The reproductive and foraging ecology of wading birds (Ciconiiformes) at Lake Okeechobee, Florida

MISSING IMAGE

Material Information

Title:
The reproductive and foraging ecology of wading birds (Ciconiiformes) at Lake Okeechobee, Florida
Physical Description:
xiv, 357 p. : ill., charts ; 28 cm.
Language:
English
Creator:
Smith, Jeffrey Phillip, 1959-
Publication Date:

Subjects

Subjects / Keywords:
Ciconiiformes -- Behavior   ( lcsh )
Birds -- Florida   ( lcsh )
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1994.
Bibliography:
Includes bibliographical references (p. 346-356).
Statement of Responsibility:
by Jeffrey Phillip Smith.
General Note:
Biographical sketch: p. 357.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 002018878
notis - AKK6314
oclc - 31348674
System ID:
AA00003638:00001

Full Text













THE REPRODUCTIVE AND FORAGING ECOLOGY OF
WADING BIRDS (CICONIIFORMES) AT LAKE OKEECHOBEE, FLORIDA









By

JEFFREY PHILLIP SMITH


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


1994













ACKNOWLEDGMENTS


I am first and foremost indebted to my graduate committee chair Michael

Collopy for selecting me to administer this project and for providing guidance and

logistic support throughout the study. Financial support for my research was provided

through the 5.5-year Lake Okeechobee Ecosystem Study contract and a graduate

research assistantship funded by the South Florida Water Management District, West

Palm Beach, Florida. Seasonal field assistants John Humphrey, Todd Morris, Michael

Plotkin and Chris Goguen, laboratory assistants Robin Sansom, Anthony Novack, Lydia

Flewelling, Kim Bjorgo, Darryl Sweetay and Gustavo Hernandez, project members Tim

Harris and Kent Williges, and university associate Howard Jelks all provided essential

field and laboratory support during the project. James Wyatt Enterprises of

Homestead, Florida, and American Aviation, Inc. of Pompano Beach, Florida, provided

aircraft services. I am particularly indebted to John Richardson for providing guidance

throughout the study in the area of GIS and computer support. Early discussions with

university associate Howard Jelks and with Wayne Hoffman, Tom Bancroft and Rick

Sawicki of the National Audubon Society, Tavernier, Florida, helped in the development

of the study. Graduate committee member Peter Frederick provided critical guidance

throughout the study with regards to wading bird ecology, and filled an important role

as committee chairman during the final year of the study. Graduate committee member

Kenneth Portier provided essential statistical guidance. Discussions with university

associate Rob Bennetts provided additional statistical guidance. Graduate committee

member Carole Mclvor provided guidance and support throughout the study and filled

ii







an important role as committee cochair from 1992-1993. I am also indebted to

committee member Wiley Kitchens for his participation and to the other staff of the

Florida Cooperative Fish and Wildlife Research Unit for logistic support during the

study. Final committee member Mark Brown also contributed to the development of the

study and provided specific guidance in the area of population modeling. I would also

like to extend thanks to the following people for their constructive reviews of early drafts

of various chapters: Nick Aumen, Susan Gray, Pete David, Chuck Hanlon and Karl

Havens of the South Florida Water Management District, and Malcolm Coulter and Tom

Sabine.

Lastly, I wish to extend special thanks to my wife Ruthe Smith and my parents

Phillip and Julie Smith for their constant love and support.














TABLE OF CONTENTS

page

ACKNOW LEDGMENTS ...................................... ii

LIST O F TABLES ........................................... vii

LIST O F FIGURES .......................................... ix

A BSTRACT ............................................... xiii

CHAPTERS

1 PROJECT INTRODUCTION .......................... 1

Impetus For the Lake Okeechobee Ecosystem Study ...... 1
Wading Birds in Southern Florida and at Lake Okeechobee .. 3
Wading Bird Project Objectives ....................... 9
Dissertation Format and Content ...................... 13

2 FORAGING HABITAT SELECTION IN RELATION TO
HYDROLOGY AND VEGETATIVE COVER .............. 14

Introduction ...................................... 14
M ethods ......................................... 18
Study Area and Subjects .......................... 18
Survey Design .................................. 20
Hydrology, Vegetation, and Analyses of Habitat
Selection ................................... 23
Results and Initial Discussion ......................... 27
Abundance of Foraging Birds and Hydrologic Trends .... 27
Patterns of Distribution and Hydrologic Trends ......... 34
Vegetation Maps and Changes During the Study ....... 40
Habitat Selection Analyses ........................ 50
General Discussion and Conclusions ................... 64
Analytical Caveats ............................... 64
Foraging Habitat Selection ........................ 67
Management Implications ......................... 74







3 FORAGING-FLIGHT DYNAMICS AND PATTERNS OF
HABITAT USE WITH RELATION TO NEST PRODUCTIVITY,
AND FORAGING SOCIABILITY AMONG FOUR NESTING
SPECIES ........................................ 82

Introduction ....................................... 82
Study Area and Methods ........................... 88
Study Area ....................... .... ........... 88
Field M ethods .................................. 89
Analytical Methods .............................. 90
Results .......................................... 95
Foraging Flight Dynamics and Patterns of Habitat Use ... 95
Foraging Sociability .............................. 122
Discussion ....................................... 126
Foraging-Flight Dynamics, Patterns of Habitat Use, .....
and Nest Productivity ......................... 126
Foraging Sociability .............................. 133

4 COLONY DYNAMICS, NEST SUCCESS AND
PRODUCTIVITY, AND CAUSES OF NEST FAILURE ...... 139

Introduction ....................................... 139
M ethods ......................................... 141
Study Species .................................. 141
Colony Locations and Composition ................. 141
Nest Monitoring ................................ 142
Nest Success and Productivity ..................... 144
Identifying Causes of Nest Failure ................... .147
Results and Discussion .............................. 148
Colony Establishment and Success ................. 148
Individual-Nest Success and Productivity .............. 158
Causes of Nest Failure ................ ...... ..180
Conclusions and Management Implications .............. 191

5 ENVIRONMENTAL DETERMINANTS OF NEST FAILURE:
LOGISTIC REGRESSION ANALYSES ................. 197

Introduction ........................................ 197
Study Area and Methods ............................ 201
Study Area .................................... 201
Nest Monitoring ..................................... 202
Datasets and Regression Analyses .................. 205
Results ................... ................ ...... 215
Summary of Nesting Activity and Climatic and
Hydrologic Trends ............................ 215
Regression Results .............................. 219
Discussion ........................................ 248
Model-Building Strategy ........................... 248
Date of Nest Initiation Effects ...................... 250
Nest Height and Cover Effects ..................... 251







Temperature Effects ............................. 253
Hydrologic Effects ............................... 254
Conclusion ....................................... 260

6 AN ENERGY-CIRCUIT POPULATION MODEL FOR
GREAT EGRETS .................................. 264

Introduction ....................................... 264
Study Area ....................................... 269
Model Formulation and Quantification ................... 269
Prey Population Dynamics ......................... 274
Prey Consumption by Wading Birds ................. 279
Great Egret Immigration and Emigration .............. 282
Recruitment of Breeding Birds and Reproductive
Dynam ics ................................... 289
Simulation Results and Discussion ..................... .294
Prey Population Dynamics ......................... 294
Foraging-Bird Population Dynamics ................. 299
Recruitment of Breeding Birds ................... .. 303
Reproductive Success and Productivity .............. 307
Conclusion ....................................... 311

7 PROJECT SUMMARY ............................... 314

Species Studied ................................. 314
Results and Observations ......................... 315
Management Recommendations .................... 329

APPENDICES

A FORAGING WADING BIRD POPULATION ESTIMATES BY
SPECIES AND SURVEY ............................ 331

B RATE COEFFICIENT CALCULATIONS FOR GREAT
EGRET POPULATION MODEL ....................... 343

REFERENCES ............................................. 346

BIOGRAPHICAL SKETCH .................................... 357













LIST OF TABLES


Table page

2-1 Vegetation types represented on annual, 9.15-m x 9.15-m
resolution, classified satellite-image maps produced by
Richardson and Harris (1994). Cover statistics for these classes
formed the basis for the 1-km x 1-km resolution classification
scheme developed to match the wading bird survey grid. 41
2-2 Identification codes and composition of vegetation community
classes assigned to 1-km x 1-km survey grid cells for habitat
selection analyses. 42
3-1 Foraging flight sample sizes (total number of birds followed to
landing) by species, year, and nesting colony. 96
3-2 Chi-square tests for each species comparing annual proportions
of foraging flights that ended on and off the lake. 98
3-3 Numbers of on and off-lake flights in relation to mean lake stage
(m NGVD), with data pooled across years. 99
3-4 Analyses of variance of flight distances (km) versus year for each
species, with medians and ranges of flight distances shown for
comparison. 100
3-5 Chi-square analysis for Great Egrets examining patterns of
habitat use in relation to mean lake stage (m NGVD), with data
pooled across years. 102
3-6 Annual mean foraging flight distances compared to lakeside
estimates of nestling-period success and nestling production by
species. 105
3-7 Snowy Egret habitat selection in relation to mean lake stage (m
NGVD), with data pooled across years. 111
3-8 Tricolored Heron habitat selection in relation to mean lake stage
(m NGVD), with data pooled across years. 116
3-9 White Ibis habitat selection in relation to mean lake stage (m
NGVD), with data pooled across years. 120
3-10 Chi-square tests examining species' tendencies to travel in
groups to foraging locations. 123
3-11 Chi-square tests examining species' tendencies to join other birds
at foraging sites, and an ANOVA examining species' differences
with regards to the size of the feeding flocks each tended to join. 125
vii







3-12 Nesting adult mean and maximum foraging-flight distances
derived from various studies. 127
4-1 Interannual nesting colony turnover rates: 1989-1992. 153
4-2 Comparison of annual peak nest and species counts at primary
interior-marsh and island colony sites. 154
4-3 Proportion of successful, failed, aborted (no eggs laid), and
uncertain-fate nesting colony initiations by species and year. 157
4-4 Peak-total and marked nest counts, and nest success and
productivity statistics by year for Great Blue Herons. 159
4-5 Peak-total and marked nest counts, and nest success and
productivity statistics by year for Great Egrets. 162
4-6 Hatched-nest survival in relation to nestling age: examination of
bias associated with using threshold ages to estimate nest
success. 165
4-7 Peak-total and marked nest counts, and nest success and
productivity statistics by year for Snowy Egrets. 168
4-8 Peak-total and marked nest counts, and nest success and
productivity statistics by year for Tricolored Herons. 169
4-9 Peak-total and marked nest counts, and nest success and
productivity statistics by year for Little Blue Herons. 174
4-10 Peak-total and marked nest counts, and nest success and
productivity statistics by year for White Ibises. 177
4-11 The fate of marked nests and causes of nest failure by species:
1989-1991. 181
5-1 Nest sample sizes considered in logistic regression analyses by
species, year, and colony. 204
5-2 Summary of independent variables and models analyzed. 209
5-3 Summary of variable and reference-model selection and
verification processes. 213
5-4 Logistic regression models identifying environmental
determinants of nest failure for each species. 220
5-5 Means and ranges of independent variables included in the full-
cycle or nestling models for each species. 224
5-6 Example of procedure for interpreting relationships when several
interactions involved; example is for interpreting effects of
minimum temperature on Great Egret nest success based on full-
cycle model. 230
6-1 Comparisons of observed nest production statistics (Chapter 4)
and simulated responses with and without forced adjustments to
annual mean clutch sizes, 1991 and 1992 egg survival rates, and
the 1990 food supply. 310













LIST OF FIGURES

Figure page

2-1 A geographic overview of the Lake Okeechobee area. 16
2-2 Aerial survey program used to document the distribution and
abundance of foraging wading birds. 21
2-3 Comparison of daily lake stage records from open-water and
interior-marsh gauging stations, shown in relation to estimates of
average daily rainfall for the Lake Okeechobee basin: January
1988 September 1992. 25
2-4 Combined-species wading bird population estimates, excluding
birds at nests, in relation to daily lake stage: August 1988 -
August 1992. 28
2-5 Population estimates for White Ibises, Great Egrets, and Small
White Ardeids (i.e., Snowy Egrets and some immature Little Blue
Herons), excluding birds at nests: August 1988 August 1992. 28
2-6 Population estimates for Glossy Ibises, Great Blue Herons, and
Small Dark Ardeids (i.e., Tricolored and adult Little Blue Herons),
excluding birds at nests: August 1988 August 1992. 29
2-7 Population estimates for Wood Storks, Roseate Spoonbills, and
Great White Herons: August 1988 August 1992. 29
2-8 Non-linear regression of combined-species wading bird
population estimates versus a) daily lake stage and b) 90-day
antecedent surface-water drying rate. 32
2-9 Select 1989 combined-species wading bird density maps for the
transect and Eagle Bay Island areas, shown in relation to plots of
mean water depth derived from the OKEEHYDRO hydrologic
model (Richardson and Hamouda 1994). 35
2-10 Select 1990 combined-species wading bird density maps for the
transect and Eagle Bay Island areas, shown in relation to plots of
mean water depth derived from the OKEEHYDRO hydrologic
model (Richardson and Hamouda 1994). 36
2-11 Select 1991 combined-species wading bird density maps for the
transect and Eagle Bay Island areas, shown in relation to plots of
mean water depth derived from the OKEEHYDRO hydrologic
model (Richardson and Hamouda 1994). 37







2-12 Select 1992 combined-species wading bird density maps for the
transect and Eagle Bay Island areas, shown in relation to plots of
mean water depth derived from the OKEEHYDRO hydrologic
model (Richardson and Hamouda 1994). 38
2-13 One km2-scale vegetative cover maps for 1989 and 1990 derived
from 9.15-m x 9.15-m resolution, classified-satellite-image maps
produced by Richardson and Harris (1994). 44
2-14 One km2-scale vegetative cover maps for 1991 and 1992 derived
from 9.15-m x 9.15-m resolution, classified-satellite-image maps
produced by Richardson and Harris (1994). 46
2-15 Wading bird foraging habitat selection in relation to vegetation
assemblages in 1989: Chi-square tests of independence with bird
densities summed across surveys. 53
2-16 Wading bird foraging habitat selection in relation to vegetation
assemblages in 1990: Chi-square tests of independence with bird
densities summed across surveys. 54
2-17 Wading bird foraging habitat selection in relation to vegetation
assemblages in 1991: Chi-square tests of independence with bird
densities summed across surveys. 55
2-18 Wading bird foraging habitat selection in relation to vegetation
assemblages in 1992: Chi-square tests of independence with bird
densities summed across surveys. 56
3-1 A geographic overview of the Lake Okeechobee area showing
the locations of nesting colonies used by the four study species,
and showing the locations of lake stage and rainfall gauging
stations. 84
3-2 Daily lake stage and total daily rainfall from January 1989 through
August 1992. 86
3-3 Annual distributions of nesting Great Egret foraging flights: 1989-
1992. 97
3-4 Plots by species of colony-specific estimates of nestling-period
success and nestling production in hatched nests versus mean
foraging-flight distances. 106
3-5 Plots by species of the percentages of marked nests that failed--
summed over 15-day periods--versus mean foraging-flight
distances. 108
3-6 Annual distributions of nesting Snowy Egret foraging flights:
1989-1992. 109
3-7 Annual distributions of nesting Tricolored Heron foraging flights:
1989-1991. 114
3-8 Annual distributions of nesting White Ibis foraging flights: 1989-
1992. 118







4-1 Nesting sites involving at least two nests or courting pairs: 1989-
1992. 143
4-2 On-lake nest counts by species in relation to lake stage: 1989-
1992. 150
4-3 Timing of nest failures, 1989-1991, in relation to trends in daily
lake stage, rainfall, and minimum air temperature. 187
5-1 A geographic overview of the Lake Okeechobee area showing
locations of nesting colonies and lake stage and rainfall gauging
stations. 203
5-2 Timing of nest failures, 1989-1991, in relation to trends in daily
lake stage, rainfall, and minimum air temperature. 216
5-3 Example of A Pearson X regression diagnostics plot for the
Great Blue Heron full-cycle model. 227
6-1 Estimated numbers of a) foraging Great Egrets and b) active
Great Egret nests at Lake Okeechobee from December 1987
through August 1992. 267
6-2 Great Egret nest success and productivity: 1989-1992. 268
6-3 Energy circuit diagram of Great Egret population model. 270
6-4 Five-year mean weekly solar insolation at Belle Glade, Florida. 276
6-5 Function used in the model to calculate the total surface-water
area (WAREA) in the transect survey region from estimates of
daily lake stage (STG). 277
6-6 Functional response of prey production (J1) to variation in relative
surface-water area (WA). 278
6-7 Functional response of Great Egret prey consumption rate (J22)
to variation in minimum daily air temperature (MINT). 281
6-8 Functional response of Great Egret immigration rate (J22) to
variation in prey biomass density (DENS). 284
6-9 Functional response of Great Egret immigration rate (J22) to
variation in daily lake stage (STG). 285
6-10 Functional response of Great Egret immigration rate (J22) to
variation in the number of foraging birds already at the lake (NB). 286
6-11 Functional response of Great Egret emigration rate (J21) to
variation in prey biomass density (DENS). 287
6-12 Functional response of Great Egret emigration rate (J21) to
variation in the surface-water drying rate (DR). 288
6-13 Functional response of Great Egret breeder recruitment rate (J8)
to change of season (MON). Response is truncated as shown by
logical functions in model. 290
6-14 Functional response of Great Egret breeder recruitment rate (J8)
to variation in daily lake stage (STG). 291







6-15 Standard simulation output of total prey biomass (BP) and prey
biomass density (DENS) in relation to primary forcing functions:
SUN proportion of maximum mean weekly solar insolation; STG
daily lake stage; WAREA total surface-water area in primary
survey region. 296
6-16 Simulation output of total prey biomass (BP) and prey biomass
density (DENS) illustrating effect of linear surface-water area
factor (WAREA or WA) on prey production (J1). 297
6-17 Monthly mean biomass density of small forage fishes (<100 mm
SL) in four macrophyte types in the west-central littoral zone of
Lake Okeechobee from December 1989 through December
1990. 298
6-18 Standard simulation output of foraging bird numbers (NB) in
relation to prey biomass density (DENS) and lake stage (STG). 301
6-19 Simulation output of the number of foraging birds (NB) illustrating
effect of modeling immigration (J22) and emigration (J21) rates
as functions of only prey biomass density (DENS). Plot of lake
stage included for reference. 302
6-20 Standard simulation output of number of breeding Great Egrets
(NBR), nestlings (NN), fledglings/first-year subadults (NF1),
second-year subadults (NF2), and third-year subadults (NF3) in
relation to lake stage (STG), prey biomass density (DENS), and
the number of foraging birds (NB). 304













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

THE REPRODUCTIVE AND FORAGING ECOLOGY OF
WADING BIRDS (CICONIIFORMES) AT LAKE OKEECHOBEE, FLORIDA

By
Jeffrey Phillip Smith

August 1994


Chairman: Michael W. Collopy
Major Department: Forest Resources and Conservation (Wildlife and Range Sciences)

I documented the distribution and abundance of nine species of wading birds at

Lake Okeechobee, Florida, from August 1988 through August 1992. The study period

spanned a two-year drought. The highest diversity and numbers of foraging birds

occurred during peak drought periods, in part indicating that the lake serves as a

regional drought refuge. Regulation of lake water levels is a priority management

issue. Lake stage set the upper and lower limits to accessible habitat; otherwise,

receding water was a strong attractor to foraging birds. Habitat selection analyses

indicated a general preference for wet-prairie habitats of intermediate stature and

structural diversity, but all species selected prey-rich beds of submerged vegetation at

low lake stages.

I documented the history of nesting from 1989-1992 through aerial surveys,

studied the nest productivity of six species in select colonies, and documented the

foraging habits of nesting adults of four species through aerial following flights. Interior-

marsh colony sites were abandoned in favor of island sites during the drought. The






widest variety of sites was used during 1992 with high lake stages, but this usually did

not translate to higher nesting success. Great Blue Herons (Ardea herodias) and Great

Egrets (Casmerodius albus) initiated more nests in high-water years. Little Blue Herons

(Eqretta caerulea) initiated the most nests in 1989 following a multi-year period of high

water. Snowy Egrets ( thula), Tricolored Herons (E. tricolor), and White Ibises

(Eudocimus albus) initiated the most nests during the peak drought year of 1990 and

the fewest in 1991 with moderate but stable water levels. All species achieved high

nest success and productivity in 1990 and the lowest levels either in 1991 or in 1992

with high water. Thus, drought-related declines in the lake stage enhanced aspects of

all species' nest productivity, but species' responses differed in subtle ways. Logistic

regression analyses confirmed a high degree of complexity with regard to the

environmental determinants of nest failure. I programmed an energy-circuit population

model for Great Egrets to conceptually organize data and further study relationships

between environmental forcing functions and the foraging and nesting ecology of one

common species.












CHAPTER 1
PROJECT INTRODUCTION

Impetus For the Lake Okeechobee Ecosystem Study


In April 1988, the South Florida Water Management District (SFWMD)

formalized a 5.5-year contract with a diverse team of researchers from the University of

Florida and the Florida Game and Fresh Water Fish Commission (FGFWFC) to

address the task of establishing a comprehensive database on the ecological dynamics

of Lake Okeechobee (Aumen 1994). Lake Okeechobee is the third largest freshwater

lake within the United States (1,732 km2 surface area; Herdendorf 1982). It drains a

12,000 km2 watershed primarily through the Kissimmee River from the northwest, and

is in turn the headwaters of the Everglades and Florida Bay to the south. Besides

supplying water to the Everglades, the lake is the primary source of water for a

burgeoning urban population and for the extensive Everglades Agricultural Area (EAA)

situated just south of the lake. It serves as a regional flood-control facility and supports

a multi-million dollar commercial and sport fishing industry. It supports a large nesting

population of the endangered Snail Kite (Rostrhamus sociabilis; Sykes 1979; Snyder et

al. 1989) and a large population of once-endangered but now recovering alligators

(Alligator mississippiensis; Mazzotti and Brandt 1994). It also provides important

foraging and nesting habitat for most of the species of wading birds that occur in

Florida (Zaffke 1984; David 1994a, b; Chapters 2, 4), often including the endangered

Wood Stork (Mycteria americana; Browder 1976; Chapter 2). The demands of these

diverse interests often conflict, and consequently management of the lake is a subject

1




2


of considerable debate. Public concern over the health of the lake peaked in 1986 with

the occurrence of unusually massive blue-green algal blooms (Fox 1987). This added

to growing frustration over waterways choked with exotic plant species such as Hydrilla

verticillata and water hyacinth (Eichhornia crassipes), and the expansion of other

nuisance species such as cattail (Typha sp.) and the exotic Melaleuca quinquenervia.

Beginning in 1973, researchers measured phosphorous loading rates in excess

of critical values established by Vollenweider (1968) for eutrophic lakes, and recently

established that total phosphorous concentrations in the pelagic zone of the lake had

doubled since 1973 (see Aumen 1994 for a review). The appearance of previously

undocumented, large scale algal blooms and the associated negative side effects--

including fish kills, foul odor and taste of lake water, and public outcry-summoned new

concern among managers within the SFWMD over the potential relationship between

lake eutrophication and the prevalence of algal blooms and other invading plants. The

long-term ecological consequences of the proliferating blooms are unknown and the

ability to monitor present and future impacts was limited due to a paucity of information

on the ecology of the lake. Residents, scientists, state officials, and managers were

particularly concerned over the possible negative impact on the trophic structure of the

lake. Fish kills related to the toxicity of algal blooms understandably elicited

widespread concern and precipitated a call for immediate action to curb inputs of

nutrients believed to be contributing to the blooms. Bearing the brunt of criticisms in

this regard were the large dairy farms located around the north end of the lake and the

sugar cane industry in the EAA. However, without specific and detailed data to support

claims, assessing the feasibility and effectiveness of proposed plans of action was at

best an uncertain task. Consequently, the SFVMD solicited the participation of

university and FGFWFC researchers to establish a baseline ecological dataset on







which future management strategies and additional specific research plans could be

based.

I became involved in the ensuing Lake Okeechobee Ecosystem Study as a

graduate student, with the responsibility of implementing a comprehensive study of the

reproductive and foraging ecology of wading birds at the lake. Other members of the

research team were responsible for studying the hydrologic and vegetation dynamics of

the ecosystem, water quality and nutrient dynamics, phytoplankton and zooplankton

community dynamics, benthic invertebrate communities, and fish community dynamics.

Most of this work will be summarized in a special Lake Okeechobee edition of Archiv fur

Hydrobiologie, to be published in late 1994 or early 1995. Field work on the project

began in May 1988 and ended in December 1992. I began surveying foraging wading

birds in August 1988. Following the preliminary 1988 fall survey season, I focused my

research efforts on the typical wading bird nesting season, January through August,

and completed four seven-month field seasons from 1989-1992. Several of the

analyses I subsequently conducted relied on the diverse datasets prepared by my

colleagues on the project-specifically the output from a hydrologic model and

vegetation maps prepared by J. Richardson and his coworkers, additional vegetation

data collected by P. Zimba and his coworkers, and data on prey populations gathered

by J. Chick and C. Mclvor, and by L. Bull and his coworkers.


Wading Birds in Southern Florida and at Lake Okeechobee


The plight of Lake Okeechobee and the potential threat of large-scale algal

blooms has only recently garnered significant attention. However, the decline of

wading bird populations in southern Florida has received considerable attention. Early

in the century, the plume-hunting trade had a severe impact on regional populations,







but since then ill-advised hydrologic manipulations and habitat degradation have played

the major role in nesting population declines as high as 90% since the 1940's (Kushlan

and White 1977; Ogden 1978a, 1994; Kushlan et al. 1984; Kushlan 1986a; Kushlan

and Frohring 1986; Walters et al. 1992; Bancroft et al. 1994; Frederick and Spalding

1994). Several recent studies have addressed the relationship of habitat availability,

hydrology, and wading bird reproductive success in various regions of southern Florida

(Kushlan et al. 1975; Kushlan 1976c, 1986a; Browder 1976; Ogden et al. 1978, 1980;

Powell 1987; Frederick and Collopy 1989a; Frederick et al. 1992; Frederick and

Spalding 1994; Bancroft et al. 1994; Ogden 1994). These and other continuing efforts

are building a solid base of information that will provide Florida managers with the

knowledge necessary for managing and maintaining healthy wading bird populations in

southern Florida. Previously, only general information was available concerning the

population dynamics of wading birds at Lake Okeechobee (Zaffke 1984; David 1994a,

b), and therefore my work filled an important gap in the regional knowledge base. In

particular, I was afforded the opportunity to augment scarce information concerning the

ecological dynamics of wading birds in a naturally occurring--albeit highly regulated--

southern Florida lacustrine ecosystem.

Wading birds comprise a particularly conspicuous component of the southern

Florida fauna. Their colonial breeding habits facilitate the study of their reproductive

efforts, and the large size and white color of several of the major species facilitates

monitoring populations through aerial surveys. Colonial-nesting wading birds are

sensitive to changes in their nesting and foraging habitat (Kerns and Howe 1967;

Kushlan 1974, 1976c; Wiese 1978; Ogden et al. 1980; Bancroft et al. 1994; Frederick

and Spalding 1994; Ogden 1994). Wading bird distribution, diversity, abundance, and

reproductive performance all reflect the characteristics and distribution of available prey







populations (Owen 1960; Dusi and Dusi 1968; Kushlan et al. 1975; Rodgers and

Nesbitt 1979; Kushlan 1976b, c, 1978a, 1981, 1986a; Browder 1976; Bancroft et al.

1994; Frederick and Spalding 1994; Ogden 1994). These links to other less

conspicuous elements in the food web of ecosystems like Lake Okeechobee, and the

high visibility of wading birds, has led many authors to conclude that wading birds are

reliable bioindicators of change in hydrological and ecological conditions in wetland

ecosystems (e.g., Custer and Osborn 1977; Curry-Lindahl 1978; Hafner and Britton

1983; Bildstein et al. 1990; Kushlan 1986b, 1993; Ogden 1994).

Lake Okeechobee has consistently supported several large, mixed-species

wading bird nesting colonies each year (Ogden 1978a; Osborn and Custer 1978;

Nesbitt et al. 1982; Zaffke 1984; David 1994a; Chapter 4). The aquatic-feeding

species that commonly nest in relatively large numbers include Great Blue Heron

(Ardea herodias), Great Egret (Casmerodius albus), Snowy Egret (Eqretta thula),

Tricolored Heron (E. tricolor), Little Blue Heron E caerulea), and White Ibis

(Eudocimus albus). Black-crowned Night Herons (Ncticorax ncticorax), Green-

backed Herons (Butorides striatus), and Glossy Ibises (Plegadis falcinellus) also nest at

the lake in small numbers, and Wood Storks have occasionally nested at or near the

lake in the past. The primarily terrestrial-feeding Cattle Egret (Bubulcus ibis) also nests

in large numbers at the lake.

The success and productivity of wading bird nesting efforts often depends on

favorable hydrologic conditions (Frederick and Spalding 1994). Colonies on Lake

Okeechobee and in marshlands typical of the Everglades are usually situated in areas

surrounded by water, probably to discourage invasion by mammalian predators (Jenni

1969; Frederick and Collopy 1994). However, the willows that wading birds at Lake

Okeechobee generally nest upon are weakened by excessive inundation (Zaffke 1984;







David 1994a), and willow reproduction is dependent on reduced water levels (Pesnell

and Brown 1977). David (1994a) documented a general decline in wading bird nesting

populations at Lake Okeechobee following the adoption of a higher lake-stage

regulation schedule in 1978. In particular, the King's Bar colony site, which had

previously been the largest perennial colony on the lake, was abandoned and has not

been used since. However, Loftus et al. (1990) and Loftus and Eklund (1994) provide

evidence that long-hydroperiod marshes generally support denser populations of

forage fishes and macroinvertebrates. Similarly, Bildstein et al. (1990) found that White

Ibises nesting in coastal colonies in South Carolina depended on high winter rainfall to

encourage high crayfish (Procambarus sp.) productivity in nearby freshwater marshes.

The adults often fed in salt marsh areas, but were forced to rely on the freshwater

marshes to provide food for their nestlings, because the young birds cannot tolerate the

high salt content of marine prey. Several studies have also indicated that wading birds

often capitalize on surface-water recessions that concentrate prey (Kushlan et al. 1975;

Kushlan 1976b; Browder 1976; Kushlan and Hunt 1979; Frederick and Spalding 1994),

and nesting adults may depend on receding water levels during the breeding season to

concentrate prey sufficiently to satisfy the needs of their growing young (Kahl 1964;

Kushlan et al. 1975; Clark 1978). However, Ogden (1994) notes that Wood Storks may

have only recently developed a strict dependence on the concentrating effects of

surface-water recessions. He suggests that hydrologic manipulations have significantly

reduced the prevalence of long-hydroperiod marshes, which has in turn led to a

substantial reduction in the overall productivity of the Everglades. The reduction in

productivity has meant that the storks must rely on the prey concentrating effects of

recessions to succeed at all, whereas prior to widespread drainage such events may

simply have helped to increase productivity to high levels. Such sensitivities render







wading birds ideal subjects for studying the effects of changing water levels at Lake

Okeechobee.

Hydroperiod and water depths are, however, not the only potentially

manageable factors that influence the ability of wading birds to exploit prey populations

and reproduce successfully. Vegetation structure, for example, is likely of critical

importance in determining both the diversity and abundance of prey (e.g., see Chick

and Mclvor 1994) and whether a bird can successfully access available prey. Sufficient

vegetative cover must be present to provide safe habitat for small fish and

invertebrates, but the birds must be able to both move about readily and visually detect

approaching predators. Thick stands of tall grass or species such as cattail and

sawgrass (Cladium iamaicense) probably physically impede access by most wading

birds (see Hoffman et al. 1994). Cattail has dramatically expanded in abundance in

areas subject to artificial eutrophication and is now considered a nuisance in many

areas (Davis 1989; Richardson and Harris 1994; Richardson et al. 1994). In contrast,

small forage fishes are generally scarce in very open habitats (Werner et al. 1977), and

those that are present are likely difficult to capture. Thus, habitats featuring

appropriate water depths and vegetation of moderate stature and structural diversity

should provide the best foraging opportunities. Nutrient-rich canals, impoundments and

bays often support extremely dense mats of water hyacinth and water lettuce (Pistia

stratiotes) or very dense stands of Hydrilla intertwined with filamentous algae, to the

point that fish and thus foraging wading birds might be excluded. However, moderate

to high densities of Hydrilla unencumbered with thick filamentous algae often harbor

high concentrations of small fishes (Mclvor and Smith 1992; Chick and Mclvor 1994),

and thick surface mats of Hydrilla and floating mats of water hyacinth may provide the

means for wading birds to access habitats too deep for wading (Chapter 2; and see







Murdich 1978). Inundation levels, nutrient inputs, and natural events such as freezes

and fires all influence the distribution and composition of vegetation communities at the

lake (Richardson et al. 1994), and therefore also influence the productivity of wading

birds.

Lake Okeechobee is almost entirely surrounded by an earthen dike, and most

water flows into and out of the lake are regulated by locks, hurricane gates, and

pumping stations. Thus, the lake is now largely isolated from adjacent wetlands.

Moreover, as mentioned previously, most of what used to be an extensive sawgrass

marsh south of the lake is now the EAA, which is covered with sugar cane and rice

fields and citrus orchards. In contrast, much of the area north and west of the lake has

been converted to dairy farms and cattle pastures. Consequently, there is a stark

contrast between the habitats available inside and outside the diked confines of the

lake. Wading birds that nest on the lake can easily access both on- and off-lake

habitats, and therefore management plans developed for the lake should recognize the

potential influence of these alternative habitats. One issue of potentially great import

concerns the possibility that highly disturbed and nutrient-enriched habitats such as

agricultural ditches and cattle ponds may support elevated populations of the

oligochaetes that serve as the first intermediate host for the nematode parasite

Eustrongylides ignotus (Spalding et al. 1993; Frederick and Spalding 1994). This

parasite has caused epizootic mortality among wading bird nestlings in the Everglades

and elsewhere (Wiese 1977; Roffe 1988; Spalding et al. 1993), and reduced the

survival and growth rates of Great Egret nestlings at Lake Okeechobee during my study

(Spalding et al. 1994).







Wading Bird Project Objectives


With the above issues and concerns in mind, the primary goals of my research

were as follows:

1) Begin to elucidate how habitat structure, hydrology, climate, and variation in the

distribution and abundance of prey influence the temporal and spatial

distribution, and nesting and foraging ecology of wading birds at the lake.

2) Assist in the development of conceptual models of the Lake Okeechobee

ecosystem, specifically focusing on the interactions of biotic components and

hydrology.

To achieve these goals, the major procedural objectives of my research were as

follows:

1) Document through regular aerial surveys the foraging distribution and abundance

of all common species of wading birds throughout the littoral zone of the lake

and in select perennial wetlands adjacent to the lake.

2) Document through aerial surveys the location, composition, and history of all

nesting colonies on and in close proximity to the lake.

3) Conduct intensive studies of marked nests in select, geographically dispersed

colonies on the lake to document the reproductive success and productivity of

all common species.

4) Monitor nestling diet through opportunistic collection of regurgitant samples

during colony checks.

5) Document nestling growth rates for focal species by gathering measurements

from representative samples of nests in each colony with marked nests.







6) Document through aerial following flights the foraging habits of nesting adults

during the brood-rearing phase.

7) Observe foraging adults primarily in habitats corresponding to those chosen by

nesting adults-identified through following flights (Objective 6)-to quantify

foraging success and prey choice.

8) Follow observations of foraging birds (Objective 7) with prey sampling to quantify

prey abundance and confirm prey species representation.

9) Integrate the above data with hydrologic and climatic datasets and annual

vegetation maps to identify the environmental conditions that lead to successful

foraging and high reproductive productivity among wading birds at the lake.

10) Program a population energetic model to conceptually integrate the various

aspects of wading bird ecology at the lake and to simulate the responses of

wading birds to fluctuating environmental conditions.

The primary working hypotheses on which I based my research program

included the following:

Distribution and Abundance, and Foraging Ecology:

1) Patterns of abundance and distribution are primarily a function of hydrology.

a) The stage level sets the ultimate spatial boundaries to foraging.

b) The magnitude and duration of drying trends determine whether prey are

concentrated or dispersed.

c) Hydroperiod and levels of inundation influence the vegetative structure of

habitats.

2) Foraging birds prefer mixes of emergent vegetation of moderate density and

height that provides refuge for prey species, but does not unduly hinder a bird's







progress through the habitat nor restrict its ability to remain vigilante against

predators and other disturbances.

3) Lakewide population trends will in part reflect environmental conditions occurring

elsewhere in Florida and the southeast. In particular, because it is a large

lacustrine system, Lake Okeechobee may constitute a significant regional

refuge during periods of drought.

4) The availability of profitable foraging habitat nearby but outside the diked

boundaries of the lake often will provide a level of independence from on-lake

hydrology.


Nesting Ecology:

1) The timing of nesting activities is a complex function of environmental cues-

including aspects of hydrology (water level, drying rates, rainfall patterns),

climate (temperature, storm frequency and severity), and prey population

dynamics-acting in conjunction with elements of social facilitation.

2) The location of nesting colonies is constrained primarily by the availability of

suitable nesting substrate-e.g., willow heads. The distribution of willow is a

function of hydroperiod, and wading birds chose colony sites surrounded by

water that is deep enough to discourage invasion by mammalian predators.

Therefore, the location of nesting colonies is a function of hydrology, specifically

lake stage and hydroperiod.

3) Colony locations are secondarily constrained in proximity to favorable foraging

grounds; again primarily a function of hydrology and the distribution of

appropriate vegetative cover.

4) The abandonment of entire colonies is primarily a function of stochastic events

such as temperature fluctuations, large-scale wind and rainfall events, and







unexpected shifts in the hydrologic regime that lead to deterioration of foraging

conditions. Excessive predation or disturbance, and disease or parasite

epidemics may also induce abandonment.

5) Mean lakeside whole-nest survival and productivity rates vary between years

primarily as a function of hydrology-lake stage, hydroperiod, and drying rates-

through its influence on the abundance, distribution, and accessibility of prey

resources. However, as with colony turnover rates, other factors such as

predation, disturbance, and disease or parasite outbreaks may significantly

impact overall nest survival rates in certain years.

6) Colony-specific rates of nest success and productivity more closely reflect local

hydrologic conditions and the health of adjacent habitats and prey populations,

as well as unique aspects of the colony site structure and local predation,

disease, and parasite phenomena.

7) Variation in the overall productivity of nesting events is manifest primarily as

changes in the mean survival rate of whole nests, rather than as changes in the

mean productivity of successful nests.

8) Nest failures and abandonments during the incubation phase are more often a

function of excessive disturbance or inclement weather, whereas failures during

the nestling phase are more regularly a function of the abundance and

accessibility of prey resources and secondarily of the prevalence of parasite and

disease outbreaks. Predation events may affect failures during both stages.







Dissertation Format and Content


Beyond this introductory chapter, this dissertation is divided into five main

chapters and a final summary chapter. I composed each of the five main chapters as

stand-alone manuscripts. Therefore, there is some redundancy in the introductory

sections and reference figures. In Chapter 2, I summarize my findings concerning the

distribution and abundance of foraging birds with relation to hydrologic influences, and

discuss an analysis of foraging habitat selection relative to vegetation community types.

In Chapter 3, I summarize my findings concerning the foraging habits of nesting adults,

which I documented through aerial following flights. In Chapter 4, I summarize my

findings concerning the history of nesting and causes of failure during the study, and

discuss general trends in reproductive effort, success, and productivity. In Chapter 5, I

discuss a comprehensive logistic regression analysis I conducted to further elucidate

relationships between nest failure rates and a suite of nest-situation, hydrology, and

climate variables. I then complete the primary presentation with Chapter 6, wherein I

discuss my effort to program a population model for Great Egrets. I do not herein

discuss in detail the results of my efforts to study nestling growth and diet or the

foraging success of birds in different habitats (and the associated prey sampling). I will

cover these topics in future manuscripts.













CHAPTER 2
FORAGING HABITAT SELECTION
IN RELATION TO HYDROLOGY AND VEGETATIVE COVER

Introduction


Wading birds are highly visible, top-level consumers and are considered

bioindicators of change in hydrological and ecological conditions in wetland ecosystems

(e.g., Custer and Osborn 1977; Curry-Lindahl 1978; Hafner and Britton 1983; Bildstein

et al. 1990; Kushlan 1986b, 1993; Ogden 1994). Nesting populations of most species

have declined dramatically over the past few decades (Ogden 1978a, 1994; Kushlan et

al. 1984) most likely due to habitat degradation and ill-advised hydrological

manipulations (Kushlan 1986a; Walters et al. 1992; Bancroft et al. 1994; Frederick and

Spalding 1994; Ogden 1994). In an effort to elucidate the causes of decline,

researchers have paid considerable attention to documenting population trends and the

nesting and foraging ecology of wading birds in Florida Bay and the Everglades region

(Kushlan et al. 1975; Browder 1976; Bancroft and Jewell 1987; Frederick and Collopy

1988, 1989a, b; Bancroft et al. 1990; Hoffman et al. 1994; Frederick et al. 1992;

Frederick and Loftus 1993; Frederick and Spalding 1994). In contrast, Lake

Okeechobee-historically the primary source of water, along with rainfall, for the

Everglades-has been little studied (but see Zaffke 1984; David 1994a, b) yet is the

third largest freshwater lake within the United States (1,732 km2 surface area;

Herdendorf 1982) and supports several large mixed-species wading bird nesting

colonies each year (Chapter 4; David 1994a).







Studies at Lake Okeechobee are likely to reveal unique aspects of wading bird

ecology and unique patterns of response to fluctuating environmental conditions for

several reasons. The lake ecosystem features a large expanse of emergent marsh

(typically about 400 km2) similar to the wet prairie marshes of the Everglades and other

palustrine systems in southern Florida (Fig. 2-1). However, the lake ecosystem differs

from Everglades-type wetlands in that it also features a large, relatively deep and

persistent open-water pool. This difference means that populations of mobile aquatic

organisms-such as the forage fishes, amphibians, and macroinvertebrates that wading

birds feed on-probably can survive during periods of even extreme drought by seeking

refuge along the edges of the central pool. Recolonization and recovery rates should

therefore be more rapid, and for this reason, overall aquatic productivity may be higher

than in strictly palustrine systems (cf. Loftus and Eklund 1994). Moreover, because

large lacustrine systems like Lake Okeechobee include a deep-pool region where prey

can seek refuge, such systems may also serve as critical regional drought refuges for

vagile species such as wading birds. The lake ecosystem also includes extensive beds

of submerged vegetation (Zimba et al. 1994) not found in shallow palustrine systems,

and sampling revealed very high concentrations of forage fish and grass shrimp

(Palaemonetes paludosus) in some of these habitats (Mclvor and Smith 1992; Chick

and Mclvor 1994). Furthermore, because the lake is almost entirely surrounded by a

large earthen dike, there is an abrupt transition from extensive marsh and open-water

habitats inside the dike to a wide variety of natural and artificial wetland habitats

outside the dike. The latter include: the Kissimmee River and Fisheating Creek

floodplains (see Fig. 2-1); diverse pocket and slough wetlands interspersed with cattle

pastures to the north and west; myriad agricultural field ditches and canals to the south

and east; and residential canals and retention ponds in several areas. The lake region












Popash Slough


Taylor Creek


Kissimmee


' ,King's Bar


Indian Prairie


Creek


River


Lak
Hic


Ig Moonshine Bay

igh emergent

aven eeg
Clewiston Sp

e
pochee Clewiston


5 km


canals


/


marsh


A geographic overview of the Lake Okeechobee area.


Slough


* Nesting colony sites
SLake stage gauging stations


Figure 2-1.







thus provides wading birds with an unusually diverse array of foraging opportunities

within a relatively limited area.

The multidisciplinary Lake Okeechobee Ecosystem Study was initiated in 1988

to provide baseline ecological data to the South Florida Water Management District

(SFWMD) and facilitate the development of an ecologically sound management

strategy for the lake. The primary goal of the project was to increase our

understanding of how hydrologic fluctuations and cultural eutrophication influence

water quality and the ecological dynamics of the vegetation and wildlife communities of

the lake. Aumen (1994) discusses the management history of the lake and the impetus

for the study in more detail. My primary responsibility was to determine how lake-stage

regulation schedules and other management practices employed by the U. S. Army

Corps of Engineers and the SFWMD influence the population dynamics of wading birds

at the lake. In this chapter, I present the results of using aerial surveys to document the

distribution and abundance of nine species of foraging wading birds on the lake from

August 1988 through August 1992. Specifically, I sought to elucidate how the

distribution and abundance of foraging wading birds changed in relation to fluctuating

water levels, and to determine whether the birds' use of foraging habitat reflected

preferences for certain vegetation assemblages.

The study period spanned both high water years and a two-year drought that

resulted in a 10-year low lake stage in May 1990. A series of fires and freezes further

contributed to pronounced interannual shifts in the composition and distribution of

vegetation assemblages (Richardson and Harris 1994). To elucidate relationships

between wading bird habitat selection and changes in hydrology and vegetation, I

integrated my wading bird foraging-dispersion data with annual vegetation maps

(Richardson and Harris 1994) and hydrologic data derived from stage recorders and a







hydrologic model (Richardson and Hamouda 1994). I thereby gained knowledge about

species-specific habitat choices and their ecological significance. I gained insight about

the relative importance to wading birds of several species of exotic and nuisance

vegetation that concern resource managers in Florida. I was also able to elucidate the

combinations of hydrology and vegetative structure that attracted the largest numbers

and variety of wading birds to the lake. Along with these studies of foraging dispersion,

I also documented the foraging efficiency and prey selection of individual birds in

different habitats, and studied the nesting ecology of common species through aerial

surveys and ground-based monitoring of marked nests (Chapters 3-5; Collopy and

Smith 1991; Mclvor and Smith 1992). The integration of such information provided the

basis for projecting management strategies that should increase the occurrence of

favorable foraging conditions on the lake and enhance the productivity of wading bird

nesting efforts.


Methods


Study Area and Subjects


A geographic overview of the Lake Okeechobee area is provided in Figure 2-1,

with primary wading bird nesting colonies, hydrologic gauging stations, and key regions,

waterways, towns and other landmarks identified. I will frequently refer to areas and

landmarks depicted on this map without reference to the figure. Aumen (1994)

provides an overview of the physiography and hydrology of the lake. Richardson and

Hamouda (1994), Richardson and Harris (1994), and Richardson et al. (1994) discuss

in detail the hydrology and vegetation community dynamics of the lake.







I tallied the following species separately on all surveys: Great Blue Heron (and

the unique Great White Heron subspecies), Great Egret, Snowy Egret, Tricolored

Heron, Little Blue Heron, White Ibis, Glossy Ibis, Wood Stork, and Roseate Spoonbill.

Cattle Egrets only rarely foraged on the lake and I did not keep track of their numbers.

It is difficult to distinguish immature Little Blue Herons and Snowy Egrets from the air.

Ground-based observations of mixed-species feeding flocks indicated that immature

Little Blue Herons always constituted less than 5% of the total number of small white

herons and egrets in such flocks (unpubl. data). During this study, I never saw more

than three or four immature Little Blue Herons together in one group, and these were

usually with other adult Little Blue Herons. Therefore, I lumped all Snowy Egrets and

unidentified small white ardeids into the category Small White Ardeids (SWA) and

conducted analyses only on the combined group. However, for convenience, and

because I am confident of the comparative rarity of immature Little Blue Herons, I will

refer to results for the SWA group as if they were for Snowy Egrets alone. I

categorized all positively identified immature Little Blue Herons as such, not as SWA.

Tricolored Herons and adult Little Blue Herons are also difficult to distinguish from the

air when in mixed flocks, and many were assigned to the category Small Dark Ardeids

(SDA). In this case, I conducted separate analyses on data for identified individuals

and for the combined group, and compared results before developing interpretations.

Great White Herons, Wood Storks, and Roseate Spoonbills were not common enough

on the lake to be included in statistical analyses of habitat selection.

On all foraging-dispersion surveys, birds were classified as either foraging,

commuting (i.e., obviously in transit, as opposed to birds that flushed as the plane

approached) or roosting (i.e., birds perched in shrubs or trees). I included only foraging

birds in analyses of habitat selection, but included all individuals in population







estimates. I tallied birds attending nests during separate weekly surveys and do not

consider these data here (see Chapter 4).


Survey Design


I employed a three-part aerial survey design to document the distribution and

abundance of foraging wading birds on and adjacent to the lake (Fig. 2-2). The survey

program included: 1) a series of belt transects to sample most of the extensive west-

side littoral region; 2) a standardized but nonsystematic survey route to cover the

eastern margin and the southern islands; and 3) five complete-count block surveys to

cover the Eagle Bay Island and adjacent marsh area at the north end of the lake, and

four other perennial wetlands adjacent to the west side of the lake. I conducted the full

set of surveys over a two-day period, usually transects on the first day and other

surveys on the second. I completed monthly survey sets from August 1988 through

early January 1989. This was the only fall season I sampled, and an initial variation in

the survey design (different lay-out and attendant visibility biases; see Mclvor and

Smith 1992) rendered the data less compatible with those from subsequent surveys.

Therefore, I did not include the early data in analyses of habitat selection, but did

include the data in plots showing trends in lakeside abundance. I adopted the new

"standard" transect survey design in late January 1989, and subsequently completed

survey sets twice per month January through August 1989-1991 (except that I was

unable to begin surveys until March in 1991 due to unforeseen difficulties associated

with renewal of a low-altitude flight permit) and monthly January through August 1992.

The January through August period corresponded to the usual wading bird nesting

season (see Chapter 4). The reduction to monthly surveys in 1992 was a contract-

related compromise that involved budget and time constraints.























& Marsh
Block


Fisheating Creek


Slough Block


East-South
Route


Lake Hicpochee Block


5 km


Figure 2-2. Aerial survey program used to document the distribution and abundance
of foraging birds.







The standard transect surveys included 37 kilometer-wide, east-west transects

that covered most of the western littoral zone (Fig. 2-2). Each survey round began at

the north end with transect #1, continued southward with odd numbered transects and

alternating start points (i.e., east or west end), and then reversed northward to cover

the remaining even numbered transects. This protocol helped ensure (pers. observe )

that flushed birds had time to return to their original foraging spots before an adjacent

transect was surveyed, and thereby reduced the chance of double counting. Transect

surveys were flown between 0700 and 1400 hrs EST in a Cessna 172 fixed-wing

aircraft at an air speed of 80-90 knots (148-167 km h"1) and an altitude of 76 m above

ground level. An aviation-grade LORAN C unit enabled navigation along latitude lines

running down the center of each transect belt. Myself and a second observer (a

different field assistant each year 1989-1991; otherwise a mix of four other trained

observers) sat on opposite sides of the plane and counted all birds observed within 150

m wide strips extending outward from the side of the plane. The count zones were

delineated as described in Norton-Griffiths (1975), and the setup rendered a 30%

sample of each transect belt. I derived final population estimates using Jolly-Seber II

techniques (Norton-Griffiths 1975). Longitude coordinates recorded with each bird or

group of birds observed provided the means for assigning all observations to 1-km x 1-

km grid cells. This enabled the production of GIS distribution maps and analyses of

habitat selection.

The East-South Route survey followed a standardized but unsystematic path

along the eastern and southern margins of the lake. The survey path included regular

diversions over the interior of wet islands and other areas where wading birds

congregated, such that all useable habitat was surveyed and all visible birds were

counted. I was the sole observer, and the flight speed and altitude conformed to







transect survey protocol. I allocated observations to only coarse-scale geographic

regions that were incompatible with the 1 km2 scale of the transect grid. I therefore did

not include these data in statistical analyses of habitat selection.

The block-surveys covered portions of five structurally varied wetland areas

outside the diked boundaries of the lake, but I chose not consider these data in this

dissertation (I intend to produce a separate manuscript summarizing the data). The

northern-most block survey also covered the Eagle Bay Island area within the

boundaries of the lake, which was an important foraging and nesting area. I was the

sole observer on all block surveys, the flight speed and altitude conformed to transect

survey protocol, and I searched each block area until I had counted all visible birds. I

mapped the locations of all birds on overlays of satellite imagery, later assigned each

bird to 1-km x 1-km grid cells, and then integrated the data with the transect-survey GIS

databases.

I generated lakeside population estimates by adding counts from the Eagle Bay

Island and East-South Route surveys to extrapolated estimates derived from the

transect surveys.


Hydrology, Vegetation, and Analyses of Habitat Selection


I sought to derive functions that effectively summarized relationships between

lakeside bird abundance and hydrologic trends. Responses to variation in the mean

lake stage were of particular interest, because this is often the parameter of concern

among managers. Stage data for the lake are geographically limited, particularly for

the emergent marsh zones. I obtained two sets of daily stage measurements from the

SFWMD that spanned the entire study period. The first series was collected at a

gauging station located in the central, open-water portion of the lake (Fig. 2-1). The







second series was collected at a station located in the interior of the emergent marsh

zone east of Moore Haven (Fig. 2-1). The two stage records matched closely at stages

above about 4.3 m NGVD, but diverged at lower stages (Fig. 2-3; NGVD = National

Geodetic Vertical Datum of 1929, essentially equivalent to height above mean sea

level; for convenience, I will henceforth omit the NGVD qualifier). The discrepancy

arises because the emergent marsh zones along the western margin become

hydrologically isolated from the open water portions of the lake at low stages

(Richardson and Hamouda 1994). At lower stages, local rainfall patterns determine

water levels in the upper-elevation marshes, and short-term fluctuations are

pronounced. Conversely, the open-water stage is a function of watershed-scale rainfall

patterns and to a lesser degree artificial manipulations implemented to meet mandated

regulation schedules. Short-term fluctuations are not pronounced because the data

reflect larger-scale hydrologic trends. I chose to use the open-water dataset for three

reasons. First, despite considerable differences in the magnitude of fluctuations and

response times, the general patterns of fluctuation shown in the two datasets were

similar; i.e. periods of rising or receding water generally matched (Fig. 2-3). Second,

the distribution of rainfall over the marsh zones was highly uneven from day to day

(pers. observ.). For this reason, I felt that the open-water data provided a better index

to basin-average conditions. Lastly and most importantly, there were large gaps (36-57

days) in the marsh dataset during the wading bird nesting season in 1989, 1991 and

1992, while the largest gap in the open-water dataset was only two days.

From the stage data, I derived additional variables to represent the mean rate of

change in surface-water levels (cm day"') over various antecedent time intervals. I

calculated the differences between the stage at survey time and that recorded 30, 60,

90 and 120 days earlier, and then divided each difference by the appropriate number of












0) CD


(GAOIAI w) a6e8s


a-






I$
0







o)
50



i i




a)






a.



9,
a)














J it
Nc










e










o-
0


(uwo) lieu!eel


a,
0



o
CD





ia
.5
0)
a,





r'-
0







0
.c




















to
a,




-E
o 0
a, 0)




co
a,
E
.ca.




Cr





0 CD
C5


oa,
00




o. a,



.So




OS



. --







days. For convenience, I will refer to the latter as "drying-rate" variables, but note that a

negative value indicates a surface-water recession or drying trend, whereas a positive

value indicates rising water. I then regressed bird counts against each individual

hydrologic parameter, and analyzed multiple regression models of bird counts versus

pairs of stage and drying-rate variables in an effort to derive the best predictive model.

I also visually compared the distributional patterns of foraging wading birds

against bimonthly GIS maps of water depth derived from the OKEEHYDRO model

discussed in Richardson and Hamouda (1994). The OKEEHYDRO model was

programmed to output mean water levels for the 15th and last day of each month, for a

series of 915-m x 915-m grid cells covering all but the central, open-water section of the

lake. Each output file was then resampled (using READOKEE; Richardson 1993) to

produce ERDAS (1991) GIS maps of water depth with a spatial resolution of 100-m x

100-m. I then used another custom program (POLYSTATS, Richardson 1993) to

calculate bimonthly mean water depths for each of the 507 1-km x 1-km grid cells that

covered my transect (489 cells) and Eagle Bay Island (18 cells) survey areas.

Richardson and Harris (1994) generated annual vegetative cover maps by

classifying SPOT satellite images using ERDAS GIS software. Each map represented

a blend of data from one spring and one fall scene. The pixel resolution was 9.15-m x

9.15-m. A total of 30 classes of vegetation (including a class for non-vegetated open

water) were represented; 27 of these occurred in the transect and Eagle Bay Island

survey areas. Because the spatial resolutions of the vegetation and bird-distribution

maps did not match, it was necessary to rescale the vegetation maps to match the bird

maps. I accomplished this in two steps. First, I overlayed my survey grid matrix on

each vegetation map and calculated a vector of class-specific relative cover statistics

for each grid cell using POLYSTATS (Richardson 1993). Second, I derived a new







vegetation classification scheme by using the nearest-centroid clustering procedure in

SAS (FASTCLUS-a clustering algorithm specifically designed for aggregating large

numbers of observations; SAS Institute Inc. 1988) to divide the 4 x 507 sets of cell-

specific cover statistics into a reduced number of new aggregate vegetation classes.

The new classification scheme then became the basis for calculating expected bird

densities in chi-square (X2) tests of independence (Sokal and Rohlf 1981: 708)

designed to determine whether foraging birds actively selected or avoided certain

vegetation assemblages. I discuss the details of these tests in the results section as I

present summary charts.


Results and Initial Discussion


Abundance of Foraging Birds and Hydrologic Trends


Total combined-species population estimates for the study are shown in relation

to mean daily lake stage in Figure 2-4. Species-specific plots are shown in Figures 2-5

- 2-7. Appendix A contains a list of the actual counts by survey and species. I did not

include birds attending nests in these totals, but the addition of these relatively small

numbers (see Chapter 4) did not appreciably alter the trends shown.

The foraging-population counts of most species followed similar trends during

the study. Counts of all species except Glossy Ibises increased dramatically in

response to the protracted surface-water recessions and low lake stages of the 1989

and 1990 drought periods. Counts were generally lower at higher lake stages--

especially above 4.4-4.5 m-in late 1988 and early 1989, and in 1992 (also see Zaffke

1984 and David 1994b). Data for the uncommon, non-nesting Wood Storks and

Roseate Spoonbills further confirmed that the drought-induced low and declining lake









60000


50000


E 40000

4-
0
30000

E
z 20000


10000


0 1 ? II I 1 411119111111 1 !_____ iiiiii I I I I n 1 I I
01Aug-88 28-Jan-89 27-Ju-89 23-Jan-90 22Jul-90 18Jan-91 17-ul-91 13-Jan-92 11-Ju-92
30-Oct-88 28-Ap-89 25-Oct-89 23-Apr-90 20-0ct-90 18-Apr-81 15-Oct-91 12-Apr-82
Date


4z
E
(D


_e
0)
3

-j
2



1


Figure 2-4. Combined-species wading bird population estimates, excluding birds at
nests, in relation to daily lake stage: August 1988 August 1992. Gaps in bird counts
indicate no surveys.


30000

25000
U)
2 20000 -

o
a 15000
-0
E
= 10000
z
5000 -

0
01-Aug-88 28-Jan-89 27J 9 2Ja 2 n-0 22-Jul-90 18-Jan-91 17-Jul91 13-Jan-92 11ul-92
30-Oct-88 28-Ap-89 25-ct-89 23-Apr-90 20-Oct-0 18-Apr-91 15-Ot-91 12-Apr-92
Date


WiVwsiP,'- White Ibis

P /Great Egret

Small White Ardeids


Figure 2-5. Population estimates for White Ibises, Great Egrets, and Small White
Ardeids (i.e., Snowy Egrets and some immature Little Blue Herons), excluding birds at
nests: August 1988 August 1992. Gaps in records indicate no surveys.














5000

4000 -

S3000 -

-2000

1000 -

0
01-Ag-88 28-Jan-89 27-A-89 2 n-9 22-JU-90 18Jan-91 17u-91 13Jan-92 11.ul-92
30-Oct-8 2845Ap89 25Oct-89 23pr-90 2-Oct-90 18Ap- 15-Oct-91 12-Ap-92
Date


:/ Glossy Ibis

Great Blue Heron


Small Dark Ardeids


Figure 2-6. Population estimates for Glossy Ibises, Great Blue Herons, and Small Dark
Ardeids (i.e., Tricolored and adult Little Blue Herons), excluding birds at nests: August
1988 August 1992.


800 /S
700
600
500
400
300 /
200 1
100 -
0
01-Aug-88 28Jan-89 27-Jul-9 23-Jan-90 22-J-90 18Jan-91 17Jul-91 13-Jan-2 11-Jul-92
30-Oct-88 28-Ap-89 25-Oct-89 23-Ap-90 20-Oct-90 18-Apr-91 15-Oct-91 12-Apr-92
Date


/ / Wood Stork

Roseate Spoonbill

Great White Heron x10


Figure 2-7. Population estimates for Wood Storks, Roseate Spoonbills, and Great
White Herons: August 1988 August 1992.







stages attracted an unusual number and variety of birds to the lake (Fig. 2-7). The

drought concentrated prey in shallow pools, canals, alligator holes, and airboat trails

(cf. Kushlan et al. 1975; Kushlan 1976b; Kushlan and Hunt 1979), and especially in

beds of submerged vegetation at the outer edges of the littoral zone (Chick and Mclvor

1994). These rich prey resources were undoubtedly an important attractor, but the

unusually high numbers and diversity of birds also suggested that the lake served as

an important regional drought refuge for wading birds. The lack of a protracted spring

recession in 1991 apparently produced less-favorable foraging conditions, despite

moderate lake stages between 3.7-4.3 m. Populations initially were higher in 1992,

probably because the year began with a moderate recession. However, consistently

high lake stages restricted access despite favorable drying trends, and numbers

declined rapidly later in the season when the recession reversed.

As mentioned previously, I was unable to begin standard transect surveys until

late-March in 1991. However, during this time I conducted higher-altitude surveys of

the block and East-South Route regions, and perused the transect area to document

large flocks. These informal surveys confirmed that numbers of most species remained

low throughout the 1991 season. They also confirmed that White and Glossy ibises

were more abundant early in the season, although the numbers were never appreciably

higher than during the 30 March survey (Figs. 2-5, 2-6).

Glossy Ibises were unique in not showing the most dramatic increase in

numbers at the lowest lake stages in 1989 and 1990 (Fig. 2-6). Instead, the species

was more abundant earlier in 1989 and as abundant early in 1990 as it was during the

peak low-water periods each season. This trend could indicate a greater preference for

moderate lake stages, perhaps because a wider variety of foraging habitats is then

available. However, the species showed a consistent bimodal trend in numbers each







year, with lower numbers during the middle of the season. The consistency of the

pattern contrasted markedly with the inconsistency of the concurrent hydrologic

conditions. As mentioned, the displayed population totals do not include birds that

were attending nests, and therefore the mid-season population estimates for common

nesting species slightly underrepresent the actual population size. However, Glossy

Ibises nested on the lake irregularly and only in small numbers during the study (pers.

observ.). Therefore, the bimodal pattern was more likely evidence that the species

usually emigrated to other nesting grounds. The same bimodal trend in numbers was

evident to a lesser degree for White Ibises, and was least pronounced in 1990 (Fig. 2-

5) when White Ibises initiated the most nests at the lake (Chapter 4).

Counts of all common species declined during March and April 1989 (Figs. 2-5,

2-6), and again none of the declines were accounted for by on-lake nest numbers. The

temporary general decline in numbers corresponded to one primary (0.11 m over 12

days) and several lesser reversals in the drying trend that began at the end of February

and continued through March (Fig. 2-4). The brief drop in Wood Stork numbers

confirmed that a nesting exodus was not involved; Wood Storks did not nest at the lake

during the study and returned when surface-water levels began to recede again (Figs.

2-4, 2-7). A freeze that occurred at the end of February--coincident with the onset of

the primary reversal--probably added to the negative impact. Cold temperatures cause

small fish to become less active and flee, burrow and hide more readily when disturbed,

which probably renders it more difficult for wading birds to secure prey (Frederick and

Loftus 1993). The combination of cold temperatures and reversals in the drying trend

contributed to the abandonment of many Great Egret nests at this time (Chapters 4, 5).

Plots of combined-species bird counts versus mean lake stage and the drying-

rate variables indicated distinctly non-linear trends (e.g., Fig. 2-8). All plots showed that











60000


50000


40000 -


30000


20000


10000 -


0-
-0.8


60000


50000


40000


30000


20000


10000


0


-0.4 0.0 0.4
90-day Drying Rate (cm day-')


4 5
Lake Stage (m NGVD)


Figure 2-8. Non-linear regression of combined-species wading bird population
estimates versus a) daily lake stage and b) 90-day antecedent surface-water drying
rate (n = 49; curves fit using SYSTAT, Wilkinson 1990).








the foraging population grew at an increasing rate as the stage dropped and the drying

rate increased. Power functions of the form Y = A X A -B fit the data best, although

negative-exponential functions of the form Y = A EXP(-B X) fit nearly as well. A

power-function regression of counts versus lake stage yielded an adjusted R2 of 0.393

(Fig. 2-8; n = 49, residual df = 47; model fit using SYSTAT, Wilkinson 1990). However,

a similar regression of counts versus the 90-day drying-rate variable yielded the highest

R2 of 0.535 (Fig. 2-8). Multiple regression models including the additive effects of stage

and different drying-rate variables yielded only marginal improvement. None of the

models fit by the SYSTAT program adequately accounted for the unusually high peak

in numbers at the height of the drought in 1990. This lent additional support for the

contention that Lake Okeechobee served as an important regional drought refuge for

wading birds. In other words, the initial response trend may have reflected the benefits

of increasingly concentrated prey, but the accelerated response at the lowest stages

probably reflected the drought-refuge effect eluded to above.

I based the choice of regression models on measures of statistical fit rather than

a specific theoretical justification. Ultimately, bird abundance probably would reach an

asymptote and eventually decline if surface-water levels continued to drop, especially if

the decline outpaced the spread of submergent vegetation. Thus, a maxima function of

the form Y = A X EXP(-B X) may be the more appropriate theoretical model, but no

data exist to test this hypothesis. Regardless, wading birds were attracted to the lake

in ever larger numbers as the lake stage dropped, primarily because prey were

increasingly concentrated in receding waterways and isolated pools. It is likely that the

benefits of concentration accrue slowly at first, and then more rapidly when prey

concentrations reach levels sufficient to accommodate certain behavioral shifts among

the foraging birds. For instance, Kushlan (1976c, 1979) found that White Ibises usually







do not preferentially select fish as prey unless they are highly concentrated by receding

water. There may be a general threshold of concentration above which wading bird

foraging is greatly enhanced because the standard evasion tactics of small fishes and

other prey cease being effective. If so, some form of non-linear wading bird population

response would be expected.


Patterns of Distribution and Hydrologic Trends


Select pairs of water depth and combined-species bird distribution maps

covering the transect and Eagle Bay Island survey areas are shown for the 1989-1992

seasons in Figures 2-9 2-12. I chose the series to highlight major shifts in distribution

that occurred during the study, primarily as a result of changing hydrologic conditions.

Most species followed similar distributional trends, with some variations discussed in

the following sections. Only the largest Great Blue Herons and Great Egrets can wade

in water near 50 cm deep; most species are constrained to depths below 20-30 cm

depending on their leg length (e.g., see Kushlan 1986a). However, thick vegetation

may provide support and allow access to deeper habitats. This often was the case in

the Fisheating Bay area where thick floating mats of water hyacinth and thick surface

mats of Hydrilla were prevalent. In addition, Great Egrets often took advantage of thick

decumbent stands of Eleocharis and Panicum for support in the deeper wet prairie

areas of Moonshine Bay. Thus, the maps include instances where birds were observed

in cells with a mean water depth greater than 50 cm. Furthermore, the 1 km2 spatial

resolution of the water depth data, and the fact that I lumped the data into only five

depth categories, obscured fine-scale topographic detail. Consequently, the maps also

include instances where birds were observed in cells with a mean depth of zero.







0 cm
0 birds
Depth


Depth and Bird Density (birds km"2) Categories
1-10 cm ] 11-50 cm m 51-100 cm 101 + cm 676+ birds
1-3 birds 4-75 birds 76-225 birds 226-675 birds


Birds


Depth


Birds


30-Jan-89: Stage = 4.32 m


19-Jun-89: Stage = 3.47 m


10-Apr-89: Stage = 3.95 m


8-Aug-89: Stage = 3.39 m


Figure 2-9. Select 1989 combined-species wading bird density maps for the transect and
Eagle Bay Island areas, shown in relation to plots of mean water depth derived from the
OKEEHYDRO hydrologic model (Richardson and Hamouda 1994).


I


. .. .


M-







Depth and Bird Density (birds km-2) Categories
1-10 cm 11-50cm 51-100 cm 101+ cm 676+ birds
1-3 birds 4-75 birds 76-225 birds 226-675 birds


Birds


Depth


Birds


25-Jan-90: Stage = 3.74 m


1-Jun-90: Stage = 3.26 m


7-Apr-90: Stage = 3.51 m


9-Aug-90: Stage = 3.50 m


Figure 2-10. Select 1990 combined-species wading bird density maps for the transect and
Eagle Bay Island areas, shown in relation to plots of mean water depth derived from the
OKEEHYDRO hydrologic model (Richardson and Hamouda 1994).


0 cm0 -c
0 birds

Depth




37
Depth and Bird Density (birds km"2) Categories
F 0 cm 1-10cm 11-50 cm m 51-100 cm m 101+ cm I 676+ birds
0 birds 1-3 birds 4-75 birds 76-225 birds 226-675 birds


Depth


Birds


Depth


Birds


13-Apr-91: Stage = 3.85 m


8-Jun-91: Stage = 3.98 m


ILI


\ I


22-Jul-91: Stage = 4.26 m


1-Sep-91: Stage = 4.81 m


Figure 2-11. Select 1991 combined-species wading bird density maps for the transect and
Eagle Bay Island areas, shown in relation to plots of mean water depth derived from the
OKEEHYDRO hydrologic model (Richardson and Hamouda 1994).


4


4277


tI~


i






- 0 cm -
0 birds
Depth


Mean Depth and Bird Density (birds km-2) Categories
1-10 cm 11-50 cm M 51-100cm 101+ cm 676+ birds
1-3 birds 4-75 birds 76-225 birds 226-675 birds


Birds


Depth


Birds

Orfi



17Ui^


18-Jan-92: Stage = 4.76 m


16-Apr-92: Stage = 4.67 m


\NI


18-Jun-92: Stage = 4.41 m


15-Aug-92: Stage = 4.76 m


Figure 2-12. Select 1992 combined-species wading bird density maps for the transect and
Eagle Bay Island areas, shown in relation to plots of mean water depth derived from the
OKEEHYDRO hydrologic model (Richardson and Hamouda 1994).


.......................... ................... .......


8







Airboat trails, which are prevalent on the lake, were conspicuous examples of localized

depressions that remained wet and attracted wading birds when most of the

surrounding marsh had dried.

With the above exceptions in mind, the maps confirm that the distribution of

birds was generally constrained to wet areas with mean water depths below 50 cm. At

high lake stages, relatively small numbers of birds were dispersed throughout the

higher-elevation, shallower areas of the littoral zone (January 1989, Fig. 2-9;

throughout 1992, Fig. 2-12). Habitats commonly used at this time included Panicum-

dominated mixed grasses and patches of Eleocharis and Rhynchospora (see

vegetation maps introduced in the next section and maps provided in Richardson and

Harris [1994] for greater detail). In contrast, the number of birds increased dramatically

as the lake stage dropped during the drought periods of 1989 and 1990 (June 1989,

Fig. 2-9; April and June 1990, Fig. 2-10). Foraging activity was progressively restricted

to the Eleocharis flats of Moonshine Bay, then to mixed lakeward-fringe habitats, and

eventually to exposed beds of submerged vegetation and open-water fringes at the

lowest lake stages. The 1989 and 1990 maps (Figs. 2-9, 2-10) indicate that areas of

concentrated foraging were most often located in either shallow habitats (1-10 cm

depths) or areas of transition from shallow to moderate depths (10-50 cm).

The 1991 maps (Fig. 2-11) indicate that, despite generally moderate lake

stages, the lack of a spring surface-water recession resulted in sparse, widely

dispersed foraging populations. In addition, comparing the last maps in the 1989,

1990, and 1992 series against others in each series confirmed that foraging birds were

less common during late surveys when surface-water levels were rising than during

early surveys when water levels were declining. This was true despite the fact that

areas of suitable depth were equally or more abundant late in the season.







Vegetation Maps and Changes During the Study.


Table 2-1 lists the original classes of vegetation (with scientific nomenclature for

those species hereafter referred to by common names) represented on the classified-

satellite-image vegetation maps produced by Richardson and Harris (1994). The

abbreviated names provided in Table 2-1 are used in Table 2-2 to describe the

composition of the FASTCLUS-derived aggregate vegetation classes assigned to each

of the survey grid cells. I arrived at the choice of 18 classes by an iterative process.

When using the FASTCLUS procedure, one must specify in advance the number of

classes desired; I compared the results of specifying between 10 and 27 classes. My

goal was to reduce the number of classes as much as possible without eliminating

recognizably distinct classes of habitat, particularly those that observations suggested

were important for foraging wading birds. I decided that 18 classes achieved the best

compromise. If I included more, some classes did not correspond to recognizable

assemblages. If I included less, some distinct and important classes were lost.

Justification for listing vegetation types in Table 2-2 that comprised less than

10% cover is based upon observations that small patches of favorable habitat often

attracted birds to otherwise seemingly inaccessible habitat complexes (e.g.,

Polygonum-dominated or Eleocharis and Nymphaea patches in a thick cattail matrix).

Even as low as 5% cover potentially equaled a patch size of 50,000 m2, which could

accommodate a sizable flock. Moreover, in many cases I included low-cover types

because they represented a closely related extension of major groups (e.g., mixed

cattail and Nvmphaea [CATNY] relative to cattail [CAT]). The codes chosen to

represent each of the final classes therefore reflect primarily dominant constituents, but

in some cases I felt it was important to emphasize the inclusion of low-cover categories







Table 2-1. Vegetation types represented on annual, 9.15-m x 9.15-m resolution,
classified satellite-image maps produced by Richardson and Harris (1994). Cover
statistics for these classes formed the basis for the 1-km x 1-km resolution classification
scheme developed to match the wading bird survey grid (see Table 2-2).



Code Vegetation Type Code Vegetation Type


ceph

mela

wil

spar

cat

saw

mixup

rhy

bog



mixgr



nym

eleo

hyac

scrp


Cephalanthus occidentalis

Melaleuca quinquenervia

willow, Salix caroliniana

Spartina baker

cattail, Typha sp.

sawgrass, Cladium jamaicensis

mixed upland complex

Rhynchospora tracvi

Pontedaria lanceolata and

Sagittaria lancifolia mix

mixed grasses, primarily

Panicum repens

Nymphaea sp.

Eleocharis celulosa

hyacinth, Eichhomia crassipes

Scirpus sp.


open

per

elper

nyel

lot

subcl



catny

phr

maid



sppan

sued

poly

cpsag


deeper open water

periphyton

Eleocharis and periphyton mix

Nymphaea and Eleocharis mix

lotus, Nelumbo lutea

submerged vegetation

and shallow open water

cattail and Nymphaea mix

Phragmites communis

maidencane, Panicum

hemitomon

Spartina and Panicum mix

successional-disturbed mix

Polygonum sp.

cattail, Pontedaria, and

Sagittaria mix








Table 2-2 Identification codes and composition of vegetation community classes
assigned to 1-km x 1-km survey grid cells for habitat selection analyses (see Table 2-1
for descriptions of constituents).


Code


N


Mean % Cover of Primary and Important Secondary Constituents


mixup-58 cat-14 mixgr-8


MU

MG

SMG

SD

SCWP

WCNP

CNWP

C

ECN

E

EP

CSNE

MCSE

RCGE

L

SUB

OSS


catny-4 bog-7



poly-8

poly-6

poly-7


mixgr-53

spar-39

sucd-67

sucd-35

wil-44

cat-25

cat-58

eleo-28

eleo-54

elper-39

saw-24

mela-57

rhy-26

Iot-49

subcl-74


sppa-5

mixgr-20

mixgr-10

cat-14

sucd-19

catny-17

catny-6

nyel-11

elper-8

eleo-18

cat-24

cat-7

cat-24

cat-12

open-7


open-47 subcl-24 scrp-10 cat-7

open-94


spar-3

sucd-11

wil-6

catny-4

cat-7

nym-5

mixgr-6

cat-11

nyel-7

peri-7

catny-7

saw-7

mixgr-12

open-7


cat-18

cat-10



wil-10

catny-8

wil-9



sucd-12

catny-2

cat-9

nyel-6

nyel-7

elper-11

subcl-6


saw-5

cat-8

mixgr-6

mixgr-8

eleo-5

eleo-5


mixgr-5

mixgr-8







saw-9







that appeared important to wading birds. The issue highlights the fact that the 1 km2

scale of spatial resolution masked some important fine-scale patterns of habitat use. In

the following accounts, I point out several instances where I thought this occurred.

The vegetation maps that resulted from the FASTCLUS procedure are shown in

Figures 2-13 and 2-14. In the following discussion, I highlight the most prominent and

pertinent changes in vegetation communities that occurred during the study. For

additional detail, readers should consult Richardson and Harris (1994) and Richardson

et al. (1994).

In 1989, cattail (C) dominated a large portion of the survey area. However, in

the southwestern section of the transect area, EP (Eleocharis and periphyton), CSNE

(cattail and sawgrass with patches of Nymphaea and Eleocharis) and RCGE

(Rhynchospora and cattail--and sawgrass--with mixed grass and Eleocharis patches)

vegetation mixes were abundant and graded into more pure Eleocharis stands (E) in

Moonshine Bay (Fig. 2-13). In the upper Indian Prairie region, mixed grasses (MG)

were prevalent and graded into E and EP mixes toward the north and into CNWP

mixes (cattail and Nymphaea with patches of willow and Polvyonum) toward the

lakeward margin. The 1989 map contained relatively few cells classified as dominated

by submerged vegetation (SUB). Instead, most outer fringe cells were classified as O

(deeper open water) and some as OSS (mixed open water and submerged vegetation

with patches of Scirpus). This reflected most closely the conditions prevalent during

the first half of the 1989 season, but a slight variation in classification protocol actually

under-represented the 'subcl' class on the original maps (Richardson and Harris 1994).

The latter had some ramifications for the habitat selection analyses (see below).

However, the ensuing drought cycle greatly enhance the spread of submerged

vegetation (Zimba et al. 1994), and the indication of greater cover for the SUB habitat













a) 8




E --

E .2


i,-
Eo_ a


IcI



o0 E



.- o
> 2
E m.)

eg.



0 C0

a a )


4-
S0.



cn ,-. 8 8

v- C .
(-


a(s


>I



O -0e
E wo





aE-
Se 8E
2 ao
Cm 8D
,"^ss









































Lu- LL a LU
; CD o; r z 2L ooCD

U -j orET w W wUC










0
W
S*8




ca Z
E,
a) E





en



.(DE




E en>




(0 0
i0e




E o
2r -' -

>w =

o 0 -
0 0u




- .-
CY)
E w.L
'- O











Ne .0".



c S
gs>










0 -co
mE E






0o -

0) 0
. T. 8 8
oS^
H.f Q-










E *oSm
CM ) 2>{









































LU LLUJ Lii
Coooo o ;iooZ
..o CC CO co wO o D U C -53
*uMWFr MEN, Now


a-


D'" M


7-







type on 1990 and 1991 maps (Figs. 2-13 and 2-14, respectively) would not have been

obscured by an adjustment to the 1989 map.

Prior to mid-1989, the main willow-dominated section of Eagle Bay Island (see

maps in Richardson and Harris [1994] for best detail) was almost entirely surrounded

by open water and sparse submerged vegetation, with a narrow fringe of Scirpus in

some areas. As the drought progressed, a wide fringe of Scirpus, cattail, grasses and

sedges, Cesbania sp., other annuals, and submerged vegetation began to develop. By

1990 the "island" had effectively doubled in size. In addition, a smaller, previously

isolated stand of Scirpus located just offshore of Eagle Bay Island began to expand.

Mats of water hyacinth and water lettuce filled open areas, and expanding stands of

Scirpus and cattail increased the area coverage of the new "island". Between these

two island regions, an expanse of lotus pushed up through a thickening Hydrilla and

Vallisneria understory. This was an important feeding area for nesting adults from

Eagle Bay Island and nearby areas (Chapter 3).

In late 1989, a severe freeze struck southern Florida and killed much of the

emergent vegetation on the lake; cattail was hit particularly hard (Richardson and Harris

1994). During the following spring and summer, a series of lightning fires were ignited.

Most of the marsh was nearly dry due to the drought, and the fires burned very hot and

cleared much of the emergent vegetation. In response, a complex mosaic of mostly

annual, successional species emerged (Richardson and Harris 1994). As a result, a

large portion of the 1990 vegetation map (Fig. 2-13) was assigned to the highly variable

successional-disturbed (SD) class, and several more distinct classes were eliminated

(i.e., MCSE Melaleuca forest grading to cattail and sawgrass with Eleocharis and

Nymphaea patches, RCGE, CNWP, and most of what was MG). Otherwise, many of

the cells in the southwestern section that were classified as E and EP in 1989 shifted in







composition to the more complex ECN class (mixed Eleocharis, cattail, and Nymphaea)

in 1990. Moreover, in many areas previously dominated by other species, patches of

Polygonum emerged; an expansion of the mixed-successional SCWP class (a complex

mix of the SD class and patches of cattail, willow, Polygonum, and other species)

reflected this trend.

By 1991, as the lake stage began to increase again, many of the aquatic

species regained their former dominance. The Eleocharis stands in the southwestern

and Moonshine Bay areas and the mixed-grass complexes on upper Indian Prairie

reemerged (Fig. 2-14). In addition, a new class of mixed grasses appeared that

included extensive patches of Spartina interspersed with Panicum and other species

(SMG). Cattail was slower to regenerate fully; the CNWP class reappeared at 1989

levels, but the heavily mixed SCWP complex-in particular flats of Polygonum-also

remained prevalent. The thick submerged beds and floating mats that developed

around Fisheating Bay, Eagle Bay Island, and King's Bar continued to thrive, but

deepening water reduced accessibility for most wading bird species.

By 1992, cattail had again regained its stronghold on the lake, and the

regeneration of Eleocharis and Panicum continued (Fig. 2-14). The exotic Melaleuca

also expanded dramatically following the drought/fire cycle. Compared to 1989, by

1992 the number of MCSE cells had doubled along the western margin near Moore

Haven and south, generally filling in where the 1989 RCGE habitats once were

common. Willow also responded favorably to the drought/rewetting cycle, and the

WCNP class (willow-dominated, with patches of cattail grading to Nvmphaea, and

Polygonum) increased five-fold over 1989 levels. Willow-dominated habitats rarely

attracted large numbers of foraging wading birds during this study (see below), but

willow serves as the primary nesting substrate for all species (Chapter 4; Zaffke 1984;







David 1994a). During the 1992 season, after the lake stage had returned to high

levels, the extensive floating mats and thick Hydrilla beds that flourished during 1990

and 1991 in Fisheating Bay and around Eagle Bay Island and King's Bar thinned and

dispersed. This eliminated one of the last remaining profitable foraging habitats on the

lake for Snowy Egrets, in particular.


Habitat Selection Analyses


Methodological details, summary graphics, and interpretation of results.

Hydrologic conditions varied each season and patterns of habitat-use changed

accordingly. I began by examining selection trends for each survey separately.

However, I concluded that early- and late-season combined-survey analyses (i.e.,

survey data combined for January-April and May-August) were sufficient to represent

major intra-annual, hydrologically-related shifts in distribution. I then compared the

semiannual results with those derived from annual combinations of data, and

culminated the perspective by combining information from all surveys in one analysis.

In this progressive fashion, I was able to identify habitat selection trends that persisted

across all temporal scales, and to isolate other trends that seemed more related to

seasonal or annual hydrologic conditions. As mentioned previously, the annual scale

of temporal resolution was most compatible with the vegetation community data. This

therefore was the lowest temporal scale for which I calculated 2 statistics, and is the

focus of attention in the accounts that follow. My presentation of results emphasizes

major and consistent study-wide trends, but also specifically highlights where changes

in lake stage led to shifts in habitat selection on both intra- and inter-annual scales.

Species-specific graphical representations of the annual X2 test results for 1989-

1992 are provided in Figures 2-15 2-18. The bars in the figures represent percent







deviations from expected (relative to abundance of vegetation types) of bird counts

summed by vegetation type across each series of annual surveys (i.e., not counts of

distinct individuals):

plotted value = (total count in given vegetation type / grand total count of birds)

(number of cells of a given vegetation type / total number of cells)

Bars extending above zero indicate higher-than-expected-use or positive selection of

the vegetation type (for practical purposes no maximum limit), whereas those extending

below zero indicate underutilization or negative selection (limit = -100%; i.e., maximum

underutilization or no use). All future references to "active" or significant, positive or

negative selection will imply X2 P < 0.001. Similarly, reference to marginal selection will

imply 0.001 < P < 0.05, and to neutral selection P > 0.05. I discuss the reasons for

choosing a conservative significance level in the discussion section.

A two-part alphanumeric code appears above each bar in Figures 2-15 2-18.

The first half of the code is a number between 0 and 10 that indicates on a relative

scale how many birds used each vegetation type during the specified year. For each

wading bird species, I summed the survey counts within vegetation types and years,

and then divided each total by the number of surveys conducted during the given year.

This resulted in estimates of the average number of each species counted in each

vegetation type on a typical survey in a given year. For each species, I pooled the

estimates for all vegetation types and all years, and from these I identified the study-

wide maximum value and to it assigned the value of 10 or 100%. I scaled all other

estimates accordingly in 10% increments with 0 = no birds, 1 = 1-10% of the maximum,

2 = 11-20%, etc. The numeric codes therefore reflect study-wide trends in

concentration of numbers in specific types of vegetation, without regard for the







abundance of vegetation types. Thus, they provide a means for judging the

comparative importance of the indicated annual selection trends. The second half of

the code is a label and is absent in most cases, indicating a significant X2 test (P <

0.001). Otherwise, a "?" indicates marginal significance and an "n" indicates no

significant difference (P > 0.05).

To properly interpret the results of the habitat selection analyses, one must

always keep in mind that the analyses did not statistically account for the interactive

effects of water depth and vegetative cover. Thus, although the label for a given

habitat type remained consistent, and the species of vegetation occupying a given cell

may have remained relatively consistent, the hydrologic characteristics of the cells were

always changing. In many cases, the classification of cells changed over the course of

the study because hydrologic conditions varied enough to cause large-scale shifts in

the vegetation. However, in some cases a change in the use of a given cell by birds

occurred even though the classification of the cell did not. Such a change could have

occurred because more favorable vegetation mixes emerged in nearby areas.

However, the change could also have occurred because the particular parcel of habitat

either dried out or flooded too deeply for it to remain an attractive foraging area.

Generally low use of upper-elevation marsh habitats at low lake stages undoubtedly

had more to do with the area being dry than with other types of vegetation being more

attractive. Conversely, typically few birds used submerged vegetation and open-water

habitats at high lake stages because the water was too deep for wading, not because

prey were absent or some feature of the vegetation was unattractive.











% Deviation from Expected
so0

700- Great Blue Heron
10





100 3

-100 2 1 1 5 1 1 2

-300

900
Small White Ardeid
700 (primarily Snowy Egret)

500
7
300

100- 2 13 3 5

_100- _2 ---0


900

Identified
700 Tricolored Heron

500

300

100- 2

-100 in in 7 in
07 07 On On
-300

900

700 White Ibis

500
6
300
21 7
100 1 3 7 1

0o 2

-300
MU SCWP CNWP ECN EP MCSE L OSS
MG WCNP C E CSNE RCGE SUB O
Vegetation


uuu

700 Glossy Ibis
6
500
4



100 1 2
0 1 0 1 0 1

-300
MU SCWP CNWP ECN EP MCSE L OSS
MG WCNP C E CSNE RCGE SUB O
n Types


Figure 2-15. Wading bird foraging habitat selection in relation to vegetation
assemblages in 1989; Chi-square tests of independence with bird densities
summed across surveys. Vegetation codes are described in Table 2-2.
Aplhanumeric codes above bars are interpreted as follows: number = study-wide
relative amount of use, 10 highest-see text for details; label = Chi-square
significance-n = P > 0.05, ? = 0.001 < P < 0.05, no label = P < 0.001.










% Deviation from Expected


900


(primarily Snowy Egret)
500

300-



10 7 3 i 1
0 0 1 1
.in -


900
70 White Ibis

500



100 6 4 1 4 3

3 00 0
: n 0 -SB1


MU SD WCNP ECN
MG SCWP C E


Small Dark Ardeid
0 (all TC/LB)

500
2




1 0 1 1

300
900
-100 1 o 1





Identified
Little Blue Heron
500

3000

100& 3

100 4
07 o on o
-300

900

700 Glossy Ibis

500

300

1001n 1 3 3 1

-100 1

-300


EP SUB 0
L OSS
Vegetation Types


Figure 2-16. Wading bird foraging habitat selection in relation to vegetation
assemblages in 1990; Chi-square tests of independence with bird densities
summed across surveys. Vegetation codes are described in Table 2-2.
Aplhanumeric codes above bars are interpreted as follows: number =
study-wide relative amount of use, 10 highest-see text for details; label =
Chi-square significance-n = P > 0.05, ? = 0.001 < P < 0.05, no label = P <
0.001.










% Deviation from Expected
900

700 Great Blue Heron
700

500

300

100 In 2 In

-100 o 1 o?

-300

900

S Small White Ardeid
(primarily Snowy Egret)
500
3
300









Identified
700 Tricolored Heron

500

300

100 1 i 2 17

-100 In M
Iin 1? on an 17
-100 1?1 1


MU SMG SCWPCNWP ECN EP MCSE SUB O
MG SD WCNP C E CSNE L OSS


700 Great Egret

500

300

100 3 1 1 1
-100m 1 1
0 0 1 1

-300

900

70 Small Dark Ardeid
(all TC/LB)
500

300 2

100D- in In 2 ?
-100 In In U 10
0 070 1
in i __ ___


700
700- Glossy Ibis

500



100 n

-100 in in
7 1 0? 0 O? 1 0
-300
MU SMG SCWP CNWP ECN EP MCSE SUB O
MG SD WCNP C E CSNE L OSS


Vegetation Types


Figure 2-17. Wading bird foraging habitat selection in relation to vegetation
assemblages in 1991; Chi-square tests of independence with bird densities
summed across surveys. Vegetation codes are described in Table 2-2.
Aplhanumeric codes above bars are interpreted as follows: number =
study-wide relative amount of use, 10 highest-see text for details; label =
Chi-square significance-n = P > 0.05, ? = 0.001 < P < 0.05, no label = P <
0.001.










% Deviation from Expected
900

S Great Blue Heron

500


1 1
300
-100 I MEN

o0 On On 1 01

-300

900

700 Small White Ardeid
(primarily Snowy Egret)
500

300
3 3
100- 4 8 14

-100 2

-300

900

Identified
Tricolored Heron
500

300 10

100 On1

S-100 1 17 1 10

-300

900

White Ibis

500

300

100 2

-100 t 1 1 7 1 1

-300
MG SD WCNP C E CSNE L OSS
SMG SCWPCNWP ECN EP MCSE SUB O


900

700 Great Egret

500

300 2 3

100 4 1

0 1 1 1 1

-300
-300 .--- .-- -. --*. ---.*

900

70, Small Dark Ardeid
(all TC/LB)
500

300


100- 1 1 31



-300 On

900

Identified
S Little Blue Heron
500

300

100 1? 2i 6 2 2 2

On 1 10
9& 110
-300

900

700 Glossy Ibis



300 2
100 1


_100
0 On 1 0 1 0? 0 1 1 0
-300
MG SD WCNP C E CSNE L OSS
SMG SCWPCNWP ECN EP MCSE SUB O


Vegetation Types



Figure 2-18. Wading bird foraging habitat selection in relation to vegetation
assemblages in 1992; Chi-square tests of independence with bird densities
summed across surveys. Vegetation codes are described in Table 2-2.
Aplhanumeric codes above bars are interpreted as follows: number =
study-wide relative amount of use, 10 highest-see text for details; label =
Chi-square significance--n = P > 0.05, ? = 0.001 < P < 0.05, no label = P <
0.001.







Habitat selection trends. All species selected wet-prairie habitats of moderate

structural complexity, particularly those featuring Eleocharis (E, EP, and often ECN),

during some portion of most years (Figs. 2-15 2-18). The only exceptions were that

Snowy Egrets (SWA) and both species of ibis underutilized these habitats in 1992 (Fig.

2-18). The largest extent of Eleocharis habitat occurred in and around the fringes of

Moonshine Bay. This region and a less extensive but more diverse area on northern

Indian Prairie were the focal points of foraging activity at moderate lake stages (3.7-4.3

m) during 1989 and the first half of 1990 (Figs. 2-9, 2-10). Most species used and

actively selected structurally similar RCGE habitats more at higher lake stages early in

1989. The center of distribution for RCGE habitats was west and south of the large

Moore Haven nesting colony (shown in Fig. 2-1 due north of the town of Moore Haven).

The close proximity of RCGE habitats to the colony may have enhanced their

attractiveness. The same relationship may have applied in 1992. The Moore Haven

colony was essentially unused in 1990 and 1991, but was reoccupied in 1992 when

water levels rose again (Chapter 4). The RCGE class was not represented as such

after 1989 due to the invasion of Melaleuca and other changes. However, in 1992 the

area still included patches of Eleocharis, Rhynchospora and grasses, and selection by

both species of ibis of the MCSE habitat type reflected a return to the area (Fig. 2-18

and Fig. 2-12: April/June 1992). The change in classification resulted from changes in

the vegetative community, but this is one case where the 1 km2 scale of resolution

failed to capture some important fine-scale details concerning patch choice. Zaffke

(1984) also commented on the importance of this southwestern region and these

habitat types at higher lake stages. In general, however, selection of Eleocharis

habitats was more irregular during 1991 and 1992 (Figs. 2-17, 2-18), as Moonshine Bay

reflooded deeply enough to exclude most species (Figs. 2-11, 2-12). Throughout the







study, Great Egrets remained the most consistent in their selection of E and EP

habitats, with Little Blue Herons ranking second in overall consistency and strength of

preference.

Most of the higher-elevation prairies and even the Eleocharis flats of Moonshine

Bay dried out when the drought reached its peak in mid-1990. Most the remaining

aquatic foraging habitat was restricted to beds of submerged vegetation and lotus, and

to open flats in bays and along the lakeward fringes of the littoral zone. As noted

above, the largest populations of foraging birds observed during the study were

attracted to the lake at this time. Activity was focused primarily in mixed submerged

habitats (SUB, OSS, and adjacent lotus [L] flats), mostly in Fisheating Bay and around

King's Bar and Eagle Bay Island (Fig. 2-10; June 1990). Relatively large numbers of all

species actively selected these habitats during the second halves of the 1989 and 1990

seasons (least true for Tricolored and Little Blue herons, which remained more

dedicated to E and EP habitats). Snowy Egrets and Great Blue Herons exhibited the

strongest, most consistent tendency to select submerged habitats, but Great Egrets

also exhibited a high affinity for submerged (SUB) and nearby mixed lotus and Hydrilla

flats (L) in 1990 (Fig. 2-16). Observations revealed that Snowy Egrets were attracted

by floating mats of water hyacinth and the large surface leaves of lotus (strong

selection for L habitats in 1990, Fig. 2-16). Both provided support in deep water (also

see Murdich 1978) and facilitated aerial foraging over prey-rich Hydrilla beds. Snowy

Egrets continued to use these habitats and herbicided mats of water hyacinth (see

below) well into 1992 after most other species had moved back toward the higher-

elevation marshes. Great Blue Herons typically focused more on open areas (0) and

open pockets in submerged beds (OSS) occupied by breeding sunfish (Lepomis sp.)

and largemouth bass (Micropterus salmoides). This trend was most notably manifest







as atypical, near-neutral selection of the most open habitat types (0) in 1989 and

especially 1990 (Figs. 2-15, 2-16). The relative density of Great Blue Herons in these

habitats was also atypically high during these times.

Ibises shifted back to MCSE habitats on the southwestern margin once surface-

water levels rose again following the 1989/90 drought. However, all other species

shifted their attention more to the mixed-transitional SCWP, CNWP, WCNP and less

often SD habitats (all of which flourished following the drought), and to the higher-

elevation MG and SMG habitats on Indian Prairie. Mixes of Eleocharis and

Rhynchospora often occurred in the first group of transitional habitat types, but patches

of Polyvonum attracted equal or greater activity. Great Egrets were particularly

attracted to reflooded Panicum-dominated mixed grass habitats in 1992 (Fig. 2-18).

Most species actively selected SCWP habitats and used them heavily throughout the

1990-1992 seasons; however, in 1989 this habitat type was uncommon and only White

Ibises showed positive selection (Figs. 2-15 2-18). CNWP and WCNP habitats were

used less consistently. Use of these habitats was generally low during the 1989 and

1990 drought periods (Figs. 2-15, 2-16), but was higher among Great Blue Herons and

Great Egrets early in 1989 (Fig. 2-15), and was higher among most species-especially

ibises-in 1991 and 1992 after the areas were reflooded (lake stages above 3.7 m;

Figs. 2-17, 2-18).

The SD class of highly disturbed habitats attracted large numbers of birds in

1990. However, this habitat class was ubiquitous and represented a wide array of

mixed-transitional assemblages, and selection was generally neutral or negative (Fig.

2-16). Glossy Ibises were the notable exception, suggesting that the species was

strongly attracted to diverse mixes of vegetation and/or areas of decaying vegetation

(the latter may support high invertebrate populations). There may have been more







statistical evidence that Glossy Ibises selected such habitats in 1990, but my

observations suggested that White Ibises were also attracted to areas of decaying

vegetation and transitional complexes. The species' near-neutral selection of SD

habitats in 1990 lent support for this contention (Fig. 2-16). Snowy Egrets also used

SD habitats heavily in 1990, with selection near neutral (Fig. 2-16). Moreover, White

Ibises, Snowy Egrets, and identified Tricolored Herons all showed significant positive

selection for SD habitats in 1991, after a significant amount of habitat regeneration had

occurred and the number of SD cells had declined. Thus, it appeared that, like Glossy

Ibises, these species were attracted to diverse mixes of vegetation. Both species of

ibis also actively selected willow-dominated WCNP habitats in 1991, habitats that

included mixtures of muddy, coarse detritus and transitional vegetation interspersed

with the willows. There was only one cell classified as SD in 1992, and only one group

of 50 White Ibis occurred there. However, because White Ibises were uncommon in

1992, the observation contributed to an overall significant positive selection trend (Fig.

2-18). The validity of the significance test was questionable, however, because the

expected frequency was so low.

Cells classified as dominated by cattail (C) were also heavily used, but as with

SD habitats, selection was usually neutral or negative (Figs. 2-15 2-18). Cattail was

most prevalent in 1989 before the freezes and fires of 1990 removed most of the thick

cover, and the volume of use was also highest in 1989. My observations suggested

that dense stands of cattail were rarely used by foraging birds; cattail simply had

invaded most areas of the lake and was mixed in with almost every other species of

vegetation (Richardson and Harris 1994). This is another example where the 1 km2

scale of spatial resolution masked finer-scale details of habitat use; birds observed

near thick cattail (or sawgrass or willow or Phragmites or Melaleuca, etc.) were usually







foraging in smaller patches of less dense habitat (e.g., patches of Eleocharis,

Rhynchospora, Panicum, Nymphaea, and Polygonum, or often in airboat trails). At

times in 1990/91, however, White Ibises were attracted to thick, decaying mats of

cattail, perhaps to feed on large numbers of detritivorous insects and snails.

All 18 classes of vegetation were used for foraging by at least one bird

sometime during the study. The MU class was the rarest-one upland area just south

and west of the Fisheating Creek inlet (maximum 7 cells in 1989; Fig. 2-13)-and was

rarely used, as expected. The densest (SMG and CSNE), most wooded (WCNP and

MCSE), and most open (0) habitats also all failed to draw significant study-wide

positive selection. However, each of these habitat types attracted at least moderate

numbers of some species and positive selection (except for O habitats) by one or more

species during at least one half-season. MCSE habitats were the rarest of the group

and attracted the second lowest number of birds overall. Nonetheless, the Eleocharis

and grass/sedge habitats at the fringes of thicker Melaleuca and cattail/sawgrass

stands along the western margin were strongly selected by Little Blue Herons early in

1989 and by ibises in 1992 when the lake stage was high. L habitats were also about

as rare as MCSE, but attracted overall strong positive selection by Snowy Egrets,

moderately positive selection by White Ibises, and overall neutral but slightly positive

selection by Great Egrets. The proximity of lotus stands and submerged beds,

particularly Hvdrilla, may have caused a spill-over effect and contributed to the

apparent but possibly coincidental selection of L habitats. However, Snowy Egrets and

cryptic and poorly documented Tricolored Herons regularly used the large, surface

leaves of lotus plants for support while foraging around the edges in thick but less

supportive Hydrilla. Open water habitats (0) were never actively selected by a species,

but most species used O cells to some degree at low lake stages-Great Egrets, Great







Blue Herons, Snowy Egrets, and White Ibises, in particular. Selection approached

neutral for both Great Blue Herons and Great Egrets at the height of the drought in late

1990. At this time, large aggregations of birds often collected in totally open areas,

apparently to feed on schooling fishes and breeding sunfish.

The southern island and marsh areas along the East-South Route (Fig. 2-2)

included some unique patches of exotic vegetation and altered landscape, but most of

the habitats used by foraging birds in these areas were well-represented in the transect

and Eagle Bay Island regions. However, an important component of Great Blue Heron

foraging habitat covered by the East-South Route was not included in the transect and

Eagle Bay Island areas. On most surveys, a substantial proportion (5-20%) of the

Great Blue Herons occurred along the open eastern fringe of the lake between Taylor

Creek and Pahokee (Fig. 2-1). Many of these birds were fishing from rocks in open

water at the edge of the lake; others waded beyond narrow fringes of Scirpus. In some

instances, over 100 birds were spaced at intervals of 50-150 m along considerable

stretches of "beach". Moreover, nearly all the Great White Herons I observed on the

lake were foraging in the same rocky-fringe habitat. The birds' targets probably

included large shad (Dorosoma sp.), golden shiners (Notemigonus crystoleuca),

sunfish, and possibly occasional schools of mullet (Mugil cephalus).

In summary, seven primary themes or patterns emerged from the habitat

selection analyses:

1) Habitats that included Eleocharis figured prominently and most consistently in the

selection preferences of all species. Nymphaea, Panicum, cattail, and periphyton

were common additional elements. This trend suggested that emergent vegetation

of moderate stature and habitats of moderate structural complexity were important

to foraging wading birds.







2) Rhynchospora, Eleocharis, and grassy patches fringing the southwestern Melaleuca

and cattail/sawgrass stands provided important foraging habitat at lake stages

above 4.3 m in 1989-particularly for Little Blue Herons-and in 1992-particularly

for ibises.

3) Mixed-grass habitats dominated by Panicum repens on Indian Prairie attracted

relatively large numbers of all species in early 1989 with lake stages above 4.3 m.

The reflooding of these habitats in 1992 drew strong positive selection from Great

and Snowy Egrets.

4) In 1991 and 1992, successional habitat complexes such as SCWP, CNWP, WCNP,

and SD-most of which emerged in the wake of the drought and fires-were heavily

used, but rarely elicited positive selection. Smaller patches of Polygonum and

those featuring diverse, mixed-stature assemblages were the primary attractors in

these habitat matrices. Ibises were also attracted to areas of decaying vegetation,

perhaps in response to high invertebrate populations.

5) Habitats dominated by cattail also ranked high in terms of volume of use, but usually

did not elicit positive selection. Again, thick cattail (and thick sawgrass) was not

the focus of attention; instead, interspersed patches of Eleocharis, Nvmphaea,

grasses, and Polygonum were the real attractors.

6) Although selection preferences varied through the study and attraction to inter-mixed

patches of preferred vegetation sometimes led to positive selection, the most

heavily wooded habitats-willow dominated WCNP and Melaleuca-dominated

MCSE-were usually underutilized relative to their abundance.

7) Habitats dominated by submerged vegetation ranked high in overall selection

preference. Maximum attraction occurred when the lake stage was low, but Snowy

Egrets, in particular, foraged over deep-water submerged vegetation habitats by







using floating mats for support and by using aerial foraging techniques. Overall,

Tricolored and Little Blue herons showed the weakest selection for submerged

habitats. The most open lakeward-fringe habitats were always underutilized, but

selection approached neutral among the larger species at the lowest lake stages.


General Discussion and Conclusions


Analytical Caveats


The habitat selection analyses could have been improved in several ways.

First, it would have been ideal to have vegetation maps that corresponded to each

individual bird survey, rather than just annual maps. The annual maps did not

adequately represent intra-seasonal changes in vegetation communities that occurred

in response to the drought, freezes, and fires. Unfortunately, the cost of the satellite

images and the processing time necessary to produce bimonthly maps were

unreasonable. Second, a finer spatial-scale of resolution for bird counts, that matched

the resolution of the vegetation maps, would have been preferable. However, this

would have been feasible over only a very restricted area, and I was primarily

interested in documenting macro- rather than microscale patterns of distribution. Third,

reclassifying a classified image--i.e., using FASTCLUS to aggregate the original

classes of vegetation-might have introduced unnecessary biases such as shifted

transition zones and hodgepodge assemblages. It might have been better to classify

the raw satellite images at a spatial scale more compatible with the bird data, but the

pixels in the images and my survey grid cells still would not have been properly aligned.

Thus, many of the survey grid cells would still have enclosed more than one type of

vegetation, and some degree of reclassification would still have been necessary.







Furthermore, the original classification was carried out by colleagues with other goals in

mind besides wading bird analyses (see Richardson and Hamouda 1994, Richardson

and Harris 1994, and Richardson et al. 1994). Regardless, I believe that visual

comparisons of the FASTCLUS derived vegetation maps (Figs. 2-13, 2-14) and the

original classified satellite maps shown in Richardson and Harris (1994) reveal little

reason to question the appropriateness of the former. The important limitation to

acknowledge, however, is that the coarse spatial scale of the bird and vegetation maps

and the coarse temporal scale of the vegetation maps did obscure fine-scale details of

habitat selection. In the results section I pointed out several specific instances where I

felt important trends were so obscured, and I emphasize that readers should not draw

strong conclusions about implied associations with individual plant species. Instead,

readers should focus on implications related to the coarse-scale structural

characteristics of the various vegetation assemblages.

A more important issue, is that habitat selection responses are simultaneously

functions of both hydrology and habitat structure. Therefore, analyses should account

for the interactive effects of hydrology and vegetation types. I first attempted to employ

log-linear models to analyze the interactive effects of grid-cell-level hydrology (data

derived from the OKEEHYDRO model; Richardson and Hamouda 1994) and vegetative

cover on the distribution of birds, in much the same way Hoffman et al. (1994) did for

their Everglades data. Unfortunately, even with 507 cells I was unable to retain a

reasonable number of vegetation, bird-density and water-depth categories and still

produce a dense enough data matrix to render an estimable model. In other words,

there were too many combinations associated with too few non-zero bird counts. Thus,

for purposes of this manuscript I was forced to consider hydrology and vegetation

separately.







The systematic and repetitive nature of my transect sampling scheme presents

a possible conflict with a primary assumption of most parametric statistical tests; i.e.,

the value of any individual observation must be independent of all other such values. It

is possible that some spatial correlation structure exists in the cell-based data that

least-squares analyses might not adequately account for. Similarly, when I extended

the analyses across multiple surveys, a temporal correlation structure might have

arisen. The flocks I observed on the lake typically covered much less than a 1 km2

area, and the spatial resolution of the surveys probably surpassed most birds' frame of

reference relative to selecting specific patches of foraging habitat. Moreover, while

patterns of habitat use definitely shifted within seasons, the available vegetation maps

did not. Accordingly, annual bird abundance datasets were most appropriate. Thus,

for purposes of developing this manuscript, I explicitly ignored potential spatial or

temporal correlation problems. However, to insure conservative interpretations, I used

only simple X2 tests of independence to establish the significance of observed habitat

selection trends, and I considered only those cases with X2 P < 0.001 as truly

significant.

I am proceeding with a more complex analysis that incorporates fully integrated

sets of hydrologic and vegetation data, and explicitly accounts for potential spatial and

temporal correlations. This manuscript therefore represents the results of employing

relatively straightforward, standard graphical and statistical techniques to analyze the

patterns of habitat selection observed during the study, and will serve as a template

against which I will compare the results of the more complex statistical approach.







Foraging Habitat Selection


Hydrologic influences. Clearly, hydrodynamics play a dominant role in

determining the distribution, accessibility, and quality of foraging habitat available to

wading birds around the lake, as well as in other regions (Kushlan et al. 1975; Browder

1976; Ogden et al. 1980; Kushlan 1986a; Powell 1987; Bancroft et al. 1990).

Hydrologic trends influence wading bird foraging by imposing limits to accessibility in

the form of deep-water and no-water extremes. At lake stages above about 4.6 m, the

lake is one continuous pool and little habitat is available to foraging birds. At the other

extreme, with lake stages below about 3.7 m, most of the emergent littoral zone dries

out and foraging is restricted to a limited but highly productive portion of the lake

consisting of outer-fringe mixes of Panicum hemitomon, cattail and Scirpus, lotus and

Nymphaea, and particularly submerged vegetation. Within these limits, the patten of

surface-water fluctuation influences prey density and therefore wading bird foraging

efficiency (Kushlan et al. 1975; Kushlan 1976a, b, 1979, 1986a; Frederick and Spalding

1994). The unique feature of the lacustrine Lake Okeechobee ecosystem is that during

even severe droughts, water and productive beds of submerged vegetation remain and

can support prey organisms and foraging birds at a time when most of the surrounding

palustrine systems cease to provide quality foraging habitat.

Hydroperiod is another aspect of hydrology that affects wading birds. The

productivity of important prey species depends on a sufficiently long hydroperiod

(Loftus et al. 1990; Loftus and Eklund 1994). Hydroperiod also determines the

distribution and health of vegetation communities (Pesnell and Brown 1977; Richardson

et al. 1994), which influences the distribution of prey (Chick and Mclvor 1994), the

structure and therefore accessibility of foraging habitats, and the distribution and health

of willow, the primary nesting substrate (David 1994a). The habitat selection data







presented here suggested that at least some species were attracted to areas featuring

complex mixes of vegetation. Seasonal water level fluctuations and interannual

differences in hydropattem, combined with periodic large-scale disturbances such as

fires and freezes, may ensure the development of diverse arrays of vegetation

(Gunderson 1994) and thus provide a wide variety of foraging opportunities for wading

birds.

Preference for emergent vegetation of moderate stature and structural

complexity. When presented with a choice of foraging habitats at moderate lake

stages, all species considered in this study tended to select patches of emergent

vegetation of moderate stature and species/structural diversity. Eleocharis-mix

habitats, particularly around and in Moonshine Bay and on northwestern Indian Prairie,

were consistently selected by most species. Rhynchospora-dominated and mixed-

grass (primarily Panicum) habitats in higher-elevation areas attracted more birds at

higher lake stages (e.g., above 4.3 m). Flats of Polygonum proliferated in the wake of

the freeze-drought-fire cycle and attracted much foraging activity from 1990 onward.

Nymphaea and sparse to moderate density cattail were common additions to the

preferred mixtures. In contrast, all species generally avoided or underutilized densely

wooded or shrubby habitats and dense stands of tall emergents like cattail and

sawgrass. The analyses occasionally indicated selection for cattail/sawgrass (C,

CSNE), willow-dominated (WCNP), or Melaleuca-dominated (MCSE) habitats, but the 1

km2 scale of resolution obscured the fact that the patches of habitat actually chosen

rarely included the areas where these species occurred in dense stands. These results

are generally consistent with information gathered on the lake by Zaffke (1984) from

1977-1981. Nearly 50% of all the White Ibises, Great Egrets, and Snowy Egrets

counted were foraging in Eleocharis or Rhynchospora-dominated habitats at moderate







lake stages. Mixed-grass habitats were used more at higher lake stages and

accounted for the second highest volume of use. Zaffke also noted that submerged

beds were used heavily during 1981 when a drought occurred and the stage dropped

to near 3.0 m. Throughout his study, habitats dominated by thick cattail, willow, and

Cephalanthus were used comparatively little.

Collopy and Jelks (1989) documented patterns of habitat use for some of the

same species (Great Blue Heron, Great Egret, White Ibis) in a matrix of mixed riparian,

open-pocket, and forested wetlands in Sarasota County, Florida. They also found that

all species tended to forage in wetlands featuring either emergent species of moderate

stature and density such as Panicum and Rhvnchospora, Pontedaria and Sagittaria,

and Polygonum, or floating emergents such as Nymphaea, Nvmphoides, and water

hyacinth. Conversely, all species tended to avoid woody habitats and dense stands of

sawgrass and Hypericum. Hoffman et al. (1994) studied the distribution of the same

species in the Everglades and adjacent areas. Their data were resolved to a coarser 4

km2 spatial scale and they considered only five classes of vegetation. Nonetheless,

their results also generally concur with those of this study. Their most consistent

finding was that all species tended to avoid "dense grass" habitats (i.e., primarily

sawgrass). Otherwise, the evidence suggested a tendency for most species to prefer

more open, heterogeneous mixes of emergent wet-prairie and slough vegetation. Such

patches were often embedded in a matrix of tree islands, and typically were associated

with transitional hydrologic conditions.

Dense, tall vegetation probably physically impedes a bird's movement and may

also reduce a bird's or a flock's ability to remain vigilant against predators and

disturbances. That the latter may be important became evident during my attempts to

approach feeding flocks on an idling airboat for the purpose of conducting







observations. The experience suggested that individuals and groups of birds-

particularly ibises--flushed more quickly from foraging positions within taller vegetation

that restricted their view of the boat than when they were in more open areas. One

might also speculate that the densest cattail and sawgrass habitats might harbor

reduced prey populations; however, throw-trap sampling in dense sawgrass at

Loxahatchee National Wildlife Refuge revealed relatively high densities of small fish,

primarily Gambusia holbrooki (C. F. Jordan pers. comm.). It therefore seems most

likely that physical constraints render these habitats less attractive to foraging birds.

Use of submerged vegetation and open-water habitats, and the lake as a

regional drought refuge. When surface-water levels dropped sufficiently to expose

beds of submerged vegetation, all species took advantage of the new foraging habitat.

Perhaps the strong selection of these habitats indicated for all species was a matter of

default, since most of the upper marshes were dry. Moreover, the large influx of birds

at this time probably in part reflected the fact that foraging conditions in the Everglades

and other surrounding wetlands were poor due to the drought (P. C. Frederick and G.

T. Bancroft pers. comm.). However, prey sampling confirmed that these habitats,

especially thick beds of Hydrilla and Vallisneria, contained very dense concentrations of

small fishes and grass shrimp (Mclvor and Smith 1992; Chick and Mclvor 1994).

Observations also revealed relatively high foraging success and caloric intake rates

(Mclvor and Smith 1992). It was not possible to determine to what degree the high

densities of prey were the result of drought-related concentration effects. Regardless,

these habitats did represent a rich food source for wading birds.

The data for the less common Wood Storks and Roseate Spoonbills provided

the strongest evidence that an unusual number and variety of wading birds were drawn

to the lake during the drought periods. Roseate Spoonbills generally remain in coastal








habitats, but small numbers of dispersing young frequently visit the Lake Okeechobee

region (Palmer 1962; and see Zaffke 1984). However, during the drought in both 1989

and 1990, an unusually large group of up to 75 immature birds remained at the lake for

up to 10 weeks. Smaller groups foraged with mixed feeding flocks typically in open

flats between mats of submerged vegetation, but the entire group established a primary

loafing and roosting area along the banks of one canal.

Wood Storks have nested in small numbers around the lake (Zaffke 1984;

David 1994a) and are regularly seen foraging in the area. However, the data gathered

in this study and by Zaffke (1984) indicate that Wood Storks are attracted to the lake in

large numbers only when the stage is dropping and below 4.6 m (nearer to 4.3 m in

most cases). Both studies indicated that the largest numbers occur during severe

drought periods (1981 during Zaffke's study and in 1990 during this study, with lake

stages below 3.0 and to 3.2 m, respectively). The large flocks of Wood Storks seen

during this study typically included both immature and adult birds. This suggests that

the lake was a more universally important resource for Wood Storks than for Roseate

Spoonbills (also see Browder 1976). The winter-spring recession of 1991/92 did not

attract storks to the lake, probably because water levels did not recede sufficiently.

However, differences in long term antecedent hydrologic conditions may have

influenced the trend (sensu Frederick 1993). The 1992 recession followed a two and a

half-year drought cycle, whereas the 1989 recession followed a three-year period of

relatively high water. The abundance of medium to large sunfishes-a frequent target

of foraging Wood Storks on the lake (unpubl. data)-probably increases with prolonged

flooding (Loftus et al. 1990). The increase in productivity would then lead to richer,

more attractive concentrations as surface-water levels receded.







All species took advantage of prey resources available within and around

exposed beds of submerged vegetation during the drought. However, Snowy Egrets

appeared to profit from submerged vegetation and floating-mat habitats even during

non-drought periods. The species' use of aerial foraging techniques-often involving

small, floating water hyacinth islands as launching platforms-was the key to its

successful use of the habitat (also see Edelson and Collopy 1990). This tendency for

Snowy Egrets to use more open habitats was also documented by Murdich (1978), who

studied several species in a large prairie-wetland ecosystem in Alachua County,

Florida. He found that Snowy Egrets, more often than Great Blue Herons, Tricolored

Herons and Great Egrets, foraged mostly in open drainage canals and pools, flooded

diked areas, and other areas of relatively low vegetation density. Similarly, Jenni

(1969) studied several species around a small lake on the University of Florida campus

in Gainesville, and commented that Snowy Egret's tended to feed "in open areas and

along edges of openings." Kent (1986a) found that Snowy Egrets foraged least

efficiently in comparison to Tricolored and Little Blue herons (Snowy Egrets estimated

to require an average daily foraging period two times longer). Snowy Egrets may

therefore experience greater pressure to reduce their foraging-niche overlap with other

species.

In this study, Great Blue Herons showed a strong tendency to forage in open

habitats, including open flats at low lake stages-sunfish bedding areas--and open

edges along canals and on the rocky east-side at a variety of lake stages. Evidence

gathered under nests and during observations of feeding flocks confirmed that Great

Blue Herons often consumed small bass and large sunfish, shad and shiners, types of

prey rarely captured by other species (unpubl. data). Otherwise, it is not surprising that

most species underutilized the most open, outer fringe habitats. Open-water areas are







less frequently populated by the smaller fishes and invertebrates that most wading

birds appear to prefer (Wemer et al. 1977), and there are no physical barriers in open

water to restrict fleeing prey, potentially making it more difficult for wading birds to

secure available prey. At the height of the drought, most species (Great Egrets, White

Ibises, Snowy Egrets, and Wood Storks, in particular) occasionally foraged in open or

sparsely vegetated outer fringe habitats, but except for Great Blue Herons, always only

in large flocks. The presence of large flocks appeared to elicit chaotic movements

among the fishes in the midst of the group, and this provided at least temporarily

profitable foraging opportunities (pers. observ.).

Jenni (1969) noted that Tricolored Herons also tended to forage along the

edges of canals and steep banks, and from floating or submerged vegetation in deeper

water. This study indicated that Snowy Egrets were more likely to forage in submerged

and mat vegetation habitats, but following the drought, Tricolored Herons continued to

select SUB habitats longer into the 1991 season than Little Blue Herons and Great

Egrets. Furthermore, although the selection analyses did not adequately reflect such

details, solitary Tricolored Herons-like Snowy Egrets-often used floating lotus leaves

to gain access to prey-rich Hvdrilla beds. Murdich (1978) found that Tricolored Herons

were the most generalized in their use of habitats in comparison to Snowy Egrets,

Great Egrets, and Great Blue Herons. According to the results of this study, this was

not the general rule at Lake Okeechobee; however, aerial surveys of both Tricolored

and Little Blue herons were biased towards those occurring in mixed flocks. Both

species, but particularly Tricolored Herons, often foraged alone in a wider variety of

habitats, but went undetected on aerial surveys. As noted by Jenni (1969) as well, I

often observed solitary Tricolored Herons foraging along the edges of canals.







Great Egrets, like Snowy Egrets, periodically used aerial foraging techniques to

access deep open-water habitats dominated by primarily Potamogeton and Hydrilla.

However, unlike Snowy Egrets, Great Egrets appeared to require the help of persistent

winds for lift and support. The 1990 spring season featured such winds, and a small

number of birds that nested in the center of Fisheating Bay appeared to profit from

aerial foraging. They fed their nestlings an unusually rich collection of larger fishes,

and their nesting success and productivity were among the highest for the study

(Collopy and Smith 1991). Edelson and Collopy (1990) found that most common

Florida wading bird species would use aerial techniques when presented with a rich

enough source of prey. In that case, such conditions were present at a hyper-eutrophic

lake where large schools of fish often gathered at the surface. Birds were able to take

advantage of numerous launching points, including elevated woody perches around the

perimeter and fish-net floats strung across the lake.


Management Implications


Management of exotic and nuisance vegetation. The control of exotic and

nuisance vegetation is a major management issue in Florida, and several species are

of concern at Lake Okeechobee. The first is Melaleuca (see Bodle et al. [1994] for a

general discussion); according to this study, wading birds do not use the trees for

nesting and do not forage within typically dense stands of the species. The further

expansion of Melaleuca following the drought and fire cycle only served to eliminate

productive Eleocharis and Rhynchospora flats that previously occurred at the margins

of the Melaleuca forests along the western margin of the lake. The fact that Melaleuca

was able to expand its range following the drought was one of the few apparent

disadvantages of the drought.







The second species of concern is cattail, a native species but one that has

expanded its coverage and dominance in many areas in response to the addition of

nutrient-laden agricultural runoff and unnatural manipulation of hydrologic regimes

(Davis 1989). Again, this study suggests that foraging and nesting wading birds do not

benefit from dense stands of cattail. Unlike for Melaleuca, however, the drought/fire

cycle at least temporarily controlled the spread of cattail in many areas, and this

provided the opportunity for productive stands of Polvyonum to develop and persist

until the lake stage returned to high levels.

Another exotic species of concern is Panicum repens, which typically dominates

the northwestern Indian Prairie region. This study indicated that wading birds,

particularly Great Egrets, frequently foraged in this area at higher lake stages (also see

Zaffke 1984). My observations suggested that when water levels in the Panicum

repens flats remained low for extended periods, thick, continuous-cover mats of dead

wrack and live plants formed. At this point, other species of vegetation were excluded,

access to the underlying surface water was restricted, and wading birds foraged in

other habitats. The fires that bumed through the Indian Prairie area in 1990 and 1991

removed the excess wrack, and regeneration after reflooding led to comparatively

sparse stands that attracted foraging wading birds. Similarly, conditions in late 1988

and early 1989 suggested that extended periods of high water also produce sparser

stands of Panicum, and therefore result in attractive foraging habitat for wading birds

once water levels decline to suitable depths. Thus, periodic low-water/fire cycles or

extended periods of high water may be necessary to ensure the suitability of Panicum

repens habitats for foraging wading birds.

The more controversial issues concem management strategies to deal with

Hydrilla and water hyacinth. Currently, the SFWMD oversees an extensive program of







herbicide spraying to control the spread of water hyacinth, Hydrilla, and water lettuce,

particularly in waterways and bays frequented by boaters and fishers (Langeland and

Joyce 1987). Sampling has indicated that thick beds of Hydrilla on the lake support

very high densities of small fishes and grass shrimp (Mclvor and Smith 1992; Chick and

Mclvor 1994). Moreover, when they are accessible, these habitats attract large

numbers of foraging wading birds. Although other native submerged species such as

Vallisneria also harbored large populations of small fish (Mclvor and Smith 1992) and

should probably be emphasized more in management plans, one cannot ignore the

benefits of Hydrilla in this regard. Moreover, only Hydrilla forms the thick, topped-out

mats that allowed wading birds access to otherwise inaccessible deep-water habitats.

Of course, it is exactly this feature that annoys boaters trying to access their favorite

fishing holes, and excessively thick and extensive beds may hinder the growth and

development of large game-fish species (Colle and Shireman 1980).

In the case of water hyacinth, my observations suggested that wading birds

rarely forage within healthy mats, but often use smaller clumps for support while

foraging at the edges and launching aerial attacks over the adjacent submerged beds

(also see Murdich 1978). My observations also indicated that several species,

particularly Snowy Egrets, flocked to large, recently-herbicided mats of water hyacinth.

Such sites typically supported activity for only a few hours or less, and birds were first

attracted within hours of the herbicide spraying. One might speculate that the food

obtained at this time was not entirely safe for consumption given the recent application

of herbicides and the tendency for wading birds to consume quantities of vegetation-

particularly abundant Lemna-with their target prey (unpubl. data). Nonetheless,

observations of foraging birds confirmed typically high capture rates and at least

moderately high caloric intake rates (unpubl. data). Thus, extensive mats of healthy








water hyacinth do not appear to benefit wading birds. However, the presence of

smaller mat-islands provides access to otherwise inaccessible open-water and

submerged vegetation habitats, and the relationship between herbicided mats and

wading bird foraging is an intriguing one that demands further study.

Hydrologic and related concems. The results of this study suggest that to

attract large populations of foraging wading birds to the lake, two primary criteria must

be met: 1) lake stages should be managed to remain at or below about 4.6 m to insure

the availability of some foraging habitat; 2) management schedules should encourage a

steady, moderate-paced spring recession that coincides with the normal Florida dry

season and usual wading bird nesting season (i.e., January June). The ideal

magnitude for a spring recession is less certain. Evidence collected during this study

suggested that the most productive foraging conditions occurred when the recession

was sufficient to expose beds of submerged vegetation to foraging birds (i.e., lake

stages below about 3.7 m). David (1994b) found that, during the 12 years prior to this

study, combined-species foraging populations usually increased in response to

dropping lake stages if the drying rate exceeded -0.15 m per 60 days. In this study, the

largest foraging populations were attracted during 1989 and 1990 at the height of the

drought, with 60-day antecedent drying rates of up to -0.51 m in June 1989 and -0.42

m in May 1990. However, the lake stage declined at a maximum rate of -0.31 m over

the 60-days prior to the end of May 1992, but foraging populations did not increase

much because the lake stage dropped below 4.6 m for only a short time. Moreover,

most species temporarily emigrated from the lake during reversals in the 1989 drying

trend, despite higher overall recession rates than in 1990. This suggests that the

steadiness of the recessions may be more important than the rate or magnitude of

decline. Further support for this contention was provided by the logistic regression







analyses of nest failure summarized in Chapter 5; the surface-water trend variable

"days-of-rising-water" consistently exerted more influence in the regression models

than "drying-rate" variables.

Most species achieved the highest per nest productivity at the height of the

drought in 1990 (Chapter 4). Snowy Egrets, Tricolored Herons, and White Ibises also

initiated the most nests 1990. However, both Great Blue Herons and Great Egrets

initiated more nests in 1989 and 1992 than in 1990 or 1991. Moreover, nest

productivity in 1989 among the latter species was only slightly lower than in 1990, and

Great Blue Herons also achieved comparable production levels in 1992. This

suggested that the combination of high winter water levels and an early spring

recession was the scenario that most attracted these species for nesting.

Little Blue Herons were also unusual in showing a steady decline in nest

numbers through the study, with the highest effort and productivity in 1989 (Chapter 4).

The fact that Little Blue Herons did not initiate more nests again in 1992 with the return

of deeper water, suggests that long-term antecedent hydrologic conditions may have

been the key to their response. During this study, Little Blue Herons relied more

heavily than other species on grass shrimp for feeding their young (Mclvor and Smith

1992), as has been demonstrated elsewhere (e.g., Rodgers 1982). Loftus et al. (1990)

and Loftus and Eklund (1994) determined that short-hydroperiod marshes in the

Everglades support depauperate populations of most aquatic organisms compared to

nearby longer-hydroperiod marshes. One of the most dramatic differences was shown

for grass shrimp (Loftus et al. 1990). Thus, extended periods of high or at least

moderate depth water may be essential for a subsequent recession to produce

advantages for nesting Little Blue Herons.







Hydrologic conditions also play a critical role in determining suitable locations for

nesting colonies. At the height of the drought most interior colonies were abandoned in

preference for island colonies still surrounded by water (Chapter 4). Two reasons for

the shift may have applied: 1) the nesting birds may have relocated closer to favorable

submerged-bed foraging grounds; 2) the move to flooded islands may have precluded

invasions by terrestrial nest predators. Regardless, all species used a wider variety of

colony sites in 1989 and especially 1992 than in 1991 and especially 1990.

Consequently, striving for maintenance of moderate lake stages (e.g., 3.7-4.3 m) rather

than lower lake stages may provide more diverse opportunities for both nesting and

foraging, and may ultimately ensure a more flexible and resilient system.

There is also the issue of the relatively consistent bimodal distribution of

numbers of ibises each year. The trend suggested that some birds attracted to the lake

to feed early and late in the year went elsewhere to nest, which in turn suggests that

nesting opportunities at the lake were in some way limited. David (1994a) documented

a long term decline in nest numbers at the lake over the period 1977-1988, with White

Ibis a prominent example. He attributed the decline to the negative influence of higher

stage-regulation schedules on the availability of foraging conditions and on the health

of willows. A decline in the health of willows-the primary nest substrate-might have

been partly responsible for the abandonment of the King's Bar nesting colony, which

historically was the most consistently occupied and largest colony on the lake. Perhaps

the ibises in particular have not found other locally available sites as appealing.

The best active management strategy will likely involve cycles of manipulation,

with planning based on a multi-year window approach rather than a static annual-

pattern approach. The typical annual pattern might include a fall-winter peak of 4.6-4.9

m drawn down 0.3-1.0 m from January through June. However, once every several







years the lake stage might be kept lower through the winter and allowed to drop to the

3.0-3.6 m range during the dry season to expose prey-rich submerged beds, invigorate

essential willow stands, and allow fires to burn away cattail and Panicum wrack, recycle

nutrients, and encourage the establishment of attractive successional vegetation

complexes. Similarly, periodically encouraging multi-year periods of higher stages

might enhance the productivity of fish and macro-invertebrate populations. Placing a

general emphasis on maintaining long-hydroperiod marshes (but not necessarily deep

water) may also be a good strategy with respect to managing for the endangered Snail

Kite. Kites apparently prosper when upper-elevation Eleocharis flats remain flooded

year round, because this insures healthy populations of their primary prey, apple snails

(Pomacea paludosa; Bennetts et al. 1994). Moroever, at lower lake stages the kites

often nest in relatively unstable cattail and Phragmites toward the outer margins of the

littoral zone, rather than in trees at the periphery of the lake. There nest success when

cattail or phragmites is the nest substrate tends to be low because many nests collapse

for lack of proper support (Bennetts et al. 1994). Because both high and low lake

stages encourage ecosystem elements favorable to wading birds, the wisest and least

costly management strategy probably will be one of benign neglect; i.e., allow natural

hydrologic fluctuations to run their course and institute significant control measures only

to avoid environmental or economic disaster.

A final critical point is that the Lake Okeechobee ecosystem must not be studied

or managed as an isolated system, particularly not where wading birds are concerned.

Wading birds are highly mobile species, and what happens at Lake Okeechobee is

always at least in part related to the quality of conditions in the Everglades and other

southern Florida wetlands, if not to conditions throughout the southeastern United

States. Again, the occurrence of so many birds on the lake in mid-1989 and mid-1990





81


might have had as much to do with foraging conditions being poor in the Everglades

and in other regions as it did with conditions being favorable at the lake. Moreover,

conditions at the lake need not be ideal for wading birds every year, as long as regional

management plans help insure the availability of suitable foraging and nesting habitat

somewhere in the southern Florida region. Accordingly, a regional management

perspective is essential.















CHAPTER 3
FORAGING FLIGHT DYNAMICS AND PATTERNS OF HABITAT USE WITH
RELATION TO NEST PRODUCTIVITY, AND FORAGING SOCIABILITY AMONG
FOUR NESTING SPECIES

Introduction


For most species of birds, the energetic and logistic demands of nesting peak

when chicks hatch and adults are challenged with the task of supplying food to their

rapidly growing chicks (e.g., Kahl 1964; Kushlan 1981). The added strain may be

particularly great for species with nidicolous (nest-bound) young, because the adults

must operate as "central-place foragers" (sensu Orians and Pearson 1979) and make

repeated trips to and from the nest each day. If food is plentiful near the nest, then the

added energetic burden may not be great, but it will increase as the distance to

profitable foraging grounds increases. If the distance is too great, adult birds might

choose to abandon their nesting effort because the energetic cost of travel exceeds the

benefits of high quality foraging habitats (Wittenberger and Hunt 1985; Bryan and

Coulter 1987). Studies of ciconiiform wading birds in the Florida Everglades have

indicated a positive correlation between colony abandonments and foraging-flight

distances (Frederick and Collopy 1988; Bancroft et al. 1990, 1994). However,

Frederick and Spalding (1994) later suggested, based on energetic calculations, that

the abandonments probably occurred because increasing flight distances reduced the

rate and total amount of prey delivery to nestlings, not because the adults were

physically taxed. Regardless, any study of the causes of reproductive failure among







wading birds should consider how variation in both the quality and cost of travel to

foraging habitats affects the species studied (Bancroft et al. 1994; Frederick and

Spalding 1994; Ogden 1994).

As part of a 4.5-year study of wading bird nesting and foraging ecology at Lake

Okeechobee, Florida (see Fig. 3-1 for a geographic overview of the region), I

documented foraging-flight characteristics for 960 individuals of four species of

commonly nesting wading birds at times when nestlings were being fed. The study

species were Great Egrets, Snowy Egrets, Tricolored Herons, and White Ibises. In this

chapter, I summarize my findings on flight times and distances, and compare these

statistics with estimates of nest success and nestling production (see Chapter 4 for a

detailed treatment of nesting ecology). The research hypothesis was that, if the

energetic demands of repeated foraging flights are an important constraint, then

increasing flight times and distances should translate to decreasing nest success and

productivity.

I also analyze the foraging flight dynamics and habitat choices of nesting adults

in relation to hydrologic trends. Managing lake water levels is a primary concern for

resource managers in the region (Aumen 1994). The lake serves as a regional flood

control and water supply facility, supports a multi-million dollar commercial and sport

fishing industry, and is an important resource for many species of both game and non-

game wildlife (Aumen 1994). The demands of these interests often conflict, and so

developing balanced approaches to management is a subject of considerable debate.

Wading birds are considered bioindicators of change in hydrological and ecological

conditions in wetland ecosystems (e.g., Custer and Osbom 1977; Curry-Lindahl 1978;

Hafner and Britton 1983; Bildstein et al. 1990; Kushlan 1986b, 1993; Ogden 1994), and

therefore understanding how wading birds respond to fluctuating hydrologic conditions








Okeechobee


Kissimmee


Creek


Okee-Tantie
Island


emergent marsh


Fisheating ,
Creek r


canals


/


Clewiston Spit


River


Lake
Hicpochee


Clewiston


Belle Glade


Skm
5 km


Figure 3-1. A geographic overview of the Lake Okeechobee area showing the
locations of nesting colonies used by the four study species, and showing the locations
of lake stage and rainfall gauging stations.


SColonies from which data were collected
* Other colonies used by one or more of the four study species
SLake stage gauging station
* Rainfall gauging stations







is often a focus of management concerns. Hydrologic trends determine water depths

and influence both the distribution of vegetation (Richardson and Harris 1994) and the

productivity of prey populations (Loftus et al. 1994), and therefore are a primary

determinant of wading bird foraging efficiency and consequently nesting success

(Bancroft et al. 1994; Frederick and Spalding 1994; Ogden 1994). The 1989-1992

study period spanned a two-year drought (Fig. 3-2) that resulted in a 10-year record low

lake stage in May-June 1990 (3.2 m NGVD; NGVD = National Geodetic Vertical Datum

of 1929, essentially equivalent to height above mean sea level; for convenience, I will

henceforth omit the NGVD qualifier). An unusually high estimated total of over 50,000

foraging wading birds congregated on the lake at this time (Chapter 2). Most nesting

species also achieved the highest reproductive success for the study in 1990 (Chapter

4). The probability of nest failure was generally lowest when surface-water levels were

declining steadily and the mean lake stage was low to moderate (3.5-4.1 m; Chapter 5).

Studying foraging-flight dynamics added another dimension to the picture and

highlighted unique aspects of the foraging ecology of nesting adults. In particular, the

study enabled a determination of whether or not the pronounced hydrologic fluctuations

contributed to shifts in foraging habitat use and changes in foraging flight distances that

might have affected the energetic of nesting.

Lake Okeechobee is nearly surrounded by a large earthen dike (Fig. 3-1).

Thus, there is an abrupt transition from extensive emergent marsh (400 km2),

submerged vegetation, and open-water habitats inside the dike to a wide variety of

natural and artificial wetland habitats outside the dike. The latter include: the

Kissimmee River and Fisheating Creek floodplains (see Fig. 3-1); diverse pocket and

slough wetlands interspersed with cattle pastures to the north and west; myriad

agricultural field ditches and canals to the south and east; and residential canals and













(wo) IleJu!edl AI!ea leBol uee/i

COD


(CAON w) a6elS aee-i Al!eO


z
0











LL
U-













Z
0
0

















z

0
r