in many s ituations it is difficult or even impossible to perform such experiments. In these cas es, 2 other tools available have proved particularly useful: observational study and th e comparative method. This work has been mainly observational and I hope that it will make a contribution to our understanding. The complexity of ecological systems ma kes management of natural resources difficult for many reasons. Before a manager can successfully apply interventions to a system, a basic understanding of the way the system will react is essential. This type of knowledge comes from understanding the mech anisms by which elements of the system interact. Before ecologists can develop mechanistic hypotheses, they must also understand the patterns which comprise the sy stems of interest. This dichotomy was explained eloquently by Wiens (1989). In trying to understand wading bird use of foraging habitat I strove to de lve into the mechanisms govern ing their interactions with the environment (both biotic and abiotic). A conceptual model of the factors influencing wading bird individual fora ging success is given in Figure 1.1. Often though, I was

PAGE 21

4 limited in what is known about the patterns of wading bird foraging habitat use. Thus, several of my objectives were aimed at desc ribing the patterns of wading bird habitat selection and use, and foraging behavior in impounded wetlands. In a few cases I was able to go deeper into glimpsing the underlyi ng proximate causes of their behavior. The objectives for this study were the following: 1. Determine the relative importance of im pounded wetlands as foraging habitat for wading bird populations breeding in the northern Indian River Lagoon ecosystem (hereafter IRL), and factors influencing habitat selection at the landscape level (Chapter 2), 2. Measure patterns (temporal a nd spatial) of wading bird prey distribution within impounded wetlands in the nor thern IRL (Chapter 3), 3. Document scale-dependent factors invol ved in wading bird foraging habitat selection and use within impounded wetlands in the nor thern IRL (Chapter 4), 4. Document the spatial and temporal pa tterns of occurrence of mixed-species foraging aggregations of wading birds w ithin impounded wetlands in the northern IRL (Chapter 4), 5. Investigate the effect of prey dens ity and group size on foraging success and distribution of foraging wading birds with in impounded wetlands in the northern IRL (Chapter 5), 6. Based on the results of the above obj ectives, formulate conservation and management recommendations to enhance wading bird habita t use and foraging success within impounded wetlands in the nor thern IRL based on the above factors (Chapter 6). I began this investigation by conducting a review of the first 11 years of data from a long-term monitoring program of wadi ng bird habitat use on the Kennedy Space Center/Merritt Island National Wildlife Refuge, hereafter referred to as KSC/MINWR (Stolen et al. 2002). This analysis revealed se veral key patterns that are of interest to managers at KSC/MINWR and require further ex planation. The first pattern of interest was the differential use of impoundments by wading birds on KSC/MINWR. Over the 11 year time period some impoundments had mu ch higher use as wading bird foraging

PAGE 22

5 habitat than others, despite a lack of obvious habitat differences expl aining the patterns. A second observation from the long-term data was that wading birds used wetland habitat without emergent vegetation less, and wetla nd habitat with emergent vegetation more, than availability would sugge st. A related pattern observa tion was that there was a seasonal shift in habitat use for most species with vegetated habitats becoming more used during the late fall and early winter (periods of higher water levels). In Chapter 2, I assessed the relative importance of impounded wetland habitat to nesting populations of wading birds in the no rthern Indian River Lagoon system. This has relevance to managers responsible for balancing the needs of different groups of wildlife (e.g., waterfowl, shorebirds, wading birds, fish). I hypothesized that some species of wading birds nesti ng within the northern IRL would select impounded wetland habitat for foraging over estuarine edge and unimpounded freshwater wetland habitat types. In chapter 3, I delved into the underlyi ng basis for habitat selection patterns by examining patterns in the dist ribution of fish between mana gement types, seasons and habitat type (vegetated versus unvegetated). I tested whether ther e was a difference in fish density or biomass between vegeta ted and unvegetated habitats within impoundments. I also modeled the density of fish as a function of habitat, season, and impoundment to explore patterns within the f actors determining density of wading bird prey. In Chapter 4, I examined patterns of habitat use and foraging ecology of wading birds within impoundments in se arch of factors important in determining foraging habitat suitability. I specifically tested whether wadi ng birds exhibited prefer ence for any one of

PAGE 23

6 3 habitat types within impoundments, namely, un vegetated, vegetated and edge habitat. I also described species-specific patterns of occurrence of mixed-species foraging aggregations on KSC/MINWR. I modele d the density of wading birds within impoundments as a function of prey density, ha bitat type, water depth, change in water depth, season and management, In Chapter 5 I addressed the question of the relative importance of prey density versus the presence of other foragers on individual foraging success. I tested 3 hypotheses. First, that wading bird foraging sites have higher prey density than nearby unused sites, second, that wading bird groups o ccur at sites with higher prey density or biomass than sites occupied by solitary forage rs, and third, that indi vidual Great Egrets or Snowy Egrets derive a benefit from foraging in aggregations that is due to the presence of other foragers (as opposed to any benefit due to enhanced prey density at sites with groups). The ultimate measure of the functional value of a habitat is the ability of individuals using the habita t to recruit new members into the population (Garshelis 2000). Many studies of wading birds have de monstrated a tight link between the ability of parents to secure food a nd their ability to su ccessfully produce o ffspring (Powell 1983, Hafner et al. 1986, Maddock and Baxter 1991, Ha fner et al. 1993, Frederick and Spalding 1994, Butler 1995, Thomas et al. 1999, Jakubas 2004). Without the direct demonstration of this link however, there is always a danger that individuals may se lect habitat that is not beneficial (Van Horne 1983, Schlaepfer 2002). My work attempted to measure components that are directly related to repr oductive success and surv ival of wading birds (Figure 1.2). The presence of a large and persistent wading bird population in my study

PAGE 24

7 area provides indirect evidence that the area provides critic al resources to wading birds (Stolen et al 2002). I addressed these research objectives and hypotheses in the impounded wetlands on the KSC/MINWR. This area encompasses a large portion of the northern IRL. Stretching for ca. 250 km along Floridas Atlantic coast, the IRL is a sub-tropical estuary with a high level of biodiversity, due to its lo cation at the junction of the warm-temperate Carolinian Province and the Tropical Caribbean Province (Gilmore 1995). Prior to recent human disturbance, the east ern shore of the IRL was ex tensively vegetated with mangrove swamps in the southern portion and irregularly flooded salt marsh habitats in the northern portion (Schmalzer 1995). Almost a ll salt marsh habitat in the northern part of the IRL was impounded for mosquito contro l by the 1970s (Brockmeyer et al. 1997). The northern part of the IRL is isolated fr om the nearest ocean inlet and has very low diurnal tidal changes (< 1 cm; Smith 1987). In this re gion, seasonal and wind-driven water level fluctuations are of much great er importance (Smith 1993). The hydrology of the northern IRL is marked by a high water period from September through December. This is followed by a gradual decline in water le vel with the lowest level occurring in late spring before the onset of summer rains. These changes greatly influence the depth of water over salt marsh habitat that is connected to the estuary, and c ontrol the extent of marsh surface covered with water (Trost 1968). A similar water-depth pattern also occurs within impoundments that are isolated from the estuary, probably as a function of rainfall, hydrostatic influence from the estu ary, and evapo-transpir ation (Stolen et al. 2002).

PAGE 25

8 KSC/MINWR contains 76 shallow impoundments in former salt marsh habitat. Human impacts on saline and brackish marshes of this region began as early as 1926 when ditches were dug to drain marsh surfaces to reduce oviposition by mosquitoes ( Aedes spp.). Following World War Two, mosquito control efforts turned to intensive spraying of DDT in the 1940s; however rapid e volution of pesticide -resistance rendered the use of DDT ineffective, and increasing public concern for risks associated with its use lead to the search for an a lternative means of mosquito control. A solution to the problem was found through impounding the salt marsh habitats that breed mosquitoes (Provost 1969). The success of impounding salt marsh for mosquito control in the IRLS culminated in the impoundment of nearly a ll of the marshes bordering the Indian River Lagoon by 1970. Alteration of the natural hydrol ogy changed the characteristics of the wetland habitats withi n, prompting a profound change in the vegetative, fish, and bird communities (Provost 1969, Rey et al. 1990, Schm alzer 1995, Brockmeyer et al. 1997). The ultimate response of wading bird populati ons to the ongoing effort to restore the hydrologic connection between isolated wetla nds and the IRL estuary has yet to be determined. This study provides baseline data to measure presumed benefits.

PAGE 26

9 Individual Foraging Success Flock Size Flock Composition Prey Density Prey Availability Interference (between foragers) Weather Season Hydrology Vegetation (structure) Prey Composition Prey Behavior Disturbance of Prey +/+ +/C ++ + +/+/+/+/C +/C C C Prey Depletion + + + + C Individual Foraging Success Individual Foraging Success Flock Size Flock Size Flock Composition Flock Composition Prey Density Prey Availability Prey Availability Interference (between foragers) Interference (between foragers) Weather Season Hydrology Vegetation (structure) Prey Composition Prey Composition Prey Behavior Prey Behavior Disturbance of Prey Disturbance of Prey +/+ +/C ++ + +/+/+/+/C +/C C C Prey Depletion Prey Depletion + + + + C Figure 1.1. A conceptual model of the fact ors influencing the foraging success of individual wading birds in shallow wetla nds. Factors are depicted as boxes, with broad environmental factors on th e left and the factor of interest (individual foraging success) on the right. Arrows depict casual relationships between factors. The symbols at ends of arrows indicate the nature of casual relationships (+ = positive influence, = negative influence, +/= direction of influence depends on level of the fact or, C = complex relationship about which insufficient information is known).

PAGE 27

10 Chapter 3 Assessment of Prey Base Chapter 5 Patch Level Individual Foraging Success Chapter 4 Impoundment Level Between and Within Foraging Site Selection Chapter 2 Landscape Level Forays From Breeding Colonies Foraging Habitat Type Selection Survival Reproduction (Persistence of Local Breeding Colonies) Chapter 3 Assessment of Prey Base Chapter 5 Patch Level Individual Foraging Success Chapter 4 Impoundment Level Between and Within Foraging Site Selection Chapter 2 Landscape Level Forays From Breeding Colonies Foraging Habitat Type Selection Survival Reproduction (Persistence of Local Breeding Colonies) Chapter 5 Patch Level Individual Foraging Success Chapter 4 Impoundment Level Between and Within Foraging Site Selection Chapter 2 Landscape Level Forays From Breeding Colonies Foraging Habitat Type Selection Survival Reproduction (Persistence of Local Breeding Colonies) Figure 1.2. Diagram depicting the relationshi p between the functional value of the northern Indian River Lagoon system for wading bird populations, and the specific topics investigat ed in this dissertation. Demographic performance (survival and reproduction) is the ultimate measure of the util ity of the habitat for wading bird populations. I addresse d several topics that influence the ability of wading birds to obtain the resources necessary to survive and successfully reproduce. The persistence of breeding co lonies in the northern Indian River Lagoon system since mon itoring began in 1987, suggests that the site has provided adequate levels of necessary resources over this time period.

PAGE 28

11 CHAPTER 2 FORAGING HABITAT SELECTION BY NE STING GREAT EGRETS AND SNOWY EGRETS Populations of colonial nesti ng herons and egrets (hereaft er ardeids) face a variety of threats, including isolation and alteration of coastal wetlands (Erwin et al. 2005). In some regions coastal wetlands are isolat ed from estuaries by impoundments that are created to control wetland hydrology or function; these ch anges can strongly affect foraging habitat availability and quality (Brockmeyer et al 1997, Erwin et al. 2005). Previous studies have shown that populations of colonial breeding wading birds can be limited by the availability of foraging habitat near their colonies (Fasola and Barbieri 1978, Gibbs 1991, Butler et al. 1992, Ogden 1994, Gibbs and Kinkel 1997, Jakubas 2004). Similarly, food availability has been shown to limit the survival of nestlings (Maddock and Baxter 1991, Butler 1995, Jakubas 2004) and in one case, their subsequent reproductive success (Thomas et al. 1999). Thus the availability and quality of foraging habitat surrounding colonies is important in determini ng reproductive success. For central place foragers like nesting egrets, distance from the colony is a factor that influences the choice of foraging habitat, and ultimately, demographic parameters (Gibbs 1991, Rosenberg and McKelvey 1999). Smith (1995a) found that nesting success and nestling production of Tricolored Herons ( Egretta tricolor ) was negatively associated with flight distance. Similarly, Simpson et al. (1987) reported that Great Blue Herons ( Ardea herodias ) that fed near colonies had greater breeding success than individuals that fed farther from colonies. In some cases colony abandonment may occur when flight

PAGE 29

12 distances exceed 25-30 km (Frederick and Co llopy 1989, Bancroft et al. 1994). Frederick and Spalding (1994) hypothesized that lost ti me rather than increased energetic demands induced by longer foraging flights was a major contributor to reduced breeding productivity or colony abandonment. Among the factors influencing wading bird habitat use, water de pth, which controls physical access to foraging habitat, and vegeta tion structure stand out as being the most significant (Chapter 4). Previ ous studies have found that pi scivorous wading birds prefer water depths below 25-30 cm (Chapter 4, Custer and Osborn 1978, Powell 1987, Gawlik 2002), and more open habitats when fora ging (Chapter 4, Breininger and Smith 1990, Hoffman et al. 1994, Surdick 1998, Bancroft et al. 2002). Coastal impounded wetlands often provide these foraging conditions (Erw in 1996). Previous studies have also suggested that estuarine wetlands may provide better foraging hab itat for piscivorous wading birds than do nearby freshwater wetland s due to differences in prey (MacCarone and Parsons 1994). Nesting colonies of wading birds provid e a tangible target for protecting and managing surrounding foraging habitat (Ogde n 1994, Kushlan 1997, Lombardini et al. 2001) and nesting ardeids can be a useful indi cator of anthropogenic impacts to wetlands (Stolen et al. 2005). However, for these bi ological elements to serve a conservation function, information on the factors influencing selection of foraging habitat is needed. The objective of this study was to quantif y foraging habitat use by Great Egrets ( Ardea alba ) and Snowy Egrets ( Egretta thula ) nesting on spoil islands within the northern part of the Indian River Lagoon system in central Fl orida (USA). This site contains extensive coastal impoundments and a large population of colonial nes ting ardeids. As such, it

PAGE 30

13 serves as a good site to i nvestigate the use of impounded coastal wetlands by colonial nesting ardeids and the potential interaction between flight distance and foraging habitat availability. I compared patterns of foraging habitat use with available habitat at three scales to assess the importance of impounded salt marsh habitat to nesting Great Egrets and Snowy Egrets in this system. The basic scientific hypothesis test ed (for each scale) was that nesting wading birds selected impounde d salt marsh for foraging habitat. I also measured foraging flight distances and f light speed, because these parameters can influence colony and foraging site select ion (e.g., Gibbs 1991). Finally, I report information on foraging groups and flock size leaving the colonies because levels of foraging scalability in egrets varies wide ly among sites (Chapter 4, Caldwell 1981, Erwin 1983b, Kersten et al. 1991, Hafner et al. 1993, Master et al. 1993) and these attributes may influence selection at loca l levels (Dall et al. 2005). Methods Study Site Follows of birds from their nests to fo raging sites were made from three mixedspecies wading bird nesting colonies locat ed within the boundaries of the Kennedy Space Center-Merritt Island National Wildlife Re fuge (KSC/MINWR) (Figure 2.1). The 55,000 ha KSC/MINWR is located in the northern portion of the Indian River Lagoon system, a subtropical estuary that stretche s for ca. 250 km from Ponce de Leon Inlet to Jupiter Inlet. This estuary is an important site for wading birds on the southeastern Atlantic coast of North America (Schikorr and Swain 1995, Se well et al. 1995). KSC/MINWR supports a large wading bird population that utilizes fr eshwater and impounded salt marsh habitats for feeding, roosting, and nesting (Smith and Breininger 1995, St olen et al. 2002). Almost all salt marsh habitat in the no rthern Indian River Lagoon was impounded for

PAGE 31

14 mosquito control by the 1970s (Brockmeyer et al. 1997). Within 15 km of the three colonies there were 90 impoundments (ave rage size 141 ha, total area 12,716 ha) containing salt marsh habitat adjacent to the es tuary. The northern portion of the Indian River Lagoon system is isolated from ocean inlets and has very low diurnal tidal changes (< 1 cm; Smith 1987). In this region, seasona l and wind-driven wate r level fluctuations are of much greater importance (Smith 1993). Habitat within impoundments is predominantly a heterogeneous mixture of ope n water and vegetated cover types, with tall marsh grass (e.g., Spartina bakeri ) and short marsh vegetation (e.g. Distichlis spicata Batis maritima ) predominating in vegetated areas (Schmalzer 1995). Efforts are currently underway to re connect isolated impounded wetlands to the estuary (Brockmeyer et al. 1997). Nesting colonies on KSC/MINWR have b een monitored since 1987 (Stolen et al. 2002). Focal colonies were selected as th e largest colony within each of three broad north-south strata of KSC/MINWR (Figure 2.1) Two of the colonies occurred on spoil islands located near dredged navigation cha nnels (Mullethead Isla nd and Banana River #14) and the third was on a large naturally occurring island which had been drag-line ditched prior to 1970 for mosquito control (B ig Island). Mullethead Island has had nesting colonies in every year since m onitoring began in 1987; Big Island has had colonies since 1993 and Banana River 14 sinc e 1994 (E. D. Stolen, unpublished data). Foraging Flight Observations I followed Great Egrets and Snowy Egrets from their nests to their foraging locations (hereafter re ferred to as follows). Follo ws were conducted using a NASA Huey helicopter between sunr ise and five hours after sunr ise. The helicopter was hovered 300 m horizontal distance from each col ony at an altitude of 150 m until a bird

PAGE 32

15 was observed leaving the colony. The helicopt er remained at least 300 m behind the bird while it was followed. Once a bird landed at a foraging location, a GPS position was recorded for the location using a Garmin 12 channel GPS receiver, and the habitat type was noted. Habitat types were defined as open water (patch es of at least 2 m of no emergent vegetation) or vegetate d. A site was considered oc cupied if at least one other wading bird occurred within 30 m of the land ing location; in these cases the landing bird was considered to have joined an aggregati on. I recorded the iden tity of other wading birds within 100 m of the subj ects landing location. If the subject was part of a group, an attempt was made to note the landing pos ition of all members of the group, but often this was not possible because the subject bird would continue flying. In these cases, only information about the subject bird was recorded. Habitat Use Analysis GPS positions of landing sites were record ed in a GIS for calculation of distances and habitat use analysis. Analysis was c onducted using the software packages ArcView 3.2 and ArcMap 8.2 (ESRI, Redlands California) Quantification of available habitat was based on a land cover map produced by the St. Johns River Water Management District using photo interpretation of 1:40,000 colorinfrared aerial photography taken in 1999 (Anonymous 2002). To calculate the total amoun t of available habita t, three concentric distance-buffers of 5, 10, and 15 km radii were produced around each of the three colonies, and the ensuing regions wee each dissolved into a single polygon where the colony-specific distance-buffers overlapped. Th ese distances were c hosen to include all wetland habitat available within the range of flight distances of Great Egrets and Snowy Egrets observed in previous studies (e.g., Frederick and Collopy 1989, Bancroft et al. 1994).

PAGE 33

16 Wetland habitats within the distance-buffers were recla ssified into three groups: impounded salt marsh, estuarine edge, and fres hwater wetland. Using the GIS, all wetland habitat occurring within the peri meter dikes surrounding impoundments was classified as impounded salt marsh habitat, a nd the area of these we tlands within each distance-buffer was summed. Some areas that were impounded have recently been restored by removal or breeching of the pe rimeter dikes (Brockmeyer et al. 1997). I included these areas in the total of impounded habitat for two reasons. Although restoration of a few of these sites occurred decades ago, most restoration was completed more recently and all restored areas had expe rienced long periods of impoundment. Thus restored sites may not yet have return ed to functioning as unimpounded wetlands. Secondly, the total area of wetlands within these impound ments was small (less than 5 percent) compared with the total for all wetlands within the 15km buffer. If these wetlands were included in another category, th ey would slightly increase the expected number of birds foraging in that habita t and decrease the expected number in impounded salt marsh. Thus, inclusion of these wetlands within the impounded salt marsh area is conservative when evaluating the hypothe sis of preference for impounded salt marsh (because it increases the pred icted proportion of birds fo raging within impoundments, thus making it harder to conclude that birds sel ected this habitat). Estuarine habitat was defined as shallo w areas along the edges of three large lagoonal basins in the study area : the Indian River, The Bana na River, and the Mosquito Lagoon (Figure 1). To calculat e the area of estuarine edge accessible to wading birds two different radii were produced along the estuarin e edge within the basins,. I used 2 buffer distances to quantify habitat for Great Egrets and Snowy Egrets separa tely. Distances for

PAGE 34

17 species-specific buffer widths were determined by examining the distribution of distances at which birds were observed during long-term foraging habitat use surveys of estuarine edge which have been conducted monthl y on KSC/MINWR since 1987 (Stolen et al. 2002). Distances were chosen that included 95% of all observations for each species as follows: Great Egret 100 m (n=5804), Snowy Egret 40 m (n=3404). The area of freshwater wetlands was calcu lated by first summing the area of all unimpounded wetlands within each distance buffe r, excluding the interiors of lakes and rivers. Then to represent available foraging ha bitat within open water of lakes and rivers, a 1-m buffer was generated within all lakes a nd rivers and the area ad ded to the total for unimpounded wetlands. Because the 2000 Land Cover and Land Use map did not map wetlands less than 0.5 acres, I used polygons from a 1990 Land Cover map of KSC/MINWR which included such features (Larson 1992). Because this map did not extend far enough west to include the entire study area, I mapped these features (where necessary) using 1999 orthor ectified color infrared photography (Anonymous 1999). Although both Great Egrets and Snowy Egrets will use forested wetlands (Palmer 1962, Hancock and Kushlan 1984) it is un clear how important this habitat is to either species. Thus I conducted habitat use analysis both w ith and without these we tlands included, to assess the effect of uncertainty over their use. I chose to include the analysis both with and without forested wetlands because planners may consider wetlands of all types as a unit when making management decisions (i.e., lumping forested and unforested wetlands together). I conducted habitat selection analysis using a resource sele ction ratio: wi =[proportion of habitat i used]/[p roportion of habitat i available] (Manly et al. 2002). This

PAGE 35

18 ratio measures the relative preference betw een habitat types al lowing comparison of preferences between habitats th at are not affected by what types are included. For each analysis, I calculated the resource selection ratio of the observed bird use of each of the three habitat categories (impounded salt marsh, freshwater and estuarine edge) for each distance buffer (5,10, and 15 km) for each specie s. Thus, I examined habitat use for 6 different scenarios for each species (i.e., 5 km radius with forested wetlands included, 5 km radius without forested wetlands included, et c.). Selection ratios were evaluated to determine if they differed from 1.0 (and thus exhibited selection or avoidance of the habitat type) using a Bonferroni partitioning of experiment-wise = 0.05. I also tested whether each selection ratio fo r each habitat type differed fr om the other 2 using methods presented in Manly et al. (2002) for each habitat use scenario. In several cases more than 1 bird was followed from a colony to a single fo raging site and in thes e cases, I included only 1 location in the analysis keeping with the assumption that individuals observed are independent (Manly et al. 2002). I assumed that individuals followed on different days were independent from other observations. To allow comparison with other habitat selection studies, I also calculated 95% conf idence intervals on the proportions of habitat used by each species for each scenario using methods previously described (Neu et al. 1974, Byers et al. 1984, Cherry 1996). Distance traveled between the nest and th e foraging site was calculated with the GIS. Due to the difficulty of observing birds exactly as they left nests, colony centers were used for the origins of all foraging flights. Duration of foraging flight was calculated as the duration betw een the detection of the bird leaving the colony and the time the bird landed at the foraging site. Th is usually resulted in a few seconds being

PAGE 36

19 truncated between when a bird left a nest a nd when it was detected leaving a colony. The average flight speed was calculated as the st raight line distance traveled divided by the time of travel; this is a minimum estimate because many birds changed headings during travel resulting in actual f light distances being greater than straight line distances calculated in the GIS. Results Foraging Flight Characteristics Thirty Great Egrets and 54 Snowy Egrets were followed from 3 nesting colonies between April 7 and June 9, 2000 (on 7 days for Great Egret and on 8 days for Snowy Egret). This period was after most nests were initiated and adults coul d be expected to be provisioning chicks (E. Stolen, unpublished data ). Great Egrets la nded at 28 and Snowy Egrets at 31 unique foraging locations (Figure 2.1). The destination of 3 Snowy Egrets and 1 Great Egret could not be determined. There were still many active nests in all 3 colonies when follows ended. The maximum numbers of nests coun ted in the entire study area were 202 Great Egrets and 165 Snow y Egrets in 11 colonies (E. Stolen, unpublished data). Over half of all Snowy Egrets followed le ft the colonies in groups (8 of the 31 separate follows). Most of th ese groups were composed of conspecifics except in 2 cases when Snowy Egrets left the colony with a Gr eat Egret. The mean and median group size of Snowy Egrets that left colonies in gr oups was 3.9 and 2.5 respectively. In addition, 4 of the Snowy Egrets followed (2 single bird s and a group of 2) joined groups in flight. Most Snowy Egrets that left colonies in groups or joined groups in flight arrived at foraging sites with those groups (2 Snowy Egrets followed di d not). All but 3 of the Snowy Egrets followed arrived at foraging s ites that were occupied by mixed-species

PAGE 37

20 foraging aggregations; 1 arrived at a site occupied by a lone Snowy Egret and 2 arrived at unoccupied sites. The mean and median si ze of mixed-species foraging aggregations joined by Snowy Egrets is summarized in tabl e 2.1. The distance, duration and speed of all Snowy Egret foraging flights fo llowed are summarized in Table 2.2. Seven of the 30 Great Egrets followed left colonies in groups (5 of 28 separate follows). Two left in groups composed of a nother single Great Egret, the others with Snowy Egrets. The mean and median group size of Great Egrets that left colonies in groups was 2.4 and 2 respectively. No Great Egret followed jo ined a group in flight. Only 3 of the 7 Great Egrets that departed colonies in groups arri ved at foraging sites with those groups (2 of the 5 follows). Twenty-two of the 30 Great Egrets followed arrived at foraging sites that were occupied by foraging aggregations (20 of 28 follows); 14 of the aggregations were composed of mixed-species foraging aggregations while 6 were composed of only conspecifics. Th e mean and median size of mixed-species foraging aggregations joined by Great Egrets is summarized in ta ble 2.1. The distance, duration and speed of the Great Egret foragi ng flights followed are summarized in Table 2.2. Over all groups followed, aggregations join ed by Great Egrets were significantly smaller than were those joined by Snowy Egrets (Mann-Whitney Test, Z = -2.011, p = 0.044). Great Egrets flew great er distances to foraging loca tions than did Snowy Egrets (Mann-Whitney U test, Z = 2.474, p = 0.013); the duration of trips from colonies to foraging sites were not statistically different between species (Mann-Whitney U test, Z = 1.689, p = 0.091).

PAGE 38

21 Habitat Use Analysis There was no difference in the frequency of use of the 3 habitat types between species ( 2 2 = 2.35, p=0.309). Great Egrets that landed in impounded salt marsh (15 of 28 sites) landed exclusively in open water and always joined a foraging aggregation. Similarly, most impounded salt marsh sites in which Snowy Egrets landed were in open water (14 of the 19). Snowy Egrets landing in impoundments always joined aggregations of other wading birds. The size of aggr egations joined by birds landing in impounded salt marsh was larger than that for other habi tats combined (Table 1) and the difference was significant for both Great Egret (M ann-Whitney Test, Z = -1.987, p = 0.047) and Snowy Egret (Mann-Whitney Test, Z = -2.02, p = 0.043). Four of the freshwater wetland sites at which Great Egrets landed were ope n water while 1 was vegetated. Similarly, 6 of the freshwater wetland sites at which Snowy Egrets landed were open and 2 were vegetated. All but 2 of the 12 estuarine edge sites at wh ich Great Egrets and Snowy Egrets landed were in open water up to 40 m from the shoreline edge. The two remaining sites (one for each species) were located on small unimpounded islands of low marsh vegetation located away from the mainland. The proportion of total area within flight distances from colonies that was wetland habitat ranged from 0.33 to 0.11 depending on f light radius, species, and whether or not forested wetlands were include d in the analysis (Table 2.3). Fifty-eight unique foraging locations were identified during follows of Great Egrets and Snowy Egrets. More than half of the foraging sites we re located within impounded salt marsh (Tables 2.4 and 2.5). When forested wetlands were included in the analysis, both species had positive selection ratios for impounded salt marsh and estuarine ha bitats and negative se lection ratios for

PAGE 39

22 freshwater habitat at all sp atial scales (with the exception of impounde d habitat for Great Egrets at the 5 km radius; Figure 2.2). Gr eat Egrets showed avoidance of freshwater habitat at the two larger spatial scales but not at the smallest scale; Snowy Egrets showed avoidance of freshwater habitat at both th e largest and smallest spatial scales, and selection for impounded habitat at the largest spa tial scale (Figure 2). These patterns of habitat selection and avoidance largely di sappeared when forested wetlands were excluded form the analysis (Figure 2.2).

PAGE 40

23 Table 2.1. Summary of the sizes of foraging aggregations joined by Great Egrets and Snowy Egrets followed from three co lonies on the Merritt Island National Wildlife Refuge. Entries in table for each species are mean number (SE) of birds in aggregations on the first row, followed by median and sample size on the next row. SpeciesImpoundedFreshwaterEstuarine EdgeCombined (all sites) Great Egret112.0 (33.1)11.0 (9.5)6.2 (3.4)70.4 (22.8) 107.5, n = 122.0, n = 33.0, n = 510.5, n = 20 Snowy Egret256.1 (82.1)24.9 (8.0)47.5 (12.5)183.4 (58.9) 100, n = 1930.0, n = 747.5, n = 247, n = 28 Table 2.2. Summary of flight distance, durat ion and speed for Great Egrets and Snowy Egrets followed from three colonies on the Merritt Island National Wildlife Refuge. Values in parentheses are standa rd errors. Sample sizes are listed in table. MeasureGreat EgretSnowy Egret Average distance (km)6.2 (0.46), n=284.7 (0.48), n=31 Median Distance (km)5.6, n=284.2, n=31 Distance range (km)1.8 10.7, n=280.7 12.5, n=31 Average duration (min)10.3 (1.0), n=288.0 (0.92), n=30 Median Duration (min)9.4, n=286.5, n=30 Duration range (min)2.0 22.8, n=281.0 18.8 Average speed (km/hr)a38.8 (1.6), n=2838.2 (1.7), n=30 a average speed calculated from dur ation and distance of each flight.

PAGE 41

24 Table 2.3. Areas of Great Egret and Snowy Egre t foraging habitats w ithin three flightradius distances from three nesting co lonies in the northern Indian River Lagoon, Florida. Areas are in hectares. Summary5 km10 km15 km Total area within distance of colonies23,44282,162152,063 Area canal and lake 1 m buffer195978 Area estuary edge 40 m buffer6861,6972,375 Area estuary edge 100 m buffer1,5323,7855,278 Area impounded salt marsh w ith forest3,5099,85712,064 Area impounded salt marsh w ithout forest2,7107,7829,448 Area freshwater wetlands w ith forest2,6208,93014,418 Area freshwater wetlands w ithout forest6322,9694,380 Proportion Great Egret habitat with forests0.330.280.21 Proportion Great Egret habitat without forests0.210.180.13 Proportion Snowy Egret habitat with forests0.290.250.19 Proportion Snowy Egret habitat without forests0.170.150.11

PAGE 42

25Table 2.4. Comparison of the proportion of bird use versus the proportion of habitat av ailability, for three types of foraging habitat by nesting Snowy Egrets. Analysis was conduc ted at three spatial scales (15, 10 a nd 5 km buffer distances). Bold text indicates evidence of selection (+ ) or avoidance (-) for a partic ular habitat type and scale. ComparrisonImpounded salt marshFreshwater wetlandsEstuarine15 km buffer Numbers of individuals followed to each type of location within buffer1984 Bird use of habitat as proprtion of all use within buffer (95 % confidence intervala) 0.61 (0.37, 0.93)0.26 (0.09, 0.55)0.13 (0.02, 0.27) Habitat as proportion of total area within buffer, forested wetlands included0.420.50.08 Habitat as proportion of total area within buffer, forested wetlands not included0.580.270.15 10 km buffer Numbers of individuals followed to each type of location within buffer1784 Bird use of habitat as proprtion of all use within buffer (95 % confidence intervala) 0.60 (0.35, 0.92)0.27 (0.09 0.55)0.13 (0.02, 0.27) Habitat as proportion of total area within buffer, forested wetlands included0.480.440.08 Habitat as proportion of total area within buffer, forested wetlands not included0.620.240.14 5 km buffer Numbers of individuals followed to each type of location within buffer1543 Bird use of habitat as proprtion of all use within buffer (95 % confidence intervala) 0.67 (0.36, 0.95)0.19 (0.030, 0.38)0.14 (0.01, 0.27) Habitat as proportion of total area within buffer, forested wetlands included0.51 0.39 0.1 Habitat as proportion of total area within buffer, forested wetlands not included0.660.170.17 a based on Cherry (1996).

PAGE 43

26Table 2.5. Comparison of the proportion of bird use versus the proportion of habitat av ailability, for three types of foraging habitat by nesting Great Egrets. Analysis was conduc ted at three spatial scales (15, 10 a nd 5 km buffer distances). Bold text indicates evidence of selection (+ ) or avoidance (-) for a partic ular habitat type and scale. ComparrisonImpounded salt marshFreshwater wetlandsEstuarine15 km buffer Numbers of individuals followed to each type of location within buffer1558 Bird use of habitat as proprtion of all use within buffer (95 % confidence intervala) 0.54 (0.29, 0.88)0.18 (0.04, 0.37)0.29 (0.01, 0.58) Habitat as proportion of total area within buffer, forested wetlands included0.38 0.46 0.17 Habitat as proportion of total area within buffer, forested wetlands not included0.490.230.28 10 km buffer Numbers of individuals followed to each type of location within buffer1448 Bird use of habitat as proprtion of all use within buffer (95 % confidence intervala) 0.54 (0.28, 0.88)0.15 (0.02, 0.31)0.31 (0.11, 0.62) Habitat as proportion of total area within buffer, forested wetlands included0.44 0.40 0.17 Habitat as proportion of total area within buffer, forested wetlands not included0.530.210.26 5 km buffer Numbers of individuals followed to each type of location within buffer423 Bird use of habitat as proprtion of all use within buffer (95 % confidence intervala) 0.44 (0.07, 0.75)0.22 (0.00, 0.37)0.33 (0.03, 0.58) Habitat as proportion of total area within buffer, forested wetlands included0.460.340.2 Habitat as proportion of total area within buffer, forested wetlands not included0.550.130.31 a based on Cherry (1996).

PAGE 44

27 ! ! ! ! ! ! ! ! !! !! ! ! ! ! !# # # # # ### # # # ### ####### # # # ## # # # # # ## # # # ####### # # # ### #A A AMullethead Banana Creek Banana River 0510 2.5 KilometersMosquito Lagoon Indian River Banana RiverAtlantic Ocean Figure 2.1. Foraging locations of Great Egrets (circles) a nd Snowy Egrets (triangles) followed from three colonies in the nort hern Indian River Lagoon, Florida.

PAGE 45

28 Great Egret, 5 km radius 0.00 0.50 1.00 1.50 2.00 2.50 3.00 IMPUNIESTIMPUNIEST Snowy Egret, 5 km radius0.00 0.50 1.00 1.50 2.00 2.50 IMPUNIESTIMPUNIEST Great Egret, 10 km radius0.0 0.5 1.0 1.5 2.0 2.5 IMPUNIESTIMPUNIEST Snowy Egret, 10 km radius 0.0 0.5 1.0 1.5 2.0 2.5 IMPUNIESTIMPUNIEST Great Egret, 15 km radius0.0 0.5 1.0 1.5 2.0 2.5 IMPUNIESTIMPUNIEST Snowy Egret, 15 km radius 0.0 0.5 1.0 1.5 2.0 2.5 IMPUNIESTIMPUNIESTForested wetlandsNo forested wetlandsForested wetlandsNo forested wetlands Forested wetlandsNo forested wetlandsForested wetlandsNo forested wetlands* * *wiwiwiwiwiwiForested wetlandsNo forested wetlandsForested wetlandsNo forested wetlands Great Egret, 5 km radius 0.00 0.50 1.00 1.50 2.00 2.50 3.00 IMPUNIESTIMPUNIEST Snowy Egret, 5 km radius0.00 0.50 1.00 1.50 2.00 2.50 IMPUNIESTIMPUNIEST Great Egret, 10 km radius0.0 0.5 1.0 1.5 2.0 2.5 IMPUNIESTIMPUNIEST Snowy Egret, 10 km radius 0.0 0.5 1.0 1.5 2.0 2.5 IMPUNIESTIMPUNIEST Great Egret, 15 km radius0.0 0.5 1.0 1.5 2.0 2.5 IMPUNIESTIMPUNIEST Snowy Egret, 15 km radius 0.0 0.5 1.0 1.5 2.0 2.5 IMPUNIESTIMPUNIESTForested wetlandsNo forested wetlandsForested wetlandsNo forested wetlands Forested wetlandsNo forested wetlandsForested wetlandsNo forested wetlands Forested wetlandsNo forested wetlandsForested wetlandsNo forested wetlands Forested wetlandsNo forested wetlandsForested wetlandsNo forested wetlands* * *wiwiwiwiwiwiForested wetlandsNo forested wetlandsForested wetlandsNo forested wetlands Forested wetlandsNo forested wetlandsForested wetlandsNo forested wetlands Figure 2.2. Great Egret and Snowy Egret foraging flight habitat resour ce selection ratios (error bars show SE) displayed by habita t type, for analyses with and without forested wetland habitats included, for th ree concentric radii of distances from colony locations. Bars marked with show evidence of selection or avoidance of habitat type. IMP= impounded salt marsh, UNI=freshwater wetlands, EST=estuarine edge habitat.

PAGE 46

29 Discussion Colonial nesting Great Egrets and Snow y Egrets showed evidence of habitat selection, but patterns were dependent on the scale of the analysis and the decision to include or exclude forested wetlands. Both species avoided freshwater habitat and used both impounded salt marsh and estuarine habita t more than expected at several spatial scales when forested wetlands were include d. When forested wetlands were excluded, both species appeared to use habitat in propor tion to availability. The difficulty in assessing habitat selection in animals due to uncertainty over the designation of what habitat is available is well-known (Johns on 1980, Alldredge and Ratti 1986, McClean et al. 1998, Manly et al. 2002). My results cauti on researchers against arbitrarily including or excluding habitats at various scales as it ma y lead to incorrect inte rpretations of habitat selection, including placing undue impor tance on some habitat types. My results are consistent with other studies that have shown that patterns of habitat selection by egrets are dependent on the scal e of the analysis (e.g., Custer and Osborn 1978, Fasola and Barbieri 1978, Gibbs et al 1987, Gibbs 1991, Gibbs and Kinkel 1997). One explanation for such a hierarchical proce ss is that decisions made at the broadest spatial scale can influence (limit) choices at more local scales (Johnson 1980). This possibility may explain the lack of apparent habitat selection by eg rets when forested wetlands were excluded from the analysis. In this study, forested wetlands tended to be more distant from colony sites, and egrets a ppeared to have selected colony sites that were surrounded by abundant suitable foraging ha bitat. Habitat selec tion within the birds flight radius around the colony appeared to be determined by factors other than coarse wetland type (i.e., forested versus non-forested).

PAGE 47

30 In this study almost all salt marsh habitat was impounded. Although my results highlighted the value of impounded salt marsh hab itats in this system, I did not assess its value as compared to unimpounded salt marsh ha bitat. Portions of the study area are the focus of intensive restoration efforts (Brock meyer et al. 1997). Th erefore, it would be appropriate to assess the que stion of selection between impounded versus unimpounded salt marsh habitat when restored areas b ecome more common in the landscape. This study suggests that egrets nesting in the northern Indian River Lagoon estuary can find suitable foraging habitat in relative close proximity to colonies to meet the energetic demands of reproducti on. All egrets I followed trav eled less than 13 km, with Great Egrets generally traveling farther than Snowy Egrets. This distance was within the 18 km radius traveled by most Great Egrets and Snowy Egrets from nesting colonies to foraging sites reported in other studies (e.g., Custer and Osborn 1978, Thompson 1978, Frederick and Collopy 1989, Bancroft et al. 1994, Smith 1995a). Moreover, the average flight duration and flight speed s recorded in this study were similar to that recorded in previous studies (Custer and Osborn 1978, Th ompson 1978). Flying farther may occur if or when habitat conditions around colonies de teriorate (Bancroft et al. 1994, Frederick and Collopy 1989, Frederick and Spalding 1994). Under such conditions, birds have to deal with trade-offs between carrying larger prey loads and the energetic demands of increased travel (Smith 1995a). For this re ason, distant habitats may not be preferred under optimal conditions, but may be importa nt under unusual circumstances (drought), or during certain stages of the nesting cycle. Most of the Great Egrets and Snowy Egrets I followed landed at sites with other wading birds present, demonstr ating the high level of social foraging under the conditions

PAGE 48

31 in my study area (see Chapter 4). The benef its of joining a group might be associated with locating prey resources (Krebs 1974, Hafner et al. 1998), or improved foraging success of group foraging individuals (Cha pter 5, Caldwell 1981, Hafner et al. 1982, Cezilly et al. 1990, Master et al 1993). However, the degree to which Great Egrets and Snowy Egrets forage socially varies wide ly between studies. For example, Smith (1995b) reported that Great Egrets and Snow y Egrets invariably joined foraging birds when followed from a nesting colony in sout h Florida. Similarly, Custer et al. (2004) found that 34% of Great Egrets followed landed at sites with other wading birds present. In contrast, Custer and Galli (2002) found that Great Egrets landed at unoccupied sites 74% of the time. In this study, joining groups might have been facilitated by the tendency of the Great Egrets and Snowy Egrets to travel to foraging sites in groups. Group flights from nesting colonies also appear s to vary widely between sites with some authors reporting levels lower than in this study (e.g., Erwin 1983b,1984; MacCarone and Parsons 1988; Smith 1995a). Most studies of ardeid foraging flights ha ve suggested that th e availability of foraging habitat surrounding the colonies is an important co mponent in the protection of wading bird nesting colonies. Nesting ardeids often switch foraging habitats rapidly in response to changes in hydrology (Smith 1995a, Smith and Collopy 1995, Custer et al. 2004). Therefore, protecting a mix of differe nt wetland types within flight distance of colonies is prudent because unpredictable di sturbances may affect some types but not others. This study suggested that such a le vel of protection, incl uding contingencies for fluctuations/changes in habitat conditions, could be met by protecting a variety of wetland habitats within 15 km of the nesting colonies of Gr eat Egrets and Snowy Egrets

PAGE 49

32 in Merritt Island. The range of conditions described in this and other studies highlights the flexibility in foraging behavior of thes e species, but also underscores the value of gathering system-specific information to help guide management decisions on their behalf.

PAGE 50

33 CHAPTER 3 THE DISTRIBUTION OF WADING BIRD PREY IN IMPOUNDED WETLANDS IN THE NORTHERN INDIAN RIVER LAGOON ESTUARY Introduction Many species of wading birds are dietary spec ialists feeding on small fish living in shallow wetland habitats (Hancock and Ku shlan 1984, Frederick 2002). Available work has demonstrated a direct connection between prey distribution a nd piscivorous wading bird foraging habitat use (Ker sten et al. 1991, Fasola et al. 1996, Master et al. 2005), and has highlighted the importance of understand ing factors influenci ng prey availability within wetlands (e.g., Gawlik 2002). Similarly, there is evidence that prey density (often considered a surrogate for availability) is an important factor determining foraging success of wading birds (e.g., Erwin et al. 1985, Cezilly et al. 1990) To understand the factors underlying a predators selection between available fo raging habitats, it is helpful to know the level of prey within each type. A few studies have attempted to describe the relative distribution of piscivorous wading bird prey between available foraging habitats (e.g., Erwin et al. 1985, MacCar one and Parsons 1994, Fasola et al. 1996, Trexler et al. 2003). However, the generality of relationshi ps will only be established as similar types of information are generated elsewhere. Stretching for ca. 250 km along Floridas A tlantic coast, the Indian River Lagoon System (IRL) is a sub-tropical estuary with a highly diverse fauna of fish (Snelson 1983) and a large and diverse population of pisc ivorous wading birds (Smith and Breininger 1995). Prior to human alteration, the eastern shore of the northern IRL was extensively

PAGE 51

34 vegetated with irregularly flooded salt marsh (S chmalzer 1995). However, almost all salt marsh habitat was impounded for mosquito control between 1954 a nd the early 1970s (Brockmeyer et al. 1997). Impounding of these wetlands has had a large impact on the fish community within, beginning with a reduc tion in the diversity and abundance of fish as a result of initial impounding (Gilmore et al. 1982, Harrington and Harrington 1982). Native IRL salt marsh had a hydroperiod char acterized by extreme changes in water depth and extent of marsh surface coverage between different seasons (Trost 1968). Impounding of marshes resulted in longer hydroperiods, greater water depths and a reduction in vegetated surface and increase in open water lacking emergent vegetation (Brockmeyer et al. 1997). These changes appa rently increased the populations of small resident fishes such as the Sheepshead Minnow ( Cyprinodon variegatus ) and the Mosquitofish ( Gambusia holbrooki ) (Brockmeyer et al. 1997, Gilmore 1998). The impounded salt marshes of the northe rn IRL are a good place to investigate patterns of piscivorous wading bird prey dist ribution. Sketchy information from the preimpoundment period on wading bird use suggests that the native marsh habitat had low wading bird use, estimated at 0.9 individuals ha-1 (Trost 1968). Since impounding, wading bird density within some impounded sa lt marsh habitat apparently increased by a factor of two (Provost 1968, Trost 1968), but several impou ndments in the northern IRL with similar habitat composition and similar management histories are used at vastly different rates by foraging wading birds (Stole n et al. 2002). With in impoundments of the northern IRL, piscivorous wading bi rds predominantly use shallow unvegetated flooded habitat and to a lesser extent, shallo w flooded wetlands with low stature, salttolerant plants (chapter 4, Breininger and Smith 1990, Smith and Breininger 1995, Stolen

PAGE 52

35 et al. 2002). However, it is unclear whethe r prey abundance or othe r factors (e.g., habitat structure, hydrology) are the key determinants of habitat preference. The simple division of available habitat into tw o categories (vegetated and unvegetated) makes quantifying patterns of piscivorous wading bird prey abundance straightforward. I designed this study primarily to test to the prediction that prey is more abundant in the unvegetated habitat type more often used by foraging wading birds. In addition, on-going restorati on and management within this system allows the opportunity to test relationships betw een hydrology and wetland connectivity with wading birds and their prey (cha pter 4, Stolen et al. 2005). Understanding of the spatial and temporal abundance and distribution of resident fish within impounded salt marsh habitat in the northern IRL would enhance the ecological foundation for management of both fish and wading bird habitats. It also would provide a better basis to compare and draw appropriate inferences and conservation insights about other wetlands in Florida. My objectives were the follwoing: 1) measure the density and bi omass of fish within both unvegetated and vegetated wading bird fora ging habitat within impounded salt marsh, 2) investigate seasonal patterns of abundance be tween habitats and management types, and 3) characterize the patchiness of fish at a sc ale relevant to the use of this resource by wading birds. This information will ultimat ely improve the understanding of the foraging habitat selection and foraging ecology of pi scivorous wading birds within this system. Methods Study Site The study site consisted of two areas of impounded salt marsh located in the northern portion of the IRL on the Kenne dy Space Center-Me rritt Island National Wildlife Refuge (KSC/MINWR, Figure 3.1) Habitat within impoundments is

PAGE 53

36 predominantly a mixture of open water and ve getated cover types, with tall marsh grass (e.g., Spartina bakeri ) and short marsh vegetation (e.g., Distichlis spicata Batis maritima ) predominating in vegetated areas (Schmalzer 1995). The unvegetated open water areas within impoundments are of 2 t ypes, both of which have sharply defined boundaries with vegetated areas. Many large round potholes, ranging in diameter from a few to several hundred meters, occur within vegetated areas of the marsh. These areas are shallow (< 0.3 m) but may be over 1 m be low the marsh surface in the centers. Large areas of shallow (<0.3 m) estuarine water th at were impounded occur along the estuarine edges of impoundments and often grade into remnant creeks towards the higher marsh. Ditches (2-5 m across) along the impoundment perimeters also contain unvegetated open water habitat, but generally these areas are too deep for foraging by wading birds. The northern portion of the IRL is isolated from ocean inlets and has very low diurnal tidal changes (< 1cm; Smith 1987). In this region, seas onal and wind-driven water level fluctuations are of much gr eater importance (Smith 1987,1993). Hydrology is marked by a high water period from September through November (due to a local seasonal sea-level increase and a summer rai ny season), followed by a gradual decline in water level with the lowest level occurri ng in early spring. These changes greatly influence the depth of water over salt mars h habitat that is connected to the estuary, controlling the extent of marsh surface c overed with water (Trost 1968). A similar pattern occurs within impoundme nts isolated from the estu ary, although water depths are generally greater than in unimpounded salt marsh (Stolen et al. 2002).

PAGE 54

37 T10K T10L T10J T10H T10D T10C T10E SHILOH 3 SHILOH 1 T27B T38I n d i a n R i v e rM o s q u i to L a g o o nA t l a n t i c O c e a nKennedy Space Center / Merritt Island National Wildlife RefugeT10I T10G 05,000 2,500Meters Figure 3.1. Map of study site showing locat ion of study impoundments on the Kennedy Space Center-Merritt Island National W ildlife Refuge. The fixed-station impoundments were: T10K, T10L, T 27B, T38, SHILOH 1 AND SHILOH 3.The random-site impoundments were : T10C, T10D, T10E, T10H, T10J, T10K, AND T10L.

PAGE 55

38 Prey Sampling Throw-trap sampling (Kushlan 1981b) was used to quantify resident marsh fish abundance. This type of sampling gear has been shown to produce accurate estimates of fish abundance in shallow wetlands similar to those in this study (Chick et al. 1992, Jordan et al. 1997, Stevens 2002). To avoid startling fish, researchers approached a sample site by walking slowly and then tossed a 1-m2 throw-trap a distance of 1 to 2 m. Once the trap landed, researchers quickly secured the edges of the trap against the substrate. Fish were then scooped from th e trap using a 40 by 30 cm dip net with 2-mm mesh. Vegetation within the trap was remove d if it impeded movement of the dip net. When the large dip net was scooped 3 times without catching a fish, a 15 by 10 cm dip net with 2-mm mesh was used, which was more effective in scraping along the edges and into the corners of the trap. The sample was completed when the smaller dip net was scooped 3 times without catching a fish. The st andard length (from anterior tip of body to base of the caudal fin) of the first 30 individuals of each species captured in each throw-trap deployment were measured to th e nearest mm. The mass of these fish was estimated using species-specific regression equa tions developed for fish captured in other impoundments on KSC/MINWR (Phil Steven s, USGS, unpublished data). Two sets of impoundments containing salt ma rsh habitat along th e edge of the IRL were sampled. The first set, termed fixe d-station sampling impound ments (Figure 3.1), consisted of 3 pairs of impoundments, chosen to meet the following criteria based on data collected during long-term m onitoring of wading birds on KSC/MINWR (Stolen et. al. 2002): 1) impoundments had to be among those in cluded in at least 5 years of monthly wading bird foraging habitat use surveys, 2) impoundments within pairs must be adjacent, and 3) within pairs the overall m ean density of wading birds observed during

PAGE 56

39 foraging habitat use surveys of one impoundment must have b een at least twice as great as the other. Once the pairs were selected, I identified at least 3 locations within each impoundment that met the following criteria: 1) locations were areas of open water at least 1 ha in size, 2) locations were adjacent to culverts that opened to the estuary 3) locations were adjacent to reli ct tidal creeks. At each of these points, I set up a 3 by 3 cell grid of 100-m2 cells marked with short PVC poles. During each quarterly sampling period, 3 cells were chosen randomly within ea ch grid and samples we re collected with a 1-m2 throw trap. The second set of impoundments, term ed random-station sampling impoundments (Figure 3.1) consisted of 7 impoundments along the Black Point Wildlif e Drive chosen to overlap with a multi-disciplinary study of the effects of impoundment on salt marsh habitat (referred to as the We tlands Initiative, see Brockmeyer 2004 for details). In these impoundments, I used a random sampling design to locate sites to compare fish density between wetland habitat with and without emergent salt marsh vegetation. At each random point, a 1-m2 throw trap was used to sample fi sh at the neares t unvegetated and vegetated habitat to the random point. Unve getated habitat was defined as a flooded area with no emergent marsh vegetation that was at least 2 m in diameter. One throw trap sample was collected at the unvegetated site, then a paired sample location was selected within the nearest flooded vegetated habitat that was contiguous to the open water habitat sampled. Here a throw trap sample was colle cted in vegetated habitat which was paired with the unvegetated sample. Paired vegetated sites were selected w ithin 1 m of the edge of the contiguous open water, and at least 5 m from the location where the open water sample was made. Vegetated habitat was de fined as a flooded area with at least 25%

PAGE 57

40 cover of emergent marsh vegetation and at least 2 m in diameter. If no open water existed within 200 m of the chosen random point, no unvegetated sample was taken. If no vegetated habitat existed within 200 m of the open water sample then no paired vegetated sample was collected. In addition to the fish data, other data recorded were: a GPS position of the sample, water depth, distance through the water at which a researcher could see the tip if a finger, a description of the substrate, surface wate r conditions, weather conditions, presence of submerged aquatic vegetation, and for vegetate d samples the portion of the trap occupied by emergent vegetation. Random-station thro w trap samples were collected quarterly July 2001-July 2002. During each quarterly sampling period, 10-15 points were chosen randomly within each impoundment. The num ber of points sampled in each habitat, season, and impoundment combination is gi ven in Table 3.1. During the study, two impoundments (T10J and T10L) remained hydrologically isolated from the estuary while the other impoundments were connected via culverts open during some periods. Analysis Density of fish for each throw-trap deployment was calculated as the number of individuals of all spec ies removed from the net. Nonparametric tests were used in comparisons of fish densities and biom ass to avoid making assumptions about distributions required for the use of parame tric tests (Conover 1980). I explored patterns of distribution of fish abundance measures (i.e., density and biomass for each prey sampling type) by plotting histograms for each variable. I attempted to use transformations (natural log, s quare root and inverse) to re duce skew and kurtosis within each variable. Patterns of co rrelation between measures of fish abundance (density and biomass) were explored using both Spearmans for raw variables and Pearsons r for

PAGE 58

41 transformed variables. I re port back-transformed means and 95% confidence intervals of abundance within season and impoundment. I calculated the mean length and mean biomass of fish (all species co mbined) for each sample, using all individuals measured. I explored patterns of these measures be tween habitat types using unpaired t-tests including all samples with measurement data I calculated the correlations between sample density and mean length and biomass fish using Spearmans I used information-theoretic model se lection methods (Burnham and Anderson 2002) to explore patterns of fish density as a function of explanatory variables (e.g., season, impoundment, and habitat). This a pproach is based on generating a set of alternative hypothese s that are stated as models relating a single response variable with a set of predictor variables. The key step in this process is in gene rating a meaningful set of candidate models based on biological insight into the system, rather than fitting a series of models that includes various systematic comb inations of predictor variables (e.g., step-wise backward elimination). Info rmation-theoretic mode l selection provides an objective means of compari ng the relative support among the models in the set, given the data. Models were formulated as general linear models (GLMs). Models were considered for interpretation of their parameters if they met the following criteria: 1) AICc of less than 10.0, 2) were included in the set of best supported models with combined Akaike weights of 0.95 (95% conf idence set) and 3) had an evidence ratio relative to the best supported model greater than 0.135 (Burnham and Anderson 2002). For the fixed-station samples, GLMs were f it using fish abundance in sample grids as the response variable and including co mbinations of the explanatory variables Impoundment (6 levels) and Season (3 levels). The set of candidate models included a

PAGE 59

42 model with each of the main ef fects alone testing the hypothesis that each factor was the only one important in explaining the variati on in fish density. The model with both factors was also included based on the hypothesi s that both effects were important in explaining the density of fish in samples. I also included the full factorial model ( main effects and the two-way interaction between f actors) to evaluate the overall fit of the models (Burnham and Anderson 2002). I used ln (fish density+1) as the response variable in GLM analyses because this transformati on produced the most normal distribution of each of the abundance variables. The resulti ng variables were still skewed right, mainly due to presence of many zeros. For the random-station samples, GLMs we re fit using fish abundance as the response variable and including combinations of the explanatory variables Habitat (vegetated or unvegetated), Impoundment (4 le vels) and Season (4 levels). I excluded samples collected in Post-nesting 2001 s eason and impoundments T10C, T10D and T10K from the analysis due to lack of sample s (see Table 3.1). The candidate model set considered included models with each factor al one, models with all pairs of two factors, models with each two-way interaction alone, a model with all two-way interactions, and several models with two-way interactions and the remaining factor as a main effect. The full factorial model including the main effects, all two-way interactions, and the threeway interaction between factor s was evaluated to assess the overall fit of the models (Burnham and Anderson 2002). I used ln(fish density+1) as the res ponse variable in GLM analyses because this transformation pr oduced the most normal distribution of each of the abundance variables. The resulting variab les were still skewed right, mainly due to presence of many zeros.

PAGE 60

43 Results Fixed-Station Sampling One hundred and fifty-five fixed-station throw trap samples were taken during 3 sampling periods in June 2000, July 2000 and January 2001; fish were present in 83 (54%) of these. Eleven species of fish were captured (2,133 individuals); 93% of individuals belonged to the top three species Sheepshead Minnow, Mosquitofish, and Sailfin Molly ( Poecilia latipinna ) (Table 3.2). Other species of nekton were rarely captured during sampling at these sites; a total of 178 Grass Shrimp, ( Paleomontes sp.) were captured in 8 samples. Estimates of fi sh density and biomass were highly correlated (Spearmans = 0.962, n = 155, p < 0.0001; the sample size was 1 less than expected because 2 samples were aggregated due to data recording error). The frequency distribution of fish density was highly skewed due to a large number of zeros (46% of throw samples; Figure 3.2). Information-theo retic model selection resulted in a single model that included both Season and Impoundme nt main effects and had an Akaike weight of 0.92 (Tables 3.33.5). Results from this model indicated that fish density was higher in January 2001 than the other seas ons and that impound ment T38 had higher density than all other impoundments (Figure 3.3). Random-Site Sampling A total of 326 unvegetated and 203 vegetate d points were sampled. In 128 cases no vegetated habitat existed within 200 m of the sample point and thus only an unvegetated sample was taken; in 5 cases no open habitat existed within 200m of the sample point so only a vegetated sample was taken. Non-fish species were captured at 26 of 326 unvegetated sites and 51 of 203 vegetate d sites sampled. Most of the non-fish prey captured were shrimp (Table 3.6). Fish were captured at 174 of the unvegetated

PAGE 61

44 sites and 180 of the vegetated sites. Fifteen species of fish were captured, but over 88% belonged to the top three species (Sheepsh ead Minnow, Mosquitofish, and Sailfin Molly, Table 3.7). There was a weak positive correlation between th e number of fish and the number of shrimp in samples (Spearmans = 0.297, p =<0.0001, n=529). There was a strong positive association between the occurr ence of shrimp and the occurrence fish in samples ( 2 2 = 19.98, p<0.0001) and only 7 of 159 samples had shrimp but no fish. Mean fish density varied substantia lly between impoundments (Figure 3.4). Seasonal densities of fish also varied greatly for both unvegetated (Table 3.8) and vegetated habitats (Tab le 3.9). Unvegetated points were more likely to have no fish than were the paired vegetated sites ( 2 1 = 44.76, p< 0.001). The density of fish was greater in vegetated than unvegetated habitat (W ilcoxon signed-rank test, p < 0.0001, z = -6.94, n=218 pairs); the 95 % confidence interval of the difference between the density of the paired vegetated and unvegetated sites was (1,6) individuals/m2. The mean fish density (individuals/m2) and 95% confidence interval for al l vegetated sites was 8.2 (6.7, 9.9) and for all unvegetated sites 2.0 (1.6, 2.4). The corr elation between the densities of fish in vegetated versus unvegetated paired-sampl es was significantly positive, but weak (Spearmans = 0.300, n = 198, p < 0.0001). Fish biomass also varied greatly by impoundment (Figure 3.5) and by season (Tables 3.10 and 3.11), and showed a similar pattern to density. The mean biomass (g/m2) and 95% confidence interval for all vegetated sites was 3.0 (2.5-3.7) and for all unvegetated sites 1.1 (0.9-1.4). Estimates of fish density and biomass were highly co rrelated for both unvegetated (Spearmans = 0.963, n = 326, p < 0.0001) and vegetated sites (Spearmans = 0.852, n = 203, p < 0.0001). There was also a weak correlati on between the biomass of fish between

PAGE 62

45 vegetated and unvegetate d sites (Spearmans = 0.190, n = 198, p < 0.007). The frequency distribution of fish density was highly skewed for unvegetated samples due to a large number of zeros (47% of unvegetated samples, 19% of vegetated samples) (Figure 3.6). The mean length and mean biomass of individual fish within samples is given in Table 3.12. The mean length of fish (by sample points) was greater at unvegetated sites (24.0 mm, SE=0.74, n=170) than at vegetate d sites (21.3 mm, SE=0.61, n=179), and the difference was significant (unpaired t-te st with unequal va riances: t=2.74, df=331.6, p=0.007). Similarly, the mean biomass per fish at unvegetated sites (0.68 g, SE=0.11, n=170), was greater than that at vege tated sites (0.52 g, SE=0.10, n=179), but the difference was only marginally significant (unpaired t-test with unequal variances: t=1.90, df=334.6, p=0.058). There was no correl ation between the mean length and density of fish at sample points (Spearmans =0-0.035, p=0.519, n=349). Nor was there a correlation between the mean biomass and de nsity of fish at sample points (Spearmans =0-0.061, p=0.257, n=349). The pattern of greate r mean length in unvegetated sites was true for 3 of the 4 most abundant species (Table 3.13). Information-theoretic model selection resu lted in two models with a combined Akaike weight of 0.99(Table 3.14). The be st-supported model included interactions between Habitat and Season, and Habitat a nd Impoundment, and had an Akaike weight of 0.82 (Tables 3.15 and 3.16). Fish density was highest in vegetated habitat in the Postnesting and Winter seasons; this pattern re versed in Pre-nesting and Nesting seasons (Figure 3.7). Marginal means of fish de nsity within impoundments are shown in Figure

PAGE 63

46 3.8. The model with all three two-way interaction terms also had some support with an Akaike weight of 0.17 (Tables 3.14, 3.17 and 3.18).

PAGE 64

47 Table 3.1. Number of points sampled during random-station fish sampling in unvegetated (unveg.) and vegetated (veg.) habita t during 5 seasons in 7 impoundments. impoundmentunveg.veg.unveg.veg.unveg.veg.unveg.veg.unveg.veg. T10C15910101010108 T10D121099105 T10E111010101109 T10H15141010101041010 T10J1515101010101021010 T10K1215 T10L131515141010103109 Post 2002 Post 2001Winter 2001Pre 2002Nesting 2002 Table 3.2. Occurrence of fish by species in throw-trap samples in 6 fixed-station sampling impoundments on KSC/MINW R measured June 2000 through January 2001. The first 2 columns give the number of samples (plots) with each species present, and the species rank by occurrence in samples. The next 2 columns give the total number of each species captured for all samples, and the species rank by total number captu red. The last 4 columns give mean length, SE and range of lengths with in species, and sample sizes for measurements. Species Number of Plots Rank by Plot Total number Rank by number Mean length (mm)SERange (mm)n Cyprinodon variegatus 5911483124.40.30(7-114)823 Elops saurus 3815739.21.41(32-51)15 Fundulus confluentus 2931128.04.00(24-32)2 Gambusia holbrooki 352358218.40.40(8-50)210 Gobiosoma robustum 11011238.01 Lucania parva 7621623.11.34(13-34)20 Menidia beryllina 10571434.31.01(21-55)68 Microgobius gulosus 11429531.81.55(20-44)24 Mugil cephalus 6710844.36.84(23-84)10 Poecilia latipinna 213132324.90.70(14-49)125 Trinectes maculatus 2941033.01.78(30-38)4 Unkown fish3869 no fish at site7272

PAGE 65

48 Table 3.3. Fixed-station fish sampling model selection results for GLM analysis with ln[1+fish density] as response variab le and season (3 levels, S) and impoundment (6 levels, L) as explanatory variables. Model MLE 2kAICc AICc n/k wiL, S1.10927.130.0060.92 S1.55432.024.89130.08 L1.64742.5715.4480.00 L*S0.721943.7616.6330.00 Table 3.4. ANOVA table for GLM with fixedstation ln(1+fish de nsity) as response variable and explanatory va riables season and location. Source Type III Sum of Squaresdf Mean SquareFSig. Corrected Model 51.4a77.355.680.00 Intercept114.631114.6388.660.00 Season28.52214.2611.030.00 Location23.9154.783.700.01 Error58.18451.29 Total234.9953 Corrected Total109.6052 a R2 = 0.47 (Adjusted R2 = 0.39)

PAGE 66

49 Table 3.5. Parameter estimates for GLM with fi xed-station ln(1+fish density) as response variable and explanatory va riables season and location. Parameter SE 95% CI Intercept2.480.47(1.52 3.43) season = JAN 20011.350.38(0.58 2.11) season = JUL 2000-0.330.39(-1.12 0.46) season = JUN 2000 0alocation = Shiloh 1-1.460.52(-2.51 -0.42) location = Shiloh 3-1.280.57(-2.43 -0.13) location = T10K -2.380.59(-3.56 -1.19) location = T10L -1.720.55(-2.84 -0.61) location = T27B -1.080.55(-2.20 0.03) location = T38 0a a This parameter is set to zero because it is redundant. Table 3.6. Occurrence of non-fish nekton in random-site throw-trap samples in 7 impoundments on KSC/MINWR measured July 2001 through July 2002. Table entries give the number of samp les (sites) with each species and total number captured by habitat. The total number of samples was 326 in unvegetated and 203 in vegetated habitats. SpeciesUnvegetatedVegetatedUnvegetatedVegetated Grass shrimp ( Palaemontes sp.) 234870248 Snapping shrimp ( Alpheus sp. ) 2252 Crayfish ( Procambrus sp.) 1212 Unknown tadpole33 Unkown Isopod11 Number of SitesTotal number

PAGE 67

50Table 3.7. Occurrence of fish by species in th row-trap samples in wetland habitat with and without emergent vegetation in 7 ran domsite impoundments on KSC/MINWR measured July 2001 through July 2002. The first 2 columns give the number of samples (sites) with each species, and th e next 2 give the species rank by occurr ence in samples (sites). The next 2 columns give the total number of each species captured for a ll samples, and the last 2 give the species rank by total number captured. SpeciesUnvegetatedVegetatedUnvegetatedVegetatedUnvegetatedVegetatedUnvegetatedVegetated Cynoscion nebulosus 2829 Cyprinodon variegatus 1251261172884923 Floridichthys carpio 1312718137 Fundulus confluentus 101175101485 Fundulus sp. 529862109 Gambusia holbrooki 7711522707133131 Gobiosoma bosc 21119411110 Gobiosoma robustum 31109251710 Gobiosoma sp. 72888498 Jordanella floridae 112113 Lucania parva 49694427525744 Menidia beryllina 166276 Menidia sp. 31109311210 Microgobius gulosus 22956521256 Mugil cephalus 310411 Poecilia latipinna 6410633763109712 Syngnathus scovelli 112113 Syngnathus sp. 19110 Unkown fish2211832129 no fish at site1522315223 Number of SitesRank by siteRank by number Total number

PAGE 68

51Table 3.8. Density of fish (individuals/m2) in wetland habitat without emergent vegetation in 7 random-site impoundments on KSC/MINWR measured July 2001 through July 2002. Post-nesting 2001 Winter 20012002 Pre-nesting 2002 Nesting 2002 Post-nesting 2002 Impoundment mean 95% CI mean95% CI mean95% CI mean95% CI mean95% CI T10C 0.2 (0.0-0.6) 0.3 (0.0-1.1) 0.0 (0.0 -0.0) 0.2 (0.0-0.8) 0.1 (0.0-0.6) T10D 0.3 (0.0-0.7) 0.1 (0.0-0.3) 0.4 (0.0 -1.2) 0.2 (0.0-0.7) 0.1 (0.0-0.3) T10E --a 1.8 (0.3-5.0) 0.5 (0.1-1.1) 1.2 (0.0-3.9) 0.1 (0.0-0.3) T10H 0.4 (0.0-1.0) 1.0 (0.0-3.0) 3.5 (1.0 -9.4) 1.5 (0.3-3.6) 1.8 (0.7-3.6) T10J 2.7 (1.3-5.2) 5.9 (2.0-15.2) 16.1 (5.0-48.0) 9.4 (5.1-16.9) 5.5 (2.4-11.6) T10K 4.6 (1.4-11.7) --a --a --a --a T10L 4.1 (1.4-10.0) 7.3 (3.9-13.1) 18.0 (6.0-51.0) 23.3 (5.4-91.5) 5.5 (1.9-13.7) a Not sampled due to change in impoundment selection following Po st-nesting season 2001.

PAGE 69

52Table 3.9. Density of fish (individuals/m2) in wetland habitat with emergent ve getation in 7 random-site impoundments on KSC/MINWR measured July 2001 through July 2002. Post-nesting 2001 Winter 2001-2002 Pre-nesting 2002 Nesting 2002 Post-nesting 2002 Impoundment mean 95% CI mean95% CI mean95% CI mean95% CI mean95% CI T10C 4.9 (1.5-13.0) 3.1 (1.3-6.3) --b --b 10.0 (3.4-26.1) T10D --b --b --b --b 2.1 (0.5-5.5) T10E --a 14.6 (7.4-28.1) 0.0 6.0 25.2 (12.450.4) T10H 0.3 (0.0-0.7) 5.3 (1.9-12.7) 0.0 1.8 (-0.8-34.3) 17.1 (8.4-33.9) T10J 3.7 (1.9-6.8) 22.7 (10.3-48.4) 11.5 (5.3-24.0) 16.2 (-1.0-289179.4) 48.2 (29.179.3) T10K 10.1 (6.7-15.0) --a --a --a --a T10L 5.7 (1.8-15.0) 23.1 (12.7-41.3) 10.1 (4.2-22.5) 2.2 (-0.8-43.1) 7.8 (2.9-18.8) a Not sampled due to change in impoundment selection following Po st-nesting season 2001. b Not sampled due to lack of flooded vegetated habitat during sample period

PAGE 70

53Table 3.10. Biomass (g/m2) of fish in wetland habitat w ithout emergent vegetation in 7 random-site impoundments on KSC/MINWR measured July 2001 through July 2002. Post-nesting 2001 Winter 20012002 Pre-nesting 2002 Nesting 2002 Post-nesting 2002 Impoundment mean 95% CI mean95% CI mean95% CI mean95% CI mean95% CI T10C 0.1 (0.0 0.2) 0.2 (0.0 0.5) 0.0 0.0 (0.0 0.1) 0.1 (0.0 0.2) T10D 0.1 (0.0 0.3) 0.2 (0.0 1.0) 0.4 (0.0 1.1) 0.6 (0.0 2.8) 0.0 (0.0 0.1) T10E --a 0.9 (0.1 2.3) 0.8 (0.1 2.0) 0.5 (0.0 1.9) 0.0 (0.0 0.1) T10H 0.2 (0.0 0.5) 0.7 (0.0 1.9) 1.9 (0.3 5.4) 0.4 (0.0 1.0) 0.2 (0.0 0.6) T10J 1.0 (0.2 2.3) 2.6 (0.6 6.9) 9.7 (3.0 28.0) 3.3 (1.3 7.2) 2.6 (0.9 6.0) T10K 2.5 (0.5 7.0) --a --a --a --a T10L 1.6 (0.5 3.4) 3.9 (1.3 9.7) 6.9 (1 .9 20.1) 9.3 (1.9 35.2) 4.2 (1.4 10.3) a Not sampled due to change in impoundment selection following Po st-nesting season 2001.

PAGE 71

54Table 3.11. Biomass (g/m2) of fish in wetland habitat with emergent ve getation in 7 random-site impoundments on KSC/MINWR measured July 2001 through July 2002. Post-nesting 2001 Winter 20012002 Pre-nesting 2002 Nesting 2002 Post-nesting 2002 Impoundment mean 95% CI mean95% CI mean95% CI mean95% CI mean95% CI T10C 2.5 (0.6 6.4) 2.5 (0.6 6.5) --b --b 3.0 (0.7 8.6) T10D --b --b --b --b 3.6 (0.6 11.9) T10E --a 7.6 (3.9 14.4) 0.0 (0.0 0.0) 0.1 (0.0 0.0) 8.8 (4.4 16.6) T10H 0.2 (0.0 0.4) 1.6 (0.6 3.1) 0.0 (0.0 0.0) 1.2 (0.0 18.3) 4.5 (1.3 12.2) T10J 0.4 (0.2 0.7) 10.6 (6.9 16.1) 3.7 (1.2 8.9) 2.6 (0.0 42900) 13.7 (7.6 24.1) T10K 2.6 (0.7 6.3) --a --a --a --a T10L 3.1 (1.4 6.1) 5.5 (2.8 10.1) 2.9 (1.1 6.4) 0.5 (0.0 3.0) 1.8 (0.4 4.5) a Not sampled due to change in impoundment selection following Po st-nesting season 2001. b Not sampled due to lack of flooded vegetated habitat during sample period.

PAGE 72

55 Table 3.12. Mean length and mean biomass of individual fish captured within randomsite throw-trap samples by species. Data is included for all samples for which these measures were made. SpeciesMean length (mm ) SEnRange (mm) Cynoscion nebulosus 73.03.002(70-76) Cyprinodon variegatus 21.30.221413(5-55) Floridichthys carpio 33.86.019(19-77) Fundulus confluentus 24.33.1018(10-52) Fundulus spp. 42.27.186(18-65) Gambusia holbrooki 19.70.141512(1-40) Gobiosoma bosci 24.42.505(19-33) Gobiosoma microgulosus 26.01 Gobiosoma robustum 29.01.3926(19-50) Gobiosoma spp. 24.31.909(16-34) Jordanella floridae 26.01 Lucania parva 21.30.27465(7-44) Menidia beryllina 36.31.8324(15-50) Menidia spp. 36.310.884(15-56) Microgobius gulosus 30.61.2460(17-57) Mugil cephalus 123.519.724(74-160) Poecilia latipinna 23.60.231353(6-65) Syngnathus scovelli 82.01 Table 3.13. Mean length of indi vidual fish in 2 habitat type s within random-site throwtrap samples, listed by species. Data is included for all samples for which these measures were made. SPECIES Mean Length (mm)SEn Range (mm) Mean Length (mm)SEn Range (mm) Cyprinodon variegatus 21.80.34631(5-55)20.90.30782(7-50) Gambusia holbrooki 21.10.25444(1-40)19.10.171068(7-39) Lucania parva 21.00.38240(7-44)21.60.38225(9-39) Poecilia latipinna 27.50.32532(6-62)21.00.28821(7-65) Unvegetated SitesVegetated Sites

PAGE 73

56 Table 3.14. Random-site model selection re sults for GLM analysis with ln[fish density+1] as response variable and ha bitat type (unvegetated/vegetated, H), season (5 levels, S), and impoundment (6 levels, I) as explanatory variables. Model MLE 2kAIC AIC AICc AICc n/kwi H*I, H*S1.091553.690.0055.520.0018.530.82 H*I, H*S, S*I1.022453.950.2758.703.1811.580.17 H*I1.21970.9817.2971.6516.1330.890.00 H*S, I1.211276.0822.3977.2621.7423.170.00 H*I, S1.211276.3622.6877.5422.0223.170.00 H,I1.36698.1944.5098.5042.9846.330.00 H,I,S1.369104.0250.33104.6949.1730.890.00 I*S, H1.2718102.9749.28105.6150.0915.440.00 H*S1.539137.0983.40137.7682.2430.890.00 I1.605140.4386.74140.6585.1355.600.00 I,S1.588143.0389.35143.5788.0534.750.00 I*S1.4717141.6888.00144.0488.5216.350.00 H1.653145.8592.17145.9490.4292.670.00 H,S1.656150.9997.30151.3095.7846.330.00 S1.925191.83138.15192.06136.5455.600.00 H*I*S (global model)0.983156.022.3364.098.578.970.01 Table 3.15. ANOVA table for GLM with ra ndom-site ln(fish density+1) as the explanatory variable and habita t*season and habitat*impoundment interactions included. Source Type III Sum of Squaresdf Mean SquareFSig. Corrected Model 237.6a1318.2815.940.00 Intercept830.551830.55724.300.00 season4.2731.421.240.30 impoundment62.77320.9318.250.00 habitat35.69135.6931.120.00 impoundment habitat32.56310.859.460.00 season habitat32.90310.979.560.00 Error302.732641.15 Total1713.00278 Corrected Total540.33277 a R2 = 0.44 (Adjusted R2 = 0.41)

PAGE 74

57 Table 3.16. Parameter estimates for GLM with random-site ln(fish density+1) as the explanatory variable and habita t*season and habitat*impoundment interactions. Parameter SE 95% CI Intercept2.340.20(1.94 2.74) season = nesting 20020.380.23(-0.08 0.83) season = post-nesting 2002-0.220.23(-0.68 0.24) season = pre-nesting 20020.490.23(0.04 0.95) season = winter 2001 0aimpoundment = T10E-1.920.23(-2.37 -1.46) impoundment = T10H-1.460.23(-1.92 -1.00) impoundment = T10J-0.260.23(-0.72 0.20) impoundment = T10L 0ahabitat = vegetated0.380.30(-0.21 0.97) habitat = unvegetated 0aimpoundment = T10E habitat = vegetated2.060.39(1.30 2.82) impoundment = T10E habitat = unvegetated 0aimpoundment = T10H habitat = vegetated0.940.38(0.21 1.68) impoundment = T10H habitat = unvegetated 0aimpoundment = T10J habitat = vegetated0.860.35(0.17 1.55) impoundment = T10J habitat = unvegetated 0aimpoundment = T10L habitat = vegetated 0aimpoundment = T10L habitat = unvegetated 0aseason = nesting 2002 habitat = vegetated-1.510.44(-2.38 -0.63) season = nesting 2002 habitat = unvegetated 0aseason = post-nesting 2002 habitat = vegetated0.520.33(-0.14 1.17) season = post-nesting 2002 habitat = unvegetated 0aseason = pre-nesting 2002 habitat = vegetated-1.050.39(-1.81 -0.29) season = pre-nesting 2002 habitat = unvegetated 0aseason = winter 2001 habitat = vegetated 0aseason = winter 2001 habitat = unvegetated 0a a This parameter is set to zero because it is redundant.

PAGE 75

58 Table 3.17. ANOVA table for GLM with ra ndom-site ln(fish density+1) as the explanatory variable and habita t*season, habitat*impoundment and season*impoundment interactions. Source Type III Sum of Squaresdf Mean SquareFSig. Corrected Model 256.3a2211.6510.460.00 Intercept706.521706.52634.330.00 season4.3431.451.300.28 impoundment61.91320.6418.530.00 Habitat25.91125.9123.270.00 impoundment Habitat20.2336.746.050.00 season Habitat34.34311.4510.280.00 season impoundment18.7192.081.870.06 Error284.022551.11 Total1713.00278 Corrected Total540.33277 a R2 = 0.47 (Adjusted R2 = 0.43)

PAGE 76

59 Table 3.18. Parameter estimates for GLM with random-site ln(fish density+1) as the explanatory variable and habita t*season, habitat*impoundment and season*impoundment interactions. Paramete r SE 95% CI Intercept2.3920.244(1.91 2.87) season = nesting 20020.5630.382(-0.19 1.32) season = post-nesting 2002-0.8470.349(-1.53 -0.16) season = pre-nesting 20020.7070.372(-0.02 1.44) season = winter 20010aimpoundment = T10E-1.6080.362(-2.32 -0.90) impoundment = T10H-1.7780.363(-2.49 -1.06) impoundment = T10J-0.510.353(-1.21 0.19) impoundment = T10L0ahabitat = vegetated0.4940.302(-0.10 1.09) habitat = unvegetated0aimpoundment = T10E habitat = vegetated1.7320.414(0.92 2.55) impoundment = T10E habitat = unvegetated0aimpoundment = T10H habitat = vegetated0.8250.395(0.05 1.60) impoundment = T10H habitat = unvegetated0aimpoundment = T10J habitat = vegetated0.8390.354(0.14 1.54) impoundment = T10J habitat = unvegetated0aimpoundment = T10L habitat = vegetated0aimpoundment = T10L habitat = unvegetated0aseason = nesting 2002 habitat = vegetated-1.5070.444(-2.38 -0.63) season = nesting 2002 habitat = unvegetated0aseason = post-nesting 2002 habitat = vegetated0.5050.328(-0.14 1.15) season = post-nesting 2002 habitat = unvegetated0aseason = pre-nesting 2002 habitat = vegetated-1.340.425(-2.18 -0.50) season = pre-nesting 2002 habitat = unvegetated0aseason = winter 2001 habitat = vegetated0aseason = winter 2001 habitat = unvegetated0aseason = nesting 2002 impoundment = T10E-0.5180.549(-1.60 0.56) season = nesting 2002 impoundment = T10H-0.1830.518(-1.20 0.84) season = nesting 2002 impoundment = T10J0.0110.533(-1.04 1.06) season = nesting 2002 impoundment = T10L0aseason = post-nesting 2002 impoundment = T10E0.3530.457(-0.55 1.25) season = post-nesting 2002 impoundment = T10H1.2880.457(0.39 2.19) season = post-nesting 2002 impoundment = T10J0.9310.457(0.03 1.83) season = post-nesting 2002 impoundment = T10L0aseason = pre-nesting 2002 impoundment = T10E-1.0760.566(-2.19 0.04) season = pre-nesting 2002 impoundment = T10H0.1890.567(-0.93 1.31) season = pre-nesting 2002 impoundment = T10J0.0960.453(-0.80 0.99) season = pre-nesting 2002 impoundment = T10L0aseason = winter 2001 impoundment = T10E0aseason = winter 2001 impoundment = T10H0aseason = winter 2001 impoundment = T10J0aseason = winter 2001 impoundment = T10L0a a This parameter is set to zero because it is redundant.

PAGE 77

60 0 10 20 30 40 50 60 70 80 01-56-1011-2021-5051-100100+Density (ind. / m2)Samples 0 10 20 30 40 50 60 70 80 01-56-1011-2021-5051-100100+Density (ind. / m2)Samples Figure 3.2. Frequency of 1-m2 throw trap samples (tosses) within 7 density categories in fixed-station fish samples from impoundments on KSC/MINWR, sampled June 2000 January 2001. 0 10 20 30 40 50 60 70 80 90 100 Shiloh 1Shiloh 3T10KT10LT27BT38 JUN_00 JUL_00 JAN_01Fish / m2 0 10 20 30 40 50 60 70 80 90 100 Shiloh 1Shiloh 3T10KT10LT27BT38 JUN_00 JUL_00 JAN_01Fish / m2 Figure 3.3. Mean fish densities measured durin g fixed-station fish sampling was greatest in January 2001. Estimates based on da ta from 3 sampling periods, June 2000 January 2001 within 6 impoundments on KSC/MINWR.

PAGE 78

61 Mean density (individuals / m2) 0 5 10 15 20 25 30 T10CT10DT10ET10HT10JT10KT10L Vegetated Unvegetate d Mean density (individuals / m2) 0 5 10 15 20 25 30 T10CT10DT10ET10HT10JT10KT10L Vegetated Unvegetate d Figure 3.4. Mean fish density was always gr eater for vegetated versus paired nearby unvegetated sites based on random-site sampling from 5 quarterly sampling periods, July 2001 July 2002. Error bars give 95% confidence intervals of back-transformed means. Mean biomass (g / m2) 0 2 4 6 8 10 12 14 T10CT10DT10ET10HT10JT10KT10L Vegetated Unve g etate d Mean biomass (g / m2) 0 2 4 6 8 10 12 14 T10CT10DT10ET10HT10JT10KT10L Vegetated Unve g etate d Figure 3.5. Mean fish biomass was usually gr eater for vegetated versus paired nearby unvegetated sites. Estimates based on ra ndom-site sampling data from 5 quarterly sampling periods, July 2001 July 2002. Erro r bars give 95% confidence intervals of back-transformed means.

PAGE 79

62 0 20 40 60 80 100 120 140 160 01-56-1011-2021-5051-100100+ Density (ind. / m2)Samples unveg count veg count Figure 3.6. Frequency of 1-m2 throw trap samples (tosses) within 7 density categories in unvegetated and vegetated flooded habita t. Samples taken July 2001 July 2002 in random-sample impoundments. 0 5 10 15 20 25 30 Winter 2001-02Pre-nesting 2002Nesting 2002Post-nesting 2002 Vegetated UnvegetatedFish / m2 0 5 10 15 20 25 30 Winter 2001-02Pre-nesting 2002Nesting 2002Post-nesting 2002 Vegetated UnvegetatedFish / m2 Figure 3.7. Estimated marginal means of fish density (back-transformed) in vegetated and unvegetated habitats within 4 im poundments on MINWR/KSC by season. Estimates are from the GLM with random-site ln(fish density+1) as the explanatory variable and habita t*season and habitat*impoundment interactions included.

PAGE 80

63 0 5 10 15 20 25 30 T10ET10 H T10JT10L Vegetated UnvegetatedFish / m2 0 5 10 15 20 25 30 T10ET10 H T10JT10L Vegetated UnvegetatedFish / m2 Figure 3.8. Estimated marginal means of fish density (back-transformed) in vegetated and unvegetated habitats within 4 impoundments on MINWR/KSC by impoundment. Estimates are from the GLM with random-site ln(fish density+1) as the explanatory variable and habitat*season and habitat*impoundment inte ractions included.

PAGE 81

64 Discussion The density of fish measured in this st udy was similar to other studies within the northern IRL (Schooly 1980, Stevens 2002, Brockm eyer 2004). The density of fish for most sites that have been sampled in this system was higher than levels reported for vegetated habitats of the Ev erglades, a nearby, ecologically similar area which also has a large population of wading birds (Loftus and Eklund 1994, Trexler et al 2002). Turner et al (1999) pointed out that the Everglades is unusual among tropical wetland systems in having unusually low fish standing stocks. Ho wever, the density of resident fish in impounded wetlands in the northern IRL was often considerably lower than the 20 individuals/m2 considered by Trexler et al. (2003) to be the historic level of small marsh resident fish in the Everglades ecosystem. This was especially true of the unvegetated habitats (Table 3.8). Other systems where wading birds forage ha ve also been observed to have much higher levels of prey density than observed in northern IRL impoundments. For example, in tidally replenished isolated pools in a New Jersey salt marsh the density of fish was an order of magnitude higher than that typical ly observed in this study (Master 1992, Master et al. 1993). The relatively lower level of prey density observed in the northern IRL could have implications for the foraging eco logy of piscivorous wading birds which may rely on concentrations of prey for suc cessful foraging and nesting (Kushlan 1976b, Bancroft et al. 1994, Frederick and Sp alding 1994, Gawlik 2002). For some impoundments (e.g., T10 E, T10L and T10 J, Table 3.9) fish densities in vegetated habitat exceeded 20 individuals/m2 in some seasons. Thus, the potential exists for foraging wading birds to locate higher prey concentrations within this system.

PAGE 82

65 The impounded salt marsh habitats in th e northern Indian River Lagoon are unique in many ways (Montague et al. 1987). Of par ticular relevance to the resident small fish community is the presence of the perimeter dikes which dampen the effects on impounded wetlands of hydrologic changes in th e estuary. Compared to the adjacent estuary, water level is more stable in impoundments and impounded marshes are flooded deeper than native marshes (personal obs ervation). The resu lting longer hydroperiod wetlands should support larger populations of small marsh-resident fish (e.g., Sheepshead Minnow, Mosquitofish, and Sailfin Molly) than would shorter hydroperiod unimpounded marshes (Loftus and Eklund 1994, Gilmore 1998, Trexler et al. 2002). In addition, the perimeter dikes serve as a barrier to predator y fish, further increasi ng the standing stocks of small fish within impoundments (Gilmor e 1998, Stevens 2002). Abundance of prey has been suggested as an explanation of why impounded wetland habitat in the northern IRL is attractive to foraging wading bird s (Breininger and Smith 1990, Schikorr and Swain 1995, Smith and Breininge r 1995, Stolen et al. 2002) Within northern IRL impoundments, there wa s a tendency for habitats in closed impoundments (T10L and T10J) to have higher de nsities of fish than those connected via culverts open to the estuary during the study period (Brockmeyer 2004). The closed impoundments had higher water levels thr oughout the study period, especially during times when the other impoundments we re drying (Appendix A). Thus, longer hydroperiods, isolation from predatory fis h, or both factors may explain the higher standing stocks of small marsh resident fi sh observed in the closed impoundments. Densities of fish were generally higher in sites with emergent salt marsh vegetation than sites without such vegetation. Highe r density of fish in vegetated versus

PAGE 83

66 unvegetated areas has been noted in other sh allow systems containing mixtures of both habitat types (Rozas and Odum 1988). Areas with vegetative struct ure may be attractive to fish because of protection against predat ors (Werner et al. 1983) and sometimes also because they contain more abundant food re sources (McIvor and Odum 1988, Rozas and Odum 1988). There was also a marked seas onal pattern to abundan ce of fish between sites with and without emergent salt ma rsh vegetation within impoundments, and the pattern suggested that some ma rsh resident fish may move in to vegetated habitats when these sites become flooded in late fall and winter. In another impoundment in the northern IRL, Stevens (2002) demonstrated th at marsh resident fish (e.g., Sheepshead Minnow, Mosquitofish, and Sailfin Molly) moved from the estuary edge to the vegetated marsh surface as rising water levels fl ooded these areas in late summer. While density of prey is obviously one important factor determining foraging success, others such as the mean size of prey may also determine the suitability of wading bird foraging habitat (Trexler et al. 1994, Werner et al. 2001 ). I found that while prey density was higher in vegetated sites, unvegetate d sites had larger prey. This could have implications for wading birds since larger pr ey have higher energy pe r capture effort and are perhaps energetically superior prey items There was also a difference in species composition between habitats with fewer Sh eepshead Minnow in vegetated sites than unvegetated. These differences in the sp ecies composition and mean size of prey between vegetated and unvegetated sites mi ght be important factors in determining foraging habitat use of piscivorous wadi ng birds in this system (Chapter 4). In the northern IRL, there was a high leve l of variation in the abundance of fish between impoundments in both vegetated and u nvegetated habitats (Figure 3.4). This

PAGE 84

67 suggests that wading bird prey is patchy at the among-impoundments sp atial scale. There was also evidence that prey was highly cl umped at sampling sites within both habitat types (Figures 3.2 and 3.6), suggesting that prey distribution was also patchy at the within-impoundment spatial scale. Fina lly, there was also considerable seasonal variability in both habitat types within impoundments (Tables 3.8-3.11). These patterns of prey distribution have important implicat ions for wading bird foraging habitat use within this system (Chapter 4).

PAGE 85

68 CHAPTER 4 EFFECTS OF HABITAT STRUCTURE AND PREY DISTRIBUTION ON WADING BIRD FORAGING HABITAT USE Introduction General ecological theory predicts that in dividuals should preferentially forage in habitats that provide higher le vels of resources contributing to individual fitness (Krebs and Kacelnik 1991). This leads to the pred iction that, all factors being equal (e.g., competitor and predator density, absence of te rritorial exclusion), foragers will select habitat based on their ability to extract crucia l resources from those habitats instead of foraging in other habitats (MacArthu r and Pianka 1966, Charnov 1976, Rosenzweig 1981). Foraging habitat selec tion and use by long-legged wa ding birds (Ciconiiformes) is of ecological interest because the ability of parents to secure food for their broods has been linked to reproductive success (Powell 1983; Hafner et al. 1986, 1993) and foraging success affects survival of both juvenile a nd adult wading birds (Frederick and Spalding 1994). Thus, foraging ecology is directly relate d to fitness, and hence, to factors that control population trends. Despite a large volume of published studies on wading birds (for recent reviews see Kushlan 1978b, Bildstein 1997) several key que stions regarding their foraging ecology remain unresolved (Erwin 1983a, Bildstein 1997, Kushlan 1997). Foremost among these are how wading birds locate their prey resour ces, and how they exploit those resources once located. Answering such questions is challenging because an individuals foraging site selection, and ultimately foraging succe ss, depends on factors occurring at different

PAGE 86

69 scales (Chapter 2, Johnson 1980), and often th e connections between scales is not well understood or documented. Understanding how wa ding birds select fora ging habitat is of particular conservation value in systems that have been modi fied in ways that affect wetland functions that determine habitat qual ity for wading birds (C hapter 3). Here I differentiate between the terms habitat selection which refers to the behavioral mechanisms leading to preferential use of some habitats over othe rs (Garshelis 2000), and habitat preference which refers to the proportional use of a habitat compared with its availability (Manly et al. 2002). Several studies have demonstrated a pos itive relationship between prey abundance and wading bird foraging rates (Kahl 1964, Er win et al. 1985, Draulans 1987, Cezilly et al. 1990). In shallow wetland systems, hydrol ogy is often the principal control on the abundance of wading bird prey populations via influences on reproductive cycles and access to wetlands (Harrington and Ha rrington 1961, Kahl 1964, Kushlan 1976b, Gilmore et al. 1982). A good example can be se en in the Everglades ecosystem in south Florida where long periods of inundation may increase fish standing stocks (Loftus and Eklund 1994). Conversely, in this same system occasional droughts may reduce piscivorous predation pressure and release nutrients, creating conditions that allow small marsh fish populations to greatly increase (Loftus and Eklund 1994, Frederick and Ogden 2001). In, managed systems, such hydrological patterns can be disrup ted with negative effects on populations of wading birds and th eir prey (Frederick and Ogden 2003). More recently, research has focused on fact ors affecting prey availability, rather than density, as paramount in determining piscivorous wading bird foraging success (Frederick and Loftus 1993, Gawlik 2002). Hydr ologic cycles for example, can affect the

PAGE 87

70 availability of wading bird pr ey through water depth; wading birds do not dive and must be able to physically reach their prey to be able to cap ture it (Kushlan 1976b, Custer and Osborn 1978, Powell 1987, Gawlik 2002). Wa ter quality (e.g., dissolved oxygen, temperature) also has a potential impact on pr ey behavior and thus availability (Kersten et al. 1991, Frederick and Loftus 1993). H ydrology is also a determining factor influencing vegetative compositi on, distribution and density. Vegetation attributes (e.g., cover, structure) affect habitat quality fo r both wading birds and th eir prey, and greatly affect prey availability. Many authors have observed that wading birds preferentially forage in sites lacking emergent vegetation over adjacent vegetated sites (Kersten et al 1991, Hoffman et al. 1994, Chaves-Ramirez and Slack 1995, Schikorr and Swain 1995, Smith et al. 1995, Smith and Breininger 1995, Surdick 1998, Bancro ft et al. 2002, Gawlik 2002, Stolen et al. 2002). In some cases, this may be because vegetative structure can influence wading bird vulnerability to predators by reduci ng visibility (Caldwell 1986). Another explanation is that increased structure greatly inhibits a foragers ability to locate and capture prey. The relationships between these factors are complex (see Chapter 1, Figure 1.1) and thus decisions by wading birds rega rding selection of foraging sites involve tradeoffs. For example, in choosing between tw o habitats with differe nt prey density, an individual must tradeoff between factors e ffecting prey availab ility (water depth, vegetative structure, presence of other foragers) and other fact ors including availability of information (e.g., locations of other foragers ), relative predation risk, and the potential size of prey between sites.

PAGE 88

71 In Chapter 2, I showed that nesting wadi ng birds preferentially used impounded salt marshes in the northern Indian River Lagoon wa tershed. There, birds foraged within 13 km of breeding colonies, well within the ma xima reported for other colonies. These results suggested that resources with in impounded habitat met their energetic requirements without their havi ng to engage in long-distance forays. Birds flew from colonies either in groups or by themselves depending on species, but almost invariably joined foraging flocks when landing. In this chapter, I continue my assessment of habitat selection by waders in the northern Indian River Lagoon by focusing on a smaller spatial scale, namely, among and within impoundments. In Chapter 3 I showed that the density of fish was greater in sites with emergent wetl and vegetation, but fish were larger in sites without emergent vegetation. Previous work demonstrated that within impoundments of the northern Indian River La goon, piscivorous wading birds predominantly use shallow flooded habitat without emergent vegetati on, and to a lesser ex tent, shallow flooded wetlands with low stature, salt-tolerant plants (Stolen et al. 2002). However, it is unclear whether prey abundance or other factors (e .g. habitat structure, hydrology) are the key determinants of habitat selection. In this chapter, I tested the hypothesis that wading birds prefer wetland habitat without emergent vegetation to habitat with emergent vegetation, and also examined their preference for the interface (e dge) between these habitat types. Next, to better understand factors that influence wading bird choice of foraging habitat, I modeled wading bird density as a f unction of prey density, habitat, season, hydrology and management. This allowed comparison of th e relative importance of these factors in determining wading bird habitat preference. Finally, I described the size and composition

PAGE 89

72 of foraging flocks within impoundments becau se their presence may have proximally influenced habitat selection (e.g., visual st imulus of foraging individuals), and also because, depending on prey density and ha bitat characteristic s, flock size and composition might affect foraging success (Chapter 5). Methods The study site consisted of areas of impounded salt marsh habitat on the 55,000 ha Kennedy Space Center-Merritt Island Nationa l Wildlife Refuge (KSC/MINWR). This site is located in the nort hern portion of the Indian River Lagoon system (IRL), a subtropical estuary which is an important site for wading birds on the southeastern Atlantic coast of North America (Schi korr and Swain 1995, Sewell et al. 1995). The northern portion of the IRL is isolated from ocean inlets and has very low diurnal tidal changes (< 1cm; Smith 1987). In this region, seasonal and winddriven water level fluctuations are of much greater importanc e (Smith 1993). Habitat within impoundments is predominantly a heterogeneous mixture of open water and vegetated cover types, with tall marsh grass (e.g., Spartina bakeri ) and short marsh vegetation (e.g., Distichlis spicata, Batis maritima ) predominating in vegetated areas (Schmalzer 1995). Three sets of data were colle cted to address the objectives of this work. These were the following: 1) wading bird density a nd distribution, 2) wa ter levels within impoundments, and 3) fish density and dist ribution. Below I describe the sampling protocols for the first 2 data types. Samp ling protocols and a deta iled summary of fish data are presented in Chapter 3. I drew upon the fish data to test hypotheses concerning habitat selection by waders in this chapter (see below). All data types were collected in 9 study impoundments located on KSC/MINWR (Figure 4.1); 7 of these impoundments overlapped with the Wetlands Initiative, a simultaneous multi-disciplinary study of the

PAGE 90

73 effects of hydrologic management on salt marsh habitat (Brockmeyer 2004). The management of these impoundments was fixed during the period of my study. Therefore, I could evaluate, via comparative tests, th e effects of different management on some wetland functions. Management types (for mo re details see Brockmeyer 2004) were the following: 7. 1) Open a management strategy in wh ich impoundment culverts are left open throughout the year allowing wate r levels to fluctuate with the levels in the estuary. Impoundments T10C and T10H. 8. 2) Rotational Impoundment Management (RIM ) a widely used seasonal mosquito control strategy that floods impoundments during the mosquito breeding season but leaves culverts open to the estuary duri ng non-breeding season (September May). Impoundments T10D and T10E. 9. 3) Wildlife Aquatic Management (WAM ) a management strategy widely employed on the refuge that was designed to provide submerged aquatic vegetation for waterfowl by gradually increasing wate r levels from spring through fall with highest water levels in October or Novemb er. Then water levels are then lowered in stages to allow waterfowl access to submerged aquatic vegetation throughout the winter and spring. In this management t ype there may be a brief period in spring during which culverts remain open, but the study WAM impoundments remained closed during the course of the Wetla nds Initiative. Impoundments T10J and T10L. 10. 4) Restored refers to an impoundment in which the perimeter dike has been completely removed in an effort to re turn the marsh to its native condition. Impoundment T10K. I used ground surveys of wading birds w ithin impoundments to address questions relating to factors influencing fora ging habitat selection at the s ite (local or patch) scale. Ground surveys were conducted weekly Janua ry July 2001 for all 9 impoundments. Surveys began 30 minutes after sunrise and we re usually completed within five hours. Surveys were conducted while driving the perimeter dikes of the impoundment and always followed the same route due to traffic re strictions on some of the dike roads. For each individual or group of wading birds the following data were recorded: location

PAGE 91

74 within predetermined zones within impoundme nts (Figures 4.2 4.4), the number and species of individuals, micr ohabitat type (unvegetated fl ooded sites, unvegetated nonflooded sites, vegetated flooded sites, perc hed), water depth (estimated from water position on legs), and activity (foraging or lo afing). Individuals we re considered to be part of a group if the distance to the next nearest wading bird was within 10 body lengths. Whenever possible, approximate locations were recorded on aerial photos for all aggregations consisting of ten or more wading birds with average inter-individual distances less than 10 body lengths. All spat ial data were recorded in a GIS to allow spatial analysis of wading bird distribution patterns. For each species, I tabulated the total number of individuals observed, the proportion of individuals within aggregations, and the water depths at foraging locations by habitat type (vegetated and unvegetate d). Survey counts for each impoundment section were divided by secti on area to calculate wading bird density. Because the interior areas of impoundments were sometimes obscured from perimeter dikes, I mapped the area of impoundments clearly visible from perimeter dikes using 1999 orthorectified color infrared photography (Anonymous 1999) and used that area in all calculations of wading bird density. Density was transforme d as ln(density+1) to help normalize data allowing means and 95% confidence intervals to be calculated on the log-transformed data; all statistics reported were back-transformed to the original units. For each species, I conducted a foraging hab itat selection analysis using a resource selection ratio: wi =[propor tion of habitat i used]/[propor tion of habitat i available] (Manly et al. 2002). This ratio measures th e relative preference between habitat types allowing comparison of preferences between hab itats that are not affected by what types

PAGE 92

75 are included. Wetland habitat within the survey area was first classified into vegetated (sites with emergent vegeta tion) and unvegetated (sites w ithout emergent vegetation) wetland habitat using a vegeta tion classification map (Brock meyer 2004). I next created a 0.5 m buffer within the perimeter of each hab itat type. The buffer areas were joined to create a 1m wide interface betw een the two habitat types called edge habitat. I subtracted the respective buffer areas from both the vegetated and unvegetated wetland habitat areas. This resulted in three wetland habita t types within the survey area: unvegetated, vegetated and edge. I conducted aerial surveys to evaluate fact ors influencing foraging habitat selection at a larger scale (e.g., between impoundment s). Aerial surveys of the 7 Wetlands Initiative impoundments were conducted July 2001 July 2002 to overlap with fish sampling (Chapter 3). Surveys were conducted 0900 1100 EST in a NASA Huey helicopter flying at an altitude of approximately 60 m, and a speed of 60 kts. These surveys were included in monthly wading bi rd foraging habitat use surveys conducted as part of long-term monitoring on KSC/MINWR (Sto len et al. 2002). During surveys, each impoundment was flown systematically such that all area within impoundments was surveyed. At each impoundment, the following data were collected: species, number of individuals, and cover type (wetland habitat with or wit hout emergent vegetation). Habitat specific wading bird density was calculated for unvege tated and vegetated habitat within each impoundment. The area of each habitat type within impoundments was calculated using a vegetation classification map (Brockmeyer 2004). When more than one survey was conducted during a season (based on periods of fish sampling), the mean density was calculated.

PAGE 93

76 Water level data for the Wetlands Initi ative impoundments was obtained from the St. Johns River Water Management District (for details of water level data collection see Brockmeyer 2004). The data consisted of hour ly recordings of wa ter level within each impoundment. For T10K (the restored impound ment) the water leve l measured in the estuary adjacent to the impoundment was used. For each survey date I calculated the mean daily water level for the study impoundment s. I calculated the daily change in water level as the difference between a days mean water level and the previous days mean water level. I then calculated the sum of the daily change in water level over the 7 days prior to the survey da te. Finally I calcu lated mean monthly water levels for examination of seasonal patterns of change in water levels. The daily impoundment water level data used for modeling wading bi rd abundance is summarized in Appendix A. Finally, I formulated a series of competi ng models that allowed me to evaluate factors that may influence wading bird foraging habitat selection at the impoundment level (Anderson and Burnham 2002, Johnson and Omland 2004). Models were formulated as general linear models (GLM) predicting wading bird density as a function of explanatory variables which included fish density, season (2 levels) management type (3 levels), habitat (vegetated or unvegetated ), water depth, and change in water depth. Table 4.1 lists the models with a brief desc ription of the factors evaluated by each. Relationships were modeled for the 5 samp ling periods during which fish density was measured (Chapter 3). Seasons used in anal ysis were nesting (Mar ch May, included 2 fish sampling periods) and non-nesting (July-Febr uary, included 3 fish sampling periods). Management types used in analysis we re: Open, RIM, and WAM (see above for

PAGE 94

77 definitions). Impoundment T10K was not included in this analysis because its management type (restored) was not replicated. I used information-theoretic model se lection methods (Burnham and Anderson 2002) to choose among the models (GLMs). M odels were considered for interpretation of their parameters if they met the following criteria: 1) AICc of less than 10.0, 2) were included in the set of best supported models with combined Akaike weights of 0.95 (95% confidence set) and 3) had an evidence rati o relative to the best supported model greater than 0.135 (Burnham and Anderson 2002). A mode l including the 6 main effects and all two-way interactions considered in any ot her model was evaluated to access the overall fit of the models (Burnham and An derson 2002). The goodness-of-fit and other diagnostics for meeting general linear mode l assumptions were investigated for all models considered for interp retation (Grafen and Hails 2002). I also applied the same model selection procedure separately for t hose individual wading bird species for which the number of individuals was at least 10% of the total for all species combined. All statistical calculations were performed using either Micr osoft Access, Microsoft Excel 2003 (Microsoft Corperation, 1985-2 003) or SPSS 12.0 (SPSS Inc. 1988-2003).

PAGE 95

78Table 4.1. Models set for analysis of f actors that effect wading bird density in foraging habitat on KSC/MINWR. ModelsDescription Pwading bird density controlled primarily by density of prey P*Hthe effects of prey density on wading bird density differ between habitat types P*Dthe effects of prey density on wadi ng bird density depend on water depth Hwading bird density controlled primarily by habitat H*Swading birds switch habitats seasonally due to factors other than prey density H*Dwading birds respond to depth differently in different habitats M*Hmanagement influences wadi ng bird habitat preference Dwading bird density controlled primarily by depth Cwading birds use change in water depth to choose foraging habitat H*Cwading birds respond to change in water depth differently bewteen habitats H, Chabitat preference an d change in water depth P, Hwading bird density controlled by density of prey and habitat H, Dwading bird density controlle d by depth of water and habitat P, Dwading bird density controlled by density of prey and water depth Mwading bird density controlled by management effects not included in any other variable P, Mwading bird density controlled by density of prey and management effects P*Mmanagement effects influences the effect of prey density a Variables: P = prey density, H = habitat, S = season, M = management, D = depth, C= change in depth

PAGE 96

79 T10K T10L T10J T10H T10D T10C T10E SHILOH 3 SHILOH 1 T27B T38I n d i a n R i v e rM o s q u i to L a g o o nA t l a n t i c O c e a nKennedy Space Center / Merritt Island National Wildlife RefugeT10I T10G 05,000 2,500Meters Figure 4.1. Map of study site showing locat ion of 9 study impoundments on the Kennedy Space Center-Merritt Island National W ildlife Refuge (T10C, T10D, T10E, T10G, T10H, T10I, T10J, T10K, and T10L ). The Wetlands Initiative subset of impoundments included: T10C, T10D, T10E, T10H, T10J, T10K, and T10L.

PAGE 97

80 T10K-2 T10K-3 T10K-1 T10K-4 T10L-7 T10L-3 T10L-4 T10L-5 T10L-6 T10L-2 T10L-1 03807601,1401,520 190 Meters Figure 4.2. Aerial photographs of study impoundments showing sections used to record locations of foraging wading birds during ground surveys. Impoundments shown are T10L and T10K.

PAGE 98

81 T10J-3 T10J-2 T10J-1 T10J-4 T10I-1 T10I-2 T10H-2 T10H-1 T10H-3 03306609901,320 165 Meters Figure 4.3. Aerial photographs of study impoundments showing sections used to record locations of foraging wading birds during ground surveys. Impoundments shown are T10H, T10I, and T10J.

PAGE 99

82 T10C-1 T10C-2 T10C-3 T10D-1 T10D-4 T10D-2 T10D-3 T10E-3 T10E-2 T10E-1 T10E-4 T10E-5 T10G-5 T10G-4 T10G-6 T10G-3 T10G-7 T10G-2 T10G-1 03206409601,280 160 Meters Figure 4.4. Aerial photographs of study impoundments showing sections used to record locations of foraging wading birds during ground surveys. Impoundments shown are T10C, T10D, T10E, and T10G.

PAGE 100

83 Results Twenty-five wading bird ground surveys of the nine study impoundments were conducted between 11 January and 19 July 200 1, yielding a total of 9,781 individuals of 13 species of wading bird (Table 4.2). Three of the species (Cattle Egret, Bulbulcus ibis Green Heron, Butorides virescens and Black-crowned Night Heron, Nycticorax nycticorax ) were sighted very infreque ntly and were not included in subsequent analyses. The mean density of wading birds observed during ground surveys was 0.26 individuals/ha and the 95% confidence interv al of the mean was (0.20, 0.32). The mean density varied over time and generally decr eased over the survey period (Figure 4.5). Density within impoundment sections also va ried greatly with some sections having much greater use than others (Figure 4.6). Wading bird densit y decreased with the proportion of open water habitat (Figure 4.7) a nd increased with the proportion of edge habitat (figure 4.8) within impoundme nt sections (see Figures 4.2-4.4). Over 80% of birds observed from dikes we re foraging in water at or below their tarsal joint, but species varied in their propensity to forage in deeper water (Table 4.3). Foraging habitat use varied by species, but most species used unvegetated wetland habitat more than vegetated wetland habi tat, with Great Blue Heron ( Ardea herodias ) and Wood Stork ( Mycteria americana ) being notable exceptions (F igure 4.9). The proportion of unvegetated wetland habitat w ithin impoundments varied between 0.06 and 0.99 (Figure 4.10). Between 7 to 28% of the observations of foraging wading bird s were within 0.5 m of the interface between unvegetated open wate r and vegetated habita t (Figure 4.11). All species greatly preferred edge habitat (hab itat within 0.5 m of the interface between vegetated and unvegetated hab itat) to both unvegetated and vegetated habitats (Figure 4.12, Table 4.4). All species except Great Bl ue Heron and Wood Stork also preferred

PAGE 101

84 unvegetated to vegetated habita t based on the proportion that was available within the impoundments (Figure 4.12, Table 4.4). Over half (59%) of all in dividual wading birds observed from dikes were in foraging aggregations ranging in size from 2 to 398 individuals (Figure 4.13). The mean group size was 43.9 (SE = 0.88, n = 9725) with a modal size of 5 (calculated with individuals included) White Ibis ( Eudocimus albus ) made up 47% of all birds observed in foraging aggregations, while Great Egrets ( Ardea alba ) Snowy Egrets ( Egretta thula ) and Tricolored Herons ( E. tricolor ) combined made up 39% (Figure 4.14). Species varied greatly in their propensity for group foraging (Figure 4.15). Locations of large foraging aggregations appeared somewhat cl umped with aggregations more common at the edges of impoundments and few within th e interiors (Figure 4.16). Large foraging aggregations also tended to be clumped temporally (Figure 4.17). Seven aerial surveys overlapped with periods when fish samples were taken within the Wetlands Initiative impoundments. The de nsity of wading birds measured during aerial surveys varied greatly between impoundments and within impoundments, the density also varied greatly between hab itat types (Figure 4.18 ). For 2 of the impoundments (T10C and T10D) the density was close to zero for both habitat types; these impoundments contained mostly unve getated habitat (Fi gure 4.10). Among the remaining 4 impoundments, the density within unvegetated habitat wa s much higher than that in vegetated habitat. The pattern of wading bird dens ity within each management type varied differently between habitat t ype and season. Impound ments with connection to the estuary (Open and RIM management types) had higher wadi ng bird density in vegetated than unvegetated foraging habita t (Figure 4.19). Closed impoundments (WAM

PAGE 102

85 management type) showed the opposite pattern with higher wading bird density in unvegetated than vegetated foraging habitat (Figure 4.19). Similarly, among seasons the pattern of wading bird density in foragi ng habitat was opposite for impoundments with connection to the estuary and those cl osed during the study period (Figure 4.20). Information-theoretic model selection resu lted in a single model with an Akaike weight of 0.91 (Table 4.5). The best-supporte d model included the interaction between fish density and habitat type (T ables 4.6 and 4.7). In this mode l the effect of prey density is different for each habitat type. Four i ndividual species (Great Egret, Snowy Egret, Tricolored Heron and White Ibis) had e nough observations to conduct model selection. The best supported models for Great Egret, Snowy Egret, Tricol ored Heron and White Ibis also included the interac tion between fish density and ha bitat type. Model selection results for these species are included in Appendix B.

PAGE 103

86 Table 4.2 Occurrence of wading birds during ground surveys of 9 impoundments on the KSC/MINWR, January July 2001. Species n Proportion of Total Proportion observed in Aggregations Great Blue Heron ( Ardea herodias ) 312 0.03 0.07 Great Egret ( A. alba ) 1,306 0.13 0.55 Snowy Egret ( Egretta thula ) 1,395 0.14 0.75 Little Blue Heron ( E. caerulea ) 398 0.04 0.49 Tricolored Heron ( E. tricolor ) 1,703 0.17 0.29 Reddish Egret ( E. rufescens ) 260 0.03 0.20 Cattle Egret ( Bulbulcus ibis ) 3 < 0.01 0.67 Green Heron ( Butorides virescens ) 20 < 0.01 0.10 Black-crowned Night Heron ( Nycticorax nycticorax ) 1 < 0.01 0.00 White Ibis ( Eudocimus albus ) 3,666 0.37 0.74 Glossy Ibis ( Plegadis falcinellus ) 238 0.02 0.86 Roseate Spoonbill ( Platalea ajaja ) 307 0.03 0.77 Wood Stork ( Mycteria americana ) 140 0.01 0.55 Unidentified 32 < 0.01 0.41

PAGE 104

87Table 4.3. Water depth at locations wher e foraging wading birds were observed in impoundments during ground surveys of impoundments on KSC/MINWR, January July 2001. Numbers are counts of observations of individuals or groups in specified habitat. SpeciesLowerMiddleUpperLowerMiddleUpper Length of LegaSourcebGreat Blue Heron (Ardea herodias)49475124 / 181Great Egret (A. alba)80128899121 / 161Snowy Egret (Egretta thula)108100436113 / 101Little Blue Heron (E. caerulea)5449154213 / 101Tricolored Heron (E. tricolor)276309115182313 / 101Reddish Egret (E. rufescens)10275112262White Ibis (Eudocimus albus)1611631068162Glossy Ibis (Plegadis falcinellus)192142n/a3Roseate Spoonbill (Platalea ajaja)31201431202Wood Stork (Mycteria americana)211332n/a3 UnvegetatedVegetated a Number gives either 1) length of tars ometatarsus / tibiotarsus or 2) total leg length measured from museum skin b Sources of leg lengths 1) E. St olen, unpublished data, 2) Powell 1987, 3) n/a indicates measure not available

PAGE 105

88Table 4.4. All species of wading birds observe d showed selection for edge habitat ove r both vegetated and unvegetated habitats. With the exception of Great Blue He ron and Wood Stork, all species also showed selection for unvegetate d and avoidance of vegetated habitat. Data from ground surveys of impoundments on KSC/MINWR, January July 2001. EdgeUnvegetatedVegetatedSpeciesp(edge)wi edgese95% CIp(open)wi opense95% CIp(veg)wi vegse95% CInB edgeB openB vegGreat Blue Heron0.289.201.05(6.68 11.72)0.420.780.07(0.62 0.93)0.300.700.08(0.52 0.88)1970.860.070.07Great Egret0.227.210.98(4.85 9.56)0.661.220.06(1.07 1.37)0.120.280.05(0.15 0.41)1920.830.140.03Snowy Egret0.072.390.27(1.73 3.04)0.671.240.03(1.17 1.31)0.260.600.03(0.52 0.68)9670.560.290.14Little Blue Heron0.134.140.71(2.45 5.83)0.651.200.06(1.06 1.34)0.230.530.06(0.38 0.68)2390.710.200.09Tricolored Heron0.093.020.64(1.48 4.57)0.831.530.05(1.42 1.65)0.080.180.04(0.08 0.28)2180.640.320.04Reddish Egret0.278.900.90(6.74 11.06)0.631.170.05(1.04 1.30)0.100.220.04(0.12 0.32)2630.860.110.02White Ibis0.175.700.39(4.78 6.63)0.651.210.03(1.14 1.27)0.170.400.03(0.34 0.47)1,0460.780.170.06Glossy Ibis0.196.280.38(5.38 7.19)0.651.200.03(1.14 1.26)0.160.380.03(0.32 0.44)1,1700.800.150.05Roseate Spoonbill0.196.120.27(5.47 6.78)0.621.150.02(1.10 1.20)0.190.450.02(0.40 0.49)2,1850.790.150.06Wood Stork0.206.681.54(2.99 10.37)0.530.970.11(0.72 1.23)0.270.630.12(0.34 0.92)740.810.120.08

PAGE 106

89 Table 4.5. Information-theoretic model selecti on results for GLM analysis with (wading bird density)1/2 as response variable, fish de nsity (P), water depth (D), and change in water depth (C) as covariates and factors: season (S, 2 levels), management (M, 2 levels), and habi tat (H, unvegetated or vegetated). Model MLE 2kAIC AIC AICc AICc n/kexp(-0.5 i)wiP*H0.135-91.231.39-89.870.00101.000.91 P, M0.155-84.338.30-82.966.91100.030.03 P0.173-82.789.85-82.267.61170.020.02 P*M0.147-83.778.85-81.118.7670.010.01 P, H0.174-81.3111.31-80.439.44130.010.01 P, D0.174-80.9211.70-80.039.84130.010.01 M0.174-80.2512.38-79.3610.51130.010.00 P*D0.175-79.0613.57-77.7012.17100.000.00 C0.193-77.2815.35-76.7613.11170.000.00 M*H0.157-79.2913.33-76.6313.2470.000.00 H0.193-76.6915.94-76.1713.70170.000.00 D0.193-76.5416.08-76.0213.85170.000.00 H*S0.175-77.3515.27-75.9913.88100.000.00 H*C0.185-75.9416.69-74.5715.30100.000.00 H, C0.194-75.4517.18-74.5615.31130.000.00 H, D0.194-74.7117.92-73.8216.05130.000.00 H*D0.195-73.2519.38-71.8817.98100.000.00 H*C, H*S, H*D, P*H, P*D, P*M0.0816-92.630.00-76.1413.7330.000.00

PAGE 107

90 Table 4.6. ANOVA table for GLM w ith (wading bi rd density)1/2 as response variable and including the fish dens ity*habitat interaction. Source Type III Sum of Squaresdf Mean SquareFSig. Corrected Model 2.99a31.006.950.00 Intercept1.9811.9813.770.00 habitat0.2910.292.010.16 fish density2.4412.4417.020.00 habitat fish density1.7811.7812.380.00 Error6.60460.14 Total17.6050 Corrected Total9.6049 a R2 = 0.312 (Adjusted R2 = 0.267) Table 4.7. Parameter estimates for GLM with (wading bird density)1/2 as response variable and including the fish density*habitat interaction. Parameter SE 95% CI Intercept0.380.12(0.14 0.62) habitat = unvegetated-0.210.15(-0.51 0.09) habitat = vegetated 0afish density0.000.01(-0.01 0.02) habitat = unvegetated fish density0.050.01(0.02 0.08) habitat = vegetated fish density 0a a This parameter is set to zero because it is redundant.

PAGE 108

91Birds / ha 0.00 0.20 0.40 0.60 0.80 1.00 1.201/4/01 1/18/01 2/1/01 2/15/01 3/1/01 3/15/01 3/29/01 4/12/01 4/26/01 5/10/01 5/24/01 6/7/01 6/21/01 7/5/01 7/19/01Birds / ha 0.00 0.20 0.40 0.60 0.80 1.00 1.201/4/01 1/18/01 2/1/01 2/15/01 3/1/01 3/15/01 3/29/01 4/12/01 4/26/01 5/10/01 5/24/01 6/7/01 6/21/01 7/5/01 7/19/01 Figure 4.5. Mean density of wading birds dur ing ground surveys of 9 impoundments on KS C/MINWR, January July 2001. Error bars show 95% confidence limits of back-transformed means.

PAGE 109

92 0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 1 2 3 1 2 3 4 1 2 3 4 5 1 2 3 45 1 2 3 1 2 1 2 3 4 1 4 1 2 3 4 6 T10CT10DT10ET10GT10IT10J T10HT10L T10KBirds / ha 0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 1 2 3 1 2 3 4 1 2 3 4 5 1 2 3 45 1 2 3 1 2 1 2 3 4 1 4 1 2 3 4 6 T10CT10DT10ET10GT10IT10J T10HT10L T10KBirds / ha Figure 4.6. Mean density of wading birds by impoundment section (see figures 4.2-4.4) during ground surveys of 9 impoundments o n KSC/MINWR, January July 2001. Error bars show 95% conf idence limits of back-transform ed means. The solid bar shows the overall mean density of 0.26 i ndividuals / ha and the dashed lines show the 95% confidence limits of that mean.

PAGE 110

93 0 0.1 0.2 0.3 0.4 0.5 0.6 0.000.100.200.300.400.500.600.700.800.901.00Proportion Open HabitatWading Bird Density (ind. / ha)y = 0.47 0.32x R2= 0.274 0 0.1 0.2 0.3 0.4 0.5 0.6 0.000.100.200.300.400.500.600.700.800.901.00Proportion Open HabitatWading Bird Density (ind. / ha) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.000.100.200.300.400.500.600.700.800.901.00Proportion Open HabitatWading Bird Density (ind. / ha)y = 0.47 0.32x R2= 0.274 Figure 4.7. A linear regression showed that as the proportion of open habitat increased the wading bird density decreased. Data collected within impoundment sections during ground surveys, on KSC/MINWR January July 2001. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.000.010.020.030.040.050.060.070.080.090.10Proportion Edge HabitatWading Bird Density (ind. / ha)y = 0.18 + 3.1x R2= 0.135 0 0.1 0.2 0.3 0.4 0.5 0.6 0.000.010.020.030.040.050.060.070.080.090.10Proportion Edge HabitatWading Bird Density (ind. / ha) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.000.010.020.030.040.050.060.070.080.090.10Proportion Edge HabitatWading Bird Density (ind. / ha)y = 0.18 + 3.1x R2= 0.135 Figure 4.8. A linear regression showed that as th e proportion of edge habitat increased the wading bird density increased. Data collected within impoundment sections during ground surveys, on KSC/MINWR January July 2001.

PAGE 111

94 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00Gr eat B lue Heron Great Egret Snowy Egret L ittle Blue Hero n Tricolored Hero n Reddish E gret Wh it e Ibis Glossy Ibis Roseate Spoonbil l Wood Stork PERCHED MUD UNKN VEG OPENProportion 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00Gr eat B lue Heron Great Egret Snowy Egret L ittle Blue Hero n Tricolored Hero n Reddish E gret Wh it e Ibis Glossy Ibis Roseate Spoonbil l Wood Stork PERCHED MUD UNKN VEG OPENProportion Figure 4.9. Habitat use by species of wading birds observed during ground surveys of foraging habitat use of 9 impoundment s on KSC/MINWR, January July 2001. 0 0.2 0.4 0.6 0.8 1 1.2 T10CT10DT10ET10GT10HT10IT10JT10KT10LProportion 0 0.2 0.4 0.6 0.8 1 1.2 T10CT10DT10ET10GT10HT10IT10JT10KT10LProportion Figure 4.10. Proportion of unvegetated habita t contained within 9 impoundments on KSC/MINWR, circa 2001.

PAGE 112

95 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Great Blue Heron Great EgretSnowy EgretLittle Blue Heron Tricolored Heron Reddish EgretWhite IbisGlossy IbisRoseate Spoonbill Wood Stork 0-0.5 m 0.5-3 m 3-10 m >10 mProportion of observations 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Great Blue Heron Great EgretSnowy EgretLittle Blue Heron Tricolored Heron Reddish EgretWhite IbisGlossy IbisRoseate Spoonbill Wood Stork 0-0.5 m 0.5-3 m 3-10 m >10 mProportion of observations Figure 4.11. Distance to edge for observati ons of wading birds of 10 species dur ing ground surveys of 9 impoundments on KSC/MINWR, January July 2001. For each species, bars give proportion of observations with in each distance to nearest edge category.

PAGE 113

96wi 0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00Great Blue Heron Great EgretSnowy EgretLittle Blue Heron Tricolored Heron Reddish EgretWhite IbisGlossy IbisRoseate Spoonbill Wood Stork wi edge wi open wi vegwi 0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00Great Blue Heron Great EgretSnowy EgretLittle Blue Heron Tricolored Heron Reddish EgretWhite IbisGlossy IbisRoseate Spoonbill Wood Stork wi edge wi open wi veg Figure 4.12. Resource selection ratios (wi) for wading birds foraging in impoundments on MINWR/KSC. Error bars give 95% CI with Bonferoni adjustment to permit multiple compar isons between habitat types within each species.

PAGE 114

97 0 500 1000 1500 2000 2500 3000 3500 4000 4500 123-1011-5051-100100+ Group SizeBirds Figure 4.13. Histogram of sizes of wading bi rd foraging aggregations observed during ground surveys of impoundments on KS C/MINWR, January July 2001. 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 Great Blue Heron Great EgretSnowy Egret Little Blue Heron Tricolored Heron Reddish Egret White IbisGlossy IbisRoseate Spoonbill Wood StorkProportion 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 Great Blue Heron Great EgretSnowy Egret Little Blue Heron Tricolored Heron Reddish Egret White IbisGlossy IbisRoseate Spoonbill Wood StorkProportion Figure 4.14. Composition of wading bird foragi ng aggregations observed during ground surveys of 9 impoundments on KSC/MINWR, January July 2001.

PAGE 115

98 Great Blue Heron 0 100 200 300 400 Ind.2-1011-100100+ Great Egret 0 200 400 600 800 Ind.2-1011-100100+ Snowy Egret 0 100 200 300 400 500 600 Ind.2-1011-100100+ Little Blue Heron 0 50 100 150 200 250 Ind.2-1011-100100+ Reddish Egret 0 50 100 150 200 250 Ind.2-1011-100100+ White Ibis 0 500 1000 1500 Ind.2-1011-100100+ Glossy Ibis 0 50 100 150 Ind.2-1011-100100+ Roseate Spoonbill 0 20 40 60 80 100 120 Ind.2-1011-100100+ Wood Stork 0 20 40 60 80 Ind.2-1011-100100+ Tricolored Heron 0 500 1000 1500 Ind.2-1011-100100+Number of birds Great Blue Heron 0 100 200 300 400 Ind.2-1011-100100+ Great Egret 0 200 400 600 800 Ind.2-1011-100100+ Snowy Egret 0 100 200 300 400 500 600 Ind.2-1011-100100+ Little Blue Heron 0 50 100 150 200 250 Ind.2-1011-100100+ Reddish Egret 0 50 100 150 200 250 Ind.2-1011-100100+ White Ibis 0 500 1000 1500 Ind.2-1011-100100+ Glossy Ibis 0 50 100 150 Ind.2-1011-100100+ Roseate Spoonbill 0 20 40 60 80 100 120 Ind.2-1011-100100+ Wood Stork 0 20 40 60 80 Ind.2-1011-100100+ Tricolored Heron 0 500 1000 1500 Ind.2-1011-100100+Number of birds Figure 4.15. The number of birds observed in eac h of 4 group sizes categories by species, during ground surveys of 9 impoundments on KSC/MINWR, January July 2001.

PAGE 116

99 ( ( ( ( ( ( (! ( ( ( ( ( ( ( ( ( ( (! ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( (! ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( (! ( ( ( ( ( ( ( 05501,1001,6502,200 275 MetersAggregation Size! (10 20! (21 50! (51 100! (101 200 Figure 4.16. Spatial distribution of wading bird foraging aggregations (10 individuals or larger) observed during ground survey s of foraging habitat use of 9 impoundments on KSC/MINWR, January July 2001.

PAGE 117

100 # # # # # # # # # # # # # # # # " " " " " "[ [ [ [ [ [ [ [ [ [ [ [ _E E E E E E E E E D D D D D D D D D D D D Di i i# # # # # # # # # # # # # # # # # # # # # #X X X X X E E E[ [ [ [ [ [ [ [ [ [ [ [ [ _X X X X X 05501,1001,6502,200 275 MetersMonth#JAN"FEB[ _MARXAPREMAYDJUNiJUL Figure 4.17. Temporal distribution of wading bird foraging aggregations (10 individuals or larger) observed during ground surveys of foraging habitat use of 9 impoundments on KSC/MINWR, January July 2001.

PAGE 118

101 Wading birds / ha 0 1 2 3 4 5 6 T10CT10DT10ET10HT10JT10KT10L Unvegetated VegetatedWading birds / ha 0 1 2 3 4 5 6 T10CT10DT10ET10HT10JT10KT10L Unvegetated Vegetated Figure 4.18. Wading bird density in 7 impound ments measured during aerial surveys concurrent with fish sampling periods. Bars give mean density over the 5 samples and error bars show standard error of mean. 0 0.2 0.4 0.6 0.8 1 1.2 unvegetated vegetatedOpenWAM RIMIndividuals/ha 0 0.2 0.4 0.6 0.8 1 1.2 unvegetated vegetatedOpenWAM RIM 0 0.2 0.4 0.6 0.8 1 1.2 unvegetated vegetatedOpenWAM RIMIndividuals/ha Figure 4.19. Mean wading bird density in 2 type s of foraging habitat measured in 6 impoundments during aerial su rveys concurrent with fish sampling periods. Bars show mean density by management type and habitat and error bars give standard error of the mean. Impoundm ent T10K was not included in this figure since its management type (restored) was not replicated.

PAGE 119

102 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 nesting winterOpenWAM RIMIndividuals/ha 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 nesting winterOpenWAM RIM 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 nesting winterOpenWAM RIMIndividuals/ha Figure 4.20. Mean wading bird density in 2 seasons measured in 6 impoundments during aerial surveys concurrent with fish samp ling periods. Bars show mean density by management type and season and erro r bars give standard error of the mean. Impoundment T10K was not included in this figure since its management type (restored) was not replicated.

PAGE 120

103 Discussion Foraging Habitat Selection I found evidence of piscivorous wading birds selecting foraging habitat at 2 scales. At the smaller scale (i.e., within impoundment s) all species except Great Blue Heron and Wood Stork showed a strong preference for we tland habitat without emergent vegetation over vegetated wetland habitat (2-8 times, s ee column B in Table 4.4). Other studies have also noted a preference by wading birds fo r foraging in wetland habitat that is more open versus habitat with emergent vege tation (Kushlan 1978b, Breininger and Smith 1990, Smith 1995a, Smith and Breininger 1995, Surdick 1998, Bancroft et al. 2002, Stolen et al. 2002). Although wading birds pref erred open water over vegetated habitat, all areas of open water were not used equally ; instead wading birds in the northern IRL were concentrated near the edges between open and vegetated habitat (Figure 4.11). When I included the area within 0.5 m of the boundary of vegetated and unvegetated habitat within impoundments as a separate category (edge habitat), there was a strong preference for edge habitat over the othe r two types (Figure 4.12, Table 4.4). One explanation for this pattern is that edges might have a highe r level of prey availability because small marsh fish hide form piscine predators within dense vegetation, but remain near open water areas with higher food resources (McIvor and Odum 1988, Rozas and Odum 1988). Another explanation of the pr eference of wading birds for edge habitat could be that water depth is more optimal (i.e., less deep) for foraging wading birds near edges rather than in deeper unve getated areas away from edges. Wading birds can only forage efficiently in water up to a certain depth (Custer and Osborn 1978, Powell 1987). Most of the wading birds observed were foraging in water that was below their tarsal joint; for most sp ecies that occur in the northern Indian River

PAGE 121

104 Lagoon this means depths of 20 cm or less (Table 4.3). Water depths within impounded wetlands in the northern IRL are elevated in late fall and early winter, followed by lower levels in spring (Smith 1993). Based on re sults of a long-term monitoring program, Stolen et al. (2002) found that wading birds in the northern IRL increased their relative use of vegetated habitat w ithin impoundments during fall an d winter, coinciding with periods of higher water levels. During these periods, much of the open water habitat is too deep for wading birds to forage in e fficiently (E. Stolen, unpublished data). However, there is also some evidence of a spatial shift in prey from open water to vegetated habitats that coin cides with changes in hydrol ogy (Harrington and Harrington 1961,1982; Gilmore 1998). In Chapter 3 I showed th at the density of fish remained fairly constant in the unvegetated habitat but increas ed markedly in vege tated habitat in postnesting and winter seasons (June December). Despite their preference fo r unvegetated open water over vegetated habitat there was a decrease in wading bird density with in impoundment subsections as the proportion of open water habitat increased. One explanation for this pattern could be that wading birds prefer to forage in open water habitat where it is mechanically easier to catch prey (prey more available) but areas with more open water may have lower prey populations. Another is that with increas ing proportion of open water, the amount of edge habitat decreased. Stolen et al. (2002) found a correlation be tween wading bird use of impoundments and the ratio of unvegetated to vegetated habitat within impoundments, a measure that is expected to be highly corre lated with the amount of edge habitat. Similarly, in this study wading bird density within impoundment subsections increased with the amount of edge habitat.

PAGE 122

105 At a broader spatial scale (i.e., among impoundments) I found that both prey density and habitat type were important in predicting wading bird foraging habitat use. The effect of hydrological connection to the estuary (i.e., management type for Wetlands Initiative impoundments) mirrored seasonal eff ects, highlighting the importance of water level in making habitat accessible to wading birds. In the impounded salt marsh habitat of the northern IRL small resident fish are the primary type of prey for wading birds (Chapter 3). These fish ar e highly clumped both spatially and temporally (Chapter 3, Schooly 1980, Gilmore 1995, Stevens 2002, Brockmeyer 2004). The model including the interaction between fish density and habitat type (unve getated vs. vegetated) was selected as the best model to explain wading bird de nsity. In this model, prey density had a positive effect on wading bird density for unve getated habitat, but in vegetated habitat wading bird density remained essentially c onstant within the range of prey density measured within impoundments (Figure 4.21). This model was also in the set of selected models for each species considered individua lly, with similar fitted parameter estimates (Appendix B). However, for Great Egret a nd Tricolored Heron, more than one model was selected, some of which included the additional variables management type and water depth (Appendix B). Among these specie s, Gawlik (2002) suggested that Great Egret, and to a lesser extent Tr icolored Heron, may be more ab le to cope with lower prey density, which might make other factors more important in determining selection of foraging habitat for these species. Although fish density was higher in vegeta ted habitat, wading birds preferred to forage in unvegetated habitat. There are several possible e xplanations for this pattern. One explanation is that the composition of prey may differ between open and vegetated

PAGE 123

106 habitats and the birds may be selecting foraging habitat ba sed on this difference. In Chapter 3 I showed that unvegetated sites had sli ghtly larger fish than unvegetated sites. Wading birds may benefit from foraging in th ese sites by capturing la rger, energetically more profitable prey items. A nother possibility might be that prey density alone may not be a good predictor of prey avai lability, especially when used to compare structurally different habitats. Many previous studies ha ve noted that prey availability is a more accurate (but much harder to measure) in dicator of habitat quality (reviewed in Sutherland 1996). Another possible explanation is that wading birds do select sites with higher prey densities, but above some threshold level th ey do not differentiate between sites with different prey densities (M iranda and Collazo 1997). Conversely, Gawlick (2002) found that wading birds stopped foraging in sites wh en the prey density fell below certain thresholds. If both vegetated and unvegetat ed sites were perceived as having equally high (or low) prey density, wading birds might prefer more open habitats for other reasons, such as better visibility of pred ators (Caldwell 1986). There is conflicting evidence about the dependence of wading bird foraging success on the concentration of their prey organisms (e.g., Kushlan 1976b, 1978b; Frederick and Spalding 1994; Gawlik 2002). This is complicated by the fact that di fferent species vary in their need for such concentrations: for example the Wood St ork is highly dependent on prey being concentrated (Ogden et al. 1976) but Great Egret is not (Powell 1987). It may be that some species of wading birds are not good at detect ing prey patches but must rely instead on environmental or social cues to provide th em with information about prey distribution (Gawlik 2002, Master et al. 2005).

PAGE 124

107 0 1 2 3 4 5 6 7 8 9 01020304050unvegetated vegetated Fish m-2Wading Birds ha-1 0 1 2 3 4 5 6 7 8 9 01020304050unvegetated vegetated Fish m-2Wading Birds ha-1 Figure 4.21. Results of model predicting wading bird density as a function of prey dens ity and habitat type for Wetlands Initia tive impoundments. Figure shows output of model within observed range of fish density.

PAGE 125

108 Mixed-Species Foraging Aggregations A large proportion of wading birds obse rved foraging in impounded salt marsh habitat in the northern Indian River Lagoon were in mixed-sp ecies foraging aggregations (Table 4.2). Moreover, over one-third were in aggregations that contained over 50 individuals (Figure 4.13). This contrasts with other sites wh ere aggregations of a highly similar community of species were smaller (e .g., Master 1992). Species that showed the greatest tendencies to forage in aggregations were Snowy Egret, White Ibis, Glossy Ibis ( Plegadis falcinellus ) and Roseate Spoonbill ( Platalea ajaja ) (Table 4.2); these species also appeared to favor medium-sized groups over large groups (Figure 4.15). Another group of species including Great Egret, Little Blue Heron ( Egretta. caerulea ), Tricolored Heron, Reddish Egret ( E. rufescens ) and Wood Stork were observed in foraging aggregations less frequently (T able 4.2). With the excepti on of Great Egret, most of these species also appear to favor medi um sized groups (Figure 4.15). Many authors have suggested that white plumage is a ssociated with group foraging (Kushlan 1977, 1978b; Caldwell 1981; Crozier a nd Gawlik 2003) but in this study some of the most conspicuous group foragers had other plumag e colors (e.g., Glossy Ibis and Roseate Spoonbill). However, it is unclear how clos ely human perception of plumage coloration matches that of birds, and these species ma y appear very conspicuous to other wading birds (Cuthill et al. 1999),. Mixed species foraging aggregations contai ned up to 10 different species of wading birds (Figure 4.14). Usually, the only prey available at the fo raging site consisted of 3 or 4 species of small marsh resident fish (Chapter 3). Thus, this divers e assemblage of birds was foraging in one location on what appeared to be a fairly uniform prey resource. This suggests that competition was not strong enough to eliminate very similar species, such

PAGE 126

109 as among various Egretta species, or that pe rhaps the individuals were benefiting from the presence of other individuals (Chapter 5). One question which I wanted to answer was if wading bird aggregations were occurring at particular sites (e.g., remnant ti dal creeks, near culverts) or were clumped temporally in areas with high prey numbers. This might be expected if such groups occurred at sites that tended to concentrate prey (Kushlan 1976b). Based on the observed patterns of occurrence of aggreg ations there did not seem to be general types of feature within impoundments that were favored by such groups. Aggregations appeared to be concentrated near the edges of vegetated and unvegetated habita t within impoundments (Figure 4.16). Groups were also somewhat clumped temporally at the impoundment level (Figure 4.17), perhaps refl ecting changes in prey availa bility as foraging conditions changed in individual impoundments.

PAGE 127

110 CHAPTER 5 BENEFITS TO INDIVIDUALS IN PISCI VOROUS WADING BIRD MIXED-SPECIES FORAGING AGGREGATIONS Introduction Aggregations of foragers are a conspicuous phenomenon that occurs in a variety of animals (Krause and Ruxton 2002) includi ng wading birds (e.g., Caldwell 1981, 1986; Erwin 1983b; Kersten et al. 1991; Hafner et al. 1993; Master et al. 1993). Several hypotheses have been proposed to explain how individuals benef it from foraging in mixed-species aggregations. These can be summarized into 4 broad categories: concentration of individuals in high-qua lity patches (Fretw ell and Lucas 1970), protection from predators (Hamilton 1971), in formation exchange among individuals (Valone and Giraldeau 1993, Dall et al. 2005) and direct enhancement of foraging success to individuals du e to the presence of other forage rs (social facilitation; Kushlan 1978b). These hypotheses are not mutually exclusive and more than 1 might be operating at any given time (Morse 1970, Powell 1985). There are numerous ways that each of th ese categories might apply to wading birds foraging in aggregations. For example, there is ample evidence that mixed-species aggregations in wading birds often form in areas of high prey density (Kushlan 1976b, 1981a; Erwin et al. 1985; Kersten et al. 1991; Smith 1995b). Other studies have shown that wading birds locate prey pa tches using the presence of other foragers; this is often referred to as local enhancement (Kre bs 1974, Kushlan 1976a, Caldwell 1981, Erwin 1983b, Smith 1995b, Gonzalez 1997). Similarly, th e risk of predation to members of a

PAGE 128

111 foraging flock might be reduced due to increas ed predator detecti on, predator confusion and risk-spreading (Hamilton 1971, Yaukey 1995, La rsen 1996); such a benefit has been shown for wading birds (Caldwell 1986). Lowe r time devoted to vigilance can also lead directly to increased foraging success if an individual feeding in an aggregation can spend more time foraging (e.g., Caraco 1979, Sullivan 1984, Elger 1989, Hino 1998). Foraging success might also be improved through gaining additional information about the location of prey within a patch (Valone 1989, Valone and Giraldeau 1993), or learning from other foragers a bout foraging tactics, or novel prey types (Krebs and Inman 1992, Beauchamp et al. 1997). For example, some species of wading birds may act as focal or core members that attract others to mixed-species foraging aggregations (Kushlan 1977, Caldwell 1981, Erwin 1983b, Mast er et al. 1993, Smith 1995b, Strong et al. 1997). In other studies, it was shown that the proximity to Snowy Egrets ( Egretta thula ) was positively correlated with foraging success for Great Egrets ( Ardea alba ), Tricolored Herons ( E. tricolor ) and Little Blue Herons ( E. caerulea ) (Caldwell 1981). Foraging groups might make prey more su sceptible to preda tion, thereby improving an individual members foraging success (Morse 1970, Powell 1985, Rodrigues et al. 1994, Hino 1998). This is often referred to as the beating effect, but social facilitation is perhaps a better label because it does not presume the mechanism by which an individual benefits from the pres ence of other foragers. This mechanism assumes that prey is mobile and cryptic or ot herwise difficult to capture, as is the case for predators of schooling fish (Morse 1970) Neill and Cullen (1974) showed that individual fish isolated from schools may beco me more susceptible to predators. It is plausible that this mechanism might apply to wading bird species feeding on schooling

PAGE 129

112 fish, and Smith (1995b) noted that wading bi rds feeding on schooling fish can cause the schools to become confused. Kushlan (1978a) demonstrated that Little Blue Herons increase their foraging success rate in th is way when foraging near White Ibises ( Eudocimus albus ). These examples underscore the challenge involved in differentiating between enhanced foraging success due to foraging in better prey patches versus improved foraging success due to the presence of othe r foragers within a patch. Measuring the benefits to an individual fo raging in mixed-species aggregat ions can be complex; careful analysis is required to dis cern causal relationships and pathways among the myriad of factors influencing foraging success (Chapter 1, Fig 1.1). This challenge is underscored by the many studies that ha ve shown the costs incurred by wading birds foraging in mixed-species aggregations (Kushlan 1976a, 1978a,c; Russell 1978; Caldwell 1980; Petit and Bildstein 1987). Further at tention to this topic is warranted because mixed-species foraging aggregations are a prominent f eature of wading bird foraging ecology. Moreover, wading birds are an important co mponent of the trophic structure of many systems (Chapter 4, Breininger and Smith 1990, Stolen et al. 2002) Yet, although many studies have evaluated this question, only three (Krebs 1974, Erwin et al. 1985, Master et al. 1993) have measured prey density levels the presumed underlying factor governing the tradeoffs between solitary and group fora ging strategies. Of th e three studies, only Master et al. (1993) assessed the tradeoffs between so litary and group foraging for Snowy Egret, Tricolored Heron and Great Egrets the species of inte rest in this study. In this chapter, I studied the foraging success of 3 species of piscivorous wading birds under various field conditions in an effort to isolate the relative cont ributions to

PAGE 130

113 enhanced foraging success due to foraging in better patches versus that due to the presence of other foragers. To achieve th is research goal, I tested 3 interrelated, hierarchical hypotheses that al lowed me to statistically pa rtition the effects of prey density and foraging with othe rs, on foraging success. First, I tested whether wading birds were selecting foraging sites with higher prey density or biomass than that at nearby unused sites. Second, I tested whether or not wading bird foraging aggregations occurred at sites with higher prey density than that at sites occupied by i ndividuals. Finally, I monitored foraging success of birds foragi ng alone and in groups, and simultaneously measured the prey density at those sites, al lowing me to address the primary question of interest, namely the relative importance and infl uence of prey density and the presence of other foragers (social facili tation) on foraging success. Methods Study Site The study site consisted of areas of impounded salt marsh habitat on the 55,000 ha Kennedy Space Center-Merritt Island Nationa l Wildlife Refuge (KSC/MINWR). This site is located in the nort hern portion of the Indian River Lagoon system (IRL), a subtropical estuary which is an important site for wading birds on the southeastern Atlantic coast of North America (Brein inger and Smith 1990, Schikorr and Swain 1995, Sewell et al. 1995, Smith and Breininger 1995, Stolen et al. 2002). Mixed-species foraging aggregations of several hundred or more individuals are common in impounded wetlands on KSC/MINWR (Chapter 4). The northern portion of the IRL is isolated from the nearest ocean inlet and has very low diur nal tidal changes (< 1cm; Smith 1987). In this region, seasonal and wind-driven water level fluctuations are of much greater importance (Smith 1993). Habitat within impoundments is predominantly a

PAGE 131

114 heterogeneous mixture of open water and vege tated cover types, with tall marsh grass (e.g., Spartina bakeri ) and short marsh vegetation (e.g., Distichlis spicata Batis maritima ) predominating in vegetated areas (Schmalzer 1995). Eighteen impoundments were chosen for observations of foragi ng wading birds on KSC/MINWR (Figure 5.1). During the first year of observation this ar ea was experiencing a sever drought and water levels were unusually low in many impoundme nts by mid summer. Rainfall and water levels during the second y ear were more typical. Foraging Behavior Observations Between January 2002 June 2003, I made observations of wading bird foraging behavior coupled with prey sampling at lo cations where the birds were foraging, to determine the effect of group size and prey density on foraging rates between and within species. I chose the 3 most abundant species of piscivorous wading birds on KSC/MINWR (Stolen et al. 2002): Great Egret, Snowy Egret, and Tricolored Heron. Observations were made from sunrise to 6 hours after sunrise. This time block encompasses the period of largest wading bird aggregations and hi ghest feeding activity in many areas studied (Hom 1983, Cezilly et al. 1990, Kersten et al. 1991). To obtain unambiguous identification of a foraging group for sampling, groups were defined as 2 or more individuals foraging with inter-indi vidual distances 10 m or less (Erwin 1983a, Wiggins 1991, Master et al. 1993). Group size classes were designated as: individual, small (2-10 individuals) medium (11-50 indivi duals), and large (> 51 individuals). To simplify the experimental design, only birds foraging in unvegetated habitat type were included; this is the most common habitat type used by wading bird aggregations on KSC/MINWR (Chapter 4, Breininger a nd Smith 1990, Stolen et al. 2002).

PAGE 132

115 During each observation session, an indi vidual or group of foraging birds was randomly selected as the sampling unit as follows. First a focal impoundment and direction of travel was ra ndomly chosen. Then, while traveling around the impoundment perimeter dike, the first individual or group encountered in the impoundment (that was of the target size and habitat class) was select ed. Wading birds were chosen for observation to equalize sample sizes among gr oup size categories. Care wa s taken that the distance of potential observation units from the impound ment perimeter dike did not influence choice of sampling unit. During behavioral observati ons individual wading birds were observed from a distance of 100 m using a 15-60x spotting scope. For each individual observed, foraging behavior was measured for a period of 1 to 3 minutes, during which the total number of strikes, the number of successful captures, and the number of steps taken was recorded. When possible, the length (estim ated based on bill length) and identity (fish versus non-fish) of prey items ingested was r ecorded. Other variables recorded for each individual observed were: microhabitat type in which the bird was foraging, group size and species composition, water depth (estimated by leg length), distance to nearest other wading bird at start of ob servation, nearest distance to other wading bird during observation, any aggressive behavior towards or from another forager, movements other than walking, distance to perimeter dike distance to observer, and location. Four variables were calculated from be havioral data for use in analyses. Captures/min and unsuccessful strikes/min were based on the number of prey items captured; this could be detected because wa ding birds usually make a distinctive headjerking motion when swallowing prey items. Strikes/min was the total number of times

PAGE 133

116 the birds bill penetrated the water incl uding both successful and unsuccessful capture attempts. Foraging success rate was calculate d as the number of captures divided by the total number of strikes made during the obs ervation period. The mean length of prey items was calculated for all individuals for whic h data were available. Estimated lengths based on the proportion of bill lengths were converted to mm based on mean measurements of museum specimens (E. Stol en, unpublished data). Independent samples t-tests (without the assumption of equal variances) were used to test for differences between the mean sizes of fish captured be tween solitary individuals and birds foraging in groups for each species. Immediately following foraging observati ons, prey abundance was estimated by making three tosses of a 1-m2 throw-trap (Kushlan 1981b). This type of sampling gear has been shown to produce accurate estimates of nekton abundance in shallow wetlands similar to those in this study (Chick et al. 1992, Jordan et al. 1997). During prey sampling, researchers first approached the samp le site by walking slowly and then tossed a 1-m2 throw trap from a distance of 1-2 m. Once the trap landed, researchers quickly secured the edges of the trap ag ainst the substrate. Fish we re then scooped from the trap using a 40 by 30 cm dip net with 2 mm mesh. Vegetation was removed within the trap if it impeded movement of the dip net. When the large dip net was scooped 3 times without catching a fish, a 15 by 10 cm dip net with 2mm mesh was used which was more effective in scraping along the edges and into the corners of the trap. The sample was completed when the smaller dip net was sc ooped 3 times without catching a fish. The first 30 individuals of each species captur ed in each throw-trap deployment were measured to the nearest mm. The mass of th ese fish was estimated using species-specific

PAGE 134

117 regression equations developed for fish captured in impoundments on KSC/MINWR (Phil Stevens, U.S. Geological Service, unpubl ished data). For each deployment of the throw trap I also recorded water depth, wa ter temperature, salinity, presence of submerged aquatic or emergent vegetati on, GPS position, and the overall vegetation of the foraging site. Following sampling at the foraging site, prey samples were also taken at a paired random location w ithin the same habitat type 100-1000 m from the foraging site. Paired random locations were chosen by first randomly selecting a compass bearing, and then sampling the nearest area of sim ilar habitat at a distance randomly chosen between 100 1000 m from the foraging observation point. If this pro cedure resulted in a site outside of the impoundment, a new random bearing was chosen. Paired random sites were sampled using the same procedures as foraging sites. For each sample site (used and paired -unused) I calculated the fish density, estimated prey biomass, and also the mean fi sh length and mean fish mass for all species combined. Within a sample, wet prey biomass was estimated by multiplying the mean biomass for each species within the sample by th e number of individuals of that species. I also included the biomass of shrimp ( Palaemonetes sp.) in samples, using 0.0817 g as the mean mass of shrimp based on data in Gilmore (1998). Contrasts between used and non-used (random) sites (e.g., density of fish, biomass of fish) were made using paired-t tests. Contrast between sites with groups and paired-random sites were made using unpaired t-test without the assumption of equal variances. All of the raw foraging rate variables were highly rightskewed and square-root transformations were used to normalize the distributions. Fish density was also highly skewed to the right; I used the transformation ln (fish density+1) to normali ze the distribution. Af ter transformation,

PAGE 135

118 these variables did not differ significantly fr om a normal distribution based on results of one-sample Kolmogorov-Smirnov tests. I used information-theoretic model se lection methods (Burnham and Anderson 2002) to choose between alternative Ge neral Linear Models (GLMs) with (captures/min)1/2 as the response variable. The analysis was based on a planned observational experimental design with capture rate (successful captures/min) as the response variable and group size and prey dens ity as an explanatory variable. This design was similar to an Analysis of Covari ance (ANCOVA) with group as the factor and prey density as the covariate. I justified the use of group size as a categorical variable because it avoided a priori assumptions about the functi onal relationship (e.g., linear, non-linear) between cap ture rates and group size, while still retaining the ability to discern changes in capture rates as a functi on of group size categories. Kushlan (1978b) hypothesized that the relation resembles an inverted parabola with foraging rates maximum at intermediate group sizes. However, given large gaps in the distribution of group sizes recorded in the study area, I lack ed the ability to di fferentiate between a linear and a quadratic model. To better understand the effect of group size, I used 2 alternate forms of the group size variable during anal yses: the categorical variable with 4 levels based on the sampling design, and a bina ry variable coding wh ether the individual was foraging in a group or alone. I also wanted to consider th e possibility that the effects of prey density might differ between group si zes (or between groups and individuals), so I included interaction terms in some models Thus, 7 models were considered with (captures/min)1/2 as the response variable and th e following explanatory variables: 11. 1) A model with prey dens ity as a covariate (PD). 12. 2) A model with the binary gr oup variable as a factor (GB).

PAGE 136

119 13. 3) A model with both prey density as a c ovariate and the binary group variable as a factor (GB, PD). 14. 4) A model with main effects and the in teraction between pr ey density and the binary group variable (GB*PD). 15. 5) A model with group cate gory as a factor (GC). 16. 6) A model with both prey density as a covariate and group ca tegory as a factor (GC, PD). 17. 7) A model with main effects and the in teraction between pr ey density and group category (GC*PD). Models were considered for interpretation of their parameters if they met the following criteria: 1) AICc of less than 10.0, 2) were included in the set of best supported models with combined Akaike weig hts of 0.90 (90% confid ence set) and 3) had an evidence ratio relative to the best supported model greater than 0.082 (Burnham and Anderson 2002). The goodness-of-fit a nd other diagnostics for meeting GLM assumptions were investigated for all mode ls; only those meeting the basic assumptions of a GLM were considered fo r interpretation (Grafen and Hails 2002). All statistical calculations were performed using either Microsoft Access, Microsoft Excel 2003 (Microsoft Corporation, 1985-2003) or SPSS 12.0 (SPSS Inc. 1988-2003).

PAGE 137

120 T10K T10L T10J T10H T10D T10C T10E SHILOH 3 SHILOH 1 T27B T38I n d i a n R i v e rM o s q u i t o L a g o o nA t l a n t i c O c e a nKennedy Space Center / Merritt Island National Wildlife Refuge 02,2504,5006,7509,000 1,125 Meters Figure 5.1 Map of study site showing locat ion of several study impoundments on the Kennedy Space Center / Merritt Island National Wildlife Refuge.

PAGE 138

121 Results Prey Density at Used and Paired Random Sites Between January and May 2001 and Marc h and May 2002, the foraging behavior of 100 wading birds foraging in aggregations ranging in size from 1 450 individuals was recorded from 63 sites. The density of fish at sites used by wading birds was not significantly greater th an at paired randomly selected unused sites (Table 5.1). The biomass of fish at sites used by wading bird s was significantly greater than at paired unused sites (Table 5.2). Th e density of fish at sites used by solitary foraging wading birds was significantly less than that at sites used by foraging groups (Table 5.1). Similarly, the mass of fish at sites used by foraging individual s was less than that at sites used by foraging groups, but the difference was not significant (Table 5.2). Both the mean length and the mean mass of fish at si tes at which wading bird s were foraging were greater than at paired random sites (Table 5.3) The correlations between fish density and the mean length or the mean biomass of fi sh at sites used by wading birds were not significant (Table 5.4). There was a significant correlation be tween the mean length and the mean biomass of fish at sites used by wading birds (Table 5.4). Behavior-Prey Sampling Information-theoretic model selection with Great Egret (captures/min)1/2 as the response variable resulted in 3 models with a combined Akaike weight of 0.89 (Table 5.5). The best-supported model included only the binary group variable, and predicted that Great Egrets had higher foraging success when in groups (Tables 5.6 and 5.7, Figure 5.2). The next best model included the bina ry group variable and the covariate prey density (Tables 5.8 and 5.9). This model pred icted a similar effect for the group variable as the simpler model, but also suggested th at the relationship betw een capture rate and

PAGE 139

122 prey density was inverse for Great Egret. The final model included the categorical group variable (Tables 5.10 and 5.11). This model also predicted that captu re rate was greater for individuals foraging in groups versus sol itary foragers. The model also indicated a tendency for capture rate to level off or slig htly decline as group size increased above the small group size category (Figure 5.3). Information-theoretic model selection with Snowy Egret (captures/min)1/2 as the response variable resulted in 4 models with a combined Akaike weight of 0.94 (Table 5.12). The best-supported model included the bi nary group variable and predicted that Snowy Egrets had higher foraging success when foraging alone than in groups (Tables 5.13 and 5.14, Figure 5.2). The next best mode l included both the covariate prey density and the binary group variable (Tables 5.15 a nd 5.16). This model predicted a similar effect for the group variable as the simpler m odel, and also suggested that capture rate increased as prey density increased for Snow y Egret. The next model selected included only the covariate prey density, however th is model did not fit the data very well (multiple coefficient of determination belo w 0.05) and thus no further consideration was given to this model. The final model incl uded the interaction between the categorical group variable and the cova riate prey density (Table s 5.17 and 5.18). This model predicted that solitary Snowy Egrets had higher capture rates than individuals foraging in a group of any size category (Figure 5.4). Information-theoretic model selection with Tricolored Heron (captures/min)1/2 as the response variable resulted in 4 models with a combined Akaike weight of 0.94 (Table 5.23). However, none of these models fit the data very well (all models had a multiple

PAGE 140

123 coefficient of determination below 0.1) and thus no further consideration was given to these models. A summary of foraging rates for Great Eg rets (Table 5.24), Snowy Egrets (Table 5.25), and Tricolored Herons (Table 5.25) are provided for comparison with other studies. There were no significant differences in the mean lengths of fish captured by solitary foragers versus those in groups for any species (Great Egret t=0.966, df=15, p=0.350; Snowy Egret t=-0.461, df=15, p=0.652; Tricolored Heron t=0.36, df=21, p=0.970); differences are shown in Figure 5.4.

PAGE 141

124 Table 5.1. Mean fish density and mean total biomass (with 95% confidence intervals) at sites used by foraging wading birds and paired random sites. Measurements are of fish/m2 and were back-transformed from estimates calculated with ln(fish density+1). n t adf ap aFish density (fish / m2) 7.25(5.15 10.06)5.81(3.93 8.42)621.38610.09 Biomass (g / m2)4.01(2.77 5.66)2.70(1.77 3.95)622.23610.01 Used SitesRandom Sites a Results of paired t-test for Ho: The difference between means at used and paired random sites was positive. Table 5.2. Mean fish density and mean total biomass (with 95% confidence intervals) at sites used by foraging individuals and groups of wading birds. Measurements are of fish / m2 and were back-transformed fr om estimates calculated with ln(fish density+1). nn t adf ap aFish density (fish / m2) 4.55(2.01 9.24)219.10(6.35 12.88)41 1.8031.90.04 Biomass (g / m2)2.74(1.245.25)214.82(3.12 7.23)41 1.4839.30.07 Individual SitesGroup Sites a Results of separate variance unpaired t-test for Ho: The difference between means at sites occupied by groups vers us sites occupied by individuals was positive. Table 5.3. Mean length and m ean mass (with 95% confidence intervals) of fish in samples at sites used by wading birds and paired-unused sites. Values in table are back-transformed to original units. Random sitesUsed sites tadfapaMean Length (cm)21.6(20.2, 23.0)23.0(21.6, 24.5)-1.89550.03 Mean Biomass (g)0.35(0.28, 0.44)0.48(0.38, 0.60)-2.57550.01 a Results of paired t-test for Ho: The difference between means at used sites and paired random sites was positive. Table 5.4. Correlations between mean length and mean mass of fish in samples at sites used by wading birds. The sample size was 120 for all comparisons. Fish density aMean fish length aMean fish biomass aFish density aPearson r1-0.020.04 p (2-tailed)0.810.64 Mean fish length aPearson r-0.02210.92 p (2-tailed)0.8090.00 Mean fish biomass aPearson r0.040.921 p (2-tailed)0.640.00 a The data was transformed as ln(fish density +1), ln(mean fish length), and ln(mean fish biomass).

PAGE 142

125 Table 5.5. Information-theoretic model selecti on results for GLM analysis of Great Egret (captures / min)1/2 as response variable, prey de nsity (PD) as covariate and binary group variable (GB) or group category (GC) as a factor. Model MLE 2kAIC AIC AICc AICc n/k wiwi / wi bestGB0.363-25.750.00-24.860.0010.330.641.000 GB, PD0.364-24.001.75-22.462.407.750.190.301 GC0.355-22.293.45-19.894.966.200.050.084 PD0.423-20.555.19-19.675.1910.330.050.075 GB*PD0.365-22.033.72-19.635.236.200.050.073 GC, PD0.356-20.575.18-17.077.795.170.010.020 GC*PD0.289-21.194.55-12.6212.243.440.000.002 Table 5.6 ANOVA table for GLM of Great Egret (captures / min)1/2 as response variable with the binary group variab le (GB) as a factor. Source Type III Sum of Squaresdf Mean SquareFSig. Corrected Model2.0612.065.370.03 Intercept9.6119.6125.040.00 Group2.0612.065.370.03 Error11.13290.38 Total32.0031 Corrected Total13.2030 R2 = 0.156 Table 5.7 Parameter estimates for GLM of Great Egret (captures / min)1/2 with the binary group variable (GB) as a factor. Parameter SE 95% CI Intercept0.930.13(0.67 1.20) Group category = individual-0.590.25(-1.11 -0.07) Group category = group 0a a This parameter is set to zero because it is redundant.

PAGE 143

126 Table 5.8 ANOVA table for GLM of Great Egret (captures / min)1/2 as response variable with prey density (PD) as covariate a nd the binary group variable (GB) as a factor. Source Type III Sum of Squaresdf Mean SquareFSig. Corrected Model2.1521.082.730.08 Intercept2.6312.636.680.02 Group2.1212.125.370.03 ln(Fish density+1)0.0910.090.230.64 Error11.04280.39 Total32.0031 Corrected Total13.2030 R2 = 0.163 Table 5.9. Parameter estimates for GLM of Great Egret (captures / min)1/2 as response variable with prey density (PD) as covariate and the binary group variable (GB) as a factor. Parameter SE 95% CI Intercept1.060.31(0.44 1.69) Group category = individual-0.600.26(-1.13 -0.07) Group category = group 0aln(fish density + 1)-0.060.12(-0.29 0.18) a This parameter is set to zero because it is redundant. Table 5.10 ANOVA table for GLM of Great Egret (captures / min)1/2 as response variable with group category (GC) as a factor. Source Type III Sum of Squaresdf Mean SquareFSig. Corrected Model2.2630.751.860.16 Intercept16.16116.1639.890.00 Group category2.2630.751.860.16 Error10.94270.41 Total32.0031 Corrected Total13.2030 R2 = 0.171

PAGE 144

127 Table 5.11 Parameter estimates for GLM of Great Egret (captures / min)1/2 as response variable with group category (GC) as a factor. Parameter SE 95% CI Intercept0.990.18(0.61 1.37) Individuals-0.650.29(-1.24 -0.05) Small groups-0.230.34(-0.93 0.46) Medium groups-0.030.32(-0.68 0.63) Large groups 0a a This parameter is set to zero because it is redundant. Table 5.12. Information-theoretic model select ion results for GLM analysis of Snowy Egret (captures / min)1/2 as response variable, prey density (PD) as covariate and binary group variable (GB) or group category (GC) as a factor. Model MLE 2kAIC AIC AICc AICc n/k wiwi / wi bestGB0.313-30.310.00-29.420.0010.330.481.00 GB, PD0.304-29.690.62-28.161.267.750.250.53 PD0.333-27.912.40-27.022.4010.330.140.30 GB*PD0.305-27.702.61-25.304.126.200.060.13 GC0.305-27.073.24-24.674.766.200.040.09 GC, PD0.296-26.523.79-23.026.405.170.020.04 GC*PD0.289-21.608.71-13.0316.393.440.000.00 Table 5.13 ANOVA table for GLM of Snowy Egret (captures / min)1/2 as response variable with the binary group variable (GB) as a factor. Source Type III Sum of Squaresdf Mean SquareFSig. Corrected Model1.2711.273.830.06 Intercept10.39110.3931.360.00 Group1.2711.273.830.06 Error9.61290.33 Total22.0731 Corrected Total10.8830 R2 = 0.117 Table 5.14 Parameter estimates for GLM of Snowy Egret (captures / min)1/2 with the binary group variable (GB) as a factor. Parameter SE 95% CI Intercept0.510.11(0.28 0.74) Group category = individual0.550.28(-0.03 1.13) Group category = group 0a a This parameter is set to zero because it is redundant.

PAGE 145

128 Table 5.15. ANOVA table for GLM of Snowy Egret (captures / min)1/2 as response variable with prey density (PD) as covariate and the binary group variable (GB) as a factor. Source Type III Sum of Squaresdf Mean SquareFSig. Corrected Model1.6920.842.570.09 Intercept1.2911.293.920.06 Group1.1911.193.640.07 ln(Fish density+1)0.4210.421.280.27 Error9.19280.33 Total22.0731 Corrected Total10.8830 R2 = 0.155 Table 5.16. Parameter estimates for GLM of Snowy Egret (captures / min)1/2 as response variable, prey density (PD) as covariat e and the binary group variable (GB) as a factor. Parameter SE 95% CI Intercept0.260.25(-0.25 0.77) Group category = individual0.530.28(-0.04 1.11) Group category = group 0aln(fish density + 1)0.110.10(-0.09 0.31) a This parameter is set to zero because it is redundant. Table 5.17. ANOVA table for GLM of Snowy Egret (captures / min)1/2 as response variable with the interac tion between prey density (PD) and the binary group variable (GB). Source Type III Sum of Squaresdf Mean SquareFSig. Corrected Model1.6930.561.660.20 Intercept1.0711.073.150.09 Group category0.2510.250.730.40 ln(prey density + 1)0.3710.371.070.31 Group category ln(prey density + 1) 0.0010.000.000.95 Error0.3710.37 Total0.001 Corrected Total9.1927 a R2 = 0.155

PAGE 146

129 Table 5.18. Parameter estimates for GLM of Snowy Egret (captures / min)1/2 as response variable with the interac tion between prey density (PD) and the binary group variable (GB). Parameter SE 95% CI Intercept0.270.29(-0.33 0.87) Individuals0.500.59(-0.71 1.71) Groups 0aln(prey density + 1)0.110.12(-0.14 0.35) Individual ln(prey density + 1) 0.010.22(-0.43 0.46) Group ln(prey density + 1)0a a This parameter is set to zero because it is redundant. Table 5.19. Information-theoretic model selectio n results for GLM analysis of Tricolored Heron (captures / min)1/2 as response variable, prey density (PD) as covariate and binary group variable (GB) or group category (GC) as a factor. Model MLE 2kAIC AIC AICc AICc n/k wiwi / wi bestPD0.363-29.620.00-28.840.0011.670.451.00 GB0.373-28.740.88-27.970.8811.670.290.65 GB, PD0.364-27.831.79-26.502.348.750.140.31 GB*PD0.355-26.832.79-24.764.097.000.060.13 GC0.365-25.903.72-23.835.017.000.040.08 GC, PD0.346-25.564.06-22.566.285.830.020.04 GC*PD0.329-22.227.40-15.0213.823.890.000.00 Table 5.20. Summary of foraging ra tes of Great Egrets in vari ous sized groups. Values in table are back-transformed mean s (with 95 % confidence limits). IndividualSmallMediumLarge Paces / min.15.16(8.75, 23.31)15.16(5.13, 30.48)14.14(1.72, 38.56)11.78(5.34, 20.73) Captures / min.0.12(0.00, 0.45)0.57(0.00, 2.28)0.78(0.59, 6.41)0.88(0.27, 1.85) Strikes / min.0.37(0.06, 0.96)0.68(0.00, 2.59)1.18(0.24, 7.10)1.62(0.85, 2.64) Success rate0.27(0.03, 0.50)0.81(0.10, 0.97)0.58(-0.56, 0.96)0.65(0.37, 0.86) Table 5.21. Summary of foraging rates of Snowy Egrets in various sized groups. Values in table are back-transformed mean s (with 95 % confidence limits). IndividualSmallMediumLarge Paces / min.42.07(23.85, 65.4140.99(20.80, 67.9721.80(2.60, 59.66)31.45(17.61, 49.29 ) Captures / min.1.13(0.01, 4.06)0.31(0.02, 0.96)0.34(0.55, 3.65)0.26(0.03, 0.69) Strikes / min.3.88(0.11, 12.98)0.82(0.14, 2.05)2.11(0.48, 4.89)1.53(0.76, 2.57) Success rate0.27(-0.02, 0.53)0.45(0.08, 0.75)0.24(-0.30, 0.71)0.27(0.05, 0.48)

PAGE 147

130 Table 5.22. Summary of foraging rates of Tricolored Herons in various sized groups. Values in table are back-transformed means (with 95 % confidence limits). IndividualSmallMediumLarge Paces / min.37.95(20.34, 61.0228.32(13.03, 49.4814.19(1.37, 75.80)20.50(4.43, 48.32) Captures / min.0.45(0.14, 0.94)0.49(0.03, 1.52)0.60(0.01, 2.75)0.38(0.02, 1.21) Strikes / min.1.79(0.63, 3.55)2.10(0.98, 3.63)1.34(0.11, 7.02)1.08(0.15, 2.85) Success rate0.51(0.00, 0.88)0.31(0.09, 0.52)0.56(-0.48, 0.99)0.44(0.02, 0.79)

PAGE 148

131 Great EgretSnowy Egret 0 0.5 1 1.5 2 2.5 3Captures / minGreat EgretSnowy Egret 0 0.5 1 1.5 2 2.5 3Captures / min Figure 5.2. Great Egrets had a hi gher capture rate when foraging in groups than when foraging alone, while Snowy Egrets had hi gher capture rate fo raging solitarily. The figure shows estimated mean capture rate for the model including the binary group variable. The values are back-transformed to the original units; error bars show 95% confidence interval.

PAGE 149

132 0 0.5 1 1.5 2 2.5Individual Small (2-10) Medium (11-50) Large (>51)Captures / min 0 0.5 1 1.5 2 2.5Individual Small (2-10) Medium (11-50) Large (>51)Captures / min Figure 5.3. Great Egrets increased their capt ure rate as group size increased. Figure shows estimated mean capture rate for the model with group category as the explanatory variable. Values are back-t ransformed to the original units; error bars show 95% confidence interval.

PAGE 150

133 Discussion Among factors known to influence the fora ging success of individual wading birds, prey availability and habitat structure are probably th e best understood (Chapter 4). However, there are additional factors that may affect individual foraging success. These include social interactio ns (Kushlan 1978c), preda tion (Caldwell 1986, Butler and Vennesland 2000, Rivers 2000), disturbance by humans (Stolen 2003), and proximity to other foraging birds (Kushlan 1978a, Caldwell 1981). Proximity to other foragers is particularly relevant for piscivorous wadi ng birds because foraging aggregations are a conspicuous element of their behavior (e.g., Chapter 4, Freder ick and Bildstein 1992, Smith 1995b, Bouton et al. 1999, Custer et al. 2004) In this chapter, I investigated the relative importance of proximity to other fo ragers by collecting co mprehensive data on prey levels and foraging activity by both so litary and group foragers. Many previous studies have found a benefit to individuals of foraging in groups, but were not able to separate site characteristics fr om the effects of proximity to other foragers (Table 5.27). I designed this study hypothesizing that an explan ation for the lack of generalized results of previous studies was that most lacked simultaneous, comprehensive data on prey levels and foraging activity by both solitary and group foragers. Consistent with past studies I found that foraging sites had higher prey density and significantly higher prey biomass than nearby unused sites. Also, I found that foraging sites occupied by groups tended to have higher prey densitie s than did sites occupied by individuals. Furthermore, fi sh were larger at sites used by wading birds. Summarizing these results, wading birds are locating patche s of prey, and groups occur at sites with higher prey levels than do solitary foragers Thus, to understand how individuals benefit from foraging in groups, one must consider the ef fects of increased prey levels as well as

PAGE 151

134 the effect of other foragers on foraging success. It is also important to remember that prey density does not always equate to av ailability, and many factors can influence availability (Gawlik 2002). In this study, I attempted to contro l for factors affecting prey availability by restricting sa mpling to one habitat type wi thin a small geographic region, thus keeping differences between sites small. In the case of Great Egrets, I found that foraging success increased for individuals in groups over that of solitary foragers (F igure 5.2). This effect was separate and stronger than the effect of prey density al one. This finding suppor ted the hypothesis that the benefits were due to so cial facilitation. In contra st, Snowy Egret foraging success increased in solitary foragers compared to group foragers. This is evidence against the social facilitation hypothesis for this species, and suggests that individual Snowy Egrets may pay a price for foraging in groups. So me have speculated that Snowy Egrets may serve as food finders for other members of mixed-species foraging aggregations (Kushlan 1977, Caldwell 1981, Master 1992, Gawlik 2002). My results suggest that the benefits of foraging alone might be short-lived, decreasi ng as other foragers join them. Snowy Egrets may offset, at least temporarily, some of the disadvantages of foraging in groups because groups in this study foraged in sites of higher prey density. Within sites, I did not find that Snowy Egrets foraging in groups were capturing larger fish (i.e., greater biomass) as another means to offset costs, but my power to detect this was low. Offsetting costs of group foraging by mean s of choosing sites with higher density, however, does not resolve larger questions regarding the evol ution of foraging strategies for the species. For example, why are Snow y Egrets so conspicuous? One answer may be that under circumstances different from those I studied, individual Snowy Egrets may

PAGE 152

135 derive benefits form the presence of othe r foragers. One possibility was raised by Caldwell (1981) who speculated that Snowy Eg rets benefited from the proximity to subordinate species (e.g., Tricol ored Heron, Little Blue Heron) that they could displace from prey within a foraging site. In th is study the high proporti on of (presumably) dominant Great Egrets in foraging groups (Chapt er 4) may have diminished this benefit. Another possibility is that information about foraging sites is more important to individuals than is maximizing short-term foraging success (Valone 1991). In support of this explanation is the fact that more than half of the Snowy Egrets followed from colonies on KSC/MINWR left in groups and almost all join ed groups at foraging sites (Chapter 2). The lack of strong evidence for a positive relationship between prey density and capture rate for both species was notewort hy. This was unexpected since many studies have suggested that wading bird populations require prey to become concentrated in order for individuals to forage efficiently in groups (Kushlan 1976b, Hafner et al. 1982, Cezilly et al. 1990, Gawlik 2002). A plausible explanation for this seeming discrepancy is that the ranges of prey de nsity recorded in this study, while not markedly high, were still within the range of maximum foraging efficiency (Krebs 1974, Miranda and Collazo 1997). If true, this suggests that at least under some circumstances, prey density is not the most important factor affecting wading bi rd foraging success. My results highlight the fact that wading birds are highly opportunistic fo ragers capable of rapid responses to changing habitat conditions (Strong et al. 1997, Gawlik 2002, Frederick and Ogden 2003). Generalized conclusions about the benef its and costs of feeding in aggregations

PAGE 153

136 must consider the context within which they occur (e.g., prey type and density, vegetative cover, water depth, turbidity, composition of aggregations). I documented that the benefit to Great Egre ts and cost to Snowy Egrets foraging in groups was in large part due to the presence of other foragers. However, the mechanism involved is unknown. The next logical step w ould be to conduct detailed behavioral observations of individual wading birds foragi ng in groups to document how they interact with other nearby foragers (Petit and Bildstein 1987). Also, observations on prey behavior at sites at which a ggregations are foraging would be very helpful. However, this could prove difficult in the often mur ky water where wading bird aggregations feed, and experimental invest igations of prey responses to pr edators would also be useful. Experimental manipulation of the habitat (e .g., prey refugia) or prey composition (e.g., cryptic versus conspicuous, shoaling versus non-shoaling) would allow specific tests of hypothesized mechanisms for the social facil itation effect. The causes and consequences of mixed-species wading bird foraging aggr egations is a rich topic deserving much further investigation.

PAGE 154

137 Table 5.23. Summary of studies investigating the benefits to wading birds of foraging in aggregations compared to foraging solitarily. For each study, the table gives the species investigated, whether or not results indicated higher foraging success of individuals in groups over those foraging al one and whether or not prey density was measured. The table also categorizes the studies based on whether they supported enhanced foraging site characteristic s versus social facilitation as the explanation for th e benefit of foraging in groups. Author Species aForaging success Explanation bPrey Density Measure d higher in groups Krebs (1974)GBHEYesSiteYes Erwin (1985)GREG, SNEGNo cNo Caldwell (1981)GREG, SNEG, TRHE, LBH E Yes dSiteNo Wiggins (1981)GREGYes eSiteNo Master et al. (1993)SNEGYes fSocial Faciltiatio n YesTRHE, GREGNo Hafner et al. (1982)LIEGYesSiteNoSQHENo Erwin et al, (1985)LIEGMixed gSocial Faciltiatio n Yes Cezilly et al. (1990) LIEGYesSite This studyGREGYesSocial Faciltiatio n YesSNEG, TRHENo a GBHE = Great Blue Heron, GREG = Great Egret, SNEG = Snowy Egret, TRHE = Tricolored Heron, LIEG = Little Egrets ( Egretta grazatta ), SQHE = Squacco Herons ( Ardeola ralloides ) b Site = benefits due to site characteristics (e.g., prey availability); Social Facilitation = benefits due to presen ce of other foragers. c However, foragers stayed in patches longer when in groups than when alone. d Enhanced foraging success was due to pr oximity to Snowy Egrets which were presumed to be better at locating prey patches. e However, solitary birds caught larger fish causing equal food intake between categories. f Also, increased variance in capture rate, decrease in success rate. g Birds in largest group showed increased fora ging success rate, but intermediate-sized groups did worse than individuals.

PAGE 155

138 CHAPTER 6 CONSERVATION IMPLICATIONS AND MANAGEMENT RECOMMENDATIONS Wading birds are intricately ti ed to their wetland habita ts, which provide resources for all aspects of their existence. It follows that the future of wading birds in Florida is tied to protection of wetlands, and where possible (e.g., managed wetlands), multi-species or integrated management. It is also clear that to formulate and implement a suitable multi-species management strategy, a better und erstanding of several features of their ecology is needed. My goal in conducting this research was to learn more about factors that influence individual pisc ivorous wading birds when they make choices about what type of foraging habitat to use, and whether or not to forage with others. In this chapter I summarize the key findings of this dissertati on and how they relate to conservation of wading birds and management of their fora ging habitat at the Kennedy Space CenterMerritt Island National Wildlife Refuge (KSC/MINWR) in the northern Indian River Lagoon system (IRL). In Chapter 2, I showed that wading bird s nesting in the no rthern IRL select foraging habitat within impounded wetlands at KSC/MINWR. This was an indication that these habitats pr ovide critical re sources for wading bird p opulations at KSC/MINWR while under the high energetic demands of nest provisioning (Powell 1983, Hafner et al. 1993). As with other studies (e.g., Klei n 1993, Klein et al. 1995, Erwin 1996, Erwin et al. 2005), I found that impounded wetlands were suitable foraging areas because they provide extensive open habitat interspersed with vegetation. Vegetated areas harbored high densities of prey and as such may aff ect the foraging habitat quality of open areas

PAGE 156

139 for wading birds. While this is the true, my results should not be construed to imply that other wetland types in northern IRL (e.g., estu arine edge and freshw ater wetlands) are not important for wading birds. Although distan ce and availability make these wetland types less attractive to wading bird s nesting in the colonies on KSC/MINWR, circumstances (e.g., droughts) might temporarily reduce th e quality of impounded wetlands and thus raise the relative value of alternative sites. I caution managers agai nst focusing solely on impounded wetlands and ignoring the condition of these other wetlands types. Because almost all of the fringing marshes of the northern IRL were impounded for mosquito control by the 1970s, it is not possible to determine the effect of impounding these wetlands on wading bird habitat preference. I could only evaluate preference under the current habitat availability context. However, because several impoundments have recently been restored (i.e., breeching or removal of perimeter dike), it would be possible to address this question by studying wading bi rd habitat use or foraging success within restored versus impounded systems. An even better approach woul d be to apply a preintervention and post-intervention study design with appropriate controls (James and McCulloch 1995) in impoundments that are sche duled to be restored in the future. In Chapter 3 I showed that impounded salt marsh habitat can produce large populations of small, resident fish under ce rtain conditions. I found that vegetated sites generally had higher densities of these fish. Previous work has shown that wading birds prefer to forage in unvegetated shallow wate r habitat over vegetate d habitat (Breininger and Smith 1990, Smith and Breininger 1995), and that wading bird use of foraging habitat at KSC/MINWR can be partly explaine d by the amount of interspersion of open and vegetated habitat (Stolen et al. 2002). My results in Chapter 4 confirmed this for

PAGE 157

140 most species. Furthermore, I found that preference for foraging in the area where unvegetated and vegetated habita ts adjoin (edge habitat) wa s strong for all species of wading birds observed. This had not been de monstrated previously, and has important implications for management of wading bird foraging habitat. In Chapter 4, I documented the high pr evalence of mixed-species wading bird foraging aggregations on KSC/MINWR. Fi nally, in Chapter 5, I showed that such aggregations occurred at site s with elevated prey levels This confirms that under prevailing conditions on KSC/ MINWR wading birds can locate foraging sites that have more fish than the average level available th roughout the landscape. I also found that at least one species, the Great Egret ( Ardea alba ), benefited from foraging in these aggregations under certain conditions. Hi gher prey densities at sites used by aggregations might offset some of the co sts of foraging in mixed-species groups for Snowy Egrets ( Egretta thula ), a species I found to forage more efficiently in solitary conditions. Further research is needed to investigate how Snowy Egrets cope with group foraging in impounded wetlands on KSC/ MINWR in light of my findings. Master (1992) reviewed th e literature on wading bird aggregations, and concluded that patchy prey distribution, high prey av ailability and low defensibility are key characteristics of sites with aggregations. He further concluded th at alternating periods of abundance and decline characterize habitats where aggregations have been observed. Similarly, Caldwell (1981) described sites where group foraging is prevalent as having prey that is abundant, seasona lly predictable, clumped, and mobile. Managers should strive to maintain these conditions with in the impounded wetlands on KSC/MINWR. Attaining such conditions, howev er, are less likely to succeed if management is focused

PAGE 158

141 on creating large concentrations of prey w ithin systems. One obvious reason has to do with factors affecting availa bility of prey, particularly water depth. Although longer hydroperiods are correlate d with increased standing stocks of resident fish within impoundments (Brockmeyer 2004), if water levels are kept above th at at which wading birds can forage (typically less than 30 cm) high levels of prey may not benefit wading birds. Instead, management should strive to create a variety of hydrologic conditions over an area with suitable habitat for bot h prey production and wading bird foraging. A strong result of my work was the findi ng that wading birds greatly prefer the edges between wetlands with and without emergent vegetation. Management of impounded wetlands on KSC/MINWR should strive to maintain and enhance such edge habitat, when possible within the context of management for other species (multi-species management). This may best be achieved by maintaining an interspersion of open water and vegetated habitats within impoundments, and by allowing seasonal variation in water levels. Since the impoundments on KSC/MI NWR have an abundance of unvegetated, shallow areas, emphasis would be best placed on maintaining and enhancing the condition of vegetated areas that are flooded fo r part of the year. Such areas may be important in production of fish populations, and also are important foraging habitat during winter when water leve ls are naturally higher thr oughout the Indian River Lagoon estuary (Stolen et al. 2002). In support of this, I found that wading bird density within impoundment sub-sections increased with the amount of vegetated habitat within. Although the model with prey density and fo raging habitat type was able to explain some of the variation in the pattern of wading bird density among impoundments (Chapter 4), I did not detect a strong relationship between prey density and wading bird

PAGE 159

142 density under the conditions that prevaile d throughout the period of observation. There was also evidence for a seasonal difference in wading bird foraging habitat use between impoundments that had more hydrologic connecti on with the estuary and those that did not. When interpreting these results it should be noted that during this period there were no rapid changes in hydrology with in these impoundments; othe r studies in the IRL have noted strong numerical respons es by wading birds to unusually dense concentrations of their prey resulting from rapid drying (Schikorr and Swain 1995, Stevens 2002). In some southern IRL impoundments, rapid drying (termed drawdowns) has been shown to increase usage by foraging wading birds (Schikorr and Swain 1995, Sewell et al. 1996), and these authors pr oposed that increasing the us e of drawdowns would be beneficial to wading bird populations in th e IRL. Drying of wetlands has also been shown to be very important for wading bird foraging habitat use in the Everglades (Bancroft et al. 1994, Frederick and Spalding 1994, Hoffman et al. 1994). While this mechanism may be useful in creating shortterm enhancement of foraging habitat for wading birds, 2 cautions are apparent. First, in impoundments, the presence of perimeter dikes is a barrier to the movement of fi sh tracking the interf ace between flooded and drying areas. In the Everglades, it is the exis tence of sustained drying that is associated with the strongest positive relationship with nesting (Frederick and Ogden 2003), and reversals of these patte rns is often associated with ne st failure (Frederick and Collopy 1989, Smith and Collopy 1995). Wading birds take advantage of concentration of fish along a drying front as water levels fall. In impoundments of the northern IRL, such patterns of drying in wetlands may be truncat ed by dikes, leaving wading birds without suitable foraging habitat. Since impoundi ng, seasonal hydrologic patterns haves

PAGE 160

143 changed, and often water and fish are cont ained in impoundments by dikes. A similar situation exists in the Water Conservation Areas in the northern Everglades where water depths are often too deep for foraging by wadi ng birds (Bancroft et al. 1994, Bancroft et al. 2002). Another consideration is fish standing stocks. There is a danger that repeated drying of marshes might reduce the standi ng stock of fish within impoundments over time. In this study, the highest density of fi sh, and the highest level of wading bird use occurred within the managed impoundments wh ich were closed during the course of the study. Within the open impoundments, intense dr ying resulted in little suitable wading bird habitat and lower densitie s of fish during some periods. Frederick and Ogden (2003) point out that in the historic Everglades system, with its longer hydroperiod and perhaps larger standing stock of small fish, some speci es of wading birds proba bly did not rely as heavily on drying of marshes to concentrate fish. Wading birds can be very useful as bi oindicators of the h ealth of estuarine ecosystems (Stolen et al. 2005). Useful a ttributes of wading bird populations as indicators within the northern IRL include their use of large spatial extent, their sensitivity to contaminants, the rapid ity with which populations can respond to perturbations, and the numerous species from which to select an indicator. These attributes provide the ability to tailor the indicator to a sp ecific ecosystem attribute. Wading bird indicators have proven capabl e of uncovering problems in ecosystems before specific causal mechanisms are known (Frederick and Ogden 2003). In addition to the use of foraging habitat, other important wading bird indicators to monitor in the

PAGE 161

144 Northern IRL include nesting success, para sites and contaminants. Managers should encourage research into these areas on KSC/MINWR Reversing population declines and increasing the like lihood of population persistence of wading birds in Florida, and elsewhere, are ultimately the goals of conservation efforts and natural resources ag encies. To help meet these challenges, explicit linkages between foraging success and re productive and survival rates need to be established (Chapter 1, Figure 1.2). It is only then that more definitive statements about the local population status and persistence pr obability can be made In the KSC/MINWR (northern IRL) wetlands complex, my work la id a strong foundation to facilitate making those linkages by defining the scales and elucid ating factors involved in the selection of foraging sites by wading birds breeding in the area. Distance to foraging sites, their quality (e.g., habitat type, prey density) and at tributes (e.g., edges, de pth) suggested that those waders could be meeting many, if not mo st, of their energetic requirements within the KSC/MINWR wetlands complex. The demo graphic relevance of the area to local population dynamics is further strengthened by the fact that the said breeding colonies have been active for over 15 year s (E. Stolen, unpublished data).

PAGE 162

APPENDIX A WATER LEVELS WITHIN IMPOUNDM ENTS AND INDIAN RIVER LAGOON ESTUARY MEASURED DURING WETLANDS INITIATIVE

PAGE 163

146 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3F eb-00 Apr 00 Jun 0 0 Aug 0 0 Oct -00 Dec-00 Feb-01 Apr0 1 Jun-01 Aug 0 1 Oct-01 Dec-01 Feb-02 Apr-02 J u n-02 A ug0 2 Oct0 2 D ec-0 2 F e b-03 C_inside D_inside E_inside H_inside L_inside J_inside Figure A.1. Mean monthly water levels within 6 impoundments on KSC/MINWR between February 2000 and February 2003.

PAGE 164

147 -0.4 -0.35 -0.3 -0.25 -0.2 -0.15 -0.1 -0.05 0 0.05F e b-00 Ap r -0 0 Ju n-00 Aug 00 Oct -00 Dec-00 Feb-01 Ap r -0 1 Ju n -01 Aug -0 1 Oct -0 1 Dec-0 1 F eb 0 2 Apr-0 2 Jun-02 Aug-0 2 Oct-02 Dec 0 2 F eb -03 CD_outside EH_outside JL_outside Figure A.2. Mean monthly water levels at 3 gaug e stations near study im poundments on KSC/MINWR between February 2000 and February 2003.

PAGE 165

148 APPENDIX B MODEL SELECTION RESULTS FOR IN DIVIDUAL WADING BIRD SPECIES Table B.1. Model selection results for GL M analysis with (Great Egret density)1/2 as response variable, fish density (P), wate r depth (D), and change in water depth (C) as covariates and fact ors: season (S, 2 levels), management (M, 2 levels,), and habitat (H, unvegetated or vegetated). Model MLE 2kAICAIC AICcAICc n/kexp(-0.5i)wiP, M0.055-143.890.00-142.530.00101.000.25 P0.053-142.441.45-141.920.60170.740.19 P*H0.055-143.140.75-141.770.75100.690.17 P*M0.047-143.630.26-140.961.5670.460.12 M0.054-140.932.96-140.052.48130.290.07 P, H0.054-140.583.31-139.692.83130.240.06 P, D0.054-140.483.41-139.592.93130.230.06 P*D0.055-138.765.13-137.395.13100.080.02 H0.063-137.176.73-136.645.88170.050.01 D0.063-136.906.99-136.386.15170.050.01 C0.063-136.837.06-136.316.22170.040.01 H*S0.055-137.146.75-135.776.75100.030.01 M*H0.057-137.706.19-135.047.4970.020.01 H, D0.064-135.318.59-134.428.11130.020.00 H, C0.064-135.298.60-134.408.13130.020.00 H*C0.065-134.899.00-133.539.00100.010.00 H*D0.065-134.109.79-132.749.79100.010.00 H*C, H*S, H*D, P*H, P*D, P*M0.0416-134.389.51-117.8924.6330.000.00

PAGE 166

149 Table B.2. Model selection results for GL M analysis with (Snowy Egret density)1/2 as response variable, fish density (P), wate r depth (D), and change in water depth (C) as covariates and fact ors: season (S, 2 levels), management (M, 2 levels,), and habitat (H, unvegetated or vegetated). Model MLE 2kAICAIC AICcAICc n/kexp(-0.5i)wiP*H0.015-212.670.00-211.300.00101.000.984 P, H0.024-201.4511.21-200.5710.74130.000.005 P*D0.015-201.1311.53-199.7711.53100.000.003 P0.023-199.7712.90-199.2512.06170.000.002 M0.024-198.7013.97-197.8113.49130.000.001 H0.023-197.9614.70-197.4413.86170.000.001 P, D0.024-198.2614.40-197.3713.93130.000.001 C0.023-197.2615.41-196.7414.56170.000.001 D0.023-197.2015.46-196.6814.62170.000.001 P, M0.025-197.5815.09-196.2115.09100.000.001 H, C0.024-196.1416.52-195.2516.05130.000.000 H, D0.024-196.1416.52-195.2516.05130.000.000 H*D0.025-195.7116.96-194.3416.96100.000.000 H*S0.025-195.2217.45-193.8617.45100.000.000 H*C0.025-194.4418.23-193.0718.23100.000.000 M*H0.027-195.7216.95-193.0518.2570.000.000 P*M0.027-194.1518.51-191.4919.8270.000.000 H*C, H*S, H*D, P*H, P*D, P*M0.0116-203.649.03-187.1524.1530.000.000

PAGE 167

150 Table B.3. Model selection results for GLM analysis with (Tricolored Heron density)1/2 response variable, fish density (P), wate r depth (D), and change in water depth (C) as covariates and fact ors: season (S, 2 levels), management (M, 2 levels,), and habitat (H, unvegetated or vegetated). Model MLE 2kAIC AIC AICc AICc n/kexp(-0.5 i)wiP*H0.025-191.820.00-190.460.00101.000.69 M0.024-186.465.36-185.584.88130.090.06aP, H0.024-185.885.94-184.995.47130.060.05aH0.023-185.066.76-184.545.92170.050.04aP0.023-185.026.80-184.505.96170.050.04aC0.023-184.477.35-183.956.51170.040.03aD0.023-184.027.80-183.506.96170.030.02 P, M0.025-184.467.36-183.107.36100.030.02 H, C0.024-183.668.16-182.777.69130.020.01 H, D0.024-183.208.62-182.318.15130.020.01 P, D0.024-183.068.76-182.178.28130.020.01 H*S0.025-182.599.23-181.239.23100.010.01 H*C0.025-181.8010.02-180.4410.02100.010.00 P*D0.025-181.8010.02-180.4410.02100.010.00 M*H0.027-182.978.85-180.3010.1670.010.00 H*D0.025-181.4810.35-180.1110.35100.010.00 P*M0.027-180.7111.11-178.0412.4270.000.00 H*C, H*S, H*D, P*H, P*D, P*M0.0216-173.9517.87-157.4732.9930.000.00 a Model results not presented below, because model did not fit data very well (multiple coefficient of determination below 0.09).

PAGE 168

151 Table B.4. Model selection results for GL M analysis with (White Ibis density)1/2 as response variable, fish density (P), wate r depth (D), and change in water depth (C) as covariates and fact ors: season (S, 2 levels), management (M, 2 levels,), and habitat (H, unvegetated or vegetated). Model MLE 2kAIC AIC AICc AICc n/kexp(-0.5 i)wiP*H0.145-89.050.00-87.690.00101.000.994 P, D0.194-74.3514.70-73.4614.23130.000.001 C0.203-73.9715.08-73.4514.24170.000.001 H0.203-73.9215.13-73.4014.29170.000.001 D0.203-73.8815.18-73.3514.34170.000.001 P, M0.185-74.4514.60-73.0914.60100.000.001 H*S0.195-73.9115.15-72.5415.15100.000.001 H*C0.195-73.9015.15-72.5415.15100.000.001 H, C0.204-72.0916.96-71.2016.49130.000.000 H, D0.204-71.9717.08-71.0916.60130.000.000 P*D0.195-72.3616.69-71.0016.69100.000.000 M*H0.177-73.3315.73-70.6617.0370.000.000 H*D0.205-69.9719.08-68.6119.08100.000.000 P*M0.187-70.7618.30-68.0919.6070.000.000 H*C, H*S, H*D, P*H, P*D, P*M0.1016-84.074.99-67.5820.1130.000.000 P0.233-66.5822.48-66.0621.63170.000.000 P, H0.234-64.5824.48-63.6924.00130.000.000 M0.234-64.5824.48-63.6924.00130.000.000 Table B.5. ANOVA table for GLM of (Great Egret density)1/2 with management type as a factor and fish density as a covariate. Source Type III Sum of Squaresdf Mean SquareFSig. Corrected Model 0.58a30.193.830.02 Intercept0.2410.244.770.03 management type0.2620.132.640.08 fish density0.2410.244.780.03 Error2.30460.05 Total4.1750 Corrected Total2.8849 a R2 = 0.200 (Adjusted R2 = 0.148)

PAGE 169

152 Table B.6. Parameter estimates for GLM of (Great Egret density)1/2 with management type as a factor and fish density as a covariate. Parameter SE 95% CI Intercept0.110.07(-0.04 0.25) management type = OPEN0.070.08(-0.10 0.24) management type = RIM-0.120.09(-0.29 0.06) management type = WAM 0afish density0.010.00(0.00 0.02) a This parameter is set to zero because it is redundant. Table B.7. ANOVA table for GLM of (Great Egret density)1/2 with fish density as a covariate. Source Type III Sum of Squaresdf Mean SquareFSig. Corrected Model 0.31a10.315.820.02 Intercept0.3110.315.800.02 fish density0.3110.315.820.02 Error2.57480.05 Total4.1750 Corrected Total2.8849 a Adjusted R2 = 0.90) Table B.8. Parameter estimates for GLM of (Great Egret density)1/2 fish density as a covariate. Parameter SE 95% CI Intercept0.100.04(0.02 0.18) fish density0.010.00(0.00 0.02) a This parameter is set to zero because it is redundant.

PAGE 170

153 Table B.9. ANOVA table for GLM of (Great Egret density)1/2 with the interaction between prey density and habitat type. Source Type III Sum of Squaresdf Mean SquareFSig. Corrected Model 0.54a30.183.550.02 Intercept0.2510.254.990.03 habitat0.0410.040.870.36 fish density0.4810.489.450.00 habitat fish density0.2210.224.390.04 Error2.34460.05 Total4.1750 Corrected Total2.8849 a R2 = 0.188 (Adjusted R2 = 0.135) Table B.10. Parameter estimates for GLM of (Great Egret density)1/2 with the interaction between prey density and habitat type. Parameter SE 95% CI Intercept0.140.07(-0.01 0.28) habitat = unvegetated-0.080.09(-0.26 0.10) habitat = vegetated 0afish density0.0040.004(-0.01 0.01) habitat = unvegetated fish density0.0180.008(0.00 0.04) habitat = vegetate d fish density 0a a This parameter is set to zero because it is redundant.

PAGE 171

154 Table B.11. ANOVA table for GLM of (Great Egret density)1/2 with the interaction between prey density and management type. Source Type III Sum of Squaresdf Mean SquareFSig. Corrected Model 0.74a50.153.060.02 Intercept0.2010.204.200.05 management type0.0520.030.520.60 fish density0.2210.224.510.04 management type fish densit y 0.1720.081.710.19 Error2.14440.05 Total4.1750 Corrected Total2.8849 a R2 = 0.26 (Adjusted R2 = 0.17) Table B.12. Parameter estimates for GLM of (Great Egret density)1/2 with the interaction between prey density and management type. Parameter SE 95% CI Intercept0.100.08(-0.06 0.25) management type = OPEN 0.030.10(-0.18 0.23) management type = RIM -0.0670.104(-0.28 0.14) management type = WAM 0afish density0.0090.005(0.00 0.02) management type = OPEN fish density0.0160.013(-0.01 0.04) management type = RIM fish density-0.0110.01(-0.03 0.01) management type = WAM* fish density 0a a This parameter is set to zero because it is redundant.

PAGE 172

155 Table B.13. ANOVA table for GLM of (Great Egret density)1/2 with management type as a factor. Source Type III Sum of Squaresdf Mean SquareFSig. Corrected Model 0.38a20.194.050.02 Intercept1.2011.2025.280.00 management type0.3820.194.050.02 Error2.70570.05 Total4.2860 Corrected Total3.0959 a R2 = 0.12 (Adjusted R2 = 0.094) Table B.14. Parameter estimates for GLM of (Great Egret density)1/2 with management type as a factor. Parameter SE 95% CI Intercept0.220.05(0.12 0.31) management type = OPEN-0.040.07(-0.17 0.10) management type = RIM-0.190.069(-0.32 -0.05) management type = WAM 0a a This parameter is set to zero because it is redundant. Table B.15. ANOVA table for GLM of (Great Egret density)1/2 with habitat type as a factor and fish density as a covariate. Source Type III Sum of Squaresdf Mean SquareFSig. Corrected Model 0.32a20.162.920.06 Intercept0.2310.234.240.05 habitat0.0110.010.120.73 fish density0.2910.295.370.03 Error2.56470.05 Total4.1750 Corrected Total2.8849 a R2 = 0.11 (Adjusted R2 = 0.073)

PAGE 173

156 Table B.16. Parameter estimates for GLM of (Great Egret density)1/2 with habitat type as a factor and fish density as a covariate. Parameter SE 95% CI Intercept0.080.07(-0.06 0.22) habitat = unvegetated0.030.07(-0.12 0.17) habitat = vegetated 0afish density0.0090.004(0.00 0.02) a This parameter is set to zero because it is redundant. Table B.17. ANOVA table for GLM of (Great Egret density)1/2 with fish density and water depth as covariates. Source Type III Sum of Squaresdf Mean SquareFSig. Corrected Model 0.31a20.162.870.07 Intercept0.0310.030.560.46 depth0.0010.000.030.86 fish density0.3010.305.550.02 Error2.57470.06 Total4.1750 Corrected Total2.8849 a R2 = 0.11 (Adjusted R2 = 0.071) Table B.18. Parameter estimates for GLM of (Great Egret density)-2 with fish density and water depth as covariates. Parameter SE 95% CI Intercept0.080.11(-0.14 0.30) depth0.050.29(-0.54 0.64) fish density0.0080.004(0.00 0.02)

PAGE 174

157 Table B.19. ANOVA table for GLM of (Snowy Egret density)1/2 with the interaction between prey density and habitat type. Source Type III Sum of Squaresdf Mean SquareFSig. Corrected Model 0.28a30.097.400.00 Intercept0.0210.021.780.19 habitat0.0010.000.210.65 fish density0.2010.2016.100.00 habitat fish density0.1810.1813.930.00 Error0.58460.01 Total1.0950 Corrected Total0.8649 a R2 = 0.33 (Adjusted R2 = 0.28) Table B.20. Parameter estimates for GLM of (Snowy Egret density)1/2 with the interaction between prey density and habitat type. Parameter SE 95% CI Intercept0.040.04(-0.03 0.11) habitat = unvegetated-0.020.04(-0.11 0.07) habitat = vegetated 0afish density0.000.00(0.00 0.01) habitat = unvegetated fish density0.020.00(0.01 0.02) habitat = vegetated fish density 0a a This parameter is set to zero because it is redundant.

PAGE 175

158 Table B.21. ANOVA table for GLM of (Tricolored Heron density)1/2 with the interaction between prey density and habitat type. Source Type III Sum of Squaresdf Mean SquareFSig. Corrected Model 0.24a30.084.110.01 Intercept0.0210.021.020.32 habitat0.0010.000.040.84 fish density0.1510.157.940.01 habitat fish density0.1510.157.930.01 Error0.88460.02 Total1.3050 Corrected Total1.1249 a R2 = 0.21 (Adjusted R2 = 0.16) Table B.22. Parameter estimates for GLM of (Tricolored Heron density)1/2 with the interaction between prey density and habitat type. Parameter SE 95% CI Intercept0.030.04(-0.06 0.12) habitat = unvegetated-0.010.05(-0.12 0.10) habitat = vegetated 0afish density0.000.00(-0.01 0.01) habitat = unvegetated fish density0.020.01(0.00 0.03) habitat = vegetated fish density 0a a This parameter is set to zero because it is redundant.

PAGE 176

159 Table B.23. ANOVA table for GLM of (White Ibis density)1/2 with the interaction between prey density and habitat type. Source Type III Sum of Squaresdf Mean SquareFSig. Corrected Model 3.24a31.087.200.00 Intercept0.2810.281.890.18 habitat0.6910.694.620.04 fish density1.8511.8512.320.00 habitat fish density2.7112.7118.080.00 Error6.90460.15 Total11.7150 Corrected Total10.1349 a R2 = 0.32 (Adjusted R2 = 0.28) Table B.24. Parameter estimates for GLM of (White Ibis density)1/2 with the interaction between prey density and habitat type. Parameter SE 95% CI Intercept0.270.12(0.02 0.52) habitat = unvegetated-0.320.15(-0.63 -0.02) habitat = vegetated 0afish density-0.010.01(-0.02 0.01) habitat = unvegetated fish density0.060.02(0.03 0.09) habitat = vegetated fish density 0a a This parameter is set to zero because it is redundant.

PAGE 177

160 LIST OF REFERENCES Alldredge, J. R., and J. T. Ratti. 1986. Comparison of some statistical techniques for analysis of resource selection. Jour nal of Wildlife Ma nagement 50:157-165. Anderson, D. R., and K. P. Burnham. 2002. A voiding pitfalls when using informationtheoretic methods. Journal of Wildlife Management 66:912-918. Anonymous. 1999. Digital Orthographic Quarte r-Quads (1999). Florida Department of Environmental Protection. http://data.labins.org/2003/index.cfm Accessed August 2004. Anonymous. 2002. SJRWMD Land Use and La nd Cover (2000). St. Johns River Water Management District. http://www.sjrwmd.com/programs/data.html Accessed October 2004. Bancroft, G. T., D. E. Gawlik, and K. Rutchey. 2002. Distribution of wading birds relative to vegetation and water depths in th e northern Everglades of Florida, USA. Waterbirds 25:265-277. _____, A. M. Strong, R. J. Sawick i, W. Hoffman, and S. D. Jewell. 1994. Relationships among wading bird foraging patterns, co lony locations, and hydrology of the Everglades. Pages 615-687 in S. M. Davis, andJ. C. Ogden, editors. Everglades: the ecosystem and its restoration. St. Lucie Press, Delray Beach, Florida, USA. Beauchamp, G., M. Belisle, and L. A. Giraldeau. 1997. Influence of conspecific attraction on the spatial dist ribution of learning foragers in a patchy habitat. Journal of Animal Ecology 66:671-682. Bildstein, K. L. 1997. Wading -bird science: a gui de for the twenty-first century. Colonial Waterbirds 20:138-142. Bouton, S. N., P. C. Frederick, M. G. Spalding, and H. McGill. 1999. Effects of chronic, low concentrations of dietary methylmerc ury on the behavior of juvenile great egrets. Environmental Toxicology and Chemistry 18:1934-1939. Breininger, D. R., and R. B. Smith. 1990. Wa terbird use of coastal impoundments and management implications in eastcentral Florida. Wetlands 10:223-241. Brockmeyer, R. E. 2004. Optimizing IRL wetla nd habitat restorat ion and management: The IRL Wetlands Initiative. St. Johns Rive r Water Management District, Palatka, Florida, USA.

PAGE 178

161 _____, J. R. Rey, R. W. Virnstein, R. G. Gilm ore, and L. Earnest. 1997. Rehabilitation of impounded estuarine wetlands by hydrologic reconnection to the Indian River Lagoon, Florida (USA). Wetlands Ecology and Management 4:93-109. Burnham, K. P., and D. R. Anderson. 2002. M odel selection and multi-model inference: a practical information-theoretic approac h. Second edition. Springer, New York. Butler, R. W. 1995. The patient predator: fo raging and population ecology of the Great Blue Heron Ardea herodias in British Columbia. Occasional. Paper No. 86, Canadian Wildlife Service. _____, and A. Poole. 1992. Great Blue Heron. in The birds of North America P. Stettenheim, and F. Gill editors. Philadel phia: The Academy of Natural Sciences; Washington, DC: The Ameri can Ornithologists Union. _____, and R. G. Vennesland. 2000. Integrating c limate change and predation risk with wading bird conservation rese arch in North America. Colonial Waterbirds 23:535540. Byers, C. R., R. K. Steinhorst, and P. R. Krausman. 1984. Clarification of a technique for analysis of utilization-av ailability data. Journal of Wildlife Management 43:10501053. Caldwell, G. S. 1980. Underlying benefits of foraging aggression in egrets. Ecology 61:996-997. _____. 1981. Attraction to tropical mixed-specie s heron flocks: proximate mechanism and consequences. Behavioral Ecology and Sociobiology 8:99-103. _____. 1986. Predation as a selective force on fo raging herons: effects of plumage color and flocking. Auk 103:494-505. Caraco, T. 1979. Time budgeting and group size: a theory. Ecology 60:611-617. Cezilly, F., V. Boy, and H. Hafner. 1 990. Group foraging in Little Egrets ( Egretta grazetta ): from field evidence to experimental investigation. Behavioural Processes 21:69-80. Charnov, E. L. 1976. Optimal foraging, the marginal value theorem. Theoretical Population Biology 9:129-136. Chaves-Ramirez, F., and R. D. Slack. 1995. Diff erential use of coastal marsh habitats by nonbreeding wading birds. Col onial Waterbirds 18:166-171. Cherry, S. 1996. A comparison of confidence inte rval methods for hab itat use-avai lability studies. Journal of Wildli fe Management 60:653-658.

PAGE 179

162 Chick, J. H., F. Jordan, J. P. Smith, and C. C. McIvor. 1992. A comparison of 4 enclosure traps and methods used to sample fishes in aquatic macrophytes. Journal of Freshwater Ecology 7:353-361. Conover, W. J. 1980. Practical nonparametric statistics. Second edition. John Wiley & Sons, New York. Crozier, G. E., and D. E. Gawlik. 2003. The use of decoys as a research tool for attracting birds. Journal of Field Ornithology 74:53-58. Custer, C. M., and J. Galli. 2002. Feeding ha bitat selection by Great Blue Herons and Great Egrets nesting in east centr al Minnesota. Waterbirds 25:115-124. _____, S. A. Suarez, and D. A. Olsen. 2004. Feed ing habitat characteristics of the Great Blue Heron and Great Egret nesting al ong the upper Mississippi River, 1995-1998. Waterbirds 27:454-468. Custer, T. W., and R. G. Osborn. 1978. Feed ing habitat use by co lonially-breeding herons, egrets, and Ibises in North Carolina. Auk 95:733-743. Cuthill, I. C., A. T. D. Bennett, J. C. Partridge, and E. J. Maier. 1999. Plumage reflectance and the objective assessment of avian sexual dichromatism. American Naturalist 153:183-200. Dall, S. R. X., L. A. Giraldeau, O. Olsson, J. M. McNamara, and D. W. Stephens. 2005. Information and its use by animals in evolutionary ecology. Trends in Ecology and Evolution 20:187-193. DeFreese, D. E. 1991. Threats to biological di versity in marine ecosystems in Florida. Coastal Management 19:73-101. Draulans, D. 1987. The effect of prey density on foraging behavior and success of adult and first-year Grey Herons ( Ardea cinerea ). Journal of Animal Ecology 56:479493. Elger, M. A. 1989. Predator vigilance and gr oup size in mammals and birds: a critical review of the empirical eviden ce. Biological Review 64:13-33. Erwin, R. M. 1983a. Feeding behavior and eco logy of colonial waterbirds: a synthesis and concluding comments. Co lonial Waterbirds 6:73-82. _____. 1983b. Feeding habitats of nesting wading bi rds: spatial use and social influences. Auk 100:96-970. _____. 1984. Feeding flights of nesting wading bi rds at a Virginia colony. Colonial Waterbirds 7:74-79.

PAGE 180

163 _____. 1996. Dependence of waterbirds and shoreb irds on shallow-water habitats in the mid-Atlantic coastal region: an ecological profile and management recommendations. Estuaries 19:213-219. _____, H. Hafner, and P. J. Dugan. 1985. Differen ces in the feeding be havior of Little Egrets ( Egretta grazetta ) in two habitats in the Camargue, France. Wilson Bulletin 97:534-538. _____, G. M. Sanders, and D. J. Prosser. 2005. Changes in lagoonal marsh morphology at selected Northeastern Atlantic coast site s of significance to migratory waterbirds. Wetlands 24:891-903. Fasola, M., and F. Barbieri. 1978. Factors a ffecting the distribution of heronries in northern Italy. Ibis 120:537-540. _____, L. Canova, and N. Saino. 1996. Rice fields support a large po rtion of herons breeding in the Mediterranean region. Col onial Waterbirds 19 (S pecial Publication 1):129-134. Fleury, B. E., and T. W. Sherry. 1995. Longterm population trends of colonial wading birds in the southeastern United States: the impact of crayfish aquaculture on Louisiana populations. Auk 112:613-632. Frederick, P. C. 1996. White Ibis. Pages 466-474 in J. A. Rodgers, Jr., H. W. Kale, II, andH. T. Smith, editors. Rare and endangered biota of Flor ida. Volume V. Birds. University Press of Florida, Gainesville, USA. _____. 2002. Chapter 19: Wading birds in the marine environment. Pages 617-656 in E. A. Schreiber, andJ. Burger, editors. Biology of marine birds. CRC Press, Boca Raton, Florida, USA. _____, and K. L. Bildstein. 1992. Foraging ecology of seven species of neotropical ibises (Threskiornithidae) during the dry seas on in the LLanos of Venezuela. Wilson Bulletin 104:1-21. _____, K. L. Bildstein, B. E. Fleury, and J. C. Ogden. 1996. Conservation of large, nomadic populations of White Ibis ( Eudocimus albus ) in the United States. Conservation Biology 10:203-216. _____, and M. W. Collopy. 1989. Nesting success of five ciconiiform species in relation to water conditions in the Fl orida Everglades. Auk 106:625-634. _____, and W. F. Loftus. 1993. Responses of ma rsh fishes and breeding wading birds to low temperatures: a possible behavioral li nk between predator and prey. Estuaries 16:216-222.

PAGE 181

164 _____, and J. C. Ogden. 1997. Philopatry a nd nomadism: contrasting long-term movement behavior and popul ation dynamics of White Ibis and Wood Storks. Colonial Waterbirds 20:316-323. _____, and _____. 2001. Pulsed breeding of long-legged wading birds and the importance of infrequent severe drought conditions in the Florida Everglades. Wetlands 21:484-491. _____, and _____. 2003. Monitoring wetland ecosystems using avian populations: seventy years of surveys in the Everglades. Pages 321-350 in D. E. Busch, andJ. C. trexler, editors. Monitori ng Ecosystems: Interdisciplinary Approaches for Evaluating Ecoregional Initiati ves. Islan Press,Washington. _____, and M. G. Spalding. 1994. Factors a ffecting reproductiv e success of wading birds (Ciconiiformes) in the Everglades ecosystem. Pages 659-691 in S. M. Davis, andJ. C. Ogden, editors. Everglades: the ecosystem and its restoration. St. Lucie Press, Delray Beach, Florida, USA. Fretwell, S. D., and H. J. Lucas, Jr. 1970. On territorial behavior and other factors influencing habitat distribution in birds. Acta Biotheoretica 19:16-36. Frohring, P. C., D. P. Voor hees, and J. A. Kushlan. 1988. History of wading bird populations in the Florida Everglades: a less on in the use of historical information. Colonial Waterbirds 11:328-335. Garshelis, D. L. 2000. Delusions in habitat evaluation: measuring use, selection, and importance. Pages 111-164 in L. Boitani, and T. K. Fuller, editors. Research techniques in animal ecology: contro versies and consequences. Columbia University Press, New York. Gawlik, D. E. 2002. The effects of prey ava ilability on the numeri cal response of wading birds. Ecological Monographs 72:329-346. Gibbs, J. P. 1991. Spatial relationships between nesting colonies a nd foraging areas of Great Blue Herons. Auk 108:764-770. Gibbs, J. P., and L. K. Kinkel. 1997. Determinan ts of the size and location of Great Blue Heron colonies. Colonial Waterbirds 20:1-7. Gibbs, J. P., S. Woodward, M. L. Hunter, and A. E. Hutchinson. 1987. Determinants of Great Blue Heron colony distribution in coas tal Maine. Auk 104:38-47. Gilmore, R. G. 1995. Environmental and bioge ographic factors influencing icthyofaunal diversity: Indian River Lagoon. Bulle tin of Marine Science 57:153-170. _____. 1998. Wetland Ecosystem Management: I ndian River Lagoon, Florida, USA. National Estuarine Program and St. Johns River Water Management District. Report Contract No. 98W230.

PAGE 182

165 _____, D. W. Cooke, and C. J. Donohoe. 1982. A comparison of the fish populations and the habitat in open and closed salt mars h impoundments in east-central Florida. Northeast Gulf Science 5:25-37. Gonzalez, J. A. 1997. Seasonal variation in th e foraging ecology of the Wood Stork in the southern LLanos of Venezuela. Condor 99:671-680. Grafen, A., and R. Hails. 2002. Modern statistics for the life sciences. Oxford University Press, Oxford. Hafner, H., V. Boy, and G. Gory. 1982. Feedi ng methods, flock size, and feeding success in the Little Egret Egretta garzetta and the Squacco Heron Ardeola ralloides in Camargue, southern France. Ardea 70:45-54. _____, P. J. Dugan, and V. Boy. 1986. Use of ar tificial and natural we tlands as feeding sites by Little Egrets ( Egretta garzetta L.) in the Camargue Southern France. Colonial Waterbirds 9:149-154. _____, P. J. Dugan, M. Kersten, O. Pineau, and J. P. Wallace. 1993. Flock feeding and food intake in Little Egrets Egretta grazetta and their effects on food provisioning and reproductive success. Ibis 135:25-32. _____, Y. Kayser, V. Boy, M. Fasola, A. Julliard, R. Pradel, and F. Cezilly. 1998. Local survival, natal dispersal, and recruitment in Little Egrets Egretta grazetta Journal of Avian Biology 29:216-227. Hamilton, W. D. 1971. Geometry forthe self ish herd. Journal of Theoretical Biology 31:295-311. Hancock, J., and J. A. Kushlan. 1984. Th e herons handbook. Croom Helm Ltd., Kent, United Kingdom. Harrington, R. W., Jr., and E. S. Harrington. 1961. Food selection among fishes invading a high subtropical salt marsh: from on set of flooding through the progress of a mosquito brood. Bulletin of Marine Science 42:646-666. _____, and _____. 1982. Effects on fishes and their forage organisms of impounding a Florida salt marsh to prevent breeding by salt marsh mosquitoes. Bulletin of Marine Science 32:523-531. Hino, T. 1998. Mutualistic and commensal orga nization of avian mixed-species foraging flocks in a forest of western Madaga scar. Journal of Avian Biology 29:17-24. Hoffman, W., G. T. Bancroft, and R. J. Sa wicki. 1994. Foraging habitat of wading birds in the water conservation areas of the Everglades. Pages 585-614 in S. M. Davis, andJ. C. Ogden, editors. Everglades: the ecosystem and its restoration. St. Lucie Press, Delray Beach, Florida, USA.

PAGE 183

166 Hom, C. W. 1983. Foraging ecology of Herons in a southern San Francisco Bay salt marsh. Colonial Waterbirds 6:37-44. Jakubas, D. 2004. The response of the Grey Heron to a rapid increase in round goby. Colonial Waterbirds 27:304-307. James, F. C., and C. E. McCulloch. 1995. The strength of inferences about causes of trends in populations. Pages 40-51 in T. E. Martin, and D. M. Finch, editors. Ecology and management of neotropical migr atory birds: a synthe sis and review of critical issues. Oxford University Press, New York. Johnson, D. H. 1980. The comparison of usage and availability measurements for evaluating resource preference. Ecology 61:65-71. Johnson, J. B., and K. S. Omland. 2004. Mode l selection in ecology and evolution. Trends in Ecology and Evolution 19:101-108. Jordan, F., S. Coyne, and J. C. Trexler. 1997. Sampling fishes in vegetated habitats: effects of habitat structure on sa mpling characteristics of the 1-m2 throw trap. Transactions of the American Fisheries Society. 126:1012-1020. Kahl, M. P. J. 1964. Food ecology of the Wood Stork ( Mycteria americana ) in Florida. Ecological Monographs 34:97-117. Kautz, R. S. 1993. Trends in Florida wildlif e habitat 1936-1987. Flor ida Scientist 56:724. Kersten, M., R. H. Britton, P. J. Dugan, and H. Hafner. 1991. Flock feeding and food intake in Little Egrets: the effects of prey distribution and behavior. Journal of Animal Ecology 60:241-252. Klein, M. L. 1993. Waterbird behavioral re sponse to human disturbances. Wildlife Society Bulletin 21:31-39. _____, S. R. Humphrey, and H. F. Percival. 1995. Effects of ecotourism on distribution of waterbirds in a wildlife re fuge. Conservation Biology 9:1454-1465. Krause, J., and G. D. Ruxton. 2002. Living in groups. Oxford University Press, Oxford. Krebs, J. R. 1974. Colonial nesting and social feeding as strategies for exploiting food resources in the Great Blue Heron ( Ardea herodias ). Behaviour 51:99-134. Krebs, J. R., and A. J. Inman. 1992. Learni ng and foraging indi viduals, groups, and populations. American Na turalist 140:S63-S84. Krebs, J. R., and A. Kacelni k. 1991. Decision-making. Pages 105-136 in J. R. Krebs, and N. B. Davies, editors. Behavioural ecol ogy: an evolutionary approach. Blackwell Scientific Publications, Oxford.

PAGE 184

167 Kushlan, J. A. 1976a. Feeding behavior of North American herons. Auk 93:86-94. _____. 1976b. Wading bird predation in a seas onally fluctuating pond. Auk 93:464-476. _____. 1977. The significance of plumage colour in the formation of feeding aggregations of ciconiiforms. Ibis 119:361-364. _____. 1978a. Commensalism in the Litt le Blue Heron. Auk 95:677-681. _____. 1978b. Feeding Ecology of Wading Birds. in A. I. Sprunt, J. C. Ogden, and S. Winckler, editors. Wading birds. Nationa l Audubon Society, Research Report No. 7. _____. 1978c. Nonrigorous foraging by r obbing egrets. Ecology 59:649-653. _____. 1981a. Resource use strategies of wa ding birds. Wilson Bulletin 93:145-163. _____. 1981b. Sampling characteristics of enclosur e fish traps. Transactions of the American Fisheries Society 110:557-562. _____. 1997. The conservation of wading bird s. Colonial Waterbirds 20:129-137. Larsen, J. K. 1996. Widgeon Anas penelope offsetting dependence on water by feeding in mixed-species flocks: a natura l experiment. Ibis 138:555-557. Larson, V. L. 1992. A method for assessing the conservation value of natural communities at a local scale. Thesis, Fl orida Institute of Technology, Melbourne, Florida, USA. Loftus, W. F., and A. Eklund. 1994. Long-term dynamics of an Everglades small-fish assemblage. Pages 461-483 in S. M. Davis, and J. C. Ogden, editors. Everglades: the ecosystem and its restoration. St. Luci e Press, Delray Beach, Florida, USA. Lombardini, K., R. E. Bennetts, and C. Tourenq. 2001. Foraging success and foraging habitat use by Cattle Egrets and Little Egrets in the Camargue, France. Condor 103:38-44. MacArthur, R. H., and E. Pianka. 1966. On the optimal use of a patchy environment. American Naturalist 100:603-609. MacCarone, A. D., and K. C. Parsons. 1988. Di fferences in flight patterns among nesting ibises and egrets. Colonial Waterbirds 11:67-71. _____, and _____. 1994. Factors affecting the use of a freshwater and an estuarine foraging site by egrets and ibises during the breeding season in New York City. Colonial Waterbirds 17:60-68. Maddock, M., and G. S. Baxter. 1991. Breeding success of egrets rela ted to rainfall: a six-year Australian study. Co lonial Waterbirds 14:133-139.

PAGE 185

168 Manly, B. F. J., L. L. McDonald, D. L. Thomas, T. L. McDonald, and W. P. Erickson. 2002. Resource selection by animals. Kluwer Academic Publishers, Dorderecht. Master, T. L. 1992. Composition, structure, and dynamics of mixed-species foraging aggregations in a southern New Jersey salt marsh. Colonial Waterbirds 15:66-74. _____, M. Frankel, and M. Russell. 1993. Benef its of foraging in mixed-species wader aggregations in a southern New Jersey salt marsh. Colonial Waterbirds 16:149-157. _____, J. K. Leiser, K. A. Bennett, J. K. Br etsch, and H. J. Wolf e. 2005. Patch selection by Snowy Egrets. Waterbirds 28:220-224. McClean, S. A., M. A. Rumble, R. M. King, and W. L. Baker. 1998. Evaluation of resource selection methods with different definitions of availability. Journal of Wildlife Management 62:793-801. McIvor, C. C., and W. E. Odum. 1988. Food, pr edation risk, and microhabitat selection in a marsh fish assemblage. Ecology 69:1341-1351. Miranda, L., and J. A. Collazo. 1997. Food habits of 4 species of wading birds (Ardeidae) in a tropical mangrove swamp. Co lonial Waterbirds 20:413-418. Montague, C. L., A. V. Zale, and H. F. Percival. 1987. Ecological effects of coastal marsh impoundments: a review. E nvironmental Management 11:743-756. Morse, D. H. 1970. Ecological aspects of some mixed-species flocks of birds. Ecological Monographs 40:119-168. Neill, S. R. S. J., and J. M. Cullen. 1974. Experiments on whether schooling by their prey affects the hunting behavi our of cephalopods and fish predators. Journal of Zoology (London) 172:549-569. Neu, C. W., C. R. Byers, and J. M. Pee k. 1974. A technique for analysis of utilizationavailability data. Journal of Wildlife Management 38:541-545. Ogden, J. C. 1994. A comparison of wading bird nesting colony dynamics (1931-1946 and 1974-1989) as an indication of ecosystem conditions in the southern Everglades. Pages 533-570 in S. M. Davis, and J. C. Ogden, editors. Everglades: the ecosystem and its restoration. St. Luci e Press, Delray Beach, Florida, USA. _____. 1996a. Snowy Egret. Pages 420-431 in J. A. Rodgers, Jr., H. W. Kale, II, and H. T. Smith, editors. Rare and endangered biot a of Florida. Volume V. birds. University Press of Florida, Gainesville, USA. _____. 1996b. Tricolored Heron. Pages 432-441 in J. A. Rodgers, Jr., H. W. Kale, II, and H. T. Smith, editors. Rare and Endangered Biota of Florida. Volume V. birds. University Press of Florida, Gainesville, USA.

PAGE 186

169 _____, J. A. Kushlan, and J. T. Tilmant. 1976. Prey selectivity by the Wood Stork. Condor 78:324-330. _____, I. A. Sprunt, J. C. Ogden, and S. Winckler. 1978. Recent population trends of colonial wading birds on the Atlantic and Gulf Coastal Plains. Pages 137-153 in A. I. Sprunt, J. C. Ogden, and S. Winckl er, editors. Wading bi rds. National Audubon Society, Research Report No. 7. Palmer, R. S. 1962. Handbook of North Ameri can birds. Volume 1.Yale University Press, New Haven. Petit, D. R., and K. L. Bildstein. 1987. Eff ect of group size and lo cation within the group on the foraging behavior of White Ibis. Condor 89:602-609. Powell, G. V. N. 1983. Food availability and reproduction by Great White Herons, Ardea herodias : a food addition study. Colo nial Waterbirds 6:139-147. _____. 1985. Sociobiology and the adaptive signifi cance of interspecific foraging flocks in the Neotropics. Ornithological Monographs 36:713-732. _____. 1987. Habitat use by wading birds in a su btropical estuary: implications of hydrography. Auk 104:740-749. Provost, M. W. 1968. Analysis by marsh and by species. Appendix to : Study of wildlife usage of salt marsh on east coast of Florida before and after impoundment for mosquito and sandfly control by C. H. Tr ost. U. S. Fish and Wildlife Service, Bureau of Sport Fisheries and Wild life. Report Contact No. 14-16-0008-623. _____. 1969. Ecological control of salt marsh mosqu itoes with side benefits to birds. in Proceedings of Tall Timbers Conference on Ecological Animal Control by Habitat Management Rey, J. R., R. A. Crossman, and T. R. Kain. 1990. Vegetation dynamics in impounded marshes along the Indian River Lagoon, Florida, USA. Environmental Management 14:397-409. Rivers, J. W. 2000. Northern Harrier pred ation of White-faced Ibis. Wilson Bulletin 112:416-417. Robertson, W. B., and G. E. Woolfenden 1992. Fl orida bird species: an annotated list. Special Publication No. 6. Florida Orn ithological Society, Gainesville, USA. Rodgers, J. A., Jr. 1996. Litt le Blue Heron. Pages 413-419 in J. A. Rodgers, Jr., H. W. Kale, II, and H. T. Smith, editors. Rare and endangered biota of Florida, V. Birds. University Press of Florida, Gainesville, USA.

PAGE 187

170 Rodrigues, M., C. G. Machado, S. M. R. Al vares, and M. Galetti. 1994. Association of the Black-goggled Tanager ( Trichothraupis melanops ) with flushers. Biotropica 26:472-475. Rosenberg, D. K., and K. S. McKelvey. 1999. Es timation of habitat se lection for centralplace foraging animals. Journal of Wildlife Management 63:1028-1038. Rosenzweig, M. L. 1981. A theory of habitat selection. Ecology 62:327-335. Rozas, L. P., and W. E. Odum. 1988. Occupa tion of submerged aquatic vegetation by fishes: testing the roles of f ood and refuge. Oecologia 77:101-106. Runde, D. E. 1991. Trends in wading bird nesting populations in Florida 1976-1978 and 1986-1989. Nongame Wildlife Section, Divisi on of Wildlife, Florida Game and Fresh Water Fish Commission, Tallahassee, USA. _____. 1996. Great Egret. Pages 404-412 in J. A. Rodgers, Jr., H. W. Kale, II, and H. T. Smith, editors. Rare and endangered biota of Florida, V. Birds. University Press of Florida, Gainesville, USA. Russell, J. K. 1978. Effects of interspeci fic dominance among egrets commensally following Roseate Spoonbills. Auk 95:608-610. Schikorr, K. E., and H. M. Swain. 1995. Wa ding birds barometer of management strategies in the Indian River Lagoon. Bulletin of Marine Science 57:215-229. Schlaepfer, M. A., M. C. Runge, and P. W. Sherman. 2002. Ecological and evolutionary traps. Trends in Ecol ogy & Evolution 17:474-480. Schmalzer, P. A. 1995. Biodivers ity of saline and brackish ma rshes of the Indian River Lagoon: historic and current patterns. Bulletin of Marine Science 57:37-48. Schooly, J. K. 1980. The structure and functi on of warm temperate estuarine fish communities. Thesis, University of Florida, Gainesville, USA. Sewell, C. E., N. D. Joiner, M. S. Robson, and M. J. Weise. 1995. Wading bird colony protection in the Indian River Lagoon. Bulletin of Marine Science 57:237-241. _____, S. Scott, and H. M. Swain. 1996. Development of management strategies for mosquito impoundments in the Indian River lagoon to enhance wading bird populations. Final Reprt submitted to The Indian River Lagoon National Estuary Program, Melbourne, Florida, and St. Luci e County Mosquito Control District, Ft. Pierce, Florida, USA. Simpson, K., J. N. M. Smith, and J. P. Ke lsall. 1987. Correlates and consequences of coloniality in Great Blue Herons. Canadian Journal of Zoology 65:572-577.

PAGE 188

171 Smith, J. P. 1995a. Foraging flights and habitat use of nesting wading birds (Ciconiiformes) at Lake Okeechobee, Fl orida. Colonial Wa terbirds 18:139-158. _____. 1995b. Foraging sociability of nesting wa ding birds (Ciconiiformes) at Lake Okeechobee, Florida. Wilson Bulletin 107:437-451. _____, and M. W. Collopy. 1995. Colony turnover nest success a nd productivity, and causes of nest failure among wading bird s (Ciconiiformes) at Lake Okeechobee, Florida (1989-1992). Arch. Hydrobiol. Sp ec. Issues Advanc. Limnol. 45:287-316. _____, J. R. Richardson, and M. W. Collopy. 1995. Foraging hab itat selection among wading birds (Ciconiiformes) at Lake Okeechobee, Florida, in relation to hydrology and vegetative cover. Arch. Hydr obiol. Spec. Issues Advanc. Limnol. 45:247-285. Smith, N. P. 1987. Introduction to the tides of Florida's Indian River Lagoon. I. water levels. Florida Scientist 50:49-61. _____. 1993. Tidal and wind-driven transport between Indian River and Mosquito Lagoon, Florida. Florida Scientist 56:235-246. Smith, R. B., and D. R. Breininger. 1995. Wading bird populations of the Kennedy Space Center. Bulletin of Marine Science 57:230-236. Snelson, F. F., Jr. 1983. Ichthyofauna of the northern part of the Indian River Lagoon System, Florida. Florida Scientist 46:187-206. Stevens, P. W. 2002. Test of salt marsh as a site of production and export of fish biomass with implications for impoundment mana gement and restoration. Dissertation, University of Florida, Gainesville, USA. Stolen, E. D. 2003. The effects of vehicle passa ge on foraging behavior of wading birds. Waterbirds 26:429-436. _____, D. R. Breininger, and P. C. Frederic k. 2005. Using waterbirds as indicators in estuarine systems: successe s and perils. Pages 409-422 in S. A. Bortone, editor. Estuarine indicators. CRC Press, Boca Raton, Florida, USA. _____, Smith, R. B., and D. R. Breininger. 2002. Analysis of wading bird use of impounded wetland habitat on Kennedy Space Center/Merritt Island National Wildlife Refuge 1987-1997. NASA T echnical Memorandum 211173, Kennedy Space Center, Florida, USA. Strong, A. M., G. T. Bancroft, and S. D. Jewell. 1997. Hydrological constraints on Tricolored Heron and Snowy Egre t resource use. Condor 99:894-905. Sullivan, K. A. 1984. The advantages of social foraging in Downy Woodpeckers. Animal Behavior 32:16-22.

PAGE 189

172 Surdick, J. A., Jr. 1998. Biotic and abiotic i ndicators of foraging site selection and foraging success of four ci coniiform species in the fr eshwater Everglades of Florida. Thesis, University of Florida, Gainesville, USA. Sutherland, W. J. 1996. From individual be haviour to population ecology. Oxford University Press, Oxford. Thomas, F., Y. Kayser, and H. Hafner. 1999. Nestling size rank in the little egret ( Egretta grazetta ) influences subsequent breeding su ccess of offspring. Behavioral Ecology and Sociobiology 45:466-470. Thompson, D. H. 1978. Feeding areas of Great Blue Herons and Great Egrets nesting within the floodplain of the upper Mississippi River. Pages 202-213 in A. I. Sprunt, J. C. Ogden, and S. Winckler, editors Wading birds. National Audubon Society, Research Report No. 7. Trexler, J. C., W. F. Loftus, and J. H. Chick. 2003. Setting and monitoring restoration goals in the absence of historical data: th e case of fishes in the Florida Everglades. Pages 351-376 in D. E. Busch, andJ. C. Trexler, editors. Monitoring ecosystems: interdisciplinary approaches for evaluati ng ecoregional initiatives. Island Press, Washington D. C., USA. _____, W. F. Loftus, F. Jordan, J. H. Chick, K. L. Kandl, T. C. McElroy, and O. L. Bass, Jr. 2002. Ecological scale and it s implications for freshwat er fishes in the Florida Everglades. Pages 153-181 in J. W. Porter, and K. G. Porter, editors. The Everglades, Florida Bay and coral reef s of the Florida Keys: an ecosystem sourcebook. CRC Press, Boca Raton, Florida, USA. _____, R. C. Tempe, and J. Travis. 1994. Size-Se lective Predation of Sailfin Mollies by 2 Species of Heron. Oikos 69:250-258. Trost, C. H. 1968. Study of wildlife usage of salt marsh on east coast of Florida before and after impoundment for mosquito and sandfly control. U. S. Fish and Wildlife Service, Bureau of Sport Fisheries an d Wildlife. Report Contact No. 14-16-0008623. Valone, T. J. 1989. Group foraging, public in formation, and patch estimation. Oikos 56:357-363. _____. 1991. Bayesian and prescient assessment: foraging with pre-harvest information. Animal Behaviour 41:569-577. _____, and L. A. Giraldeau. 1993. Patch estimation by group foragers: what information is used? Animal Behaviour 45:721-728. Van Horne, B. 1983. Density as a misleading indicator of habitat quality. Journal of Wildlife Management 47:893-901.

PAGE 190

173 Wiens, J. A. 1989. The ecology of bird communities. Cambridge University Press, Cambridge Wiggins, D. A. 1991. Foraging success and a ggression in solitary and group-feeding Great Egrets ( Casmerodius albus ). Colonial Waterbirds 14:176-179. Werner, E. E., J. F. Gilliam, D. J. Hall, and G. G. Mittelbach. 1983. An Experimental test of the effects of predation risk on habitat use in fish. Ecology 64:1540-1548. Werner, S. J., M. E. Tobin, and P. B. Fior anelli. 2001. Great Egret preference for catfish size classes. Wate rbirds 24:381-385. Wood, D. A. 1996. Floridas endangered specie s, threatened species and species of special concern. Florida Game and Fres hwater Fish Commission, Tallahassee, USA. Yaukey, P. H. 1995. Effects of food suppl ementation and predator simulation on nuthatches and parids within mixed-sp ecies flocks. Wilson Bulletin 107:542-547.

PAGE 191

174 BIOGRAPHICAL SKETCH Eric Douglas Stolen was born in Ann Ar bor, Michigan. Growing up he loved the outdoors and spent much time wandering in the woods near his home. From an early age Eric was interested in plants and animals a nd enjoyed learning to identify local species. Erics grandfather, Neal S. Tosi, was very en couraging in this regard. Likewise Erics mother, Sandra, and father, Douglas spent mu ch time in the outdoors with Eric when he was young. Erics parents were also very sc holarly and inquisitive, traits which they managed to pass on to both their offspring. Eric also learned to cherish the companionship of others through many hours spent playing with his younger sister, Brenda. For several years when Eric was in high school, the family moved overseas and Eric spent a year in Saudi Arabia and anot her in Switzerland. This time also allowed trips to other parts of the world including east Africa and Russia, and these experiences broadened Erics understanding of the worl d. Eric was a good scholar but often a distracted student, but managed to comple te his education, earn ing a diploma from Manistee High School in Manist ee, Michigan. Following a y ear at the University of Michigan (majoring in chemistry), Eric followed his family to Florida and completed an A.S. degree from Brevard Community College Following a few years at Florida State University studying molecular biology, Eric completed a B.S. in biology from the University of Central Florida. It was duri ng this time that Eric developed a passion for birds, mainly with the help of a few very friendly bird watchers in the local Audubon Society. Next Eric completed a M.S., also at the University of Central Florida, studying

PAGE 192

175 under the great ornithologist Walter K. Taylor Erics thesis was on the foraging and roosting behavior of the Black Vulture in ce ntral Florida. As a young graduate student, Eric met his soul mate, Megan, and the two we re eventually married. It was during the completion of Erics thesis that he took a full-time position working on the Life Science Support Contract at the Kennedy Space Cent er, working for Dynamac Corporation, a contractor to NASA. In 1998, Er ic decided to return to schoo l to work on a doctorate and was able to talk his employer into letti ng him do so while still working at the Space Center; this proved to be both a blessing and a curse. Eric enrolled the help of Dr. Jaime Collazo, a professor at North Carolina Stat e University who was conducting research on shorebird ecology at the Merrit t Island National Wildlife Ref uge (which spatially is an overlay on the Kennedy Space Center). Jaime was able talk Dr. Franklin Percival at the University of Florida into agreeing to chair Erics supervisory committee. During the completion of Erics studies at the University of Florida, Eric and Megan collectively achieved their greatest contribu tions to the future when Ethan Douglas Stolen was born in November of 2000 and then Erin Leslie Stolen was born in October of 2003.


Permanent Link: http://ufdc.ufl.edu/UFE0013802/00001

Material Information

Title: Habitat Selection and Foraging Success of Wading Birds in Impounded Wetlands in Florida
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0013802:00001

Permanent Link: http://ufdc.ufl.edu/UFE0013802/00001

Material Information

Title: Habitat Selection and Foraging Success of Wading Birds in Impounded Wetlands in Florida
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0013802:00001


This item has the following downloads:


Full Text







HABITAT SELECTION AND FORAGING SUCCESS OF WADING BIRDS
IN IMPOUNDED WETLANDS IN FLORIDA














By

ERIC DOUGLAS STOLEN


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


2006































Copyright 2006

by

Eric Douglas Stolen
































This document is dedicated to my late father W. Douglas Stolen who believed in me and
my late father-in-law Leslie Kinney who inspired me.










ACKNOWLEDGMENTS

First and foremost, I want to thank Dr. Jaime Collazo for serving as my co-advisor,

mentor, colleague, and friend. Jaime made the process of working on this dissertation

meaningful and helped me develop professionally in many, many ways. Likewise, as my

other co-advisor Dr. Franklin Percival was a constant source of help and encouragement.

He also taught me a great deal about the "philosophy" part of the degree. I also had the

pleasure of getting to know the other members of my committee during various stages of

my studies. Dr. Wiley Kitchens was an inspiring teacher and a solid source of advice,

especially in the initial stages of planning and implementing my field work. Dr. Clay

Montague was also a great teacher and truly made me think about how to solve problems

in science. Dr. Peter Frederick was an invaluable source of knowledge about wading bird

ecology and also a very practical guide to navigating the process of earning my degree.

His insightful comments on earlier drafts greatly improved this manuscript.

I was fortunate to have the help of many very able field technicians who spent

countless hours knee-deep in mud under the baking sun. Special thanks go to Francisco

Collazo, Alison Pevler, Elizabeth Labunsky, James Borgmeyer, Heather Eijanga and Jap

Eijanga. Others also helped with field work on many days, including Zoe Donaldson,

Ashley Below, Ryan Woods, Geoff Carter, Donna Oddy, and Ron Brockmeyer, to name

a few.

Geoff Carter gave me valuable editorial advice and spent hours reading various

drafts of this dissertation; I am extremely fortunate to share an office with such a great






naturalist and writer. Brean Duncan spent many hours helping with sticky GIS problems

and was always ready to help no matter how busy he was. Dave Breininger got me

started on wading birds and treated me as a colleague even before I was one. Dave also

has been very patient in allowing me to pursue my research interests at the expense of

publications documenting our monitoring program (a situation I will at last have the time

to remedy). My bosses Carlton Hall, Ross Hinkle and Doug Britt were very supportive

and encouraging throughout. My NASA managers, Kelly Gorman and Burt Summerfield

were also very supportive of my work on wading bird foraging ecology and have the

vision to see the value of knowledge about our natural resources.

I am also grateful for the support and encouragement by the staff at the Merritt

Island National Wildlife Refuge, especially Marc Epstein, Ron Height, and Ralph Loyd.

My mother, Sandra Morgan, and my mother-in-law, Regina Kinney, helped

innumerable times taking care of my children while I was occupied at the computer. If it

were not for that help I never could have finished writing. Finally, I want to deeply thank

my wife Megan and my children Ethan Stolen and Erin Stolen, whose support was

irreplaceable and whose patience was inexhaustible. I did it all for them.










TABLE OF CONTENTS

Page

ACKNOW LEDGM ENTS ............................................................................................... iv

LIST OF TABLES .......................................................................................................... viii

LIST OF FIGURES ........................................ .............................................................. xiii

ABSTRACT..................................................................................................................... xvi

CHAPTER

1 INTRODUCTION ....................................................................................................1

2 FORAGING HABITAT SELECTION BY NESTING GREAT EGRETS AND
SNOW Y EGRETS.................................. ............................................................11

M eth o d s .................................................... ........................................ .............................. 13
Study Site.......................................................... ..............................................13
Foraging Flight Observations.............................................................................. 14
Habitat Use Analysis ........................................................................................ 15
R esu lts........ ..................... ........ ........... ... ... ...................... .............................19
Foraging Flight Characteristics ..............................................................................19
Habitat Use Analysis ................................................... ..............................21
D iscu ssion ..................................... .............................................. ..........................29

3 THE DISTRIBUTION OF WADING BIRD PREY IN IMPOUNDED
WETLANDS IN THE NORTHERN INDIAN RIVER LAGOON ESTUARY........33

Introduction.................................................................................................................33
M eth o d s ............................................. ............. ..... .......................................................3 5
Study Site........................................................................................................35
Prey Sampling ......................................................... ...................................38
Analysis ........................................................................... ................. ..... 40
R e su lts ...................... ......... ...... ............ ... .......... .. ......... ...... ...... ................. ...........4 3
Fixed-Station Sampling .................................................................................... 43
Random-Site Sampling..................................................... ............................43
Discussion................... ............ ............ ................. .......................... 64





vi






4 EFFECTS OF HABITAT STRUCTURE AND PREY DISTRIBUTION ON
WADING BIRD FORAGING HABITAT USE .......................................................68

Introduction ......................................................................... .................... .................... 68
M methods ........................................ ........ ..................... ............. .......... ....... .......72
R esu lts.........................................................................................................................83
D iscu ssion ..................................... ...................... ......... .......... ........................... 103
Foraging H habitat Selection...............................................................................103
Mixed-Species Foraging Aggregations............................................................ 108

5 BENEFITS TO INDIVIDUALS IN PISCIVOROUS WADING BIRD MIXED-
SPECIES FORAGING AGGREGATIONS............................................................ 110

Introduction.............................................................................. ............................... 110
Methods .........................................................................................................................113
Study S ite....................................... ........................................................... 113
Foraging Behavior Observations.......................................................................114
R esults....................................................................................................................... 12 1
Prey Density at Used and Paired Random Sites................................................121
Behavior-Prey Sampling ........................ ....................................................121
D iscussion............................................ .............................. .. ....... ................. 133

6 CONSERVATION IMPLICATIONS AND MANAGEMENT
RECO M M EN D A TION S.........................................................................................138

APPENDIX

A WATER LEVELS WITHIN IMPOUNDMENTS AND INDIAN RIVER
LAGOON ESTUARY MEASURED DURING WETLANDS INITIATIVE .........145

B MODEL SELECTION RESULTS FOR INDIVIDUAL WADING BIRD
SPE C IE S ............................................................ ................................................... 148

LIST O F R EFER EN C ES ............................................................................................160

BIO G R A PH IC A L SK ETCH ......................... ......................... ....................................174










LIST OF TABLES


Table page

2.1. Summary of the sizes of foraging aggregations joined by Great Egrets and Snowy
Egrets follow ed from three colonies ................................... ..................................23

2.2. Summary of flight distance, duration and speed for Great Egrets and Snowy
Egrets followed from three colonies .........................................................................23

2.3. Areas of Great Egret and Snowy Egret foraging habitats within three flight-radius
distances from three nesting colonies...................................................... ...............24

2.4. Comparison of the proportion of bird use versus the proportion of habitat
availability, for three types of foraging habitat by nesting Snowy Egrets ..............25

2.5. Comparison of the proportion of bird use versus the proportion of habitat
availability, for three types of foraging habitat by nesting Great Egrets. ................26

3.1. Number of points sampled during random-station fish sampling in unvegetated
(unveg.) and vegetated (veg.) habitat during 5 seasons in 7 impoundments. ..........47

3.2. Occurrence offish by species in throw-trap samples in 6 fixed-station sampling
im poundm ents on KSC/M INW R. ............................................................ ..............47

3.3. Fixed-station fish sampling model selection results for GLM analysis with
ln[l+fish density] as response variable and season................................................48

3.4. ANOVA table for GLM with fixed-station ln(l+fish density) as response variable
and explanatory variables season and location. ...............................................48

3.5. Parameter estimates for GLM with fixed-station ln(l+fish density) as response
variable and explanatory variables season and location.........................................49

3.6. Occurrence of non-fish nekton in random-site throw-trap samples in 7
impoundments on KSC/M INW R ............................................................................. 49

3.7. Occurrence offish by species in throw-trap samples in wetland habitat with and
without emergent vegetation in 7 random-site impoundments..............................50

3.8. Density offish (individuals/m2) in wetland habitat without emergent vegetation in
7 random-site impoundments on KSC/MINWR ...................................................51






3.9. Density offish (individuals/m2) in wetland habitat with emergent vegetation in 7
random-site impoundments on KSC/MINWR......................................................52

3.10. Biomass (g/m2) of fish in wetland habitat without emergent vegetation in 7
random-site impoundments on KSC/MINWR.......................................................53

3.11. Biomass (g/m2) of fish in wetland habitat with emergent vegetation in 7 random-
site impoundments on KSC/MINWR. .................................................................54

3.12. Mean length and mean biomass of individual fish captured within random-site
throw-trap samples by species.............................................................................55

3.13. Mean length of individual fish in 2 habitat types within random-site throw-trap
sam ples, listed by species....................................................................................... 55

3.14. Random-site model selection results for GLM analysis with ln[fish density+l] as
response variable and habitat type. .......................................................................56

3.15. ANOVA table for GLM with random-site ln(fish density+1) as the explanatory
variable and habitat* season and habitat* impoundment interactions included........56

3.16. Parameter estimates for GLM with random-site In(fish density+1) as the
explanatory variable and habitat*season and habitat*impoundment interactions...57

3.17. ANOVA table for GLM with random-site ln(fish density+1) as the explanatory
variable and habitat* season, habitat* impoundment and season* impoundment ......58

3.18. Parameter estimates for GLM with random-site ln(fish density+1) as the
explanatory variable and habitat*season, habitat*impoundment and....................59

4.1. Models set for analysis of factors that effect wading bird density in foraging
habitat on KSC/MINWR. .......................... ................................................ 78

4.2 Occurrence of wading birds during ground surveys of 9 impoundments on the
KSC/M INW R, January July 2001. ..................................................................... 86

4.3. Water depth at locations where foraging wading birds were observed in
impoundments during ground surveys of impoundments on KSC/MINWR...........87

4.4. All species of wading birds observed showed selection for edge habitat over both
vegetated and unvegetated habitats........................................ ......................................88

4.5. Information-theoretic model selection results for GLM analysis with (wading bird
density)1/2 as response variable, fish density (P), water depth (D).........................89

4.6. ANOVA table for GLM with (wading bird density)1/2 as response variable and
including the fish density*habitat interaction. .......................................................90






4.7. Parameter estimates for GLM with (wading bird density)1/2 as response variable
and including the fish density*habitat interaction.................................................90

5.1. Mean fish density and mean total biomass (with 95% confidence intervals) at sites
used by foraging wading birds and paired random sites ......................................124

5.2. Mean fish density and mean total biomass (with 95% confidence intervals) at sites
used by foraging individuals and groups of wading birds ...................................124

5.3. Mean length and mean mass (with 95% confidence intervals) of fish in samples at
sites used by wading birds and paired-unused sites ...............................................124

5.4. Correlations between mean length and mean mass offish in samples at sites used
by wading birds. The sample size was 120 for all comparisons .........................124

5.5. Information-theoretic model selection results for GLM analysis of Great Egret
(captures / min)1/2 as response variable, prey density (PD) as covariate.................125

5.6 ANOVA table for GLM of Great Egret (captures / min)1/2 as response variable
with the binary group variable (GB) as a factor ................................................... 125

5.7 Parameter estimates for GLM of Great Egret (captures / min)1/2 with the binary
group variable (G B) as a factor...........................................................................125

5.8 ANOVA table for GLM of Great Egret (captures / min)1/2 as response variable
with prey density (PD) as covariate and the binary group variable (GB)..............126

5.9. Parameter estimates for GLM of Great Egret (captures / min)/2 as response
variable with prey density (PD) as covariate and the binary group variable .........126

5.10 ANOVA table for GLM of Great Egret (captures / min)1/2 as response variable
with group category (GC) as a factor .................................................................. 126

5.11 Parameter estimates for GLM of Great Egret (captures / min)12 as response
variable with group category (GC) as a factor. .................................................127

5.12. Information-theoretic model selection results for GLM analysis of Snowy Egret
(captures / min)12 as response variable, prey density (PD) as covariate............ 127

5.13 ANOVA table for GLM of Snowy Egret (captures / min)1/2 as response variable
with the binary group variable (GB) as a factor..................................................127

5.14 Parameter estimates for GLM of Snowy Egret (captures / min)1/2 with the binary
group variable (G B) as a factor. .................................................... ....................... 127

5.15. ANOVA table for GLM of Snowy Egret (captures / min)1/2 as response variable
with prey density (PD) as covariate and the binary group variable (GB)............128






5.16. Parameter estimates for GLM of Snowy Egret (captures / min)2 as response
variable, prey density (PD) as covariate and the binary group variable (GB) .......128

5.17. ANOVA table for GLM of Snowy Egret (captures / min)12 as response variable
with the interaction between prey density (PD)............................ ...................128

5.18. Parameter estimates for GLM of Snowy Egret (captures / min)/2 as response
variable with the interaction between prey density (PD).......................................129

5.19. Information-theoretic model selection results for GLM analysis of Tricolored
Heron (captures / min) 12 as response variable, prey density (PD).........................129

5.20. Summary of foraging rates of Great Egrets in various sized groups ......................129

5.21. Summary of foraging rates of Snowy Egrets in various sized groups....................129

5.22. Summary of foraging rates of Tricolored Herons in various sized groups.............130

5.23. Summary of studies investigating the benefits to wading birds of foraging in
aggregations compared to foraging solitarily....................................................137

B. 1. Model selection results for GLM analysis with (Great Egret density) ................. 148
B.2. Model selection results for GLM analysis with (Snowy Egret density)12.............149

B.3. Model selection results for GLM analysis with (Tricolored Heron density)/2........150

B.4. Model selection results for GLM analysis with (White Ibis density) ....................151

B.5. ANOVA table for GLM of (Great Egret density)1/2 with management type as a
factor and fish density as a covariate. ................................................................151

B.6. Parameter estimates for GLM of (Great Egret density)1/2 with management type
as a factor and fish density as a covariate. ..........................................................152

B.7. ANOVA table for GLM of (Great Egret density)12 with fish density as a
cov ariate. ........................................ .. .............. ........ ...... ................ ................152

B. 8. Parameter estimates for GLM of (Great Egret density)/2 fish density as a
covariate. ........................................................................................................... 152

B.9. ANOVA table for GLM of (Great Egret density)1/2 with the interaction between
prey density and habitat type...................................................... ........................... 153

B. 10. Parameter estimates for GLM of (Great Egret density)"2 with the interaction
between prey density and habitat type. .............................................................153

B. 11. ANOVA table for GLM of (Great Egret density)12 with the interaction between
prey density and management type. ....................................................................154






B. 12. Parameter estimates for GLM of (Great Egret density)1/2 with the interaction
between prey density and management type .......................................................154

B. 13. ANOVA table for GLM of (Great Egret density)1/2 with management type as a
factor ....................... ... ..................... ............ ...... ............. ................... ............ 55

B. 14. Parameter estimates for GLM of (Great Egret density)1/2 with management type
as a fact r. ........................................ ........................ .................................... ...... 15 5

B. 15. ANOVA table for GLM of (Great Egret density)1/2 with habitat type as a factor
and fish density as a covariate......................................................................... 155

B. 16. Parameter estimates for GLM of (Great Egret density)1/2 with habitat type as a
factor and fish density as a covariate. ......................................................................156

B. 17. ANOVA table for GLM of (Great Egret density)1/2 with fish density and water
depth as covariates. ............................................................ ........................... 156

B. 18. Parameter estimates for GLM of (Great Egret density)-2 with fish density and
water depth as covariates................................ .................. ........................156

B. 19. ANOVA table for GLM of (Snowy Egret density)12 with the interaction
between prey density and habitat type. .................................................................157

B.20. Parameter estimates for GLM of (Snowy Egret density)1/2 with the interaction
between prey density and habitat type. ...........................................................157

B.21. ANOVA table for GLM of (Tricolored Heron density)12 with the interaction
between prey density and habitat type. ............................. ...........................158

B.22. Parameter estimates for GLM of (Tricolored Heron density)1/2 with the
interaction between prey density and habitat type. ............................................... 158

B.23. ANOVA table for GLM of (White Ibis density)1/2 with the interaction between
prey density and habitat type.................................................. ........................... 159

B.24. Parameter estimates for GLM of (White Ibis density)1/2 with the interaction
between prey density and habitat type. .............................................................159










LIST OF FIGURES


Figure page

1.1. A conceptual model of the factors influencing the foraging success of individual
w ading birds in shallow w wetlands ................................... ............... ..................9

1.2. Diagram depicting the relationship between the functional value of the northern
Indian River Lagoon system for wading bird populations.......................................10

2.1. Foraging locations of Great Egrets (circles) and Snowy Egrets (triangles) followed
from three colonies in the northern Indian River Lagoon, Florida ........................27

2.2. Great Egret and Snowy Egret foraging flight habitat resource selection ratios..........28

3.1. Map of study site showing location of study impoundments .....................................37

3.2. Frequency of 1-m2 throw trap samples (tosses) within 7 density categories .............60

3.3. Mean fish densities measured during fixed-station fish sampling..............................60

3.4. Mean fish density was always greater for vegetated versus paired nearby
unvegetated sites based on random-site sampling from 5 quarterly sampling.........61

3.5. Mean fish biomass was usually greater for vegetated versus paired nearby
unvegetated sites ................................................... ......................................... 61

3.6. Frequency of 1-m2 throw trap samples (tosses) within 7 density categories in
unvegetated and vegetated flooded habitat .......................................................62

3.7. Estimated marginal means offish density (back-transformed) in vegetated and
unvegetated habitats within 4 impoundments on MINWR/KSC by season ............62

3.8. Estimated marginal means of fish density (back-transformed) in vegetated and
unvegetated habitats within 4 impoundments on MINWR/KSC...........................63

4.1. Map of study site showing location of 9 study impoundments. .................................79

4.2. Aerial photographs of study impoundments showing sections used to record
locations of foraging wading birds.......................................................................80

4.3. Aerial photographs of study impoundments showing sections used to record
locations of foraging wading birds....................................................................... 81






4.4. Aerial photographs of study impoundments showing sections used to record
locations of foraging w ading birds..........................................................................82

4.5. Mean density of wading birds during ground surveys of 9 impoundments on
KSC/M INW R, January July 2001...................................................................... 91

4.6. Mean density of wading birds by impoundment section (see figures 4.2-4.4)
during ground surveys of 9 impoundments on KSC/MINWR.................................92

4.7. A linear regression showed that as the proportion of open habitat increased the
w ading bird density decreased. ....................................... ... .................. ...............93

4.8. A linear regression showed that as the proportion of edge habitat increased the
wading bird density increased. ...................................................................................93

4.9. Habitat use by species of wading birds observed during ground surveys of
foraging habitat use of 9 impoundments on KSC/MINWR.....................................94

4.10. Proportion of unvegetated habitat contained within 9 impoundments on
K SC/M IN W R, circa 2001. ................................... ................ ............................94

4.11. Distance to edge for observations of wading birds of 10 species during ground
surveys of 9 impoundments on KSC/MINWR, January July 2001 ....................95

4.12. Resource selection ratios (wi) for wading birds foraging in impoundments on
M IN W R /K SC ..................................................................................... .. ................. 96

4.13. Histogram of sizes of wading bird foraging aggregations observed during ground
surveys of impoundments on KSC/MINWR, January July 2001. ........................97

4.14. Composition of wading bird foraging aggregations observed during ground
surveys of 9 impoundments on KSC/MINWR, January July 2001. ...................97

4.15. The number of birds observed in each of 4 group sizes categories by species.........98

4.16. Spatial distribution of wading bird foraging aggregations ....................................99

4.17. Temporal distribution of wading bird foraging aggregations...............................100

4.18. Wading bird density in 7 impoundments measured during aerial surveys
concurrent with fish sampling periods ...............................................................101

4.19. Mean wading bird density in 2 types of foraging habitat .......................................101

4.20. Mean wading bird density in 2 seasons measured in 6 impoundments during
aerial surveys concurrent with fish sampling periods ............................................102

4.21. Results of model predicting wading bird density as a function of prey density
and habitat type for Wetlands Initiative impoundments ......................................107






5.1 Map of study site showing location of several study impoundments on the
Kennedy Space Center / Merritt Island National Wildlife Refuge ......................120

5.2. Great Egrets had a higher capture rate when foraging in groups than when
foraging alone, while Snowy Egrets had higher capture rate foraging solitarily...131

5.3. Great Egrets increased their capture rate as group size increased ..........................132

A. 1. Mean monthly water levels within 6 impoundments on KSC/MINWR between
February 2000 and February 2003. ....................................................................... 146

A.2. Mean monthly water levels at 3 gauge stations near study impoundments on
KSC/MINWR between February 2000 and February 2003.................................147










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

HABITAT SELECTION AND FORAGING SUCCESS OF WADING BIRDS IN
IMPOUNDED WETLANDS IN FLORIDA

By

Eric Douglas Stolen

May 2006

Chair: Franklin Percival
Major Department: Wildlife Ecology and Conservation

Most wading birds (Ciconiiformes) in Florida are tied to wetland habitats, and thus,

to wetland protection and management. I investigated factors that influence piscivorous

wading bird's foraging habitat selection and use at three spatial scales, and reasons why

individuals choose to forage with others. Field work was conducted within the northern

Indian River Lagoon, a sub-tropical estuary in east central Florida.

Based on follows from colonies, I found that nesting Great Egrets (Ardea alba) and

Snowy Egrets (Egretta thula) preferred estuarine wetland habitat and avoided freshwater

wetlands. Interpretation of selection patterns depended on scale and habitats included.

All birds followed foraged within 13 km of breeding colonies, well within the maxima

reported for other colonies, suggesting that resources within impounded habitat met their

energetic requirements.

To facilitate testing hypotheses regarding wading bird habitat selection, I measured

wading bird prey density in paired vegetated and unvegetated sites within impounded salt






marsh. Over 88% of the fish captured belonged to three species (Cyprinodon variegatus,

Gambusia. holbrooki, and Poecilia latipinna). Fish were highly clumped spatially, and

density and biomass varied substantially between impoundments and seasons for both

habitats.

Habitat selection within impoundments matched previous studies. Most species

preferred unvegetated to vegetated habitat and all strongly preferred the area within 0.5 m

of the boundary between types. Prey density had a strong effect on wading bird habitat

use in unvegetated but not in vegetated wetland sites.

I measured the success rate of wading birds foraging alone and in mixed-species

aggregations, and also measured the prey density at the foraging sites and nearby unused

sites. Foraging sites had higher biomass (but not density) than the average level available

throughout the landscape. Foraging groups occurred at the sites with higher prey density

(but not biomass) than solitary foragers. Great Egrets benefited from foraging in groups,

Snowy Egrets did not.

This study underscored the importance of edge habitat for foraging wading birds in

impounded marshes, supported the social facilitation hypothesis regarding group

foraging, and provided baseline data to measure responses from restoration and

management of marsh habitats in Florida.


xvii









CHAPTER 1
INTRODUCTION

Florida is home to 16 regularly occurring species within the order Ciconiiformes,

birds commonly referred to as the long-legged wading birds (Robertson and Woolfenden

1992). Due to a long history of direct harvesting, habitat loss, and habitat alteration, most

species of wading birds in Florida have experienced two distinct reductions in population

sizes during the last 2 centuries (Ogden et al. 1978, Frohring et al. 1988, Ogden 1994).

Following an initial decline due to plume hunting in the late nineteenth century, most

species showed healthy increases following legal protection up to the 1930's. However,

a second decline associated with extensive alteration of wetlands, resulted in great

reductions to populations of most species in south Florida (Ogden 1994). Recently,

population declines have been documented throughout the state for many species (e.g.,

Frederick 1996; Ogden 1996a, 1996b). Comparison of 2 statewide nesting colony

surveys conducted in 1976-1978 and 1986-1989 indicated that the numbers of wading

birds breeding in the state declined, and populations became more fragmented, between

the two periods (Runde 1991). Preliminary results of a third statewide nesting colony

survey conducted in 1998-1999 show that the decline in numbers of breeding wading

birds and fragmentation of wading bird populations continues (J. Swan and J. Dodge,

Florida Fish and Wildlife Conservation Commission, personal communication). Thus,

many species of wading birds are listed as species of special concern by the Florida Fish

and Wildlife Conservation Commission (Wood 1996).











The interpretation of population declines in Florida is made more difficult by the

tremendous shifts in the distribution of wading birds during the last several decades in the

southeastern United States (Ogden 1994, Fleury and Sherry 1995, Frederick et al. 1996,

Frederick and Ogden 1997). These changes have been attributed to changes in

management of wetlands (especially hydrology) in the regions involved (Bancroft et al.

1994, Frederick and Spalding 1994, Ogden 1994). The ability of wading birds to shift

their use of foraging habitat in response to changing conditions over smaller temporal and

spatial scales also makes it difficult to track wading bird populations. It can be difficult

to separate such changing patterns of habitat use from changes in the status of

populations. Thus, there is a need for a more mechanistic understanding of factors

influencing wading bird habitat selection and use.

Wading birds are intricately tied to their wetland habitats, which provide resources

for all aspects of their existence. Unfortunately, wetland loss in Florida has occurred at a

staggering rate with over half of all marshlands (or 1.57 million ha) destroyed by 1987

(Kautz 1993). Many experts agree that the greatest current threat to wading bird

populations in Florida is the continued loss and alteration of wetland habitat in the state

(Frederick and Spalding 1994; Hoffman et al. 1994; Ogden 1994, 1996a; 1996b; Rodgers

1996; Runde 1996; Kushlan 1997). With a constantly expanding human population, it

does not seem likely that the pressure on natural wetland habitats in Florida will ease

soon (DeFreese 1991, Gilmore 1995). It is clear that the future of wading birds in Florida

is tied to protection of wetlands, and where possible (e.g., managed wetlands) multi-

species or integrated management. It is also clear that to formulate and implement a










suitable multi-species management strategy, a better understanding of several features of

their ecology is needed.

My goal in conducting this research was to augment our understanding of wading

bird selection of foraging habitat at different scales. In particular, I wished to learn more

about factors that influence individual piscivorous wading birds when they make choices

about what type of foraging habitat to use, and whether or not to forage with others. I

believe that science is essentially a way of gaining information about nature, and that

there are several different methods that we can use to gain such knowledge. Often,

controlled, replicated, manipulative experiments are viewed as the best way to gain

reliable scientific information. But in many situations it is difficult or even impossible to

perform such experiments. In these cases, 2 other tools available have proved

particularly useful: observational study and the comparative method. This work has been

mainly observational and I hope that it will make a contribution to our understanding.

The complexity of ecological systems makes management of natural resources

difficult for many reasons. Before a manager can successfully apply interventions to a

system, a basic understanding of the way the system will react is essential. This type of

knowledge comes from understanding the mechanisms by which elements of the system

interact. Before ecologists can develop mechanistic hypotheses, they must also

understand the patterns which comprise the systems of interest. This dichotomy was

explained eloquently by Wiens (1989). In trying to understand wading bird use of

foraging habitat I strove to delve into the mechanisms governing their interactions with

the environment (both biotic and abiotic). A conceptual model of the factors influencing

wading bird individual foraging success is given in Figure 1.1. Often though, I was










limited in what is known about the patterns of wading bird foraging habitat use. Thus,

several of my objectives were aimed at describing the patterns of wading bird habitat

selection and use, and foraging behavior in impounded wetlands. In a few cases I was

able to go deeper into glimpsing the underlying proximate causes of their behavior. The

objectives for this study were the following:

1. Determine the relative importance of impounded wetlands as foraging habitat for
wading bird populations breeding in the northern Indian River Lagoon ecosystem
(hereafter IRL), and factors influencing habitat selection at the landscape level
(Chapter 2),

2. Measure patterns (temporal and spatial) of wading bird prey distribution within
impounded wetlands in the northern IRL (Chapter 3),

3. Document scale-dependent factors involved in wading bird foraging habitat
selection and use within impounded wetlands in the northern IRL (Chapter 4),

4. Document the spatial and temporal patterns of occurrence of mixed-species
foraging aggregations of wading birds within impounded wetlands in the northern
IRL (Chapter 4),

5. Investigate the effect of prey density and group size on foraging success and
distribution of foraging wading birds within impounded wetlands in the northern
IRL (Chapter 5),

6. Based on the results of the above objectives, formulate conservation and
management recommendations to enhance wading bird habitat use and foraging
success within impounded wetlands in the northern IRL based on the above factors
(Chapter 6).

I began this investigation by conducting a review of the first 11 years of data from

a long-term monitoring program of wading bird habitat use on the Kennedy Space

Center/Merritt Island National Wildlife Refuge, hereafter referred to as KSC/MINWR

(Stolen et al. 2002). This analysis revealed several key patterns that are of interest to

managers at KSC/MINWR and require further explanation. The first pattern of interest

was the differential use of impoundments by wading birds on KSC/MINWR. Over the

11 year time period some impoundments had much higher use as wading bird foraging










habitat than others, despite a lack of obvious habitat differences explaining the patterns.

A second observation from the long-term data was that wading birds used wetland habitat

without emergent vegetation less, and wetland habitat with emergent vegetation more,

than availability would suggest. A related pattern observation was that there was a

seasonal shift in habitat use for most species with vegetated habitats becoming more used

during the late fall and early winter (periods of higher water levels).

In Chapter 2, I assessed the relative importance of impounded wetland habitat to

nesting populations of wading birds in the northern Indian River Lagoon system. This

has relevance to managers responsible for balancing the needs of different groups of

wildlife (e.g., waterfowl, shorebirds, wading birds, fish). I hypothesized that some

species of wading birds nesting within the northern IRL would select impounded wetland

habitat for foraging over estuarine edge and unimpounded freshwater wetland habitat

types.

In chapter 3, I delved into the underlying basis for habitat selection patterns by

examining patterns in the distribution of fish between management types, seasons and

habitat type (vegetated versus unvegetated). I tested whether there was a difference in

fish density or biomass between vegetated and unvegetated habitats within

impoundments. I also modeled the density of fish as a function of habitat, season, and

impoundment to explore patterns within the factors determining density of wading bird

prey.

In Chapter 4, I examined patterns of habitat use and foraging ecology of wading

birds within impoundments in search of factors important in determining foraging habitat

suitability. I specifically tested whether wading birds exhibited preference for any one of










3 habitat types within impoundments, namely, unvegetated, vegetated and edge habitat. I

also described species-specific patterns of occurrence of mixed-species foraging

aggregations on KSC/MINWR. I modeled the density of wading birds within

impoundments as a function of prey density, habitat type, water depth, change in water

depth, season and management,

In Chapter 5 I addressed the question of the relative importance of prey density

versus the presence of other foragers on individual foraging success. I tested 3

hypotheses. First, that wading bird foraging sites have higher prey density than nearby

unused sites, second, that wading bird groups occur at sites with higher prey density or

biomass than sites occupied by solitary foragers, and third, that individual Great Egrets or

Snowy Egrets derive a benefit from foraging in aggregations that is due to the presence of

other foragers (as opposed to any benefit due to enhanced prey density at sites with

groups).

The ultimate measure of the functional value of a habitat is the ability of

individuals using the habitat to recruit new members into the population (Garshelis

2000). Many studies of wading birds have demonstrated a tight link between the ability

of parents to secure food and their ability to successfully produce offspring (Powell 1983,

Hafner et al. 1986, Maddock and Baxter 1991, Hafner et al. 1993, Frederick and Spalding

1994, Butler 1995, Thomas et al. 1999, Jakubas 2004). Without the direct demonstration

of this link however, there is always a danger that individuals may select habitat that is

not beneficial (Van Home 1983, Schlaepfer 2002). My work attempted to measure

components that are directly related to reproductive success and survival of wading birds

(Figure 1.2). The presence of a large and persistent wading bird population in my study











area provides indirect evidence that the area provides critical resources to wading birds

(Stolen et al 2002).

I addressed these research objectives and hypotheses in the impounded wetlands on

the KSC/MINWR. This area encompasses a large portion of the northern IRL.

Stretching for ca. 250 km along Florida's Atlantic coast, the IRL is a sub-tropical estuary

with a high level of biodiversity, due to its location at the junction of the warm-temperate

Carolinian Province and the Tropical Caribbean Province (Gilmore 1995). Prior to recent

human disturbance, the eastern shore of the IRL was extensively vegetated with

mangrove swamps in the southern portion and irregularly flooded salt marsh habitats in

the northern portion (Schmalzer 1995). Almost all salt marsh habitat in the northern part

of the IRL was impounded for mosquito control by the 1970's (Brockmeyer et al. 1997).

The northern part of the IRL is isolated from the nearest ocean inlet and has very low

diurnal tidal changes (< 1 cm; Smith 1987). In this region, seasonal and wind-driven

water level fluctuations are of much greater importance (Smith 1993). The hydrology of

the northern IRL is marked by a high water period from September through December.

This is followed by a gradual decline in water level with the lowest level occurring in late

spring before the onset of summer rains. These changes greatly influence the depth of

water over salt marsh habitat that is connected to the estuary, and control the extent of

marsh surface covered with water (Trost 1968). A similar water-depth pattern also

occurs within impoundments that are isolated from the estuary, probably as a function of

rainfall, hydrostatic influence from the estuary, and evapo-transpiration (Stolen et al.

2002).


















KSC/MINWR contains 76 shallow impoundments in former salt marsh habitat.

Human impacts on saline and brackish marshes of this region began as early as 1926

when ditches were dug to drain marsh surfaces to reduce oviposition by mosquitoes

(Aedes spp.). Following World War Two, mosquito control efforts turned to intensive

spraying of DDT in the 1940's; however rapid evolution of pesticide-resistance rendered

the use of DDT ineffective, and increasing public concern for risks associated with its use

lead to the search for an alternative means of mosquito control. A solution to the

problem was found through impounding the salt marsh habitats that breed mosquitoes

(Provost 1969). The success of impounding salt marsh for mosquito control in the IRLS

culminated in the impoundment of nearly all of the marshes bordering the Indian River

Lagoon by 1970. Alteration of the natural hydrology changed the characteristics of the

wetland habitats within, prompting a profound change in the vegetative, fish, and bird

communities (Provost 1969, Rey et al. 1990, Schmalzer 1995, Brockmeyer et al. 1997).

The ultimate response of wading bird populations to the ongoing effort to restore the

hydrologic connection between isolated wetlands and the IRL estuary has yet to be

determined. This study provides baseline data to measure presumed benefits.
























Hydrology
Interference + Prey
(b(between Depletion
+/ foragers)

Vegetation Individual
(stmture) Prey Foraging
Availability Success



Weather Prey
Prey Behavior
C Density
C
C. [

Season + Disturbance
Season ,_ e P y

Composition





Figure 1.1. A conceptual model of the factors influencing the foraging success of
individual wading birds in shallow wetlands. Factors are depicted as boxes,
with broad environmental factors on the left and the factor of interest
(individual foraging success) on the right. Arrows depict casual relationships
between factors. The symbols at ends of arrows indicate the nature of casual
relationships (+ = positive influence, = negative influence, +/- = direction of
influence depends on level of the factor, C = complex relationship about
which insufficient information is known).





















Survival
Reproduction
(Persistence of Local Breeding
Colonies)


Chapter 2
Landscape Level
Forays From Breeding Colonies
Foraging Habitat Type Selection


Chapter 4 Chapter 3
Impoundment Level Assessment of
Between and Within Foraging Site Selection Prey Base


Chapter 5
Patch Level
Individual Foraging Success




Figure 1.2. Diagram depicting the relationship between the functional value of the
northern Indian River Lagoon system for wading bird populations, and the
specific topics investigated in this dissertation. Demographic performance
(survival and reproduction) is the ultimate measure of the utility of the habitat
for wading bird populations. I addressed several topics that influence the
ability of wading birds to obtain the resources necessary to survive and
successfully reproduce. The persistence of breeding colonies in the northern
Indian River Lagoon system since monitoring began in 1987, suggests that the
site has provided adequate levels of necessary resources over this time period.









CHAPTER 2
FORAGING HABITAT SELECTION BY NESTING GREAT EGRETS AND SNOWY
EGRETS

Populations of colonial nesting herons and egrets (hereafter ardeids) face a variety

of threats, including isolation and alteration of coastal wetlands (Erwin et al. 2005). In

some regions coastal wetlands are isolated from estuaries by impoundments that are

created to control wetland hydrology or function; these changes can strongly affect

foraging habitat availability and quality (Brockmeyer et al. 1997, Erwin et al. 2005).

Previous studies have shown that populations of colonial breeding wading birds can be

limited by the availability of foraging habitat near their colonies (Fasola and Barbieri

1978, Gibbs 1991, Butler et al. 1992, Ogden 1994, Gibbs and Kinkel 1997, Jakubas

2004). Similarly, food availability has been shown to limit the survival of nestlings

(Maddock and Baxter 1991, Butler 1995, Jakubas 2004) and in one case, their subsequent

reproductive success (Thomas et al. 1999). Thus the availability and quality of foraging

habitat surrounding colonies is important in determining reproductive success.

For central place foragers like nesting egrets, distance from the colony is a factor

that influences the choice of foraging habitat, and ultimately, demographic parameters

(Gibbs 1991, Rosenberg and McKelvey 1999). Smith (1995a) found that nesting success

and nestling production of Tricolored Herons (Egretta tricolor) was negatively associated

with flight distance. Similarly, Simpson et al. (1987) reported that Great Blue Herons

(Ardea herodias) that fed near colonies had greater breeding success than individuals that

fed farther from colonies. In some cases, colony abandonment may occur when flight










distances exceed 25-30 km (Frederick and Collopy 1989, Bancroft et al. 1994). Frederick

and Spalding (1994) hypothesized that lost time rather than increased energetic demands

induced by longer foraging flights was a major contributor to reduced breeding

productivity or colony abandonment.

Among the factors influencing wading bird habitat use, water depth, which controls

physical access to foraging habitat, and vegetation structure stand out as being the most

significant (Chapter 4). Previous studies have found that piscivorous wading birds prefer

water depths below 25-30 cm (Chapter 4, Custer and Osborn 1978, Powell 1987, Gawlik

2002), and more open habitats when foraging (Chapter 4, Breininger and Smith 1990,

Hoffman et al. 1994, Surdick 1998, Bancroft et al. 2002). Coastal impounded wetlands

often provide these foraging conditions (Erwin 1996). Previous studies have also

suggested that estuarine wetlands may provide better foraging habitat for piscivorous

wading birds than do nearby freshwater wetlands due to differences in prey (MacCarone

and Parsons 1994).

Nesting colonies of wading birds provide a tangible target for protecting and

managing surrounding foraging habitat (Ogden 1994, Kushlan 1997, Lombardini et al.

2001) and nesting ardeids can be a useful indicator of anthropogenic impacts to wetlands

(Stolen et al. 2005). However, for these biological elements to serve a conservation

function, information on the factors influencing selection of foraging habitat is needed.

The objective of this study was to quantify foraging habitat use by Great Egrets (Ardea

alba) and Snowy Egrets (Egretta thula) nesting on spoil islands within the northern part

of the Indian River Lagoon system in central Florida (USA). This site contains extensive

coastal impoundments and a large population of colonial nesting ardeids. As such, it









serves as a good site to investigate the use of impounded coastal wetlands by colonial

nesting ardeids and the potential interaction between flight distance and foraging habitat

availability. I compared patterns of foraging habitat use with available habitat at three

scales to assess the importance of impounded salt marsh habitat to nesting Great Egrets

and Snowy Egrets in this system. The basic scientific hypothesis tested (for each scale)

was that nesting wading birds selected impounded salt marsh for foraging habitat. I also

measured foraging flight distances and flight speed, because these parameters can

influence colony and foraging site selection (e.g., Gibbs 1991). Finally, I report

information on foraging groups and flock size leaving the colonies because levels of

foraging scalability in egrets varies widely among sites (Chapter 4, Caldwell 1981, Erwin

1983b, Kersten et al. 1991, Hafner et al. 1993, Master et al. 1993) and these attributes

may influence selection at local levels (Dall et al. 2005).

Methods

Study Site

Follows of birds from their nests to foraging sites were made from three mixed-

species wading bird nesting colonies located within the boundaries of the Kennedy Space

Center-Merritt Island National Wildlife Refuge (KSC/MINWR) (Figure 2.1). The 55,000

ha KSC/MINWR is located in the northern portion of the Indian River Lagoon system, a

subtropical estuary that stretches for ca. 250 km from Ponce de Leon Inlet to Jupiter Inlet.

This estuary is an important site for wading birds on the southeastern Atlantic coast of

North America (Schikorr and Swain 1995, Sewell et al. 1995). KSC/MINWR supports a

large wading bird population that utilizes freshwater and impounded salt marsh habitats

for feeding, roosting, and nesting (Smith and Breininger 1995, Stolen et al. 2002).

Almost all salt marsh habitat in the northern Indian River Lagoon was impounded for









mosquito control by the 1970's (Brockmeyer et al. 1997). Within 15 km of the three

colonies there were 90 impoundments (average size 141 ha, total area 12,716 ha)

containing salt marsh habitat adjacent to the estuary. The northern portion of the Indian

River Lagoon system is isolated from ocean inlets and has very low diurnal tidal changes

(< 1 cm; Smith 1987). In this region, seasonal and wind-driven water level fluctuations

are of much greater importance (Smith 1993). Habitat within impoundments is

predominantly a heterogeneous mixture of open water and vegetated cover types, with

tall marsh grass (e.g., Spartina bakeri) and short marsh vegetation (e.g. Distichlis spicata,

Batis maritima) predominating in vegetated areas (Schmalzer 1995). Efforts are

currently underway to reconnect isolated impounded wetlands to the estuary

(Brockmeyer et al. 1997).

Nesting colonies on KSC/MINWR have been monitored since 1987 (Stolen et al.

2002). Focal colonies were selected as the largest colony within each of three broad

north-south strata of KSC/MINWR (Figure 2.1). Two of the colonies occurred on spoil

islands located near dredged navigation channels (Mullethead Island and Banana River

#14) and the third was on a large naturally occurring island which had been drag-line

ditched prior to 1970 for mosquito control (Big Island). Mullethead Island has had

nesting colonies in every year since monitoring began in 1987; Big Island has had

colonies since 1993 and Banana River 14 since 1994 (E. D. Stolen, unpublished data).

Foraging Flight Observations

I followed Great Egrets and Snowy Egrets from their nests to their foraging

locations (hereafter referred to as "follows"). Follows were conducted using a NASA

Huey helicopter between sunrise and five hours after sunrise. The helicopter was

hovered 300 m horizontal distance from each colony at an altitude of 150 m until a bird









was observed leaving the colony. The helicopter remained at least 300 m behind the bird

while it was followed. Once a bird landed at a foraging location, a GPS position was

recorded for the location using a Garmin 12 channel GPS receiver, and the habitat type

was noted. Habitat types were defined as open water (patches of at least 2 m of no

emergent vegetation) or vegetated. A site was considered occupied if at least one other

wading bird occurred within 30 m of the landing location; in these cases the landing bird

was considered to have joined an aggregation. I recorded the identity of other wading

birds within 100 m of the subject's landing location. If the subject was part of a group,

an attempt was made to note the landing position of all members of the group, but often

this was not possible because the subject bird would continue flying. In these cases, only

information about the subject bird was recorded.

Habitat Use Analysis

GPS positions of landing sites were recorded in a GIS for calculation of distances

and habitat use analysis. Analysis was conducted using the software packages ArcView

3.2 and ArcMap 8.2 (ESRI, Redlands California). Quantification of available habitat was

based on a land cover map produced by the St. Johns River Water Management District

using photo interpretation of 1:40,000 color-infrared aerial photography taken in 1999

(Anonymous 2002). To calculate the total amount of available habitat, three concentric

distance-buffers of 5, 10, and 15 km radii were produced around each of the three

colonies, and the ensuing regions wee each dissolved into a single polygon where the

colony-specific distance-buffers overlapped. These distances were chosen to include all

wetland habitat available within the range of flight distances of Great Egrets and Snowy

Egrets observed in previous studies (e.g., Frederick and Collopy 1989, Bancroft et al.

1994).










Wetland habitats within the distance-buffers were reclassified into three groups:

impounded salt marsh, estuarine edge, and freshwater wetland. Using the GIS, all

wetland habitat occurring within the perimeter dikes surrounding impoundments was

classified as impounded salt marsh habitat, and the area of these wetlands within each

distance-buffer was summed. Some areas that were impounded have recently been

restored by removal or breeching of the perimeter dikes (Brockmeyer et al. 1997). I

included these areas in the total of impounded habitat for two reasons. Although

restoration of a few of these sites occurred decades ago, most restoration was completed

more recently and all restored areas had experienced long periods of impoundment. Thus

restored sites may not yet have returned to functioning as unimpounded wetlands.

Secondly, the total area of wetlands within these impoundments was small (less than 5

percent) compared with the total for all wetlands within the 15-km buffer. If these

wetlands were included in another category, they would slightly increase the expected

number of birds foraging in that habitat and decrease the expected number in impounded

salt marsh. Thus, inclusion of these wetlands within the impounded salt marsh area is

conservative when evaluating the hypothesis of preference for impounded salt marsh

(because it increases the predicted proportion of birds foraging within impoundments,

thus making it harder to conclude that birds selected this habitat).

Estuarine habitat was defined as shallow areas along the edges of three large

lagoonal basins in the study area: the Indian River, The Banana River, and the Mosquito

Lagoon (Figure 1). To calculate the area of estuarine edge accessible to wading birds two

different radii were produced along the estuarine edge within the basins,. I used 2 buffer

distances to quantify habitat for Great Egrets and Snowy Egrets separately. Distances for










species-specific buffer widths were determined by examining the distribution of distances

at which birds were observed during long-term foraging habitat use surveys of estuarine

edge which have been conducted monthly on KSC/MINWR since 1987 (Stolen et al.

2002). Distances were chosen that included 95% of all observations for each species as

follows: Great Egret 100 m (n=5804), Snowy Egret 40 m (n=3404).

The area of freshwater wetlands was calculated by first summing the area of all

unimpounded wetlands within each distance buffer, excluding the interiors of lakes and

rivers. Then to represent available foraging habitat within open water of lakes and rivers,

a 1-m buffer was generated within all lakes and rivers and the area added to the total for

unimpounded wetlands. Because the 2000 Land Cover and Land Use map did not map

wetlands less than 0.5 acres, I used polygons from a 1990 Land Cover map of

KSC/MINWR which included such features (Larson 1992). Because this map did not

extend far enough west to include the entire study area, I mapped these features (where

necessary) using 1999 orthorectified color infrared photography (Anonymous 1999).

Although both Great Egrets and Snowy Egrets will use forested wetlands (Palmer 1962,

Hancock and Kushlan 1984) it is unclear how important this habitat is to either species.

Thus I conducted habitat use analysis both with and without these wetlands included, to

assess the effect of uncertainty over their use. I chose to include the analysis both with

and without forested wetlands because planners may consider wetlands of all types as a

unit when making management decisions (i.e., lumping forested and unforested wetlands

together).

I conducted habitat selection analysis using a resource selection ratio: wi

=[proportion of habitat i used]/[proportion of habitat i available] (Manly et al. 2002). This










ratio measures the relative preference between habitat types allowing comparison of

preferences between habitats that are not affected by what types are included. For each

analysis, I calculated the resource selection ratio of the observed bird use of each of the

three habitat categories (impounded salt marsh, freshwater and estuarine edge) for each

distance buffer (5,10, and 15 km) for each species. Thus, I examined habitat use for 6

different scenarios for each species (i.e., 5 km radius with forested wetlands included, 5

km radius without forested wetlands included, etc.). Selection ratios were evaluated to

determine if they differed from 1.0 (and thus exhibited selection or avoidance of the

habitat type) using a Bonferroni partitioning of experiment-wise a= 0.05. I also tested

whether each selection ratio for each habitat type differed from the other 2 using methods

presented in Manly et al. (2002) for each habitat use scenario. In several cases more than

1 bird was followed from a colony to a single foraging site and in these cases, I included

only 1 location in the analysis keeping with the assumption that individuals observed are

independent (Manly et al. 2002). I assumed that individuals followed on different days

were independent from other observations. To allow comparison with other habitat

selection studies, I also calculated 95% confidence intervals on the proportions of habitat

used by each species for each scenario using methods previously described (Neu et al.

1974, Byers et al. 1984, Cherry 1996).

Distance traveled between the nest and the foraging site was calculated with the

GIS. Due to the difficulty of observing birds exactly as they left nests, colony centers

were used for the origins of all foraging flights. Duration of foraging flight was

calculated as the duration between the detection of the bird leaving the colony and the

time the bird landed at the foraging site. This usually resulted in a few seconds being









truncated between when a bird left a nest and when it was detected leaving a colony. The

average flight speed was calculated as the straight line distance traveled divided by the

time of travel; this is a minimum estimate because many birds changed headings during

travel resulting in actual flight distances being greater than straight line distances

calculated in the GIS.

Results

Foraging Flight Characteristics

Thirty Great Egrets and 54 Snowy Egrets were followed from 3 nesting colonies

between April 7 and June 9, 2000 (on 7 days for Great Egret and on 8 days for Snowy

Egret). This period was after most nests were initiated and adults could be expected to be

provisioning chicks (E. Stolen, unpublished data). Great Egrets landed at 28 and Snowy

Egrets at 31 unique foraging locations (Figure 2.1). The destination of 3 Snowy Egrets

and 1 Great Egret could not be determined. There were still many active nests in all 3

colonies when follows ended. The maximum numbers of nests counted in the entire

study area were 202 Great Egrets and 165 Snowy Egrets in 11 colonies (E. Stolen,

unpublished data).

Over half of all Snowy Egrets followed left the colonies in groups (8 of the 31

separate follows). Most of these groups were composed of conspecifics except in 2 cases

when Snowy Egrets left the colony with a Great Egret. The mean and median group size

of Snowy Egrets that left colonies in groups was 3.9 and 2.5 respectively. In addition, 4

of the Snowy Egrets followed (2 single birds and a group of 2) joined groups in flight.

Most Snowy Egrets that left colonies in groups or joined groups in flight arrived at

foraging sites with those groups (2 Snowy Egrets followed did not). All but 3 of the

Snowy Egrets followed arrived at foraging sites that were occupied by mixed-species











foraging aggregations; 1 arrived at a site occupied by a lone Snowy Egret and 2 arrived at

unoccupied sites. The mean and median size of mixed-species foraging aggregations

joined by Snowy Egrets is summarized in table 2.1. The distance, duration and speed of

all Snowy Egret foraging flights followed are summarized in Table 2.2.

Seven of the 30 Great Egrets followed left colonies in groups (5 of 28 separate

follows). Two left in groups composed of another single Great Egret, the others with

Snowy Egrets. The mean and median group size of Great Egrets that left colonies in

groups was 2.4 and 2 respectively. No Great Egret followed joined a group in flight.

Only 3 of the 7 Great Egrets that departed colonies in groups arrived at foraging sites

with those groups (2 of the 5 follows). Twenty-two of the 30 Great Egrets followed

arrived at foraging sites that were occupied by foraging aggregations (20 of 28 follows);

14 of the aggregations were composed of mixed-species foraging aggregations while 6

were composed of only conspecifics. The mean and median size of mixed-species

foraging aggregations joined by Great Egrets is summarized in table 2.1. The distance,

duration and speed of the Great Egret foraging flights followed are summarized in Table

2.2.

Over all groups followed, aggregations joined by Great Egrets were significantly

smaller than were those joined by Snowy Egrets (Mann-Whitney Test, Z = -2.011, p =

0.044). Great Egrets flew greater distances to foraging locations than did Snowy Egrets

(Mann-Whitney U test, Z = 2.474, p = 0.013); the duration of trips from colonies to

foraging sites were not statistically different between species (Mann-Whitney U test, Z =

1.689, p = 0.091).










Habitat Use Analysis

There was no difference in the frequency of use of the 3 habitat types between

species (x22 = 2.35, p=0.309). Great Egrets that landed in impounded salt marsh (15 of

28 sites) landed exclusively in open water and always joined a foraging aggregation.

Similarly, most impounded salt marsh sites in which Snowy Egrets landed were in open

water (14 of the 19). Snowy Egrets landing in impoundments always joined aggregations

of other wading birds. The size of aggregations joined by birds landing in impounded

salt marsh was larger than that for other habitats combined (Table 1) and the difference

was significant for both Great Egret (Mann-Whitney Test, Z = -1.987, p = 0.047) and

Snowy Egret (Mann-Whitney Test, Z = -2.02, p = 0.043). Four of the freshwater wetland

sites at which Great Egrets landed were open water while 1 was vegetated. Similarly, 6

of the freshwater wetland sites at which Snowy Egrets landed were open and 2 were

vegetated. All but 2 of the 12 estuarine edge sites at which Great Egrets and Snowy

Egrets landed were in open water up to 40 m from the shoreline edge. The two remaining

sites (one for each species) were located on small unimpounded islands of low marsh

vegetation located away from the mainland.

The proportion of total area within flight distances from colonies that was wetland

habitat ranged from 0.33 to 0.11 depending on flight radius, species, and whether or not

forested wetlands were included in the analysis (Table 2.3). Fifty-eight unique foraging

locations were identified during follows of Great Egrets and Snowy Egrets. More than

half of the foraging sites were located within impounded salt marsh (Tables 2.4 and 2.5).

When forested wetlands were included in the analysis, both species had positive selection

ratios for impounded salt marsh and estuarine habitats and negative selection ratios for

























22


freshwater habitat at all spatial scales (with the exception of impounded habitat for Great

Egrets at the 5 km radius; Figure 2.2). Great Egrets showed avoidance of freshwater

habitat at the two larger spatial scales but not at the smallest scale; Snowy Egrets showed

avoidance of freshwater habitat at both the largest and smallest spatial scales, and

selection for impounded habitat at the largest spatial scale (Figure 2). These patterns of

habitat selection and avoidance largely disappeared when forested wetlands were

excluded form the analysis (Figure 2.2).
















Table 2.1. Summary of the sizes of foraging aggregations joined by Great Egrets and
Snowy Egrets followed from three colonies on the Merritt Island National
Wildlife Refuge. Entries in table for each species are mean number (SE) of
birds in aggregations on the first row, followed by median and sample size on
the next row.
Species Impounded Freshwater Estuarine Edge Combined (all sites)

Great Egret 112.0 (33.1) 11.0(9.5) 6.2(3.4) 70.4(22.8)
107.5, n = 12 2.0, n = 3 3.0, n = 5 10.5, n = 20

Snowy Egret 256.1 (82.1) 24.9 (8.0) 47.5 (12.5) 183.4 (58.9)
100, n = 19 30.0, n = 7 47.5, n = 2 47, n = 28


Table 2.2. Summary of flight distance, duration and speed for Great Egrets and Snowy
Egrets followed from three colonies on the Merritt Island National Wildlife
Refuge. Values in parentheses are standard errors. Sample sizes are listed in
table.


Measure
Average distance (km)

Median Distance (km)


Great Egret
6.2 (0.46), n=28


5.6, n=28


Snowy Egret
4.7 (0.48), n=31


4.2, n=31


Distance range (km)

Average duration (min)

Median Duration (min)

Duration range (min)

Average speed (km/hr)a


1.8 10.7, n=28

10.3 (1.0), n=28


9.4, n=28


2.0 22.8, n=28


38.8 (1.6), n=28


0.7- 12.5, n=31

8.0 (0.92), n=30


6.5, n=30

1.0- 18.8


38.2 (1.7), n=30


a average speed calculated from duration and distance of each flight.


















Table 2.3. Areas of Great Egret and Snowy Egret foraging habitats within three flight-
radius distances from three nesting colonies in the northern Indian River
Lagoon, Florida. Areas are in hectares.
Summary 5 km 10 km 15 km

Total area within distance of colonies 23,442 82,162 152,063

Area canal and lake 1 m buffer 19 59 78

Area estuary edge 40 m buffer 686 1,697 2,375

Area estuary edge 100 m buffer 1,532 3,785 5,278

Area impounded salt marsh with forest 3,509 9,857 12,064

Area impounded salt marsh without forest 2,710 7,782 9,448

Area freshwater wetlands with forest 2,620 8,930 14,418

Area freshwater wetlands without forest 632 2,969 4,380


Proportion Great Egret habitat with forests 0.33 0.28 0.21

Proportion Great Egret habitat without forests 0.21 0.18 0.13

Proportion Snowy Egret habitat with forests 0.29 0.25 0.19

Proportion Snowy Egret habitat without forests 0.17 0.15 0.11













Table 2.4. Comparison of the proportion of bird use versus the proportion of habitat availability, for three types of foraging habitat by
nesting Snowy Egrets. Analysis was conducted at three spatial scales (15, 10 and 5 km buffer distances). Bold text
indicates evidence of selection (+) or avoidance (-) for a particular habitat type and scale.
Comparrison Impounded salt marsh Freshwater wetlands Estuanne


15 km buffer
Numbers of individuals followed to each type of location within buffer 19
Bird use of habitat as proportion of all use within buffer (95 % confidence interval) 0.61
Habitat as proportion of total area within buffer, forested wetlands included 0.42
Habitat as proportion of total area within buffer, forested wetlands not included 0.58

10 km buffer
Numbers of individuals followed to each type of location within buffer 17
Bird use of habitat as proportion of all use within buffer (95 % confidence interval) 0.60
Habitat as proportion of total area within buffer, forested wetlands included 0.48
Habitat as proportion of total area within buffer, forested wetlands not included 0.62

5 km buffer
Numbers of individuals followed to each type of location within buffer 15
Bird use of habitat as proportion of all use within buffer (95 % confidence interval) 0.67
Habitat as proportion of total area within buffer, forested wetlands included 0.51
Habitat as proportion of total area within buffer, forested wetlands not included 0.66
a based on Cherry (1996).


(0.37, 0.93)


(0.35, 0.92)


(0.36, 0.95)


0.26 (0.09, 0.55)
0.5
0.27


0.27 (0.09 0.55)


(0.030, 0.38)


0.13 (0.02, 0.27)
0.08
0.15


0.13 (0.02, 0.27)
0.08
0.14


0.14 (0.01, 0.27)
0.1
0.17













Table 2.5. Comparison of the proportion of bird use versus the proportion of habitat availability, for three types of foraging habitat by
nesting Great Egrets. Analysis was conducted at three spatial scales (15, 10 and 5 km buffer distances). Bold text
indicates evidence of selection (+) or avoidance (-) for a particular habitat type and scale.


Comparison
15 km buffer
Numbers of individuals followed to each type of location within buffer
Bird use of habitat as proportion of all use within buffer (95 % confidence interval)
Habitat as proportion of total area within buffer, forested wetlands included
Habitat as proportion of total area within buffer, forested wetlands not included

10 km buffer
Numbers of individuals followed to each type of location within buffer
Bird use of habitat as proportion of all use within buffer (95 % confidence interval)
Habitat as proportion of total area within buffer, forested wetlands included
Habitat as proportion of total area within buffer, forested wetlands not included

5 km buffer
Numbers of individuals followed to each type of location within buffer
Bird use of habitat as proportion of all use within buffer (95 % confidence interval)
Habitat as proportion of total area within buffer, forested wetlands included
Habitat as proportion of total area within buffer, forested wetlands not included
a based on Cherry (1996)


Impounded salt marsh Freshwater wetlands Estuanne


(0.29, 0.88)


0.54 (0.28, 0.88)
0.44
0.53


(0.07, 0.75)


0.18 (0.04, 0.37)
0.46 -
0.23


0.15 (0.02, 0.31)
0.40-
0.21


2
0.22 (0.00, 0.37)
0.34
0.13


0.29 (0.01, 0.58)
0.17
0.28


0.31 (0.11, 0.62)
0.17
0.26


3
0.33 (0.03, 0.58)
0.2
0.31




























Atlantic Oce


Figure 2.1. Foraging locations of Great Egrets (circles) and Snowy Egrets (triangles)
followed from three colonies in the northern Indian River Lagoon, Florida.













2.5

2.0

1.5-

1.0 -

0.5-

0.0-



2.5-

2.0-

1.5-

1.0-

0.5-

0.0



3.00

2.50

2.00

1.50

1.00

0.50

0.00


IMP UNI EST IMP UNI EST IMP UNI EST
Forested wetlands No forested wetlands Forested wetlands


IMP UNI EST
No forested wetlands


IMP UNI EST IMP UNI EST IMP UNI EST IMP UNI EST
Forested wetlands No forested wetlands Forested wetlands No forested wetlands



Figure 2.2. Great Egret and Snowy Egret foraging flight habitat resource selection ratios
(error bars show SE) displayed by habitat type, for analyses with and without
forested wetland habitats included, for three concentric radii of distances from
colony locations. Bars marked with show evidence of selection or
avoidance of habitat type. IMP=impounded salt marsh, UNI=freshwater
wetlands, EST=estuarine edge habitat.


IMP UNI EST IMP UNI EST IMP UNI EST IMP UNI EST
Forested wetlands No forested wetlands Forested wetlands No forested wetlands










Discussion

Colonial nesting Great Egrets and Snowy Egrets showed evidence of habitat

selection, but patterns were dependent on the scale of the analysis and the decision to

include or exclude forested wetlands. Both species avoided freshwater habitat and used

both impounded salt marsh and estuarine habitat more than expected at several spatial

scales when forested wetlands were included. When forested wetlands were excluded,

both species appeared to use habitat in proportion to availability. The difficulty in

assessing habitat selection in animals due to uncertainty over the designation of what

habitat is available is well-known (Johnson 1980, Alldredge and Ratti 1986, McClean et

al. 1998, Manly et al. 2002). My results caution researchers against arbitrarily including

or excluding habitats at various scales as it may lead to incorrect interpretations of habitat

selection, including placing undue importance on some habitat types.

My results are consistent with other studies that have shown that patterns of habitat

selection by egrets are dependent on the scale of the analysis (e.g., Custer and Osbom

1978, Fasola and Barbieri 1978, Gibbs et al. 1987, Gibbs 1991, Gibbs and Kinkel 1997).

One explanation for such a hierarchical process is that decisions made at the broadest

spatial scale can influence (limit) choices at more local scales (Johnson 1980). This

possibility may explain the lack of apparent habitat selection by egrets when forested

wetlands were excluded from the analysis. In this study, forested wetlands tended to be

more distant from colony sites, and egrets appeared to have selected colony sites that

were surrounded by abundant suitable foraging habitat. Habitat selection within the birds

flight radius around the colony appeared to be determined by factors other than coarse

wetland type (i.e., forested versus non-forested).










In this study almost all salt marsh habitat was impounded. Although my results

highlighted the value of impounded salt marsh habitats in this system, I did not assess its

value as compared to unimpounded salt marsh habitat. Portions of the study area are the

focus of intensive restoration efforts (Brockmeyer et al. 1997). Therefore, it would be

appropriate to assess the question of selection between impounded versus unimpounded

salt marsh habitat when restored areas become more common in the landscape.

This study suggests that egrets nesting in the northern Indian River Lagoon estuary

can find suitable foraging habitat in relative close proximity to colonies to meet the

energetic demands of reproduction. All egrets I followed traveled less than 13 km, with

Great Egrets generally traveling farther than Snowy Egrets. This distance was within the

18 km radius traveled by most Great Egrets and Snowy Egrets from nesting colonies to

foraging sites reported in other studies (e.g., Custer and Osborn 1978, Thompson 1978,

Frederick and Collopy 1989, Bancroft et al. 1994, Smith 1995a). Moreover, the average

flight duration and flight speeds recorded in this study were similar to that recorded in

previous studies (Custer and Osborn 1978, Thompson 1978). Flying farther may occur if

or when habitat conditions around colonies deteriorate (Bancroft et al. 1994, Frederick

and Collopy 1989, Frederick and Spalding 1994). Under such conditions, birds have to

deal with trade-offs between carrying larger prey loads and the energetic demands of

increased travel (Smith 1995a). For this reason, distant habitats may not be preferred

under optimal conditions, but may be important under unusual circumstances (drought),

or during certain stages of the nesting cycle.

Most of the Great Egrets and Snowy Egrets I followed landed at sites with other

wading birds present, demonstrating the high level of social foraging under the conditions










in my study area (see Chapter 4). The benefits of joining a group might be associated

with locating prey resources (Krebs 1974, Hafner et al. 1998), or improved foraging

success of group foraging individuals (Chapter 5, Caldwell 1981, Hafner et al. 1982,

Cezilly et al. 1990, Master et al. 1993). However, the degree to which Great Egrets and

Snowy Egrets forage socially varies widely between studies. For example, Smith

(1995b) reported that Great Egrets and Snowy Egrets invariably joined foraging birds

when followed from a nesting colony in south Florida. Similarly, Custer et al. (2004)

found that 34% of Great Egrets followed landed at sites with other wading birds present.

In contrast, Custer and Galli (2002) found that Great Egrets landed at unoccupied sites

74% of the time. In this study, joining groups might have been facilitated by the

tendency of the Great Egrets and Snowy Egrets to travel to foraging sites in groups.

Group flights from nesting colonies also appears to vary widely between sites with some

authors reporting levels lower than in this study (e.g., Erwin 1983b,1984; MacCarone and

Parsons 1988; Smith 1995a).

Most studies of ardeid foraging flights have suggested that the availability of

foraging habitat surrounding the colonies is an important component in the protection of

wading bird nesting colonies. Nesting ardeids often switch foraging habitats rapidly in

response to changes in hydrology (Smith 1995a, Smith and Collopy 1995, Custer et al.

2004). Therefore, protecting a mix of different wetland types within flight distance of

colonies is prudent because unpredictable disturbances may affect some types but not

others. This study suggested that such a level of protection, including contingencies for

fluctuations/changes in habitat conditions, could be met by protecting a variety of

wetland habitats within 15 km of the nesting colonies of Great Egrets and Snowy Egrets




























32


in Merritt Island. The range of conditions described in this and other studies highlights

the flexibility in foraging behavior of these species, but also underscores the value of

gathering system-specific information to help guide management decisions on their

behalf.









CHAPTER 3
THE DISTRIBUTION OF WADING BIRD PREY IN IMPOUNDED WETLANDS IN
THE NORTHERN INDIAN RIVER LAGOON ESTUARY

Introduction

Many species of wading birds are dietary specialists feeding on small fish living in

shallow wetland habitats (Hancock and Kushlan 1984, Frederick 2002). Available work

has demonstrated a direct connection between prey distribution and piscivorous wading

bird foraging habitat use (Kersten et al. 1991, Fasola et al. 1996, Master et al. 2005), and

has highlighted the importance of understanding factors influencing prey availability

within wetlands (e.g., Gawlik 2002). Similarly, there is evidence that prey density (often

considered a surrogate for availability) is an important factor determining foraging

success of wading birds (e.g., Erwin et al. 1985, Cezilly et al. 1990). To understand the

factors underlying a predator's selection between available foraging habitats, it is helpful

to know the level of prey within each type. A few studies have attempted to describe the

relative distribution of piscivorous wading bird prey between available foraging habitats

(e.g., Erwin et al. 1985, MacCarone and Parsons 1994, Fasola et al. 1996, Trexler et al.

2003). However, the generality of relationships will only be established as similar types

of information are generated elsewhere.

Stretching for ca. 250 km along Florida's Atlantic coast, the Indian River Lagoon

System (IRL) is a sub-tropical estuary with a highly diverse fauna offish (Snelson 1983)

and a large and diverse population of piscivorous wading birds (Smith and Breininger

1995). Prior to human alteration, the eastern shore of the northern IRL was extensively










vegetated with irregularly flooded salt marsh (Schmalzer 1995). However, almost all salt

marsh habitat was impounded for mosquito control between 1954 and the early 1970's

(Brockmeyer et al. 1997). Impounding of these wetlands has had a large impact on the

fish community within, beginning with a reduction in the diversity and abundance of fish

as a result of initial impounding (Gilmore et al. 1982, Harrington and Harrington 1982).

Native IRL salt marsh had a hydroperiod characterized by extreme changes in water

depth and extent of marsh surface coverage between different seasons (Trost 1968).

Impounding of marshes resulted in longer hydroperiods, greater water depths and a

reduction in vegetated surface and increase in open water lacking emergent vegetation

(Brockmeyer et al. 1997). These changes apparently increased the populations of small

resident fishes such as the Sheepshead Minnow (Cyprinodon variegatus) and the

Mosquitofish (Gambusia holbrooki) (Brockmeyer et al. 1997, Gilmore 1998).

The impounded salt marshes of the northern IRL are a good place to investigate

patterns of piscivorous wading bird prey distribution. Sketchy information from the pre-

impoundment period on wading bird use suggests that the native marsh habitat had low

wading bird use, estimated at 0.9 individuals ha-1 (Trost 1968). Since impounding,

wading bird density within some impounded salt marsh habitat apparently increased by a

factor of two (Provost 1968, Trost 1968), but several impoundments in the northern IRL

with similar habitat composition and similar management histories are used at vastly

different rates by foraging wading birds (Stolen et al. 2002). Within impoundments of

the northern IRL, piscivorous wading birds predominantly use shallow unvegetated

flooded habitat and to a lesser extent, shallow flooded wetlands with low stature, salt-

tolerant plants (chapter 4, Breininger and Smith 1990, Smith and Breininger 1995, Stolen









et al. 2002). However, it is unclear whether prey abundance or other factors (e.g., habitat

structure, hydrology) are the key determinants of habitat preference. The simple division

of available habitat into two categories (vegetated and unvegetated) makes quantifying

patterns of piscivorous wading bird prey abundance straightforward. I designed this

study primarily to test to the prediction that prey is more abundant in the unvegetated

habitat type more often used by foraging wading birds.

In addition, on-going restoration and management within this system allows the

opportunity to test relationships between hydrology and wetland connectivity with

wading birds and their prey (chapter 4, Stolen et al. 2005). Understanding of the spatial

and temporal abundance and distribution of resident fish within impounded salt marsh

habitat in the northern IRL would enhance the ecological foundation for management of

both fish and wading bird habitats. It also would provide a better basis to compare and

draw appropriate inferences and conservation insights about other wetlands in Florida.

My objectives were the follwoing: 1) measure the density and biomass of fish within both

unvegetated and vegetated wading bird foraging habitat within impounded salt marsh, 2)

investigate seasonal patterns of abundance between habitats and management types, and

3) characterize the patchiness of fish at a scale relevant to the use of this resource by

wading birds. This information will ultimately improve the understanding of the foraging

habitat selection and foraging ecology of piscivorous wading birds within this system.

Methods

Study Site

The study site consisted of two areas of impounded salt marsh located in the

northern portion of the IRL on the Kennedy Space Center-Merritt Island National

Wildlife Refuge (KSC/MINWR, Figure 3.1). Habitat within impoundments is












predominantly a mixture of open water and vegetated cover types, with tall marsh grass

(e.g., Spartina bakeri) and short marsh vegetation (e.g., Distichlis spicata, Batis

maritima) predominating in vegetated areas (Schmalzer 1995). The unvegetated open

water areas within impoundments are of 2 types, both of which have sharply defined

boundaries with vegetated areas. Many large round "potholes", ranging in diameter from

a few to several hundred meters, occur within vegetated areas of the marsh. These areas

are shallow (< 0.3 m) but may be over 1 m below the marsh surface in the centers. Large

areas of shallow (<0.3 m) estuarine water that were impounded occur along the estuarine

edges of impoundments and often grade into remnant creeks towards the higher marsh.

Ditches (2-5 m across) along the impoundment perimeters also contain unvegetated open

water habitat, but generally these areas are too deep for foraging by wading birds.

The northern portion of the IRL is isolated from ocean inlets and has very low

diurnal tidal changes (< 1cm; Smith 1987). In this region, seasonal and wind-driven

water level fluctuations are of much greater importance (Smith 1987,1993). Hydrology is

marked by a high water period from September through November (due to a local

seasonal sea-level increase and a summer rainy season), followed by a gradual decline in

water level with the lowest level occurring in early spring. These changes greatly

influence the depth of water over salt marsh habitat that is connected to the estuary,

controlling the extent of marsh surface covered with water (Trost 1968). A similar

pattern occurs within impoundments isolated from the estuary, although water depths are

generally greater than in unimpounded salt marsh (Stolen et al. 2002).





















































Figure 3.1. Map of study site showing location of study impoundments on the Kennedy
Space Center-Merritt Island National Wildlife Refuge. The fixed-station
impoundments were: T10K, T10L, T27B, T38, SHILOH 1 AND SHILOH
3.The random-site impoundments were: T10C, T10D, T10E, T10H, T10J,
T10K, AND T10L.









Prey Sampling

Throw-trap sampling (Kushlan 1981b) was used to quantify resident marsh fish

abundance. This type of sampling gear has been shown to produce accurate estimates of

fish abundance in shallow wetlands similar to those in this study (Chick et al. 1992,

Jordan et al. 1997, Stevens 2002). To avoid startling fish, researchers approached a

sample site by walking slowly and then tossed a 1-m2 throw-trap a distance of 1 to 2 m.

Once the trap landed, researchers quickly secured the edges of the trap against the

substrate. Fish were then scooped from the trap using a 40 by 30 cm dip net with 2-mm

mesh. Vegetation within the trap was removed if it impeded movement of the dip net.

When the large dip net was scooped 3 times without catching a fish, a 15 by 10 cm dip

net with 2-mm mesh was used, which was more effective in scraping along the edges and

into the comers of the trap. The sample was completed when the smaller dip net was

scooped 3 times without catching a fish. The standard length (from anterior tip of body

to base of the caudal fin) of the first 30 individuals of each species captured in each

throw-trap deployment were measured to the nearest mm. The mass of these fish was

estimated using species-specific regression equations developed for fish captured in other

impoundments on KSC/MINWR (Phil Stevens, USGS, unpublished data).

Two sets of impoundments containing salt marsh habitat along the edge of the IRL

were sampled. The first set, termed fixed-station sampling impoundments (Figure 3.1),

consisted of 3 pairs of impoundments, chosen to meet the following criteria based on data

collected during long-term monitoring of wading birds on KSC/MINWR (Stolen et. al.

2002): 1) impoundments had to be among those included in at least 5 years of monthly

wading bird foraging habitat use surveys, 2) impoundments within pairs must be

adjacent, and 3) within pairs the overall mean density of wading birds observed during










foraging habitat use surveys of one impoundment must have been at least twice as great

as the other. Once the pairs were selected, I identified at least 3 locations within each

impoundment that met the following criteria: 1) locations were areas of open water at

least 1 ha in size, 2) locations were adjacent to culverts that opened to the estuary 3)

locations were adjacent to relict tidal creeks. At each of these points, I set up a 3 by 3

cell grid of 100-m2 cells marked with short PVC poles. During each quarterly sampling

period, 3 cells were chosen randomly within each grid and samples were collected with a

1-m2 throw trap.

The second set of impoundments, termed random-station sampling impoundments

(Figure 3.1) consisted of 7 impoundments along the Black Point Wildlife Drive chosen to

overlap with a multi-disciplinary study of the effects of impoundment on salt marsh

habitat (referred to as the Wetlands Initiative, see Brockmeyer 2004 for details). In these

impoundments, I used a random sampling design to locate sites to compare fish density

between wetland habitat with and without emergent salt marsh vegetation. At each

random point, a 1-m2 throw trap was used to sample fish at the nearest unvegetated and

vegetated habitat to the random point. Unvegetated habitat was defined as a flooded area

with no emergent marsh vegetation that was at least 2 m in diameter. One throw trap

sample was collected at the unvegetated site, then a paired sample location was selected

within the nearest flooded vegetated habitat that was contiguous to the open water habitat

sampled. Here a throw trap sample was collected in vegetated habitat which was paired

with the unvegetated sample. Paired vegetated sites were selected within 1 m of the edge

of the contiguous open water, and at least 5 m from the location where the open water

sample was made. Vegetated habitat was defined as a flooded area with at least 25%









cover of emergent marsh vegetation and at least 2 m in diameter. If no open water

existed within 200 m of the chosen random point, no unvegetated sample was taken. If

no vegetated habitat existed within 200 m of the open water sample then no paired

vegetated sample was collected.

In addition to the fish data, other data recorded were: a GPS position of the sample,

water depth, distance through the water at which a researcher could see the tip if a finger,

a description of the substrate, surface water conditions, weather conditions, presence of

submerged aquatic vegetation, and for vegetated samples the portion of the trap occupied

by emergent vegetation. Random-station throw trap samples were collected quarterly

July 2001-July 2002. During each quarterly sampling period, 10-15 points were chosen

randomly within each impoundment. The number of points sampled in each habitat,

season, and impoundment combination is given in Table 3.1. During the study, two

impoundments (T10J and T10L) remained hydrologically isolated from the estuary while

the other impoundments were connected via culverts open during some periods.

Analysis

Density offish for each throw-trap deployment was calculated as the number of

individuals of all species removed from the net. Nonparametric tests were used in

comparisons of fish densities and biomass to avoid making assumptions about

distributions required for the use of parametric tests (Conover 1980). I explored patterns

of distribution of fish abundance measures (i.e., density and biomass for each prey

sampling type) by plotting histograms for each variable. I attempted to use

transformations (natural log, square root and inverse) to reduce skew and kurtosis within

each variable. Patterns of correlation between measures of fish abundance (density and

biomass) were explored using both Spearman's p for raw variables and Pearson's r for










transformed variables. I report back-transformed means and 95% confidence intervals of

abundance within season and impoundment. I calculated the mean length and mean

biomass of fish (all species combined) for each sample, using all individuals measured. I

explored patterns of these measures between habitat types using unpaired t-tests

including all samples with measurement data. I calculated the correlations between

sample density and mean length and biomass fish using Spearman's p.

I used information-theoretic model selection methods (Burnham and Anderson

2002) to explore patterns of fish density as a function of explanatory variables (e.g.,

season, impoundment, and habitat). This approach is based on generating a set of

alternative hypotheses that are stated as models relating a single response variable with a

set of predictor variables. The key step in this process is in generating a meaningful set

of candidate models based on biological insight into the system, rather than fitting a

series of models that includes various systematic combinations of predictor variables

(e.g., step-wise backward elimination). Information-theoretic model selection provides

an objective means of comparing the relative support among the models in the set, given

the data. Models were formulated as general linear models (GLMs). Models were

considered for interpretation of their parameters if they met the following criteria: 1)

AAICc of less than 10.0, 2) were included in the set of best supported models with

combined Akaike weights of 0.95 (95% confidence set) and 3) had an evidence ratio

relative to the best supported model greater than 0.135 (Burnham and Anderson 2002).

For the fixed-station samples, GLMs were fit using fish abundance in sample grids

as the response variable and including combinations of the explanatory variables

Impoundment (6 levels) and Season (3 levels). The set of candidate models included a










model with each of the main effects alone testing the hypothesis that each factor was the

only one important in explaining the variation in fish density. The model with both

factors was also included based on the hypothesis that both effects were important in

explaining the density offish in samples. I also included the full factorial model ( main

effects and the two-way interaction between factors) to evaluate the overall fit of the

models (Burnham and Anderson 2002). I used ln(fish density+1) as the response variable

in GLM analyses because this transformation produced the most normal distribution of

each of the abundance variables. The resulting variables were still skewed right, mainly

due to presence of many zeros.

For the random-station samples, GLMs were fit using fish abundance as the

response variable and including combinations of the explanatory variables Habitat

(vegetated or unvegetated), Impoundment (4 levels) and Season (4 levels). I excluded

samples collected in Post-nesting 2001 season and impoundments T10C, T10D and T10K

from the analysis due to lack of samples (see Table 3.1). The candidate model set

considered included models with each factor alone, models with all pairs of two factors,

models with each two-way interaction alone, a model with all two-way interactions, and

several models with two-way interactions and the remaining factor as a main effect. The

full factorial model including the main effects, all two-way interactions, and the three-

way interaction between factors was evaluated to assess the overall fit of the models

(Burnham and Anderson 2002). I used ln(fish density+1) as the response variable in

GLM analyses because this transformation produced the most normal distribution of each

of the abundance variables. The resulting variables were still skewed right, mainly due to

presence of many zeros.









Results

Fixed-Station Sampling

One hundred and fifty-five fixed-station throw trap samples were taken during 3

sampling periods in June 2000, July 2000 and January 2001; fish were present in 83

(54%) of these. Eleven species offish were captured (2,133 individuals); 93% of

individuals belonged to the top three species Sheepshead Minnow, Mosquitofish, and

Sailfin Molly (Poecilia latipinna) (Table 3.2). Other species ofnekton were rarely

captured during sampling at these sites; a total of 178 Grass Shrimp, (Paleomontes sp.)

were captured in 8 samples. Estimates of fish density and biomass were highly correlated

(Spearman's p = 0.962, n = 155, p < 0.0001; the sample size was 1 less than expected

because 2 samples were aggregated due to data recording error). The frequency

distribution offish density was highly skewed due to a large number of zeros (46% of

throw samples; Figure 3.2). Information-theoretic model selection resulted in a single

model that included both Season and Impoundment main effects and had an Akaike

weight of 0.92 (Tables 3.3- 3.5). Results from this model indicated that fish density was

higher in January 2001 than the other seasons and that impoundment T38 had higher

density than all other impoundments (Figure 3.3).

Random-Site Sampling

A total of 326 unvegetated and 203 vegetated points were sampled. In 128 cases

no vegetated habitat existed within 200 m of the sample point and thus only an

unvegetated sample was taken; in 5 cases no open habitat existed within 200m of the

sample point so only a vegetated sample was taken. Non-fish species were captured at 26

of 326 unvegetated sites and 51 of 203 vegetated sites sampled. Most of the non-fish

prey captured were shrimp (Table 3.6). Fish were captured at 174 of the unvegetated










sites and 180 of the vegetated sites. Fifteen species offish were captured, but over 88%

belonged to the top three species (Sheepshead Minnow, Mosquitofish, and Sailfin Molly,

Table 3.7). There was a weak positive correlation between the number offish and the

number of shrimp in samples (Spearman's p = 0.297, p =<0.0001, n=529). There was a

strong positive association between the occurrence of shrimp and the occurrence fish in

samples (x22 = 19.98, p<0.0001) and only 7 of 159 samples had shrimp but no fish.

Mean fish density varied substantially between impoundments (Figure 3.4).

Seasonal densities offish also varied greatly for both unvegetated (Table 3.8) and

vegetated habitats (Table 3.9). Unvegetated points were more likely to have no fish than

were the paired vegetated sites (21 = 44.76, p< 0.001). The density offish was greater in

vegetated than unvegetated habitat (Wilcoxon signed-rank test, p < 0.0001, z = -6.94,

n=218 pairs); the 95 % confidence interval of the difference between the density of the

paired vegetated and unvegetated sites was (1,6) individuals/m2. The mean fish density

(individuals/m2) and 95% confidence interval for all vegetated sites was 8.2 (6.7, 9.9) and

for all unvegetated sites 2.0 (1.6, 2.4). The correlation between the densities of fish in

vegetated versus unvegetated paired-samples was significantly positive, but weak

(Spearman's p = 0.300, n = 198, p < 0.0001). Fish biomass also varied greatly by

impoundment (Figure 3.5) and by season (Tables 3.10 and 3.11), and showed a similar

pattern to density. The mean biomass (g/m2) and 95% confidence interval for all

vegetated sites was 3.0 (2.5-3.7) and for all unvegetated sites 1.1 (0.9-1.4). Estimates of

fish density and biomass were highly correlated for both unvegetated (Spearman's p=

0.963, n = 326, p < 0.0001) and vegetated sites (Spearman's p= 0.852, n = 203, p <

0.0001). There was also a weak correlation between the biomass of fish between











vegetated and unvegetated sites (Spearman's p = 0.190, n = 198, p < 0.007). The

frequency distribution of fish density was highly skewed for unvegetated samples due to

a large number of zeros (47% of unvegetated samples, 19% of vegetated samples) (Figure

3.6).

The mean length and mean biomass of individual fish within samples is given in

Table 3.12. The mean length of fish (by sample points) was greater at unvegetated sites

(24.0 mm, SE=0.74, n=170) than at vegetated sites (21.3 mm, SE=0.61, n=179), and the

difference was significant (unpaired t-test with unequal variances: t=2.74, df=331.6,

p=0.007). Similarly, the mean biomass per fish at unvegetated sites (0.68 g, SE=0.11,

n=170), was greater than that at vegetated sites (0.52 g, SE=0.10, n=179), but the

difference was only marginally significant (unpaired t-test with unequal variances:

t=1.90, df=334.6, p=0.058). There was no correlation between the mean length and

density offish at sample points (Spearman's p=0-0.035, p=0.519, n=349). Nor was there

a correlation between the mean biomass and density of fish at sample points (Spearman's

p=0-0.061, p=0.257, n=349). The pattern of greater mean length in unvegetated sites was

true for 3 of the 4 most abundant species (Table 3.13).

Information-theoretic model selection resulted in two models with a combined

Akaike weight of 0.99(Table 3.14). The best-supported model included interactions

between Habitat and Season, and Habitat and Impoundment, and had an Akaike weight

of 0.82 (Tables 3.15 and 3.16). Fish density was highest in vegetated habitat in the Post-

nesting and Winter seasons; this pattern reversed in Pre-nesting and Nesting seasons

(Figure 3.7). Marginal means offish density within impoundments are shown in Figure































46


3.8. The model with all three two-way interaction terms also had some support with an

Akaike weight of 0.17 (Tables 3.14, 3.17 and 3.18).
















Table 3.1. Number of points sampled during random-station fish sampling in unvegetated
(unveg.) and vegetated (veg.) habitat during 5 seasons in 7 impoundments.
Post 2001 Winter 2001 Pre 2002 Nesting 2002 Post 2002
impoundment unveg. veg. unveg. veg. unveg. veg. unveg. veg. unveg. veg.
TIOC 15 9 10 10 10 10 10 8
T10D 12 10 9 9 10 5
T10E 11 10 10 10 1 10 9
T10H 15 14 10 10 10 10 4 10 10
T10J 15 15 10 10 10 10 10 2 10 10
T10K 12 15
T10L 13 15 15 14 10 10 10 3 10 9

Table 3.2. Occurrence of fish by species in throw-trap samples in 6 fixed-station
sampling impoundments on KSC/MINWR measured June 2000 through
January 2001. The first 2 columns give the number of samples (plots) with
each species present, and the species' rank by occurrence in samples. The
next 2 columns give the total number of each species captured for all samples,
and the species' rank by total number captured. The last 4 columns give mean
length, SE and range of lengths within species, and sample sizes for
measurements.
Mean
Number of Rank by Total Rank by length
Species Plots Plot number number (mm) SE Range (mm) n
Cyprinodon variegatus 59 1 1483 1 24.4 0.30 (7-114) 823
Elops saurus 3 8 15 7 39.2 1.41 (32-51) 15
Fundulus confluentus 2 9 3 11 28.0 4.00 (24-32) 2
Gambusia holbrooki 35 2 358 2 18.4 0.40 (8-50) 210
Gobiosoma robustum 1 10 1 12 38.0 1
Lucania parva 7 6 21 6 23.1 1.34 (13-34) 20
Menidia beryllina 10 5 71 4 34.3 1.01 (21-55) 68
Microgobius gulosus 11 4 29 5 31.8 1.55 (20-44) 24
Mugil cephalus 6 7 10 8 44.3 6.84 (23-84) 10
Poecilia latipinna 21 3 132 3 24.9 0.70 (14-49) 125
Trinectes maculatus 2 9 4 10 33.0 1.78 (30-38) 4
Unkown fish 3 8 6 9
no fish at site 72 72




















Table 3.3. Fixed-station fish sampling model selection results for GLM analysis with
ln[l+fish density] as response variable and season (3 levels, S) and
impoundment (6 levels, L) as explanatory variables.

Model MLE a2 k AICc AAICc n/k wi

L, S 1.10 9 27.13 0.00 6 0.92

S 1.55 4 32.02 4.89 13 0.08

L 1.64 7 42.57 15.44 8 0.00

L*S 0.72 19 43.76 16.63 3 0.00

Table 3.4. ANOVA table for GLM with fixed-station ln(l+fish density) as response
variable and explanatory variables season and location.
Type III
Sum of Mean
Source Squares df Square F Sig.
Corrected Model 51.4a 7 7.35 5.68 0.00
Intercept 114.63 1 114.63 88.66 0.00
Season 28.52 2 14.26 11.03 0.00
Location 23.91 5 4.78 3.70 0.01
Error 58.18 45 1.29
Total 234.99 53
Corrected Total 109.60 52
a R2 = 0.47 (Adjusted R = 0.39)
















Table 3.5. Parameter estimates for GLM with fixed-station ln(l+fish density) as response
variable and explanatory variables season and location.

Parameter P SE 95% CI P3

Intercept 2.48 0.47 (1.52- 3.43)

season = JAN 2001 1.35 0.38 (0.58 2.11)

season = JUL 2000 -0.33 0.39 (-1.12 0.46)

season = JUN 2000 Oa

location = Shiloh 1 -1.46 0.52 (-2.51 -0.42)

location = Shiloh 3 -1.28 0.57 (-2.43 -0.13)

location = T10K -2.38 0.59 (-3.56 -1.19)

location = T10L -1.72 0.55 (-2.84 -0.61)

location = T27B -1.08 0.55 (-2.20 0.03)

location = T38 0a
a This parameter is set to zero because it is redundant.

Table 3.6. Occurrence of non-fish nekton in random-site throw-trap samples in 7
impoundments on KSC/MINWR measured July 2001 through July 2002.
Table entries give the number of samples (sites) with each species and total
number captured by habitat. The total number of samples was 326 in
unvegetated and 203 in vegetated habitats.
Number of Sites Total number
Species Unvegetated Vegetated Unvegetated Vegetated
Grass shrimp (Palaemontes sp.) 23 48 702 48
Snapping shrimp (Alpheus sp) 2 2 5 2
Crayfish (Procambrus sp.) 1 2 1 2
Unknown tadpole 3 3
Unkown Isopod 1 1


















Table 3.7. Occurrence of fish by species in throw-trap samples in wetland habitat with and without emergent vegetation in 7 random-
site impoundments on KSC/MINWR measured July 2001 through July 2002. The first 2 columns give the number of
samples (sites) with each species, and the next 2 give the species' rank by occurrence in samples (sites). The next 2
columns give the total number of each species captured for all samples, and the last 2 give the species' rank by total
number captured.


Species
Cynoscion nebulosus
Cyprinodon variegatus
Floridichthys carpio
Fundulus confluentus
Fundulus sp.
Gambusia holbrooki
Gobiosoma bosc
Gobiosoma robustum
Gobiosoma sp.
Jordanella floridae
Lucania parva
Menidia beryllina
Menidia sp.
Microgobius gulosus
Mugil cephalus
Poecilia latipinna
Syngnathus scovelli
Syngnathus sp.
Unkown fish
no fish at site


Number of Sites
Unvegetated Vegetated
2
125 126
1 3
10 11
5 2
77 115
2 1
3 1
7 2


1
49
16
3
22
3
64
1

2
152


69

1
9

106

1
2
23


Rank by site
Unvegetated Vegetated
8
1 1
12 7
7 5
9 8
2 2
11 9
10 9
8 8


Total number
Unvegetated Vegetated
2
728 849
1 8
10 14
6 2
707 1331
4 1
25 1
8 4


1
275
27
3
52
4
763
1

3
152


Rank by number
Unvegetated Vegetated
9
2 3
13 7
8 5
10 9
3 1
11 10
7 10
9 8


257

1
12

1097

1
2
23


























Table 3.8. Density of fish (individuals/m 2) in wetland habitat without emergent vegetation in 7 random-site impoundments on
KSC/MINWR measured July 2001 through July 2002.


Impoundment

T10C

T10D

T10E

T10H

T10J

T10K


mI

0.:

0.:


Post-nesting
2001

ean 95% CI

2 (0.0-0.6)

3 (0.0-0.7)


a


0.4

2.7

4.6


(0.0-1.0)

(1.3-5.2)

(1.4-11.7)


Winter 2001-
2002

mean 95% CI

0.3 (0.0-1.1)

0.1 (0.0-0.3)

1.8 (0.3-5.0)

1.0 (0.0-3.0)

5.9 (2.0-15.2)

a


Pre-nesting 2002


mean

0.0

0.4

0.5

3.5

16.1

a


95% CI

(0.0-0.0)

(0.0-1.2)

(0.1-1.1)

(1.0-9.4)

(5.0-48.0)


Nesting 2002


mean

0.2

0.2

1.2

1.5

9.4

a


95% CI

(0.0-0.8)

(0.0-0.7)

(0.0-3.9)

(0.3-3.6)

(5.1-16.9)


Post-nesting
2002

mean 95% CI

0.1 (0.0-0.6)

0.1 (0.0-0.3)

0.1 (0.0-0.3)

1.8 (0.7-3.6)

5.5 (2.4-11.6)

a


4.1 (1.4-10.0) 7.3 (3.9-13.1)


18.0 (6.0-51.0)


a Not sampled due to change in impoundment selection following Post-nesting season 2001.


23.3 (5.4-91.5) 5.5 (1.9-13.7)


T10L























Table 3.9. Density of fish (individuals/m2) in wetland habitat with emergent vegetation in 7 random-site impoundments on
KSC/MINWR measured July 2001 through July 2002.
Post-nesting Winter 2001-2002 Pre-nesting Nesting 2002 Post-nesting
2001 2002 2002
Impoundment mean 95% CI mean 95% CI mean 95% CI mean 95% CI mean 95% CI

T10C 4.9 (1.5-13.0) 3.1 (1.3-6.3) --b b 10.0 (3.4-26.1)

T10D -b b b b 2.1 (0.5-5.5)
(12.4-
T10E -a 14.6 (7.4-28.1) 0.0 6.0 25.2 50.4)

T10H 0.3 (0.0-0.7) 5.3 (1.9-12.7) 0.0 1.8 (-0.8-34.3) 17.1 (8.4-33.9)
(29.1-
T10J 3.7 (1.9-6.8) 22.7 (10.3-48.4) 11.5 (5.3-24.0) 16.2 (-1.0-289179.4) 48.2 79.3)

T10K 10.1 (6.7-15.0) -a -a a a

T10L 5.7 (1.8-15.0) 23.1 (12.7-41.3) 10.1 (4.2-22.5) 2.2 (-0.8-43.1) 7.8 (2.9-18.8)
a Not sampled due to change in impoundment selection following Post-nesting season 2001.
b Not sampled due to lack of flooded vegetated habitat during sample period
























Table 3.10. Biomass (g/m2) of fish in wetland habitat without emergent vegetation in 7 random-site impoundments on KSC/MINWR
measured July 2001 through July 2002.
Post-nesting Winter 2001- Pre-nesting 2002 Nesting 2002 Post-nesting 2002
2001 2002
Impoundment mean 95% CI mean 95% CI mean 95% CI mean 95% CI mean 95% CI

T10C 0.1 (0.0- 0.2) 0.2 (0.0- 0.5) 0.0 0.0 (0.0- 0.1) 0.1 (0.0- 0.2)

T10D 0.1 (0.0- 0.3) 0.2 (0.0- 1.0) 0.4 (0.0- 1.1) 0.6 (0.0- 2.8) 0.0 (0.0- 0.1)

T10E -a 0.9 (0.1 2.3) 0.8 (0.1 2.0) 0.5 (0.0- 1.9) 0.0 (0.0-0.1)

T10H 0.2 (0.0- 0.5) 0.7 (0.0- 1.9) 1.9 (0.3- 5.4) 0.4 (0.0- 1.0) 0.2 (0.0- 0.6)

T10J 1.0 (0.2- 2.3) 2.6 (0.6- 6.9) 9.7 (3.0- 28.0) 3.3 (1.3 7.2) 2.6 (0.9- 6.0)

T10K 2.5 (0.5-7.0) -a __a _a _a

T10L 1.6 (0.5 3.4) 3.9 (1.3 9.7) 6.9 (1.9 20.1) 9.3 (1.9 35.2) 4.2 (1.4 10.3)
a Not sampled due to change in impoundment selection following Post-nesting season 2001.























Table 3.11. Biomass (g/m2) of fish in wetland habitat with emergent vegetation in 7 random-site impoundments on KSC/MINWR
measured July 2001 through July 2002.
Post-nesting Winter 2001- Pre-nesting 2002 Nesting 2002 Post-nesting 2002
2001 2002
Impoundment mean 95% CI mean 95% CI mean 95% CI mean 95% CI mean 95% CI

T10C 2.5 (0.6- 6.4) 2.5 (0.6- 6.5) -b -b 3.0 (0.7- 8.6)

T10D -b b b b 3.6 (0.6-11.9)

T10E -a 7.6 (3.9- 14.4) 0.0 (0.0- 0.0) 0.1 (0.0- 0.0) 8.8 (4.4- 16.6)

T10H 0.2 (0.0- 0.4) 1.6 (0.6- 3.1) 0.0 (0.0- 0.0) 1.2 (0.0- 18.3) 4.5 (1.3- 12.2)

T10J 0.4 (0.2- 0.7) 10.6 (6.9- 16.1) 3.7 (1.2 8.9) 2.6 (0.0 42900) 13.7 (7.6- 24.1)

T10K 2.6 (0.7- 6.3) -a __a _a _a

T10L 3.1 (1.4-6.1) 5.5 (2.8- 10.1) 2.9 (1.1-6.4) 0.5 (0.0- 3.0) 1.8 (0.4- 4.5)
a Not sampled due to change in impoundment selection following Post-nesting season 2001.
b Not sampled due to lack of flooded vegetated habitat during sample period.
















Table 3.12. Mean length and mean biomass of individual fish captured within random-
site throw-trap samples by species. Data is included for all samples for which
these measures were made.
Species Mean length (mm) SE n Range (mm)
Cynoscion nebulosus 73.0 3.00 2 (70-76)
Cyprinodon variegatus 21.3 0.22 1413 (5-55)
Floridichthys carpio 33.8 6.01 9 (19-77)
Fundulus confluentus 24.3 3.10 18 (10-52)
Fundulus spp. 42.2 7.18 6 (18-65)
Gambusia holbrooki 19.7 0.14 1512 (1-40)
Gobiosoma bosci 24.4 2.50 5 (19-33)
Gobiosoma microgulosus 26.0 1
Gobiosoma robustum 29.0 1.39 26 (19-50)
Gobiosoma spp. 24.3 1.90 9 (16-34)
Jordanellafloridae 26.0 1
Lucaniaparva 21.3 0.27 465 (7-44)
Menidia beryllina 36.3 1.83 24 (15-50)
Menidia spp. 36.3 10.88 4 (15-56)
Microgobius gulosus 30.6 1.24 60 (17-57)
Mugil cephalus 123.5 19.72 4 (74-160)
Poecilia latipinna 23.6 0.23 1353 (6-65)
Syngnathus scovelli 82.0 1

Table 3.13. Mean length of individual fish in 2 habitat types within random-site throw-
trap samples, listed by species. Data is included for all samples for which
these measures were made.
Unvegetated Sites Vegetated Sites


Mean Mean
Length Range Length Range
SPECIES (mm) SE n (mm) (mm) SE n (mm)
Cyprinodon variegatus 21.8 0.34 631 (5-55) 20.9 0.30 782 (7-50)
Gambusia holbrooki 21.1 0.25 444 (1-40) 19.1 0.17 1068 (7-39)
Lucaniaparva 21.0 0.38 240 (7-44) 21.6 0.38 225 (9-39)
Poecilia latipinna 27.5 0.32 532 (6-62) 21.0 0.28 821 (7-65)


v


v











Table 3.14. Random-site model selection results for GLM analysis with In[fish
density+1] as response variable and habitat type (unvegetated/vegetated, H),
season (5 levels, S), and impoundment (6 levels, I) as explanatory variables.
Model MLE a2 k AIC AAIC AICc AAICc n/k wi
H*I, H*S 1.09 15 53.69 0.00 55.52 0.00 18.53 0.82
H*I, H*S, S*I 1.02 24 53.95 0.27 58.70 3.18 11.58 0.17
H*I 1.21 9 70.98 17.29 71.65 16.13 30.89 0.00
H*S, I 1.21 12 76.08 22.39 77.26 21.74 23.17 0.00
H*I, S 1.21 12 76.36 22.68 77.54 22.02 23.17 0.00
H,I 1.36 6 98.19 44.50 98.50 42.98 46.33 0.00
H,I,S 1.36 9 104.02 50.33 104.69 49.17 30.89 0.00
*S, H 1.27 18 102.97 49.28 105.61 50.09 15.44 0.00
H*S 1.53 9 137.09 83.40 137.76 82.24 30.89 0.00
I 1.60 5 140.43 86.74 140.65 85.13 55.60 0.00
I,S 1.58 8 143.03 89.35 143.57 88.05 34.75 0.00
I*S 1.47 17 141.68 88.00 144.04 88.52 16.35 0.00
H 1.65 3 145.85 92.17 145.94 90.42 92.67 0.00
H,S 1.65 6 150.99 97.30 151.30 95.78 46.33 0.00
S 1.92 5 191.83 138.15 192.06 136.54 55.60 0.00
H*I*S (global model) 0.98 31 56.02 2.33 64.09 8.57 8.97 0.01



Table 3.15. ANOVA table for GLM with random-site ln(fish density+1) as the
explanatory variable and habitat*season and habitat*impoundment
interactions included.
Type III
Sum of Mean
Source Squares df Square F Sig.
Corrected Model 237.6a 13 18.28 15.94 0.00
Intercept 830.55 1 830.55 724.30 0.00
season 4.27 3 1.42 1.24 0.30
impoundment 62.77 3 20.93 18.25 0.00
habitat 35.69 1 35.69 31.12 0.00
impoundment habitat 32.56 3 10.85 9.46 0.00
season habitat 32.90 3 10.97 9.56 0.00
Error 302.73 264 1.15
Total 1713.00 278
Corrected Total 540.33 277
aR = 0.44 (Adjusted R = 0.41)













Table 3.16. Parameter estimates for GLM with random-site In(fish density+1) as the
explanatory variable and habitat*season and habitat*impoundment
interactions.


Parameter P
Intercept 2.34
season = nesting 2002 0.38
season = post-nesting 2002 -0.22
season = pre-nesting 2002 0.49
season = winter 2001 Oa
impoundment = T10E -1.92
impoundment = T10H -1.46
impoundment = T10J -0.26
impoundment = T10L Oa
habitat = vegetated 0.38
habitat = unvegetated Oa
impoundment = T10E habitat = vegetated 2.06
impoundment = T10E habitat = unvegetated Oa
impoundment = T10H habitat = vegetated 0.94
impoundment = T10H habitat = unvegetated Oa
impoundment = T10J habitat = vegetated 0.86
impoundment = T10J habitat = unvegetated Oa
impoundment = T10L habitat = vegetated 0O
impoundment = T10L habitat = unvegetated Oa
season = nesting 2002 habitat = vegetated -1.51
season = nesting 2002 habitat = unvegetated Oa
season = post-nesting 2002 habitat = vegetated 0.52
season = post-nesting 2002 habitat = unvegetated Oa
season = pre-nesting 2002 habitat = vegetated -1.05
season = pre-nesting 2002 habitat = unvegetated Oa
season = winter 2001 habitat = vegetated Oa
season = winter 2001 habitat = unvegetated Oa
a This parameter is set to zero because it is redundant.


SE
0.20
0.23
0.23
0.23


0.23
0.23
0.23


0.30


95% CI P
(1.94- 2.74)
(-0.08 0.83)
(-0.68 0.24)
(0.04- 0.95)


(-2.37 --1.46)
(-1.92- -1.00)
(-0.72 0.20)


(-0.21 0.97)


0.39 (1.30- 2.82)


0.38 (0.21 1.68)


0.35





0.44


0.33


(0.17- 1.55)





(-2.38 -0.63)


(-0.14- 1.17)


0.39 (-1.81 --0.29)

























Table 3.17. ANOVA table for GLM with random-site In(fish density+1) as the
explanatory variable and habitat*season, habitat*impoundment and
season*impoundment interactions.
Type III
Sum of Mean
Source Squares df Square F Sig.
Corrected Model 256.3a 22 11.65 10.46 0.00
Intercept 706.52 1 706.52 634.33 0.00
season 4.34 3 1.45 1.30 0.28
impoundment 61.91 3 20.64 18.53 0.00
Habitat 25.91 1 25.91 23.27 0.00
impoundment Habitat 20.23 3 6.74 6.05 0.00
season Habitat 34.34 3 11.45 10.28 0.00
season impoundment 18.71 9 2.08 1.87 0.06
Error 284.02 255 1.11
Total 1713.00 278
Corrected Total 540.33 277
aR = 0.47 (Adjusted R = 0.43)













Table 3.18. Parameter estimates for GLM with random-site In(fish density+1) as the
explanatory variable and habitat*season, habitat*impoundment and
season*impoundment interactions.


Parameter
Intercept
season = nesting 2002
season = post-nesting 2002
season = pre-nesting 2002
season = winter 2001
impoundment = T OE
impoundment = T1 OH
impoundment = T 1OJ
impoundment = T 1OL
habitat = vegetated
habitat = unvegetated
impoundment = T10E habitat = vegetated
impoundment = T10E habitat = unvegetated
impoundment = T1 OH habitat = vegetated
impoundment = T10H habitat = unvegetated
impoundment = T10J habitat = vegetated
impoundment = T10J habitat = unvegetated
impoundment = T10L habitat = vegetated
impoundment = T10L habitat = unvegetated
season = nesting 2002 habitat = vegetated
season = nesting 2002 habitat = unvegetated
season = post-nesting 2002 habitat = vegetated
season = post-nesting 2002 habitat = unvegetated
season = pre-nesting 2002 habitat = vegetated
season = pre-nesting 2002 habitat = unvegetated
season = winter 2001 habitat = vegetated
season = winter 2001 habitat = unvegetated
season = nesting 2002 impoundment = T10E
season = nesting 2002 impoundment = T10H
season = nesting 2002 impoundment = T10J
season = nesting 2002 impoundment = T10L
season = post-nesting 2002 impoundment = T OE
season = post-nesting 2002 impoundment = T1 OH
season = post-nesting 2002 impoundment = T 1OJ
season = post-nesting 2002 impoundment = T OL
season = pre-nesting 2002 impoundment = T10E
season = pre-nesting 2002 impoundment = T1OH
season = pre-nesting 2002 impoundment = T10J
season = pre-nesting 2002 impoundment = T10L
season = winter 2001 impoundment = T1OE
season = winter 2001 impoundment = T1 OH
season = winter 2001 impoundment = T1OJ
season = winter 2001 imooundment = T1OL


2.392
0.563
-0.847
0.707
0O
-1.608
-1.778
-0.51
0a
0.494
0O
1.732
Oa
0.825
0"
0.839
0O
0O
0O
-1.507
0
Oa
0.505
0O
-1.34
0O
0O
Oa
-0.518
-0.183
0.011
0"
0.353
1.288
0.931
Oa
-1.076
0.189
0.096
0O
0O
Oa
0"
0"


a This parameter is set to zero because it is redundant.


SE
0.244
0.382
0.349
0.372

0.362
0.363
0.353


95% CI P
(1.91 2.87)
(-0.19- 1.32)
(-1.53- -0.16)
(-0.02- 1.44)

(-2.32- -0.90)
(-2.49- -1.06)
(-1.21 -0.19)


0.302 (-0.10- 1.09)

0.414 (0.92- 2.55)

0.395 (0.05- 1.60)

0.354 (0.14- 1.54)




0.444 (-2.38 --0.63)

0.328 (-0.14-1.15)

0.425 (-2.18- -0.50)


0.549
0.518
0.533

0.457
0.457
0.457

0.566
0.567
0.453


(-1.60- 0.56)
(-1.20- 0.84)
(-1.04- 1.06)

(-0.55 1.25)
(0.39- 2.19)
(0.03- 1.83)

(-2.19- 0.04)
(-0.93- 1.31)
(-0.80 0.99)











80

70

60

50

S40

30

20

10

0


II .I


0 1-5 6-10 11-20 21-50 51-100 100+
Density (ind. / m2)

Figure 3.2. Frequency of 1-m2 throw trap samples (tosses) within 7 density categories in
fixed-station fish samples from impoundments on KSC/MINWR, sampled
June 2000 January 2001.


60

50
40


Shilh 1


Shiloh3


T10K


T10L


T27B


Figure 3.3. Mean fish densities measured during fixed-station fish sampling was greatest
in January 2001. Estimates based on data from 3 sampling periods, June 2000
January 2001 within 6 impoundments on KSC/MINWR.


SJUL_00
0 JAN 01


































T10C TIOD T10E T10H T1OJ T10K T10L



Figure 3.4. Mean fish density was always greater for vegetated versus paired nearby
unvegetated sites based on random-site sampling from 5 quarterly sampling
periods, July 2001 July 2002. Error bars give 95% confidence intervals of
back-transformed means.


T10C


T1OD


T10E


T1OH


T1OJ


T1OK


T1OL


Figure 3.5. Mean fish biomass was usually greater for vegetated versus paired nearby
unvegetated sites. Estimates based on random-site sampling data from 5 quarterly
sampling periods, July 2001 July 2002. Error bars give 95% confidence intervals of
back-transformed means.


0 Vegetated
O Unvegetated













o unveg count
E veg count









Irirli -E^


0 I I I I I I I
0 1-5 6-10 11-20 21-50 51-100 100+
Density (ind. / m2)

Figure 3.6. Frequency of 1-m2 throw trap samples (tosses) within 7 density categories in
unvegetated and vegetated flooded habitat. Samples taken July 2001 July
2002 in random-sample impoundments.


30
-Vegetated
25 Unvegetated

20 T /


Winter 2001-02 Pre-nesting 2002 Nesting 2002 Post-nesting 2002

Figure 3.7. Estimated marginal means of fish density (back-transformed) in vegetated and
unvegetated habitats within 4 impoundments on MINWR/KSC by season.
Estimates are from the GLM with random-site ln(fish density+1) as the
explanatory variable and habitat*season and habitat*impoundment
interactions included.


140

120

100

80

60

40

20








































T10E T10H T10J T10L


Figure 3.8. Estimated marginal means of fish density (back-transformed) in vegetated and
unvegetated habitats within 4 impoundments on MINWR/KSC by
impoundment. Estimates are from the GLM with random-site ln(fish
density+1) as the explanatory variable and habitat*season and
habitat*impoundment interactions included.










Discussion

The density of fish measured in this study was similar to other studies within the

northern IRL (Schooly 1980, Stevens 2002, Brockmeyer 2004). The density offish for

most sites that have been sampled in this system was higher than levels reported for

vegetated habitats of the Everglades, a nearby, ecologically similar area which also has a

large population of wading birds (Loftus and Eklund 1994, Trexler et al. 2002). Turner et

al (1999) pointed out that the Everglades is unusual among tropical wetland systems in

having unusually low fish standing stocks. However, the density of resident fish in

impounded wetlands in the northern IRL was often considerably lower than the 20

individuals/m2 considered by Trexler et al. (2003) to be the historic level of small marsh

resident fish in the Everglades ecosystem. This was especially true of the unvegetated

habitats (Table 3.8).

Other systems where wading birds forage have also been observed to have much

higher levels of prey density than observed in northern IRL impoundments. For example,

in tidally replenished isolated pools in a New Jersey salt marsh the density of fish was an

order of magnitude higher than that typically observed in this study (Master 1992, Master

et al. 1993). The relatively lower level of prey density observed in the northern IRL

could have implications for the foraging ecology of piscivorous wading birds which may

rely on concentrations of prey for successful foraging and nesting (Kushlan 1976b,

Bancroft et al. 1994, Frederick and Spalding 1994, Gawlik 2002). For some

impoundments (e.g., T10 E, T10L and T10 J, Table 3.9) fish densities in vegetated

habitat exceeded 20 individuals/m2 in some seasons. Thus, the potential exists for

foraging wading birds to locate higher prey concentrations within this system.










The impounded salt marsh habitats in the northern Indian River Lagoon are unique

in many ways (Montague et al. 1987). Of particular relevance to the resident small fish

community is the presence of the perimeter dikes which dampen the effects on

impounded wetlands of hydrologic changes in the estuary. Compared to the adjacent

estuary, water level is more stable in impoundments and impounded marshes are flooded

deeper than native marshes (personal observation). The resulting longer hydroperiod

wetlands should support larger populations of small marsh-resident fish (e.g., Sheepshead

Minnow, Mosquitofish, and Sailfin Molly) than would shorter hydroperiod unimpounded

marshes (Loftus and Eklund 1994, Gilmore 1998, Trexler et al. 2002). In addition, the

perimeter dikes serve as a barrier to predatory fish, further increasing the standing stocks

of small fish within impoundments (Gilmore 1998, Stevens 2002). Abundance of prey

has been suggested as an explanation of why impounded wetland habitat in the northern

IRL is attractive to foraging wading birds (Breininger and Smith 1990, Schikorr and

Swain 1995, Smith and Breininger 1995, Stolen et al. 2002)

Within northern IRL impoundments, there was a tendency for habitats in closed

impoundments (T10L and T10J) to have higher densities of fish than those connected via

culverts open to the estuary during the study period (Brockmeyer 2004). The closed

impoundments had higher water levels throughout the study period, especially during

times when the other impoundments were drying (Appendix A). Thus, longer

hydroperiods, isolation from predatory fish, or both factors may explain the higher

standing stocks of small marsh resident fish observed in the closed impoundments.

Densities of fish were generally higher in sites with emergent salt marsh vegetation

than sites without such vegetation. Higher density of fish in vegetated versus










unvegetated areas has been noted in other shallow systems containing mixtures of both

habitat types (Rozas and Odum 1988). Areas with vegetative structure may be attractive

to fish because of protection against predators (Werner et al. 1983) and sometimes also

because they contain more abundant food resources (Mclvor and Odum 1988, Rozas and

Odum 1988). There was also a marked seasonal pattern to abundance of fish between

sites with and without emergent salt marsh vegetation within impoundments, and the

pattern suggested that some marsh resident fish may move into vegetated habitats when

these sites become flooded in late fall and winter. In another impoundment in the

northern IRL, Stevens (2002) demonstrated that marsh resident fish (e.g., Sheepshead

Minnow, Mosquitofish, and Sailfin Molly) moved from the estuary edge to the vegetated

marsh surface as rising water levels flooded these areas in late summer.

While density of prey is obviously one important factor determining foraging

success, others such as the mean size of prey may also determine the suitability of wading

bird foraging habitat (Trexler et al. 1994, Werner et al. 2001). I found that while prey

density was higher in vegetated sites, unvegetated sites had larger prey. This could have

implications for wading birds since larger prey have higher energy per capture effort and

are perhaps energetically superior prey items. There was also a difference in species

composition between habitats with fewer Sheepshead Minnow in vegetated sites than

unvegetated. These differences in the species composition and mean size of prey

between vegetated and unvegetated sites might be important factors in determining

foraging habitat use of piscivorous wading birds in this system (Chapter 4).

In the northern IRL, there was a high level of variation in the abundance offish

between impoundments in both vegetated and unvegetated habitats (Figure 3.4). This

























67


suggests that wading bird prey is patchy at the among-impoundments spatial scale. There

was also evidence that prey was highly clumped at sampling sites within both habitat

types (Figures 3.2 and 3.6), suggesting that prey distribution was also patchy at the

within-impoundment spatial scale. Finally, there was also considerable seasonal

variability in both habitat types within impoundments (Tables 3.8-3.11). These patterns

of prey distribution have important implications for wading bird foraging habitat use

within this system (Chapter 4).









CHAPTER 4
EFFECTS OF HABITAT STRUCTURE AND PREY DISTRIBUTION ON WADING
BIRD FORAGING HABITAT USE

Introduction

General ecological theory predicts that individuals should preferentially forage in

habitats that provide higher levels of resources contributing to individual fitness (Krebs

and Kacelnik 1991). This leads to the prediction that, all factors being equal (e.g.,

competitor and predator density, absence of territorial exclusion), foragers will select

habitat based on their ability to extract crucial resources from those habitats instead of

foraging in other habitats (MacArthur and Pianka 1966, Charnov 1976, Rosenzweig

1981). Foraging habitat selection and use by long-legged wading birds (Ciconiiformes)

is of ecological interest because the ability of parents to secure food for their broods has

been linked to reproductive success (Powell 1983; Hafner et al. 1986, 1993) and foraging

success affects survival of both juvenile and adult wading birds (Frederick and Spalding

1994). Thus, foraging ecology is directly related to fitness, and hence, to factors that

control population trends.

Despite a large volume of published studies on wading birds (for recent reviews see

Kushlan 1978b, Bildstein 1997) several key questions regarding their foraging ecology

remain unresolved (Erwin 1983a, Bildstein 1997, Kushlan 1997). Foremost among these

are how wading birds locate their prey resources, and how they exploit those resources

once located. Answering such questions is challenging because an individual's foraging

site selection, and ultimately foraging success, depends on factors occurring at different










scales (Chapter 2, Johnson 1980), and often the connections between scales is not well

understood or documented. Understanding how wading birds select foraging habitat is of

particular conservation value in systems that have been modified in ways that affect

wetland functions that determine habitat quality for wading birds (Chapter 3). Here I

differentiate between the terms "habitat selection" which refers to the behavioral

mechanisms leading to preferential use of some habitats over others (Garshelis 2000),

and "habitat preference" which refers to the proportional use of a habitat compared with

its availability (Manly et al. 2002).

Several studies have demonstrated a positive relationship between prey abundance

and wading bird foraging rates (Kahl 1964, Erwin et al. 1985, Draulans 1987, Cezilly et

al. 1990). In shallow wetland systems, hydrology is often the principal control on the

abundance of wading bird prey populations via influences on reproductive cycles and

access to wetlands (Harrington and Harrington 1961, Kahl 1964, Kushlan 1976b,

Gilmore et al. 1982). A good example can be seen in the Everglades ecosystem in south

Florida where long periods of inundation may increase fish standing stocks (Loftus and

Eklund 1994). Conversely, in this same system occasional droughts may reduce

piscivorous predation pressure and release nutrients, creating conditions that allow small

marsh fish populations to greatly increase (Loftus and Eklund 1994, Frederick and Ogden

2001). In, managed systems, such hydrological patterns can be disrupted with negative

effects on populations of wading birds and their prey (Frederick and Ogden 2003).

More recently, research has focused on factors affecting prey availability, rather

than density, as paramount in determining piscivorous wading bird foraging success

(Frederick and Loftus 1993, Gawlik 2002). Hydrologic cycles for example, can affect the











availability of wading bird prey through water depth; wading birds do not dive and must

be able to physically reach their prey to be able to capture it (Kushlan 1976b, Custer and

Osbom 1978, Powell 1987, Gawlik 2002). Water quality (e.g., dissolved oxygen,

temperature) also has a potential impact on prey behavior and thus availability (Kersten

et al. 1991, Frederick and Loftus 1993). Hydrology is also a determining factor

influencing vegetative composition, distribution and density. Vegetation attributes (e.g.,

cover, structure) affect habitat quality for both wading birds and their prey, and greatly

affect prey availability.

Many authors have observed that wading birds preferentially forage in sites lacking

emergent vegetation over adjacent vegetated sites (Kersten et al. 1991, Hoffman et al.

1994, Chaves-Ramirez and Slack 1995, Schikorr and Swain 1995, Smith et al. 1995,

Smith and Breininger 1995, Surdick 1998, Bancroft et al. 2002, Gawlik 2002, Stolen et

al. 2002). In some cases, this may be because vegetative structure can influence wading

bird vulnerability to predators by reducing visibility (Caldwell 1986). Another

explanation is that increased structure greatly inhibits a forager's ability to locate and

capture prey. The relationships between these factors are complex (see Chapter 1, Figure

1.1) and thus decisions by wading birds regarding selection of foraging sites involve

tradeoffs. For example, in choosing between two habitats with different prey density, an

individual must tradeoff between factors effecting prey availability (water depth,

vegetative structure, presence of other foragers) and other factors including availability of

information (e.g., locations of other foragers), relative predation risk, and the potential

size of prey between sites.










In Chapter 2, I showed that nesting wading birds preferentially used impounded salt

marshes in the northern Indian River Lagoon watershed. There, birds foraged within 13

km of breeding colonies, well within the maxima reported for other colonies. These

results suggested that resources within impounded habitat met their energetic

requirements without their having to engage in long-distance forays. Birds flew from

colonies either in groups or by themselves depending on species, but almost invariably

joined foraging flocks when landing. In this chapter, I continue my assessment of habitat

selection by waders in the northern Indian River Lagoon by focusing on a smaller spatial

scale, namely, among and within impoundments. In Chapter 3 I showed that the density

offish was greater in sites with emergent wetland vegetation, but fish were larger in sites

without emergent vegetation. Previous work demonstrated that within impoundments of

the northern Indian River Lagoon, piscivorous wading birds predominantly use shallow

flooded habitat without emergent vegetation, and to a lesser extent, shallow flooded

wetlands with low stature, salt-tolerant plants (Stolen et al. 2002). However, it is unclear

whether prey abundance or other factors (e.g. habitat structure, hydrology) are the key

determinants of habitat selection.

In this chapter, I tested the hypothesis that wading birds prefer wetland habitat

without emergent vegetation to habitat with emergent vegetation, and also examined their

preference for the interface (edge) between these habitat types. Next, to better

understand factors that influence wading bird choice of foraging habitat, I modeled

wading bird density as a function of prey density, habitat, season, hydrology and

management. This allowed comparison of the relative importance of these factors in

determining wading bird habitat preference. Finally, I described the size and composition









of foraging flocks within impoundments because their presence may have proximally

influenced habitat selection (e.g., visual stimulus of foraging individuals), and also

because, depending on prey density and habitat characteristics, flock size and

composition might affect foraging success (Chapter 5).

Methods

The study site consisted of areas of impounded salt marsh habitat on the 55,000 ha

Kennedy Space Center-Merritt Island National Wildlife Refuge (KSC/MINWR). This

site is located in the northern portion of the Indian River Lagoon system (IRL), a

subtropical estuary which is an important site for wading birds on the southeastern

Atlantic coast of North America (Schikorr and Swain 1995, Sewell et al. 1995). The

northern portion of the IRL is isolated from ocean inlets and has very low diurnal tidal

changes (< 1cm; Smith 1987). In this region, seasonal and wind-driven water level

fluctuations are of much greater importance (Smith 1993). Habitat within impoundments

is predominantly a heterogeneous mixture of open water and vegetated cover types, with

tall marsh grass (e.g., Spartina bakeri) and short marsh vegetation (e.g., Distichlis

spicata, Batis maritima) predominating in vegetated areas (Schmalzer 1995).

Three sets of data were collected to address the objectives of this work. These were

the following: 1) wading bird density and distribution, 2) water levels within

impoundments, and 3) fish density and distribution. Below I describe the sampling

protocols for the first 2 data types. Sampling protocols and a detailed summary offish

data are presented in Chapter 3. I drew upon the fish data to test hypotheses concerning

habitat selection by waders in this chapter (see below). All data types were collected in 9

study impoundments located on KSC/MINWR (Figure 4.1); 7 of these impoundments

overlapped with the Wetlands Initiative, a simultaneous multi-disciplinary study of the










effects of hydrologic management on salt marsh habitat (Brockmeyer 2004). The

management of these impoundments was fixed during the period of my study. Therefore,

I could evaluate, via comparative tests, the effects of different management on some

wetland functions. Management types (for more details see Brockmeyer 2004) were the

following:

7. 1) Open a management strategy in which impoundment culverts are left open
throughout the year allowing water levels to fluctuate with the levels in the estuary.
Impoundments T10C and T10H.

8. 2) Rotational Impoundment Management (RIM) a widely used seasonal mosquito
control strategy that floods impoundments during the mosquito breeding season but
leaves culverts open to the estuary during non-breeding season (September May).
Impoundments T10D and T10E.

9. 3) Wildlife Aquatic Management (WAM) a management strategy widely
employed on the refuge that was designed to provide submerged aquatic vegetation
for waterfowl by gradually increasing water levels from spring through fall with
highest water levels in October or November. Then water levels are then lowered
in stages to allow waterfowl access to submerged aquatic vegetation throughout the
winter and spring. In this management type there may be a brief period in spring
during which culverts remain open, but the study WAM impoundments remained
closed during the course of the Wetlands Initiative. Impoundments T10J and
T10L.

10. 4) Restored refers to an impoundment in which the perimeter dike has been
completely removed in an effort to return the marsh to its native condition.
Impoundment T10K.

I used ground surveys of wading birds within impoundments to address questions

relating to factors influencing foraging habitat selection at the site (local or patch) scale.

Ground surveys were conducted weekly January July 2001 for all 9 impoundments.

Surveys began 30 minutes after sunrise and were usually completed within five hours.

Surveys were conducted while driving the perimeter dikes of the impoundment and

always followed the same route due to traffic restrictions on some of the dike roads. For

each individual or group of wading birds the following data were recorded: location










within predetermined zones within impoundments (Figures 4.2 4.4), the number and

species of individuals, microhabitat type (unvegetated flooded sites, unvegetated non-

flooded sites, vegetated flooded sites, perched), water depth (estimated from water

position on legs), and activity (foraging or loafing). Individuals were considered to be

part of a group if the distance to the next nearest wading bird was within 10 body lengths.

Whenever possible, approximate locations were recorded on aerial photos for all

aggregations consisting often or more wading birds with average inter-individual

distances less than 10 body lengths. All spatial data were recorded in a GIS to allow

spatial analysis of wading bird distribution patterns.

For each species, I tabulated the total number of individuals observed, the

proportion of individuals within aggregations, and the water depths at foraging locations

by habitat type (vegetated and unvegetated). Survey counts for each impoundment

section were divided by section area to calculate wading bird density. Because the

interior areas of impoundments were sometimes obscured from perimeter dikes, I mapped

the area of impoundments clearly visible from perimeter dikes using 1999 orthorectified

color infrared photography (Anonymous 1999) and used that area in all calculations of

wading bird density. Density was transformed as In(density+l) to help normalize data

allowing means and 95% confidence intervals to be calculated on the log-transformed

data; all statistics reported were back-transformed to the original units.

For each species, I conducted a foraging habitat selection analysis using a resource

selection ratio: wi =[proportion of habitat i used]/[proportion of habitat i available]

(Manly et al. 2002). This ratio measures the relative preference between habitat types

allowing comparison of preferences between habitats that are not affected by what types










are included. Wetland habitat within the survey area was first classified into vegetated

(sites with emergent vegetation) and unvegetated (sites without emergent vegetation)

wetland habitat using a vegetation classification map (Brockmeyer 2004). I next created

a 0.5 m buffer within the perimeter of each habitat type. The buffer areas were joined to

create a Im wide interface between the two habitat types called edge habitat. I subtracted

the respective buffer areas from both the vegetated and unvegetated wetland habitat

areas. This resulted in three wetland habitat types within the survey area: unvegetated,

vegetated and edge.

I conducted aerial surveys to evaluate factors influencing foraging habitat selection

at a larger scale (e.g., between impoundments). Aerial surveys of the 7 Wetlands

Initiative impoundments were conducted July 2001 July 2002 to overlap with fish

sampling (Chapter 3). Surveys were conducted 0900 1100 EST in a NASA Huey

helicopter flying at an altitude of approximately 60 m, and a speed of 60 kts. These

surveys were included in monthly wading bird foraging habitat use surveys conducted as

part of long-term monitoring on KSC/MINWR (Stolen et al. 2002). During surveys, each

impoundment was flown systematically such that all area within impoundments was

surveyed. At each impoundment, the following data were collected: species, number of

individuals, and cover type (wetland habitat with or without emergent vegetation).

Habitat specific wading bird density was calculated for unvegetated and vegetated habitat

within each impoundment. The area of each habitat type within impoundments was

calculated using a vegetation classification map (Brockmeyer 2004). When more than

one survey was conducted during a season (based on periods offish sampling), the mean

density was calculated.











Water level data for the Wetlands Initiative impoundments was obtained from the

St. Johns River Water Management District (for details of water level data collection see

Brockmeyer 2004). The data consisted of hourly recordings of water level within each

impoundment. For T10K (the restored impoundment) the water level measured in the

estuary adjacent to the impoundment was used. For each survey date I calculated the

mean daily water level for the study impoundments. I calculated the daily change in

water level as the difference between a day's mean water level and the previous days

mean water level. I then calculated the sum of the daily change in water level over the 7

days prior to the survey date. Finally I calculated mean monthly water levels for

examination of seasonal patterns of change in water levels. The daily impoundment

water level data used for modeling wading bird abundance is summarized in Appendix A.

Finally, I formulated a series of competing models that allowed me to evaluate

factors that may influence wading bird foraging habitat selection at the impoundment

level (Anderson and Burnham 2002, Johnson and Omland 2004). Models were

formulated as general linear models (GLM) predicting wading bird density as a function

of explanatory variables which included fish density, season (2 levels) management type

(3 levels), habitat (vegetated or unvegetated), water depth, and change in water depth.

Table 4.1 lists the models with a brief description of the factors evaluated by each.

Relationships were modeled for the 5 sampling periods during which fish density was

measured (Chapter 3). Seasons used in analysis were nesting (March May, included 2

fish sampling periods) and non-nesting (July-February, included 3 fish sampling periods).

Management types used in analysis were: Open, RIM, and WAM (see above for


















definitions). Impoundment T10K was not included in this analysis because its

management type (restored) was not replicated.

I used information-theoretic model selection methods (Bumham and Anderson

2002) to choose among the models (GLMs). Models were considered for interpretation

of their parameters if they met the following criteria: 1) AAICc of less than 10.0, 2) were

included in the set of best supported models with combined Akaike weights of 0.95 (95%

confidence set) and 3) had an evidence ratio relative to the best supported model greater

than 0.135 (Burnham and Anderson 2002). A model including the 6 main effects and all

two-way interactions considered in any other model was evaluated to access the overall

fit of the models (Bumham and Anderson 2002). The goodness-of-fit and other

diagnostics for meeting general linear model assumptions were investigated for all

models considered for interpretation (Grafen and Hails 2002). I also applied the same

model selection procedure separately for those individual wading bird species for which

the number of individuals was at least 10% of the total for all species combined. All

statistical calculations were performed using either Microsoft Access, Microsoft Excel

2003 (Microsoft Corperation, 1985-2003) or SPSS 12.0 (SPSS Inc. 1988-2003).





















Table 4.1. Models set for analysis of factors that effect wading bird density in foraging habitat on KSC/MINWR.
Models Description
P wading bird density controlled primarily by density of prey
P*H the effects of prey density on wading bird density differ between habitat types
P*D the effects of prey density on wading bird density depend on water depth
H wading bird density controlled primarily by habitat
H*S wading birds switch habitats seasonally due to factors other than prey density
H*D wading birds respond to depth differently in different habitats
M*H management influences wading bird habitat preference
D wading bird density controlled primarily by depth
C wading birds use change in water depth to choose foraging habitat
H*C wading birds respond to change in water depth differently bewteen habitats
H, C habitat preference and change in water depth
P, H wading bird density controlled by density of prey and habitat
H, D wading bird density controlled by depth of water and habitat
P, D wading bird density controlled by density of prey and water depth
M wading bird density controlled by management effects not included in any other variable
P, M wading bird density controlled by density of prey and management effects
P*M management effects influences the effect of prey density
aVariables: P= prey density, H = habitat, S = season, M = management, D = depth, C= change in depth






















































Figure 4.1. Map of study site showing location of 9 study impoundments on the Kennedy
Space Center-Merritt Island National Wildlife Refuge (T10C, T10D, T10E,
T10G, T10H, T10I, T10J, T10K, and T10L). The Wetlands Initiative subset
of impoundments included: T10C, T10D, T10E, T10H, T10J, T10K, and
T10L.









80





































TI I





[1- r
ife^ ^ l- I B IBIB


4.2. Aerial photographs of study impoundments showing sections used to record
locations of foraging wading birds during ground surveys. Impoundments
shown are T1OL and T1OK.

























*

.CI
'It le~ "Uci


Figure 4.3. Aerial photographs of study impoundments showing sections used to record
locations of foraging wading birds during ground surveys. Impoundments
shown are T10H, T10I, and T1OJ.


P' .:






















ME


Figure 4.4. Aerial photographs of study impoundments showing sections used to record
locations of foraging wading birds during ground surveys. Impoundments
shown are T1OC, T10D, T10E, and TO1G.









Results

Twenty-five wading bird ground surveys of the nine study impoundments were

conducted between 11 January and 19 July 2001, yielding a total of 9,781 individuals of

13 species of wading bird (Table 4.2). Three of the species (Cattle Egret, Bulbulcus ibis,

Green Heron, Butorides virescens, and Black-crowned Night Heron, Nycticorax

nycticorax) were sighted very infrequently and were not included in subsequent analyses.

The mean density of wading birds observed during ground surveys was 0.26

individuals/ha and the 95% confidence interval of the mean was (0.20, 0.32). The mean

density varied over time and generally decreased over the survey period (Figure 4.5).

Density within impoundment sections also varied greatly with some sections having

much greater use than others (Figure 4.6). Wading bird density decreased with the

proportion of open water habitat (Figure 4.7) and increased with the proportion of edge

habitat (figure 4.8) within impoundment sections (see Figures 4.2-4.4).

Over 80% of birds observed from dikes were foraging in water at or below their

tarsal joint, but species varied in their propensity to forage in deeper water (Table 4.3).

Foraging habitat use varied by species, but most species used unvegetated wetland habitat

more than vegetated wetland habitat, with Great Blue Heron (Ardea herodias) and Wood

Stork (Mycteria americana) being notable exceptions (Figure 4.9). The proportion of

unvegetated wetland habitat within impoundments varied between 0.06 and 0.99 (Figure

4.10). Between 7 to 28% of the observations of foraging wading birds were within 0.5 m

of the interface between unvegetated open water and vegetated habitat (Figure 4.11). All

species greatly preferred edge habitat (habitat within 0.5 m of the interface between

vegetated and unvegetated habitat) to both unvegetated and vegetated habitats (Figure

4.12, Table 4.4). All species except Great Blue Heron and Wood Stork also preferred