Citation
Aquatic Vertebrate Usage of Littoral Habitat Prior to Extreme Habitat Modification in Lake Tohopekaliga, Florida

Material Information

Title:
Aquatic Vertebrate Usage of Littoral Habitat Prior to Extreme Habitat Modification in Lake Tohopekaliga, Florida
Copyright Date:
2008

Subjects

Subjects / Keywords:
Aquatic habitats ( jstor )
Frogs ( jstor )
Lakes ( jstor )
Sirens ( jstor )
Snakes ( jstor )
Species ( jstor )
Turtles ( jstor )
Vegetation ( jstor )
Vertebrates ( jstor )
Water depth ( jstor )
Lake Tohopekaliga ( local )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Embargo Date:
12/18/2004

Downloads

This item has the following downloads:


Full Text












AQUATIC VERTEBRATE USAGE OF LITTORAL HABITAT PRIOR TO EXTREME
HABITAT MODIFICATION IN LAKE TOHOPEKALIGA, FLORIDA















By

ANN MARIE MUENCH


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE

UNIVERSITY OF FLORIDA


2004

































Copyright 2004

by

Ann Marie Muench


































This document is dedicated to my parents, Joseph F. and Mary K. Muench, whose love
and support have strongly contributed to my academic, professional, and personal
growth.















ACKNOWLEDGMENTS

I would like to thank my major advisor, Wiley Kitchens, for taking me on as a

graduate student. He was a critical source of expert advice and encouragement, and was

always accessible for consultation. I also thank my committee members, Madan Oli and

Lauren Chapman, whose academic instruction and critical analysis of this thesis are much

appreciated. My coworkers also contributed much to my education, and I am thankful.

Funding for this research was provided by the Florida Fish and Wildlife

Conservation Commission (FFWCC). From this agency, Duke Hammond helped

immensely with the direction of the study, and provided critical feedback on progress

reports that we provided to the commission. The staff of the Kissimmee, FL, office of

the FFWCC was helpful in facilitating our field work at Lake Tohopekaliga. Bobbi Jo

Cromwell from the Osceola County Department of Parks and Recreation allowed us to

store all of the crayfish and minnow traps on Makinson Island in Lake Toho.

The field work for this study was conducted through the time of many dedicated

students and technicians from the Florida Cooperative Fish and Wildlife Research Unit.

These stalwart coworkers included (in alphabetical order) Scott Berryman, Stephen

Brooks, Janell Brush, Brenda Calzada, John Davis, Jamie Duberstein, Bruno Ferreira,

Joey Largay, Kristianna Lindgren, Samantha Musgrave, April Norem, Derek Piotrowicz,

Laura Pfenninger, Erik Powers, Vanessa Rumancik, John David Semones, Micheala

Spears, Chris Tonsmeire, Paul Traylor, Zach Welch, and Christa Zweig.
















TABLE OF CONTENTS

page

A C K N O W L E D G M E N T S ................................................................................................. iv

LIST OF TABLES ............. ................... ......... .... ... ............... .. vii

LIST O F FIG U RE S ......................................................... ......... .. ............. viii

A B ST R A C T .......... ..... ...................................................................................... x

CHAPTER

1 M A IN IN TR OD U CTION .................................................. .............................. .

L ake E cosy stem .....................................................................................................1
Stu dy A rea ......................................................................... . 2
R research Objectives.............................................. 8

2 DESCRIPTIONS OF FOCAL SPECIES .................................. ...............10

A quatic V ertebrate H habitat ................................................................................... 10
Fish Species ................................. ........................................... 11
C entrarch id s (su n fish ) ................................................................................... 1 1
E x o tic c catfish ................................................................ ............................... 1 3
H erpetofaunal Species ............................................................13
A m phibians ............................................. 13
R reptiles .................................................................................... ................... 15

3 AS SEMBLAGE WITHIN THE Pontederia cordata COMMUNITY ................. 17

In tro du ctio n ...............17..............................................
Field Methods ......... ......... ......... .. ...............17
T rap D description s .............................................................17
W hole-Lake Sam pling .............. ........ ................ ............... ............... 20
A analysis M methods ..................... .................. ......................... 23
Trap C om prisons ...................... ............ .......................................... 23
Species R richness .................................. ...........................................23
Assemblage Composition ............................ ........ ............25
Influence of Temporal Gradients on Assemblage .......... ..................... 25
Proportion of Habitat Utilized by Focal Species ........................... ............. 27


v









R e su lts ...........................................................................................2 9
T rap C om prison s ......................... ............................ .. ......... .... ............29
S p e cie s R ich n e ss ........................................................................................... 3 1
A ssem blage Com position.................................... ......................... ............ ... 32
Influence of Temporal Gradients on Assemblage.............................................34
Proportion of Habitat Utilized by Focal Species....................................40
Discussion ..................................................41
T rap C om p arison s .............................. .... ...................... .. ........ .... ............4 1
S p e cie s R ich n e ss ........................................................................................... 4 1
A ssem blage C om position ................................................................................ 42
Influence of Temporal Gradients on Assemblage.................. .............. 44
Proportion of Habitat Utilized by Focal Species............................. ..............45

4 ASSEMBLAGE ACROSS VEGETATION COMMUNITIES ..............................64

In tro d u ctio n ............. ........... ... .................. ................. ................ 6 4
F ield M ethods ....................................................... 64
G rid and W eb Sam pling ............................................. ............................. 64
W hole-Lake Sam pling....................... ......................... ................. .. ............. 66
A analysis M ethods ............... .......... .. .......... .. ...... .. ................. .... ... ........... 67
Population Estimates and Movement for Herpetofaunal Species .....................67
Capture Success for Focal Species ......... .............. ................. ... .............. 68
R e su lts ....................... .. .. .............. .. .......................... ................ 7 0
Population Estimates for Herpetofaunal Species .............................................70
Capture success for Focal Species.................................... ....................... 72
D iscu ssion .................. ....... .... ....... ...... ......................... 74
Population Estimates for Herpetofaunal Species .........................................74
Capture Success for Focal Species .............. ..................... .............. 75

5 SUMMARY AND CONCLUSIONS................... .. .................. ...............88

Review of Aquatic Vertebrate Community Dynamics in Lake Tohopekaliga ...........88
A ir T em perature ........................................ ................. .... ..... .. 89
L ak e S tag e ...................................................... ................ 8 9
W ater D epth ............ ...... ....... ..... .............. .. ......... ..... ........... 90
V egetation C om m unity ............................................... ............................ 90
Population Size E stim ates ........................................... ............................ 90
Lake Tohopekaliga Habitat Enhancement.................... ............ ............... 91
Future Aquatic Vertebrate M monitoring Plans .................................. ............... 95

L IST O F R E FE R E N C E S ...................... ........................................................ .......... 97

BIOGRAPH ICAL SKETCH ....................... .......... ......... .................................. 105















LIST OF TABLES


Table page

3-1 All species captured in 2002, with species codes used in subsequent figures.........30

3-2 Species capture frequencies for the 0.6 and 1.3 cm mesh minnow traps ................31

3-3 Indicator species analysis results........................................ .......................... 33

3-4 Stress and instability results from all NMS ordinations.............................34

3-5 Percent of variance explained (r2) by environmental variables for each axis in the
vertebrate NMS with detection/nondetection data............... ....... ............... 35

3-6 Percent of variance explained (r2) for each axis by species in the vertebrate NMS
w ith detection/nondetection data ......................................................... ............... 35

3-7 Percent of variance explained (r2) by environmental variables for each axis in the
vertebrate N M S w ith count data........................................ ........................... 36

3-8 Percent of variance explained (r2) for each axis by species in the vertebrate NMS
w ith count data. .......................................................................36

3-9 Percent of variance explained (r2) by environmental variables for each axis in the
fish NM S with detection/nondetection data.......................................................37

3-10 Percent of variance explained (r2) for each axis by species in the fish NMS with
detection/nondetection data.......................................................... ................37

3-11 Percent of variance explained (r2) by environmental variables for each axis in the
fish N M S w ith count data............................................... .............................. 38

3-12 Percent of variance explained (r2) for each axis by species in the fish NMS with
c o u n t d ata .......................................................................... 3 8

3-13 Percent of variance explained (r2) by environmental variables for each axis in the
herpetofaunal NM S with count data.................................... ......................... 39

3-14 Percent of variance explained (r2) for each axis by species in the herpetofaunal
NMS with count data..................... ................... ............ 39

4-1 Grid sizes and population estimates by mark recapture methods. .........................71
















LIST OF FIGURES


Figure page

3-1 Crayfish and minnow trap in P. cordata habitat. .............................................. 47

3-2 Locations of 2002 P. cordata sampling transects in Lake Tohopekaliga ...............48

3-3 2002 Vertebrate species richness estimates by sample date..................................49

3-4 2002 Fish species richness estimates by sample date......................................... 49

3-5 2002 Herpetofaunal species richness estimates by sample date ............................ 50

3-6 NMS ordination of sample units in vertebrate species space using
detection/nondetection data.......................................................... ............... 51

3-7 NMS ordination of vertebrate species in sample unit space using
detection/nondetection data.......................................................... ................52

3-8 NMS ordination of sample units in vertebrate species space using count data........53

3-9 NMS ordination of vertebrate species in sample unit space using count data .........54

3-10 NMS ordination of sample units in fish species space using detection/
n on d election d ata ............ ........................................................................... .. ....... .. 5 5

3-11 NMS ordination of fish species in sample unit space using detection/nondetection
d a ta .............................................................................. 5 6

3-12 NMS ordination of sample units in fish species space using count data..................57

3-13 NMS ordination of fish species in sample unit space using count data .................58

3-14 NMS ordination of sample units in herpetofaunal species space using count data..59

3-15 NMS ordination of herpetofaunal species in sample unit space using count data ...60

3-16 Average and range of lake stage values by cluster................................................61

3-17 Average and range of air temperature values by cluster. .......................................61

3-18 Site occupancy estimates for focal fish species in spring 2002. ...........................62









3-19 Site occupancy estimates for focal fish species in fall 2002. ..................................62

3-20 Site occupancy estimates for focal herpetofaunal species in spring 2002. ..............63

3-21 Site occupancy estimates for focal herpetofaunal species in fall 2002 ..................63

4-1 Locations of 2003 sampling transects in Lake Tohopekaliga. ..............................80

4-2 Diagram of weekly trap placement at specified depths........................................81

4-3 Mean maximum distances traveled with variances and maximum distances,
based on results of mark-recapture sampling. .......................................................81

4-4 Number of trap sites sampled in each vegetation community per sample occasion. 82

4-5 Salamander capture success by vegetation community..............................83

4-6 Salamander capture success by water depth. ................................... ..................... 83

4-7 Frog capture success by vegetation community........................................... 84

4-8 Frog capture success by water depth ..................................................................... 84

4-9 Snake capture success by vegetation community. ................................................85

4-10 Snake capture success by water depth............................................. ...............85

4-11 Turtle capture success by vegetation community. ................................................86

4-12 Turtle capture success by water depth................................... ....................... 86

4-13 Fish capture success by vegetation community. ................... ................... .......... 87

4-14 Fish capture success by water depth. ............................................ ............... 87















Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science

AQUATIC VERTEBRATE USAGE OF LITTORAL HABITAT PRIOR TO EXTREME
HABITAT MODIFICATION IN LAKE TOHOPEKALIGA, FLORIDA
By

Ann Marie Muench

December 2004

Chair: Wiley Kitchens
Major Department: Wildlife Ecology and Conservation

Lake Tohopekaliga is a large, shallow lake in central Florida that is part of the

Kissimmee chain of lakes. Cultural eutrophication and lake stabilization over the past

several decades have facilitated the formation of a densely vegetated, often monotypic,

littoral zone. Lake managers conducted an enhancement project in 2004 to improve

largemouth bass (Micropterus salmoides) habitat. This project included an extreme water

level drawdown and concurrent mechanical removal of 7.3 million cubic meters of

organic sediment and vegetation from the shoreline. Following the drawdown, herbicidal

treatments will keep the lake vegetation in an early state of succession in order to prolong

the effects of the enhancement. Little is known about potential impacts of these

procedures on wildlife, including vegetation, avian, herpetofaunal, and even fish

communities. This study examines the status of the reptile, amphibian and fish

communities in the two years prior to the lake enhancement to provide baseline data for

future assessments.









Funnel traps were used for all sampling, allowing a suite of vertebrate species to be

examined. In 2002, sampling was conducted in the Pontederia cordata pickerelweedd)

zone of the lake. Cluster analysis, indicator species analysis, and nonmetric

multidimensional scaling ordinations were used to examine temporal changes in the

species composition of the assemblages. Environmental variables such as lake stage and

average air temperatures played large roles in structuring the aquatic vertebrate

communities through species richness and assemblage composition. Fish assemblages

were most correlated with air temperature, while herpetofaunal assemblages mainly

showed association with lake stage. Site occupancy estimates showed that many of the

herpetofaunal species are present throughout the pickerelweed habitat in both the spring

and fall seasons, while fish showed more fluctuation in seasonal presence.

Spatial sampling took place in 2003. Sampling was conducted across vegetation

communities and water depths. Both variables captured varying trends in the presence of

the focal species, which included fully aquatic salamanders (Siren spp., Amphiuma

means), water snakes (mainly Nerodia spp.), small kinostemid turtles (Kinosternon

baurii, S.i ntu,,hei ,1n odoratus), large aquatic frogs (Rana spp.), juvenile centrarchids

(especially Micropterus salmoides) and exotic catfish (Hoplosternum littorale).

Attempted population density estimates for the more abundant herpetofaunal species

ended in failure. Inappropriate trapping grid size and spacing for several species at one

time led to poor capture probabilities and large variances in population size estimates.














CHAPTER 1
MAIN INTRODUCTION

Lake Ecosystem

The productive littoral environment in a lake system is dynamic, since the aquatic

habitat has strong terrestrial influences and the terrestrial habitat has strong aquatic

influences. Biological diversity is high in the ecotone due to biotic and abiotic properties

that distinguish it from adjacent ecosystems, such as vegetation species, soil properties,

and water chemistry (Lachavanne 1997). Water level fluctuations are the main

determinants of the width of the littoral zone. In lentic systems with gently sloping

shorelines a wide band of macrophytes provides patches of heterogeneous habitat for a

diverse assemblage of faunal species. Animal species that have specific requirements for

different life stages depend on the proximity of supralittoral (never flooded), eulittoral

(occasionally flooded), and infralittoral (always flooded) habitats. Since the ratio of

water surface to volume is much higher in the littoral zone than in the deeper pelagic

region, environmental variables such as light, air temperature, wind and water flow

(waves) have much more critical roles in shaping the gradient (Pieczynska 1990,

Pieczynska and Zalewski 1997).

Excess inputs of nutrients beyond that naturally found in a particular lake system

leads to eutrophication. This condition encourages surplus sedimentation and vegetation

growth, changing the landscape of the original littoral zone and subsequently altering the

biological communities within that habitat and the lake as a whole. Eventual extinction

of the lake may result from decades of nutrient pollution due to sewage discharge and









drainage from agricultural and urban lands. Internal recycling of nutrients within a water

body keeps it from recovering even when the inputs are reduced (Cooke et al. 1993). In

order to preserve the lake for the longest time possible, rehabilitation efforts are often

made to counter the results of eutrophication, for example drawdowns, dredging and

mechanical vegetation removal (Hasler 1947, Cooke et al. 1993).

Study Area

Lake Tohopekaliga (7,612 ha, 18,810 acres) is located in Osceola County, Florida,

within the Upper Kissimmee Basin. This physiographic area is known as the Osceola

Plain, which is flat and has very few distinguishing topographical characteristics. The

elevation in the plain ranges from 18-30 m (60-95 ft) National Vertical Geodetic Datum

(NGVD), but rarely reaches maximum heights (Harper 1921). Originating from

prehistoric ocean bottom, the sediment mainly consists of coastal sands. Numerous

shallow lakes in the region, including Lake Tohopekaliga, were formed by dissolution of

the carbonate-containing substrates (limestone) in depressed areas (Schiffer 1998).

Freshwater wetlands in this area include cypress sloughs, wet prairies, river swamps,

floodplains, mixed forested wetlands, and marshes. Pine flatwoods dominate the upland

community (HDR Engineering, Inc. 1989).

The Lake Tohopekaliga Subbasin (211.6 square km, 131.2 square miles), within the

Upper Kissimmee Basin, receives water input from the Shingle Creek (184.2 square km,

118.0 square miles) and East Lake Tohopekaliga (81.7 square km, 48.4 square miles)

Subbasins. Precipitation, overland flows, and to some extent groundwater from the

underlying Surficial Aquifer also provides the lake with important water sources. While

evapotranspiration is a strong factor in withdrawal of water from the lake, outflow from

Lake Tohopekaliga occurs at its southernmost point, where the South Port Canal connects









it to Cypress Lake. Water from the Upper Kissimmee Basin flows southward through the

Kissimmee Chain of Lakes, through the channelized Kissimmee River, to Lake

Okeechobee, east and west coast estuaries and South Florida (HDR Engineering, Inc.

1989, Schiffer 1998).

Human disturbance of this hydrologic system began in the mid-nineteenth century,

with local efforts to drain wetlands. In 1882 Hamilton Disston began channelizing the

watersheds in the upper basin by constructing inter-lake canals. The major results of this

endeavor were lowered lake levels, drying of lake edges and inter-lake slough wetlands,

as well as rapid transit of nutrient-laden surface waters downstream. Wetlands stretching

between Lake Tohopekaliga and East Lake Tohopekaliga were strongly impacted. In the

1920's, Shingle Creek (a major source of water for Lake Tohopekaliga) was channelized,

bypassing water around the swamps and marshes within that subbasin. Many landowners

also dug ditches and canals to drain wetlands and improve their pastureland. In 1947, the

Central and Southern Florida Flood Control Project was implemented by the U.S. Army

Corps of Engineers in response to major flooding in the Kissimmee Basin. As a result of

this plan, the Kissimmee River was channelized, Disston's inter-lake canals were

improved, and water control structures were built throughout the area. The goal of these

actions was to use the chain of lakes for water management, to provide room for water

during the wet season and to store water during the dry season. This entailed stabilization

of water levels, which historically fluctuated up to 3 m (10 ft), to a 0.6-1.2 m (2-4 ft)

range. This reduction in fluctuation subsequently allowed landowners and private

citizens to build on historic lake bottom and drained wetlands within the floodplain,









further strengthening the need for tight flood control (Blake 1980, HDR Engineering, Inc.

1989).

Lake Tohopekaliga and surrounding lakes have suffered many water quality

problems due to the intense hydrologic modifications. The constructed canals, especially

in the Shingle Creek area, allowed direct conveyance of stormwater runoff and sewage

into the lakes without the benefit of filtration through wetlands. Urban and agricultural

land use continued to expand, contributing more and more overland pollution. The

agricultural land in the area is mainly utilized as pastureland, and historically dairy farms

provided significant inputs of nutrients. Although eutrophication within the lakes was

rapidly increasing, water level stabilization prevented natural fluctuations from mitigating

the problem. Since the lake levels were unnaturally restricted from periodic flooding and

drying events, thick stands of vegetation began invading the littoral zone, which in turn

led to organic sediment buildup and decreased water quality. As of 1988, no wastewater

discharges have been permitted to the lakes. However, non-point source urban and

agricultural runoff and septic tank leakage remain major contributors to eutrophication in

the Kissimmee chain of lakes (HDR Engineering, Inc. 1989).

As mentioned, eutrophication has caused major vegetation changes to the littoral

zone of Lake Tohopekaliga. Dense monotypic expanses of aquatic vegetation began to

dominate the gradually sloping shoreline, formerly characterized by sandy substrate and

sparse vegetation. Nuisance species such as Pontederia cordata pickerelweedd) and

Typha domingensis (cattail) formed wide bands of habitat around the lake. Pontederia

cordata and associated species created floating mats on the lakeward edges of the littoral

zone that rose and fell with the water level. Exotic species such as Hydrilla verticillata









(hydrilla), Eichornia crassipes (water hyacinth), Alternanthera philoxeroides (alligator

weed), and Panicum repens (torpedo grass) also benefited from increased nutrients and

high boat traffic between waterways and became a major focus of lake managers through

the years. Turnover in the vegetation community produced an organic muck substrate

within the littoral zone. Documented faunal responses to this changing habitat have

included declines in fisheries, especially sport and forage fish species, and invertebrates

(HDR Engineering, Inc. 1989).

In 1968, the first fish population surveys in Lake Tohopekaliga were conducted by

the State of Florida Fish and Wildlife Conservation Commission (FFWCC, formerly

Florida Game and Fresh Water Fish Commission, FGFWFC) with rotenone,

electroshocking, and trammel nets (Wegener 1969). Management recommendations

included drawdowns every 5-7 years in order to oxidize the increasingly organic substrate

and provide benefits to the growing fish population upon reflooding of littoral habitats.

In 1971, Lake Tohopekaliga underwent its first extreme drawdown. The water stage was

dropped about 2.1 m (7 ft) from high pool stage (16.8 m, 55.0 ft NGVD). Drought

conditions kept water levels below 15.8 m (52.0 ft) NGVD (low pool stage) until one

year after the initiation of the drawdown. The dewatering of the littoral zone attained the

management goals, with increased acreage of desirable plant species, greater production

of fish and fish-food organisms (invertebrates) per acre, and increased sportfishing

success (Wegener and Williams 1974).

Improvements made to Lake Tohopekaliga's littoral zone were short lived, due to

continued input of nutrients (e.g., about 53 million liters (14 million gallons) of sewage

waste per day discharged into the lake (Wegener and Williams 1974)) and water level









stabilization. An offshore berm was forming at the low-pool water line that was thought

to be acting as a barrier to fish and invertebrates at low water levels (Moyer et al. 1987).

A second drawdown was conducted in 1979 and also had beneficial results, although by

1986 the habitat had degraded once more to sub-optimal fishery habitat. The organic

berm in the lakeward portion of the littoral zone was increasingly becoming a

management issue not able to be addressed by drawdowns alone (Moyer et al. 1987).

When the third drawdown was performed in 1987 a pilot muck removal project was

included. Along 19 km (12 miles) of shoreline, 164,830 cubic meters (225,000 cubic

yards) of muck were mechanically removed from the organic berm. This was considered

"an unprecedented large-scale restoration project to improve littoral habitat" (Moyer et al.

1993, Appendix 4, page 2). All research pointed to highly positive results from the

drawdown and muck-scraping procedures. Fishery surveys have continued since 1968,

and now include roving creel surveys, blocknet/rotenone sampling, electrofishing,

experimental gill nets, and shallow water sampling with Wegener rings. Other wildlife

monitoring included snail kite (Rostrhamus sociabilis) individual and nest counts, limited

aquatic plant sampling, and littoral zone invertebrate community monitoring (Moyer et

al. 1993).

Another lake enhancement project had been planned for early 2002 in Lake

Tohopekaliga, however logistical constraints caused postponement until early 2004

(Florida Fish and Wildlife Conservation Commission 2003). The project originally

included an extreme drawdown from 16.8 m (55.0 ft) NGVD to 14.9 m (49 ft) NGVD

beginning in November 2003, as well as the mechanical removal of about 5.4 million

cubic meters (6.8 million cubic yards) of muck and vegetation from the majority of the









lake's shoreline (Florida Fish and Wildlife Conservation Commission 2003). The

subsequent estimate of the actual volume scraped was 7.3 million cubic meters (8-million

cubic yards), with 1,351 ha (3,339 acres) of shoreline habitat removed. The entire width

of the littoral zone was targeted for removal, not just the organic berm. The Pontederia

cordata-dominated habitat underwent widespread elimination throughout the lake.

Twenty-nine in-lake disposal islands were created from much of the scraped material

(Florida Fish and Wildlife Conservation Commission 2004). Once the water levels

recover, heavy applications of herbicides will be used to keep the habitat in an early state

of succession, allowing lake managers to selectively allow regrowth of desirable

vegetation. Currently, the target conditions post-enhancement are undefined.

The main objective of the Lake Tohopekaliga habitat enhancement project is the

removal large expanses of undesirable macrophyte stands, particularly Pontederia

cordata and Typha domingensis, as well as the organic substrate (muck) associated with

this dense vegetation (Florida Fish and Wildlife Conservation Commission 2003).

Reptiles, amphibians and many juvenile fish species are known to occupy structurally

complex lentic habitats and utilize the muck and thick vegetation for foraging, cover, and

also reproduction (e.g., amphibians). Lake enhancement techniques (both mechanical

vegetation and muck removal and subsequent herbicide applications) modify these

resources, changing the habitat suitability for aquatic vertebrates. High mortality during

the scraping process and migration during the drawdown will likely also alter the

community structure and dynamics. The effort to sustain high species diversity in the

lake ecosystem may be important to the stability of the system, and by examining the









consequences of these restoration techniques managers can better evaluate their worth to

wildlife and fishery species.

Research Objectives

While some positive responses have been documented for the fishery of Lake

Tohopekaliga following past enhancement projects, many wildlife guilds have been

neglected. There is limited quantitative knowledge of vegetation responses to mechanical

removal and large-scale herbicidal treatment. It is also uncertain how wetland birds are

affected. There are still unanswered questions regarding aquatic vertebrates that utilize

the thick vegetation and organic sediment, including reptiles, amphibians and fish.

Herpetofaunal responses to enhancement activities have not been studied in the past, even

though they are pervasive in the habitat. Although fishery science claims that the

eutrophic littoral habitat is unsuitable for centrarchids (i.e. sport fish), conventional

sampling methods may be incapable of detecting them in highly vegetated areas (Parker

1970, Allen et al. 2003).

The current study is part of a larger project evaluating the wildlife response to

habitat enhancement in Lake Tohopekaliga. Also included in this project are vegetation

(see Welch 2004) and avian monitoring studies. The research presented in this thesis

examines the aquatic vertebrate community in the littoral zone of Lake Tohopekaliga

prior to the 2003 drawdown and mechanical vegetation and muck scraping activities.

The littoral zone is defined here as the area occupied by emergent vegetation. However,

there is particular emphasis given to the pickerelweed zone due to its extensive removal

during the lake enhancement. The large-scale wildlife habitat investigation will continue

for at least three years after enhancement activities to examine responses to the

modifications by the different guilds.









With a large-scale habitat modification, quantifying the effect on a whole suite of

species provides maximum information. While most of the species have common

biological or ecological traits, they also constitute a variety of habitat requirements based

on food sources, reproduction methods, and movement patterns. For this reason

community metrics within the land-water ecotone are of main concern, as represented by

species richness and community composition. Species-specific site occupancy and

capture frequencies also facilitate understanding of habitat utilization by focal vertebrate

species. The main objectives of this research are to

1. Characterize the vertebrate faunal makeup of Lake Tohopekaliga's littoral zone
prior to the 2004 lake enhancement project,

2. Estimate parameters such as density and activity/home range for focal species in
the P. cordata habitat,

3. Estimate site occupancy rates for focal species within the littoral zone, as an
estimate of the proportion of the area that the species inhabits,

4. Document how temporally changing variables including lake stage, water level
fluctuation, air temperature, and rainfall shape the aquatic vertebrate community in
the P. cordata zone, and

5. Investigate the influence of spatial variables such as water depth and vegetation
community on the herpetofauna and fish within the landscape.














CHAPTER 2
DESCRIPTIONS OF FOCAL SPECIES

Aquatic Vertebrate Habitat

Wetland communities of reptiles and amphibians show much diversity in

ecological function. Often being the largest and most abundant vertebrates in this habitat,

they have important places in the food webs of lakes (Iverson 1982). Some species

provide terrestrial links while others are fully aquatic and never leave the littoral zone

(Joly and Morand 1997). Fish species also rely on both the littoral, pelagic, and to some

extent flooded terrestrial (nursery) habitats as they undergo shifts with life stage (Werner

2002). There has been a worldwide decline in biodiversity, particularly seen in

amphibian species. A variety of human disturbances have been identified, including

climate change, habitat loss and fragmentation, introduced species, pollution, acid rain,

and disease (Reaser 2000). Florida in particular has been severely impacted by

destruction of wetlands, channelization of streams, manipulated hydrologic cycles, and

rapid human growth (HDR Engineering, Inc. 1989, Pough et al. 2001). Alteration of

freshwater habitats has been a problem for many aquatic species. Animals that are long-

lived or have delayed sexual maturity, low reproductive rates, or poor dispersal or

colonization abilities are particularly vulnerable to habitat destruction (Klemens 2000).

Purposeful habitat modification should preserve conditions necessary for aquatic animals

to complete their life cycles, including appropriate nesting/spawning, foraging, and cover

habitats.









Species of aquatic vertebrates that are most at risk due to their habitat

requirements are emphasized here. Most herpetological research has been conducted on

breeding populations of amphibians, large charismatic reptiles, or single species and

guilds (but see Bancroft et al. 1983). Fishery science remains focused mainly on

sportfish at the individual or population level (Miranda and Dibble 2002). Resident

littoral zone species make up the assemblage of interest for this study and represent

several different orders of animals with a variety of life history traits. Fish guilds, such as

juvenile centrarchids (especially Lepomis spp., Micropterus salmoides) and exotic catfish

(Hoplosternum littorale), are focused upon. Documentation of the presence of these

species in heavily vegetated littoral habitats in Florida is very poor, probably due to

inadequate sampling techniques. Reptile and amphibian species of interest include fully

aquatic salamanders (Siren spp., Amphiuma means), water snakes (mainly Nerodia spp.),

small kinosternid turtles (Kinosternon baurii, Si.e inlheil %// odoratus), and large aquatic

frogs (Rana spp.). Minimal research has been conducted on the effects of lake

management techniques on these herpetofaunal species. Most of these species and guilds

have a common reliance upon vegetated wetlands for at least some part of their life

cycles. They also are often preyed upon by the same species, including alligators,

wading and predatory birds, large predatory fish and aquatic snakes, and together

represent many segments of the food web in the lake ecosystem.

Fish Species

Centrarchids (sunfish)

Most species in the family Centrarchidae in Lake Tohopekaliga are sportfish.

Foremost among them in Florida lakes is the largemouth bass (Micropterus salmoides).

This species is the primary target for benefit by the Lake Tohopekaliga enhancement.









Bluegills (Lepomis macrochirus), redear sunfish (Lepomis microlophus), black crappie

(Pomoxis nigromaculatus), warmouths (Lepomis gulosus), spotted sunfish (Lepomis

punctatus), and dollar sunfish (Lepomis marginatus) are also considered sportfish in

Florida. Enneacanthus glorious, (bluespotted sunfish), has a maximum total length of

80 mm, and is therefore only considered a forage fish species (Hoyer and Canfield 1994).

Most of these species depend upon the vegetated littoral zone during juvenile stages and

for spawning. Vegetated habitats provide juvenile sunfish with protection from larger

predators and abundant food supplies (Werner and Hall 1988, Chapman et al. 1996,

Miranda et al. 2000). The phenomenon of ontogenetic habitat shifts is particularly well

studied in bluegills. This species moves between the littoral to the pelagic zone

throughout its life cycle. The littoral zone provides nesting habitat, as well as a preferred

environment for juvenile bluegills from approximately 12-83 mm standard length due to

size-specific predation risks (Werner and Hall 1988).

It is claimed that these species have no access to Lake Tohopekaliga's littoral

zone due to physical and chemical barriers. The floating mats ofPontederia cordata,

resulting from the eutrophic status of the lake, are thought to form a physical barrier for

centrarchids, limiting adult access to shallow water spawning sites. Even if the fish could

penetrate this barrier, physicochemical characteristics of the dense vegetation would not

permit survival (Moyer et al. 1995, Allen and Tugend 2002, and Allen et al. 2003).

Traditional methods of fish sampling in high-macrophyte littoral habitats in Lakes

Tohopekaliga and Kissimmee, Florida, have yielded few or no centrarchid species

(Moyer et al. 1993, Allen and Tugend 2002). However, since common fish sampling

methods, including electrofishing and rotenone/blocknet, do not perform well in heavily









vegetated habitats (Parker 1970, Moyer et al. 1995, Allen and Tugend 2002), many

suppositions upon which lake enhancement projects depend are theoretical.

Exotic catfish

Hoplosternum littorale is an exotic species in Florida, originating in South

America. This armored catfish was first found within the United States in South Florida

in 1995, and was presumably released through the aquarium trade or aquaculture.

Various life history and behavioral traits, including aerial respiration, large body size,

high environmental tolerances, and nest-guarding behaviors, are responsible for rapid

expansion of its range in Florida (Nico et al. 1996). This species is currently nesting in

and pervasive throughout the littoral zone in Lake Tohopekaliga (personal observation).

Ptei gqpliI hIiy\ spp. (suckermouth or sailfin catfish) has also been captured in Lake

Tohopekaliga, although only on a few occasions. This species was probably released into

Florida through the aquarium trade (Page 1994). These two species may pose significant

ecological threats to native food webs and aquatic plant communities. While the Florida

Fish and Wildlife Conservation Commission conducts yearly monitoring by

electrofishing, this study is the first known report of these species this far north in the

Kissimmee chain of lakes.

Herpetofaunal Species

Amphibians

Rana grylio (pig frog) is a highly aquatic species, rarely being seen on shore.

They are usually associated with dense marsh vegetation. While leopard frogs, including

Rana sphenocephala (Florida leopard frog), prefer habitats with standing water, larger

individuals can inhabit somewhat dryer on-shore habitats and use larger home ranges,

relying on plant shade, dew and soil moisture for survival (Dole 1965). Adult leopard









frogs and their tadpoles are also noticeably absent from sandy, unvegetated shorelines

(Dole 1965, Alford and Crump 1982, Bancroft et al. 1983). These two large frog species

have differences in length of larval development, with pig frogs taking more than a year

to metamorphose and leopard frogs taking only two to three months (Bancroft et al.

1983). This, along with year-round breeding in Florida, results in a variety of size classes

of tadpoles throughout the year.

Siren lacertina (Greater siren) and Amphiuma means (Two-toed amphiuma) are

two of the largest species of salamanders in the world (Petranka 1998). Amphiumas

depend on lungs for aerial respiration, while sirens have external gills as well. Although

they may have lengths greater than 76 cm (Conant and Collins 1998), diminutive limbs in

both species are thought to limit overland dispersal. These salamanders burrow into

organic sediment when their habitats become dry and may remain alive for up to three to

five years in underground burrows without food until water comes back to the habitat

(Martof 1969, Etheridge 1990). Bancroft et al. (1983) found that the density of

amphiumas and greater sirens in Lake Conway, Florida, increased with sediment depth.

They also reported that neither species inhabited sandy, unvegetated shorelines. Sirens

have compressed tails that may help to propel them in vegetated open water as well as

emergent vegetation habitats. Amphiumas on the other hand have long cylindrical tails

and are thought to be limited to shallow water (Bancroft et al. 1983). Sirens feed mainly

on mollusks, insects, crayfish and filamentous algae, as well as some other vegetation.

Amphiumas eat fish, crayfish, salamanders, frogs and a wide variety of other species

(Petranka 1998).









Reptiles

The striped mud turtle (Kinosternon baurii) and common musk turtle

(.Siie nithe/i % odoratus) are both small species (maximum carapace lengths of 12.2 cm

and 13.7 cm respectively) that prefer shallow water wetlands (Conant and Collins 1998).

They are both omnivorous, feeding upon animals and some plants opportunistically.

However, mud turtles are attracted to fast-moving prey while musk turtles search out

more sedentary organisms as they crawl along the substrate in search of prey (Mahmoud

1968). Striped mud turtles usually occur in water greater than 60 cm deep, with lower

water levels or rainfall triggering terrestrial activity (Wygoda 1979, Ernst et al. 1994).

On the other hand, common musk turtles are highly aquatic, not leaving water unless

nesting. This species seems to prefer water depths less than 60 cm, but have been seen in

up to 9 m of water (Ernst et al. 1994). Bancroft et al. (1983) found about 20% of all

captured common musk turtles in the littoral zone, and the rest (usually larger

individuals) in open water habitat. According to Mahmoud (1969), S. odoratus is found

in lakes as well as riverine habitats with gravel or sandy substrates.

Nerodiafasciatapictiventris (Florida water snake) is most often encountered in

the shallowest regions of inhabited wetlands (Ernst and Ernst 2003). They are observed

often in disturbed and white sand littoral habitats (Bancroft et al. 1983). This species eats

mainly fish until they reach a total length of 50 cm, at which point they switch to preying

upon frogs (Mushinsky et al. 1982). Nerodiafloridana (Florida green water snake) is the

largest North American water snake, with total lengths approaching two meters (Conant

and Collins 1998). They are inhabitants of quiet water wetlands and sometimes venture

out into open water (Ernst and Ernst 2003). Bancroft et al. (1983) found them to be

pervasive throughout the littoral zone, and while the dense vegetation seems to be






16


preferred, the species of vegetation may not be very important. Some individuals were

captured up to 40 m from the edge of the littoral zone in open water, while several

terrestrial sightings occurred during winter months. Sediment depths of 11-20 cm

yielded the most individuals, and sandy beach habitats were avoided by Florida green

water snakes (Bancroft et al. 1983). They feed mostly upon fish, but also on frogs,

salamanders, tadpoles, small turtles and invertebrates (Mushinsky and Hebrard 1977,

Ernst and Ernst 2003)














CHAPTER 3
ASSEMBLAGE WITHIN THE Pontederia cordata COMMUNITY

Introduction

The objective of this section is to investigate the temporal variation of community

composition and dynamics. The four main research questions are 1) is there temporal

variation in the aquatic vertebrate assemblage, 2) does community composition change

over time, 3) what environmental factors seem to be influencing the temporal variation in

the assemblage and individual focal species, and 4) how are the focal species dispersed

through the habitat. Key environmental variables that change over the course of a year

include lake stage, water level fluctuation, air temperature, and rainfall. Each of these

will be examined for their influence on the vertebrate assemblage. The thick P. cordata

pickerelweedd) habitat was the prime target for mechanical removal during the lake

enhancement process and therefore was the focus of sampling effort. This protocol was

also used to select focal species (which species were present in the habitat and most

detectable with the traps) and evaluate trap-sampling methods. All of this information

will facilitate monitoring in the future, regarding how, when and where to sample in order

to capture the community dynamics and variances associated with the lake's ever-

changing environment.

Field Methods

Trap Descriptions

As previously mentioned, fishery surveys conducted in the Kissimmee Chain of

Lakes include roving creel surveys, blocknet/rotenone sampling, electrofishing,









experimental gill nets, and Wegener rings (Moyer et al. 1993). These methods collect

information on a variety of fish species, but sport fish are the typical target of research.

Traditional herpetofaunal sampling techniques include visual surveys and hand or dip-net

collecting (Bury and Corn 1991), pitfall and funnel traps in combination with drift fences

(Corn 1994), and use of seines or dredges for removing floating vegetation along with the

animals inhabiting it (Bancroft et al. 1983). None of these methods are appropriate for

the extremely thick, rooted vegetation in the littoral zone. Turtle traps exist, such as hoop

nets and floating traps for basking turtles (Lagler 1943); however large turtles are not

central to this research since they are not restricted to the littoral zone. PVC pipes have

also been used as passive traps for treefrogs (Moulton et al. 1996), and audio surveys are

often used for breeding ranid frogs (Zimmerman 1994). However, a single, all-

encompassing technique was desired for this community study, and the answer came

from funnel traps. Recently, several researchers have noticed the benefits of capturing

aquatic organisms in thick vegetation with crayfish and minnow style funnel traps (Darby

et al. 2001, Sorensen 2003, Johnson and Barichivich 2004). Without the use of either

bulky drift fences or bait, these traps have been successful in capturing a wide variety of

reptiles, amphibians, fish and some invertebrates. Funnel traps were used for all

sampling during this study.

The minnow and crayfish traps were all constructed of 1.3 cm (0.5 in) mesh, dark

green vinyl-coated hardware cloth (Figure 3-1). The crayfish traps, similar to those

described by Darby et al. (2001), were positioned on the substrate, or as near to the

substrate as the vegetation would allow. They were approximately 80 cm (30 in) tall

including a "chimney" extending from the body of the trap, allowing the top to be above









the water surface. At the base were three entry funnels leading into the trap, with each

opening about 6 cm (2.5 in) in diameter, but the exact size varied slightly due to

handmade construction. The modified minnow traps were 60 cm (24 in) long rectangular

traps, which were approximately 25 cm (10 in) deep and 18 cm (7 in) high. At each end

there was one entry funnel, with an opening approximately 9 cm (3.5 in) wide and 6 cm

(2.5 in) tall. Floats made of Styrofoam pool toys ("Wacky Noodles") were attached to

the minnow traps to allow them to float halfway out of the water, with the funnels about

even with the water surface, based on the design by Casazza et al. (2000). The funnels

permitted animal access into the trap, but discouraged escape by making the exits harder

to find than the entrances. By allowing the traps to remain partially above the water, the

animals had access to air and mortality was reduced. Both nocturnal and diurnal species

were accessible to capture since traps could be deployed without time constraint. The

traps were not baited, however once an animal was captured in the trap other animals

may have been attracted to it.

The dimensions of the traps restricted the assemblage of animal species captured.

The traps did not confine young individuals or small species of fish, frogs, snakes and

salamanders due to the 1.3 cm (0.5 in) mesh size. Also, individuals larger than the funnel

diameter were excluded. To compare the difference in species captured with 1.3 cm (0.5

in) versus 0.6 cm (0.25 in) mesh, 18 commercially-manufactured minnow traps, similar

to the "eelpots" used in Casazza et al. (2000), were deployed at randomly assigned trap

points from 11/5/2003 to 1/8/2004. These traps are cylindrical, about 60 cm (24 in) long

and 23 cm (9 in) in diameter, with the funnel openings about 5 cm (2 in) in diameter.

They were also fitted with floatation. The hardware cloth was bare metal, not vinyl-









coated. Comparisons of species and number of captures were made between the 0.6 cm

mesh minnow traps and the 1.3 cm mesh modified minnow traps from the same trap

points during this sampling period. We expected to capture more species with the

smaller mesh size since small species and younger individuals could escape from the

larger mesh, but be retained by the 0.6 cm holes.

Whole-Lake Sampling

To gain information regarding temporal habitat utilization by the aquatic vertebrate

assemblage, sampling was conducted around the periphery of the whole lake to maximize

the inference of the results to the system. For the whole-lake sampling, 18 sites were

randomly selected from the less developed, southern two-thirds of the lake (Figure 3-2).

At each site, a transect was established with three trap locations placed perpendicular to

shore and spaced approximately 10 m (33 ft) apart, except in disturbed stretches of

habitat with barriers within this distance (e.g., commercial airboat trails). One crayfish

and one minnow trap were placed at each trap location, attached to a PVC pole for extra

stability. The result of this trapping arrangement was uniform sampling effort at each

transect. The trap locations were placed in the most lakeward portion of dense P. cordata

when possible (mainly in the 0.6-0.9 m (2-3 ft) depth zone at 13.8 m (55 ft) NGVD). The

transect sites varied in proximity to the ecotone between the open water habitat and the

vegetation. Most transects had thick stands of Typha or more diverse floating mats

between the relatively monotypic sections of pickerelweed and open water. The band of

emergent macrophytes at these locations was comparatively broad. On the other hand, at

some transects the traps were relatively close to this ecotone due to narrowness of the

pickerelweed zone at these locations, well established commercial airboat trails, or

herbicide applications near the transects providing large unvegetated areas. One transect









fell in an area where the substrate had previously been scraped, and the vegetation

consisted mainly ofHydrilla verticillata (hydrilla) and very few emergent macrophytes.

The whole-lake trap survey was conducted year-round, pending suitable water

levels (greater than approximately 16 m, 52.5 ft NGVD). Below this point, there was not

enough water for the trapped animals and rodents and birds were inadvertently captured.

Sampling throughout the year 2002 was as follows:

* January 24 Traps were deployed to randomly selected transects and sampling
began.

* May 2 Insufficient water levels in the pickerelweed zone caused traps to be
removed and sampling suspended.

* June 12 Redeployment of traps to select transects with sufficient water depth
resulted in decreased trapping effort until July 24.

* July 24 All traps were back in place in fixed sampling locations.

* December 3 Traps were removed from pickerelweed zone due to low water
associated with the attempted 2002 drawdown.

When active, the traps stayed in place day and night and were typically checked

once weekly. Despite efforts to keep samples spaced seven days apart, the time interval

was not always consistent due to logistical issues (e.g., airboat problems, rough weather).

At each sampling occasion, two or three observers traveled to each transect in an airboat

and checked the traps for their contents. All animals were brought back to the boat to be

worked up. Reptiles and amphibians were weighed individually with Pesola spring

scales and certain length measurements were taken, depending on the species. Fish were

identified and grouped according to species for each trap. All individuals of each species

per trap were weighed together in order to obtain the total biomass of the fish species

caught. After being worked up, the animals were released at the transects where they

were captured. The types of data collected with these methods include species detection-









nondetection, number of individuals captured on each sample occasion, biomass, and

reptile and amphibian length measurements.

This sampling protocol was intended to continue in the exact same locations post-

lake enhancement (2003) in order to compare community traits before and after the

modifications. However when the drawdown and muck removal was postponed for

another year, it was no longer beneficial to keep sampling since there was not a

before/after comparison to be made. Variables such as water temperature and dissolved

oxygen were not measured directly since this was not the initial focus of the study.

Alternative environmental variables were obtained using Internet resources. Lake stages

and rainfall were taken from the South Florida Water Management District's DBHYDRO

browser (http://glades.sfwmd.gov/pls/dbhydro_pro_plsql/). The lake stage was the mean

daily average taken from the headwater of Station S61 (the water control structure in the

south part of the Lake Tohopekaliga leading to Lake Cypress via the South Port Canal) in

feet NGVD. Lake stage was recorded for the day of each sample occasion. Water

fluctuation for one sample is the difference of the water level at that sample minus the

water level at the previous sample occasion. Rainfall was also recorded at Station S61

and precipitation totals for each sample were added up from the day of the previous

sample occasion until the day before the new sample occasion. Air temperature data was

gathered from the National Oceanic and Atmospheric Administration's National Climatic

Data Center's website (http://www.ncdc.noaa.gov/servlets/ULCD). The weather station

location was the Orlando International Airport (MCO) in Orlando, Florida. This is

located approximately six kilometers (10 miles) from the north shore of Lake

Tohopekaliga. Average temperatures were calculated for every sample occasion by









averaging the daily average temperatures from the day of the previous sample occasion

until the day before the new sample occasion. Minimum and maximum temperatures

were also recorded for each sample period.

Analysis Methods

Trap Comparisons

To determine the utility of traps with smaller mesh size for this study, the species

and number of captures for the 0.6 cm (0.25 in) mesh commercial minnow traps and the

1.3 cm (0.5 in) mesh modified minnow traps were compared. The 0.6 cm mesh traps

were randomly placed at only a portion of the whole-lake trap sites, along with a 1.3 cm

mesh crayfish and minnow trap. For this reason, data from both minnow traps were

compared for just the trap sites with both mesh types.

Species Richness

Sampling was carried out with a repeated measures protocol, potentially resulting

in lack of independence between samples. However, assuming random movement of

individuals and species through the habitat over space and time, sampling over time did

not result in repeated captures of the same individuals. This transient nature of the

species and utilization of non-parametric procedures for most analyses are believed

remove potential bias due to repeated samples. To determine the presence of temporal

variation in the aquatic vertebrate assemblage, total species richness was calculated for

each sample occasion in 2002. Fish and herpetofaunal species richness were also

individually estimated for each sample occasion. Program COMDYN4 was used with

detection-nondetection data to estimate richness (Hines et al. 1999), taking into account

species detection probabilities. It uses a model (Mh) that allows each species to have a

different detection probability (the probability of detecting at least one individual of the









species). Since most species detection probabilities are less than one, raw count data can

result in underestimations of richness. In a similar manner, the term "presence/absence

data" can also be misleading since lack of detection provides no evidence of a species'

absence from the trap site. For this reason I instead use the term "detection/nondetection

data" throughout this thesis. Equal sampling effort is necessary for each occasion.

Assumptions of this method are 1) population closure for species, 2) independence of

captures and 3) individual species capture probabilities stay constant during sampling

(Burnham and Overton 1979). However, this method is robust to deviations from these

assumptions. Even when the assumptions are violated the model-based richness

estimates are less biased than counts of species (Nichols et al. 1998).

Data from seventeen of the eighteen transects were used in the richness analysis.

The one transect that was located in the previously scraped habitat was removed from the

analysis in order to focus solely on variations within the P. cordata habitat. Since species

capture data were fairly sparse for each transect per sample occasion, the transects were

randomly assigned to six groups that represent sample replicates across space. They were

randomly grouped in order to remove effects of shoreline characteristics at different

transects. Richness was not estimated for sample occasions with reduced trap effort

(sample occasions 14-20). Linear regressions were performed using SPSS (SPSS Inc.

2001) in order to determine significant predictors of the vertebrate, fish and herpetofaunal

species richness. Richness estimates from each sample occasion were used in these

analyses. There were three outliers greater than two standard deviations from the mean,

which were removed for the herpetofaunal regression analyses. Average air temperature









(C), lake stage (m), rainfall (cm), and water level fluctuation (cm) were used as the

independent variables.

Assemblage Composition

Species richness estimates the number of species present, but indicates nothing

about community composition. To compare the presence of vertebrate species over time,

sample occasions were assigned to clusters using hierarchical cluster analysis. This was

run using PC-Ord software (McCune and Mefford 1999) with detection-nondetection

data of species for each sample occasion. Sorensen's distance measure with the flexible

beta (beta=-0.25) linkage method was used. Indicator species analysis (McCune and

Grace 2002) was applied to determine the most appropriate number of clusters and the

best species to represent those groups. Any groups comprised of a single sample

occasion were removed from the indicator species analysis. A Monte Carlo procedure

was run 1000 times with randomized data to calculate a p-value for each species, which

tested the null hypothesis that their indicator values were no larger than would be

expected by chance. The optimum number of groups was selected by the indicator

species analysis that yielded the most species with statistically significant indicator

values (McCune and Grace 2002).

Influence of Temporal Gradients on Assemblage

Multivariate ordination was used to establish what temporally changing

environmental factors were influencing the variation in assemblage composition.

Nonmetric multidimensional scaling (NMS) is an ordination technique that uses ranked

distances between sample units to reduce dimensions and allow description of the

community in relation to environmental gradients. The distances represent dissimilarity

between sample units in terms of species composition. This method was chosen because









it is particularly useful for non-normal data and many sampling events with no captures

(McCune and Grace 2002). Sample units, i.e. individual sample occasions, are plotted in

species space using an iterative search for the optimal placement for the sample units.

Optimal placement is determined by the maximum possible reduction in stress, which is a

measure of dissimilarity between the original data matrix and the reduced-dimension

final ordination. PC-ORD software was used for all NMS analyses (McCune and

Mefford 1999).

NMS was run for the entire vertebrate assemblage for all sample occasions using

detection-nondetection and count data separately. Fish and herpetofaunal assemblages

were then analyzed separately in the same fashion to determine if the environmental

gradients affected them differently. Outliers were identified using the outlier analysis

provided in PC-Ord, with the criteria being greater than two standard deviations from the

mean (McCune and Mefford 1999). All outliers were removed from the analyses.

General relativizations by row were conducted on the raw count data to equalize common

and uncommon species and lower the coefficient of variation (CV) of the row totals.

Relativizations were followed by square root transformations to balance the relative

importance of the species without altering their ranks.

Sorensen's distance measure was used to calculate dissimilarity matrices for the

ordinations. Starting configurations were created by random number seeds, which were

generated by the time of day. Fifty runs were conducted with the real data to find the

solution with the lowest stress. Fifty Monte Carlo randomized runs were performed to

select the appropriate number of dimensions that best represent the variation in the data.

Comparisons between the runs with real data and randomized data give a probability that









final stress in the ordination could be found by chance. After the first 50 runs, the number

of dimensions was determined and the final NMS was rerun using the random number

seed from the initial ordination. From this the final stress and instability (fluctuation of

stress per iteration) were evaluated.

Measured environmental variables, including lake stage, stage fluctuation, total

rainfall, and average, maximum and minimum average air temperatures over the sample

period, were included in the ordination graphs. The ordinations were plotted with

environmental variables as biplots, indicating the strength of correlations of variables

with the synthetic axes. Only the environmental variables with r2>0.2 (percent of

variance represented) are shown in the ordination plots. Sample units were color coded

by their membership to the groups defined by the cluster analysis, representing different

species composition.

Proportion of Habitat Utilized by Focal Species

Using the program PRESENCE (MacKenzie et al. 2002), detection/nondetection

data were analyzed to determine site occupancy rates for all species with enough captures

to get reasonable estimates. This method allows for numerous, representative, randomly

selected sites transectss) within the much larger area of interest (P. cordata zone) to be

sampled for the presence of species. The inference gained from sampling these sites can

then be applied to the pickerelweed zone Lake Tohopekaliga. The main function of this

method is to determine habitat usage for species with low detection probabilities (<1).

Detectability is an important factor when sampling secretive aquatic organisms in thick

vegetation. The program calculates (1) a "naive estimate," which is simply the

proportion of sites where the species was caught (considered biased low), (2) species-

specific detection probabilities based on the capture data, and (3) the "proportion of sites









occupied" (PSO) which is the naive estimate corrected for detection probability

(MacKenzie et al. 2002).

This method assumes closure of species to changes in occupancy status over the

course of sampling. However, if the species have large activity ranges and the

movements are assumed to be random, the closure assumption may be relaxed

(MacKenzie et al. 2002). The sample occasions were divided into two groups. The first

is from the start of sampling at the beginning of February until the traps were removed at

the beginning of May. The second group is from the beginning of August, when water

levels allowed full sampling effort, to the end of sampling in November. Between these

groups the lake stage became so low that there was no water in the P. cordata zone,

which surely caused a violation of the closure assumption for this method. This required

the split of sample occasions into groups that are assumed closed to species immigration

or emigration. The first (spring) group includes 13 sample occasions for the herpetofauna

and 11 sample occasions for the fish species, since fish were not recorded for the first

sample and the last sample in the group had water depths too shallow to capture fish. The

second (fall) group has 16 sample occasions for all species.

Parameters were estimated for each species for both groups using the single season

models in PRESENCE. The data were analyzed using models with both constant and

survey specific detection probabilities. Results from the model with the lowest Akaike's

Information Criterion (AIC) value were reported for each species. If the AIC values were

within two points of each other, the simpler, constant detection probability model was

selected.









Results

Trap Comparisons

All reptile, amphibian and fish species captured during the 2002 whole-lake sampling,

along with species codes used in the figures are listed in Table 3-1. Due to restrictive

funnel sizes, most fish species (especially centrarchids) were represented by juvenile life

stages, except small species such as mollies and killifish. On the other hand, adult

individuals characterized the majority of the reptile and amphibian species, since most

young individuals could escape through the mesh. Table 3-2 shows the species and

number of captures for the two types of minnow traps. Eleven vertebrate species were

captured with the 0.6 cm (0.25 in) mesh minnow traps, while 19 species were captured in

the 1.3 cm (0.5 in) mesh minnow traps at the same sample locations. Three species were

unique to the 0.6 cm mesh traps on these occasions: black swamp snake (Seminatrix

pygaea), flagfish (Jordanellafloridae), and mosquitofish (Gambusia spp). Only three

reptile or amphibian species were captured: black swamp snake, Florida leopard frog

(Rana sphenocephala), and pig frog (Rana grylio). More tadpoles were captured with the

0.6 cm mesh (n=31) than with the 1.3 cm mesh (n=2). Nine species were unique to the

1.3 cm mesh traps: striped mud turtle (Kinosternon baurii), striped crayfish snake

(Regina alleni), Florida water snake (Nerodiafasciatapictiventris), Florida green water

snake (Nerodiafloridana), siren (Siren spp), redfin pickerel (Esox americanus), armored

catfish (H. littorale), dollar sunfish (Lepomis marginatus), spotted sunfish (Lepomis

punctatus), largemouth bass (Micropterus salmoides), and redear sunfish (Lepomis

microlophus). Seven combined reptile and amphibian species were caught with these

traps. Species common to both traps were leopard frog, pig frog, bluegill (Lepomis

macrochirus), bluespotted sunfish (Enneacanthus gloriouss, golden topminnow










Table 3-1. All species captured in 2002, with species codes used in subsequent figures.
Fish Species Scientific Name Family Species Code
Armored catfish Hoplosternum littorale Callichthyidae HOPLI
Black crappie Pomoxis nigromaculatus Centrarchidae POMNI
Bluegill Lepomis macrochirus Centrarchidae LEPMAC
Bluespotted sunfish Enneacanthus glorious Centrarchidae ENNGL
Bowfin Amia calva Amiidae AMICA
Chain pickerel Esox niger Esocidae ESONI
Chubsucker Erimyzon spp. Catostomidae ERIMY
Dollar sunfish Lepomis marginatus Centrarchidae LEPMAR
Flagfish Jordanella floridae Cyprinodontidae JORFL
Gar Lepisosteus spp. Lepisosteidae LEPIS
Golden shiner Notemigonus crysoleucas Cyprinidae NOTCR
Golden topminnow Fundulus chrysotus Fundulidae FUNCH
Largemouth bass Micropterus salmoides Centrarchidae MICSA
Pterygoplichthys Pterygoplichthys spp. Loricariidae PTERY
Redear sunfish Lepomis microlophus Centrarchidae LEPMI
Redfin pickerel Esox americanus Esocidae ESOAM
Sailfin molly Poecilia latipinna Poeciliidae POELA
Seminole killifish Fundulus seminolis Fundulidae FUNSE
Spotted sunfish Lepomis punctatus Centrarchidae LEPPU
Warmouth Lepomis gulosus Centrarchidae LEPGU
Herpetofaunal Species Scientific Name Family Species Code
Amphiuma Amphiuma means Amphiumidae AMPME
Cottonmouth Agkistrodon piscivorous conanti Viperidae AGKPICO
Fl. banded water snake Nerodia fasciata pictiventris Colubridae NERFAPI
Fl. green water snake Nerodia floridana Colubridae NERFL
Fl. snapping turtle Chelydra serpentina osceola Chelydridae CHESEOS
Fl. softshell turtle Apalone ferox Trionychidae APAFE
Leopard frog Rana sphenocephala Ranidae RANSP
Mud snake Farancia abacura abacura Colubridae FARABAB
Peninsula cooter Pseudemys floridana peninsularis Emydidae PSEFLPE
Pig frog Rana grylio Ranidae RANGR
Siren Siren spp. Sirenidae SIREN
Stinkpot Sternotherus odoratus Kinosternidae STEOD
Striped crayfish snake Regina alleni Colubridae REGAL
Striped mud turtle Kinosternon baurii Kinosternidae KINBA
Tadpole-leopard frog Rana sphenocephala Ranidae TADRANSP
Tadpole-pig frog Rana grylio Ranidae TADRANGR










Table 3-2. Species capture frequencies for the 0.6 and 1.3 cm mesh minnow traps.
Vertebrate Species 0.6 cm mesh 1.3 cm mesh
Armored catfish 0 2
Black swamp snake 1 0
Bluegill 2 1
Bluespotted sunfish 18 31
Dollar sunfish 0 2
Flagfish 147 0
Florida green water snake 0 4
Florida water snake 0 1
Gambusia 75 0
Golden topminnow 17 1
Largemouth bass 0 1
Florida leopard frog 4 2
Pig frog 2 7
Redearsunfish 0 1
Redfin pickerel 0 1
Sailfin molly 56 5
Siren 0 1
Spotted sunfish 0 2
Striped crayfish snake 0 1
Striped mud turtle 0 1
Tadpoles 31 2
Warmouth 2 6


(Fundulus chrysotus), sailfin molly (Poecilia latipinna), and warmouth (Lepomis

gulosus).

Species Richness

Vertebrate species richness (Figure 3-3) was negatively correlated with average air

temperature (r2=0.330, df=l, p=0.001) and rainfall (r2=0.173, df=l, p=0.028). Average

air temperature was the only significant predictor of richness for the fish (Figure 3-4),

(r2=0.316, df=l, p=0.002), with higher species richness estimates occurring with lower

air temperatures. The estimated richness of the herpetofaunal assemblage (Figure 3-5)

was negatively correlated with lake stage (r2=0.327, df=l, p=0.003), rainfall (r2=0.166,

df=l, p=0.043), and water level fluctuation (r2=0.211, df=l, p=0.021).









Assemblage Composition

Six clusters were chosen to represent the 34 sample occasions. Thirteen indicator

species were determined with p<0.05 (Table 3-3). Indicator values are given for these

species, with 100 representing perfect indication of that group based on relative

abundances and frequency of occurrence. A zero indicates complete absence of a species

from a particular group. Since these cryptic species are quite mobile (with respect to the

traps) and dependent upon detection for quantification, indicator values are relatively low

compared to vegetation studies where virtually all species are detectable. Group 1 is

identified by several species, including Amia calva (bowfin), Fundulus chrysotus (golden

topminnow), Lepomis macrochirus (bluegill), Lepomis marginatus (dollar sunfish),

Nerodiafasciata pictiventris (Florida water snake), Rana sphenocephala (Florida leopard

frog), and R. sphenocephala and Rana grylio (pig frog) tadpoles. The second group

consisted only of sample occasion #13 (an extremely low water sample), and was

therefore omitted. Hoplosternum littorale (armored catfish) and Regina alleni (striped

crayfish snake) are the indicator species for Group 3. There were no significant indicator

species for Group 4, but Amphiuma means amphiumaa) and Lepisosteus spp. (gar) show

the highest indicator values with 28 and 22 respectively. Si. intihel i/ odoratus

(common musk turtle) was the sole indicator species for Group 5. The two pickerel

species, Esox americanus (redfin) and Esox niger (chain), were the only two species

indicating Group 6.

Some species were captured during every sample occasion, including

Enneacanthus glorious (bluespotted sunfish), Lepomis gulosus (warmouth), and Siren

spp. (siren). Hoplosternum littorale and Kinosternon baurii (striped mud turtle) were

found on almost every sample occasion. It is unclear why the armored catfish is an










indicator species for Group 3, when it has an indicator value of 22 for all groups but one.

Also, the bluegill had indicator values of 47 for both Group 1 and Group 6, although it

was assigned to Group 1 with p=0.028.

Table 3-3. Indicator species analysis results. Significant indicator species are
highlighted (p<0.05) and displayed with associated indicator values and the
clusters to which the species was assigned.
Species Code Cluster Max Indicator Value Mean Standard Deviation Probability
AGKPICO 1 25 15 7.4 0.211
AMICA 1 48.5 18 9.86 0.017
AMPME 1 27.9 24.8 4.25 0.234
APAFE 5 9.1 15.3 7.72 1
CHESEOS 4 12.5 15.3 7.49 0.696
ENNGL 1 20 20 0.63 1
ERIMY 6 27.9 23.1 7.1 0.201
ESOAM 6 40.9 22.6 7.6 0.022
ESONI 6 43.3 18.1 10.46 0.022
FARABAB 6 21.3 18.2 10.31 0.29
FUNCH 1 43.8 21.3 8.26 0.04
HOPLI 3 22.2 21.3 0.82 0.039
JORFL 1 34.1 16.8 9.59 0.077
KINBA 1 20.5 20.6 0.73 0.672
LEPGU 1 20 20 0.63 1
LEPIS 4 22.3 24.9 3.41 0.76
LEPMAC 1 47.1 21.9 8.23 0.028
LEPMAR 1 36.8 23.8 7.01 0.05
LEPMI 6 33.8 22.3 8.45 0.1
LEPPU 5 12 22.2 8.8 0.976
MICSA 6 28.8 23 7.01 0.178
NERFAPI 1 40.8 22.7 7.03 0.018
NERFL 3 21.6 21.3 0.83 0.295
NONE 1 31.7 19.9 9.88 0.107
NOTCR 6 28.6 16.5 8.92 0.059
POELA 1 28.8 24.6 4.2 0.115
POMNI 1 25 14.9 7.44 0.196
PSEFLPE 5 9.1 15.1 7.52 1
PTERY 5 18.7 16.9 9.76 0.179
RANGR 1 22.4 23.4 1.3 0.92
RANSP 1 100 17.4 9.61 0.001
REGAL 3 72.6 17.8 10.18 0.002
SIREN 1 20 20 0.63 1
STEOD 5 29.8 24.4 2.19 0.001
TADRANGR 1 34.6 24.4 5.45 0.009
TADRANSP 1 75 17 9.16 0.001









Influence of Temporal Gradients on Assemblage

Stable three-dimensional ordinations were produced with all NMS analyses except

the herpetofaunal analysis with detection/nondetection data. The final solutions were

based on the criteria of stress being reduced by at least 5% with each additional

dimension. The final stress values were lower with the real data than was found by the

Monte Carlo randomized runs (p<0.05), which indicates that there was real structure

found in the data. The final stress values for all ordinations (except for herpetofaunal

detection/nondetection) were between 12 and 18 (Table 3-4), which are common values

for ecological data and depict a fair portrayal of the data (McCune and Grace 2002).

Table 3-4. Stress and instability results from all NMS ordinations
Assemblage Data Type Final Stress Instability Iterations
Vertebrate Detection/nondetection 17.65 0.00254 500
Vertebrate Counts 13.27 0.00049 69
Fish Detection/nondetection 15.31 0.00095 29
Fish Counts 12.93 0.00016 49
Herpetofauna Counts 14.08 0.00045 38


For the NMS with vertebrate detection/nondetection data, Axes 1 and 3 best

explained 61% of the variance found in the assemblage composition. Environmental

variables with r2>0.2 were shown as biplots on the plots and include lake stage and air

temperature measures (Figure 3-6, Table 3-5). Axis 1 was correlated with lake stage

(r2=0.349). Axis 3 was most correlated with both lake stage and average air temperature,

(r2=0.354 and 0.259 respectively). Sixteen species were correlated with either Axis 1 or

3 with r2>0.2 (Figure 3-7, Table 3-6), 10 of them being indicator species.










Table 3-5. Percent of variance explained (r2) by environmental variables for each axis in
the vertebrate NMS with detection/nondetection data. Variables with r2 > 0.2
are highlighted.
Axis 1 2 3
Variable r2 r2 r2
Stage (m) 0.349 0.021 0.345
Fluc (cm) 0.039 0.009 0.183
Rain (cm) 0.037 0.006 0.029
Max (C) 0.038 0.362 0.127
Min (C) 0.002 0.255 0.229
Ave (C) 0.009 0.375 0.259

Table 3-6. Percent of variance explained (r2) for each axis by species in the vertebrate
NMS with detection/nondetection data. Indicator species with r2 > 0.2 are
highlighted in blue, while all other species with r2 > 0.2 are highlighted in
yellow.
Axis 1 2 3 Axis 1 2 3
Species r2 r2 r2 Species r2 r2
AGKPICO 0.073 0.002 0.094 LEPMI 0.035 0.56 0.018
AMICA 0.027 0.077 0.327 LEPPU 0.064 0.027 0
AMPME 0.002 0.171 0.01 MICSA 0.291 0.227 0.242
APAFE 0.035 0.004 0.001 NERFAPI 0.227 0.167 0.257
CHESEOS 0.021 0 0.063 NERFL 0.003 0 0
ENNGL 0.381 0.019 0.002 NONE 0.057 0.004 0.102
ERIMY 0.031 0.536 0.007 NOTCR 0.011 0.05 0.017
ESOAM 0.033 0.186 0.353 POELA 0.188 0.018 0.248
ESONI 0.126 0.166 0.089 POMNI 0.007 0.011 0.067
FARABAB 0.049 0 0.21 PSEFLPE 0.005 0.085 0.001
FUNCH 0.019 0.419 0.547 PTERY 0.002 0.081 0.038
HOPLI 0.376 0.003 0.155 RANGR 0.149 0.035 0
JORFL 0.087 0.047 0.15 RANSP 0.218 0.02 0.331
KINBA 0 0.07 0.016 REGAL 0.01 0.117 0.037
LEPGU 0.381 0.019 0.002 SIREN n/a n/a n/a
LEPIS 0.36 0.007 0.083 STEOD 0.021 0.006 0.018
LEPMAC 0.002 0.311 0.53 TADRANGR 0.016 0.127 0.311
LEPMAR 0.019 0.04 0.422 TADRANSP 0.07 0.039 0.287


With vertebrate assemblage count data, Axes 1 and 3 had the highest r2. Together

these axes represent 63% of the variance in the species composition. Axis 2 also had an

r2>0.2, and was best correlated with air temperature measures. As with the

detection/nondetection analysis, lake stage and air temperature measures showed the










highest correlation with these axes (Figure 3-8, Table 3-7). Air temperature is correlated

with Axis 1, with maximum air temperature having the highest r2 of 0.353. Axis 3 was

most highly correlated with lake stage (r2=0.458) and average air temperature (r2=0.375).

Seventeen species had an r2>0.2 for at least one of the axes, nine of them being indicator

species (Figure 3-9, Table 3-8).


Table 3-7.


Percent of variance explained (r2) by environmental variables for each axis in
e ht vertebrate NMS with count data Variab d


Axis 1 2 3
Variable r r
Stage (m) 0.146 0.010 0.458
Fluc (cm) 0.013 0.006 0.108
Rain (cm) 0.001 0.001 0.044
Max (C) 0.353 0.299 0.232
Min (C) 0.261 0.346 0.281
Ave (C) 0.298 0.379 0.375


--.--- ----- ..--- -. -.----


Table 3-8.


Percent of variance explained (r2) FOr each axis by species in the vertebrate
NMS with count data. Indicator species with r2 > 0.2 are highlighted in blue,
while all other species with / > 02 are highlighted in yellow


Axis 1 2 3 Axis 1 2 3
Species r2 r2 r2 Species r2 r2
AGKPICO 0.16 0.005 0.095 LEPMI 0.241 0.331 0.139
AMICA 0.059 0.089 0.192 LEPPU 0.06 0.02 0.018
AMPME 0.183 0.124 0.211 MICSA 0.13 0.184 0.048
APAFE 0.044 0.021 0.017 NERFAPI 0.166 0.001 0.373
CHESEOS 0 0.002 0.087 NERFL 0.068 0.024 0.004
ENNGL 0.809 0.342 0.108 NONE 0.031 0.011 0.164
ERIMY 0.217 0.024 0.092 NOTCR 0.124 0.136 0.022
ESOAM 0.014 0.455 0.09 POELA 0.133 0.023 0.341
ESONI 0.051 0.108 0.044 POMNI 0.022 0.005 0.087
FARABAB 0.041 0.028 0.018 PSEFLPE 0.065 0 0.038
FUNCH 0.002 0.291 0.575 PTERY 0.063 0.005 0.115
HOPLI 0.05 0.285 0.653 RANGR 0.125 0.157 0
JORFL 0.186 0.04 0.172 RANSP 0.092 0.038 0.345
KINBA 0.43 0.003 0.005 REGAL 0.039 0.039 0.005
LEPGU 0 0.104 0.013 SIREN 0.428 0.46 0.172
LEPIS 0.005 0.158 0.288 STEOD 0.268 0.182 0.288
LEPMAC 0.079 0.423 0.386 TADRANGR 0.147 0.159 0.008
LEPMAR 0.023 0.182 0.4 TADRANSP 0.137 0.065 0.24










The fish assemblage alone with detection/nondetection data resulted in the first two

axes representing 63% of the variance explained. Average temperature was most highly

correlated with Axis 1 (r2=0.291), and also with Axis 2 (r2=0.368), while lake stage was

correlated with Axis 2 (r2=0.275), (Figure 3-10, Table 3-9). For these two axes, there

were a total of 11 species with an r2>0.2, with six indicator species (Figure 3-11, Table 3-

10).

Table 3-9. Percent of variance explained (r2) by environmental variables for each axis in
the fish NMS with detection/nondetection data. Variables with r2 > 0.2 are
highlighted.
Axis 1 2 3
Variable r2 r
Stage (m) 0.042 0.275 0.105
Fluc (cm) 0.008 0.044 0.062
Rain (cm) 0.001 0.031 0.018
Max(C) 0.250 0.177 0.115
Min(C) 0.205 0.330 0.106
Ave (C) 0.291 0.368 0.083

Table 3-10. Percent of variance explained (r2) for each axis by species in the fish NMS
with detection/nondetection data. Indicator species with r2 > 0.2 are
highlighted in blue, while all other species with r2 > 0.2 are highlighted in
yellow.
Axis 1 2 3 Axis 1 2 3
Species r2 r2 r2 Species r2 r2
AMICA 0.152 0.296 0.12 LEPMAC 0.361 0.646 0
ENNGL n/a n/a n/a LEPMAR 0.406 0.229 0.168
ERIMY 0.149 0.089 0.223 LEPMI 0.118 0.205 0.436
ESOAM 0.287 0.223 0.083 LEPPU 0.002 0.009 0.021
ESONI 0.195 0.066 0.013 MICSA 0.706 0.049 0.001
FUNCH 0.486 0.572 0.003 NOTCR 0.072 0.026 0.018
HOPLI 0.058 0.251 0.149 POELA 0.007 0.263 0.264
JORFL 0.038 0.181 0.137 POMNI 0.034 0.009 0.004
LEPGU n/a n/a n/a PTERY 0.253 0.003 0.046
LEPIS 0.021 0.339 0.004


For the fish ordination with count data, Axes 2 and 3 represent the most variation in

the assemblage data with a cumulative r2 of 0.76. Again, air temperature and stage are










the environmental variables most correlated with these axes (Figure 3-12, Table 3-11).

Axis 2 was most highly correlated with maximum air temperature (r2=0.380), while lake

stage (r2=0.287) and average air temperature (r2=0.333) are most correlated with Axis 3.

Fourteen of the fish species had an r2>0.2 for these two axes, with six of them being

indicator species (Figure 3-13, Table 3-12).

Table 3-11. Percent of variance explained (r2) by environmental variables for each axis
in the fish NMS with count data. Variables with r2 > 0.2 are highlighted.
Axis 1 2 3
Variable
Stage (m) 0.070 0.005 0.287
Fluc (cm) 0.008 0.001 0.064
Rain (cm) 0.003 0.000 0.024
Max (C) 0.242 0.380 0.189
Min (C) 0.346 0.303 0.243
Ave (C) 0.319 0.354 0.333


Table 3-


12. Percent of variance explained (r2) for each axis by species in the fish NMS
with count data. Indicator species with r2 > 0.2 are highlighted in blue, while
,211 .I, 24t \ t. t, 1 t 11 T


a ot er 1 ispec1es witl / L u.z. arie 1111111t 111d n yIe w.
Axis 1 2 3 Axis 1 2 3
Species Species
AMICA 0 0.008 0.445 LEPIS 0.359 0.044 0.07
ENNGL 0.171 0.57 0.239 LEPMAC 0.085 0.308 0.503
ERIMY 0.221 0.367 0.135 LEPMAR 0.033 0.185 0.624
ESOAM 0.159 0.165 0.372 LEPMI 0.227 0.359 0.164
ESONI 0.002 0.252 0.156 LEPPU 0.074 0.274 0.006
FUNCH 0.113 0.106 0.757 MICSA 0 0.64 0.294
HOPLI 0.108 0.03 0.194 NOTCR 0 0.152 0.058
JORFL 0.004 0.046 0.303 POELA 0.17 0 0.423
LEPGU 0.191 0.337 0.223 POMNI 0.054 0.077 0.016


The herpetofaunal assemblage detection/nondetection data were analyzed with

NMS, but results yielded only a one-dimensional solution. Stress on the final run was

51.36, which represents an unacceptable amount of variation from the original data set.

Values of stress greater than 20 indicate that the solution may be misleading, while

ordinations with stress values over 40 represent very little of the structure in the original










data matrix (McCune and Grace 2002). Due to these outcomes, NMS was considered

unsuccessful for the reptile and amphibian assemblage with detection/nondetection alone.

Count data for the reptiles and amphibians did yield a successful ordination. The

first and third axes explain a total of 69% of the variation in the assemblage composition.

Lake stage is highly correlated with Axis 1 (r2=0.267) and Axis 3 (r2=0.389), and water

level fluctuation is correlated with Axis 3 (r2=0.204), (Figure 3-14, Table 3-13). Eight

species were correlated with Axes 1 and 3 with r2>0.2, six of them being indicator

species (Figure 3-15, Table 3-14).

Table 3-13. Percent of variance explained (r2) by environmental variables for each axis
in the herpetofaunal NMS with count data. Variables with r2 > 0.2 are
highlighted.
Axis 1 2 3
Variable r2 r2 r2
Stage (m) 0.267 0.081 0.389
Fluc (cm) 0.006 0.002 0.204
Rain (cm) 0.000 0.001 0.057
Max (C) 0.052 0.020 0.031
Min (C) 0.022 0.034 0.091
Ave(C) 0.058 0.016 0.073

Table 3-14. Percent of variance explained (r2) for each axis by species in the
herpetofaunal NMS with count data. Indicator species with r2 > 0.2 are
highlighted in blue, while all other species with r2 > 0.2 are highlighted in
yellow.
Axis 1 2 3 Axis 1 2 3
Species r2 r2 r2 Species r2 r2
AGKPICO 0.041 0.188 0.012 PSEFLPE 0.042 0.079 0.04
AMPME 0.014 0.291 0.094 RANGR 0.025 0.229 0.229
APAFE 0.087 0 0.017 RANSP 0.185 0.061 0.263
CHESEOS 0.154 0.018 0.036 REGAL 0.384 0.055 0.109
FARABAB 0.004 0.024 0.409 SIREN 0.018 0.031 0.047
KINBA 0.083 0.215 0.351 STEOD 0.437 0.091 0.033
NERFAPI 0.275 0.023 0.498 TADRANGR 0.002 0.095 0.311
NERFL 0.097 0.258 0.03 TADRANSP 0.107 0.275 0.057









To determine the separation of the sample occasion clusters along lake stage and

average air temperature gradients means and ranges of each are shown in Figures 3-16

and 3-17 respectively. Clusters one and two have lake stage ranges lower and non-

overlapping with clusters four, five and six. Stage values for cluster 3 span much of the

ranges for every other group. Air temperature values are very similar for clusters one and

six, both being below and non-overlapping with clusters two, three and four. Cluster five

overlaps with most of the ranges of all other groups.

Proportion of Habitat Utilized by Focal Species

Figures 3-18 and 3-19 show the occupancy rate estimates for the fish species.

Several species exhibited a decrease in site occupancy from the spring to the fall season,

most pronounced in M. salmoides, L. macrochirus and L. microlophus. Lepomis

microlophus, L. gulosus and E. glorious occurred at all transects during the spring.

Lepisosteidae spp., H. littorale and L. gulosus were present in all transects in the fall.

Lepomispunctatus had invalid occupancy estimates during the spring due to low

detection probabilities, with the same problem for L. macrochirus and L. microlophus in

the fall. In the spring, the only fish species that used survey-specific detection probability

models for occupancy estimation were E. glorious and L. macrochirus, which indicates

that detection of these two species varied between sample occasions. All species were

modeled with time-constant detection probabilities in the fall, except E. glorious, P.

latipinna and Lepisosteidae spp.

Estimates of the proportion of sites occupied for the eight focal herpetofaunal

species are presented in Figures 3-20 and 3-24. Site occupancy was similar in the spring

and fall for most species. There were not enough data to make valid estimates for A.

means in the spring and for R. sphenocephala and N. fasciatapictiventris in the fall. All









species were modeled with the time constant detection probability except N. floridana

and R. sphenocephala in the spring. With the exception ofR. sphenocephala and A.

means, most species were estimated to be in 90-100% of all transects at some point in the

year.

Discussion

Trap Comparisons

The 1.3 cm (0.5 in) mesh traps were better at capturing the species of interest,

including reptiles, amphibians, centrarchids and exotic catfish. Although it had been

assumed that many species would be under-represented with the larger mesh, the few

unique species caught in the 0.6 cm (0.25 in) mesh traps were not of particular interest to

this study. The fish species included mosquitofish and flagfish, which are not suspected

of being impacted by the removal of the vegetation from the littoral zone. The one black

swamp snake was actually stuck about halfway through the mesh, indicating that even 0.6

cm mesh may not be small enough to capture this species representatively. The size and

shape of the traps and openings may also have affected capture probabilities of different

species. Also, the bare metal hardware cloth may be more visible to the animals than the

dark green vinyl-coated hardware cloth of the larger mesh traps. In comparing

practicality of sampling with each type of trap, the 0.6 cm commercially manufactured

traps were much less durable and prone to breaking during use, as well as being more

expensive ($16.16 each for 0.6 cm vs. $11.00 each for 1.3 cm). Given these factors, the

1.3 cm mesh traps were more useful than the 0.6 cm for sampling in this habitat.

Species Richness

Ectothermic fish communities respond to water temperature due to its effect on

metabolism and growth rates, especially for juveniles (Holt 2002). This relationship









determines the length of time that species will benefit from residing in a given habitat.

Water temperature is closely associated with air temperature due to the shallow aquatic

habitat. Therefore, it is not surprising that the average air temperature is negatively

correlated with fish species richness. In the summer season, water temperature may

exceed tolerance limits for sensitive fish such as select species of sunfish, causing them

to leave the habitat and resulting in fewer fish species. Temperature also indirectly

determines the amount of dissolved oxygen in the water; however this variable was not

measured in the field.

Lake stage, rainfall, and water level fluctuations were significant predictors of

herpetofaunal richness, all of which have a relationship to water depth at the fixed trap

locations. The gradually sloping contour of Lake Tohopekaliga causes small increases in

lake stage to flood broad expanses of previously dry habitat, allowing species to enter

into new habitat for spawning or foraging. Alternatively, moderate drops in lake stage

may cause a rapid decrease in the area of the littoral zone inundated with water, causing

species to emigrate or burrow, or else risk being trapped by unsuitable conditions.

Receding water levels may bring in more terrestrial species into the previously aquatic

habitat, such as cottonmouths, Florida water snakes, and Florida leopard frogs.

Alternatively, deeper water in the habitat may restrict species that prefer shallow water

and promote more aquatic species such as the common musk turtle.

Assemblage Composition

Since animal species move throughout the habitat and are detected with

probabilities less than one, and animal assemblages tend to be transient in nature, the

groups determined by cluster analysis are not very discrete with respect to species









composition. This resulted in fairly low indicator values, but with several being

significant nonetheless.

The four sample occasions in Group 1 had the lowest lake stage values and were

tied with Group 6 for the lowest average air temperatures. The Florida water snake and

Florida leopard frog are the herpetofaunal indicator species for this group, being the more

terrestrial of the focal reptile and amphibian species. Bowfins, golden topminnows,

bluegill and dollar sunfish are the fish indicator species. Bowfins are one of the most

tolerant freshwater fish species and are often called mudfishh" (Boschhung et al. 1995),

so it is not unreasonable that this species would occupy this shallow vegetated habitat.

Golden topminnows prefer to occupy lakes with abundant vegetation (Hoyer and

Canfield 1994), and therefore may be more tolerant of dense vegetation communities

than other species. It is not as easy to explain why the two sunfish species were

indicators for these shallow water depths. They may have been stranded by receding

waters and captured more easily in traps with puddles of water remaining.

Group 2 only had one sample occasion attributed to it. This sample was the last

one in late April before the traps were removed due to lack of water in the habitat.

Although not included in the indicator species analysis, this sample included no fish

species.

Only three sample occasions were clustered into Group 3. During these hot

summer samples, the lake stage was rapidly rising, and trap effort was reduced pending

appropriate water depths at the trap sites. The striped crayfish snake was the main

indicator species for this group. Being a specialist predator on crayfish (Godley 1980),









prey availability in recently flooded habitats may have been the driving factor for their

indicator values on these occasions.

Eight samples fell into Group 4, which had no indicator species attributed to it.

These sample occasions occurred in early fall, with the highest lake stages and average

air temperatures.

Group 5 occurred mainly in the summer months (eight samples), but also contained

two samples in February. Water depths were moderately high. Air temperatures were

low in February and high in summer, spanning a wide range of temperatures. The sole

indicator species for this group was the common musk turtle. This species is known for

being highly aquatic, leaving the water only to nest (Wygoda 1979, Gibbons et al. 1983).

It also is active for the widest temperature ranges of any other kinosternid species in

North America, being able to retreat to deeper waters to buffer the effects of air

temperature extremes (Mahmoud 1969, Ernst 1986).

The last cluster, Group 6, included seven samples, with six occurring February

through April and the other one in November. Low air temperatures and moderately high

but dropping lake stages characterize these sample occasions. The two Esox spp. (pike)

are the indicators for this group. These species breed from February to March in the

south, spawning in densely vegetated habitats less than 50 cm deep (Billard 1996). This

may explain their presence in the habitat during these environmental conditions.

Influence of Temporal Gradients on Assemblage

Detection/nondetection data were used for ordinations, and are generally

recommended when comparing habitat distributions of species (Hayek 1994), and when

sample unit heterogeneity is large (McCune and Grace 2002). Counts were also used to

compare results obtained by the two types of data, but although agreement between the









two provides more support, lack of detection probabilities make count data less valuable.

For the vertebrate and fish ordinations, results were similar for both types of data. The

herpetofaunal ordination was unsuccessful with detection/ nondetection data, and

therefore counts were used solely.

Average air temperature and lake stage came out as the most important variables

correlated with the axes representing variation in species composition in the vertebrate

and fish assemblages. As mentioned before, temperature influences growth rates for

young fish, as well as the amount of dissolved oxygen in the water. Both of these factors

limits the time that fish are able to occupy a habitat. Physical access to heavily vegetated

habitat is also limited by water depth, which is controlled by lake stage. This determines

the volume of water the animals have to move through, as well as the effect of vegetation

density in the water column. However, for herpetofaunal species alone air temperature is

not associated with variation in the species composition. In this case, lake stage is most

important, with water level fluctuation also showing a correlation with one of the axes.

Lake stage probably dictates movements of species that do not show site fidelity, in

response to habitat requirements and prey availability. For species that are not known to

move long distances, for example sirens and amphiumas, low water levels trigger

burrowing activities (Aresco 2001), thereby reducing capture opportunities.

Proportion of Habitat Utilized by Focal Species

Due to the large-scale removal of pickerelweed from the littoral zone of Lake

Tohopekaliga during enhancement activities, it was important to investigate the spatial

distribution of species in this habitat. Site occupancy analyses were used to estimate the

proportion of this habitat type that was used throughout the year by various fish, reptile

and amphibian species. While some species are temporally and spatially pervasive in the









habitat (e.g., warmouths, bluespotted sunfish, Florida green water snakes, sirens, striped

mud turtles, and pig frogs), others seem to use the Pontederia cordata zone

intermittently. Of the fish species, sailfin mollies, spotted sunfish, chubsuckers, and

largemouth bass were found in a moderate proportion of transects (30-70%) in both the

spring and fall. Bluegill and redear sunfish were both in a high proportion of sites

(>80%) in the spring, but were found in less than 20% of the transects in the fall. This

trend may be due to juvenile fish using the littoral zone for foraging and predator

avoidance during the spring when suitable physicochemical conditions permit survival

(Werner and Hall 1979, Crowder and Cooper 1982, Werner and Hall 1988, Chapman et

al. 1996). Unmeasured environmental characteristics such as low dissolved oxygen may

have kept the sunfish out of the thick vegetation after the summer low-water spell

(Miranda and Hodges 2000). Gars and armored catfish went from about 65-80% of the

sites in the spring to 100% occupancy in the fall. These two species are far more tolerant

of harsh environmental conditions than most sunfish due in part to their capacity for

aerial respiration (Boschung et al. 1995, Brauner et al. 1995). For the armored catfish,

the greater presence in the fall may be due to the breeding season and sufficiently high

water levels for nesting (Mol 1993).

Most reptile and amphibian species occupied a similar proportion of sites in both

the spring and fall. Florida leopard frogs and Florida water snakes were only captured

when water levels were very low, which restricted reasonable estimates of site occupancy

to the spring season. Since the heavily vegetated littoral zone is known as prime habitat

for several of these species due to life history requirements, it is not surprising to find

most of the focal species in such a high proportion of the sites (>70% occupancy).
































Figure 3-1. Crayfish and minnow trap in P. cordata habitat.
















































Figure 3-2. Locations of 2002 P. cordata sampling transects in Lake Tohopekaliga

























2102 3102 4102
2/1/02 3/1/02 4/1/02


5/1/02 6/1/02 7/1/02 8/1/02 9/1/02 10/1/02 11/1/02 12/1/02


Date


Figure 3-3. 2002 Vertebrate species richness estimates by sample date, with points
representing richness for the time between the last sample occasion and the
sample date.


2/1/02 3/1/02 4/1/02


5/1/02 6/1/02 7/1/02 8/1/02 9/1/02 10/1/02 11/1/02 12/1/02 1/1/03


Date
Figure 3-4. 2002 Fish species richness estimates by sample date, with points representing
richness for the time between the last sample occasion and the sample date.


III
0 ** .. .


*. ; .. .**'"+" -7 "."
: : : 0 ~+.,""


0o 1
1/1/02


1/1/03


0 +
1/1/02







50



40





S30




20





10
S. .............................
.






0
0 i i i i i i ii

1/1/02 2/1/02 3/1/02 4/1/02 5/1/02 6/1/02 7/1/02 8/1/02 9/1/02 10/1/02 11/1/02 12/1/02 1/1/03

Date


Figure 3-5. 2002 Herpetofaunal species richness estimates by sample date, with points
representing richness for the time between the last sample occasion and the
sample date.








51



18 Cluster

A1
@2
30 31 3
33
4
S5
15 34 28 6
V 2V Stage(m)
CO 2 Ave(C) 16
.) 2 V 26 2 25 3 1
X V
< 3
32
21
13 27
S V
24 0
v, 6 6




35 7
o0 36
10
12 11 A
A A

Axis 1


Figure 3-6. NMS ordination of sample units in vertebrate species space using
detection/nondetection data. Points represent sample occasions and distances
between points show the relative differences in species composition. The
length of each line is proportional to the strength of the correlation between
the environmental gradient and the synthetic axes.













CHESEOS
+


PTERY
+


POELA
+4-


NERFAPI NONE
++


JORFL


Stage(m)


APAFE
+


TADRANGR
+ NOTCR MICSA
+ +
LEPM ESOAM ESONI
W +


FARABAB
+ LEPMACFUNCH
++


AMICA


RANSP
+


AGKPICO
+


POMNI
+


TADRANSP


Axis 1



Figure 3-7. NMS ordination of vertebrate species in sample unit space using
detection/nondetection data. Points represent average species positions with
respect to sample units. The length of each line is proportional to the strength
of the correlation between the environmental gradient and the synthetic axes.








53




23
31 Cluster
30 A
3
4
17 Stage(m) 29 V1 5
Ave(C) 32 6
18 V 27
18 2
26 V
V 22
28 33 V
19 2 20
CO 16 V 24
34
S25 V




335
V 5



35
4 6 80 9

7 10
0A
15
36
11 12


Axis 1


Figure 3-8. NMS ordination of sample units in vertebrate species space using count data.
Points represent sample occasions and distances between points show the
relative differences in species composition. The length of each line is
proportional to the strength of the correlation between the environmental
gradient and the synthetic axes.














CHESEOS
+


Stage(m)


ENN
REGAL LI


PTERY
+


Ave(C)

LEPISHOPLI STEOD
\-r ++
S AMPME

KINBA
GL -lfI+ TADRANGR
EPPU +


+ -'1ICSA
ERIMY' ESOAM
LEPMI 'ESONI +
+ +


LtIVI/KI
+
LEPMAC
+


FUNCH


POMNI
+


FARABAB
+
POELA
+
NONE NERFAPI
+ +
AMICA


JORFL
+


RANSP TADRANSP
+


Axis 1



Figure 3-9. NMS ordination of vertebrate species in sample unit space using count data.
Points represent average species positions with respect to sample units. The
length of each line is proportional to the strength of the correlation between
the environmental gradient and the synthetic axes.


PSEFLPE
+


APAFE
+


NOTCR
+


AGKPICO
+













23
V19
17


3
v 16


22
2729Ave(C)33
_427Stage(m 34
34
20

2
31 28
V


15 32
V


7 64
0 00
A

8


30 I


Cluster
S1
3
4
S5
0 6


Axis 1


Figure 3-10. NMS ordination of sample units in fish species space using
detection/nondetection data. Points represent sample occasions and distances
between points show the relative differences in species composition. The
length of each line is proportional to the strength of the correlation between
the environmental gradient and the synthetic axes.


























i ERIMY
.A + POELA
x
< LEPMAR
ESOAM
POMNI ESONI + LEPMI
+ +
NOTCR
+





FUNCH
LEPMAC
+
AMICA JORFL
+ +

Axis 1




Figure 3-11. NMS ordination of fish species in sample unit space using
detection/nondetection data. Points represent average species positions with
respect to sample units. The length of each line is proportional to the strength
of the correlation between the environmental gradient and the synthetic axes.















9
o0
7
10
A
6
8 4 5
36
A


33 Ave(C)
26 34
V V


Cluster
S1
3
4
S5
0 6


Axis 2


Figure 3-12. NMS ordination of sample units in fish species space using count data.
Points represent sample occasions and distances between points show the
relative differences in species composition. The length of each line is
proportional to the strength of the correlation between the environmental
gradient and the synthetic axes.

























0)
*.U ESOAM POMNI
< + +

+ +

ERIMY
ENNGL-
+
LEPGU
LEPPU



Stage(m)

Ave(C) HOPLI EPIS
++

Axis 2



Figure 3-13. NMS ordination offish species in sample unit space using count data.
Points represent average species positions with respect to sample units. The
length of each line is proportional to the strength of the correlation between
the environmental gradient and the synthetic axes.




























28 34
'V


5
23
V Fluct(cm)
25
Stage(m) 3

32
33 30 V


Cluster
A 1
S2
3
4
S5
0 6


Axis 1


Figure 3-14. NMS ordination of sample units in herpetofaunal species space using count
data. Points represent sample occasions and distances between points show
the relative differences in species composition. The length of each line is
proportional to the strength of the correlation between the environmental
gradient and the synthetic axes.









60





"NP Cluster
A 1
*2
FARABAB 2
3
PSEFLPE 4
S5
06
NERFAPI
+ LADRANSP
APAFE +

+

+
A GPICo

TADRANGR



+
"RANGR


++
SIREN 4




Fluct(cm)
Stage(m)



REGAL




+
CHESEOS

Axis 1



Figure 3-15. NMS ordination of herpetofaunal species in sample unit space using count

data. Points represent average species positions with respect to sample units.

The length of each line is proportional to the strength of the correlation

between the environmental gradient and the synthetic axes.












































Figure 3-16.





30

28

o 26
0

24

22
E
I- 20

g) 18
03

> 16

14

12


1 2 3 4 5 6

Cluster


Average and range of lake stage values by cluster.











0


1 2 3 4 5 6


Cluster


Figure 3-17. Average and range of air temperature values by cluster.








62




1.1

1.0 0 0 0

0.9

0.8

Z 0.7

0.6
F--
o 0.5

o 0.4

0.3
*
0.2

0.1

0.0 -

-0 .1 ....
POELA LEPPU ERIMY LEPIS MICSA HOPLI LEPMAC LEPMI LEPGU ENNGL

FISH SPECIES

NAIVE ESTIMATE
O PROPORTION OF SITES OCCUPIED WITH STANDARD ERROR



Figure 3-18. Site occupancy estimates for focal fish species in spring 2002.



1.1

1.0 -0 0 0

0.9

0.8

0.7 -

0.6

0 0.5 -

S0.4 -

S0.3

0.2 -

0.1 -

0.0 0 0

-0.1
POELA LEPPU ERIMY LEPIS MICSA HOPLI LEPMAC LEPMI LEPGU ENNGL

FISH SPECIES

NAIVE ESTIMATE
O PROPORTION OF SITES OCCUPIED WITH STANDARD ERROR



Figure 3-19. Site occupancy estimates for focal fish species in fall 2002.

















1.0 0 0 0 0

0.9

0.8



0.6 { .






0.2
0.1
of-
O 0.5
0 0.4

0.3

0.2

0.1

0.0 -

-0.1 .
RANSP STEOD AMPME RANGR NERFAPI KINBA SIREN NERFL

HERPETOFAUNAL SPECIES

NAIVE ESTIMATE
O PROPORTION OF SITES OCCUPIED WITH STANDARD ERROR



Figure 3-20. Site occupancy estimates for focal herpetofaunal species in spring 2002.


0 0
o o


RANSP STEOD AMPME RANGR NERFAPI KINBA SIREN NERFL

HERPETOFAUNAL SPECIES

* NAIVE ESTIMATE
O PROPORTION OF SITES OCCUPIED WITH STANDARD ERROR


Site occupancy estimates for focal herpetofaunal species in fall 2002.


Figure 3-21.














CHAPTER 4
ASSEMBLAGE ACROSS VEGETATION COMMUNITIES

Introduction

The objective of this section is to investigate the influence of vegetation type and

water depth on spatial variation in the aquatic vertebrate community. The three main

research questions are 1) are we able to estimate population parameters such as

abundance and density for the focal species using trapping grids or webs, 2) is there

spatial variation in the aquatic vertebrate assemblage, and 3) do the vegetation

communities or water depths influence the spatial variation for the individual focal

species?

Field Methods

Grid and Web Sampling

A pilot mark-recapture protocol was employed in the summer of 2002 in order to

estimate activity ranges, abundances and densities of species of interest. Trap points

were arranged in a square grid pattern (White et al. 1982). In order to have uniform

sampling effort within the grids, one minnow and one crayfish trap was placed at each

point. All newly captured reptiles and amphibians were weighed and measured, then

tagged with Passive Integrated Transponders (PIT tags), which are small microchips

inserted under the skin to permit individual identification when scanned. Each animal

was released at the capture location after being worked up. Traps were checked daily for

PIT tagged individuals, and all new individuals were measured and tagged during each

sampling event. In order to satisfy the assumption of closure for analysis and minimize









temporal variation in detection probabilities, the grids were sampled for 5-7 days. This

technique was used only for the most abundant species, since detection probabilities

would be too small to make accurate estimates of any others. These species included A.

means, Siren spp., N. floridana, and K. baurii.

The first grid consisted of 49 trap points, each three m (9.8 ft) apart, in a seven by

seven grid ("GRID1"), and sampled for seven consecutive days in July. Next, 100 trap

points were placed in a 10 by 10 grid ("GRID2A"), five meters (16.4 ft) apart, and

sampled for six consecutive days in August. The same grid was then sampled for five

non-consecutive days ("GRID2B"), using frozen sardines as bait, until it was decided that

the bait was logistically impractical. The last design was 100 trap points ("GRID3"),

spaced three meters (9.8 ft) apart, in a 10 by 10 grid. This was sampled for five times

over seven days, due to logistic problems. Attempts to sample several times within a day

were also made, but were discontinued immediately due to low numbers of captures.

In addition to these grids, a trapping web was attempted in September 2002. The

web is a variation of point-transect distance sampling, typically used for small terrestrial

animals. It is designed to have lines of traps radiating out from a center point, forming a

gradient of sampling effort and detection probabilities. Data from each concentric ring is

grouped according to distance intervals (Anderson et al. 1983). This web consisted of

eight radiating arms of 12 trap points each, placed at three-meter (9.8 ft) intervals, for a

total of 96 trap points. The benefit of this method is that it uses only the initial captures,

so the recapture rate is irrelevant. The assumptions of this method are that all animals

near the center of the web are captured, the size of the web is large relative to the

movements of the animals, distances are measured accurately from the center, and









individual captures are independent events. Sampling continues daily until no new

animals are caught at the center of the web, indicating 100% detection at the center

(Anderson et al. 1983) Since it was evident that the assumptions were not met in this

web, sampling was discontinued after six sample occasions.

Whole-Lake Sampling

After the postponement of the fall 2002 drawdown, we began sampling again with

modifications to the 2002 temporal sampling protocol. The goal was to investigate

differences in vertebrate habitat usage of vegetation communities beyond the P. cordata

zone, as well as varying water depths. Eighteen new locations were randomly chosen for

transects (Figure 4-1), since the sampling of 2002 had disturbed the habitat in some of the

previous locations. Sampling sites in Goblet's Cove were included and none were placed

in the disturbed stretch of shoreline in the southern part of the lake. Transects were

placed at least 200 m (656 ft) apart. At each transect, there were four trap points, each

with a minnow and crayfish trap. However, instead of placing the trap sites at fixed

locations in the habitat as in 2002, they were placed at fixed depths and moved with the

water level. When the water was rising or remaining stationary, each transect had four

trap points located at 15, 30, 45, and 60 cm (6, 12, 18, 24 in) deep (Figure 4-2). During

falling water levels, the trap points were placed at 30, 45, 60, and 75 cm (12, 18, 24, and

30 in) deep, except when falling lake stages were not predicted. The traps were still

checked once weekly, and at each sampling occasion, the traps were moved to the

appropriate depth. This resulted in trapping animals at specific depth ranges over the

course of the week, with similar water depths between sample occasions. For example if

a trap point was located at the 30 cm (12 in) depth and the water level rose several

centimeters during the week, the trap was considered to be sampling the 30-45 cm (12-18









in) water depth. For falling water levels, the 30 cm (12 in) traps were sampling the 15-30

cm (6-12 in) depth. Percent cover of vegetation species was also estimated for a 2 m (6.5

ft) radius around every trap site on each sampling occasion.

Continuous sampling was conducted from 1/30/2003 to 1/5/2004, during which

time traps were located up in the shallow grassy habitat at high water levels

(characterized mainly by Luziolafluitans and Panicum repens), through the thick

emergent habitat (with Pontederia cordata and Typha domingensis), down to more open

water zones (with Hydrilla verticillata and floating leaf species Nuphar luteum and

Nymphaea odorata) at lower water levels. Besides the added environmental variables,

this new protocol also allowed us to sample year-round, instead of having to remove the

traps at moderately low water levels. Sampling ended on 1/5/2004 at about 15.5 m (50.8

ft) NGVD, with lake stage dropping due to the 2003 drawdown.

Analysis Methods

Population Estimates and Movement for Herpetofaunal Species

Trapping grids are known to exhibit "edge effects" due to animals near the edges of

the grid moving in and out of the sampled area. To account for this phenomenon a

boundary strip is typically estimated and added to the grid area to estimate an effective

sampling area. Wilson and Anderson (1985) propose using the mean maximum distance

moved (MMDM) by animals recaptured at least once to estimate the activity range of the

species. Although lacking a solid theoretical explanation, this method works well in

simulations. The alternative method, nested grid design, requires a large data set for

estimation of density (Williams et al. 2002). Since our data were fairly sparse with

recaptures, we used the MMDM method to calculate the effective areas of the trapping

grids for each species.









Low numbers of recaptures were attained for each grid; so in order to calculate the

MMDM for each species movement distances were pooled from all grids and the

trapping web. To calculate the diagonal distances within the trapping grids, the

Pythagorean theorem was used: [a2+b2=c2], where sides a and b are sides of known

lengths. For diagonal distances in the trapping web, [a2=b2+c2-2bc(cosA)] was used,

where A is the degree measure between sides b and c of known length (Larson et al.

1994). For each species, MMDM and its variance were calculated using formulas from

Wilson and Anderson (1985). The widths of the boundary strips were estimated as half

the MMDM for each species. The effective grid areas and associated variances were then

calculated for each grid per species (Wilson and Anderson 1985).

Program MARK (White and Burnham 1999) was used to estimate population sizes

for each species per trapping grid. Estimates were obtained using models M(o), (constant

capture probabilities), and M(t), (time dependent capture probabilities). The Akaike's

Information Criteria (AIC) were compared to determine which model best fit the small

data set. Density was then calculated for each grid per species by dividing the population

size by the effective sampling area (Wilson and Anderson 1985).

Capture Success for Focal Species

Basic analyses were conducted on the 2003 data to look for trends in the data

associated with the main sampling variables involved in this protocol. All trap points

were divided into groups according to vegetation community (see results section below

for descriptions) and water depth at each trap location. Ten species were examined, two

each of salamanders (Siren spp., A. means), snakes (both Nerodia spp.), turtles (K. baurii

and S. odoratus), frogs (both Rana spp.) and fish (H. littorale and M. salmoides). The

reptile and amphibian pairs represent species with similar life history traits but which









have slightly different habitat requirements or preferences. They were also the most

frequently captured reptile and amphibian species in this study. The two fish species

were chosen to characterize opposite ends of the spectrum of habitat selection. While the

armored catfish is a generalist species with great tolerance for low dissolved oxygen, high

temperatures, thick vegetation and other extreme environmental variables (e.g., Nico and

Fuller 1999), largemouth bass and other sunfishes are thought to be highly intolerant of

these same habitat characteristics (e.g., Allen and Tugend 2002).

Capture success was calculated as the number of captures per species divided by

the total number of trap points for the particular variable of interest. This usually resulted

in a very low frequency, due to the large number of trap points and low detectability of

species. An arcsine squareroot transformation was applied to all success values, in order

to spread the ends of the scale, while improving normality for the proportion data

(McCune and Grace 2002). This allowed the relative values to show up more clearly

while reducing the effect of large sample units. The assumption of equal detectability of

the different species between habitats or water depths may be violated, but detection

probabilities cannot be calculated for this particular analysis. However, uniform

sampling methods were used over space transectss) and time (sampling occasions) to

reduce variability in detection.

For this sampling protocol, dependence of trap placement upon water depth (i.e.

lake stage) resulted in unequal sampling for vegetation communities. In addition, the

vast number of trap sites (n= 3,426) and sparse nature of the data made multivariate

analyses virtually impossible. For example, dividing data into groups (either subjectively

or with cluster analysis) depending on sample occasions, water depth or vegetation









communities would neglect important differences in the other variables and/or result in

groups of vastly unequal numbers of trap points. Attempted NMS analyses of all trap

sites (ungrouped) yielded no results due to the great number of zeroes in the matrices.

The data were not even appropriate for most univariate analyses. For example, chi-

square analyses of capture success would indicate whether there were significant

differences in the counts of focal species between vegetation types or water depths,

however the large sample sizes invariably lead to significant differences. Repeated-

measures analysis of variance was considered to test the differences between water

depths over time, however the data were too sparse to divide the counts between both

sample occasions and water depths. Species richness was also inestimable because of the

frequent nondetection of species and unequal sample sizes. As a result of the

complicated nature of the data, the analyses were largely descriptive in nature. These

descriptions of habitat usage rely mainly on comparisons of capture success across two

categorical environmental variables: water depth and vegetation community.

Results

Population Estimates for Herpetofaunal Species

Each species showed different movement distances over the sampling grids and

web. Figure 4-3 shows the mean maximum distances traveled and variances, along with

the associated widths and variances of the boundary strips for each species. Maximum

distances traveled for A. means, Siren spp., N. floridana, and K. baurii were 18 m (59 ft),

24 m (79 ft), 34 m (112 ft), and 59 m (194 ft) respectively. These movements are fairly

large relative to the sizes of the grids, with 72 m (236 ft) being the absolute maximum

distance between any two traps during all sampling. This indicates that the assumption of

closure was violated. Table 4-1 contains the percent increase in the size of each grid










when the boundary widths for each species were added to the sizes of the grids. While

the effective sampling areas are fairly acceptable for the amphiumas and sirens (Wilson

and Anderson 1985), Florida green water snakes and striped mud turtles add excessive

area to the original sampling areas. After recognizing the fact that closure was violated in

these grids, population sizes and densities were estimated with unreliable accuracy, but a

best attempt was made given the data.

Estimates of population size and variances are shown in Table 4-1. Several times

the capture history data were so sparse that estimates could not be calculated for some

grids and species. The estimates generated with sufficient data often have large variances

due to low recapture probabilities. The null model of no variation in detection

Table 4-1. Grid sizes and population estimates by mark recapture methods.
Parameter AMPME SIREN NERFL KINBA
GRID1 Actual size of grid (ha) 0.0324 0.0324 0.0324 0.0324
Est. effective sampling area and variance (ha) 0.078 (0.40) 0.08 (0.93) 0.12 (1.79) 0.16 (9.17)
Percent of original grid (%) 239 264 376 493
Est. population size and variance (# indivs.) n/a n/a 16(184) 7 (30)
Density estimate (#/ha) n/a n/a 131 44
GRID2A Actual size of grid (ha) 0.2025 0.2025 0.2025 0.2025
Est. effective sampling area and variance (ha) 0.30 (1.66) 0.32 (3.71) 0.39 (6.08) 0.46 (27.83)
Percent of original grid (%) 149 157 192 225
Est. population size and variance (# indivs.) n/a n/a n/a n/a
Density estimate (#/ha) n/a n/a n/a n/a
GRID2B Actual size of grid (ha) 0.2025 0.2025 0.2025 0.2025
Est. effective sampling area and variance (ha) 0.30 (1.66) 0.32 (3.71) 0.39 (6.08) 0.46 (27.83)
Percent of original grid (%) 149 157 192 225
Est. population size and variance (# indivs.) 14 (37) n/a 26 (99) 9 (5)
Density estimate (#/ha) 49 n/a 68 20
GRID3 Actual size of grid (ha) 0.0729 0.0729 0.0729 0.0729
Est. effective sampling area and variance (ha) 0.14 (0.72) 0.15 (1.65) 0.19 (2.94) 0.24 (14.27)
Percent of original grid (%) 187 201 267 332
Est. population size and variance (# indivs.) 19(117) n/a 25(24) 26(525)
Density estimate (#/ha) 142 n/a 133 108









probability was always selected over the time-varying capture probability model.

Density estimates based on these abundances are also included in Table 4-1.

Capture success for Focal Species

Figure 4-4 shows the types of vegetation communities that were sampled

throughout 2003. Each sampling occasion corresponds to one weekly sample, which

includes 72 trap points (4 trap points for each of 18 transects). Occasionally there were

less than 72 samples for a given sample occasion, usually because data recording for a

sample was inadvertently neglected, traps went unchecked due to dangerous weather, or

traps were missing. Each trap point was subjectively categorized in the field into

different vegetation communities, based on the dominant species present. The "Grass"

community is the closest shoreward, and is characterized by Panicum repens, Luziola

fluitans, Juncus effusus, and Eleocharis spp. Lakeward from this is the "Rooted-HE",

which refers to the herbaceous emergent species, especially Pontederia cordata and

Typha domingensis. The community termed "G/HE" is the border of the grass and

rooted herbaceous emergent zones, which had enough samples to be a separate category.

"Floating-HE" is the floating mat community, consisting ofP. cordata, Bidens spp.,

Ludwigia leptocarpa and a variety of other species. Out past the herbaceous emergent

communities are the deeper "Outward" communities. These include floating-leaf

emergents (Nelumbo lutea, Nymphaea odorata, andNuphar luteum), submersed plants

(Hydrilla verticillata), and deep emergents (Paspalidium geminatum). Trap points that

were on the borders between distinct vegetation communities, or were part of

communities with too few samples to have its own category, were classified as "Mixed".

Since transects were randomly chosen and the trap points were moved with the water

level, there was no way to collect equal numbers of samples from each community.









Figures 4-5 and 4-6 show results from capture success comparisons for

salamanders. Both species tend to be captured more frequently in the outer herbaceous

emergent vegetation communities and community edges, as well as greater water depths.

Pig frogs seemed to show a preference for grassy habitats and community edges,

decreasing towards more outward vegetation communities, while leopard frogs did not

show much of a pattern (Figure 4-7). However, much stronger trends appeared with

water depth (Figure 4-8). Both species, especially the leopard frogs, showed an inverse

relationship to water depth.

The two snake species showed divergent trends in capture rate. In Figure 4-9,

Florida green water snakes (Nerodiafloridana) appeared more in the rooted pickerelweed

communities, while the Florida water snakes (Nerodiafasciatapictiventris) did not have

quite as strong a tendency to be in specific habitats. Water depth seemed to be more

important in Florida water snake occurrence (Figure 4-10), with most being trapped in

shallow water and decreasing steadily with depth. The Florida green water snake did not

have such a trend. Figures 4-11 and 4-12 show that common musk turtles were captured

more frequently in vegetation habitats furthest from shore, as well as deeper water

depths. Striped mud turtles on the other hand did not have strong trends, but peak in

rooted pickerelweed habitats and intermediate water depths.

Largemouth bass did not show a strong affinity for any certain habitat, however

they were captured slightly more frequently in rooted herbaceous emergent and

borderline communities (Figure 4-13). They also appeared most in intermediate water

depths, (Figure 4-14), e.g.45-60 cm (18-24 in) deep. Bass captured in the traps were

juvenile fish, with total lengths in 2003 ranging from 5.1-14.0 cm (2-5.5 in), (n=60).









The armored catfish were found mostly in border vegetation communities (Figure 4-13).

They also were positively correlated with water depth (Figure 4-14).

Discussion

Population Estimates for Herpetofaunal Species

The small activity ranges estimated for the amphiumas and sirens were similar to

what have been found in other studies (Gehlbach and Kennedy 1978, Sorensen 2004).

Maximum distances were higher in this case, (18 m (59 ft) vs. 5 m (16 ft) for amphiumas

and 24 m (79 ft) vs. 10 m (33 ft) for greater sirens (Sorensen 2004)). These estimates

suggest that lack of movement in these animals make them susceptible to mortality

during muck removal operations. The sizes of the grids were probably not large enough

to make reasonable activity range estimates for the Florida green water snake or striped

mud turtle. Bancroft et al. (1983) documented a Florida green water snake moving 223

m (731 ft) in less than two hours. As another example, Nerodia taxispilota (brown water

snakes) have been documented moving distances greater than 1 km (0.62 miles) (Mills et

al. 1995). Mahmoud (1969) found maximum distances for several species of kinosternid

turtles, including 525.5 m (1,723.6 ft)for S. odoratus, 435.3 m (1,427.8 ft) for

Kinosternonflavescens (yellow mud turtle), 408.4 m (1,340 ft) for Kinosternon

subrubrum (Mississippi mud turtle), and 93.9 m (308 ft) for Si. ,i/hel ii'n carinatus

(Mississippi musk turtle).

Due to low recapture rates and large movements of individuals, poor density

estimates were attained with these protocols. Even when the simplest models were

utilized to estimate population sizes (the estimated number of animals with no account of

area sampled), variances were unacceptably high. Even if the estimates had been

reasonable, the study would have been limited to a small number of species, a narrow









window of opportunity when the pickerelweed zone was completely inundated with

water, and non-random locations which possessed a wide enough band of habitat to

contain the grids within a relatively homogeneous habitat. Therefore, the mark-recapture

grids will not be utilized for post-enhancement sampling.

Capture Success for Focal Species

Vegetation communities offer different tradeoffs to animal species that inhabit

them. Variations in predator efficiency, prey type and abundance, or abiotic properties

associated with dissimilar macrophyte types strongly determine their use to aquatic

vertebrates (Miranda et al. 2000). Physical properties of plant species such as branching,

leaf shape and number, plant biomass and position throughout the water column affect

animals' ability to maneuver in the habitat, as well as physical and chemical properties

such as dissolved oxygen, nutrient levels, water temperature, light penetration and current

(Chick and Mclvor 1994). Welch (2004) conducted a thorough ecological investigation

into the vegetation communities of Lake Tohopekaliga prior to the 2004 enhancement.

One finding was that the soils associated with the intermediate littoral zone depths and

Pontederia cordata communities were highly organic and low in bulk density. On the

other hand, the shallow grassy communities and the various deeper water communities

had soils higher in bulk density, and therefore were sandier in composition. Substrate

alone may provide benefits or detriments to animal species, depending on their specific

life history traits.

Water depth at a given location has strong influences on vertebrate species

distribution and habitat use. It is the main determinant of the boundaries of the littoral

ecotone, limiting emergent aquatic vegetation growth to the limits of the water

fluctuation. The gradual slope of the shoreline causes small increases or decreases in









lake stage to flood or dry out broad expanses habitat, altering its use to different species.

Water depth also establishes the volume of water that aquatic organisms have to move

through, and can provide enough space for the presence of a thermocline (Miranda et al.

2000).

The aquatic salamander species are known for burrowing in the organic sediment

associated with dense vegetation as refuge from predators and drought (Etheridge 1990,

Conant and Collins 1998). In fact many of these large salamanders have been uncovered

during muck removal operations around Florida, even when there was water still

covering deeper areas of the lake (Aresco 2001). The findings of this study indicate the

same pattern, with highest capture success occurring in densely vegetated communities.

As mentioned previously, these communities are also most associated with low bulk

density and high organic composition of the soils (Welch 2004). This indicates that not

only may the dense vegetation provide ample forage and cover for these creatures, but

also the organic sediment (muck) is preferred for burrowing. While deeper water depths

yielded more sirens and amphiumas, it is possible that they are simply more active in

deeper water, increasing detection probabilities.

Higher capture success in shallow water sites was expected for leopard frogs, since

all individuals captured in 2002 occurred in April and November when water levels were

low. This species is known to travel relatively far from aquatic habitats, given proper

cover and shade from terrestrial vegetation, depending on soil moisture and dew to

prevent desiccation (Dole 1965, Conant and Collins 1998). Leopard frogs are

particularly dense in herbaceous vegetation around lakes, with plenty of protection and

food sources in the grasses (Kilby 1936). On the other hand, pig frogs are highly aquatic,









often being associated with emergent and floating vegetation (Conant and Collins 1998).

Pig frogs were captured the most in the "mixed" vegetation community. This category

was mainly represented by floating vegetation (water hyacinth) and borders between

communities, (mainly between the floating mats of emergent vegetation and submersed

vegetation). The dominance of pig frogs found in this community may indicate that

border communities provide a tradeoff between predator avoidance and prey availability.

Decreases in capture success with increased water depth and relative distance from shore

indicate that water depth was very influential in determining the presence of both species

of ranid frogs. The leopard frog in particular has a very strong decreasing trend with

water depth, which is consistent with its more terrestrial nature.

Water depth was also an important factor in the Florida water snakes' habitat

preference, which was expected since in the 2002 sampling most individuals were caught

in April and November, both during relatively low water periods. Water depth does not

seem to have as strong an influence on Florida green water snakes. To illustrate this

difference, Seigel et al. (1995) found that during a three-year drought in Ellenton Bay,

South Carolina, many N. fasciata left the habitat only seven days after it dried out, while

N. floridana never left in large numbers. While the abundance of snakes was generally

lower, it was five years before N. floridana was captured after the drought. As discussed

previously, water snakes typically show little site fidelity and are capable of long-range

movements (Bancroft et al. 1983, Mills et al. 1995). Several of the species show

ontogenetic niche shifts with age and size, often changing diet and habitat preference at a

certain size (Mushinsky et al. 1982, Mushinsky and Miller 1993). Florida water snakes

in particular are known for feeding on fish when young, and then at 50 cm (20 in) snout-









vent length they begin to feed almost exclusively on frogs (Mushinsky et al. 1982). The

traps used in this study mainly catch adult snakes, so prey (i.e. anuran) availability in

shallow habitats may result in this species' preference for shallower water depths.

Alternatively, Florida green water snakes are not so specialized, being caught with and

regurgitating a variety of fish, frogs, and even large sirens. While both species were

captured more frequently in the emergent vegetation communities, the water depth seems

to influence the presence of these species the most.

The turtle species have different life history traits that may explain observed

differences in capture success. For example, common musk turtles are highly aquatic,

rarely leaving the water except to nest. When water levels drop, at least in ponds, they

follow the water down and then burrow into the sediment to avoid desiccation (Wygoda

1979, Gibbons et al. 1983). Mahmoud (1969) suggests that Scite ii,,I/ it'l spp are more

dependent upon water depth than Kinosternon spp. The former appears to prefer water

depths greater than 30 cm and have been found in up to seven meters of water. As shown

in this study, there is a sharp increase in capture success associated with both lakeward

vegetation communities and deeper water depths. Alternatively, striped mud turtles are

much more terrestrial, usually dispersing over land during drought or heavy rainfall to

find alternative habitats (Bennet 1972, Wygoda 1979). On the other hand, Gibbons et al.

(1983) found that Kinosternon subrubrum experienced no increased emigration due to

drought in Ellenton Bay, South Carolina, since it is a fairly terrestrial species and is not

negatively affected by dry conditions. Maximum activity of striped mud turtles occurred

in 15 cm in Oklahoma (Mahmoud 1969). In this study, peak captures occur in rooted









emergent vegetation and intermediate water depths. All species of kinosternid turtles are

thought to prefer vegetated habitats to unvegetated ones (Mahmoud 1969).

The captures of young largemouth bass in all parts of the littoral habitat were

contrary to common fisheries doctrine. One would expect them to occur almost

exclusively in the open water/submersed habitats, due to physicochemical requirements,

as well as the physical barrier of the organic berm formed by the floating vegetation mats

(Moyer et al. 1995, Allen and Tugend 2002, Allen et al. 2003). However, relatively high

water levels evidently allow young bass and other centrarchid species to enter grassy

habitats, as well as inhabit the pickerelweed zone. Miranda et al. (2000) described

vegetated aquatic habitats as a mosaic of microhabitats within larger seemingly

inhospitable macrophyte stands. Although from a human's perspective the habitat may

seem uniformly unsuitable, fish can move both horizontally and vertically to find pockets

of suitable physical (e.g., temperature) and chemical (e.g., dissolved oxygen) water

conditions for survival. Perhaps this explains the bass' ability to move through this

landscape relatively unscathed. The armored catfish on the other hand, have a high

tolerance for poor water quality due to their ability to breathe air (Brauner et al. 1995).

They were still found at deeper water depths, but his was likely due to breeding

requirements in the littoral zone for greater than 0.3 m water depths (Hostache and Mol

1998).









































Figure 4-1. Locations of 2003 sampling transects in Lake Tohopekaliga.


60 cm
45 On
30 cm
15 cm


































Figure 4-2. Diagram of weekly trap placement at specified depths.


Striped mud turtle



Fl.green water snake



Siren



Amphiuma


-


-- U

- U


0 5 10 15 20 25 30 35 40 45 50 55 60 65 70

Distance (m)

W Maximum mean distance moved and variance
U Maximum distances moved per species


Figure 4-3. Mean maximum distances traveled with variances and maximum distances,
based on results of mark-recapture sampling.







82




75
> 70 -
C 65
E 60 II GRASS
0 55 (n=799)
L) G/HE
c 50 -
o (n=230)
45 ROOTED-HE
40 (n=1174)
S35 I FLOATING-HE
> (n=714)
= 30
U30 MIXED
L 25 (n=162)
20 OUTWARD
(n=347)
15-
E 10
S5-

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Sample Occasion


Figure 4-4. Number of trap sites sampled in each vegetation community per sample
occasion.
















O 0.35
Q_

S0.30


-. 0.25


0 0.20


U- 0.15
c
1-
5 0.10
0

0.05
CO
0-



. 0.00
U)


Figure 4-5. Salamander capture success by vegetation community.




o





P 0.15 -





E 0.10





S0.05


Co
09














E 0.10 -
E 0 15 30 45 60


Water Depth (cm)
I SIREN
0-










I AMPHIUMA



Figure 4-6. Salamander capture success by water depth.


G G/HE R-HE F-HE M O

Vegetation Community

I SIREN
W AMPHIUMA















0
1?

.. 0.25
Q_



. 0.20




S0.15

E
0
S0.10
cr



0.05


0)
c 0.00
o G G/HE R-HE F-HE M O

Vegetation Community

S PIG FROG
I SOUTHERN LEOPARD FROG



Figure 4-7. Frog capture success by vegetation community.


o 0 15 30 45 60

Water Depths (cm)

SPIG FROG
I SOUTHERN LEOPARD FROG



Figure 4-8. Frog capture success by water depth.
















. 0.25








0-
"| 0.20


a-
o
- 0.15
03
E
0
S0.10
-0




2 0.05


"II
U)
. 0.00
o G G/HE R-HE F-HE M O

Vegetation Communities

SFLORIDA GREEN WATER SNAKE
II FLORIDA WATER SNAKE



Figure 4-9. Snake capture success by vegetation community.


o 0 15 30 45

Water Depth (cm)

SFLORIDA GREEN WATER SNAKE
I FLORIDA WATER SNAKE



Figure 4-10. Snake capture success by water depth.



































2 G G/HE R-HE F-HE M O
<
Vegetation Communities
SSTRIPED MUD TURTLE
II COMMON MUSK TURTLE


Figure 4-11. Turtle capture success by vegetation community.


o 0 15 30 45

Water Depths (cm)

SSTRIPED MUD TURTLE
I COMMON MUSK TURTLE


Figure 4-12. Turtle capture success by water depth.















1?

o. 0.20
0-




P 0.15
03



0
a)


E 0.10

U)
r-

0.05





c 0.00
U)
Co


Figure 4-13. Fish capture success by vegetation community.


0 15 30 45 60

Water Depths (cm)

LARGEMOUTH BASS
I ARMORED CATFISH


Figure 4-14. Fish capture success by water depth.


G G/HE R-HE F-HE M O

Vegetation Communities

LARGEMOUTH BASS
II ARMORED CATFISH


.o
. 0.12
0-


S0.10
I)


0-
E 0.08
C-)
0





0.04
o
1-
E 0.06

UI)

0.04
1-
o

S0.02


0)
r 0.00
Co
<














CHAPTER 5
SUMMARY AND CONCLUSIONS

Review of Aquatic Vertebrate Community Dynamics in Lake Tohopekaliga

This study has documented the conditions of the fish and herpetofaunal

communities in the littoral zone of Lake Tohopekaliga prior to extreme habitat

modifications performed in 2004. The littoral zone in this eutrophic lake defined here

includes the entire vegetated shoreline, from the grassy vegetation community on the

shore that is occasionally inundated with water to the lakeward band of emergent

vegetation that is always flooded. Animals captured in this habitat represent species that

are dependent in some way upon its unique characteristics for their survival, including

still water, cover from predators or light, nesting substrate, organic sediment in which to

burrow or forage, and appropriate prey. Perhaps the most useful information from this

study so far has been the species list, which includes juvenile centrarchids and exotic

catfish. While it is no surprise to find most of the herpetofaunal species in this type of

landscape, the quantity and quality of fish captured was not expected. Previous research

indicates that sunfish cannot and do not inhabit heavily vegetated littoral habitats, while

the two species of exotic catfish captured have not been documented at all this far north

in Florida. Species such as warmouths, bluespotted sunfish, Florida green water snakes,

sirens, striped mud turtles, and pig frogs were found in all parts of the habitat throughout

the year, indicating that they may be threatened by the habitat alterations more than other

species.









Research efforts in 2002 were focused on the Pontederia cordata zone only, which

was the species most considered a nuisance by lake managers. The widespread removal

of this species threatens to disrupt the wildlife community structure in a large area of the

lake, making research of this specific vegetation community a necessary component of

the study. In 2003, sampling was conducted across vegetation communities within the

littoral zone in an attempt to understand how different focal species are distributed

throughout the habitat. The following sections will provide a review of the observed

effects of these environmental variables on species richness, community composition and

distribution of focal species.

Air Temperature

Average air temperatures over the course of each sample occasion accounted for

much variation in the aquatic vertebrate community. Both fish and total vertebrate

richness were negatively correlated with average air temperature over time. NMS

analyses also indicated that this variable explained large percentages of variance in the

species composition of the fish and vertebrate communities. Vertebrate assemblages

represented by Regina alleni and Esox spp. were associated with high air temperatures,

while groups with Nerodiafasciata pictiventris and Rana sphenocephala as indicator

species were found at lower air temperatures.

Lake Stage

Only herpetofaunal species richness was found to decline with increased lake

stages. This variable explained much of the variation in fish, herpetofaunal, and

combined vertebrate assemblages. In periods of high lake stages, .Sicil n, iih' odoratus

and Esox spp. were the indicator species for sample occasion clusters. N. fasciata




Full Text

PAGE 1

AQUATIC VERTEBRATE USAGE OF LITT ORAL HABITAT PRIOR TO EXTREME HABITAT MODIFICATION IN L AKE TOHOPEKALIGA, FLORIDA By ANN MARIE MUENCH A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2004

PAGE 2

Copyright 2004 by Ann Marie Muench

PAGE 3

This document is dedicated to my parents, Joseph F. and Mary K. Muench, whose love and support have strongly contributed to my academic, professional, and personal growth.

PAGE 4

iv ACKNOWLEDGMENTS I would like to thank my major advisor, Wiley Kitchens, for taking me on as a graduate student. He was a critical source of expert advice and encouragement, and was always accessible for consultation. I also thank my committee members, Madan Oli and Lauren Chapman, whose academic instruction and cr itical analysis of this thesis are much appreciated. My coworkers also contributed much to my education, and I am thankful. Funding for this research was provide d by the Florida Fish and Wildlife Conservation Commission (FFWCC). From this agency, Duke Hammond helped immensely with the direction of the study, and provided critical feedback on progress reports that we provided to the commission. The staff of the Kissimmee, FL, office of the FFWCC was helpful in facilitating our fi eld work at Lake Tohopekaliga. Bobbi Jo Cromwell from the Osceola County Department of Parks and Recreation allowed us to store all of the crayfish and minnow trap s on Makinson Island in Lake Toho. The field work for this study was conducte d through the time of many dedicated students and technicians from the Florida C ooperative Fish and Wildlife Research Unit. These stalwart coworkers included (in al phabetical order) Scott Berryman, Stephen Brooks, Janell Brush, Brenda Calzada, John Davis, Jamie Duberstein, Bruno Ferreira, Joey Largay, Kristianna Lindgren, Samantha Musgrave, April Norem, Derek Piotrowicz, Laura Pfenninger, Erik Powers, Vanessa Rumancik, John David Semones, Micheala Spears, Chris Tonsmeire, Paul Traylor, Zach Welch, and Christa Zweig.

PAGE 5

v TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES............................................................................................................vii LIST OF FIGURES.........................................................................................................viii ABSTRACT....................................................................................................................... ..x CHAPTER 1 MAIN INTRODUCTION............................................................................................1 Lake Ecosystem............................................................................................................1 Study Area....................................................................................................................2 Research Objectives......................................................................................................8 2 DESCRIPTIONS OF FOCAL SPECIES...................................................................10 Aquatic Vertebrate Habitat.........................................................................................10 Fish Species................................................................................................................11 Centrarchids (sunfish).........................................................................................11 Exotic catfish.......................................................................................................13 Herpetofaunal Species................................................................................................13 Amphibians..........................................................................................................13 Reptiles................................................................................................................15 3 ASSEMBLAGE WITHIN THE Pontederia cordata COMMUNITY.......................17 Introduction.................................................................................................................17 Field Methods.............................................................................................................17 Trap Descriptions................................................................................................17 Whole-Lake Sampling.........................................................................................20 Analysis Methods.......................................................................................................23 Trap Comparisons...............................................................................................23 Species Richness.................................................................................................23 Assemblage Composition....................................................................................25 Influence of Temporal Gradients on Assemblage...............................................25 Proportion of Habitat Utilized by Focal Species.................................................27

PAGE 6

vi Results........................................................................................................................ .29 Trap Comparisons...............................................................................................29 Species Richness.................................................................................................31 Assemblage Composition....................................................................................32 Influence of Temporal Gradients on Assemblage...............................................34 Proportion of Habitat Utilized by Focal Species.................................................40 Discussion...................................................................................................................41 Trap Comparisons...............................................................................................41 Species Richness.................................................................................................41 Assemblage Composition....................................................................................42 Influence of Temporal Gradients on Assemblage...............................................44 Proportion of Habitat Utilized by Focal Species.................................................45 4 ASSEMBLAGE ACROSS VEGETATION COMMUNITIES.................................64 Introduction.................................................................................................................64 Field Methods.............................................................................................................64 Grid and Web Sampling......................................................................................64 Whole-Lake Sampling.........................................................................................66 Analysis Methods.......................................................................................................67 Population Estimates and Movement for Herpetofaunal Species.......................67 Capture Success for Focal Species......................................................................68 Results........................................................................................................................ .70 Population Estimates for Herpetofaunal Species................................................70 Capture success for Focal Species.......................................................................72 Discussion...................................................................................................................74 Population Estimates for Herpetofaunal Species................................................74 Capture Success for Focal Species......................................................................75 5 SUMMARY AND CONCLUSIONS.........................................................................88 Review of Aquatic Vertebrate Community Dynamics in Lake Tohopekaliga...........88 Air Temperature..................................................................................................89 Lake Stage...........................................................................................................89 Water Depth.........................................................................................................90 Vegetation Community.......................................................................................90 Population Size Estimates...................................................................................90 Lake Tohopekaliga Habitat Enhancement..................................................................91 Future Aquatic Vertebrate Monitoring Plans.............................................................95 LIST OF REFERENCES...................................................................................................97 BIOGRAPHICAL SKETCH...........................................................................................105

PAGE 7

vii LIST OF TABLES Table page 3-1 All species captured in 2002, with species codes used in subsequent figures.........30 3-2 Species capture frequencies for th e 0.6 and 1.3 cm mesh minnow traps.................31 3-3 Indicator species analysis results..............................................................................33 3-4 Stress and instability resu lts from all NMS ordinations...........................................34 3-5 Percent of variance explained (r2) by environmental variables for each axis in the vertebrate NMS with det ection/nondetection data...................................................35 3-6 Percent of variance explained (r2) for each axis by species in the vertebrate NMS with detection/nondetection data..............................................................................35 3-7 Percent of variance explained (r2) by environmental variables for each axis in the vertebrate NMS with count data...............................................................................36 3-8 Percent of variance explained (r2) for each axis by species in the vertebrate NMS with count data.........................................................................................................36 3-9 Percent of variance explained (r2) by environmental variables for each axis in the fish NMS with detection/nondetection data.............................................................37 3-10 Percent of variance explained (r2) for each axis by species in the fish NMS with detection/nondetection data......................................................................................37 3-11 Percent of variance explained (r2) by environmental variables for each axis in the fish NMS with count data.........................................................................................38 3-12 Percent of variance explained (r2) for each axis by species in the fish NMS with count data.................................................................................................................38 3-13 Percent of variance explained (r2) by environmental variables for each axis in the herpetofaunal NMS with count data.........................................................................39 3-14 Percent of variance explained (r2) for each axis by species in the herpetofaunal NMS with count data................................................................................................39 4-1 Grid sizes and population estimat es by mark recapture methods............................71

PAGE 8

viii LIST OF FIGURES Figure page 3-1 Crayfish and minnow trap in P. cordata habitat......................................................47 3-2 Locations of 2002 P. cordata sampling transects in Lake Tohopekaliga................48 3-3 2002 Vertebrate species richne ss estimates by sample date.....................................49 3-4 2002 Fish species richness estimates by sample date...............................................49 3-5 2002 Herpetofaunal species ric hness estimates by sample date..............................50 3-6 NMS ordination of sample units in vertebrate species space using detection/nondetection data......................................................................................51 3-7 NMS ordination of vertebrate species in sample unit space using detection/nondetection data......................................................................................52 3-8 NMS ordination of sample units in vert ebrate species space using count data........53 3-9 NMS ordination of vertebrate species in sample unit space using count data.........54 3-10 NMS ordination of sample units in fish species space using detection/ nondetection data......................................................................................................55 3-11 NMS ordination of fish species in sa mple unit space using detection/nondetection data........................................................................................................................... 56 3-12 NMS ordination of sample units in fish species space using count data..................57 3-13 NMS ordination of fish species in sample unit space using count data...................58 3-14 NMS ordination of sample units in herp etofaunal species space using count data..59 3-15 NMS ordination of herpetofaunal specie s in sample unit space using count data...60 3-16 Average and range of lake stage values by cluster...................................................61 3-17 Average and range of air te mperature values by cluster..........................................61 3-18 Site occupancy estimates for focal fish species in spring 2002...............................62

PAGE 9

ix 3-19 Site occupancy estimates for focal fish species in fall 2002....................................62 3-20 Site occupancy estimates for focal herpetofaunal species in spring 2002...............63 3-21 Site occupancy estimates for focal herpetofaunal species in fall 2002....................63 4-1 Locations of 2003 sampling transects in Lake Tohopekaliga..................................80 4-2 Diagram of weekly trap pl acement at specified depths............................................81 4-3 Mean maximum distances traveled with variances and maximum distances, based on results of mark-recapture sampling...........................................................81 4-4Number of trap sites sampled in each vegetation community per sample occasion.82 4-5 Salamander capture success by vegetation community............................................83 4-6 Salamander capture success by water depth............................................................83 4-7 Frog capture success by vegetation community.......................................................84 4-8 Frog capture success by water depth........................................................................84 4-9 Snake capture success by vegetation community....................................................85 4-10 Snake capture success by water depth......................................................................85 4-11 Turtle capture success by vegetation community....................................................86 4-12 Turtle capture success by water depth......................................................................86 4-13 Fish capture success by vegetation community.......................................................87 4-14 Fish capture success by water depth........................................................................87

PAGE 10

x Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science AQUATIC VERTEBRATE USAGE OF LITT ORAL HABITAT PRIOR TO EXTREME HABITAT MODIFICATION IN L AKE TOHOPEKALIGA, FLORIDA By Ann Marie Muench December 2004 Chair: Wiley Kitchens Major Department: Wildlife Ecology and Conservation Lake Tohopekaliga is a large, shallow lake in central Flor ida that is part of the Kissimmee chain of lakes. Cultural eutrophi cation and lake stabilization over the past several decades have facilitated the forma tion of a densely vege tated, often monotypic, littoral zone. Lake managers conducted an enhancement project in 2004 to improve largemouth bass ( Micropterus salmoides ) habitat. This project included an extreme water level drawdown and concurrent mechanical removal of 7.3 million cubic meters of organic sediment and vegetation from the s horeline. Following the drawdown, herbicidal treatments will keep the lake vegetation in an early state of succession in order to prolong the effects of the enhancement. Little is known about potential impacts of these procedures on wildlife, incl uding vegetation, avian, herpetofaunal, and even fish communities. This study examines the status of the reptile, amphibian and fish communities in the two years prior to the lake enhancement to provide baseline data for future assessments.

PAGE 11

xi Funnel traps were used for all sampling, allowi ng a suite of vertebrate species to be examined. In 2002, sampling was conducted in the Pontederia cordata (pickerelweed) zone of the lake. Cluster analysis, indicator species analysis, and nonmetric multidimensional scaling ordinations were us ed to examine temporal changes in the species composition of the assemblages. Envi ronmental variables such as lake stage and average air temperatures played large role s in structuring the aquatic vertebrate communities through species richness and a ssemblage composition. Fish assemblages were most correlated with air temperature, while herpetofaunal assemblages mainly showed association with lake stage. Site occupancy estimates showed that many of the herpetofaunal species are pres ent throughout the pickerelweed habitat in both the spring and fall seasons, while fish showed more fluctuation in seasonal presence. Spatial sampling took place in 2003. Sampling was conducted across vegetation communities and water depths. Both variables captured varying trends in the presence of the focal species, which included fully aquatic salamanders ( Siren spp., Amphiuma means ), water snakes (mainly Nerodia spp.), small kinosternid turtles ( Kinosternon baurii, Sternotherus odoratus ), large aquatic frogs ( Rana spp.), juvenile centrarchids (especially Micropterus salmoides ) and exotic catfish ( Hoplosternum littorale ). Attempted population density estimates for the more abundant herpetofaunal species ended in failure. Inappropriate trapping grid size and spacing for several species at one time led to poor capture proba bilities and large variances in population size estimates.

PAGE 12

1 CHAPTER 1 MAIN INTRODUCTION Lake Ecosystem The productive littoral environment in a lake system is dynamic, since the aquatic habitat has strong terrestrial influences and the terrestrial habita t has strong aquatic influences. Biological diversity is high in the ecotone due to biotic and abiotic properties that distinguish it from adjacent ecosystems, such as vegetation species, soil properties, and water chemistry (Lachavanne 1997). Water level fluctuations are the main determinants of the width of the littoral zone In lentic systems with gently sloping shorelines a wide band of macrophytes provide s patches of heterogeneous habitat for a diverse assemblage of faunal sp ecies. Animal species that have specific requirements for different life stages depend on the proximity of supralittoral (never flooded), eulittoral (occasionally flooded), and infralittoral (alway s flooded) habitats. Since the ratio of water surface to volume is much higher in the littoral zone than in the deeper pelagic region, environmental variables such as lig ht, air temperature, wind and water flow (waves) have much more critical roles in shaping the grad ient (Pieczynska 1990, Pieczynska and Zalewski 1997). Excess inputs of nutrients beyond that natu rally found in a particular lake system leads to eutrophication. This condition encour ages surplus sediment ation and vegetation growth, changing the landscape of the original littoral zone and subsequently altering the biological communities within that habitat and the lake as a whole. Eventual extinction of the lake may result from decades of nut rient pollution due to sewage discharge and

PAGE 13

2 drainage from agricultural and urban lands. Internal recyclin g of nutrients within a water body keeps it from recovering even when the inputs are reduced (Cooke et al. 1993). In order to preserve the lake for the longest ti me possible, rehabilita tion efforts are often made to counter the results of eutrophication, for example drawdowns, dredging and mechanical vegetation removal (Hasler 1947, Cooke et al. 1993). Study Area Lake Tohopekaliga (7,612 ha, 18,810 acres) is located in Osceola County, Florida, within the Upper Kissimmee Basin. This phys iographic area is known as the Osceola Plain, which is flat and has very few disti nguishing topographical ch aracteristics. The elevation in the plain ranges from 18-30 m (60-95 ft) National Vertical Geodetic Datum (NGVD), but rarely reaches maximum hei ghts (Harper 1921). Originating from prehistoric ocean bottom, the sediment mainly consists of coastal sands. Numerous shallow lakes in the region, in cluding Lake Tohopekaliga, were formed by dissolution of the carbonate-containing substrates (limes tone) in depressed areas (Schiffer 1998). Freshwater wetlands in this area include cypress sloughs, wet prairies, river swamps, floodplains, mixed forested wetlands, and marshe s. Pine flatwoods dominate the upland community (HDR Engineering, Inc. 1989). The Lake Tohopekaliga Subbasin (211.6 square km, 131.2 square miles), within the Upper Kissimmee Basin, receives wate r input from the Sh ingle Creek (184.2 square km, 118.0 square miles) and East Lake Tohopeka liga (81.7 square km, 48.4 square miles) Subbasins. Precipitation, overland flows, and to some extent groundwater from the underlying Surficial Aquifer also provides the lake with impor tant water sources. While evapotranspiration is a strong f actor in withdrawal of wate r from the lake, outflow from Lake Tohopekaliga occurs at its southernmost point, where the Sout h Port Canal connects

PAGE 14

3 it to Cypress Lake. Water from the Uppe r Kissimmee Basin flows southward through the Kissimmee Chain of Lakes, through the channelized Kissimmee River, to Lake Okeechobee, east and west coast estuaries a nd South Florida (HDR Engineering, Inc. 1989, Schiffer 1998). Human disturbance of this hydrologic syst em began in the mid-nineteenth century, with local efforts to drain wetlands. In 1882 Hamilton Disston began channelizing the watersheds in the upper basin by constructing in ter-lake canals. The ma jor results of this endeavor were lowered lake levels, drying of lake edges and interlake slough wetlands, as well as rapid transit of nutrient-laden surface waters downstream. Wetlands stretching between Lake Tohopekaliga and East Lake Toh opekaliga were strongly impacted. In the 1920Â’s, Shingle Creek (a major source of wa ter for Lake Tohopeka liga) was channelized, bypassing water around the swamps and marshes within that subbasin. Many landowners also dug ditches and canals to drain wetlands and improve their pastureland. In 1947, the Central and Southern Florida Flood Control Project was implemented by the U.S. Army Corps of Engineers in response to major floodi ng in the Kissimmee Ba sin. As a result of this plan, the Kissimmee River was cha nnelized, DisstonÂ’s inte r-lake canals were improved, and water control structures were built throughout the area. The goal of these actions was to use the chain of lakes for wa ter management, to provide room for water during the wet season and to stor e water during the dry season. This entailed stabilization of water levels, which historically fluctuated up to 3 m (10 ft), to a 0.6-1.2 m (2-4 ft) range. This reduction in fluctuation subs equently allowed landowners and private citizens to build on historic lake bottom and drained we tlands within the floodplain,

PAGE 15

4 further strengthening the need for tight fl ood control (Blake 1980, HDR Engineering, Inc. 1989). Lake Tohopekaliga and surrounding lakes have suffered many water quality problems due to the intense hydrologic modifica tions. The constructed canals, especially in the Shingle Creek area, allowed direct conveyance of stormwater runoff and sewage into the lakes without the be nefit of filtration through wetlands. Urban and agricultural land use continued to expand, contributing more and more overland pollution. The agricultural land in the area is mainly utilized as pasturela nd, and historically dairy farms provided significant inputs of nutrients. Although eutrophication within the lakes was rapidly increasing, water level stabilization pr evented natural fluctuations from mitigating the problem. Since the lake levels were unnaturally restricted fr om periodic flooding and drying events, thick stands of vegetation bega n invading the littoral zone, which in turn led to organic sediment buildup and decrease d water quality. As of 1988, no wastewater discharges have been permitted to the lakes. However, non-point source urban and agricultural runoff and septic tank leakage remain major contributors to eutrophication in the Kissimmee chain of lakes (HDR Engineering, Inc. 1989). As mentioned, eutrophication has caused major vegetation changes to the littoral zone of Lake Tohopekaliga. Dense monotypic expanses of aquatic vegetation began to dominate the gradually sloping shoreline, fo rmerly characterized by sandy substrate and sparse vegetation. Nuisance species such as Pontederia cordata (pickerelweed) and Typha domingensis (cattail) formed wide bands of habitat around the lake. Pontederia cordata and associated species crea ted floating mats on the lakewa rd edges of the littoral zone that rose and fell with the water level. Exotic species such as Hydrilla verticillata

PAGE 16

5 (hydrilla), Eichornia crassipes (water hyacinth), Alternanthera philoxeroides (alligator weed), and Panicum repens (torpedo grass) also benef ited from increased nutrients and high boat traffic between waterways and became a major focus of lake managers through the years. Turnover in the vegetation comm unity produced an organic muck substrate within the littoral zone. Documented faunal responses to this changing habitat have included declines in fisheries, especially s port and forage fish species, and invertebrates (HDR Engineering, Inc. 1989). In 1968, the first fish population surveys in Lake Tohopekaliga were conducted by the State of Florida Fish and Wildlife C onservation Commission (FFWCC, formerly Florida Game and Fresh Water Fish Commission, FGFWFC) with rotenone, electroshocking, and trammel nets (Wegen er 1969). Management recommendations included drawdowns every 5-7 years in order to oxidize the increasingly organic substrate and provide benefits to the growing fish population upon reflooding of littoral habitats. In 1971, Lake Tohopekaliga underwent its firs t extreme drawdown. The water stage was dropped about 2.1 m (7 ft) from high pool stage (16.8 m, 55.0 ft NGVD). Drought conditions kept water levels below 15.8 m (52.0 ft) NGVD (low pool stage) until one year after the initiation of th e drawdown. The dewatering of th e littoral zone attained the management goals, with increased acreage of desirable plant species, greater production of fish and fish-food organisms (invertebra tes) per acre, and in creased sportfishing success (Wegener and Williams 1974). Improvements made to Lake TohopekaligaÂ’s li ttoral zone were short lived, due to continued input of nutrients (e.g., about 53 million liters (14 million gallons) of sewage waste per day discharged into the lake (Wegener and Williams 1974)) and water level

PAGE 17

6 stabilization. An offshore berm was forming at the low-pool water li ne that was thought to be acting as a barrier to fish and inverteb rates at low water levels (Moyer et al. 1987). A second drawdown was conducted in 1979 and al so had beneficial results, although by 1986 the habitat had degraded once more to s ub-optimal fishery habitat. The organic berm in the lakeward portion of the littoral zone was increasingly becoming a management issue not able to be addre ssed by drawdowns alone (Moyer et al. 1987). When the third drawdown was performed in 1987 a pilot muck removal project was included. Along 19 km (12 miles) of shoreline, 164,830 cubic meters (225,000 cubic yards) of muck were mechanically removed fr om the organic berm. This was considered “an unprecedented large-scale restoration project to improve littoral habitat” (Moyer et al. 1993, Appendix 4, page 2). All research point ed to highly positive results from the drawdown and muck-scraping procedures. Fi shery surveys have continued since 1968, and now include roving creel surveys, blocknet/rotenone sampling, electrofishing, experimental gill nets, and shallow water sa mpling with Wegener rings. Other wildlife monitoring included snail kite ( Rostrhamus sociabilis) individual and nest counts, limited aquatic plant sampling, and littoral zone i nvertebrate community monitoring (Moyer et al. 1993). Another lake enhancement project had been planned for early 2002 in Lake Tohopekaliga, however logistical constr aints caused postponement until early 2004 (Florida Fish and Wildlife Conservation Commission 2003). The project originally included an extreme drawdown from 16.8 m (55.0 ft) NGVD to 14.9 m (49 ft) NGVD beginning in November 2003, as well as the mechanical removal of about 5.4 million cubic meters (6.8 million cubic yards) of muck and vegetation from the majority of the

PAGE 18

7 lakeÂ’s shoreline (Florida Fish and Wi ldlife Conservation Commission 2003). The subsequent estimate of the actual volume sc raped was 7.3 million cubic meters (8-million cubic yards), with 1,351 ha (3,339 acres) of s horeline habitat remove d. The entire width of the littoral zone was targeted for re moval, not just the organic berm. The Pontederia cordatadominated habitat underwent widespread elimination throughout the lake. Twenty-nine in-lake disposal islands were cr eated from much of the scraped material (Florida Fish and Wildlife Conservati on Commission 2004). On ce the water levels recover, heavy applications of herbicides will be used to keep the habitat in an early state of succession, allowing lake managers to selectively allow regr owth of desirable vegetation. Currently, the target conditions post-enhancement are undefined. The main objective of the Lake Tohopeka liga habitat enhancement project is the removal large expanses of undesira ble macrophyte stands, particularly Pontederia cordata and Typha domingensis as well as the organic substrate (muck) associated with this dense vegetation (Florida Fish a nd Wildlife Conservati on Commission 2003). Reptiles, amphibians and many juvenile fish species are known to occupy structurally complex lentic habitats and u tilize the muck and thick vege tation for foraging, cover, and also reproduction (e.g., amphibi ans). Lake enhancement techniques (both mechanical vegetation and muck removal and subsequent herbicide applications) modify these resources, changing the habitat suitability fo r aquatic vertebrates. High mortality during the scraping process and migration during the drawdown will likely also alter the community structure and dynamics. The effort to sustain high species diversity in the lake ecosystem may be important to the st ability of the system, and by examining the

PAGE 19

8 consequences of these restoration techniques managers can better eval uate their worth to wildlife and fishery species. Research Objectives While some positive responses have been documented for the fishery of Lake Tohopekaliga following past enhancement pr ojects, many wildlife guilds have been neglected. There is limited qua ntitative knowledge of vegetati on responses to mechanical removal and large-scale herbicidal treatment. It is also uncertain how wetland birds are affected. There are still unanswered questions regarding aquatic vert ebrates that utilize the thick vegetation and organic sediment, including reptiles, amphibians and fish. Herpetofaunal responses to enhancement activitie s have not been studied in the past, even though they are pervasive in the habitat. Although fishery science claims that the eutrophic littoral habitat is unsuitable for centrarchids (i.e. sport fish), conventional sampling methods may be incapable of detec ting them in highly vegetated areas (Parker 1970, Allen et al. 2003). The current study is part of a larger pr oject evaluating the wildlife response to habitat enhancement in Lake Tohopekaliga. Al so included in this project are vegetation (see Welch 2004) and avian monitoring studies. The research presented in this thesis examines the aquatic vertebrate community in the littoral zone of Lake Tohopekaliga prior to the 2003 drawdown and mechanical ve getation and muck scraping activities. The littoral zone is defined here as the ar ea occupied by emergent vegetation. However, there is particular emphasis gi ven to the pickerelweed zone due to its extensive removal during the lake enhancement. The large-scale wildlife habitat investigation will continue for at least three years after enhancement activities to examine responses to the modifications by the different guilds.

PAGE 20

9 With a large-scale habitat modification, qua ntifying the effect on a whole suite of species provides maximum information. While most of the species have common biological or ecological traits, th ey also constitute a variety of habitat requirements based on food sources, reproduction methods, and movement patterns. For this reason community metrics within the land-water ecoto ne are of main concern, as represented by species richness and commun ity composition. Species-specific site occupancy and capture frequencies also facili tate understanding of habitat utilization by focal vertebrate species. The main objectives of this research are to 1. Characterize the vertebrate faunal makeup of Lake TohopekaligaÂ’s littoral zone prior to the 2004 lake enhancement project, 2. Estimate parameters such as density and activity/home range for focal species in the P. cordata habitat, 3. Estimate site occupancy rates for focal species within the littoral zone, as an estimate of the proportion of the area that the species inhabits, 4. Document how temporally changing variab les including lake stage, water level fluctuation, air temperature, and rainfall sh ape the aquatic vertebrate community in the P. cordata zone, and 5. Investigate the influence of spatial vari ables such as water depth and vegetation community on the herpetofauna and fish within the landscape.

PAGE 21

10 CHAPTER 2 DESCRIPTIONS OF FOCAL SPECIES Aquatic Vertebrate Habitat Wetland communities of reptiles and amphibians show much diversity in ecological function. Often being the largest and most abundant vertebrates in this habitat, they have important places in the food we bs of lakes (Iverson 1982). Some species provide terrestrial links while others are fully aquatic and never leave the littoral zone (Joly and Morand 1997). Fish species also rely on both the littoral, pelagic, and to some extent flooded terrestrial (nurse ry) habitats as they undergo sh ifts with life stage (Werner 2002). There has been a worl dwide decline in biodivers ity, particularly seen in amphibian species. A variety of human disturbances have been identified, including climate change, habitat loss and fragmenta tion, introduced species, pollution, acid rain, and disease (Reaser 2000). Florida in particular has been severely impacted by destruction of wetlands, channelization of st reams, manipulated hydrologic cycles, and rapid human growth (HDR Engineering, In c. 1989, Pough et al. 2001). Alteration of freshwater habitats has been a problem for many aquatic species. Animals that are longlived or have delayed sexual maturity, low reproductive rates, or poor dispersal or colonization abilities are particularly vulnera ble to habitat destruction (Klemens 2000). Purposeful habitat modification should preser ve conditions necessary for aquatic animals to complete their life cycles, including a ppropriate nesting/spawning, foraging, and cover habitats.

PAGE 22

11 Species of aquatic vertebrates that ar e most at risk due to their habitat requirements are emphasized here. Most he rpetological research has been conducted on breeding populations of amphibians, large ch arismatic reptiles, or single species and guilds (but see Bancroft et al. 1983). Fishery science remains focused mainly on sportfish at the individual or population le vel (Miranda and Dibble 2002). Resident littoral zone species make up the assemblage of interest for this study and represent several different orders of animals with a variet y of life history traits. Fish guilds, such as juvenile centrarchids (especially Lepomis spp., Micropterus salmoides ) and exotic catfish ( Hoplosternum littorale ), are focused upon. Documenta tion of the presence of these species in heavily vegetated littoral habitats in Florida is very poor, probably due to inadequate sampling techniques. Reptile and amphibian species of in terest include fully aquatic salamanders ( Siren spp., Amphiuma means ), water snakes (mainly Nerodia spp.), small kinosternid turtles ( Kinosternon baurii, Sternotherus odoratus ), and large aquatic frogs ( Rana spp.). Minimal research has been conducted on the effects of lake management techniques on these herpetofaunal species. Most of th ese species and guilds have a common reliance upon vegetated wetlands for at least some part of their life cycles. They also are often preyed upon by the same species, including alligators, wading and predatory birds, large predator y fish and aquatic sn akes, and together represent many segments of the food web in the lake ecosystem. Fish Species Centrarchids (sunfish) Most species in the family Centrarchi dae in Lake Tohopekaliga are sportfish. Foremost among them in Florida lakes is the largemouth bass ( Micropterus salmoides ). This species is the primary target for be nefit by the Lake Tohopekaliga enhancement.

PAGE 23

12 Bluegills (Lepomis macrochirus) redear sunfish (Lepomis microlophus) black crappie ( Pomoxis nigromaculatus ), warmouths ( Lepomis gulosus) spotted sunfish ( Lepomis punctatus), and dollar sunfish ( Lepomis marginatus ) are also considered sportfish in Florida. Enneacanthus gloriosus (bluespotted sunfish), has a maximum total length of 80 mm, and is therefore only c onsidered a forage fish speci es (Hoyer and Canfield 1994). Most of these species depend upon the vegetate d littoral zone during juvenile stages and for spawning. Vegetated habitats provide juve nile sunfish with protection from larger predators and abundant food supplies (W erner and Hall 1988, Chapman et al. 1996, Miranda et al. 2000). The phenom enon of ontogenetic habitat shif ts is particularly well studied in bluegills. This species move s between the littoral to the pelagic zone throughout its life cycle. The lit toral zone provides nesting hab itat, as well as a preferred environment for juvenile bluegills from a pproximately 12-83 mm standard length due to size-specific predation ri sks (Werner and Hall 1988). It is claimed that these species have no access to Lake TohopekaligaÂ’s littoral zone due to physical and chemical barriers. The floating mats of Pontederia cordata, resulting from the eutrop hic status of the lake, are thought to form a physical barrier for centrarchids, limiting adult access to shallow water spawning sites. Even if the fish could penetrate this barrier, physicochemical char acteristics of the dense vegetation would not permit survival (Moyer et al. 1995, Allen and Tugend 2002, and Allen et al. 2003). Traditional methods of fish sampling in high-macrophyte littoral habitats in Lakes Tohopekaliga and Kissimmee, Florida, have yielded few or no centrarchid species (Moyer et al. 1993, Allen and Tugend 2002). However, since common fish sampling methods, including electrofishing and rotenone/b locknet, do not perform well in heavily

PAGE 24

13 vegetated habitats (Parker 1970, Moyer et al. 1995, Allen and Tugend 2002), many suppositions upon which lake enhancemen t projects depend are theoretical. Exotic catfish Hoplosternum littorale is an exotic species in Fl orida, originating in South America. This armored catfish was first found within the United Stat es in South Florida in 1995, and was presumably released thr ough the aquarium trade or aquaculture. Various life history and beha vioral traits, including aeria l respiration, large body size, high environmental tolerances, and nest-guard ing behaviors, are re sponsible for rapid expansion of its range in Florid a (Nico et al. 1996). This sp ecies is currently nesting in and pervasive throughout the littoral zone in Lake Tohopekaliga (personal observation). Pterygoplichthys spp. (suckermouth or sailfin catfish) has also been captured in Lake Tohopekaliga, although only on a few occasions. This species was proba bly released into Florida through the aquarium trade (Page 1994) These two species may pose significant ecological threats to native food webs and aquatic plant communities. While the Florida Fish and Wildlife Conservation Comm ission conducts yearly monitoring by electrofishing, this study is th e first known report of these sp ecies this far north in the Kissimmee chain of lakes. Herpetofaunal Species Amphibians Rana grylio (pig frog) is a highly aquatic speci es, rarely being seen on shore. They are usually associated with dense ma rsh vegetation. While leopard frogs, including Rana sphenocephala (Florida leopard frog), prefer habi tats with standing water, larger individuals can inhabit somewh at dryer on-shore habitats an d use larger home ranges, relying on plant shade, dew and soil moisture for survival (Dole 1965). Adult leopard

PAGE 25

14 frogs and their tadpoles are also noticeably absent from sandy, unvegetated shorelines (Dole 1965, Alford and Crump 1982, Bancroft et al. 1983). These two large frog species have differences in length of larval developm ent, with pig frogs taking more than a year to metamorphose and leopard frogs taking only two to three months (Bancroft et al. 1983). This, along with year-round br eeding in Florida, results in a variety of size classes of tadpoles throughout the year. Siren lacertina (Greater siren) and Amphiuma means (Two-toed amphiuma) are two of the largest sp ecies of salamanders in the wo rld (Petranka 1998). Amphiumas depend on lungs for aerial respira tion, while sirens have extern al gills as well. Although they may have lengths greater than 76 cm (Conant and Collins 1998), diminutive limbs in both species are thought to limit overland disp ersal. These salamanders burrow into organic sediment when their habitats become dry and may remain alive for up to three to five years in underground burrows without food until water comes back to the habitat (Martof 1969, Etheridge 1990). Bancroft et al. (1983) found that the density of amphiumas and greater sirens in Lake Conwa y, Florida, increased with sediment depth. They also reported that neither species i nhabited sandy, unvegetated shorelines. Sirens have compressed tails that may help to prope l them in vegetated open water as well as emergent vegetation habitats. Amphiumas on the other hand have l ong cylindrical tails and are thought to be limited to shallow water (Bancroft et al. 1983). Sirens feed mainly on mollusks, insects, crayfish and filamentous algae, as well as some other vegetation. Amphiumas eat fish, crayfish, salamanders, fr ogs and a wide variety of other species (Petranka 1998).

PAGE 26

15 Reptiles The striped mud turtle ( Kinosternon baurii ) and common musk turtle ( Sternotherus odoratus ) are both small species (maximum carapace lengths of 12.2 cm and 13.7 cm respectively) that prefer shallo w water wetlands (Conant and Collins 1998). They are both omnivorous, feeding upon animal s and some plants opportunistically. However, mud turtles are attracted to fast -moving prey while musk turtles search out more sedentary organisms as they crawl along the substrate in search of prey (Mahmoud 1968). Striped mud turtles usually occur in water greater than 60 cm deep, with lower water levels or rainfall tri ggering terrestrial activity (W ygoda 1979, Ernst et al. 1994). On the other hand, common musk turtles are highly aquatic, not leaving water unless nesting. This species seems to prefer water de pths less than 60 cm, but have been seen in up to 9 m of water (Ernst et al. 1994). Ba ncroft et al. (1983) found about 20% of all captured common musk turtles in the littora l zone, and the rest (usually larger individuals) in open water habita t. According to Mahmoud (1969), S. odoratus is found in lakes as well as riverine habitats with gravel or sandy substrates. Nerodia fasciata pictiventris (Florida water snake) is most often encountered in the shallowest regions of inhabited wetland s (Ernst and Ernst 2003). They are observed often in disturbed and white sand littoral habita ts (Bancroft et al. 1983) This species eats mainly fish until they reach a total length of 50 cm, at which point they switch to preying upon frogs (Mushinsky et al. 1982). Nerodia floridana (Florida green water snake) is the largest North American water snake, with to tal lengths approaching two meters (Conant and Collins 1998). They are inhabitants of qui et water wetlands and sometimes venture out into open water (Ernst and Ernst 2003). Bancroft et al. (1983) found them to be pervasive throughout the littora l zone, and while the dense vegetation seems to be

PAGE 27

16 preferred, the species of vege tation may not be very importa nt. Some individuals were captured up to 40 m from the edge of the lit toral zone in open water, while several terrestrial sightings occurre d during winter months. Se diment depths of 11-20 cm yielded the most individuals, and sandy beach habitats were avoided by Florida green water snakes (Bancroft et al 1983). They feed mostly upon fish, but also on frogs, salamanders, tadpoles, small turtles and invertebrates (Mushinsky and Hebrard 1977, Ernst and Ernst 2003)

PAGE 28

17 CHAPTER 3 ASSEMBLAGE WITHIN THE Pontederia cordata COMMUNITY Introduction The objective of this section is to inves tigate the temporal va riation of community composition and dynamics. The four main rese arch questions are 1) is there temporal variation in the aquatic vertebrate assemb lage, 2) does community composition change over time, 3) what environmental factors seem to be influencing the temporal variation in the assemblage and individual focal species, and 4) how are the focal species dispersed through the habitat. Key envi ronmental variables that change over the course of a year include lake stage, water leve l fluctuation, air temperature, and rainfall. Each of these will be examined for their influence on the vertebrate assemblage. The thick P. cordata (pickerelweed) habitat was the prime target for mechanical removal during the lake enhancement process and therefore was the focu s of sampling effort. This protocol was also used to select focal species (which sp ecies were present in the habitat and most detectable with the traps) and evaluate trap -sampling methods. All of this information will facilitate monitoring in the future, regard ing how, when and where to sample in order to capture the community dynamics and vari ances associated with the lakeÂ’s everchanging environment. Field Methods Trap Descriptions As previously mentioned, fishery surv eys conducted in the Kissimmee Chain of Lakes include roving creel surveys, bl ocknet/rotenone sampling, electrofishing,

PAGE 29

18 experimental gill nets, and Wegener rings (Moyer et al. 1993). These methods collect information on a variety of fish species, but spor t fish are the typical target of research. Traditional herpetofaunal sampling techniques include visual surveys and hand or dip-net collecting (Bury and Corn 1991), pitfall and funne l traps in combination with drift fences (Corn 1994), and use of seines or dredges fo r removing floating vege tation along with the animals inhabiting it (Bancroft et al. 1983). None of these methods are appropriate for the extremely thick, rooted vegeta tion in the littoral zone. Turt le traps exist, such as hoop nets and floating traps for basking turtles (Lagler 1943); however la rge turtles are not central to this research since they are not re stricted to the littoral zone. PVC pipes have also been used as passive tr aps for treefrogs (Moulton et al 1996), and audio surveys are often used for breeding ranid frogs (Zi mmerman 1994). However, a single, allencompassing technique was desired for th is community study, and the answer came from funnel traps. Recently, several research ers have noticed the be nefits of capturing aquatic organisms in thick vegetation with crayfish and minnow style funnel traps (Darby et al. 2001, Sorensen 2003, Johnson and Barich ivich 2004). Without the use of either bulky drift fences or bait, these traps have b een successful in capturing a wide variety of reptiles, amphibians, fish and some invert ebrates. Funnel traps were used for all sampling during this study. The minnow and crayfish traps were all constructed of 1.3 cm (0.5 in) mesh, dark green vinyl-coated hardware cloth (Figure 31). The crayfish traps, similar to those described by Darby et al. ( 2001), were positioned on the subs trate, or as near to the substrate as the vegetation would allow. Th ey were approximately 80 cm (30 in) tall including a “chimney” extending from the body of the trap, allowing the top to be above

PAGE 30

19 the water surface. At the base were three en try funnels leading into the trap, with each opening about 6 cm (2.5 in) in diameter, but the exact size varied slightly due to handmade construction. The modified minnow tr aps were 60 cm (24 in) long rectangular traps, which were approximately 25 cm (10 in) deep and 18 cm (7 in) high. At each end there was one entry funnel, with an opening approximately 9 cm (3.5 in) wide and 6 cm (2.5 in) tall. Floats made of Styrofoam pool toys (“Wacky Noodles”) were attached to the minnow traps to allow them to float halfwa y out of the water, wi th the funnels about even with the water surface, based on the design by Casazza et al. (2000). The funnels permitted animal access into the trap, but discouraged escape by making the exits harder to find than the entrances. By allowing the traps to remain partially above the water, the animals had access to air and mortality was re duced. Both nocturnal and diurnal species were accessible to capture since traps could be deployed without time constraint. The traps were not baited, however once an anim al was captured in the trap other animals may have been attracted to it. The dimensions of the traps restricted the assemblage of animal species captured. The traps did not confine young individuals or small species of fish, frogs, snakes and salamanders due to the 1.3 cm (0.5 in) mesh size. Also, individuals larger than the funnel diameter were excluded. To compare the di fference in species captured with 1.3 cm (0.5 in) versus 0.6 cm (0.25 in) mesh, 18 commerc ially-manufactured minnow traps, similar to the “eelpots” used in Casazza et al. (2000) were deployed at randomly assigned trap points from 11/5/2003 to 1/8/2004. These traps are cylindrical, about 60 cm (24 in) long and 23 cm (9 in) in diameter, with the funnel openings about 5 cm (2 in) in diameter. They were also fitted with floatation. Th e hardware cloth was bare metal, not vinyl-

PAGE 31

20 coated. Comparisons of species and number of captures were made between the 0.6 cm mesh minnow traps and the 1.3 cm mesh modi fied minnow traps from the same trap points during this sampling period. We exp ected to capture more species with the smaller mesh size since small species and younger individuals could escape from the larger mesh, but be retained by the 0.6 cm holes. Whole-Lake Sampling To gain information regarding temporal ha bitat utilization by the aquatic vertebrate assemblage, sampling was conducted around the pe riphery of the whole lake to maximize the inference of the results to the system. For the whol e-lake sampling, 18 sites were randomly selected from the less developed, southe rn two-thirds of the lake (Figure 3-2). At each site, a transect was established with three trap locations placed perpendicular to shore and spaced approximately 10 m (33 ft) apart, except in disturbed stretches of habitat with barriers wi thin this distance (e.g., commercial airboat trails). One crayfish and one minnow trap were placed at each tr ap location, attached to a PVC pole for extra stability. The result of this trapping arra ngement was uniform sampling effort at each transect. The trap locations were placed in the most lakeward portion of dense P. cordata when possible (mainly in the 0.6-0.9 m (2-3 ft) depth zone at 13.8 m (55 ft) NGVD). The transect sites varied in proximity to the ecotone between the open water habitat and the vegetation. Most transect s had thick stands of Typha or more diverse floating mats between the relatively monotypic sections of pickerelweed an d open water. The band of emergent macrophytes at these locations was comparatively br oad. On the other hand, at some transects the traps were relatively clos e to this ecotone due to narrowness of the pickerelweed zone at these locations, well established commercial airboat trails, or herbicide applications near the transects pr oviding large unvegetated areas. One transect

PAGE 32

21 fell in an area where the substrate had pr eviously been scraped, and the vegetation consisted mainly of Hydrilla verticillata (hydrilla) and very few emergent macrophytes. The whole-lake trap survey was c onducted year-round, pending suitable water levels (greater than approximately 16 m, 52.5 ft NGVD). Below this point, there was not enough water for the trapped animals and rode nts and birds were in advertently captured. Sampling throughout the year 2002 was as follows: January 24 – Traps were deployed to ra ndomly selected transects and sampling began. May 2 – Insufficient water levels in the pickerelweed zone caused traps to be removed and sampling suspended. June 12 – Redeployment of traps to select transects with su fficient water depth resulted in decreased trapping effort until July 24. July 24 – All traps were back in place in fixed sampling locations. December 3 – Traps were removed from pickerelweed zone due to low water associated with the attempted 2002 drawdown. When active, the traps stayed in place day and night and were typically checked once weekly. Despite efforts to keep sample s spaced seven days apart, the time interval was not always consistent due to logistical issues (e.g., air boat problems, rough weather). At each sampling occasion, two or three observers traveled to each transect in an airboat and checked the traps for their contents. All animals were brought b ack to the boat to be worked up. Reptiles and amphibians were weighed individually with Pesola spring scales and certain length measurements were taken, depending on the species. Fish were identified and grouped according to species for each trap. All individuals of each species per trap were weighed together in order to obtain the total biomass of the fish species caught. After being worked up, the animals we re released at the transects where they were captured. The types of data collected with these methods include species detection-

PAGE 33

22 nondetection, number of individuals capture d on each sample occasion, biomass, and reptile and amphibian length measurements. This sampling protocol was intended to c ontinue in the exact same locations postlake enhancement (2003) in order to compar e community traits be fore and after the modifications. However when the drawdown and muck removal was postponed for another year, it was no longer beneficial to keep sampling since there was not a before/after comparison to be made. Variab les such as water temperature and dissolved oxygen were not measured directly since this was not the initial focus of the study. Alternative environmental variables were obtai ned using Internet resources. Lake stages and rainfall were taken from the South Fl orida Water Management DistrictÂ’s DBHYDRO browser ( http://glades.sfwmd.gov/pls/dbhydro_pro_plsql/ ). The lake stage was the mean daily average taken from the headwater of Sta tion S61 (the water control structure in the south part of the Lake Tohopekaliga leading to Lake Cypress via the South Port Canal) in feet NGVD. Lake stage was recorded for the day of each sample occasion. Water fluctuation for one sample is the difference of the water level at that sample minus the water level at the previous sample occasion. Rainfall was also recorded at Station S61 and precipitation totals for each sample we re added up from the day of the previous sample occasion until the day before the new sample occasion. Air temperature data was gathered from the National Oceanic and Atmo spheric AdministrationÂ’s National Climatic Data CenterÂ’s website ( http://www.ncdc.noaa .gov/servlets/ULCD ). The weather station location was the Orlando Inte rnational Airport (M CO) in Orlando, Florida. This is located approximately six kilometers (10 miles) from the north shore of Lake Tohopekaliga. Average temperatures were calculated for every sample occasion by

PAGE 34

23 averaging the daily average temperatures from the day of the previous sample occasion until the day before the new sample occas ion. Minimum and maximum temperatures were also recorded for each sample period. Analysis Methods Trap Comparisons To determine the utility of traps with smaller mesh si ze for this study, the species and number of captures for the 0.6 cm (0.25 in ) mesh commercial minnow traps and the 1.3 cm (0.5 in) mesh modified minnow trap s were compared. The 0.6 cm mesh traps were randomly placed at only a portion of the whole-lake trap sites, along with a 1.3 cm mesh crayfish and minnow trap. For this reason, data from both minnow traps were compared for just the trap sites with both mesh types. Species Richness Sampling was carried out with a repeated measures protocol, potentially resulting in lack of independence be tween samples. However, assuming random movement of individuals and species through the habitat over space and time, sampling over time did not result in repeated captures of the same individuals. Th is transient nature of the species and utilization of non-parametric pr ocedures for most analyses are believed remove potential bias due to re peated samples. To determin e the presence of temporal variation in the aquatic vertebrate assembla ge, total species richness was calculated for each sample occasion in 2002. Fish and he rpetofaunal species richness were also individually estimated for each sample o ccasion. Program COMDYN4 was used with detection-nondetection data to estimate richness (Hines et al. 1999), taking into account species detection probabilitie s. It uses a model (Mh) that allows each species to have a different detection probability (the probability of detecting at leas t one individual of the

PAGE 35

24 species). Since most species detection proba bilities are less than one raw count data can result in underestimations of richness. In a similar manner, the term “presence/absence data” can also be misleading since lack of detection provides no evidence of a species’ absence from the trap site. For this reas on I instead use the term “detection/nondetection data” throughout this thesis. Equal sampli ng effort is necessary for each occasion. Assumptions of this method are 1) populati on closure for species, 2) independence of captures and 3) individual species capture pr obabilities stay cons tant during sampling (Burnham and Overton 1979). However, this me thod is robust to de viations from these assumptions. Even when the assumptions are violated the model-based richness estimates are less biased than counts of species (Nichols et al. 1998). Data from seventeen of the eighteen transe cts were used in the richness analysis. The one transect that was located in the prev iously scraped habitat was removed from the analysis in order to focus solely on variations within the P. cordata habitat. Since species capture data were fairly sparse for each tr ansect per sample occas ion, the transects were randomly assigned to six groups that represent sample replicates across space. They were randomly grouped in order to remove effects of shoreline characteristics at different transects. Richness was not estimated for sa mple occasions with reduced trap effort (sample occasions 14-20). Linear regre ssions were performed using SPSS (SPSS Inc. 2001) in order to determine significant predictors of the vertebrate, fish and herpetofaunal species richness. Richness estimates from each sample occasion were used in these analyses. There were three outliers greater th an two standard deviations from the mean, which were removed for the herpetofaunal re gression analyses. Average air temperature

PAGE 36

25 (oC), lake stage (m), rainfall (cm), and water level fluctuation (cm) were used as the independent variables. Assemblage Composition Species richness estimates the number of species present, but indicates nothing about community composition. To compare the pr esence of vertebrate species over time, sample occasions were assigned to clusters us ing hierarchical cluste r analysis. This was run using PC-Ord software (McCune and Mefford 1999) with detection-nondetection data of species for each sample occasion. Sore nsen's distance measure with the flexible beta (beta=-0.25) linkage method was used. Indicator species analysis (McCune and Grace 2002) was applied to determine the most appropriate number of clusters and the best species to represent those groups. Any groups comprised of a single sample occasion were removed from the indicator sp ecies analysis. A Monte Carlo procedure was run 1000 times with randomized data to cal culate a p-value for each species, which tested the null hypothesis that their indicator values were no larger than would be expected by chance. The optimum number of groups was selected by the indicator species analysis that yielded the most speci es with statistically significant indicator values (McCune and Grace 2002). Influence of Temporal Gr adients on Assemblage Multivariate ordination was used to establish what temporally changing environmental factors were influencing th e variation in assemblage composition. Nonmetric multidimensional scaling (NMS) is an ordination technique that uses ranked distances between sample un its to reduce dimensions a nd allow description of the community in relation to environmental gradie nts. The distances represent dissimilarity between sample units in terms of species composition. This method was chosen because

PAGE 37

26 it is particularly useful for non-normal data and many sampling events with no captures (McCune and Grace 2002). Sample units, i.e. individual sample occasions, are plotted in species space using an iterat ive search for the optimal placement for the sample units. Optimal placement is determined by the maximum possible reduction in stress, which is a measure of dissimilarity between the origin al data matrix and the reduced-dimension final ordination. PC-ORD software was used for all NMS analyses (McCune and Mefford 1999). NMS was run for the entire vertebrate a ssemblage for all sample occasions using detection-nondetection and count data separa tely. Fish and herpetofaunal assemblages were then analyzed separately in the same fashion to determine if the environmental gradients affected them differently. Outliers were identified using the outlier analysis provided in PC-Ord, with the cr iteria being greater than two st andard deviations from the mean (McCune and Mefford 1999). All outliers were removed from the analyses. General relativizations by row were conducte d on the raw count data to equalize common and uncommon species and lower the coefficien t of variation (CV) of the row totals. Relativizations were followed by square root transformati ons to balance the relative importance of the species without altering their ranks. SorensenÂ’s distance measure was used to calculate dissimilarity matrices for the ordinations. Starting configur ations were created by random number seeds, which were generated by the time of day. Fifty runs we re conducted with the real data to find the solution with the lowest stre ss. Fifty Monte Carlo randomi zed runs were performed to select the appropriate number of dimensions that be st represent the variation in the data. Comparisons between the runs with real data and randomized data give a probability that

PAGE 38

27 final stress in the ordination could be found by chance. After the first 50 runs, the number of dimensions was determined and the fi nal NMS was rerun usi ng the random number seed from the initial ordination. From this the final stress and inst ability (fluctuation of stress per iteration) were evaluated. Measured environmental variables, includi ng lake stage, stage fluctuation, total rainfall, and average, maximum and minimum average air temperatures over the sample period, were included in the or dination graphs. The ordinations were plotted with environmental variables as biplots, indicati ng the strength of corre lations of variables with the synthetic axes. Only th e environmental variables with r2>0.2 (percent of variance represented) are shown in the ordina tion plots. Sample un its were color coded by their membership to the groups defined by the cluster analysis, representing different species composition. Proportion of Habitat Ut ilized by Focal Species Using the program PRESENCE (MacKenz ie et al. 2002), detection/nondetection data were analyzed to determine site occ upancy rates for all species with enough captures to get reasonable estimates. This method allows for numerous, representative, randomly selected sites (transects) within the much larger area of interest ( P. cordata zone) to be sampled for the presence of species. The in ference gained from sampling these sites can then be applied to the pickerelweed zone La ke Tohopekaliga. The main function of this method is to determine habitat usage for speci es with low detection probabilities (<1). Detectability is an important factor when sampling secretive aquati c organisms in thick vegetation. The program calculates (1) a “nave estimate,” which is simply the proportion of sites where the species was ca ught (considered biased low), (2) speciesspecific detection probabilities based on the capture data, and (3) the “proportion of sites

PAGE 39

28 occupied” (PSO) which is the nave estim ate corrected for detection probability (MacKenzie et al. 2002). This method assumes closure of species to changes in occupancy status over the course of sampling. However, if the sp ecies have large activity ranges and the movements are assumed to be random, the closure assumption may be relaxed (MacKenzie et al. 2002). The sample occasions were divided into two groups. The first is from the start of sampling at the beginning of February until the traps were removed at the beginning of May. The second group is fr om the beginning of August, when water levels allowed full sampling effort, to the end of sampling in November. Between these groups the lake stage became so lo w that there was no water in the P. cordata zone, which surely caused a violation of the closur e assumption for this method. This required the split of sample occasions into groups th at are assumed closed to species immigration or emigration. The first (spring) group incl udes 13 sample occasions for the herpetofauna and 11 sample occasions for the fish species, since fish were not recorded for the first sample and the last sample in the group had wa ter depths too shallow to capture fish. The second (fall) group has 16 sample occasions for all species. Parameters were estimated for each species for both groups using the single season models in PRESENCE. The data were analy zed using models with both constant and survey specific detection probabilities. Resu lts from the model with the lowest Akaike’s Information Criterion (AIC) value were reported for each species. If the AIC values were within two points of each other, the simple r, constant detection probability model was selected.

PAGE 40

29 Results Trap Comparisons All reptile, amphibian and fish species captured during the 2002 whole-lake sampling, along with species codes used in the figures ar e listed in Table 3-1. Due to restrictive funnel sizes, most fish species (especially centrarchids) were represented by juvenile life stages, except small species such as mollie s and killifish. On the other hand, adult individuals characterized the majority of the reptile and amphibian species, since most young individuals could escape through the me sh. Table 3-2 shows the species and number of captures for the two types of minnow traps. Eleven vertebrate species were captured with the 0.6 cm (0.25 in) mesh minnow traps, while 19 species were captured in the 1.3 cm (0.5 in) mesh minnow traps at the sa me sample locations. Three species were unique to the 0.6 cm mesh traps on th ese occasions: black swamp snake ( Seminatrix pygaea ), flagfish ( Jordanella floridae ), and mosquitofish ( Gambusia spp). Only three reptile or amphibian species were captured: black swamp snake, Florida leopard frog ( Rana sphenocephala ), and pig frog ( Rana grylio ). More tadpoles were captured with the 0.6 cm mesh (n=31) than with the 1.3 cm mesh (n=2). Nine species were unique to the 1.3 cm mesh traps: striped mud turtle ( Kinosternon baurii ), striped crayfish snake ( Regina alleni ), Florida water snake ( Nerodia fasciata pictiventris ), Florida green water snake ( Nerodia floridana ), siren ( Siren spp), redfin pickerel ( Esox americanus ), armored catfish ( H. littorale ), dollar sunfish ( Lepomis marginatus ), spotted sunfish ( Lepomis punctatus ), largemouth bass ( Micropterus salmoides ), and redear sunfish ( Lepomis microlophus ). Seven combined reptile and amphibi an species were caught with these traps. Species common to both traps were leopard frog, pig frog, bluegill ( Lepomis macrochirus ), bluespotted sunfish ( Enneacanthus gloriosus ), golden topminnow

PAGE 41

30 Table 3-1. All species captured in 2002, with species codes used in subsequent figures. Fish Species Scientific Name Family Species Code Armored catfish Hoplosternum littorale Callichthyidae HOPLI Black crappie Pomoxis nigromaculatus Centrarchidae POMNI Bluegill Lepomis macrochirus Centrarchidae LEPMAC Bluespotted sunfish Enneacanthus gloriosus Centrarchidae ENNGL Bowfin Amia calva Amiidae AMICA Chain pickerel Esox niger Esocidae ESONI Chubsucker Erimyzon spp. Catostomidae ERIMY Dollar sunfish Lepomis marginatus Centrarchidae LEPMAR Flagfish Jordanella floridae Cyprinodontidae JORFL Gar Lepisosteus spp. Lepisosteidae LEPIS Golden shiner Notemigonus crysoleucas Cyprinidae NOTCR Golden topminnow Fundulus chrysotus Fundulidae FUNCH Largemouth bass Micropterus salmoides Centrarchidae MICSA Pterygoplichthys Pterygoplichthys spp. Loricariidae PTERY Redear sunfish Lepomis microlophus Centrarchidae LEPMI Redfin pickerel Esox americanus Esocidae ESOAM Sailfin molly Poecilia latipinna Poeciliidae POELA Seminole killifish Fundulus seminolis Fundulidae FUNSE Spotted sunfish Lepomis punctatus Centrarchidae LEPPU Warmouth Lepomis gulosus Centrarchidae LEPGU Herpetofaunal Species Scientific Name Family Species Code Amphiuma Amphiuma means Amphiumidae AMPME Cottonmouth Agkistrodon piscivorous conanti Viperidae AGKPICO Fl. banded water snake Nerodia fasciata pictiventris Colubridae NERFAPI Fl. green water snake Nerodia floridana Colubridae NERFL Fl. snapping turtle Chelydra serpentina osceola Chelydridae CHESEOS Fl. softshell turtle Apalone ferox Trionychidae APAFE Leopard frog Rana sphenocephala Ranidae RANSP Mud snake Farancia abacura abacura Colubridae FARABAB Peninsula cooter Pseudemys floridana peninsularis Emydidae PSEFLPE Pig frog Rana grylio Ranidae RANGR Siren Siren spp. Sirenidae SIREN Stinkpot Sternotherus odoratus Kinosternidae STEOD Striped crayfish snake Regina alleni Colubridae REGAL Striped mud turtle Kinosternon baurii Kinosternidae KINBA Tadpole-leopard frog Rana sphenocephala Ranidae TADRANSP Tadpole-pig frog Rana grylio Ranidae TADRANGR

PAGE 42

31 Table 3-2. Species capture frequencies for the 0.6 and 1.3 cm mesh minnow traps. Vertebrate Species 0.6 cm mesh 1.3 cm mesh Armored catfish 0 2 Black swamp snake 1 0 Bluegill 2 1 Bluespotted sunfish 18 31 Dollar sunfish 0 2 Flagfish 147 0 Florida green water snake 0 4 Florida water snake 0 1 Gambusia 75 0 Golden topminnow 17 1 Largemouth bass 0 1 Florida leopard frog 4 2 Pig frog 2 7 Redear sunfish 0 1 Redfin pickerel 0 1 Sailfin molly 56 5 Siren 0 1 Spotted sunfish 0 2 Striped crayfish snake 0 1 Striped mud turtle 0 1 Tadpoles 31 2 Warmouth 2 6 ( Fundulus chrysotus ), sailfin molly ( Poecilia latipinna ), and warmouth ( Lepomis gulosus ). Species Richness Vertebrate species richness (Figure 3-3) was negatively correlated with average air temperature (r2=0.330, df=1, p=0.001) and rainfall (r2=0.173, df=1, p=0.028). Average air temperature was the only si gnificant predictor of richne ss for the fish (Figure 3-4), (r2=0.316, df=1, p=0.002), with higher species richness estimates occurring with lower air temperatures. The estimated richness of the herpetofaunal assemblage (Figure 3-5) was negatively correlated with lake stage (r2=0.327, df=1, p=0.003), rainfall (r2=0.166, df=1, p=0.043), and water level fluctuation (r2=0.211, df=1, p=0.021).

PAGE 43

32 Assemblage Composition Six clusters were chosen to represent th e 34 sample occasions. Thirteen indicator species were determined with p<0.05 (Table 33). Indicator values are given for these species, with 100 representing perfect indi cation of that group based on relative abundances and frequency of occurrence. A zer o indicates complete absence of a species from a particular group. Since these cryptic sp ecies are quite mobile (with respect to the traps) and dependent upon detection for quantif ication, indicator values are relatively low compared to vegetation studies where virtua lly all species are de tectable. Group 1 is identified by several species, including Amia calva (bowfin), Fundulus chrysotus (golden topminnow), Lepomis macrochirus (bluegill), Lepomis marginatus (dollar sunfish), Nerodia fasciata pictiventris (Florida water snake), Rana sphenocephala (Florida leopard frog), and R. sphenocephala and Rana grylio (pig frog) tadpoles. The second group consisted only of sample occasion #13 (an extremely low water sample), and was therefore omitted. Hoplosternum littorale (armored catfish) and Regina alleni (striped crayfish snake) are the indicator species for Group 3. There were no significant indicator species for Group 4, but Amphiuma means (amphiuma) and Lepisosteus spp. (gar) show the highest indicator values with 28 and 22 respectively. Sternotherus odoratus (common musk turtle) was the sole indicator species for Group 5. The two pickerel species, Esox americanus (redfin) and Esox niger (chain), were the only two species indicating Group 6. Some species were captured duri ng every sample occasion, including Enneacanthus gloriosus (bluespotted sunfish), Lepomis gulosus (warmouth), and Siren spp. (siren). Hoplosternum littorale and Kinosternon baurii (striped mud turtle) were found on almost every sample occasion. It is unclear why the armored catfish is an

PAGE 44

33 indicator species for Group 3, when it has an in dicator value of 22 for all groups but one. Also, the bluegill had indicator values of 47 for both Group 1 and Group 6, although it was assigned to Group 1 with p=0.028. Table 3-3. Indicator species analysis results. Significant indicator species are highlighted (p<0.05) and displayed with associated indicator values and the clusters to which the species was assigned. Species Code Cluster Max Indicator Value Mean Standard Deviation Probability AGKPICO 1 25 15 7.4 0.211 AMICA 1 48.5 18 9.86 0.017 AMPME 1 27.9 24.8 4.25 0.234 APAFE 5 9.1 15.3 7.72 1 CHESEOS 4 12.5 15.3 7.49 0.696 ENNGL 1 20 20 0.63 1 ERIMY 6 27.9 23.1 7.1 0.201 ESOAM 6 40.9 22.6 7.6 0.022 ESONI 6 43.3 18.1 10.46 0.022 FARABAB 6 21.3 18.2 10.31 0.29 FUNCH 1 43.8 21.3 8.26 0.04 HOPLI 3 22.2 21.3 0.82 0.039 JORFL 1 34.1 16.8 9.59 0.077 KINBA 1 20.5 20.6 0.73 0.672 LEPGU 1 20 20 0.63 1 LEPIS 4 22.3 24.9 3.41 0.76 LEPMAC 1 47.1 21.9 8.23 0.028 LEPMAR 1 36.8 23.8 7.01 0.05 LEPMI 6 33.8 22.3 8.45 0.1 LEPPU 5 12 22.2 8.8 0.976 MICSA 6 28.8 23 7.01 0.178 NERFAPI 1 40.8 22.7 7.03 0.018 NERFL 3 21.6 21.3 0.83 0.295 NONE 1 31.7 19.9 9.88 0.107 NOTCR 6 28.6 16.5 8.92 0.059 POELA 1 28.8 24.6 4.2 0.115 POMNI 1 25 14.9 7.44 0.196 PSEFLPE 5 9.1 15.1 7.52 1 PTERY 5 18.7 16.9 9.76 0.179 RANGR 1 22.4 23.4 1.3 0.92 RANSP 1 100 17.4 9.61 0.001 REGAL 3 72.6 17.8 10.18 0.002 SIREN 1 20 20 0.63 1 STEOD 5 29.8 24.4 2.19 0.001 TADRANGR 1 34.6 24.4 5.45 0.009 TADRANSP 1 75 17 9.16 0.001

PAGE 45

34 Influence of Temporal Gr adients on Assemblage Stable three-dimensional ordinations were produced with all NMS analyses except the herpetofaunal analysis w ith detection/nondetection data. The final solutions were based on the criteria of stress being redu ced by at least 5% with each additional dimension. The final stress values were lowe r with the real data than was found by the Monte Carlo randomized runs (p<0.05), which indicates that there was real structure found in the data. The final stress values fo r all ordinations (except for herpetofaunal detection/nondetection) were between 12 and 18 (Table 3-4), which are common values for ecological data and depict a fair portrayal of the da ta (McCune and Grace 2002). Table 3-4. Stress and instability results from all NMS ordinations Assemblage Data Type Final Stress Instability Iterations Vertebrate Detection/ nondetection17.65 0.00254 500 Vertebrate Counts 13.27 0.00049 69 Fish Detection/nondetection15.31 0.00095 29 Fish Counts 12.93 0.00016 49 Herpetofauna Counts 14.08 0.00045 38 For the NMS with vertebrate detecti on/nondetection data, Axes 1 and 3 best explained 61% of the variance found in th e assemblage composition. Environmental variables with r2>0.2 were shown as biplots on the pl ots and include la ke stage and air temperature measures (Figure 3-6, Table 3-5) Axis 1 was correlated with lake stage (r2=0.349). Axis 3 was most correlated with bo th lake stage and average air temperature, (r2=0.354 and 0.259 respectively). Sixt een species were correlated with either Axis 1 or 3 with r2>0.2 (Figure 3-7, Table 3-6), 10 of them being indicator species.

PAGE 46

35 Table 3-5. Percent of variance explained (r2) by environmental variables for each axis in the vertebrate NMS with detection/n ondetection data. Variables with r2 > 0.2 are highlighted. Axis 1 2 3 Variable r2 r2 r2 Stage(m) 0.349 0.021 0.345 Fluc(cm) 0.039 0.009 0.183 Rain(cm) 0.037 0.006 0.029 Max(C) 0.038 0.362 0.127 Min(C) 0.002 0.255 0.229 Ave(C) 0.009 0.375 0.259 Table 3-6. Percent of variance explained (r2) for each axis by species in the vertebrate NMS with detection/nondetection da ta. Indicator species with r2 > 0.2 are highlighted in blue, while all other species with r2 > 0.2 are highlighted in yellow. Axis 1 2 3 Axis 1 2 3 Species r2 r2 r2 Species r2 r2 r2 AGKPICO 0.073 0.002 0.094 LEPMI 0.035 0.56 0.018 AMICA 0.027 0.077 0.327 LEPPU 0.064 0.027 0 AMPME 0.002 0.171 0.01 MICSA 0.291 0.227 0.242 APAFE 0.035 0.004 0.001 NERFAPI 0.227 0.167 0.257 CHESEOS 0.021 0 0.063 NERFL 0.003 0 0 ENNGL 0.381 0.019 0.002 NONE 0.057 0.004 0.102 ERIMY 0.031 0.536 0.007 NOTCR 0.011 0.05 0.017 ESOAM 0.033 0.186 0.353 POELA 0.188 0.018 0.248 ESONI 0.126 0.166 0.089 POMNI 0.007 0.011 0.067 FARABAB 0.049 0 0.21 PSEFLPE 0.005 0.085 0.001 FUNCH 0.019 0.419 0.547 PTERY 0.002 0.081 0.038 HOPLI 0.376 0.003 0.155 RANGR 0.149 0.035 0 JORFL 0.087 0.047 0.15 RANSP 0.218 0.02 0.331 KINBA 0 0.07 0.016 REGAL 0.01 0.117 0.037 LEPGU 0.381 0.019 0.002 SIREN n/a n/a n/a LEPIS 0.36 0.007 0.083 STEOD 0.021 0.006 0.018 LEPMAC 0.002 0.311 0.53 TADRANGR0.016 0.127 0.311 LEPMAR 0.019 0.04 0.422 TADRANSP0.07 0.039 0.287 With vertebrate assemblage count da ta, Axes 1 and 3 had the highest r2. Together these axes represent 63% of the variance in the species composition. Axis 2 also had an r2>0.2, and was best correlated with air temperature measures. As with the detection/nondetection analysis lake stage and air temper ature measures showed the

PAGE 47

36 highest correlation with these axes (Figure 3-8, Table 3-7). Air temperature is correlated with Axis 1, with maximum air temperature having the highest r2 of 0.353. Axis 3 was most highly correlated with lake stage (r2=0.458) and average air temperature (r2=0.375). Seventeen species had an r2>0.2 for at least one of the axes nine of them being indicator species (Figure 3-9, Table 3-8). Table 3-7. Percent of variance explained (r2) by environmental variables for each axis in the vertebrate NMS with count data. Variables with r2 > 0.2 are highlighted. Axis 1 2 3 Variable r2 r2 r2 Stage(m) 0.146 0.010 0.458 Fluc(cm) 0.013 0.006 0.108 Rain(cm) 0.001 0.001 0.044 Max(C) 0.353 0.299 0.232 Min(C) 0.261 0.346 0.281 Ave(C) 0.298 0.379 0.375 Table 3-8. Percent of variance explained (r2) for each axis by species in the vertebrate NMS with count data. Indicator species with r2 > 0.2 are highlighted in blue, while all other species with r2 > 0.2 are highlighted in yellow. Axis 1 2 3 Axis 1 2 3 Species r2 r2 r2 Species r2 r2 r2 AGKPICO 0.16 0.005 0.095 LEPMI 0.241 0.331 0.139 AMICA 0.059 0.089 0.192 LEPPU 0.06 0.02 0.018 AMPME 0.183 0.124 0.211 MICSA 0.13 0.184 0.048 APAFE 0.044 0.021 0.017 NERFAPI 0.166 0.001 0.373 CHESEOS 0 0.002 0.087 NERFL 0.068 0.024 0.004 ENNGL 0.809 0.342 0.108 NONE 0.031 0.011 0.164 ERIMY 0.217 0.024 0.092 NOTCR 0.124 0.136 0.022 ESOAM 0.014 0.455 0.09 POELA 0.133 0.023 0.341 ESONI 0.051 0.108 0.044 POMNI 0.022 0.005 0.087 FARABAB 0.041 0.028 0.018 PSEFLPE 0.065 0 0.038 FUNCH 0.002 0.291 0.575 PTERY 0.063 0.005 0.115 HOPLI 0.05 0.285 0.653 RANGR 0.125 0.157 0 JORFL 0.186 0.04 0.172 RANSP 0.092 0.038 0.345 KINBA 0.43 0.003 0.005 REGAL 0.039 0.039 0.005 LEPGU 0 0.104 0.013 SIREN 0.428 0.46 0.172 LEPIS 0.005 0.158 0.288 STEOD 0.268 0.182 0.288 LEPMAC 0.079 0.423 0.386 TADRANGR 0.147 0.159 0.008 LEPMAR 0.023 0.182 0.4 TADRANSP 0.137 0.065 0.24

PAGE 48

37 The fish assemblage alone with detection/ nondetection data resulted in the first two axes representing 63% of the variance expl ained. Average temperature was most highly correlated with Axis 1 (r2=0.291), and also with Axis 2 (r2=0.368), while lake stage was correlated with Axis 2 (r2=0.275), (Figure 3-10, Table 3-9). For these two axes, there were a total of 11 species with an r2>0.2, with six indicator sp ecies (Figure 3-11, Table 310). Table 3-9. Percent of variance explained (r2) by environmental variables for each axis in the fish NMS with detection/nondet ection data. Variables with r2 > 0.2 are highlighted. Axis 1 2 3 Variable r2 r2 r2 Stage(m) 0.042 0.275 0.105 Fluc(cm) 0.008 0.044 0.062 Rain(cm) 0.001 0.031 0.018 Max(C) 0.250 0.177 0.115 Min(C) 0.205 0.330 0.106 Ave(C) 0.291 0.368 0.083 Table 3-10. Percent of variance explained (r2) for each axis by species in the fish NMS with detection/nondetection data Indicator species with r2 > 0.2 are highlighted in blue, while all other species with r2 > 0.2 are highlighted in yellow. Axis 1 2 3 Axis 1 2 3 Species r2 r2 r2 Species r2 r2 r2 AMICA 0.152 0.296 0.12 LEPMAC 0.361 0.646 0 ENNGL n/a n/a n/a LEPMAR 0.406 0.229 0.168 ERIMY 0.149 0.089 0.223 LEPMI 0.118 0.205 0.436 ESOAM 0.287 0.223 0.083 LEPPU 0.002 0.009 0.021 ESONI 0.195 0.066 0.013 MICSA 0.706 0.049 0.001 FUNCH 0.486 0.572 0.003 NOTCR 0.072 0.026 0.018 HOPLI 0.058 0.251 0.149 POELA 0.007 0.263 0.264 JORFL 0.038 0.181 0.137 POMNI 0.034 0.009 0.004 LEPGU n/a n/a n/a PTERY 0.253 0.003 0.046 LEPIS 0.021 0.339 0.004 For the fish ordination with count data, Axes 2 and 3 represent the most variation in the assemblage data with a cumulative r2 of 0.76. Again, air temperature and stage are

PAGE 49

38 the environmental variables most correlated wi th these axes (Figure 3-12, Table 3-11). Axis 2 was most highly correlated with maximum air temperature (r2=0.380), while lake stage (r2=0.287) and average air temperature (r2=0.333) are most correlated with Axis 3. Fourteen of the fish species had an r2>0.2 for these two axes, with six of them being indicator species (Fi gure 3-13, Table 3-12). Table 3-11. Percent of variance explained (r2) by environmental variables for each axis in the fish NMS with count data. Variables with r2 > 0.2 are highlighted. Axis 1 2 3 Variable r2 r2 r2 Stage(m) 0.070 0.005 0.287 Fluc(cm) 0.008 0.001 0.064 Rain(cm) 0.003 0.000 0.024 Max(C) 0.242 0.380 0.189 Min(C) 0.346 0.303 0.243 Ave(C) 0.319 0.354 0.333 Table 3-12. Percent of variance explained (r2) for each axis by species in the fish NMS with count data. Indicator species with r2 > 0.2 are highlighted in blue, while all other species with r2 > 0.2 are highlighted in yellow. Axis 1 2 3 Axis 1 2 3 Species r2 r2 r2 Species r2 r2 r2 AMICA 0 0.008 0.445 LEPIS 0.359 0.044 0.07 ENNGL 0.171 0.57 0.239 LEPMAC 0.085 0.308 0.503 ERIMY 0.221 0.367 0.135 LEPMAR 0.033 0.185 0.624 ESOAM 0.159 0.165 0.372 LEPMI 0.227 0.359 0.164 ESONI 0.002 0.252 0.156 LEPPU 0.074 0.274 0.006 FUNCH 0.113 0.106 0.757 MICSA 0 0.64 0.294 HOPLI 0.108 0.03 0.194 NOTCR 0 0.152 0.058 JORFL 0.004 0.046 0.303 POELA 0.17 0 0.423 LEPGU 0.191 0.337 0.223 POMNI 0.054 0.077 0.016 The herpetofaunal assemblage detection/ nondetection data were analyzed with NMS, but results yielded only a one-dimensi onal solution. Stress on the final run was 51.36, which represents an unacceptable amount of variation from the original data set. Values of stress greater than 20 indicate that the solution may be misleading, while ordinations with stress values over 40 represent very little of the structure in the original

PAGE 50

39 data matrix (McCune and Grace 2002). Due to these outcomes, NMS was considered unsuccessful for the reptile and amphibian a ssemblage with detection/nondetection alone. Count data for the reptiles and amphibians did yield a successful ordination. The first and third axes explain a total of 69% of the variation in the assemblage composition. Lake stage is highly corr elated with Axis 1 (r2=0.267) and Axis 3 (r2=0.389), and water level fluctuation is correlated with Axis 3 (r2=0.204), (Figure 3-14, Table 3-13). Eight species were correlated w ith Axes 1 and 3 with r2>0.2, six of them being indicator species (Figure 3-15, Table 3-14). Table 3-13. Percent of variance explained (r2) by environmental variables for each axis in the herpetofaunal NMS with count data. Variables with r2 > 0.2 are highlighted. Axis 1 2 3 Variable r2 r2 r2 Stage(m) 0.267 0.081 0.389 Fluc(cm) 0.006 0.002 0.204 Rain(cm) 0.000 0.001 0.057 Max(C) 0.052 0.020 0.031 Min(C) 0.022 0.034 0.091 Ave(C) 0.058 0.016 0.073 Table 3-14. Percent of variance explained (r2) for each axis by species in the herpetofaunal NMS with count data. Indicator species with r2 > 0.2 are highlighted in blue, while all other species with r2 > 0.2 are highlighted in yellow. Axis 1 2 3 Axis 1 2 3 Species r2 r2 r2 Species r2 r2 r2 AGKPICO 0.041 0.188 0.012 PSEFLPE 0.042 0.079 0.04 AMPME 0.014 0.291 0.094 RANGR 0.025 0.229 0.229 APAFE 0.087 0 0.017 RANSP 0.185 0.061 0.263 CHESEOS 0.154 0.018 0.036 REGAL 0.384 0.055 0.109 FARABAB 0.004 0.024 0.409 SIREN 0.018 0.031 0.047 KINBA 0.083 0.215 0.351 STEOD 0.437 0.091 0.033 NERFAPI 0.275 0.023 0.498 TADRANGR 0.002 0.095 0.311 NERFL 0.097 0.258 0.03 TADRANSP 0.107 0.275 0.057

PAGE 51

40 To determine the separation of the sample occasion clusters along lake stage and average air temperature gradients means and ranges of each are shown in Figures 3-16 and 3-17 respectively. Clusters one and two have lake stage ranges lower and nonoverlapping with clusters four, five and six. Stage values for cluster 3 span much of the ranges for every other group. Air temperature va lues are very similar for clusters one and six, both being below and non-overlapping with cl usters two, three and four. Cluster five overlaps with most of the ra nges of all other groups. Proportion of Habitat Ut ilized by Focal Species Figures 3-18 and 3-19 show the occupancy rate estimates for the fish species. Several species exhibited a decrease in site occupancy from the spring to the fall season, most pronounced in M. salmoides, L. macrochirus and L. microlophus. Lepomis microlophus, L. gulosus and E. gloriosus occurred at all transects during the spring. Lepisosteidae spp., H. littorale and L. gulosus were present in all transects in the fall. Lepomis punctatus had invalid occupancy estimates during the spring due to low detection probabilities, w ith the same problem for L. macrochirus and L. microlophus in the fall. In the spring, the only fish species th at used survey-specific detection probability models for occupancy estimation were E. gloriosus and L. macrochirus, which indicates that detection of these two species varied between sample occasions. All species were modeled with time-constant detectio n probabilities in the fall, except E. gloriosus, P. latipinna and Lepisosteidae spp. Estimates of the proportion of sites occ upied for the eight focal herpetofaunal species are presented in Figures 3-20 and 3-24. Site occ upancy was similar in the spring and fall for most species. There were not enough data to make valid estimates for A. means in the spring and for R. sphenocephala and N. fasciata pictiventris in the fall. All

PAGE 52

41 species were modeled with the time constant detection probability except N. floridana and R. sphenocephala in the spring. With the exception of R. sphenocephala and A. means most species were estimated to be in 90-1 00% of all transects at some point in the year. Discussion Trap Comparisons The 1.3 cm (0.5 in) mesh traps were bette r at capturing the sp ecies of interest, including reptiles, amphibia ns, centrarchids and exotic catfish. Although it had been assumed that many species would be under-rep resented with the larger mesh, the few unique species caught in the 0.6 cm (0.25 in) mesh traps were not of particular interest to this study. The fish species included mosquito fish and flagfish, which are not suspected of being impacted by the removal of the vege tation from the littoral zone. The one black swamp snake was actually stuck about halfwa y through the mesh, indicating that even 0.6 cm mesh may not be small enough to capture th is species representa tively. The size and shape of the traps and openings may also have affected capture proba bilities of different species. Also, the bare metal hardware cloth may be more visible to the animals than the dark green vinyl-coated hardwa re cloth of the larger mesh traps. In comparing practicality of sampling with each type of trap, the 0.6 cm commercially manufactured traps were much less durable and prone to breaking during use, as well as being more expensive ($16.16 each for 0.6 cm vs. $11.00 each for 1.3 cm). Given these factors, the 1.3 cm mesh traps were more useful than the 0.6 cm for sampling in this habitat. Species Richness Ectothermic fish communities respond to water temperature due to its effect on metabolism and growth rates, especially fo r juveniles (Holt 2002) This relationship

PAGE 53

42 determines the length of time that species will benefit from residing in a given habitat. Water temperature is closely associated with air temperature due to the shallow aquatic habitat. Therefore, it is not surprising th at the average air temperature is negatively correlated with fish species richness. In the summer season, water temperature may exceed tolerance limits for sensitive fish such as select species of sunfish, causing them to leave the habitat and resulting in fewer fish species. Temperature also indirectly determines the amount of dissolved oxygen in the water; however this variable was not measured in the field. Lake stage, rainfall, and water level fluc tuations were signifi cant predictors of herpetofaunal richness, all of which have a re lationship to water dept h at the fixed trap locations. The gradually sloping contour of Lake Tohopekaliga causes small increases in lake stage to flood broad expanses of previ ously dry habitat, allo wing species to enter into new habitat for spawning or foraging. A lternatively, moderate drops in lake stage may cause a rapid decrease in the area of th e littoral zone inundated with water, causing species to emigrate or burrow, or else risk being trapped by unsuitable conditions. Receding water levels may bring in more terr estrial species into th e previously aquatic habitat, such as cottonmouths, Florida water snakes, and Florida leopard frogs. Alternatively, deeper water in the habitat may restrict species that prefer shallow water and promote more aquatic species su ch as the common musk turtle. Assemblage Composition Since animal species move throughout the habitat and are detected with probabilities less than one, and animal assemb lages tend to be transient in nature, the groups determined by cluster analysis are not very discrete with respect to species

PAGE 54

43 composition. This resulted in fairly low indicator values, but with several being significant nonetheless. The four sample occasions in Group 1 had the lowest lake stage values and were tied with Group 6 for the lowest average air temperatures. The Florida water snake and Florida leopard frog are the herp etofaunal indicator species fo r this group, being the more terrestrial of the focal rep tile and amphibian species. Bowfins, golden topminnows, bluegill and dollar sunfis h are the fish indicator species. Bowfins are one of the most tolerant freshwater fish speci es and are often called "mud fish" (Boschhung et al. 1995), so it is not unreasonable that this species w ould occupy this shallow vegetated habitat. Golden topminnows prefer to occupy la kes with abundant vegetation (Hoyer and Canfield 1994), and therefore may be more tolerant of dense vegetation communities than other species. It is not as easy to explain why the two sunfish species were indicators for these shallow water depths. They may have been stranded by receding waters and captured more easily in trap s with puddles of water remaining. Group 2 only had one sample occasion attributed to it. This sample was the last one in late April before the traps were rem oved due to lack of water in the habitat. Although not included in the indicator species analysis, this sample included no fish species. Only three sample occasions were cl ustered into Group 3. During these hot summer samples, the lake stage was rapidly ri sing, and trap effort was reduced pending appropriate water depths at th e trap sites. The striped crayfish snake was the main indicator species for this gr oup. Being a specialist predator on crayfish (Godley 1980),

PAGE 55

44 prey availability in recently flooded habitats may have been the driving factor for their indicator values on these occasions. Eight samples fell into Group 4, which had no indicator species attributed to it. These sample occasions occurred in early fall with the highest lake stages and average air temperatures. Group 5 occurred mainly in the summer months (eight samples), but also contained two samples in February. Water depths were moderately high. Air temperatures were low in February and high in summer, spanni ng a wide range of temperatures. The sole indicator species for this group was the comm on musk turtle. This species is known for being highly aquatic, leaving the water onl y to nest (Wygoda 1979, Gibbons et al. 1983). It also is active for the widest temperatur e ranges of any other kinosternid species in North America, being able to retreat to deeper waters to buffer the effects of air temperature extremes (Mahmoud 1969, Ernst 1986). The last cluster, Group 6, included seven samples, with six occurring February through April and the other one in November. Low air temperatures and moderately high but dropping lake stages characterize these sample occasions. The two Esox spp. (pike) are the indicators for this group. These speci es breed from February to March in the south, spawning in densely vegetated habitats less than 50 cm deep (Billard 1996). This may explain their presence in the habita t during these environmental conditions. Influence of Temporal Gr adients on Assemblage Detection/nondetection data were used for ordinations, and are generally recommended when comparing habitat distribu tions of species (Hayek 1994), and when sample unit heterogeneity is large (McCune and Grace 2002). Counts were also used to compare results obtained by the two types of data, but although agreement between the

PAGE 56

45 two provides more support, lack of detection probabilities make count data less valuable. For the vertebrate and fish ordinations, result s were similar for both types of data. The herpetofaunal ordination was unsuccessful with detection/ nondetection data, and therefore counts were used solely. Average air temperature and lake stage cam e out as the most important variables correlated with the axes representing varia tion in species composition in the vertebrate and fish assemblages. As mentioned before temperature influences growth rates for young fish, as well as the amount of dissolved ox ygen in the water. Both of these factors limits the time that fish are able to occupy a habitat. Physical access to heavily vegetated habitat is also limited by water depth, which is controlled by lake stage. This determines the volume of water the animals have to move through, as well as the effect of vegetation density in the water column. However, for herpetofaunal species alone air temperature is not associated with variation in the species composition. In th is case, lake stage is most important, with water level fluctuation also s howing a correlation with one of the axes. Lake stage probably dictates movements of sp ecies that do not show site fidelity, in response to habitat requirements and prey ava ilability. For species that are not known to move long distances, for example sirens a nd amphiumas, low water levels trigger burrowing activities (Aresco 2001), ther eby reducing captu re opportunities. Proportion of Habitat Ut ilized by Focal Species Due to the large-scale removal of picker elweed from the littoral zone of Lake Tohopekaliga during enhancement activities, it was important to investigate the spatial distribution of species in this habitat. Site occupancy analys es were used to estimate the proportion of this habitat type that was used throughout the year by various fish, reptile and amphibian species. While some species ar e temporally and spatially pervasive in the

PAGE 57

46 habitat (e.g., warmouths, blues potted sunfish, Florida green wa ter snakes, sirens, striped mud turtles, and pig frogs), others seem to use the Pontederia cordata zone intermittently. Of the fish species, sailfin mollies, spotted sunfish, chubsuckers, and largemouth bass were found in a moderate propo rtion of transects ( 30-70%) in both the spring and fall. Bluegill and redear sunfis h were both in a high proportion of sites (>80%) in the spring, but were found in less than 20% of the transects in the fall. This trend may be due to juvenile fish using the littoral zone for foraging and predator avoidance during the spring when suitable physicochemical conditions permit survival (Werner and Hall 1979, Crowder and Cooper 1982, Werner and Hall 1988, Chapman et al. 1996). Unmeasured environmental charact eristics such as low dissolved oxygen may have kept the sunfish out of the thick vegetation after the summer low-water spell (Miranda and Hodges 2000). Gars and armore d catfish went from about 65-80% of the sites in the spring to 100% occ upancy in the fall. These two species are far more tolerant of harsh environmental conditions than most sunfish due in part to their capacity for aerial respiration (Boschung et al. 1995, Braune r et al. 1995). For the armored catfish, the greater presence in the fall may be due to the breeding season and sufficiently high water levels for nesting (Mol 1993). Most reptile and amphibian species occupi ed a similar proportion of sites in both the spring and fall. Florida leopard frogs a nd Florida water snakes were only captured when water levels were very low, which restri cted reasonable estimates of site occupancy to the spring season. Since the heavily vege tated littoral zone is known as prime habitat for several of these species due to life histor y requirements, it is not surprising to find most of the focal species in such a high proportion of the s ites (>70% occupancy).

PAGE 58

47 Figure 3-1. Crayfish and minnow trap in P. cordata habitat.

PAGE 59

48 Figure 3-2. Locations of 2002 P. cordata sampling transects in Lake Tohopekaliga

PAGE 60

49 Date 1/1/02 2/1/02 3/1/02 4/1/02 5/1/02 6/1/02 7/1/02 8/1/02 9/1/02 10/1/02 11/1/02 12/1/02 1/1/03 Species Richness with 95% CI 0 5 10 15 20 25 30 35 40 45 50 55 Figure 3-3. 2002 Vertebrate sp ecies richness estimates by sample date, with points representing richness for the time between the last sample occasion and the sample date. Date 1/1/02 2/1/02 3/1/02 4/1/02 5/1/02 6/1/02 7/1/02 8/1/02 9/1/02 10/1/02 11/1/02 12/1/02 1/1/03 Species Richness with 95% CI 0 5 10 15 20 25 30 35 40 Figure 3-4. 2002 Fish species ri chness estimates by sample date, with points representing richness for the time between the last sample occasion and the sample date.

PAGE 61

50 Date 1/1/02 2/1/02 3/1/02 4/1/02 5/1/02 6/1/02 7/1/02 8/1/02 9/1/02 10/1/02 11/1/02 12/1/02 1/1/03 Species Richness with 95% CI 0 10 20 30 40 Figure 3-5. 2002 Herpetofaunal species richne ss estimates by sample date, with points representing richness for the time between the last sample occasion and the sample date.

PAGE 62

51 10 11 12 13 15 16 17 18 19 2 20 21 22 23 24 25 26 27 28 29 3 30 31 32 33 34 35 36 4 5 6 7 8 9 Stage(m) A ve(C) A xis 1Axis 3 Cluster 1 2 3 4 5 6 Figure 3-6. NMS ordination of sample uni ts in vertebrate species space using detection/nondetection data. Points re present sample occasions and distances between points show the relative diffe rences in species composition. The length of each line is proportional to th e strength of the correlation between the environmental gradient and the synthetic axes.

PAGE 63

52 Figure 3-7. NMS ordination of vertebrate species in sample unit space using detection/nondetection data. Points represent average species positions with respect to sample units. The length of each line is proportional to the strength of the correlation between the environmental gradient and the synthetic axes. AGKPICO AMICA AMPME APAFE CHESEOS ENNGL ERIMY ESOAM ESONI FARABAB FUNCH HOPLI JORFL KINBA LEPGU LEPIS LEPMAC LEPMAR LEPMI LEPPU MICSA NERFAPI NERFL NONE NOTCR POELA POMNI PSEFLPE PTERY RANGR RANSP REGAL SIREN STEOD TADRANGR TADRANSP Stage(m) Ave(C) Axis 1Axis 3

PAGE 64

53 Figure 3-8. NMS ordination of sample units in vertebrate species space using count data. Points represent sample occasions and distances between points show the relative differences in species comp osition. The length of each line is proportional to the strength of the correlation between the environmental gradient and the synthetic axes. 10 11 12 15 16 17 18 19 2 20 21 22 23 24 25 26 27 28 29 3 30 31 32 33 34 35 36 4 5 6 7 8 9 Stage(m) Ave(C) Axis 1Axis 3 Cluster 1 3 4 5 6

PAGE 65

54 Figure 3-9. NMS ordination of ve rtebrate species in sample unit space using count data. Points represent average species positions with respect to sample units. The length of each line is proportional to th e strength of the correlation between the environmental gradient and the synthetic axes. AGKPICO AMICA AMPME APAFE CHESEOS ENNGL ERIMY ESOAM ESONI FARABAB FUNCH HOPLI JORFL KINBA LEPGU LEPIS LEPMAC LEPMAR LEPMI LEPPU MICSA NERFAPI NERFL NONE NOTCR POELA POMNI PSEFLPE PTERY RANGR RANSP REGAL SIREN STEOD TADRANGR TADRANSP Stage(m) Ave(C) Axis 1Axis 3

PAGE 66

55 Figure 3-10. NMS ordination of sample units in fish species space using detection/nondetection data. Points re present sample occasions and distances between points show the relative diffe rences in species composition. The length of each line is proportional to th e strength of the correlation between the environmental gradient and the synthetic axes. Ave(C) 10 11 12 15 16 17 18 19 2 20 21 22 23 24 25 26 27 28 29 3 30 31 32 33 34 35 36 4 5 6 7 8 9 Stage(m) Axis 1Axis 2 Cluster 1 3 4 5 6

PAGE 67

56 Figure 3-11. NMS ordination of fish species in sample unit space using detection/nondetection data. Points represent average species positions with respect to sample units. The length of each line is proportional to the strength of the correlation between the environmental gradient and the synthetic axes. AMICA ENNGL ERIMY ESOAM ESONI FUNCH HOPLI JORFL LEPGU LEPIS LEPMAC LEPMAR LEPMI LEPPU MICSA NOTCR POELA POMNI PTERY Stage(m) Ave(C) Axis 1Axis 2

PAGE 68

57 Figure 3-12. NMS ordination of sample units in fish species space using count data. Points represent sample occasions and distances between points show the relative differences in species comp osition. The length of each line is proportional to the strength of the correlation between the environmental gradient and the synthetic axes. 10 11 12 15 16 17 18 19 2 20 21 22 23 24 25 26 27 28 29 3 30 31 32 33 34 35 36 4 5 6 7 8 9 Stage(m) Ave(C) Axis 2Axis 3 Cluster 1 3 4 5 6

PAGE 69

58 Figure 3-13. NMS ordination of fish species in sample unit space using count data. Points represent average species positions with respect to sample units. The length of each line is proportional to th e strength of the correlation between the environmental gradient and the synthetic axes. AMICA ENNGL ERIMY ESOAM ESONI FUNCH HOPLI JORFL LEPGU LEPIS LEPMAC LEPMAR LEPMI LEPPU MICSA NOTCR POELA POMNI Stage(m) Ave(C) Axis 2Axis 3

PAGE 70

59 Figure 3-14. NMS ordination of sample units in herpetofaunal species space using count data. Points represent sample occasions and distances between points show the relative differences in species composition. The length of each line is proportional to the strength of the correlation between the environmental gradient and the synthetic axes. 10 11 12 13 15 16 17 18 19 2 20 21 22 23 24 25 26 27 28 29 3 30 31 32 33 34 35 36 4 5 6 7 8 9 Stage(m) Fluct(cm) Axis 1Axis 3 Cluster 1 2 3 4 5 6

PAGE 71

60 Figure 3-15. NMS ordination of herpetofauna l species in sample unit space using count data. Points represent average species pos itions with respect to sample units. The length of each line is proportional to the strength of the correlation between the environmental grad ient and the synthetic axes. AGKPICO AMPME APAFE CHESEOS FARABAB KINBA NERFAPI NERFL PSEFLPE RANGR RANSP REGAL SIREN STEOD TADRANGR TADRANSP Stage(m) Fluct(cm) Axis 1Axis 3 Cluster 1 2 3 4 5 6

PAGE 72

61 Cluster 123456 Lake Stage (m) 16.0 16.1 16.2 16.3 16.4 16.5 16.6 16.7 16.8 Figure 3-16. Average and range of lake stage values by cluster. Cluster 123456 Average Air Temperature (OC) 12 14 16 18 20 22 24 26 28 30 Figure 3-17. Average and range of air temperature values by cluster.

PAGE 73

62 FISH SPECIES POELALEPPUERIMYLEPISMICSAHOPLILEPMACLEPMILEPGUENNGL PROPORTION -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 NAIVE ESTIMATE PROPORTION OF SITES OCCUPIED WITH STANDARD ERROR Figure 3-18. Site occupancy estimates for focal fish species in spring 2002. FISH SPECIES POELALEPPUERIMYLEPISMICSAHO PLILEPMACLEPMILEPGUENNGL PROPORTION -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 NAIVE ESTIMATE PROPORTION OF SITES OCCUPIED WITH STANDARD ERROR Figure 3-19. Site occupancy estimate s for focal fish species in fall 2002.

PAGE 74

63 HERPETOFAUNAL SPECIES RANSPSTEODAMPMERANGRNERFAPIKINBASIRENNERFL PROPORTION -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 NAIVE ESTIMATE PROPORTION OF SITES OCCUPIED WITH STANDARD ERROR Figure 3-20. Site occupancy estimates for focal herpetofaunal sp ecies in spring 2002. HERPETOFAUNAL SPECIES RANSPSTEODAMPMERANGRNERFAPIKINBASIRENNERFL PROPORTION -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 NAIVE ESTIMATE PROPORTION OF SITES OCCUPIED WITH STANDARD ERROR Figure 3-21. Site occupancy estimates fo r focal herpetofauna l species in fall 2002.

PAGE 75

64 CHAPTER 4 ASSEMBLAGE ACROSS VEGETATION COMMUNITIES Introduction The objective of this section is to inves tigate the influence of vegetation type and water depth on spatial variati on in the aquatic vertebrate community. The three main research questions are 1) are we able to estimate population parameters such as abundance and density for the focal species us ing trapping grids or webs, 2) is there spatial variation in the aquatic vertebra te assemblage, and 3) do the vegetation communities or water depths influence the spatial variation for the individual focal species? Field Methods Grid and Web Sampling A pilot mark-recapture protocol was employed in the summer of 2002 in order to estimate activity ranges, abundances and densi ties of species of in terest. Trap points were arranged in a square grid pattern (W hite et al. 1982). In order to have uniform sampling effort within the grids, one minnow and one crayfish trap was placed at each point. All newly captured reptiles and amphi bians were weighed and measured, then tagged with Passive In tegrated Transponders (PIT tags), which are small microchips inserted under the skin to permit individual identification when scanned. Each animal was released at the cap ture location after being worked up. Traps were checked daily for PIT tagged individuals, and all new individua ls were measured and tagged during each sampling event. In order to satisfy the assu mption of closure for analysis and minimize

PAGE 76

65 temporal variation in detecti on probabilities, the grids were sampled for 5-7 days. This technique was used only for the most abunda nt species, since dete ction probabilities would be too small to make accurate estimat es of any others. These species included A. means Siren spp., N. floridana and K. baurii The first grid consisted of 49 trap points, each three m (9.8 ft) apart, in a seven by seven grid ("GRID1"), and sampled for seven consecutive days in July. Next, 100 trap points were placed in a 10 by 10 grid ("GR ID2A"), five meters (16.4 ft) apart, and sampled for six consecutive days in August. The same grid was then sampled for five non-consecutive days ("GRID2B"), using frozen sardines as bait, until it was decided that the bait was logistically imp ractical. The last design wa s 100 trap points ("GRID3"), spaced three meters (9.8 ft) apart, in a 10 by 10 grid. This was sampled for five times over seven days, due to logistic problems. A ttempts to sample several times within a day were also made, but were discontinued imme diately due to low numbers of captures. In addition to these grids, a trapping web was attempted in September 2002. The web is a variation of point-transect distance sampling, typically used for small terrestrial animals. It is designed to have lines of tr aps radiating out from a center point, forming a gradient of sampling effort and detection probab ilities. Data from each concentric ring is grouped according to distance intervals (Anderson et al. 1983). This web consisted of eight radiating arms of 12 trap points each, placed at threemeter (9.8 ft) intervals, for a total of 96 trap points. The benefit of this me thod is that it uses onl y the initial captures, so the recapture rate is irrelevant. The a ssumptions of this method are that all animals near the center of the web are captured, the size of the web is large relative to the movements of the animals, distances are measured accurately from the center, and

PAGE 77

66 individual captures are i ndependent events. Sampling continues daily until no new animals are caught at the center of the we b, indicating 100% dete ction at the center (Anderson et al. 1983) Since it was evident that the assumptions were not met in this web, sampling was discontinued after six sample occasions. Whole-Lake Sampling After the postponement of the fall 2002 drawdown, we began sampling again with modifications to the 2002 temporal sampling protocol. The goal was to investigate differences in vertebrate habitat usage of vegetation communities beyond the P. cordata zone, as well as varying water depths. Ei ghteen new locations were randomly chosen for transects (Figure 4-1), since th e sampling of 2002 had disturbed the habitat in some of the previous locations. Sampling sites in Goblet Â’s Cove were included and none were placed in the disturbed stretch of shoreline in the southern part of the lake. Transects were placed at least 200 m (656 ft) apart. At each transect, there were four trap points, each with a minnow and crayfish trap. However, instead of placing the trap sites at fixed locations in the habitat as in 2002, they were placed at fixed depths and moved with the water level. When the water was rising or remaining stationary, each transect had four trap points located at 15, 30, 45, and 60 cm ( 6, 12, 18, 24 in) deep (Figure 4-2). During falling water levels, the trap points were placed at 30, 45, 60, and 75 cm (12, 18, 24, and 30 in) deep, except when falling lake stages were not predicted. The traps were still checked once weekly, and at each sampling occasion, the traps were moved to the appropriate depth. This result ed in trapping animals at sp ecific depth ranges over the course of the week, with similar water depths between sample occasions. For example if a trap point was located at the 30 cm (12 in) depth and the water level rose several centimeters during the week, the trap was cons idered to be sampling the 30-45 cm (12-18

PAGE 78

67 in) water depth. For falling wa ter levels, the 30 cm (12 in) traps were sampling the 15-30 cm (6-12 in) depth. Percent cover of vegeta tion species was also estimated for a 2 m (6.5 ft) radius around every trap site on each sampling occasion. Continuous sampling was conducted from 1/30/2003 to 1/5/2004, during which time traps were located up in the shallow grassy habitat at high water levels (characterized mainly by Luziola fluitans and Panicum repens ), through the thick emergent habitat (with Pontederia cordata and Typha domingensis ), down to more open water zones (with Hydrilla verticillata and floating leaf species Nuphar luteum and Nymphaea odorata ) at lower water levels. Besides the added environmental variables, this new protocol also allowed us to samp le year-round, instead of having to remove the traps at moderately low water levels. Sampling ended on 1/5/2004 at about 15.5 m (50.8 ft) NGVD, with lake stage droppi ng due to the 2003 drawdown. Analysis Methods Population Estimates and Movement for Herpetofaunal Species Trapping grids are known to exhibit “edge eff ects” due to animals near the edges of the grid moving in and out of the sample d area. To account for this phenomenon a boundary strip is typically estimated and added to the grid area to estimate an effective sampling area. Wilson and Anderson (1985) propose using the mean maximum distance moved (MMDM) by animals recaptured at least once to estimate the activity range of the species. Although lacking a solid theoretica l explanation, this method works well in simulations. The alternative method, nested grid design, requires a large data set for estimation of density (Williams et al. 2002). Since our data were fairly sparse with recaptures, we used the MMDM method to calc ulate the effective areas of the trapping grids for each species.

PAGE 79

68 Low numbers of recaptures were attained fo r each grid; so in or der to calculate the MMDM for each species movement distances were pooled from all grids and the trapping web. To calculate the diagonal distances within the trapping grids, the Pythagorean theorem was used: [a2+b2=c2], where sides a and b are sides of known lengths. For diagonal distan ces in the trapping web, [a2=b2+c2-2bc(cosA)] was used, where A is the degree measure between side s b and c of known length (Larson et al. 1994). For each species, MMDM and its variance were calculated using formulas from Wilson and Anderson (1985). The widths of th e boundary strips were estimated as half the MMDM for each species. The effective grid areas and associated variances were then calculated for each grid per species (Wilson and Anderson 1985). Program MARK (White and Burnham 1999) was used to estimate population sizes for each species per trapping grid. Estimates were obtained using models M(o), (constant capture probabilities), and M( t), (time dependent capture pr obabilities). The AkaikeÂ’s Information Criteria (AIC) were compared to determine which model best fit the small data set. Density was then calculated for each grid per species by dividing the population size by the effective sampling area (Wilson and Anderson 1985). Capture Success for Focal Species Basic analyses were conducted on the 2003 data to look for trends in the data associated with the main sampling variables involved in this protocol. All trap points were divided into groups according to vegetation community (see results section below for descriptions) and water depth at each tr ap location. Ten species were examined, two each of salamanders ( Siren spp ., A. means ), snakes (both Nerodia spp.), turtles ( K. baurii and S. odoratus ), frogs (both Rana spp.) and fish ( H. littorale and M. salmoides ). The reptile and amphibian pairs represent species with similar life history traits but which

PAGE 80

69 have slightly different habitat requirements or preferences. They were also the most frequently captured reptile and amphibian sp ecies in this study. The two fish species were chosen to characterize oppos ite ends of the spectrum of habitat selection. While the armored catfish is a generalist species with great tolerance for low dissolved oxygen, high temperatures, thick vegetation and other extr eme environmental variables (e.g., Nico and Fuller 1999), largemouth bass and other sunfishe s are thought to be highly intolerant of these same habitat characteristic s (e.g., Allen and Tugend 2002). Capture success was calculated as the num ber of captures per species divided by the total number of trap points for the particular variable of interest. This usually resulted in a very low frequency, due to the large num ber of trap points and low detectability of species. An arcsine squareroot transformation was applied to all success values, in order to spread the ends of the scale, whil e improving normality for the proportion data (McCune and Grace 2002). This allowed the relative values to s how up more clearly while reducing the effect of la rge sample units. The assumpti on of equal detectability of the different species between habitats or wa ter depths may be violated, but detection probabilities cannot be calculated for this particular analysis. However, uniform sampling methods were used over space (trans ects) and time (sampling occasions) to reduce variability in detection. For this sampling protocol, dependence of trap placement upon water depth (i.e. lake stage) resulted in unequal sampling for vegetation communities. In addition, the vast number of trap sites (n= 3,426) and spar se nature of the data made multivariate analyses virtually impossible. For example, dividing data into groups (either subjectively or with cluster analysis) depending on sa mple occasions, water depth or vegetation

PAGE 81

70 communities would neglect important differences in the other variables and/or result in groups of vastly unequal numbers of trap poi nts. Attempted NMS analyses of all trap sites (ungrouped) yielded no re sults due to the great number of zeroes in the matrices. The data were not even appr opriate for most univariate analyses. For example, chisquare analyses of capture success would indicate whether there were significant differences in the counts of focal species between vegetation type s or water depths, however the large sample sizes invariably lead to significant differences. Repeatedmeasures analysis of variance was considered to test the differences between water depths over time, however the data were t oo sparse to divide the counts between both sample occasions and water depths. Species richness was also inestimable because of the frequent nondetection of sp ecies and unequal sample si zes. As a result of the complicated nature of the data, the analyses were largely descriptive in nature. These descriptions of habitat usage rely mainly on comparisons of capt ure success across two categorical environmental variables: water depth and vegetation community. Results Population Estimates for Herpetofaunal Species Each species showed different movement distances over the sampling grids and web. Figure 4-3 shows the mean maximum distances traveled and variances, along with the associated widths and variances of the boundary strips for each species. Maximum distances traveled for A. means Siren spp., N. floridana and K. baurii were 18 m (59 ft), 24 m (79 ft), 34 m (112 ft), and 59 m (194 ft) respectively. These m ovements are fairly large relative to the sizes of the grids, with 72 m (236 ft) being the absolute maximum distance between any two traps during all samp ling. This indicates that the assumption of closure was violated. Table 4-1 contains th e percent increase in the size of each grid

PAGE 82

71 when the boundary widths for each species were added to the sizes of the grids. While the effective sampling areas are fairly acceptable for the amphiumas and sirens (Wilson and Anderson 1985), Florida green water snak es and striped mud turtles add excessive area to the original sampling areas. After recogn izing the fact that closure was violated in these grids, population sizes and densities we re estimated with unreliable accuracy, but a best attempt was made given the data. Estimates of population size and variances are shown in Table 4-1. Several times the capture history data were so sparse that estimates could not be calculated for some grids and species. The estimates generated with sufficient data often have large variances due to low recapture probabilities. The null model of no variation in detection Table 4-1. Grid sizes a nd population estimates by mark recapture methods. Parameter AMPME SIREN NERFL KINBA GRID1 Actual size of grid (ha) 0.0324 0.0324 0.0324 0.0324 Est. effective sampling area and variance (ha)0.078 (0.40)0.08 (0.93) 0.12 (1.79)0.16 (9.17) Percent of original grid (%) 239 264 376 493 Est. population size and variance (# indivs.) n/a n/a 16 (184) 7 (30) Density estimate (#/ha) n/a n/a 131 44 GRID2A Actual size of gr id (ha) 0.2025 0.2025 0.2025 0.2025 Est. effective sampling area and variance (ha)0.30 (1.66)0.32 (3.71) 0.39 (6.08)0.46 (27.83) Percent of original grid (%) 149 157 192 225 Est. population size and variance (# indivs.) n/a n/a n/a n/a Density estimate (#/ha) n/a n/a n/a n/a GRID2B Actual size of gr id (ha) 0.2025 0.2025 0.2025 0.2025 Est. effective sampling area and variance (ha)0.30 (1.66)0.32 (3.71) 0.39 (6.08)0.46 (27.83) Percent of original grid (%) 149 157 192 225 Est. population size and variance (# indivs.) 14 (37) n/a 26 (99) 9 (5) Density estimate (#/ha) 49 n/a 68 20 GRID3 Actual size of grid (ha) 0.0729 0.0729 0.0729 0.0729 Est. effective sampling area and variance (ha)0.14 (0.72)0.15 (1.65) 0.19 (2.94)0.24 (14.27) Percent of original grid (%) 187 201 267 332 Est. population size and variance (# indivs.) 19 (117) n/a 25 (24) 26 (525) Density estimate (#/ha) 142 n/a 133 108

PAGE 83

72 probability was always selected over the tim e-varying capture probability model. Density estimates based on these abunda nces are also included in Table 4-1. Capture success for Focal Species Figure 4-4 shows the types of vegeta tion communities that were sampled throughout 2003. Each sampling occasion corresponds to one weekly sample, which includes 72 trap points (4 trap points for each of 18 transects). O ccasionally there were less than 72 samples for a given sample occa sion, usually because data recording for a sample was inadvertently neglected, traps went unchecked due to dangerous weather, or traps were missing. Each trap point was s ubjectively categorized in the field into different vegetation communities, based on the dominant species present. The “Grass” community is the closest shor eward, and is characterized by Panicum repens, Luziola fluitans, Juncus effusus, and Eleocharis spp. Lakeward from this is the “Rooted-HE”, which refers to the herbaceous emergent species, especially Pontederia cordata and Typha domingensis The community termed “G/HE” is the border of the grass and rooted herbaceous emergent zones, which ha d enough samples to be a separate category. “Floating-HE” is the floating mat community, consisting of P. cordata, Bidens spp., Ludwigia leptocarpa and a variety of other species. Out past the herbaceous emergent communities are the deeper “Outward” communities. These include floating-leaf emergents ( Nelumbo lutea, Nymphaea odorata, and Nuphar luteum ), submersed plants ( Hydrilla verticillata ), and deep emergents ( Paspalidium geminatum ). Trap points that were on the borders between distinct ve getation communities, or were part of communities with too few samples to have its own category, were classified as “Mixed”. Since transects were randomly chosen and the trap points were moved with the water level, there was no way to collect equal num bers of samples from each community.

PAGE 84

73 Figures 4-5 and 4-6 show results fr om capture success comparisons for salamanders. Both species tend to be capture d more frequently in the outer herbaceous emergent vegetation communities and community edges, as well as greater water depths. Pig frogs seemed to show a preference fo r grassy habitats and community edges, decreasing towards more outward vegetation communities, while leopard frogs did not show much of a pattern (Figure 4-7). Howe ver, much stronger trends appeared with water depth (Figure 4-8). Both species, espe cially the leopard frogs, showed an inverse relationship to water depth. The two snake species showed divergent trends in capture rate. In Figure 4-9, Florida green water snakes ( Nerodia floridana ) appeared more in the rooted pickerelweed communities, while the Florida water snakes ( Nerodia fasciata pictiventris ) did not have quite as strong a tendency to be in specific habitats. Water depth seemed to be more important in Florida water snake occurrence (Figure 4-10), with most being trapped in shallow water and decreasing steadily with dept h. The Florida green water snake did not have such a trend. Figures 4-11 and 4-12 show that common musk turtles were captured more frequently in vegetation habitats furt hest from shore, as well as deeper water depths. Striped mud turtles on the other ha nd did not have strong trends, but peak in rooted pickerelweed habitats and intermediate water depths. Largemouth bass did not show a strong affi nity for any certain habitat, however they were captured slightly more frequently in rooted herbaceous emergent and borderline communities (Figure 4-13). They also appeared most in intermediate water depths, (Figure 4-14), e.g.45-60 cm (18-24 in) deep. Bass ca ptured in the traps were juvenile fish, with total lengths in 2003 ra nging from 5.1-14.0 cm (2-5.5 in), (n=60).

PAGE 85

74 The armored catfish were found mostly in bo rder vegetation communities (Figure 4-13). They also were positively correlate d with water depth (Figure 4-14). Discussion Population Estimates for Herpetofaunal Species The small activity ranges estimated for the amphiumas and sirens were similar to what have been found in other studies (Gehlbach and Kennedy 1978, Sorensen 2004). Maximum distances were higher in this case, (18 m (59 ft) vs. 5 m (16 ft) for amphiumas and 24 m (79 ft) vs. 10 m (33 ft) for greater sirens (Sorensen 2004)). These estimates suggest that lack of movement in these an imals make them susceptible to mortality during muck removal operations. The sizes of the grids were proba bly not large enough to make reasonable activity range estimates fo r the Florida green water snake or striped mud turtle. Bancroft et al (1983) documented a Florida green water snake moving 223 m (731 ft) in less than two hours. As another example, Nerodia taxispilota (brown water snakes) have been documented moving distances greater than 1 km ( 0.62 miles) (Mills et al. 1995). Mahmoud (1969) found maximum distan ces for several species of kinosternid turtles, including 525.5 m (1,723.6 ft) for S. odoratus 435.3 m (1,427.8 ft) for Kinosternon flavescens (yellow mud turtle), 408.4 m (1,340 ft) for Kinosternon subrubrum (Mississippi mud turtle ), and 93.9 m (308 ft) for Sternotherus carinatus (Mississippi musk turtle). Due to low recapture rates and large movements of individuals, poor density estimates were attained with these protocols. Even when the simplest models were utilized to estimate population sizes (the estimated number of animals with no account of area sampled), variances were unacceptably high. Even if the estimates had been reasonable, the study would have been limite d to a small number of species, a narrow

PAGE 86

75 window of opportunity when the pickerelweed zone was completely inundated with water, and non-random locations which po ssessed a wide enough band of habitat to contain the grids within a rela tively homogeneous habitat. Therefore, the mark-recapture grids will not be utilized for post-enhancement sampling. Capture Success for Focal Species Vegetation communities offer different tradeoffs to animal species that inhabit them. Variations in predator efficiency, pr ey type and abundance, or abiotic properties associated with dissimilar macrophyte types strongly determine their use to aquatic vertebrates (Miranda et al. 2000) Physical properties of plan t species such as branching, leaf shape and number, plant biomass and pos ition throughout the water column affect animalsÂ’ ability to maneuver in the habitat, as well as phys ical and chemical properties such as dissolved oxygen, nutrient levels, water temperature, light pe netration and current (Chick and McIvor 1994). Welch (2004) c onducted a thorough ecol ogical investigation into the vegetation communities of Lake T ohopekaliga prior to the 2004 enhancement. One finding was that the soils associated with the intermediate littoral zone depths and Pontederia cordata communities were highly organic and low in bulk density. On the other hand, the shallow grassy communities and the various deeper water communities had soils higher in bulk density, and theref ore were sandier in composition. Substrate alone may provide benefits or detriments to animal species, dependi ng on their specific life history traits. Water depth at a given location has st rong influences on vertebrate species distribution and habitat use. It is the main determinant of the boundaries of the littoral ecotone, limiting emergent aquatic vegetati on growth to the limits of the water fluctuation. The gradual slope of the shoreline causes sma ll increases or decreases in

PAGE 87

76 lake stage to flood or dry out broad expanses ha bitat, altering its use to different species. Water depth also establishes th e volume of water that aqua tic organisms have to move through, and can provide enough space for the pr esence of a thermocline (Miranda et al. 2000). The aquatic salamander species are known for burrowing in the organic sediment associated with dense vegetation as refuge from predators and drought (Etheridge 1990, Conant and Collins 1998). In fact many of these large salamanders have been uncovered during muck removal operations around Flor ida, even when there was water still covering deeper areas of the lake (Aresco 2001) The findings of this study indicate the same pattern, with highest capture success occurring in densely vegetated communities. As mentioned previously, these communities are also most associated with low bulk density and high organic composition of the so ils (Welch 2004). This indicates that not only may the dense vegetation provide ample fo rage and cover for these creatures, but also the organic sediment (muck) is prefe rred for burrowing. While deeper water depths yielded more sirens and amphiumas, it is possi ble that they are si mply more active in deeper water, increasing de tection probabilities. Higher capture success in shallow water si tes was expected for leopard frogs, since all individuals captured in 2002 occurred in Ap ril and November when water levels were low. This species is known to travel rela tively far from aquatic habitats, given proper cover and shade from terrestrial vegetati on, depending on soil moisture and dew to prevent desiccation (Dole 1965, Conant a nd Collins 1998). Leopard frogs are particularly dense in herbaceous vegetation ar ound lakes, with plenty of protection and food sources in the grasses (Kilby 1936). On the other hand, pig fr ogs are highly aquatic,

PAGE 88

77 often being associated with emergent and floating vegetation (Conant and Collins 1998). Pig frogs were captured the most in the “m ixed” vegetation community. This category was mainly represented by floating vegetati on (water hyacinth) and borders between communities, (mainly between the floating mats of emergent vegetation and submersed vegetation). The dominance of pig frogs found in this community may indicate that border communities provide a tradeoff between predator avoidance and prey availability. Decreases in capture success with increased water depth and relative distance from shore indicate that water depth was very influentia l in determining the presence of both species of ranid frogs. The leopard frog in particular has a very strong decreasing trend with water depth, which is consistent with its more terrestrial nature. Water depth was also an important factor in the Florida water snakes’ habitat preference, which was expected since in th e 2002 sampling most i ndividuals were caught in April and November, both during relativel y low water periods. Water depth does not seem to have as strong an influence on Florid a green water snakes. To illustrate this difference, Seigel et al. (1995) found that during a three-year dr ought in Ellenton Bay, South Carolina, many N. fasciata left the habitat only seven days after it dried out, while N. floridana never left in large numbers. While the abundance of snakes was generally lower, it was five years before N. floridana was captured after the drought. As discussed previously, water snakes typically show little site fidelity and are capable of long-range movements (Bancroft et al. 1983, Mills et al 1995). Several of the species show ontogenetic niche shifts with ag e and size, often changing diet and habitat preference at a certain size (Mushinsky et al. 1982, Mushinsky and Miller 1993). Fl orida water snakes in particular are known for feeding on fish when young, and then at 50 cm (20 in) snout-

PAGE 89

78 vent length they begin to feed almost ex clusively on frogs (Mushi nsky et al. 1982). The traps used in this study mainly catch adult sn akes, so prey (i.e. a nuran) availability in shallow habitats may result in this species preference for shallower water depths. Alternatively, Florida green water snakes ar e not so specialized, being caught with and regurgitating a variety of fish, frogs, and ev en large sirens. While both species were captured more frequently in the emergent vegetation communities, the water depth seems to influence the presence of these species the most. The turtle species have different life hi story traits that may explain observed differences in capture success. For exampl e, common musk turtles are highly aquatic, rarely leaving the wate r except to nest. When water le vels drop, at least in ponds, they follow the water down and then burrow into th e sediment to avoid desiccation (Wygoda 1979, Gibbons et al. 1983). Ma hmoud (1969) suggests that Sternotherus spp are more dependent upon water depth than Kinosternon spp. The former appears to prefer water depths greater than 30 cm and have been found in up to seven meters of water. As shown in this study, there is a sharp increase in capture success associated with both lakeward vegetation communities and deeper water depths Alternatively, striped mud turtles are much more terrestrial, usually dispersing over land during drought or heavy rainfall to find alternative habitats (B ennet 1972, Wygoda 1979). On th e other hand, Gibbons et al. (1983) found that Kinosternon subrubrum experienced no increased emigration due to drought in Ellenton Bay, South Carolina, since it is a fairly terrestrial species and is not negatively affected by dry conditions. Maximu m activity of striped mud turtles occurred in 15 cm in Oklahoma (Mahmoud 1969). In this study, peak captures occur in rooted

PAGE 90

79 emergent vegetation and intermediate water dept hs. All species of ki nosternid turtles are thought to prefer vegetated habitats to unvegetated ones (Mahmoud 1969). The captures of young largemouth bass in a ll parts of the littoral habitat were contrary to common fisherie s doctrine. One would expect them to occur almost exclusively in the open water/submersed habi tats, due to physicochemical requirements, as well as the physical barrier of the organic berm formed by the floating vegetation mats (Moyer et al. 1995, Allen and Tugend 2002, Allen et al. 2003). However, relatively high water levels evidently allow young bass and ot her centrarchid species to enter grassy habitats, as well as inhabit the pickerelweed zone. Mira nda et al. (2000) described vegetated aquatic habitats as a mosaic of microhabitats within larger seemingly inhospitable macrophyte stands. Although from a humanÂ’s pe rspective the habitat may seem uniformly unsuitable, fish can move bot h horizontally and vertically to find pockets of suitable physical (e.g., temperature) and chemical (e.g., dissolved oxygen) water conditions for survival. Perhaps this explai ns the bassÂ’ ability to move through this landscape relatively unscathed. The armo red catfish on the other hand, have a high tolerance for poor water quality due to their ab ility to breathe air (B rauner et al. 1995). They were still found at deeper water de pths, but his was likely due to breeding requirements in the littoral zone for greater than 0.3 m water depths (Hostache and Mol 1998).

PAGE 91

80 60 cm 45 cm 30 cm 15 cm 60 cm 45 cm 30 cm 15 cm Figure 4-1. Locations of 2003 sampli ng transects in Lake Tohopekaliga.

PAGE 92

81 15 cm 30 cm 45 cm Crayfish traps are positioned on the substrate Minnow traps float on surface of water 60 cm 15 cm 30 cm 45 cm Crayfish traps are positioned on the substrate Minnow traps float on surface of water 60 cm 15 cm 30 cm 45 cm Crayfish traps are positioned on the substrate Minnow traps float on surface of water 60 cm Figure 4-2. Diagram of weekly trap placement at specified depths. Distance (m) 0510152025303540455055606570 Amphiuma Siren Fl.green water snake Striped mud turtle Maximum mean distance moved and variance Maximum distances moved per species Figure 4-3. Mean maximum distances travel ed with variances and maximum distances, based on results of mark-recapture sampling.

PAGE 93

82 Sample Occasion 024681012141618202224262830323436384042444648 Number of Each Vegetation Community 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 GRASS (n=799) G/HE (n=230) ROOTED-HE (n=1174) FLOATING-HE (n=714) MIXED (n=162) OUTWARD (n=347) Figure 4-4. Number of trap sites sample d in each vegetation community per sample occasion.

PAGE 94

83 Vegetation Community GG/HER-HEF-HEMO Arcsine Squareroot Transformed (#captures/# trap points) 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 SIREN AMPHIUMA Figure 4-5. Salamander capture success by vegetation community. Water Depth (cm) 0 15 30 45 60 Arcsine Squareroot Transformed (# captures/ # trap points) 0.00 0.05 0.10 0.15 0.20 SIREN AMPHIUMA Figure 4-6. Salamander capture success by water depth.

PAGE 95

84 Vegetation Community GG/HER-HEF-HEMO Arcsine Squareroot Transformed (# captures/ # trap points) 0.00 0.05 0.10 0.15 0.20 0.25 PIG FROG SOUTHERN LEOPARD FROG Figure 4-7. Frog capture su ccess by vegetation community. Water Depths (cm) 0 15 30 45 60 Arcsine Squareroot Transformed (# captures/ # trap points) 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 PIG FROG SOUTHERN LEOPARD FROG Figure 4-8. Frog capture success by water depth.

PAGE 96

85 Vegetation Communities GG/HER-HEF-HEMO Arcsine Squareroot Transformed (# captures/ # trap points) 0.00 0.05 0.10 0.15 0.20 0.25 FLORIDA GREEN WATER SNAKE FLORIDA WATER SNAKE Figure 4-9. Snake capture success by vegetation community. Water Depth (cm) 0 15 30 45 60 Arcsine Squareroot Transformed (# captures/ # trap points) 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 FLORIDA GREEN WATER SNAKE FLORIDA WATER SNAKE Figure 4-10. Snake capture success by water depth.

PAGE 97

86 Vegetation Communities GG/HER-HEF-HEMO Arcsine Squareroot Transformed (# capture/ #trap points) 0.00 0.05 0.10 0.15 0.20 0.25 0.30 STRIPED MUD TURTLE COMMON MUSK TURTLE Figure 4-11. Turtle capture success by vegetation community. Water Depths (cm) 0 15 30 45 60 Arcsine Squareroot Transformed (# captures/ # trap points) 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 STRIPED MUD TURTLE COMMON MUSK TURTLE Figure 4-12. Turtle captu re success by water depth.

PAGE 98

87 Vegetation Communities GG/HER-HEF-HEMO Arcsine Squareroot Transformed (# captures/ # trap points) 0.00 0.05 0.10 0.15 0.20 LARGEMOUTH BASS ARMORED CATFISH Figure 4-13. Fish capture su ccess by vegetation community. Water Depths (cm) 0 15 30 45 60 Arcsine Squareroot Transformed (# captures/ # trap points) 0.00 0.02 0.04 0.06 0.08 0.10 0.12 LARGEMOUTH BASS ARMORED CATFISH Figure 4-14. Fish capture success by water depth.

PAGE 99

88 CHAPTER 5 SUMMARY AND CONCLUSIONS Review of Aquatic Vertebrate Communi ty Dynamics in Lake Tohopekaliga This study has documented the conditi ons of the fish and herpetofaunal communities in the littoral zone of Lake Tohopekaliga prior to extreme habitat modifications performed in 2004. The littoral zone in this eutrophic lake defined here includes the entire vegetated shoreline, fr om the grassy vegetation community on the shore that is occasionally inundated with water to the lakeward band of emergent vegetation that is always floode d. Animals captured in this ha bitat represent species that are dependent in some way upon its unique ch aracteristics for their survival, including still water, cover from predators or light, nest ing substrate, organic sediment in which to burrow or forage, and appropriate prey. Perh aps the most useful information from this study so far has been the species list, which includes juvenile centrarchids and exotic catfish. While it is no surprise to find most of the herpetofaunal species in this type of landscape, the quantity and quality of fish cap tured was not expected. Previous research indicates that sunfish cannot and do not inhabit heavily vegetated littoral habitats, while the two species of exotic catfish captured have not been documented at all this far north in Florida. Species such as warmouths, bl uespotted sunfish, Florida green water snakes, sirens, striped mud turtles, and pig frogs were found in all parts of the habitat throughout the year, indicating that they may be threatened by the habitat alterations more than other species.

PAGE 100

89 Research efforts in 2002 were focused on the Pontederia cordata zone only, which was the species most considered a nuisance by lake managers. The widespread removal of this species threatens to disrupt the wildli fe community structure in a large area of the lake, making research of this specific ve getation community a n ecessary component of the study. In 2003, sampling was conducted ac ross vegetation communities within the littoral zone in an attempt to understand how different fo cal species are distributed throughout the habitat. The following secti ons will provide a revi ew of the observed effects of these environmental variables on species richness, community composition and distribution of focal species. Air Temperature Average air temperatures over the course of each sample occasion accounted for much variation in the aquatic vertebrate co mmunity. Both fish and total vertebrate richness were negatively correlated with average air temperature over time. NMS analyses also indicated that this variable explained large percentages of variance in the species composition of the fish and vertebra te communities. Vertebrate assemblages represented by Regina alleni and Esox spp. were associated w ith high air temperatures, while groups with Nerodia fasciata pictiventris and Rana sphenocephala as indicator species were found at lowe r air temperatures. Lake Stage Only herpetofaunal species richness was found to decline with increased lake stages. This variable expl ained much of the variation in fish, herpetofaunal, and combined vertebrate assemblages. In periods of high lake stages, Sternotherus odoratus and Esox spp. were the indicator species fo r sample occasion clusters. N. fasciata

PAGE 101

90 pictiventris and R. sphenocephala characterized assemblages occurring with low lake stages. Water Depth Water depths from 0-75 cm were samp led for aquatic vertebrates. Sirens, amphiumas, armored catfish and especially common musk turtles showed increasing capture success with water depth. On the ot her hand, pig frogs, leopard frogs and Florida water snakes showed strong decreases in capture success as depth increased. The main drawback to this sampling protocol was th e limiting size of the cr ayfish traps, with restricted sampling beyond 75 cm deep. Species occurrence beyond this depth would have provided valuable information that would be relevant to potenti al responses to the lake drawdown and scraping activities. Vegetation Community Both salamander species tended to be captu red mostly in the emergent vegetation, but probably had a stronger association with the decreased bulk density in the soils of these vegetation communities. Pig frogs were found to occupy the more shoreward vegetations disproportionately. Florida green water snakes were more often found in the rooted grasses and pickerelweed than in th e outermost communities. While striped mud turtles peaked in the rooted emergent vege tation zone, common musk turtle increased in frequency in the more lakeward communities. Population Size Estimates Due to unsuitable sizes and trap spacing of the grids and web, large movements and low recapture rates of the focal species resu lted in poor population size estimates. Large variances indicate the unreliability of the result s. This attempt illustrates the difficulty of using one method to sample several different species that have di ssimilar life history

PAGE 102

91 traits. Even if reasonable estimates had result ed from these pilot studies, the limitation of the protocol to only four species and such a specific habitat ( P. cordata zone while inundated with water) narrows the focus of the study beyond much use to lake managers. Lake Tohopekaliga Habitat Enhancement The fall 2003 drawdown has been successfu lly implemented and an estimated 7.3 million cubic meters of muck and vegetation have been removed (scraped) from the littoral zone. Bulldozers and dump trucks re moved a total area of 1,351 ha of habitat (Florida Fish and Wildlife Conservation Commissi on 2004). Repercussions or benefits to the various wildlife guilds inhabitin g the lake have yet to be determined. As noted in this study, there are multiple species that utili ze the heavily vegetated littoral habitat throughout the year, whether by preference or necessity. While there remain intact stretches of habitat that have not been scra ped, the majority of the shoreline has been radically altered. The main reason for the extreme nature of this project was to attain maximum benefits to the largemouth bass population for the longest period of time possible. High costs associated with the enhancement procedures prohibit frequent drawdowns and scraping efforts. Due to these factors, managers feel pressure to remove the most vegetation and muck as possible at one time in order to produce long-term benefits (Allen and Tugend 2002). If it were not for monetary restrictions, perhaps alternatives such as limiting removal of mu ck and vegetation to a smaller portion of the shoreline, or using mechanical removal as a method of increasing the patchiness of the vegetated area would be prefe rred. In these cases, adult bass would still have access to shallow spawning sites, while species requi ring more complex habitats would not be displaced.

PAGE 103

92 Direct mortality will doubtless cause change s in the abundance of several species, most notably for the sirens and amphiuma s. During just a few hours of bulldozer activity, several dozen sirens and amphiumas and one pig frog were unearthed from beneath the vegetation (personal observation) The machinery also crushed one large snapping turtle in the dry pickerelweed z one. Scraping was conducted from the grassy vegetation communities out into the floati ng leaf/submersed vegetation communities, removing virtually all of the emergent vegeta tion across the ecotonal gradient. The depth of sediment removal was established by white sand substrate and absence of all root structures. Sediment and ve getation was piled into large windrows and subsequently loaded into dump trucks and deposited in uplan d disposal areas or in -lake spoil islands. Although most salamanders that were aestiva ting beneath the soil surface seemed to survive the initial pass with the equipment, most probably were crushed by the weight of the debris or physically removed from the lake altogether. Bulldozer operators reported numerous small turtles being uncovered, in addition to the “muck eels” (i.e., large salamanders). In agreemen t with this study, Aresco ( 2001) also reported aestivating sirens and amphiumas uncovered by bulldozer operations in Lake Jackson, Leon County, Florida, even though open-water ha bitat was still present past the vegetated zone. Due to the terrestrial nature of K. baurii, N. fascia ta pictiventris, and R. sphenocephala, and the highly aquatic characteristics of S. odoratus these species may not have been in the littoral zone of Lake Tohope kaliga during the drawdown and may have experienced less mortality associated with the bulldozers. In addition to mortality due to enhancem ent operations, ensuing lack of suitable habitat will further alter vertebrate communities. A suite of herbicides is being utilized to

PAGE 104

93 stop natural vegetation succession so that managers may selectively allow regrowth of specific species at subjectively desirable (i.e. more sparse) de nsities. As a result, at any point in the former littoral zone, the ha bitat will have underg one several physical alterations. Complete absence of emergent vegetation and or ganic sediment will prohibit recolonization by many of the herpetofaunal species examin ed in this study. For example, species such as sirens, amphiumas, striped mud turtles, and Florida green water snakes will likely avoid the enhanced shorel ines for these reasons. Juvenile sunfish may also limit their use of this habitat due to lack of prey and cover. Bancroft et al. (1983) found that Florida green water snakes, greater sirens, southe rn leopard frogs, striped mud turtles and common musk turtles were nega tively impacted by vegetation loss resulting from grass carp introductions and shoreline development in Lake Conway, Florida. Alternatively, other species such as amphiumas and Florida water snakes did not seem to have adverse reactions to th e resulting loss of vegetation and increase in sandy beach shoreline. In addition to the obvious lack of muck and vegetation (except for the exotic hydrilla), several more subtle changes will have taken place. For any given lake stage, any affected location will have deeper water due to the removal of the sediment and root structure. As seen in this study, increased water depths are associated with lower herpetofaunal species richness. In particular aquatic frogs and Flor ida water snakes may be less inclined to inhabit these deeper habitats, while the co mmon musk turtle may increase presence. Pelagic fish species may replace juvenile or more littoral zone fish species. Increased wave action and water cu rrent will result from vegetation removal in the shoreline habitat. In these large lakes, strong wi nds and thunderstorms blowing

PAGE 105

94 across the open water often cause large wa ves and whitecaps. Light penetration and water temperature will also increase due to vegetation loss. This may influence fish species richness, which was show n to decrease with increased air temperatures in this study. Fish species with higher environmenta l tolerances such as H. littorale, Esox spp., and Lepisosteus spp. may increase in abundance relative to Lepomis spp. Most of these factors will not only impact ad ult habitat preference for severa l native species of fish and herpetofauna, but will certai nly have strong effects on nes ting potential for fish and amphibians, several species of which require still water and/or emergent vegetation stems. Even though this expansive vegetated habitat is a result of anthropogenic eutrophication in central Florid a lakes, it may provide an a lternative habitat for species affected by decades of extensive wetland de struction. Wetland isolation resulting from human modifications reduces co rridor travel between habita ts and abundances for species such as sirens that rely on wetlands fo r survival (Snodgrass et al. 1999). Through the previous century, channelization of natura l streams, wetland drainage to improve pastureland, and water level stabilization within the Upper Kissimmee Basin have eliminated and fragmented many habitats used by wetland animal species. As a consequence of these landscape manipulations the lush, vegetated, lake littoral zones may provide refuge for displaced aquatic verteb rates in this area, regardless of the fact that the eutrophic lake edges are not natural features. There is also the concern that the extr eme techniques used during the enhancement project were unnecessary and based upon false suppositions. Following a similar drawdown and muck-scraping project in 1995 in Lake Kissimmee, Flor ida, Allen et al.

PAGE 106

95 (2003) found no increase in harvestable largem outh bass or angler catch rate in the following six years compared to pre-enhancemen t. Due to the lack of response from the adult bass population, the rese archers concluded that ma nagers should view lake enhancement as a way to improve recreational (i .e. fishing boat) access in the lake rather than setting the stated goal of improving bass fishing. Mi randa and Dibble (2002) stated the need for fishery scientists to focus at al l organizational levels to properly manage for bass. Relying on just the individual and population response is not enough to understand and manipulate the behavior of the species or manage an ecosystem. Ecological interactions at the community and ecosystem levels must also be considered before single-species management techniques are impos ed on a habitat, since there are important interactions between other sp ecies and the abiotic environment that may influence the success of the species. Future Aquatic Vertebrate Monitoring Plans While this study has provided baseline information on species presence and identified the influence of temporally and spatially changing environmental variables on the vertebrate communities, it is only the be ginning of the necessary research. The continuation of the current project on Lake Tohopekaliga will provide missing information needed to determine the changes in the wildlife communities. In late 2004, pending suitable water depths in the littoral z one, traps will be redeployed at the original transects sampled in 2002. The fixed-location trapping protocol will be implemented and will continue indefinitely. Comparisons will be made in site occupancy and temporal community utilization of this habitat, be tween the preand post-enhancement littoral zone. Although the spatial sampling protoc ol (used in 2003) provided information on habitat use beyond the pickerelweed zone, it will not be continued. The dependence of

PAGE 107

96 trap locations on lake stage results in little investigator control over sampling effort in different vegetation communities, and therefor e is limited in its use as a long-term sampling technique. In addition to whole-lake sampling, three study areas have been defined which each contained four 400 m long sections that underwent different treatments, including scraping and herbicide application, only scrapi ng, only herbicide application and control. Since the mark-recapture grids were unsucce ssful in attaining reliable population estimates, an alternative sampling protocol will be used in these study plots. Fixed transects similar to the whole-lake transects will be laid out in the 12 plots. Sampling will take place in the spring and fall seasons when lake stages will allow sufficient water in the habitat. In both seasons traps will be checked for 3 months, yielding 12 weekly sampling occasions. Captured individuals of the focal herpetofaunal species will be PIT tagged to determine movement patterns within and among treatment plots. With this protocol repeated samples in the same locations will allow estimation of detection probabilities, facilitating estimation of site occupancy and community diversity measures. Assemblage composition and association with measured environmen tal variables will also be analyzed. Comparisons of the aquatic vertebrate communities between treatment plots will reveal effects of the vegetation and sediment remo val by herbicidal treatments and mechanical removal.

PAGE 108

97 LIST OF REFERENCES Alford, R. A., and M. L. Crump. 1982. Habitat partitioning among size classes of larval Southern leopard frogs, Rana utricularia Copeia 1982:367-373. Allen, M. S., and K. I. Tugend. 2002. Eff ects of a large-scale habitat enhancement project on habitat quality for age-0 largem outh bass at Lake Kissimmee, Florida. Pages 265-276 in D. P. Philipp and M. S. Ri dgeway. Black bass: ecology, conservation, and management. Americ an Fisheries Society, Symposium 31, Bethesda, Maryland, USA. Allen, M. S., K. I. Tugend, and M. J. Mann. 2003. Largemouth bass abundance and angler catch rates following a habitat enhancement project at Lake Kissimmee, Florida. North American Journa l of Fisheries Management 23:845-855. Anderson, D. R., K. P. Burnham, G. C. White and D. L. Otis. 1983. Density estimation of small-mammal populations using a trapping web and distance sampling methods. Ecology 64:674-680. Aresco, M. J. 2001. Siren lacertina (greater siren). Aestivation chamber. Herpetological Review 32:32-33. Bancroft, G. T., S. J. Godley, D. T. Gross, N. N. Rojas, and D. A. Sutphen. 1983. Largescale operations management test of use of the white amur for control of problem aquatic plants; the herpetof auna of Lake Conway: spec ies accounts: miscellaneous papers A-85-3, U.S. Army Engineer Waterways Experiment Station, CE, Vicksburg, MS. Bennet, D. H. 1972. Notes on the terre strial wintering of mud turtles ( Kinosternon subrubrum ). Herpetologica 28:245-247. Billard, R. 1996. Reproduction of pike: ga metogenesis, gamete biology and early development. Pages 13-43 in J. F. Craig, editor. Pike : biology and exploitation. Chapman & Hall, London, England. Blake, N. M. 1980. Land into water water into land, a history of water management in Florida. University Presses of Fl orida, Tallahassee, Florida, USA. Boschung, H. T. Jr., J. C. Williams, D. W. Gotsha ll, D. K. Caldwell, and M. C. Caldwell. 1995. National Audubon Society field guide to North American fishes, whales, and dolphins. Alfred A. Knopf, Inc., New York, USA.

PAGE 109

98 Brauner, C. J., C. L. Ballantyne, D. J. Randall, and A. L. Val. 1995. Air breathing in the armoured catfish ( Hoplosternum littorale ) as an adaptation to hypoxic, acidic, and hydrogen sulphide rich wa ters. Canadian Journal of Zoology 73:739-744. Burnham, K. P., and W. S. Overton. 1979. Robust estimation of population size when capture probabilities vary among animals. Ecology 60:927-936. Bury, R. B., and P. S. Corn. 1991. Sampli ng methods for amphibians in streams in the Pacific Northwest. Gen. Tech. Rep. PNW-GTR-257. Portland, Oregon: U.S. Department of Agriculture, Forest Servi ce, Pacific Northwest Research Station. Casazza, M. L., G. D. White, and C. J. Gr egory. 2000. A funnel trap modification for surface collection of aquatic amphibians and reptiles. Herpetological Review 31:91-92. Chapman, L. J., C. A. Chapman, and M. Ch andler. 1996. Wetland ecotones as refugia for endangered fishes. Biol ogical Conservation 78:263-270. Chick, J. H., and C. C. McIvor. 1994. Pa tterns in abundance and composition of fishes among beds of different macrophytes: view ing a littoral zone as a landscape. Canadian Journal of Fisheries and Aquatic Sciences 51: 2873-2882. Conant, R., and J. T. Collins. 1998. A field guide to reptiles and amphibians of eastern and central North America, third edi tion, expanded. Houghton Mifflin Company, New York, New York, USA. Cooke, G. D., E. B. Welch, S. A. Peterson, and P. R. Newroth. 1993. Restoration and management of lakes and reservoirs, second edition. Lewis Publishers, Boca Raton, Florida, USA. Corn, P. S. 1994. Straight-line drift fences and pitfall traps. Pages 109-117 in W. R. Heyer, M. A. Donnelly, R. W. McDiarmid, L. C. Hayek, and M. S. Foster, editors. Measuring and monitoring biol ogical diversity: standard methods for amphibians. Smithsonian Institution, Washington D.C., USA. Crowder, L. B., and W. E. Cooper. 1982. Habitat structural complexity and the interaction between bluegill a nd their prey. Ecology 63:1802-1813. Darby, P. C., P. L. Valentine-Darby, H. F. Percival, and W. M. Kitchens. 2001. Collecting Florida applesnails ( Pomacea paludosa ) from wetland habitats using funnel traps. Wetlands 21:308-311. Dole, J. W. 1965. Spatial relations in natural populations of the leopard frog, Rana pipiens Schreber, in northern Michigan. The American Midland Naturalist 74: 464-478. Ernst, C. H. 1986. Ecology of the turtle, Sternotherus odoratus in Southeastern Pennsylvania. Journal of Herpetology 20:341-352.

PAGE 110

99 Ernst, C. H., and E. Ernst. 2003. Snakes of the United States and Canada. Smithsonian Institute Press, Washington DC, USA. Ernst, C. H., J. E. Lovich, and R. W. Bar bour. 1994. Turtles of the United States and Canada. Smithsonian Institute Press, Washington DC, USA. Etheridge, K. 1990. The energetics of estivating sirenid salamanders ( Siren lacertina and Pseudobranchus striatus ). Herpetologica 46:407-414. Florida Fish and Wildlife Conservation Commission. 2003. 2004 Lake Tohopekaliga habitat enhancement project, a fishery mana gement program. Freshwater Fisheries Division, Kissimmee, Florida, USA. Florida Fish and Wildlife Conservation Commission. 2004. Kissi mmee Chain of Lakes highlights, August 13, 2004. Aquatic Ha bitat Conservation and Restoration Section, Kissimmee, Florida, USA. Gehlbach, F. R., and S. E. Kennedy. 1978. Population ecology of a highly productive aquatic salamander ( Siren intermedia ). Southwestern Naturalist 23: 423-430. Gibbons, J. W., J. L. Greene, and J. D. Congdon. 1983. Drought-related responses of aquatic turtle populations. J ournal of Herpetology 17:242-246. Godley, J. S. 1980. Foraging ecolo gy of the striped swamp snake, Regina alleni in Southern Florida. Ecol ogical Monographs 50:411-436. Godley, J. S. 1983. Observations on the courtship, nests and young of Siren intermedia in southern Florida. The Amer ican Midland Naturalist 110:215-219. Harper, R. M. 1921. Geography of cent ral Florida. Florida State Geological Survey, 13th Annual Report. Hasler, A. D. 1947. Eutrophication of lakes by domestic drainage. Ecology 28:383–395 Hayek, L. C. 1994. Analysis of amphi bian biodiversity data. Pages 207-269 in W. R. Heyer, M. A. Donnelly, R. W. McDiarmid, L. C. Hayek, and M. S. Foster, editors. Measuring and monitoring biol ogical diversity: standard methods for amphibians. Smithsonian Institution, Washington D.C., USA. HDR Engineering, Inc. 1989. Technical report for the deve lopment of a surface water improvement and management plan for Lake Tohopekaliga/East Lake Tohopekaliga, Final Report. South Florida Water Management District, West Palm Beach, Florida, USA. Contract No. 88-475-0961. Hines, J. E., T. Boulinier, J. D. Nichols, J. R. Saur, and K. H. Pollock. 1999. COMDYN: software to study the dynamics of animal communities using a capture-recapture approach. Bird Study 46 Supplement, 209-217.

PAGE 111

100 Holt, G. J. 2002. Human impacts. Pages 222-242 in L. A. Fuiman and R. G. Werner, editors. Fishery science: the unique cont ributions of early life stages. Blackwell Science, Ltd., Osney Mead, Oxford, UK. Hostache, G., and J. H. Mol. 1998. Reproduc tive biology of the neotropical armoured catfish Hoplosternum littorale (Siluriformes-Callichthyida e): a synthesis stressing the role of the floating bubble nest Aquatic Living Resources 11:173-185. Hoyer, M. V., and D. E. Canfield, Jr. 1994. Handbook of common freshwater fish in Florida lakes. University of Fl orida, Gainesville, Florida, USA. Iverson, J. B. 1982. Biomass in turtle populations: a neglected subject. Oecologia 55:69-76. Johnson, S. A., and W. J. Barichivich. 2004. A simple technique for trapping Siren lacertina Amphiuma means and other aquatic vertebrate s. Journal of Freshwater Ecology 19:263-269. Joly, P., and A. Morand. 1997. Amphibian di versity in land-water ecotones. Pages 161182 in Lachavanne, J. B., and R. Juge, editors Biodiversity in land-inland water ecotones Man and the Biosphere Series. Volume 18. UNESCO, Paris, France, and Parthenon Publishing, Carnforth, England. Kilby, J. D. 1936. A biological analysis of the food and feeding habits of Rana sphenocephala (Cope) and Hyla cinerea (Schneider). M.S. Thesis. University of Florida, Gainesville, Florida, USA. Klemens, M. W. 2000. Turtle Conservation. Smithsonian Institution Press, Washington, DC, USA. Lachavanne, J. B. 1997. Why study biodiversit y in land-inland water ecotones? Pages 1-45 in Lachavanne, J.B., and R. Juge, edito rs. Biodiversity in Land-Inland Water Ecotones Man and the Biosphere Series. Volume 18. UNESCO, Paris, France, and Parthenon Publishing, Carnforth, England. Lagler, K. F. 1943. Methods of collecti ng freshwater turtles. Copeia 1943:21-25. Larson, R. E., R. P. Hostetler, and B. H. Edwards. 1994. Calculus with analytic geometry, fifth edition. D.C. Heath and Company, Massachusetts, USA. MacKenzie, D. I., J. D. Nichols, G. B. L achman, S. Droege, J. A. Royle, and C. A. Langtimm. 2002. Estimating site occupanc y rates when detec tion probabilities are less than one. Ecology 83:2248-2255. Mahmoud, I. Y. 1968. Feeding behavior in ki nosternid turtles. Herpetologica 24: 300305.

PAGE 112

101 Mahmoud, I. Y. 1969. Comparative ecology of the kinosternid tur tles of Oklahoma. The Southwestern Naturalist 14:31-66. Martof, B. S. 1969. Prolonged inanition in Siren lacertina Copeia 1969:285-288. McCune, B. and J. B. Grace. 2002. Analysis of ecological communities. MjM Software Design, Gleneden Beach, Oregon, USA. McCune, B., and M. J. Mefford. 1999. Multivariate analysis of ecological data, version 4.20. MjM Software, Gleneden Beach, Oregon, USA. Mills, M. S., C. J. Hudson, and H. J. Berna. 1995. Spatial ecology and movements of the brown water snake ( Nerodia taxispilota ). Herpetologica 51:412-423. Miranda, L. E., and E. D. Dibble. 2002. An ecological foundation for black bass management. Pages 433-453 in D. P. Philipp and M. S. Ridgeway. Black bass: ecology, conservation, and management. Am erican Fisheries Society, Symposium 31, Bethesda, Maryland, USA. Miranda, L. E., and K. B. Hodges. 2000. Role of aquatic vegetation coverage on hypoxia and sunfish abundance in bays of a eutrophic reservoir. Hydrobiologia 427:51-57. Miranda, L. E., M. P. Driscoll, and M. S. Allen. 2000. Transient physiochemical microhabitats facilitate fish survival in inhospitable aquatic plant stands. Freshwater Biology 44:617-628. Mol, J. H. 1993. Structure and function of floating bubble nests of three armoured catfishes (Callichthyidae) in relation to the aquatic environment. Pages 167-197 i n P. E. Ouboter, editor. Freshwater Ec osystems of Suriname. Monographiae Biologicae 70. Klewer, Do rdrecht, the Netherlands. Moulton, C. A., W. J. Fleming, and B. R. Nerney. 1996. The use of PVC pipes to capture hylid frogs. Herpet ological Review 27:186-187. Moyer, E. J., M. W. Hulon, R. S. Butler, D. C. Arwood, C. Michael, and C. A. Harris. 1987. Kissimmee Chain of Lakes studi es, completion report for Lake Tohopekaliga investigations. State of Florida Game and Fresh Water Fish Commission, Tallahassee, Florida, USA. Moyer, E. J., M. W. Hulon, J. Buntz, R. W. Hujik, J. J. Sweatman, C. S. Michael, D. C. Arwood, and A. C. Jasent. 1993. Comp letion report for Kissimmee Chain of Lakes studies (1987-1992). State of Florida Game and Fresh Water Fish Commission, Tallahassee, Florida, USA. Moyer, E. J., M. W. Hulon, J. J. Sweatman, R. S. Butler, and V. P. Williams. 1995. Fishery responses to habitat restoration in Lake Tohopekaliga, Florida. North American Journal of Fish eries Management 15:591-595.

PAGE 113

102 Mushinsky, H. R., and J. J. Hebrard. 1977. Food partitioning by five species of water snakes in Louisiana. Herpetologica 33:162-166. Mushinsky, H. R., J. J. Hebrard, and D. S. Vodopich. 1982. Ontogeny of water snake foraging ecology. Ecology 63:1624-1629. Mushinsky, H. R., and D. E. Miller. 1993. Predation on water snakes: ontogenetic and interspecific considerations. Copeia 1993:660-665. Nichols, J. D., T. Boulinier, J. E. Hines, K. H. Pollock, and J. R. Saur. 1998. Estimating rates of local species extinction, colonizat ion, and turnover in animal communities. Ecological Appli cations 8:1213-1225. Nico, L. G., S. J. Walsh, and R. H. Robins 1996. An introduced population of the South American Callichthyid catfish Hoplosternum littorale in the Indian River Lagoon system, Florida. Flor ida Scientist 59:189-200. Nico, L. G., and P. L. Fuller. 1999. Spatial a nd temporal patterns of nonindigenous fish introductions in the United States. Fisheries 24:16-27. Page, L. M. 1994. Identification of sailfin catfishes introduced to Florida. Florida Scientist 57:171-172. Parker, J. O., Jr. 1970. Surfacing of dead fish following application of rotenone. Transactions of the American Fisheries Society 99:805-807. Petranka, J. W. 1998. Salamanders of th e United States and Canada. Smithsonian Institute Press, Washington, DC, USA Pieczynska, E. 1990. Lentic a quatic-terrestrial ecotones: th eir structure, functions and importance. Pages 103-140 in : R. J. Naiman and H. Decamps, editors. The ecology and management of aquatic-terrest rial ecotones. The Parthenon Publishing Group, Paris, France. Pieczynska, E, and M. Zalewski. 1997. Ha bitat complexity in land-inland water ecotones. Pages 61-108 in : Lachavanne, J. B., and R. J uge, editors. Biodiversity in land-inland water ecotones Man and the Biosphere Series. Volume 18. UNESCO, Paris, France, and Parthenon P ublishing, Carnforth, England. Pough, F. H., R. M. Andrews, J. E. Cadle, M. L. Crump, A. H. Savitzky, and K. D. Wells. 2001. Herpetology, second edition. Prentice-Hall, Inc., Upper Saddle River, New Jersey, USA. Reaser, J. K. 2000. Amphibian declines: an issue overview. Federal Taskforce on Amphibian Declines and Deformities, Washington, DC, USA. Schiffer, D. M. 1998. Hydrology of central Florida lakes: a primer. U.S. Geological Survey Circular 1137, Denver, Colorado, USA.

PAGE 114

103 Seigel, R. A., J. W. Gibbons, and T. K. Lynch. 1995. Temporal changes in reptile populations: effects of a severe drought on aquatic snakes. Herpetologica 51: 424434. Snodgrass, J. W., J. W. Ackerman, A. L. Br yan, Jr., and J. Burger. 1999. Influence of hydroperiod, isolation, and heterospecif ics on the distribution of aquatic salamanders ( Siren and Amphiuma ) among depression wetlands. Copeia 1999:107113. Sorensen, K. 2003. Trapping success and population analysis of Siren lacertina and Amphiuma means M.S. Thesis. University of Flor ida, Gainesville, Florida, USA. Sorensen, K. 2004. Population characteristics of Siren lacertina and Amphiuma means in north Florida. Southe astern Naturalist 3:249-258. SPSS Inc. 2001. SPSS 11.0 for Windows. Chicago, Illinois, USA. Wegener, W. 1969. Lake Tohopekaliga fisher y investigation. Florida Game and Fresh Water Fish Commission, Tallaha ssee, Florida, USA. Wegener, W., and V. Williams. 1974. Extreme lake drawdown, a working fish management technique. Dingell-Johnson Federal Aid Project F-29-R. Florida Game and Fresh Water Fish Commissi on, Tallahassee, Florida, USA. Welch, Z. C. 2004. Littoral vegetation of Lake Tohopekaliga: community descriptions prior to a large-scale fisheries habita t-enhancement project. M.S. Thesis. University of Florida, Gainesville, Florida, USA. Werner, R. G. 2002. Habitat requirements. Pages 161-182 in L. A. Fuiman and R. G. Werner, editors. Fishery science: the uni que contributions of early life stages. Blackwell Science, Ltd., Osney Mead, Oxford, UK. Werner, E. E., and D. J. Hall. 1979. Fo raging efficiency and habitat switching in competing sunfishes. Ecology 60:256-264. Werner, E. E., and D. J. Hall. 1988. Ontoge netic habitat shifts in bluegill: the foraging rate-predation risk tradeoff. Ecology 69: 1352-1366. White, G. C., D. R. Anderson, K. P. Burnham, and D. L. Otis. 1982. Capture-recapture and removal methods for sampling clos ed populations. Los Alamos National Laboratory Report LA-8787-NERP, Los Alamos, New Mexico, USA. White, G. C., and K. P. Burnham. 1999. Program MARK: Survival estimation from populations of marked animals. Bird Study 46 Supplement, 120-138. Williams, B. K., J. D. Nichols, and M. J. Conroy. 2002. Analysis and management of animal populations. Academic Pre ss, San Diego, California, USA.

PAGE 115

104 Wilson, K. R., and D. R. Anderson. 1985. Eval uation of two density estimators of small mammal population size. J ournal of Mammalogy 66:13-21. Wygoda, M. L. 1979. Terrestrial activity of striped mud turtles, Kinosternon baurii (Reptilia, Testudines, Kinos ternidae) in West-central Florida. Journal of Herpetology 13:469-480. Zimmerman, B. L. 1994. Audio-st rip transects. Pages 92-97 i n W. R. Heyer, M. A. Donnelly, R. W. McDiarmid, L. C. Hayek, and M. S. Foster, editors. Measuring and monitoring biological diversity: standard methods for amphibians. Smithsonian Institution, Washington D.C, USA.

PAGE 116

105 BIOGRAPHICAL SKETCH Ann Marie Muench was born in Louisvil le, KY, in 1977. Through middle and high school she also lived in Cumberland, MD, and Jacksonville Beach, FL. After high school, she attended the Univer sity of North Florida in Jacksonville, FL, where in 1998 she received a Bachelor of Science degree in biology. After graduation she worked as a Metals Analyst at Environmental Conserva tion Laboratories, Inc., an environmental testing laboratory in Jacksonv ille, FL. Ann Marie decided to pursue her career goal of catching reptiles and amphibian s, and applied to the Depart ment of Wildlife Ecology and Conservation at the University of Florida fo r graduate school. For the next three years she examined the ecology of the aquatic ve rtebrate community in Lake Tohopekaliga, FL, as a graduate assistant with the Florida Cooperative Fish and Wildlife Research Unit. She received her Master of Science degr ee in December, 2004, and subsequently moved to Black Mountain, NC.