GOPHER TORTOISE (Gopherus polyphemus) HABITAT SELECTION, HEALTH, AND FECUNDITY

By C. LEANN WHITE

A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2009 1

© 2009 C. LeAnn White

2

To Matt and Ev

3

ACKNOWLEDGMENTS My deepest thanks go to all of the field team members that assisted with data collection for this project, especially Caro Perez-Heydrich, Melissa Clark, Katie Jackson, Victor Kasper, Jessica Coan, Kristin Miller, and Sarah Chaney. I thank John Wooding for collecting an amazing set of habitat data without which my project would not have been possible. I thank Caro Perez-Heydrich for providing comic relief, friendship, and teaching me everything I know about tracking tortoises. I also thank Melissa Clark for introducing hip-hop night and for being an all-around good sport when working day and night for weeks on end. I am particularly indebted to Lori Wendland for giving me my first wildlife disease position and teaching me much of what I know about gopher tortoises and upper respiratory tract disease. I thank my committee members Jorge Hernandez, Joel Brendemuhl, George Tanner, Mary Brown, and Mary Christman for their guidance and support. In particular, I am grateful to Jorge Hernandez for continuing my education in study design and epidemiology. I thank Joel Brendemuhl for steering me through the complexities of animal nutrition. I thank George Tanner for imparting his knowledge of wildlife-habitat relationships. I am especially grateful for Mary Brown’s encouragement and overall positive outlook on life. I am forever grateful to Mary Christman for seeing my potential and helping guide me through the world of statistics. I’m a better person for it. I thank the entire Brown Laboratory especially Dina Demcovitz and Barbara Crenshaw for logistical and moral support. I thank Jan Stevens, Fiona Maunsell, and Ayman Allam for their friendship and support throughout my graduate career. Finally, my thanks go to Matthew Reetz for never letting go, even in the darkest of hours.

4

TABLE OF CONTENTS page ACKNOWLEDGMENTS ...............................................................................................................4  LIST OF TABLES...........................................................................................................................7  LIST OF FIGURES .......................................................................................................................10  ABSTRACT...................................................................................................................................11  CHAPTER 1

INTRODUCTION: THE RELATIONSHIP BETWEEN HABITAT SELECTION, HEALTH STATUS, AND REPRODUCTION IN GOPHER TORTOISES.........................13 

2

GOPHER TORTOISE HABITAT SELECTION IN THREE HABITAT TYPES IN NORTH AND CENTRAL FLORIDA ...................................................................................16  Introduction.............................................................................................................................16  Methods ..................................................................................................................................18  Study Site Selection.........................................................................................................18  Sampling Scheme and Sample Collection.......................................................................19  Monitoring of Tortoise Movement ..................................................................................19  Habitat Use Measurements..............................................................................................20  Habitat Availability Measurements.................................................................................21  Data Analysis...................................................................................................................21  Results.....................................................................................................................................22  Discussion...............................................................................................................................23  Conclusions.............................................................................................................................25 

3

EXAMINING HABITAT AT TWO SPATIAL SCALES TO DETERMINE ITS INFLEUNCE ON NUTRITIONAL STATUS IN GOPHER TORTOISES ..........................32  Introduction.............................................................................................................................32  Methods ..................................................................................................................................34  Study Populations for Population Level Habitat Characteristics ....................................34  Study Populations for Individual Level Habitat Characteristics .....................................36  Field Techniques and Assessment of Tortoise Nutritional Status...................................37  Individual Level Habitat Measurements .........................................................................38  Population Level Habitat Measurements.........................................................................39  Data Analysis...................................................................................................................39  Results.....................................................................................................................................41  Individual Habitat Use Summary Statistics.....................................................................41  Population Level Habitat Summary Statistics.................................................................41  Canonical Correlation and ANOVA Results...................................................................42  Discussion...............................................................................................................................43  5

Effects of Reproduction...................................................................................................44  Effects of Food Intake and Rainfall Patterns...................................................................45  Effects of Disease ............................................................................................................48  Conclusions.............................................................................................................................49  4

BIOCHEMICAL, MORPHOLOGICAL, AND HABITAT FACTORS ASSOCIATED WITH FECUNDITY IN GOPHER TORTOISES .................................................................73  Introduction.............................................................................................................................73  Methods ..................................................................................................................................77  Study Populations and Field Methods.............................................................................77  Statistical Modeling of Fecundity ...................................................................................78  Results.....................................................................................................................................83  Summary Statistics ..........................................................................................................83  Statistical Modeling Results ............................................................................................84  Discussion...............................................................................................................................86  Factors Associated with Being Gravid ............................................................................87  Factors Associated with Clutch Size ...............................................................................92  Conclusions.............................................................................................................................94 

5

EVALUATION OF THE EFFECTS OF PLASMA VITAMIN AND MINERAL LEVELS ON FECUNDITY IN GOPHER TORTOISES ....................................................109  Introduction...........................................................................................................................109  Methods ................................................................................................................................112  Plasma Sampling ...........................................................................................................112  Data Analysis.................................................................................................................113  Results...................................................................................................................................114  Discussion.............................................................................................................................115  Zinc................................................................................................................................115  Selenium ........................................................................................................................117  Manganese.....................................................................................................................118  Conclusions...........................................................................................................................120 

6

CONCLUSIONS AND FUTURE DIRECTIONS ...............................................................127  Summary and Implications of Findings................................................................................127  Gopher Tortoise Habitat Selection ................................................................................127  Relationship between Habitat and Nutritional Status in Gopher Tortoises...................128  Effects of Habitat, Nutrition, and Disease on Gopher Tortoise Reproduction..............130  Effects of Vitamin and Minerals on Gopher Tortoise Reproduction ............................133  Future Research ....................................................................................................................135 

LIST OF REFERENCES.............................................................................................................138  BIOGRAPHICAL SKETCH .......................................................................................................149 

6

LIST OF TABLES Table

page

2-1

Mean distance traveled by gopher tortoises in 2007 at three study sites in north and central Florida. ...................................................................................................................30 

2-2

Results of comparison of distance traveled by gopher tortoises in 2007 at three study sites in north and central Florida........................................................................................30 

2-3

Mean percent cover of vegetation categories in habitat used by gopher tortoises and habitat available to tortoises in 2007 at three study sites in north and central Florida. .....30 

2-4

Pearson correlation coefficients for vegetation predictor values at the BS site used in the comparison of gopher tortoise habitat use and availability in 2007. ...........................30 

2-5

Pearson correlation coefficients for vegetation predictor values at the CentFL site used in the comparison of gopher tortoise habitat use and availability in 2007. ...............31 

2-6

Pearson correlation coefficients for vegetation predictor values at the OR site used in the comparison of gopher tortoise habitat use and availability in 2007. ...........................31 

3-1

Summary of individual gopher habitat use at four study sites in north and central Florida, 2006 and 2007. .....................................................................................................56 

3-2

Summary of individual gopher habitat use by sex in four study sites in north and central Florida, 2006 and 2007. .........................................................................................57 

3-3

Summary of population level habitat availability 11 study sites in north and central Florida, 2004–2006. ...........................................................................................................58 

3-4

Mean species richness at 11 study sites in north and central Florida, 2004–2006. ...........60 

3-5

Results of ANOVA on the effect of factors and interactions on selected blood parameters used in the population habitat level canonical correlation. .............................60 

3-6

Tortoise blood parameters for 11 sites in north and central Florida, 2004–2006. .............62 

3-7

Summary statistics of blood parameters by sex for gopher tortoises sampled in north and central Florida, 2004–2006. ........................................................................................67 

3-8

Summary statistics of blood parameters by year for gopher tortoises sampled in north and central Florida. ............................................................................................................69 

3-9

Comparison of gopher tortoise GLUC levels by site with a Tukey adjustment of the means comparison..............................................................................................................71 

4-1

A priori models representing hypotheses of factors associated with fecundity of female gopher tortoises in north and central Florida, 2004–2006. ....................................99  7

4-2

Summary statistics for clutch size produced by female gopher tortoises in north and central Florida, 2004–2006. ...............................................................................................99 

4-3

Proportion of gravid females and mean clutch sizes for gravid females in five gopher tortoise populations in north and central Florida, 2004–2006. ..........................................99 

4-4

Summary statistics for plasma chemistry parameters and morphometric measurements used in analysis of fecundity of gopher tortoises in north and central Florida, 2004–2006. .........................................................................................................100 

4-5

Results of simple logistic regressions relating presence-absence of eggs and predictor variables for north and central Florida gopher tortoises, 2004–2006.. ............................101 

4-6

Spearman correlation coefficients for predictor variables used in analysis of fecundity of gopher tortoises in north and central Florida, 2004–2006...........................102 

4-7

Pairwise correlation between the number of eggs and the predictor variables for female tortoises in north and central Florida with at least one egg. ................................104 

4-8

Eigenvalues from the principal components analysis performed on study year and vegetation parameters measured in gopher tortoise study sites in north and central Florida, 2004–2006. .........................................................................................................105 

4-9

The first two eigenvectors from the principal components analysis performed on study year and vegetation parameters measured in gopher tortoise study sites in north and central Florida, 2004–2006. ......................................................................................105 

4-10

Model estimates for the binomial and Poisson component of the final ZIP model of fecundity in gopher tortoises from north and central Florida, 2004–2006. .....................106 

4-11

Female gopher tortoises for which the ZIP model misclassified their gravid status. ....106 

4-12

Summary statistics for plasma chemistry parameters and morphometric measurements for gravid versus non-gravid female gopher tortoises in north and central Florida, 2004–2006. .............................................................................................107 

5-1

Summary statistics for plasma vitamin and mineral levels of non-gravid female, gravid female, and male gopher tortoises in north and central Florida, 2004–2006. ......124 

5-2

Spearman correlation coefficients, probability, and number of observations for vitamin and mineral predictor variables used in analysis of fecundity of female gopher tortoises in north and central Florida, 2004–2006. ..............................................125 

5-3

Results of the logistic regression and canonical discriminant analysis performed on predictor variables selected by stepwise discriminant analysis for gravid female gopher tortoises in north and central Florida, 2004–2006. ..............................................126 

8

5-4

Comparison of vitamin and mineral plasma levels in non-gravid female and male gopher tortoises in north and central Florida, 2004–2006. ..............................................126 

5-5

Comparison of vitamin and mineral plasma levels in gravid female and male gopher tortoises in north and central Florida, 2004–2006. ..........................................................126 

9

LIST OF FIGURES Figure

page

1-1

Conceptual model of the relationship between habitat selection, health, and fitness in gopher tortoise ...................................................................................................................15 

2-1

Rotated factor pattern of the six vegetation categories used to compare gopher tortoise habitat use and availability in 2007. .....................................................................29 

2-2

Overlay plot of Factor 1 versus Factor 2 generated from the factor analysi......................29 

3-1

Location of 11 gopher tortoise study sites in Florida ........................................................51 

3-2

Size class distributions of four study sites selected for individual habitat measurements.....................................................................................................................52 

3-3

The effect of site*sex on P levels (ln transformed) in gopher tortoises sampled during 2004 –2006 from 11 study sites in north and central Florida ............................................53 

3-4

The effect of site*sex on Ca levels in gopher tortoises sampled during 2004–2006 from 11 study sites in north and central Florida ................................................................53 

3-5

The effect of site*year on K levels in gopher tortoises sampled during 2004–2006 in north and central Florida....................................................................................................54 

3-6

The effect of site*year on Na levels in gopher tortoises sampled during 2004–2006 in north and central Florida....................................................................................................54 

3-7

The effect of site*year on TP levels in gopher tortoises sampled during 2004–2006 in north and central Florida....................................................................................................55 

3-8

The effect of site*year on γGT levels (ln transformed) in gopher tortoises sampled during 2004–2006 in north and central Florida. ................................................................55 

4-1

Location of the five gopher tortoise study sites in Florida. ...............................................97 

4-2

Frequency graph of the clutch size produced by female gopher tortoises in north and central Florida during 2004–2006......................................................................................98 

5-1

Side by side boxplots of vitamin and mineral plasma levels for non-gravid and gravid female gopher tortoises in north and central Florida, 2004–2006. ..................................121 

10

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy GOPHER TORTOISE (Gopherus polyphemus) HABITAT SELECTION, HEALTH, AND FECUNDITY By C. LeAnn White December 2009 Chair: Mary B. Brown Major: Veterinary Medical Sciences Habitat alterations that have resulted in a loss of diversity or abundance of plant species could potentially decrease the nutritional status of animals over time, thereby affecting their growth rates, survival, or reproduction. The overarching goal of this project was to examine whether alterations to gopher tortoise (Gopherus polyphemus) habitats could negatively impact their nutritional status or fecundity. To answer this question we examined (1) individual habitat selection to determine if gopher tortoises were selective in their use of resources when habitat type varied, (2) whether habitat parameters could be used to predict gopher tortoise nutritional status, and (3) the influence of nutritional status (determined by blood biochemistry and vitamin and mineral assays) on reproduction in gopher tortoises. Our results indicated that tortoises largely used broad vegetation categories (grasses, forbs, legumes, and woody species) in proportion to their availability. This finding could potentially have different consequences for tortoises in different sites or habitat types. For example, legumes are thought to be particularly important for re-conditioning females after egg laying, but if these items are less available in some habitats they may also constitute less of their diet. Interestingly, both clutch sizes and percent gravid females were lowest in this study at the site with lowest percent cover of legumes. We also found that the habitat measurements used in this 11

study were not able to account for a significant amount of the variability in gopher tortoise blood parameters. However, there were significant main effects and interactions for site, year, and sex, for many of the gopher tortoise blood parameters, suggesting the importance of the reproductive cycle and geographic location on blood parameters in this species. Female gopher tortoises with lower body condition scores and lower plasma phosphorus levels were less likely to have eggs than females with lower body condition scores. Not unexpectedly, clutch size in gopher tortoises increased with carapace length. We also identified three minerals (zinc, manganese, and selnium) that differed significantly between gravid and non-gravid female gopher tortoises. Although the importance of these minerals to reproduction has been demonstrated in other species as we initially positied, more data are needed to determine whether levels in non-reproductive females truly represent deficiencies or toxicity in this species. Future studies will need to address the extended reproductive cycle of gopher tortoises as it may be difficult to determine a time of the year when blood parameters are not at least partially influenced by reproduction. Sampling blood parameters and forage availability at multiple times during the year may help address this problem as well as increase our understanding of whether nutrients acquired prior to or after brumation (period of inactivity when metabolism slows but animals are still awake) influence reproduction.

12

CHAPTER 1 INTRODUCTION: THE RELATIONSHIP BETWEEN HABITAT SELECTION, HEALTH STATUS, AND REPRODUCTION IN GOPHER TORTOISES The fitness of an animal is measured by its contribution to the breeding population of the next generation (Meffe and Carroll 1997). Fitness is a theoretical evolutionary concept with direct applications in conservation biology because it includes parameters such as growth rates, survival, and reproduction that affect the persistence of wildlife populations (Audesirk et al. 2005). Although many factors (e.g., genetics, weather, predation) could potentially alter these fitness parameters, habitat loss and degradation have undoubtedly affected biodiversity of vertebrates and invertebrates worldwide (Vitousek 1994, Vitousek et al. 1997). Often the mechanisms by which habitat changes may affect a fitness parameter like survival are obvious such as highway-induced mortality of wildlife (Aresco 1999, Dodd et al. 2004). However, habitat changes may also have subtle effects or take longer periods of time to cause noticeable changes in fitness parameters. For example, anthropogenic habitat conversions (e.g., to pastureland, agriculture or urban areas) that result in a loss of plant diversity could potentially decrease the nutritional status of animals over time, thereby affecting growth rates, survival, or reproduction of animals. The overarching objective of this study was to examine the potential consequences of habitat quality on reproduction in gopher tortoises (Gopherus polyphemus), a long-lived species important for overall upland ecosystem health. Animals can only use resources that are available to them. Therefore, even if an animal is selective in its use of resources, its nutritional status could vary because of habitat quality parameters such as forage availability. Since nutritional status has been shown to affect reproduction in various laboratory and wildlife species (Bolton et al. 1992, Underwood and Suttle 1999, Wobeser 2006), it may serve as an intermediate effect between habitat quality and reproductive output in gopher tortoises (Figure 1-1). In this study 13

we examined individual habitat selection to determine if gopher tortoises were selective in their use of resources when habitat type varied (Chapter 2). We also examined whether habitat parameters could be used to predict gopher tortoise nutritional status (Chapter 3). Lastly, we examined the influence of nutritional status (determined by blood biochemistry and vitamin and mineral assays) on reproduction in gopher tortoises (Chapters 4 and 5). We also considered the potential for other factors such as disease to affect gopher tortoise nutritional status and reproduction (Figure 1-1), specifically Upper Respiratory Tract Disease (URTD). URTD is the best-characterized disease in gopher tortoises and has been associated with several gopher tortoise die-offs (Gates et al. 2002, Seigel et al. 2003). The potential for URTD to affect various parameters was considered throughout this study and is addressed in appropriate chapters. The gopher tortoise is a keystone upland species whose burrowing activities and burrows themselves are important for upland ecosystems. However, due to continued declines over the past century the gopher tortoise is currently listed as threatened or endangered throughout most its range. The primary cause of these declines is considered to be the result of the cumulative effects of habitat loss and disease (Auffenberg and Franz 1982, Berish et al. 2000, Gates et al. 2002, Seigel et al. 2003). The remaining natural and relocated populations of gopher tortoises now reside in habitats of variable quality. Therefore, determining whether the health status and reproductive output of these animals is being impacted by their habitat will be extremely important for management recommendations aimed at conservation of the this species and upland ecosystems as a whole.

14

Disease (extrinsic factor) -URTD status

Habitat

Health Status

-Site level availability -Individual habitat use

-Blood parameters associated with nutrition

Fitness -Fecundity

Intrinsic/Host Factors

Other Extrinsic Factors

-Genetics -Degenerative effects of aging

-Intra/Interspecific competition -Predation -Population demography

Figure 1-1. Conceptual model of the relationship between habitat selection, health, and fitness in gopher tortoises. The solid lines indicate factors and pathways that will be investigated in this study. The dashed boxes and lines represent other intrinsic and extrinsic parameters and pathways that can also influence health status and fitness but are not addressed in this study.

15

CHAPTER 2 GOPHER TORTOISE HABITAT SELECTION IN THREE HABITAT TYPES IN NORTH AND CENTRAL FLORIDA Introduction General characteristics of gopher tortoise habitats include sandy, well-drained soils with open-canopy or early successional areas that provide high ground level light intensities and high herbaceous plant diversity (Auffenberg and Franz 1982). These conditions historically existed in upland sandhills, pine flatwoods, scrubby flatwoods, xeric hammocks, dry prairies, and coastal dunes. However, many of these habitats have been altered and degraded by decades of fire suppression and anthropogenic changes such as conversion to pasture or agricultural lands and urbanization. Fire suppression for example, can degrade gopher tortoise habitats by allowing increased woody growth and decreased light penetration that reduces herbaceous plant growth and suitable areas for thermoregulation (basking) and nesting. Foraging livestock or monocultures planted for food production may also result in decreased habitat quality for gopher tortoises by decreasing plant diversity due to competitive release of unpalatable species during grazing (e.g., woody species) or soil disturbances that increase encroachment of exotic species (e.g., cogon grass). The effects can persist even after these areas are no longer actively used for production because ecological succession will generally accelerate resulting in closed canopy areas unsuitable for gopher tortoises. Gopher tortoises may relocate to marginal habitat as long as it has open canopy and sufficient biomass (Aresco and Guyer 1999). The consequences of relocating to new habitats is not well understood, but it has been suggested that the excess energy expenditure by gopher tortoises searching for suitable burrow sites and creating new burrows may decrease energy available for reproduction (Aresco and Guyer 1999). Any habitat change that influences an animal’s ability to locate resources such as food, mates, and suitable areas for nesting and thermoregulation has the potential to affect fitness 16

parameters such survival, growth, and reproductive rates. Studies have shown that habitat changes such as decreased biomass and increased canopy cover result in increased home range sizes (Auffenberg and Iverson 1979) and decreased site fidelity for gopher tortoises (Aresco and Guyer 1999). Further, since they are considered generalist herbivores and eat a variety of plant species, herbaceous diversity may also be important to gopher tortoises. Tortoise populations now reside in areas of varying degrees of habitat quality due to relocation or translocation programs or because of historical management practices (e.g., wildlife suppression) in naturally occurring populations. Examining tortoise habitat selection in various types of habitat will increase our understanding of how tortoises respond to habitat change. To determine whether an animal is selective in its use of resources, habitat characteristics are often classified into discrete categories. Habitat selection is then defined as the use of a particular type of habitat category more or less often than would be expected by chance (Johnson 1980). Habitat availability in many studies is defined at the level of the site or home range of the animal. However, this definition of availability assumes the entire study site is available to it and that the distribution of habitat within the home range of the animal represents free choice (Arthur et al. 1996). Our interest is more local in that we consider habitat utilization at the level of the individual animal’s movements within the available environment and define utilization as a function of the choices that the animal makes as it moves through the landscape. Habitat was defined in this study as the number of species and percent cover of five vegetation categories (forb, grass, legume, woody species, and bare ground/litter) occurring ≤ 1 m from the ground (accessible to gopher tortoises). We defined habitat use as the vegetation encountered by gopher tortoises during daily movements outside of their burrows. Habitat availability was measured at a random point within a fixed radius circle whose center was defined by a habitat use point,

17

which allowed habitat availability to change continuously as the animal moved throughout its environment. The objective of this study was to compare gopher tortoise habitat selection in three habitat types. We tested the hypothesis that gopher tortoises selectively use habitats in order to balance their nutritional needs. Gopher tortoise movement will increase when the amount and quality of available foraging habitat is decreased. Methods Study Site Selection Gopher tortoise habitat use and availability was measured on three study sites in north and central Florida. Study sites were selected to provide the greatest potential differences in habitat characteristics on the basis of previously performed surveys (M.B. Brown unpubl. data). The study site at Big Shoals State Park (BS) in Hamilton County is a 15.2 ha former pine plantation surrounded by a hardwood hammock. The gopher tortoise population density at BS is approximately 4.8 tortoises/ha. The site is dominated by bahiagrass (Panicum notatum), which makes up almost 50% of the groundcover. The overstory is primarily comprised of loblolly (Pinus taeda) and longleaf (Pinus palustris) pine trees. The study site at Ordway-Swisher Biological Station (OR), Putnam County, is 61.0 ha sandhill habitat. The population density for OR is approximately 2.5 tortoises/ha. The most commonly observed groundcover species at OR is wiregrass (Aristida spp), silkgrass (Pityopsis graminifolia), and Florida alicia (Chapmannia floridana). The Central Florida (CentFL) study area is a privately owned 6.9 ha site in Orange County with the highest population density (29.1 tortoises/ha) of the three sites included in this study. The CentFL site is a former citrus grove, and the groundcover is currently dominated by Mexican clover (Richardia brasiliensis) and natalgrass (Melinis repens) with a patchy distribution of slash pine (Pinus elliottii) and sand live oak (Quercus geminata) trees. According 18

to habitat surveys performed during 2004–2006 (M.B. Brown unpubl. data) the OR study site had higher plant diversity (mean species/plot = 13.67) than the BS (mean species/plot = 6.95) or CentFL (mean species/plot = 7.31) study sites. However, the 2004–2006 habitat also indicated that the overall herbaceous biomass was lower at the OR study site (total bare ground/litter percent cover = 66.8%, sd = 18.2) compared to the other two study sites (BS total bare ground/litter percent cover = 33.4%, sd = 16.2; CentFL total bare ground/ litterpercent cover = 47.8%, sd = 17.5). Sampling Scheme and Sample Collection Habitat use and availability was measured at the three study sites during May–August 2007. This sampling period was selected because the late spring and summer months represent the highest activity period for gopher tortoises as they must gain nutrients needed for reproduction and develop body reserves needed to sustain them through winter dormancy. Since approximately four weeks were required to capture and track 20 tortoises, each site was sampled once during the 2007 sample period. OR was sampled in June, BS was sampled in July, and CentFL was sampled in August of 2007. Monitoring of Tortoise Movement A total of 68 tortoises were captured in 2007 using pitfall traps set directly in front of burrow openings. Traps were checked daily. Dye packs (diameter = 3-5 cm) were constructed by filling nylon pouches (i.e., pantyhose material) with fluorescent powder dye. These pouches were then attached to the drilled identification holes located on the tortoises rear carapacial scutes (Cagle 1939). The movement trails of tortoises with dye packs were tracked nightly for 10-14 days using a hand held UV light. Tracking techniques did not disturb tortoise movement because tortoises are diurnal. Tortoise movement trails were recorded as continuous polylines

19

(created by recording one waypoint/second) in the field with a GPS attached to a handheld PDA with ArcPad v. 6.2 (Environmental Systems Research Institute, Redlands, CA). Habitat Use Measurements Habitat use was defined as the vegetation encountered by gopher tortoises during daily movements outside of their burrows. If tortoises did not leave the apron (unvegetated sandy area) of the burrow (as indicated by their dye trails) these movements were regarded as thermoregulation activities (e.g., sunning) and were not used for habitat use measurements. Vegetation sampling began following a three day acclimation period after capture and release of the tortoise to allow them to resume normal behaviors and activities. Vegetation sampling was conducted at 5-m intervals (hereafter referred to as sample plot) along the movement trails. At each sample plot, a 1-meter point frame with 10 pins spaced at 10-cm intervals was placed perpendicularly to the trail. Since the carapace length of adult tortoises is generally between 20 and 30 cm, the 1-m length point frame placed perpendicularly to the trail allowed measurement of the percent cover of plants within the foraging radius (50 cm) of an adult tortoise. Plants in contact with the pins on the point frame were identified to lowest taxon possible (genus or species) to determine the number of species encountered along each trail. For analyses, we combined the plants into broad vegetation categories (grasses, forbs, legumes, woody species, bare ground/litter). Percent cover (# of point intercepts/total points x 100) of each of the five categories was calculated for each trail. Although legumes (Fabaceae) are a subset of forbs, they were categorized as their own vegetation category in this study because of their high nutritional value and potential importance to gopher tortoise diets (Garner and Landers 1981). The total number of plant species for each trail was also included in the analysis.

20

Habitat Availability Measurements Each habitat use sample plot along the movement trail was paired with a habitat availability plot located 5 m away to represent habitat that was available to the tortoise but was not selected. Each habitat availability plot was located using a random number chart to select a compass direction (0-360°; omitting any points that landed back on the movement trail). Vegetation was characterized at habitat availability plots using the same protocol for habitat use plots along movement trails. Our measurements of habitat availability resulted in a paired availability measurement for each habitat use measurement. This definition allows habitat availability to change for each habitat use choice, so that each availability plot represents habitat that the tortoise could have chosen but did not at each sampling plot along the trail. This study design is useful for quantitative comparisons among habitat characteristics at small scales and does not rely on habitat in the study region that is not accessible if animals cannot move throughout the entire study area (Arthur et al. 1996). Data Analysis To determine if the distance traveled by tortoises differed among sites, total distance traveled per night per tortoise was log transformed to more closely approximate a normal distribution and then compared among sites using an Analysis of Variance (ANOVA). Differences of least squares means were used to determine significant differences between site means. Pearson product-moment correlation was used to determine the relationship between the vegetation predictor variables at the three study sites. Conditional logistic regression for matched pairs data was used to analyze gopher tortoise selection of habitat. Tortoise and day were the strata so that matched pairs of habitat use for each tortoise*day were compared to 21

habitat availability for each tortoise*day. A forward selection procedure was performed (significance for entry and staying in the model was p ≤ 0.05) to determine which combination of vegetation variables was useful for predicting gopher tortoise habitat use. All statistical analyses were performed using SAS v. 9.2 (SAS Institute©, Cary, NC). A factor analysis using a varimax orthogonal factor rotation (JMP v. 7. SAS Institute Inc., Cary, NC, 1989-2007) was used to examine whether habitat use plots had similar compositions to the habitat availability plots. Results The movement of 68 gopher tortoises was monitored during 2007 at three study sites in north and central Florida. Although the movement of each tortoise was monitored for 10-14 days post-capture at each site, the actual number of tracking days per tortoise ranged from 1-14 days (mean = 8 days) due to variation in individual activity patterns. The mean distance traveled per day for the entire study group (n = 68) was 56.9 m (sd = 79.9 m). The mean distance traveled at each study site 56.8 m for BS, 43.4 m for CentFL, and 71.2 m for OR (Table 2-1). The distance traveled at OR and CentFL varied significantly (p < 0.05) from one another (Table 2-2). We measured a total of 252 days of movement from 30 females (BS = 10; CentFL = 7; OR = 13), 256 days of movement from 34 males (BS = 13; CentFL = 10; OR = 11), and 23 days of movement of four juveniles (tortoises < 21.0 cm; all from the CentFL site). The overall mean distance traveled was 50.5 m for females (sd = 55.5, range = 0.3-313.2), 64.8 m for males (sd = 99.3, range = 0.2-637.8), and 40.0 m for juveniles (sd = 52.5, range = 0.8-180.2). The mean distance traveled did not differ significantly (p = 0.4) among males, females, or juveniles. The mean percent cover for each vegetation category was calculated for habitats used by tortoises and habitat available to tortoises at each study site (Table 2-3). The BS site had the highest percent cover of grass in both used and available plots (Table 2-3). The OR site had the 22

highest percent cover of bare ground/litter in both used and available plots (Table 2-3). The CentFL site had the highest percent cover of forbs in both used and available plots (Table 2-3). Percent cover of legumes and woody species was < 10% in both used and available plots at all three study sites. There was significant correlation (correlation coefficient ≥ 0.30) among several of the vegetation categories, but the relationships varied by study site. At all three sites grass was negatively correlated with bare ground/litter (Table 2-4 through 2-6). At BS grass was also negatively correlated woody species (Table 2-4) and at the CentFL site grass was negatively correlated with forbs (Table 2-5). At the BS site, woody species were positively correlated with legumes and at the OR site woody species were negatively associated with bare ground/litter. At the CentFL and OR sites, forbs were negatively correlated with bare ground/litter (Table 2-5 and 2-6). At CentFL, grass was negatively correlated with forbs (Table 2-5). BS was the only study site for which vegetation variables were found to significantly differ between use and availability. The analyses indicated that woody habitat was used significantly less by tortoises than was available (odds ratio of 0.93; 95% CI = 0.86-0.99). The factor analysis results indicated that the habitat use plots had similar compositions to the habitat availability plots. Discussion Gopher tortoise habitat use was compared to habitat availability in three habitats in north and central Florida to determine if habitat selection differed among habitat types. The OR study site is a natural sandhill that is actively managed by prescribed fire and had the highest plant diversity of the three study sites. The BS study site is a former pine plantation surrounded by a hardwood hammock and the groundcover is currently dominated by bahiagrass. The CentFL study site is a former citrus grove whose groundcover is dominated by Mexican clover and natalgrass. 23

At the OR and CentFL sites, gopher tortoise use of the five vegetation categories (bare ground/litter, forbs, grass, legumes, and woody species, and number of plant species) did not differ significantly from availability. The BS site was the only site in this study for which gopher tortoises were selective in their use of habitat. Although the mean percent cover of woody species in available plots at BS was only 2.54% (sd = 6.00), the habitat used by tortoises at this site had significantly lower (p < 0.05; odds ratio = 0.93, 95% CI = 0.86-0.99) woody vegetation. Although increased canopy cover has been associated with reduced herbaceous biomass in gopher tortoise habitats (Landers and Speake 1980), woody species in this study were measured only at the groundcover level. In our study, the negative association with woody species may be due to the fact that woody species generally contain high amounts of lignin that is difficult to digest as well as secondary metabolites (Bryant et al. 1991) that make them unpalatable to tortoises. Movement trails at BS were more obvious than at the other two sites with tortoises following small (< 0.5 m wide), visible paths through tall, dense mats of bahiagrass (C. L. White pers. obs.). Therefore, woody species may have been less common in habitats used by gopher tortoises at the BS site because the disturbance created by tortoises repeatedly using these paths decreased woody species establishment. The finding that tortoise habitat use did not differ from availability may have varying consequences in different sites or habitat types. For example percent cover of legumes was lower at the BS site compared to OR and CentFL. Since tortoises at this site used legumes in proportion to their availability, BS tortoises used legumes less than at the other two sites. Legumes are an important component of gopher tortoise diets as they are high in protein and more easily digested than grass and woody species. A reduction in the amount of these nutrient-

24

rich forage materials could potentially affect nutrient demanding processes such as reproduction and growth. The mean distance traveled by tortoises at the OR site was highest of all three study sites (Table 2-1). Distance traveled by tortoises can vary seasonally, with long-distance movements by males often occurring during late July and early August when active spermatogenesis is occurring (Taylor 1982). Females may also occasionally move long distances in search of suitable nesting sites, but at least one study found that females rarely relocated during the spring (McRae et al. 1981). The OR site was sampled in June, therefore, seasonal variation in movement patterns caused by the reproductive cycle were probably not the primary driving force behind the increased distances traveled by tortoises at this site. Tortoise home range sizes have been shown to increase as herbaceous biomass decreases (Auffenberg and Iverson 1979). Since the OR site had the highest percent of bare ground/litter, the forage at this site may have had a patchier distribution compared to the other two sites and could have resulted in tortoises moving greater distances in search of food. Alternatively, the CentFL site had the highest percent cover of forbs in both used and available habitat of the three study sites (Table 2-5). The distance traveled by tortoises at the CentFL site was significantly lower than the OR site (Tables 2-1, 2-2). Therefore, tortoises at this site may have traveled less because of increased availability of nutritional forage (e.g., forbs). Conclusions Gopher tortoises are selective in their use of resources and seasonal shifts in their diet have been documented in previous studies (Garner and Landers 1981, Mushinsky et al. 2003). Many gopher tortoise forage species are at their highest nutritional value and digestibility (i.e., relatively low fiber and high protein and minerals) during the spring. As grasses become more 25

fibrous in the summer, gopher tortoises shift to higher quality forage plants such as legumes when available (Garner and Landers 1981). The ability to shift resources throughout the active period (April–November) may have important consequences on gopher tortoises as demonstrated by the seasonal growth patterns of juveniles. Peak season growth of juveniles has been shown to occur during peak availability of legumes (summer). During the fall months when higher quality forage is less available, the gopher tortoise diet often consists largely of mature grasses and other plants high in fiber. This period also corresponds with a pronounced reduction in tortoise growth (Landers et al. 1982). On sites such as BS that are dominated by a single species of grass, tortoises may have to rely on bahiagrass as a primary food source throughout the year, whereas on a site such as CentFL tortoises may have prolonged access to higher quality forage such as forbs. Even though this study did not demonstrate that tortoises were selective in their use of habitat, comparing the nutritional status or reproductive status of animals at these sites would give insight into whether the differences in habitat were likely to have negative impacts on gopher tortoises. The conditional logistic regression tested whether tortoises were selective in their use of habitat, but an underlying assumption of this test was that the distribution of vegetation categories varied between habitat use and availability plots. The results from our factor analysis, however, demonstrated that the composition of the habitat use plots did not differ from availability plots (Figure 2-1 and 2-2), suggesting that gopher tortoises generally use resources in proportion to their availability. However, one limitation of this study was the low percent cover for most vegetation categories (Table 2-3). For example, gopher tortoises may prefer legumes, but since legumes made up < 5% of the groundcover at all three study sites (Table 2-3), the likelihood of them appearing in a significant number of habitat use or availability plots was low.

26

This study also recorded habitat used by gopher tortoises from continuous movement trails that were sampled at regular intervals. Although vegetation along the movement trails was undoubtedly eaten (bites marks were observed on vegetation during habitat sampling, C. L. White pers. obs.), some of the habitat plots may have created noise in the “used habitat” data because they represented habitat through which the gopher tortoise traveled on their way to another burrow or between foraging bouts. Although direct observations have been previously used to examine foraging habits in gopher tortoises, these studies have had either small sample sizes (n = 17 total foraging observations; Mushinsky et al. 2003) or combined direct observations with scat analyses (Garner and Landers 1981, MacDonald and Mushinsky 1988). One of the primary problems with using scat analyses to determine diet selection in herbivorous species is that less digestible, more fibrous material (e.g., wiregrass) may make up a larger proportion of scat material than more easily digestible material. Using direct observations to determine gopher tortoise foraging habits are of limited practicality because gopher tortoises are estimated to spend only about 10% of their time above ground (Auffenberg and Iverson 1979, Mushinsky et al. 2003). Since tortoises on the sites in this study generally returned to their burrows when humans approached, we attempted to collect foraging habit data by observing tortoises with binoculars from a tree stand in the summer of 2006. This method was unsuccessful for several reasons, including that the density of vegetation often made it impossible to determine the type of vegetation the tortoise was eating. Furthermore, gopher tortoises do not necessarily forage on a daily basis. Since we were only able to observe 1-3 burrows at a time, depending on the density of tortoises, even when we spent the entire day in the tree stand there were many days without a single observation.

27

In a study of habitat selection by wood turtles (Clemmys insculpta; Compton et al. 2002) activity areas were defined for individuals based on the distance from successive locations (determined by radiotelemetry performed every other day) to the running mean of the previous five locations. Once an empirically selected distance was met (e.g., 100 m), a new activity area and running mean were started. Activity areas with less than five points were considered travel points and were deleted. Using this technique for future studies of gopher tortoise habitat use may help researchers eliminate some of the noise in tortoise movement by separating travel points from areas where tortoises spend larger amounts of time. However, monitoring tortoise movement with the methods described in the present study is advantageous in that it provided continuous data on tortoise movements. One interesting finding from our data is that distance traveled at a well managed sandhill site (OR) was greater than in two ruderal habitats. This warrants exploration to determine its effect, if any, on gopher tortoise health, survival, or reproduction.

28

forb 0.5

legume

0.0

grass

spp litter -0.5 woody

-0.5

0.0

0.5

Factor 1

Figure 2-1. Rotated factor pattern of the six vegetation categories used to compare gopher tortoise habitat use and availability in 2007.

4

Factor2

3 2 1 0 -1 -2 -3 -2

-1

0

1

2

3

Factor1 Figure 2-2. Overlay plot of Factor 1 versus Factor 2 generated from the factor analysis. Blue dots = used habitat, Red dots = available habitat.

29

Table 2-1. Mean distance (m) traveled by gopher tortoises in 2007 at three study sites in north and central Florida. Site N Mean Distance (m) SD Minimum Maximum BS 23 56.8 69.8 0.2 477.2 CentFL 21 43.4 69.0 0.3 458.2 OR 24 71.2 99.1 0.4 637.8 Table 2-2. Results of comparison of distance traveled by gopher tortoises in 2007 at three study sites in north and central Florida. Site to Site Comparison Chi-square p-value BS verus CentFL 3.07 0.08 BS versus OR 0.96 0.33 CentFL versus OR 6.92 0.01 Table 2-3. Mean percent cover of vegetation categories in habitat used by gopher tortoises and habitat available to tortoises in 2007 at three study sites in north and central Florida. Site Vegetation Category Mean Used (SD) Mean Available (SD) BS Bare ground/litter 43.71 (20.86) 41.12 (17.42) Forb 2.43 (4.94) 3.02 (4.56) Grass 52.22 (20.86) 52.83 (18.44) Legume 0.30 (1.08) 0.48 (1.65) Woody 1.31 (3.49) 2.54 (6.00) Number of Species 3.70 (2.82) 3.90 (2.77) CentFL Bare ground/litter 47.75 (18.01) 44.03 (16.00) Forb 24.32 (16.64) 23.35 (14.69) Grass 25.56 (18.57) 30.26 (15.79) Legume 2.21 (4.97) 2.20 (5.72) Woody 0.05 (0.30) 0.06 (0.36) Number of Species 4.20 (2.85) 4.31 (2.50) OR Bare ground/litter 58.19 (13.76) 57.81 (15.14) Forb 1.63 (2.81) 1.45 (2.18) Grass 33.09 (11.39) 32.34 (12.39) Legume 2.11 (3.59) 2.08 (3.38) Woody 5.20 (6.02) 6.31 (7.13) Number of Species 7.45 (5.07) 7.25 (5.23) Table 2-4. Pearson correlation coefficients for vegetation predictor values at the BS site used in the comparison of gopher tortoise habitat use and availability in 2007. Bare ground/Litter Forb Grass Legume Woody Species Bare ground/Litter 1.00 -0.26 -0.93 -0.002 0.07 0.07 Forb -0.26 1.00 0.03 0.01 -0.08 0.20 Grass -0.93 0.03 1.00 -0.17 -0.33 -0.20 Legume -0.002 0.01 -0.17 1.00 0.38 0.28 Woody 0.07 -0.08 -0.33 0.38 1.00 0.24 Species 0.07 0.20 -0.20 0.28 0.24 1.00 30

Table 2-5. Pearson correlation coefficients for vegetation predictor values at the CentFL site used in the comparison of gopher tortoise habitat use and availability in 2007. Bare ground/Litter Forb Grass Legume Woody Species Bare ground/Litter 1.00 -0.42 -0.57 -0.14 0.12 0.06 Forb -0.42 1.00 -0.46 -0.11 -0.12 -0.02 Grass -0.57 -0.46 1.00 -0.07 -0.08 -0.08 Legume -0.14 -0.11 -0.07 1.00 0.16 0.14 Woody 0.12 -0.12 -0.08 0.16 1.00 0.36 Species 0.06 -0.02 -0.08 0.14 0.36 1.00 Table 2-6. Pearson correlation coefficients for vegetation predictor values at the OR site used in the comparison of gopher tortoise habitat use and availability in 2007. Bare ground/Litter Forb Grass Legume Woody Species Bare ground/Litter 1.00 -0.24 -0.82 -0.25 -0.48 -0.14 Forb -0.24 1.00 0.04 -0.02 0.07 0.07 Grass -0.82 0.04 1.00 0.02 -0.01 0.05 Legume -0.25 -0.02 0.02 1.00 -0.02 0.13 Woody -0.48 0.07 0.01 -0.02 1.00 0.11 Species -0.14 0.07 0.05 0.13 0.11 1.00

31

CHAPTER 3 EXAMINING HABITAT AT TWO SPATIAL SCALES TO DETERMINE ITS INFLEUNCE ON NUTRITIONAL STATUS IN GOPHER TORTOISES Introduction Extrinsic factors such as nutrition and the environment are well established as critical components of human health (Cohen et al. 2007, Fraker 2000, Fuduka et al. 2008, Jolly 2000, Keusch 2003). However, these factors are not as well understood in the context of wildlife health and conservation (Wobeser 2006). Determining which factors affect wildlife health will likely become increasingly important, particularly to species with declining populations. Habitat factors have been a driving force in the reduction of biodiversity worldwide (Wilson 1992) and one mechanism by which they may affect wildlife health is through increased disease transmission rates in small, fragmented populations (e.g., bighorn sheep; Boyce and Weisenberger 2005). Habitat loss and degradation could also alter wildlife health in more subtle ways. For example, a reduction in availability of food resources or specific nutrients could slowly decrease the health status of wildlife over time, resulting in increased susceptibility to infections or reduced reproduction (Wobeser 2006). Since anthropogenic habitat change will undoubtedly continue as the human population grows, the number of wildlife populations residing in altered habitats is likely to increase. Therefore, studies that address how the health status of animals is affected by their habitat will be important for future wildlife conservation efforts. Although it is often difficult to prioritize wildlife species in terms of conservation research needs, the gopher tortoise is both a keystone species (Eisenberg 1983) and ecosystem engineer (sensu Jones et al. 1994). The gopher tortoise’s role as an ecosystem engineer occurs through burrowing activities which enhance seed dispersal (MacDonald and Mushinsky 1988) and creation of open areas needed for early successional plant species in upland communities 32

(Kaczor and Hartnett 1990). Gopher tortoises are considered keystone upland species because their burrows provide protection from fire, predators, and temperature extremes for numerous upland vertebrate and invertebrate species (Jackson and Milstey 1989). Since the gopher tortoise likely has a disproportionate effect on overall upland ecosystem function, conservation efforts aimed at this species will benefit entire communities of upland plants and animals. Habitat changes such as urbanization, fragmentation, and degradation have been cited as the primary contributors to the > 80% decline in gopher tortoises over the past century (Auffenberg and Franz 1982). Loss in the quantity of gopher tortoise habitat has occurred through activities such as development of uplands for urban land use and subsequent fragmentation of the landscape. When development activities were associated with incidental take permits (allowing entombment or killing of tortoises on development sites in exchange for money used to protect occupied tortoise habitat elsewhere), the link between habitat loss and decreased tortoise populations was obvious. However, the effects of other measures used to mitigate habitat loss such as translocation and relocation of tortoises to new habitats are unknown for most projects (Dodd and Seigel 1991). The effect of releasing tortoises in habitats that are degraded by fire suppression (which can encourage hardwood tree growth, closed canopy structure, and reduced herbaceous groundcover) or land use activities such as livestock grazing or agriculture is difficult to determine since success of relocation/translocation projects is measured by the ability of populations to establish self-sustaining populations (Dodd and Seigel 1991). In a long-lived species such as the gopher tortoise, it may take years of monitoring to determine population declines. Examining the health of gopher tortoises and its relationship with habitat factors in different habitats may be an alternate, timelier method to determine whether tortoise health has been negatively impacted by habitat degradation. Therefore, the objective of

33

this study was to examine the effects of individual and population level habitat characteristics on gopher tortoise nutritional status. We tested the hypothesis that population level habitat characteristics indicative of siteaveraged forage availability can explain variability in tortoise nutritional status among populations. This hypothesis assumes constant population level habitat effects on nutritional status in tortoises. However, individual tortoises may use different habitats within the same sites. Therefore, individual habitat use may be a better predictor of variability in individual tortoise nutritional status. Methods Study Populations for Population Level Habitat Characteristics Population level habitat characteristics were measured during May–August, 2004–2006 on 11 study areas located in north and central Florida (Figure 3-1). The study sites at Big Shoals State Park (BS), Ordway-Swisher Biological Station (OR), and a privately owned Central Florida (CentFL) site are described in detail in Chapter 2. The study site at Camp Blanding/Fort Blanding (CB), Clay County, is a 16.0 ha reclaimed titanium mine that was reforested with row plantings of pine. The site is dominated by bahiagrass (mean percent cover = 33%), with the next most common plant, Elliott’s milkpea (Galactia elliottii), making up an average percent cover of only 3.6%. The gopher tortoise population density at CB is approximately 2.5 tortoises/ha. The study site at Cecil Field/Branan Field Wildlife and Environmental Area (CF), Duval County, Florida, is a 43.2 ha area primarily consisting of pine flatwoods and sandhill habitats with a tortoise population density of 4.2 tortoises/ha. Wiregrass (Aristida stricta), twinflower (Dyschoriste oblongifolia), shiny blueberry (Vaccinium myrsinites), soft milkpea (Galactia mollis), and silkgrass are among the most commonly occurring groundcover species at the site. Slash pine and turkey oak (Quercus laevis) are the predominant canopy species. The 34

study site at Flying Eagle Wildlife Management Area (FE), Citrus County, is a 39.9 ha improved pasture area. The most commonly occurring plant at the site is bahiagrass, which makes up over 50% of the groundcover. Other commonly occurring species include Elliott’s milkpea and horseweed (Conyza canadensis). The gopher tortoise population density at FE is approximately 2.5 tortoises/ha. The study site at Fort Cooper State Park (FC), Citrus County, is a 24.7 ha sandhill community. Due to years of fire suppression much of the site consists of a dense layer of oak shrubs that has reduced the herbaceous groundcover at this site. Therefore, bare ground and litter make up 54% of the ground cover and woody species make up 25% of the ground cover. The most commonly occurring ground cover plant is wiregrass (Aristida stricta and Sporobolus junceus; mean percent cover = 12.3%). The gopher tortoise population density at FC is approximately 4.4 tortoises/ha. The study area at Gold Head Branch State Park (GH), Clay County, is 48.1 ha sandhill community. The most commonly observed groundcover species at this site is wiregrass, silkgrass, and Florida alicia. The gopher tortoise population at GH is 1.8 tortoises/ha, the lowest for all sites included in this study. The study site at Green Swamp Wildlife Management Area (GS), Sumter County, is a 20.8 ha sandhill community. Wiregrass, twinflower (Dyschoriste oblongifolia), and silkgrass are among the most commonly occurring ground cover species at the site. The gopher tortoise population density at GS is approximately 4.4 tortoises/ha. The study site at Oldenburg Wildlife and Environmental Area (OL), Hernando County, is a 29.9 ha area that consists of two distinct habitat types. The first is a former sandhill area that is densely wooded due to fire suppression and the other habitat type is an open area that is consistently mowed due to a powerline that runs the length of the study area. The powerline area is dominated by bahiagrass. The two most common species in the woody area are wiregrass and Panicum grass species. The gopher tortoise population density at OL is approximately 3.1

35

tortoises/ha. The study site at Tenoroc Fish Management Area (TE), Polk County, is a 5.3 ha reclaimed phosphate mine. Grass species dominate this site, with the most common grass species including bahiagrass, smut grass (Sporobolus indicus), centipedegrass (Eremochloa ophiuroides), and cogongrass (Imperata cylindrica). The gopher tortoise population density at TE is approximately 23.4 tortoises/ha (M. B. Brown unpubl. data). Florida Fish and Wildlife Conservation Commission (FWC) permitted relocations of tortoises took place at CB, CentFL, FE, GS, and TE study sites. Unauthorized release of tortoises (i.e., by the public) are known to have taken place at the CF study site (Diemer-Berish et al. 2000, Wendland 2007) and are suspected at the FC, GH, and OL study sites (DiemerBerish et al. 2000, Wendland 2007). Study Populations for Individual Level Habitat Characteristics The study sites selected for determining individual habitat use by gopher tortoises were BS, CF, OR, and CentFL. Individual habitat use was monitored on OR and CentFL in both 2006 and 2007, at CF in 2006 only, and BS in 2007 only. Study sites were selected to provide the greatest potential differences in nutritional status of tortoises on the basis of preliminary observations, including differences in habitat characteristics, fecundity, and size class distributions (an indication of the survival rate of each life stage). Fecundity was based on the number of eggs per female, which was calculated based on all radiographed females of gravid size (>21.0 cm). The average clutch size is approximately 5-6 eggs for female gopher tortoises (Butler and Hull 1996, Diemer and Moore 1994). The lowest reported mean clutch size in the literature is 3.8 (Wright 1982) and the highest is 8.9 eggs (Ashton et al. 2007). The OR site is a sandhill community that is actively managed with prescribed fire, has average female fecundity (mean = 5.44 eggs/female), and typical or expected size class distributions (Figure 3-2; see Alford 1980). CentFL is a former citrus grove with large amounts 36

of two exotic plant species (natalgrass and Mexican clover), has above average female fecundity (mean = 7.63 eggs/clutch), and abnormal size class distributions (more juveniles and subadults than expected; Figure 3-2). CF is a pine flatwoods and sandhill community, has below average female fecundity (mean = 3.54 eggs/clutch), and typical or expected size class distributions (Figure 3-2). However, CF also has a high seroprevalence of Upper Respiratory Tract Disease (URTD, ~75%; Wendland 2007). Because disease may impact the nutritional status and overall health of individuals within a population, during 2007 BS was sampled instead of CF. BS is a former pine plantation surrounded by a hardwood hammock and the groundcover is dominated by bahiagrass. BS has below average female fecundity (mean = 2.50 eggs/clutch), low seroprevalence of URTD (< 10%; Wendland 2007) and abnormal size class distributions (very few juveniles; Figure 3-2). The mean clutch size of females at the BS site (2.5 eggs) was lower than the lowest reported clutch of 3.8 eggs (Wright 1982), which could potentially be the result of differences in forage availability. Field Techniques and Assessment of Tortoise Nutritional Status During May–August, 2004–2006, 592 tortoises were captured at the 11 study sites using pitfall traps set directly in front of burrow openings. Complete physical examinations and health assessments were performed on each tortoise. Examinations included an evaluation of the animal’s behavior, attitude and posture, assessment of all external body systems (eyes, nares, oral cavity, skin/toenails, shell, etc.), detailed descriptions and drawings of any abnormalities or clinical signs of disease, complete morphometric measurements, and photographic documentation of the face, carapace, plastron, and any abnormalities. The body condition (BC) of all captured tortoises was assessed by calculating the ratio of body mass to shell volume ((body mass/(carapace length x maximum shell height x maximum shell width); Nagy et al. 2002). 37

Using appropriate sterile technique, blood (up to 2.5 cc) was collected from the brachial vein of tortoises and submitted to the Clinical Pathology Laboratory, College of Veterinary Medicine, University of Florida for plasma biochemistry panels. Eighteen blood parameters were used to assess nutritional status included: anion gap (ANION), alkaline phosphatase (ALP), aspartate aminotransferase (AST), blood urea nitrogen (BUN), calcium (Ca), cholesterol (CHOL), chloride (CHL), creatinine kinase (CK), gamma glutamyl transferase (γGT), glucose (GLU), packed cell volume (PCV), phosphorus (P), potassium (K), magnesium (Mg), sodium (Na), total carbon dioxide (TCO 2 ), total protein (TP), and uric acid (URIC). Diagnostic assays for URTD in tortoises are relatively well-established and include direct culture of M. agassizii, detection of mycoplasmal chromosomal DNA by polymerase chain reaction (PCR), and detection of anti-mycoplasma antibodies by enzyme-linked immunosorbent assay (Brown et al. 2002). Culture/PCR and ELISA were performed on all sampled tortoises. However, given the fastidious growth requirements, slow growth rates, and low sensitivity for culture and PCR for M. agassizii (Wendland 2007), only serological status as determined by ELISA (sensitivity = 0.98, specificity = 0.98; positive titer ≥ 64) was used to assess disease status in our study. Individual Level Habitat Measurements The habitat used by 135 animals was measured at BS, CentFL, CF, and OR during 2006– 2007 according to protocol described in detail in Chapter 2. Fluorescent powder dye packs constructed with nylon pouches were attached to identification holes located on the rear carapacial scutes. Fluorescent dye trails were tracked nightly for 10-14 days using a hand held UV light. Tortoise movement trails were recorded as continuous polylines (one waypoint/second) in ArcPad v. 6.2 (Environmental Systems Research Institute, Redlands, CA). Vegetation sampling was conducted at 5-m intervals along the movement trails using a 1-meter 38

point frame placed perpendicularly on the trail. Percent cover (# of point intercepts/total points x 100) of vegetation was calculated for each trail. Plants were combined into 5 broad vegetation categories (bare ground/litter, forbs, grasses, legumes, and woody species) for analysis. Population Level Habitat Measurements Surveys were conducted annually during 2004–2006 on the 11 study sites to determine habitat structure and vegetation composition at the study site level. Twenty-five randomly selected plots at each site were used to determine canopy coverage and percent ground cover of individual species (M. B. Brown unpubl. data). Plants were then combined into the same five vegetation categories described above (bare ground/litter, forbs, grasses, forbs, legumes, and woody species) for analysis. Species richness was determined for each plot by determining the number of species in the plot. Data Analysis The first dataset was made up of 18 health-related blood parameters measured on 592 tortoises and vegetation parameters measured at the population level. The second dataset was a subset of tortoises (n = 135) from the first dataset for which the 18 health-related blood parameters did not change but vegetation parameters were measured at the individual level. Therefore, both of our datasets consisted of multiple response variables (i.e., 18 health-related blood parameters) and multiple predictor variables (i.e., five vegetation categories measured at either the individual level or population level). We performed separate canonical correlation analyses on each dataset, the first analysis (dataset n = 592) examined the relationship between tortoise blood parameters and population-level habitat parameters and the second analysis (dataset n = 135) examined the relationship of gopher tortoise blood parameters with individuallevel habitat parameters. To reduce the number of blood parameters initially entered into the analyses, we performed three preliminary canonical correlation analyses by grouping blood 39

parameters into subsets that roughly corresponded to nutrition (TP, Ca, P, GLUC, CHOL, Mg), hydration (BUN, URIC, Na, K, CHL, ANION, TCO 2 ), and health (ALP, AST, γGT, CK, PCV) models. We also included morphometric variables BC and Clength (carapace length) in each of the three a priori models. From the preliminary canonical correlation analyses, we selected blood parameters with the largest structural loadings between the blood parameters and their canonical variables to form the final models. The final model for the canonical analysis of gopher tortoises blood parameters and population-level habitat parameters (dataset n = 592) included ALP, AST, BC, Ca, CK, γGT, GLUC, K, Mg, Na, P, TCO 2 , TP, and URIC. The final model for the canonical analysis of gopher tortoise blood parameters and individual-level habitat (dataset n = 135) parameters included BUN, URIC, Na, ANION, PCV, TP, Clength, and BC. The CF site had a high seroprevalence for URTD and disease could potentially impact nutritional status. Therefore, we also performed a canonical analysis of gopher tortoise blood parameters and individual level habitat on a dataset that excluded all data from the CF site (dataset n = 101). We also performed an analysis of variance (ANOVA) on the dataset with population level habitat characteristics (n = 592) to determine if site (11 study sites), sex (male, female), or year (2004–2006), and interactions among these variables (site*year, and site*sex) had a significant effect on the tortoise blood parameters. Juveniles were omitted from this analysis because they were not sampled on every site and we were, therefore, missing site*sex combinations for those sites. The ANOVA used blood parameters selected by the final population level canonical correlation model (ALP, AST, BC, Ca, CK, γGT, GLUC, K, Mg, Na, P, TCO 2 , TP, and URIC) and included post-hoc testing using a Tukey adjustment of the means comparison. To better approximate a normal distribution, the following variables were log transformed before being

40

entered into the ANOVA: ALP, AST, CK, γGT, P, and URIC. All statistical analyses described above were done using SAS v. 9.2 (SAS Institute©, Cary, NC). Results Individual Habitat Use Summary Statistics The habitat used by individual gopher tortoises was measured at the BS, CentFL, CF, and ORD study sites. Gopher tortoise use of grass was highest at BS (mean = 51.89, sd = 17.43) and lowest at CentFL (mean = 18.92, sd = 10.05). Forb use was low at all study sites (< 5%; Table 3-1) except at CentFL, where the mean percent cover on movement trails was 25.1% (Table 3-1). Legume use was ≤ 1% at all sites except ORD, where the mean percent cover on movement trails was 5.2%. Woody species use was greatest at the CF site (mean = 10.95, sd = 10.01). Litter use was similar at all 4 sites (~50%; Table 3-1). Habitat use of the five vegetation categories was similar for males, females, and juveniles (Table 3-2). Population Level Habitat Summary Statistics Habitat measurements at the population level were measured at BS, CB, CentFL, CF, FC, FE, GH, GSW, OL, OR, and TE. Mean percent cover of grass over the three year study period (2004–2006) at the 11 sites ranged from 10.8% (FC site) to 60.2% (BS site; Table 3-3). Mean percent cover of forbs was < 8% at all study sites except for the TE and CentFL sites, where the means were 13.2% and 18.9%, respectively. Mean percent cover of legumes was < 6% at all study sites except for the FE site, where the mean was 18.2% due to the large amounts of Elliott’s milkpea at this site. Mean percent cover of woody species ranged from < 1% at the BS site to > 20% at the FC site. The amount of bare ground/litter was > 30% for all 11 sites except FE where the mean was 18.8%. The canopy cover at the sites ranged from 1.5% (TE) to 46.8% (FC). GS and CF had the highest species richness among the study sites, and CB (mean = 5.80, sd = 2.77) and FE (mean = 5.14, sd = 2.43) had the lowest (Table 3-4). 41

Canonical Correlation and ANOVA Results Results of the canonical correlation analysis indicated that the variance of the blood parameters captured by population-level habitat measurements (3%) and the R2 (0.34) for the first canonical variate was low. The correlation analysis indicated that the variance of blood parameters captured by individual-level habitat (dataset n = 135) use (9%) and the R2 (0.47) for the first canonical variate was also low. The variance of blood parameters captured by individual-level habitat use (7%) and the R2 (0.50) for the first canonical variate were also not significantly improved by excluding CF from the individual level habitat dataset (dataset n = 101). Therefore, neither individual habitat use or population level habitat data as measured in this study was able to account for a significant amount of variability in gopher tortoise health related blood parameters. The ANOVA performed on the dataset with population level habitat characteristics (n = 592) indicated that there was a significant interaction (p ≤ 0.05) between site and year for Ca, K, γGT, Na, TCO 2 , TP, and URIC (Table 3-5). There was a significant interaction (p ≤ 0.05) between site and sex for ALP, Ca, P, and TCO 2 (Table 3-5). There was also a significant interaction between year and sex for K. For those blood parameters for which the above interactions were not significant, there were significant differences (p ≤ 0.05) among sites for AST, CK, GLUC, and Mg (Table 3-5). There were significant differences (p ≤ 0.05) among years for ALP, AST, CK, and P (Table 3-5). There were significant differences (p ≤ 0.05) among males and females for CK and TP (Table 3-5). Summary statistics were calculated for blood parameters selected by the final population level canonical correlation model by site (Table 3-6), sex (Table 3-7), and year (Table 3-8).

42

Discussion In this study we examined the relationship between habitat parameters (at the individual habitat use level or at the population site level) and gopher tortoise nutritional status. The canonical correlation results indicate that the habitat measurements used in this study (at the population or individual level) were not able to account for a significant amount of the variability in gopher tortoise blood parameters. One explanation for this finding is that our habitat measurements may have been too coarse to capture differences in nutritional availability important at the tortoise’s level. For example, our population level habitat measurements indicated that percent cover of grass was approximately 30% for a pine flatwoods and sandhill site (CF), former citrus grove (CentFL), former sandhill site with a mowed powerline in the center of habitat (OL), and reclaimed titanium mine (CB). However, the dominant grass species were wiregrass and silkgrass at CF, natalgrass at CentFL, and bahiagrass at OL and CB. Therefore, even though the percent cover of grass and other vegetation categories measured at the population level may have been similar, the nutritional availability among the sites may still have differed due to differences in species availability or factors such as rainfall, season, or soil nutrient availability. These other factors may have contributed to the “site” effect that was observed for many of the blood parameters in the ANOVA (Table 3-5). Individual tortoise habitat use measurements were performed in addition to the site level habitat measurements in this study. However, due to the cryptic nature of tortoise behavior (they usually returned to their burrows if we came too close), we were not able to document the plant species and amount of specific plant species consumed by tortoises. Instead, we used habitat measurements taken along gopher tortoise movement trails to determine habitat used by tortoises. Tortoises were continuously monitored for 10-14 days and foraging undoubtedly occurred along the movement trails. In fact, tortoises with dye packs were occasionally observed 43

foraging while we performed other field activities during the daytime (e.g., checking pitfall traps and examining new tortoises; C. L. White pers. obs.). However, the measurements along the movement trails still only represented a “snapshot” in time of habitat being used by the tortoises at each site and contained vegetation through which the tortoise traveled on their way to or from foraging areas and other burrows. Effects of Reproduction In gopher tortoises, vitellogenesis (yolk accumulation) and ovarian follicular growth begins in late summer to early fall, slows over the winter months, and is completed the following spring (Iverson 1980). Eggs are then shelled in the oviducts in mid-April and generally laid in May or June. Mating primarily occurs in late summer and early fall (Taylor 1982, Diemer and Moore 1994). In this project, the study sites were sampled in approximately the same order each year to accommodate simultaneous data collection on fecundity in gopher tortoises at five of the 11 sites (BS, CentFL, CF, GH, OR; see Chapters 4 and 5). Since reproduction can cause significant changes in plasma biochemistry parameters, our sampling scheme likely contributed to the significant effects of site and site*sex interactions for several blood parameters. For example, increases in Ca and P levels have been shown to coincide with egg production and vitellogenesis in desert tortoises (Gopherus agassizii; Campbell 1996). Similarly, P levels in this study were highest in female gopher tortoises on the sites that were sampled in May and June for fecundity (except for CF; Figure 3-3). Although somewhat consistently higher in females than males among all sites (Figure 3-4), the effects of reproduction on Ca levels are more difficult to interpret than P from our results since females from sites sampled in May–June (CentFL, GH, OR), July (OL), and August (CB) had elevated Ca levels. The significant effect of site on Mg levels in gopher tortoises may also be partially explained by the study’s sampling scheme. Since the largest store of Mg is in the skeleton 44

(Robbins 1993), when bone dissolution occurs to provide Ca and P for egg calcification Mg levels may also increase. Indeed, the three sites with the highest mean plasma levels of Mg (BS, CentFL, OR) were sites sampled in May–June when females have just finished calcifying eggs (Table 3-6). However, since there was not a sex effect on Mg levels and males obviously do not lay eggs, other factors may also have contributed to variations among sites for this blood parameter. The enzyme CK can be used to test for muscle cell damage or exertion and increases can result from trauma, disease, or resisting restraint during blood collection (Campbell 1996). We captured male and female gopher tortoises using the same techniques, but females had significantly higher CK levels than males (Table 3-7). Although the two sites with the highest CK levels (CB and TE; Table 3-6) were both reclaimed mining sites, they were also sampled in late July and August during 2004–2006 making it difficult to determine whether increased CK levels were seasonal effect or caused by other aspect of the sites not measured in the current study. Effects of Food Intake and Rainfall Patterns The enzymes ALP and AST are widely distributed in the body and plasma activity of these enzymes and are not considered to be organ specific in reptiles. ALP activity is high in the kidney of reptiles, but significant plasma increases do not occur with renal disease because the enzyme is released in urine rather than blood from damaged renal cells (Campbell 1996). However, levels of ALP and AST may be useful as a general indicator of overall increases in metabolic activity in gopher tortoises because increases in these enzymes have been observed in desert tortoises after they have emerged from brumation and have begun eating (Dickinson et al. 2002). In our study, we found significant effects of site and year on levels of ALP and AST in gopher tortoises, which could indicate annual or site level differences in forage availability that 45

resulted in differences in metabolic activities for gopher tortoises. For example, both ALP and AST levels were the lowest during 2006 (Table 3-8), which was the driest of the three study years. Our results indicated that there was a significant effect of site on gopher tortoise GLUC levels. Increased GLUC levels in desert tortoises have been attributed to increased nutrient intake (Christopher et al. 1999). Although our population level habitat data did not indicate that overall forage availability was lower on the TE site compared to other sites, post-hoc tests showed that GLUC levels at the TE site were significantly lower than most of the other study sites (Table 3-9). The normal blood glucose concentration of most reptile ranges between 60100 mg/dL, and the mean for this site was 53.9 mg/dL (Table 3-6) suggesting that some animals at this site may have been hypoglycemic which can be caused by malnutrition, hepatobiliary diseases, or septicemia (Campbell 1996). Rainfall patterns have been correlated with hydration and nutritional status in desert tortoises. In years with below average rainfall, desert tortoise URIC and TP levels decreased and electrolytes (Na and K) increased (Dickinson et al. 2002). During this study, rainfall was significantly below average in 2006 (Florida Climate Center: http://coaps.fsu.edu/climate_center, accessed 15 September 2009), leading to drought conditions statewide. However, since the study sites were located throughout north and central Florida, rainfall may also have differed among individual sites within years. Therefore, it is possible that difference in rainfall patterns contributed to the site or site*year effects for several of the blood parameters in this study. For example, although gopher tortoise K levels were at their highest during 2006 (Table 3-8), differences in specific plant availability or rainfall among sites may have contributed to the site*year effect for this parameter. Most sites remained within the normal plasma K range for

46

turtles (2-6 mEg/L; Campbell 1996), except for 2006 when two sites exceeded this range (GS and TE; Figure 3-5). Gopher tortoise Na levels were increased in 2005 and 2006 for most sites compared to 2004 (Figure 3-6), but the majority of individuals at all sites were within the normal range for tortoises (120-150 mEq/L; Campbell 1996). TP levels were also at their lowest for several sites in 2006, but there was a considerable amount of overlap among years and sites (Figure 3-7), and tortoises remained within the normal range of 3-7 g/dL (Campbell 1996) for all three study years. γGT is an enzyme found in particularly high concentrations in the liver, bile ducts, and kidney and functions to transfer amino acids across the cellular membrane (Latimer et al. 2003). Although interpretation of this variable in relation to health or forage availability is difficult due to the lack of information about what factors influence this enzyme in tortoises, increased γGT levels in other species are generally associated with increased liver activity. Increased γGT levels were observed on almost all sites during the driest study year (2006; Figure 3-8). Since at least one other blood parameter suggested that gopher tortoises may have been nutritionally stressed by the dry conditions in 2006 (increased K levels), increases in γGT levels in gopher tortoises may be an indication of liver glycogen breakdown as a result of poor forage availability. In desert tortoises, ingestion of nutrients and increased hydration status may be an explanation for increases in ALP, AST, GLUC, P, TP, and URIC levels (Christopher et al. 1999). The results from our canonical correlation analyses indicated that habitat factors could not explain a significant amount of the variability in these variables. As discussed above, variability in site or regional level factors such as rainfall or reproduction may have influenced many of these variables. We could detect no consistent pattern, however, that would help determine the cause for the significant effects of site, year, site*year on these particular parameters.

47

Another factor that may also have contributed to the site, year, or site*year effects in our blood parameters was differences in management practices among the sites. Five of the study sites were burned during the study period (BS = winter 2004–2005, CF = summer 2006, GH = winter 2004–2005, GSW = winter 2004 –2005, and OR = winter 2004–2005) and one of the sites was roller chopped (OL = winter 2003–2004). Fire and roller chopping can benefit groundcover species by decreasing hardwood and saw palmetto (Serenoa repens) growth and can increase nutrient cycling through the consumption of aboveground biomass and litter (Abrahamson and Hartnett 1990). Since these management techniques can also cause plant species to send up new shoots that may be higher in nutrient and water availability, they may have contributed to a site*year interaction for many of the blood parameters commonly used to diagnose hydration status (e.g., Na, K, URIC, TCO 2 ; Christopher et al. 1999). However, the timing and intensity of management techniques can influence their effects on herbaceous plant growth and it may take multiple growing seasons to see peak production (Buckner and Landers 1979). Therefore, even though the dates that management practices took place during this study were known, they likely interacted with other site level characteristics such as time since last burn or rainfall, making it difficult to determine their specific effects on tortoise blood parameters. Effects of Disease Disease can negatively impact the nutritional status and overall health of individuals and may be reflected in various biochemistry parameters. For example, desert tortoises with clinical signs of URTD had significantly (p ≤ 0.05) lower levels of P and significantly (p ≤ 0.05) higher levels of Na and CHOL than clinically healthy tortoises (Jacobson et al. 1991). In our study, P levels seemed to be primarily correlated with the reproductive cycle, except for at the CF site, which incidentally also had a very high seroprevalence for URTD (Wendland 2007). Therefore, it seems plausible that disease could affect the P levels in gopher tortoises. 48

Site, season, and sex-specific references intervals for blood parameters have been established for desert tortoise populations in the Mojave Desert (Christopher et al. 1999). When examining animals with clinical disease, researchers found that laboratory tests had low sensitivity and high specificity for assessing morbidity and mortality in desert tortoises (Christopher et al. 2003). No direct association between blood data and specific disorder (URTD, cutaneous dyskeratosis, and urinary tract disease) was found, but researchers noted marked seasonal variation and a tendency for tortoises to have intermittent but multiple abnormalities over time (Christopher et al. 2003). Establishing reference intervals for gopher tortoise blood parameters will increase our understanding of the relationship between disease and specific blood parameters in gopher tortoises. Conclusions Nutritional studies for wildlife have traditionally involved foraging observations. Indeed several studies have examined the foraging ecology of gopher tortoises using scat, direct observations, and occasionally boluses taken directly from the digestive tract (Garner and Landers 1981, MacDonald and Mushinsky 1988, Mushinsky et al. 2003). However, in a study that relied solely on direct observations to determine tortoise diets, only 17 observations were used from a 21 month period (Mushinsky et al. 2003), illustrating these methods are limited by the relatively small amount of time that tortoises spend away from their burrows (Auffenberg and Iverson 1979) and their tendency to return to their burrows once they are disturbed (Mushinsky et al. 2003). Direct observation methods are also limited by vegetation structure and density that can severely limit observer visibility of animals. Variable tortoise densities also can limit the number of animals visible at any one time. Scat analyses are limited by the skill level needed for identification of partially digested food material and by the absence of highly digestible matter in (McInnis et al. 1983). Based on these limitations, these methods may not be 49

appropriate for large-scale studies that attempt to determine how multiple sites with varying habitat parameters may influence an organisms’ nutritional status. Seasonal and annual variations in desert tortoise hematology and plasma biochemistry parameters have been correlated with seasonal and annual variations in rainfall (Christopher et al. 1999, Dickinson et al. 2002), perhaps because rainfall determines annual forage abundance in desert habitats (Wallis et al. 1999). Therefore, our assumption that availability of certain forage species or vegetation groups may be correlated with blood parameters in gopher tortoises is valid. However, this study demonstrated that significant differences among sites exist for many gopher tortoise blood parameters that cannot be accounted for by broad habitat measurements (e.g., percent cover of grasses, forbs, and legumes). This finding is important for management of gopher tortoises because it suggests that gopher tortoise health may be affected by multiple site level variables such as management techniques, soil nutrient availability, and species composition. Due to the extended reproductive cycle of gopher tortoises and the potential effects of other seasonal events such as brumation, future studies may need to concentrate on a smaller number of study sites and measure blood parameters and habitat and nutrient availability (from forage and soil samples) multiple times per year to further examine the relationship between gopher tortoise health and habitat quality.

50

Figure 3-1. Location of 11 gopher tortoise study sites in Florida.

51

Figure 3-2. Size class distributions of four study sites selected for individual habitat measurements. Typical or expected size class distributions were present at the OR and CF sites. The BS and CentFl sites had abnormal size class distributions with very few juveniles found at the BS site and more juveniles and subadults than expected present at the CentFL site.

52

Figure 3-3. The effect of site*sex on P levels (ln transformed) in gopher tortoises sampled during 2004–2006 from 11 study sites in north and central Florida. Sexcode 1 = females and sexcode 2 = males.

Figure 3-4. The effect of site*sex on Ca levels in gopher tortoises sampled during 2004–2006 from 11 study sites in north and central Florida. Sexcode 1 = females and sexcode 2 = males. 53

Figure 3-5. The effect of site*year on K levels in gopher tortoises sampled during 2004–2006 in north and central Florida.

Figure 3-6. The effect of site*year on Na levels in gopher tortoises sampled during 2004–2006 in north and central Florida.

54

Figure 3-7. The effect of site*year on TP levels in gopher tortoises sampled during 2004–2006 in north and central Florida.

Figure 3-8. The effect of site*year on γGT levels (ln transformed) in gopher tortoises sampled during 2004–2006 in north and central Florida.

55

Table 3-1. Summary of individual gopher habitat use at four study sites in north and central Florida, 2006 and 2007. Habitat Parameter (% Cover) N Mean SD Median Minimum Maximum Bare ground/Litter BS (2007 only) 22 43.51 17.41 40.19 11.58 73.16 CentFL 37 54.15 14.80 52.22 25.80 95.00 CF (2006 only) 34 50.00 15.78 55.15 22.31 72.86 ORD 42 48.95 15.19 48.29 20.00 78.33 Forb BS (2007 only) 22 2.99 2.14 2.59 0.00 8.42 CentFL 37 25.13 12.30 24.50 0.00 52.70 CF (2006 only) 34 2.66 1.50 2.66 0.00 5.46 ORD 42 4.69 5.90 1.95 0.00 25.00 Grass BS (2007 only) 22 51.89 17.43 57.68 18.96 80.00 CentFL 37 18.92 10.05 20.00 0.00 43.33 CF (2006 only) 34 35.88 11.02 37.49 12.54 57.14 ORD 42 35.43 9.04 35.29 15.00 55.56 Legume BS (2007 only) 22 0.48 0.88 0.00 0.00 3.87 CentFL 37 1.01 2.28 0.00 0.00 11.05 CF (2006 only) 34 0.70 1.37 0.20 0.00 7.50 ORD 42 5.19 5.36 3.08 0.00 21.67 Woody BS (2007 only) 22 1.14 1.37 0.63 0.00 4.50 CentFL 37 0.18 0.68 0.00 0.00 3.90 CF (2006 only) 34 10.95 10.01 9.50 0.00 37.19 ORD 42 5.74 3.39 5.26 0.00 18.15

56

Table 3-2. Summary of individual gopher habitat use by sex in four study sites in north and central Florida, 2006 and 2007. Habitat Parameter (% Cover) N Mean SD Median Minimum Maximum Litter Female 56 49.74 14.97 48.55 22.31 78.33 Male 61 50.51 15.30 50.00 11.58 81.67 Juvenile 18 47.22 20.26 40.91 20.00 95.00 Forb Female 56 9.98 13.63 2.80 0.00 52.70 Male 61 9.10 10.32 5.00 0.00 40.00 Juvenile 18 9.38 13.16 3.81 0.00 45.60 Grass Female 56 32.53 16.30 32.70 0.00 71.79 Male 61 34.12 15.89 31.87 1.70 80.00 Juvenile 18 35.92 13.43 40.33 10.00 57.14 Legume Female 56 2.08 3.47 0.65 0.00 14.64 Male 61 2.33 4.19 0.56 0.00 21.67 Juvenile 18 1.67 4.15 0.00 0.00 16.67 Woody Female 56 5.58 7.85 2.36 0.00 37.19 Male 61 3.55 5.23 1.88 0.00 26.64 Juvenile 18 6.44 7.83 2.24 0.00 27.92

57

Table 3-3. Summary of population level habitat availability 11 study sites in north and central Florida, 2004–2006. Summary values represent percent cover. Habitat Parameter N Mean SD Median Minimum Maximum Bare ground/Litter BS 70 33.36 16.17 32.50 0.00 72.25 CB 70 60.94 16.04 63.50 27.50 89.38 CentFL 70 47.83 17.50 47.50 13.13 97.50 CF 70 43.87 27.71 47.50 2.50 100.00 FC 70 59.77 14.71 60.00 25.00 95.63 FE 70 18.83 19.80 15.00 0.00 88.75 GH 70 64.77 14.44 66.25 24.00 98.13 GS 70 33.46 18.74 30.00 5.00 91.25 OL 70 53.71 25.45 56.25 2.50 98.75 OR 70 66.75 18.18 65.00 11.38 98.13 TE 70 34.68 17.32 35.69 0.00 70.00 Forb BS 70 4.90 5.32 3.63 0.00 25.00 CB 70 1.07 2.87 0.13 0.00 16.88 CentFL 70 18.92 13.30 18.0 0.00 71.25 CF 70 3.93 3.72 2.50 0.00 18.00 FC 70 1.87 1.78 1.31 0.00 8.75 FE 70 5.58 9.96 0.88 0.00 55.00 GH 70 3.86 3.19 3.13 0.00 13.75 GS 70 7.65 5.65 8.00 0.00 25.00 OL 70 6.36 8.78 3.25 0.00 53.13 OR 70 1.73 1.88 0.88 0.00 8.13 TE 70 13.16 10.96 10.81 0.00 40.63 Grass BS 70 60.16 16.77 60.00 22.63 94.38 CB 70 30.64 16.41 25.94 0.75 70.00 CentFL 70 27.71 15.78 25.94 0.00 62.50 CF 70 27.54 21.12 21.56 2.63 86.88 FC 70 10.85 9.47 8.69 0.00 40.00 FE 70 56.16 27.64 61.88 2.75 93.75 GH 70 14.99 10.97 13.13 0.63 55.63 GS 70 31.30 14.65 31.25 1.88 67.50 OL 70 27.65 25.85 18.75 0.00 90.00 OR 70 19.85 14.01 17.50 1.38 76.25 TE 70 41.75 23.58 40.00 0.25 95.63 Legume BS 70 0.16 0.71 0.00 0.00 5.75 CB 70 3.91 4.02 3.19 0.00 15.50 CentFL 70 1.38 2.40 0.13 0.00 15.13 CF 70 1.72 2.26 0.81 0.00 12.75 FC 70 1.97 2.96 0.88 0.00 18.75 FE 70 18.16 23.31 5.38 0.00 79.38 58

Table 3-2. Continued. Habitat Parameter N Legume GH 70 GS 70 OL 70 OR 70 TE 70 Woody BS 70 CB 70 CentFL 70 CF 70 FC 70 FE 70 GH 70 GS 70 OL 70 OR 70 TE 70 Canopy BS 70 CB 70 CentFL 70 CF 70 FC 70 FE 70 GH 70 GS 70 OL 70 OR 70 TE 70 GH 70 GS 70 OL 70 OR 70 TE 70 Woody BS 70 CB 70 CentFL 70 CF 70 FC 70 FE 70 GH 70

Mean

SD

Median Minimum Maximum

3.26 4.34 3.65 3.20 5.50

4.82 5.27 6.92 3.73 7.18

1.50 2.44 0.94 1.50 3.31

0.00 0.00 0.00 0.00 0.00

25.00 27.50 44.38 15.75 38.13

0.90 1.95 1.83 4.86 1.49 4.08 18.33 15.61 21.38 12.98 2.06 3.89 8.95 8.83 15.99 12.25 6.10 9.95 5.80 6.30 1.56 4.71

0.00 0.00 0.00 13.63 19.38 0.44 7.00 14.13 0.75 4.00 0.00

0.00 0.00 0.00 0.75 0.00 0.00 0.00 0.00 0.00 0.00 0.00

9.38 28.13 26.25 67.50 52.50 21.25 38.13 56.25 42.50 30.00 28.75

32.31 23.57 19.83 32.11 46.77 10.60 29.27 5.97 36.18 41.96 1.50 3.26 4.34 3.65 3.20 5.50

21.17 14.91 19.88 18.11 19.16 19.14 15.22 12.70 25.61 12.70 11.06 4.82 5.27 6.92 3.73 7.18

30.78 24.21 13.50 30.04 46.20 3.53 29.00 2.11 31.51 41.20 1.50 1.50 2.44 0.94 1.50 3.31

0.00 0.00 0.00 0.00 1.00 0.00 5.00 0.00 2.08 15.08 0.00 0.00 0.00 0.00 0.00 0.00

85.35 59.35 78.00 87.17 98.41 76.00 66.00 44.13 94.64 86.18 48.00 25.00 27.50 44.38 15.75 38.13

0.90 1.95 1.83 4.86 1.49 4.08 18.33 15.61 21.38 12.98 2.06 3.89 8.95 8.83

0.00 0.00 0.00 13.63 19.38 0.44 7.00

0.00 0.00 0.00 0.75 0.00 0.00 0.00

9.38 28.13 26.25 67.50 52.50 21.25 38.13

59

Table 3-4. Mean species richness at 11 study sites in north and central Florida, 2004–2006. Site N Mean SD Median Minimum Maximum BS 70 6.95 3.14 7.0 1 17 CB 70 5.80 2.77 6.0 1 14 CentFL 70 7.31 3.79 7.0 2 18 CF 70 18.06 8.66 20.0 3 36 FC 70 16.19 7.56 18.0 2 29 FE 70 5.14 2.43 5.0 1 11 GH 70 15.71 6.56 17.0 4 29 GS 70 22.24 11.10 25.0 1 39 OL 70 13.17 7.32 12.5 2 28 OR 70 13.67 6.77 16.0 2 27 TE 70 8.53 3.34 8.0 2 16 Table 3-5. Results of ANOVA on the effect of study site (11 study populations), month (May– August), year (2004–2006), and sex, site*month, site*year, site*sex, and month*sex interactions on selected blood parameters that were used in the population habitat level canonical correlation. Parameter df F P ALP (U/L) Site 10 23.52