1 SURVIVAL PROBABILITIES AND DENSITY OF FOUR SYMPATRIC SPECIES OF FRESHWATER TURTLES IN FLORIDA By GABRIELLE ELAIN E HRYCYSHYN A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2007
2 2007 Gabrielle Elaine Hrycyshyn
3 To the turtles of Wekiwa Springs State Park
4 ACKNOWLEDGMENTS I would like to thank the park rangers and staff at Wekiwa Springs State Park for helping me with this study. The park ki ndly provided free camping and use of its canoes, and their support has been essential in continuing this stud y. I would like to thank the organizations that have provided funding for this project over the years in the form of grants for food and travel for volunteers: Penn State University, th e University of Florida, the Un iversity of North Florida and Freed-Hardeman University. Permits for this st udy were obtained yearly from WSSP. Animal use was approved by the University of Florid a, IACUC Protocol number E307. All volunteers were approved to handle these wild turtles throug h their various institutions and trained by senior volunteers. Mostly I would like to thank the over 100 volunteers who have contributed to this study over the yearswithout their help this st udy would not have been possible. A special thank you to my most dedicated volunteers: Jo sh Brown, Jon Freezer, Eric Munscher, Jessica Munscher, and Andy Weber. Finally, my deepest gr atitude to the founders of this study, Dr. J. Brian Hauge and Dr. Brian Butterfield, and my major advisor Dr. Karen A. Bjorndal, and my committee members, Dr. Alan B. Bolten and Dr. Benjamin Bolker.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ..........6 LIST OF FIGURES................................................................................................................ .........7 ABSTRACT....................................................................................................................... ..............8 CHAPTER 1 INTRODUCTION................................................................................................................. .10 2 MATERIALS AND METHODS...........................................................................................12 Study Area and Design.......................................................................................................... .12 Statistical Methods and Analyses...........................................................................................13 3 RESULTS...................................................................................................................... .........21 4 DISCUSSION................................................................................................................... ......35 Density........................................................................................................................ ............35 Biomass........................................................................................................................ ...........37 Survival....................................................................................................................... ............40 Realized Population Growth Rate..........................................................................................43 Sex Ratios..................................................................................................................... ..........45 Habitat Alteration............................................................................................................. ......48 APPENDIX A CORMACK-JOLLY-SEBER MODELS...............................................................................50 B PRADELS MODELS OF THREE SPECIES.......................................................................51 C PRADELS MODELS OF Pseudemys nelsoni ......................................................................52 D REGRESSIONS OF ESTIMATES........................................................................................53 LIST OF REFERENCES............................................................................................................. ..55 BIOGRAPHICAL SKETCH.........................................................................................................61
6 LIST OF TABLES Table page 2-1 Sampling periods, number of days spent ca tching turtles, and numbers of the four most common species of turtles caught.............................................................................18 3-1 Geometric means followed by 95% conf idence intervals of six estimates........................26 3-2 Comparison of geom etric mean densities..........................................................................27 A-1 Cormack-Jolly-Seber models for each of the four most commonly caught species of turtles at Wekiwa Springs State Pa rk from March 2000 to November 2005.....................50 B-1 Pradels reverse-time survival and lambda models for P peninsularis S minor minor and S odoratus ........................................................................................................51 C-1 Pradels reverse-time survival and lambda models for P nelsoni .....................................52 D-1 Regression line propert ies of seven estimates...................................................................53
7 LIST OF FIGURES Figure page 2-1 The lagoon at Wekiwa Springs State Park, Apopka, Florida............................................19 2-2 The numbering system used to notch the marginal scutes on Emydid turtles...................20 3-1 Estimated annual recapture probabilities...........................................................................28 3-2 Estimated density (turtles/ha)............................................................................................2 9 3-3 Mean body mass (kg)........................................................................................................ .30 3-4 Estimated biomass (kg/ha).................................................................................................3 1 3-5 Estimated apparent survival probabilities..........................................................................32 3-6 Estimated realized population growth rate ( )...................................................................33 3-7 Percentages of male and female turtles..............................................................................34
8 Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science SURVIVAL PROBABILITIES AND DENSITY OF FOUR SYMPATRIC SPECIES OF FRESHWATER TURTLES IN FLORIDA By Gabrielle Elaine Hrycyshyn August 2007 Chair: Karen A. Bjorndal Major: Interdisciplinary Ecology Turtles are important, and of ten neglected, components of their ecosystems, and many turtle species are endangered. I studied four sp ecies of common freshwater turtles over a five year period in Wekiwa Springs State Park (WS SP), Apopka, Florida. Thes e four turtle species were the Peninsula Cooter ( Pseudemys peninsularis ), Florida Red-bellied Cooter ( P. nelsoni ), Loggerhead Musk Turtle ( Sternotherus minor minor ) and Stinkpot ( S. odoratus ). For each species I estimated annual recapt ure probabilities, population dens ity, biomass, annual survival probabilities and realized populati on growth rate using capture-m ark-recapture data from March 2000 to November 2005. My density estimates showed significan t declines throughout the study for both Pseudemys species, and non-significant in creases in density for both Sternotherus species. These findings are not directly corroborated by the estimated re alized population growth rates, which suggest that P. peninsularis populations are increasing non-significantly, and P. nelsoni populations are declining non-significantly. I postu late that the apparent disagreement between the realized population growth rate and density estimates aris es from the need to account for temporal variability in estimates of rea lized population growth rate.
9 Overall my findings suggest that none of th ese turtle species are in danger, and the apparent declines in both Pseudemys species appear to stabiliz e toward the end of the study. There are numerous environmental factors that could impact the future health of the spring ecosystem and these turtle populations. Th ese include invasion by the aquatic plant Hydrilla verticillata which necessitates management activities that many disturb the turtles, increased nitrate pollution and increased wa ter usage and therefore decreas ed spring output. I recommend monitoring of these turtle populations to ensure that should population declines occur appropriate management actions can be taken.
10 CHAPTER 1 INTRODUCTION Turtles and tortoises are long-liv ed and iteroparous with late sexual maturity. Changes in population structure and abundance often take years to manifest and it can be difficult to ascertain the causes of population fluctuations (Gibbons, 1997; Bjorndal et al., 2005). Studies of various turtle populations have found that high survival rates am ong juveniles and adults are the key to maintenance of healthy populations (Cro use et al., 1987; Congdon et al., 1993). However, determining these variables for any given turtle population in the wild can be difficult and require numerous years of study (Gibbons, 1997; Heppell, 1998; Bjornd al et al., 2005). Despite the difficulties of long-term studies turtles must be studied thoroughly. A disproportionate number of turtle species worl dwide are facing population declines, possible local and regional extirp ation, and extinction (Gibbons et al., 2000). The life history traits of turtles make their populations le ss able to rebound from the eff ects of habitat destruction or alteration, invasive species, and exploitation (Garber and Burger, 1995; Marchand and Litvaitis, 2004). To understand the dynamics of these decl ining populations, information on non-impacted or less-impacted populations of the same or similar species is necessary. Mark-recapture studies are one of the prefe rred methods for studying long-lived species such as turtles, as these methods can help elucidate the population demographic variables of survivorship, maturation, migra tion and fecundity. These four f actors interact to produce the observed population demographics, or th e realized population growth rate ( ) (Williams et al., 2001). Coupled with information on population size, realized population growth rates can be used to evaluate population trends in the past and provide a basis for estimating future population trends.
11 In this study I present five years of mark-r ecapture data on four species of freshwater turtles in Wekiwa Springs State Park, Florida. Th is spring, while protected, is still influenced by human water usage in the area, decreased water up take by the aquifer that feeds it (Saint Johns River Water Management District, 2006), and n itrate pollution from nearby septic tanks and fertilized lawns (Toth and For tich, 2002). The goal of this study wa s to characterize the current demographics of the sub-adult a nd adult portion of these four fres hwater turtle populations in the spring. I determined the density of each turtle species and average adult survival probabilities. This information, coupled with estimates of realized population growth rates, provided information on the relative health and stability of these populations. I analyzed sex ratios and size-structure of each population to provide fu rther insights into the composition of each population. Finally, I estimated total biomass for each species. These data provide useful insights into the demographics of natural pop ulations of freshwater turtles.
12 CHAPTER 2 MATERIALS AND METHODS Study Area and Design Turtles were captured in the swimming area and main lagoon of Wekiwa Spring (Figure 21), located in Wekiwa Springs State Park (WSSP), Apopka, Florida (Orange and Seminole counties) and the adj acent spring run (28o42 N, 81o27 W). WSSP was purchased by the state of Florida in 1969; the spring has been as a recreational area si nce 1941. Rock Springs Run State Preserve (RSRSP), purchased by the state of Florida in 1983, borders WSSP and creates a protected area of over 8,500 ha. The area represents a typical central Florida spring with wet lowlands surrounding the spring, surrounded by dry sand hill uplands mainta ined by frequent ground fires. The area directly around the spring, which expels appr oximately 164 million L of water a day, has been modified with concrete walls and ladders to facilitate swimming and ge neral recreational use. This small swimming area then opens into an ap proximately 1.32 ha lagoon that is used for canoeing and fishing but not swimming. At the end of the lagoon the water moves into the Wekiwa Springs Run, in about 0.65 ha of which tu rtles were also captured. This spring run is approximately 1 km long and then joins Rock Springs Run to become the headwaters of the Wekiva River. This study used the robust design format proposed by Pollock (1982) wherein primary periods of capture (denoted by m onth and year in this study) were broken into secondary periods (days within each month). The study of the freshwater turtle community in WSSP began in March 2000 and continued semi-annually throug h November 2005 (Table 2-1). Seven native turtle species were captu red: Peninsula Cooter ( Pseudemys peninsularis ), Florida Red-bellied Cooter ( P. nelsoni ), Loggerhead Musk Turtle ( Sternotherus minor minor ), Stinkpot ( S.
13 odoratus ), Florida Softshell ( Apalone ferox ), Florida Snapping Turtle ( Chelydra serpentina osceola ), and Florida Chicken Turtle ( Deirochelys reticularia chrysea ). The first four species were by far the most abundant and are the subject of this paper. Additionally, 9 introduced Redeared Sliders ( Trachemys scripta elegans ) were captured and rem oved from the study site. Nearly all of the turtle s were caught by hand while snorkeling; of the four main species only 2 S. odoratus were trapped. All captured turtles were pl aced in a canoe and brought to a central location for processing. Maximum straight measurements we re recorded for carapace length, plastron length, carapace width and shell height Individuals were then sexe d using common secondary sex characteristics following Erns t et al. (1994). After mass was measured, each individual was marked with a variation of Cagles (1939) met hod, wherein each marginal scute excluding the nuchal was used (Figure 2-2). Marks were made with a handheld hack saw, making small and shallow but clearly distinguishable line marks. Any notable characteristics of each turtle were recorded, including unusual markings or damage. Total time from capture to release varied but averaged 1 hour. Turtles were released from a central location back into the lagoon, except for those individuals captured in the spring run, whic h were returned to the spring run and kept separate from turtles captured in the lagoon at all times. No turt les died while being captured and handled. Statistical Methods and Analyses I characterized the populations of the four most common species of turtles caught at WSSP; small sample sizes and virtually no recap tures of the other th ree species prevented analysis. Results presented are only for sub-adul ts and adults; juveniles were determined by a minimum plastron length cut-off for each species and these individuals were dropped from the study due to insufficient numbers for statistical or modeling analysis. Mi nimum plastron length
14 was determined as the smallest common plastron length at which males exhibit secondary sex characteristics and can be differentiated from fema les. These turtles are not necessarily sexually mature, merely distinguishable as bel onging to one sex or the other. For the Pseudemys species the cut-off was a plastron length of 100 mm (18 P. peninsularis and 6 P. nelsoni removed), 40 mm for S. minor minor (24 individuals removed) and 35 mm for S. odoratus (0 individuals removed). After removal of juveniles, sex ratios were calculated which do not include multiple recaptures of the same individual within a sampli ng period. Sex ratios were determined for each sampling period as well as for the entire study, and significance was analyzed using a 2 test with a null hypothesis that males and females occurred in equal numbers. Yates (1934) correction for continuity was applied because d.f. = 1, and si gnificance was assessed at the P < 0.05 level. Cormack-Jolly-Seber models were run usi ng program MARK to determine apparent survival ( ) and recapture ( ) rates for each population between each primary period (White and Burnham, 1999). Goodness-of-fit tests were cond ucted using the program U-CARE, which also provided an estimate of dispersion parameter (Choquet et al., 2003). After adjusting for overdispersion, a total of 9 possible Cormack-Jolly -Seber (CJS) models was generated, which included apparent survival and r ecapture as either time specific (t) (where time indicates primary period), time specific with the addition of diffe rences between the sexes but no interaction (t+sex), and time specific with interaction with differences in sex (t*sex) (Appendix A). These 9 models were then averaged based on model wei ghts using the model-averaging procedures in MARK. This procedure provided averaged estim ates of apparent survival and recapture probabilities between each primary period, with unconditional standard errors (Burnham and Anderson, 2002; Iverson et al., 2006).
15 The resulting recapture probabilities and the ac tual numbers of male and female sub-adult and adult turtles captured duri ng each primary period were su bsequently used to estimate abundance as described by Williams et al. ( 2002) and Iverson et al. (2006). Abundance per primary period i was calculated using Equation 2-1, where Ni is abundance at time i, determined by the ratio of individuals captured at time i, ni, divided by the recapture rate at time i, pi. Standard error of abundance was approximated using Equation 2-2. i i in Np (2-1) 1/2 2 2 (1) SE()var() iii ii iinnp Np pp (2-2) Density was estimated in two steps to account for differences in densities between the lagoon and spring run habitats. First, the estim ated abundance of each sex was multiplied by the percentage of individuals captured in the lagoon ( P. peninsularis 96.1%; P. nelsoni 84.5%; S. minor minor 43.1%; S. odoratus 85.5%), and then divided by 1. 32 ha for the total size of the lagoon. Second, the estimated abundance of eac h sex was multiplied by the percentage of individuals captured in the spri ng run (100% percent captured in lagoon) and then divided by 0.65 ha for the size of the spring run sampled. These estimates were then added together to provide total density estimates within the sampling area. Biomass for each species was calculated for each sex based on the mean mass of that sex during the sampling period, multiplied by the sex-specific density estimates of that period (Congdon et al., 1986). This method accounts for the large sexual dimorphism in the Pseudemys species and assumes that the sex ratio of turtle s captured within a sampling period reflects the actual sex ratio of the population during the sampling period, as thes e are reflected in the density estimates used. Coefficients of variation (CV, or relative error) were ca lculated using Equation
16 2-3, in which Di stands for the density of that sex at time i and Mi stands for the average mass of that sex at time i. 22 SE()SE() CV ii Sex iiDM DM (2-3) Pradels (1996) reverse time capture-mark-r ecapture models were implemented in MARK to model realized popu lation growth rate ( ). Analysis of averaged recapture rates from CJS models showed time but not sex dependence for all species; therefore recapture rate was held as time specific. Averaged CJS apparent survival rates for P peninsularis S. minor minor and S. odoratus showed variation over time but no signifi cant effect of sex, so for these species apparent survival rate was held as time specifi c. Therefore, 5 possible Pradels survival and models were generated (Appendix B). P nelsoni showed some significa nt differences between male and female survival rates as well as variation over time so survival rates were held as both time and sex specific, resulting in a total of 15 potential Pradels survival and models (Appendix C). Realized population gr owth rate was not constrained. The generated models were then averaged as previously described to provide estimates of realized population growth rate between each primary period with unconditional standard errors. Geometric means were calculated for recaptur e probabilities, density, biomass, apparent annual survival probabilities, and realized populat ion growth rate. The 95% confidence intervals for these means were generated using the natural log of each set of estimates and the standard error of the mean of the log-tr ansformed estimates. Significant differences between means were assessed using overlap of 95% conf idence intervals. Over all trends in the above estimates were assessed using inverse-variance we ighted linear regression, usi ng the inverse of the standard error of each estimate in the da taset, in SigmaPlot 9.0 (Systat Software Inc., 2004) (Appendix D).
17 Overall trends in sex ratios (males/female) were assessed using linear regression weighted by the natural log of the number of individuals that each ratio was derived from, in SigmaPlot 9.0 (Appendix D).
18 Table 2-1. Sampling periods, number of days sp ent catching turtles, a nd numbers of the four most common species of turtles caught (exc luding within-sampling period recaptures) per sampling period at Wekiwa Springs Stat e Park. Numbers caught are divided into males (M) and females (F); not all males a nd females are sexually mature. Individuals that could not be assigned to a sex have b een excluded from this table. Total values represent the number of turt les excluding recaptures ever captured throughout the entire course of the study. Species P. peninsularis P. nelsoni S. minor minor S. odoratus Sampling period # days M F M F M F M F March 2000 4 37 37 15 8 0 0 2 1 May 2000 3 46 24 22 20 4 3 16 12 March 2001 4 57 59 10 12 2 4 4 3 March 2002 5 66 62 32 31 2 3 2 2 May 2002 6 54 47 42 54 7 3 13 7 March 2003 5 48 37 16 18 17 23 33 30 May 2003 6 48 50 9 21 15 13 26 23 March 2004 5 49 56 20 25 39 47 28 16 May 2004 5 58 44 34 57 36 43 8 13 March 2005 6 42 37 21 24 27 24 31 15 August 2005 5 34 36 23 22 14 9 16 10 November 2005 3 13 22 27 27 10 9 8 7 Total 57 272 249 146 161 137 154 157 117
19 Figure 2-1. The lagoon at Wekiwa Springs State Park, Apopka, Fl orida where turtles for this study were marked, recaptured and releas ed from March 2000 to November 2005.
20 Figure 2-2. The numbering system used to notch the marginal scutes on Emydid turtles, which typically have 12 marginal scutes on each side, disregarding the nuchal scute. Kinosternid turtles typically have 11 margin al scutes on each side and thus 400 is the highest possible scute number. [Adapted from Cagle, F. R. 1939. A system for marking turtles for future iden itification. Copeia 1939:170-173.]
CHAPTER 3 RESULTS From March 2000 through November 2005, the total number of individuals captured for each species, followed by the percentage of males was: 534 P peninsularis (52.6%), 317 P nelsoni (48.3%), 295 S minor minor (47.8%), 276 S odoratus (57.2%). Numbers of individual turtles captured per primary period are shown in Table 2-1. Total number of individuals of less common species included 11 A. ferox (only captured from March 2000 to March 2002), 11 C. serpentina osceola and 3 D. reticularia chrysea Annual recapture probab ilities from the CJS models were clearly time-specific for the Pseudemys species (Figure 3-1), and showed no effect of sex in any species. Geometric annual recapture probability means indicate that P. peninsularis was most likely to be recaptured, and S. odoratus the least likely (Table 3-1). Each species showed trends toward increasing annual recapture probabilities with the exception of male S. minor minor which exhibited a nonsignificant decreasing trend (A ppendix D). The only species w ith significantly increasing recapture probabilities over the course of the study was P. nelsoni (male slope = 0.008, R2 = 88.3%, P < 0.05; female slope = 0.007, R2 = 86.8%, P < 0.05). Density estimates were calculated using the recapture probabilities for each sex from the CJS models. Although recapture pr obabilities did not differ signi ficantly between the sexes in most cases (Figure 3-1) the actual number of tu rtles captured of each sex did vary (Table 3-1). Inverse-variance weighted linear regression showed that P. peninsularis density declined significantly throughout the study for both males (slope = -2.171, R2 = 95.6%, P < 0.05) and females (slope = -2.143, R2 = 95.7%, P < 0.05). The geometric mean density of sub-adult and adult of P. peninsularis was 280.4 (95% C.I. 225.1, 349.3) turt les/ha for the entire study period, and there were no significant differences between density of males and females (Table 3-1).
Density of P. peninsularis peaked from March 2001 through Ma y 2002 (1-3, Figure 3-2), with a geometric mean density for this time period of 421.5 (95% C.I. 338.1, 525.4) turtles/ha. The geometric mean density after May 2002 (4-9, Figure 3-2) was 228.7 (95% C.I. 206.7, 253.2) turtles/ha, significantly lower than found at the start of study. Most P. peninsularis were captured in the lagoon, with only 3.93% individuals ev er captured in the sp ring run (Table 3-2). The density of P nelsoni declined significantly throughout my study for both males (slope = -1.579, R2 = 90.0%, P < 0.05) and females (slope = -2.617, R2 = 95.6%, P < 0.01). The geometric mean density of sub-adult and adult of P nelsoni was 175.3 (95% C.I. 119.8, 256.4) turtles/ha for the entire study period. The lowest estimate of density occurred during the March 2005 to August 2005 sampling period (Figure 3-2). Ma les tended to be found at lower densities than females, although the difference in their ove rall densities was not si gnificant (Table 3-1). More P. nelsoni were found in the spring run (15.46%) th an found for its congener (Table 3-2). Density estimates for S. minor minor varied widely due to low recapture probabilities, and did not exhibit any significant linear trends for males or females. Geometric mean density for this species was 327.2 (95% C.I. 153.4, 698.2) turt les/ha for the entire study period (Table 3-1). The highest population density reported occurred during the May 2003 to March 2004 sampling period, immediately following one of the lowest estimates from March 2003 to May 2003 (Figure 3-2). The majority of S. minor minor were captured in the spring run, accounting for 56.95% of the total individuals captured, which is considerably higher than any other species addressed in this study (Table 3-2). S. odoratus density estimates also varied wi dely throughout the study (Figure 3-2), although a non-significant increas ing trend was observed for both males and females (Appendix D). The geometric mean density was 320.5 (95% C. I. 202.7, 507.0) turtles/ha, virtually identical
to the estimate for its congener (Table 3-1). Approximately 14.49% of the S odoratus captured were found in the spring run, similar to proportion of P. nelsoni (Table 3-2). Mean masses of males and females (Figur e 3-3) showed that each species except S. minor minor exhibits sexual dimorphism, with females be ing significantly larger than males (Table 31). P. peninsularis females were significantly larger than P. nelsoni females, although males of either species were not significantly di fferent in mean mass. Both sexes of S. odoratus were significantly smaller than S. minor minor of either sex. Over the course of the study males and females of both Pseudemys species declined in mean mass. This declining trend was significant in P. peninsularis (male slope = -0.006, R2 = 99.6%, P < 0.01; female slope = -0.008, R2 = 99.7%, P = 0.05) and non-significant in P. nelsoni (Appendix D). Neither species of Sternotherus displayed any temporal trends in me an mass over the course of the study. Estimated biomass (Figure 3-4) for each species of turtle was calculated by multiplying density estimates (Figure 3-2) by mean mass of that sex during each sampling period (Figure 33). P peninsularis makes the greatest biomass contributi on to the turtle community due to the large estimated population size and th e greater mass of females, and only P. peninsularis had significantly different geometric mean bioma sses between the sexes (Table 3-1). Biomass declined significantly for males and females of both P. peninsularis (male slope = -4.712, R2 = 93.3%, P < 0.05; female slope = -10.202, R2 = 94.7%, P < 0.05) and P. nelsoni (male slope = 3.154, R2 = 86.5.3%, P < 0.05; female slope = -7.694, R2 = 88.7%, P < 0.05) over the course of the study. There were no linear trends for biomass of S. minor minor and a non-significant increase in biomass of S. odoratus throughout the study (Appendix D). Apparent annual survival probabi lities (Figure 3-5) did not sh ow significant sex effects in any species (Table 3-1), although the models (t*sex) (t) and (t*sex) (t+sex) accounted for
90.0% of the averaged weight for th e survival values calculated for P nelsoni (Appendix A). Geometric mean apparent annual survival probab ility point estimates were variable and ranged from 45.2% for female S. minor minor to 72.4% for male P. peninsularis (Table 3-1). Linear trends in apparent annual survival probabilities were difficult to ascertain for all species due to high standard errors. Linear trends were non-significant for all species except P. nelsoni which showed a significant declining trend for fe males throughout the study (slope = -0.014, R2 = 96.5%, P < 0.05), and a non-significant decline in male survival probabilities (Appendix D). Pradels reverse-time survival and models were used to as certain realized population growth rate over the course of the study (Fi gure 3-6). Based on the CJS models, recapture probabilities were held as time-specific for all sp ecies, and apparent surviv al probabilities were held as time-specific for each species except P nelsoni while varied (Appendix B and C). No species had significantly differen t geometric mean realized popul ation growth rates for males and females, and every species except S. odoratus had geometric means with an upper 95% confidence interval that reached or exceeded 1. 000, the criterion for population growth (Table 31). Every species except P. nelsoni exhibited a non-significa nt trend of increasing values throughout the study, while P. nelsoni had non-significantly declining values (Appendix D). Despite this trend, P nelsoni attained the highest single value of the Pseudemys species, which occurred during the March 2004 to May 2004 primary period, of approximately 6.81 (standard error 4.59) for males and females. Although percentages of males and females vari ed between sampling pe riods (Figure 3-7), only one sampling period for P. peninsularis two sampling periods for the P. nelsoni and one sampling period for the S. odoratus were significantly different fr om a 1:1 sex ratio. Over the duration of the study, consid ering each individual once, P. peninsularis P. nelsoni and S. minor
minor had male/female ratios not significantly di fferent from 1:1 (1.09, 0.91 and 0.89, respectively). S. odoratus had a significantly unequal sex ratio of 1.34 males/females ( 2 = 5.551, d.f. = 1, P < 0.05). Both species of Pseudemys had non-significant temporal trends towards female-dominated sex ratios (Appendix D). Sternotherus species showed no temporal linear trends in sex ratio.
26Table 3-1. Geometric means followed by 95% confiden ce intervals of six estima tes: annual recapture ( ) probability, density, mean mass, biomass, apparent annual survival ( ) probability, and realized population growth rate ( ). Estimates are from four species of turtles captured at Wekiwa Springs State Park from March 2000 to November 2005. Species P. peninsularis P. nelsoni S. minor minor S. odoratus Estimate Units Male Female Male Female Male Female Male Female Annual probability -0.274 0.235, 0.320 0.269 0.230, 0.315 0.249 0.146, 0.425 0.226 0.131, 0.392 0.155 0.110, 0.219 0.116 0.087, 0.154 0.065 0.043, 0.098 0.087 0.061, 0.124 Density turtles/ ha 143.2 114.5, 179.2 136.6 108.9, 171.2 72.3 48.5, 107.7 101.5 68.7, 149.9 148.2 73.4, 299.0 175.2 76.9, 399.1 201.3 125.6, 322.7 114.6 71.1, 184.6 Mean mass kg 1.762 1.662, 1.867 4.225 4.059, 4.399 1.872 1.732, 2.024 3.041 2.708, 3.416 0.110 0.100, 0.121 0.106 0.094, 0.120 0.046 0.042, 0.051 0.067 0.062, 0.072 Biomass kg/ha 246.4 185.8, 326.9 572.3 443.2, 739.0 135.2 86.3, 211.9 298.3 191.5, 464.8 16.0 7.5, 34.0 20.4 8.4, 49.4 9.1 5.4, 15.2 7.5 4.4, 12.7 Annual probability -0.724 0.583, 0.900 0.650 0.506, 0.835 0.471 0.282, 0.786 0.612 0.344, 1.000 0.546 0.314, 0.949 0.452 0.230, 0.889 0.681 0.417, 1.000 0.461 0.187, 1.000 -0.740 0.478, 1.147 0.668 0.445, 1.003 0.727 0.335, 1.575 0.755 0.348, 1.638 1.281 0.923, 1.777 1.280 0.923, 1.776 0.880 0.799, 0.968 0.880 0.800, 0.969
27 Table 3-2. Comparison of geometric mean dens ities and 95% confidence intervals for four species of turtle captured at Wekiwa Springs State Park from March 2000 to November 2005 in the lagoon and spring run. Lagoon Spring run Species Male Female Male Female P. peninsularis 132.3 (105.7, 165.5) 126.1 (100.6, 158.1) 11.0 (8.8, 13.7) 10.5 (8.4, 13.1) P. nelsoni 52.7 (35.4, 78.6) 74.0 (50.1, 109.3) 19.6 (13.1, 29.2) 27.5 (18.6, 40.6) S. minor minor 40.2 (19.9, 81.1) 47.5 (20.9, 108.3) 108.0 (53.5, 217.9) 127.7 (56.1, 290.9) S. odoratus 149.8 (93.5, 240.1) 85.3 (52.9, 137.4) 51.6 (32.2, 82.6) 29.3 (18.2, 47.3)
28 Figure 3-1. Estimated annual recapture probabiliti es and associated standard error of turtle populations in Wekiwa Springs State Pa rk from March 2000 to November 2005, obtained from averaging results of Cormack-Jolly-Seber models in Program MARK Models were constrained to be time-specific with or without the effect of sex for both survival and recapture probabilities. Estim ates include only turtles above minimum size requirements for sexing (see text). Th e x-axis numbers denote sampling periods: 1) May 2000, 2) March 2001, 3) March 2002, 4) May 2002, 5) March 2003, 6) May 2003, 7) March 2004, 8) May 2004, 9) March 2005. A) Pseudemys peninsularis B) P. nelsoni C) Sternotherus minor minor and D) S. odoratus Black circles with narrow error bar caps indicate females, and white triangles with wide error bar caps represent males.
29 Figure 3-2. Estimated density (tur tles/ha) and associated standard error of turtle populations in Wekiwa Springs State Park from March 2000 to November 2005. Estimates include only turtles above minimum size requirement s for sexing (see text). The x-axis numbers denote sampling periods: 1) Ma y 2000 to March 2001, 2) March 2001 to March 2002, 3) March 2002 to May 2002, 4) May 2002 to March 2003, 5) March 2003 to May 2003, 6) May 2003 to March 2004, 7) March 2004 to May 2004, 8) May 2004 to March 2005, 9) March 2005 to August 2005. A) Pseudemys peninsularis B) P. nelsoni C) Sternotherus minor minor and D) S. odoratus Black circles with narrow error bar caps indicate females, and white triangles with wide error bar caps represent males.
30 Figure 3-3. Mean body mass (kg) and standard error of male and female turtles caught at Wekiwa Springs State Park from March 2000 to November 2005. Estimates include only turtles above minimum size requirement s for sexing (see text). The x-axis numbers denote sampling periods: 1) Marc h 2000, 2) May 2000, 3) March 2001, 4) March 2002, 5) May 2002, 6) March 2003, 7) May 2003, 8) March 2004, 9) May 2004, 10) March 2005, 11) August 2005, 12) November 2005. A) Pseudemys peninsularis B) P. nelsoni C) Sternotherus minor minor and D) S. odoratus Black circles with narrow error ba r caps indicate females, and white triangles with wide error bar caps represent males.
31 Figure 3-4. Estimated biomass (kg/ha) with st andard error of turtle populations in Wekiwa Springs State Park from March 2000 to November 2005. Estimates include only turtles above minimum size requirements fo r sexing (see text). Mean mass of each sex was used to determine contribution of each sex to biomass to account for sexual dimorphism in size. The x-axis values denot e sampling periods as specified in Figure 3-2. A) Pseudemys peninsularis B) P. nelsoni C) Sternotherus minor minor and D) S. odoratus Black circles with narrow error bar caps indicate females, and white triangles with wide error bar caps represent males.
32 Figure 3-5. Estimated apparent survival probabilities and associ ated standard error of turtle populations in Wekiwa Springs State Pa rk from March 2000 to November 2005, obtained from averaging results of Cormack-Jolly-Seber models in Program MARK. Models were constrained to be time-specific with or without the effect of sex for both survival and recapture probabilities. Estim ates include only turtles above minimum size requirements for sexing (see text). The xaxis values denote sampling periods as specified in Figure 3-2. A) Pseudemys peninsularis B) P. nelsoni C) Sternotherus minor minor and D) S. odoratus Black circles with narrow error bar caps indicate females, and white triangles with wide error bar caps represent males.
33 Figure 3-6. Estimated realiz ed population growth rate ( ) and associated standard error of turtle populations in Wekiwa Springs State Pa rk from March 2000 to November 2005, obtained from averaging results of Pradels reverse-time survival and lambda models in Program MARK. Models were constr ained to have time-specific recapture probabilities for all species (no effects of sex), while survival was held to be timespecific for all species except P nelsoni where survival was time-specific, with or without the effects of sex, and lambda was not cons trained. Estimates include only turtles above minimum size requirements fo r sexing (see text). The x-axis values denote sampling periods as sp ecified in Figure 3-2. A) Pseudemys peninsularis B) P. nelsoni C) Sternotherus minor minor and D) S. odoratus Black circles with narrow error bar caps indicate females, and white triangles with wide error bar caps represent males.
34 Figure 3-7. Percentages of male and female turtles caught at We kiwa Springs State Park from March 2000 to November 2005. Only turtle s above minimum size requirements for sexing are included (see text). A) Pseudemys peninsularis B) P. nelsoni C) Sternotherus minor minor and D) S. odoratus An (*) in the column indicates that a 2 test for deviation from a 1: 1 sex ratio was significant at the P < 0.05 level, d.f. = 1, using Yates (1934) correction for continuity. Black bars indicate females, white bars represent males.
35 CHAPTER 4 DISCUSSION Density The interaction of the population demographi c parameters of survival, emigration, fecundity and maturation rate produce observable population density changes (Ozgul et al., 2004; Reid et al., 2004). Wildlife managers often seek estimates of population density, yet accurately estimating this parameter remains difficult (Gibbons et al., 1997). I was able to produce reliable estimates of population density, based on the numb er of individuals captured in each primary period and the rigorous estimation of recapture ra te provided by the Cormack-Jolly-Seber model in MARK (McDonald and Amstr up, 2001; Iverson et al., 2006). The geometric mean density for sub-adult and adult P. peninsularis was 280.4 (95% C.I. 225.1, 349.3) turtles/ha for the en tire study period, compared to P nelsoni with 175.3 (95% C.I. 119.8, 256.4) turtles/ha. P. peninsularis occurred at densities significantly higher than P. nelsoni during three sampling periods (Figure 3-1). Bo th species were found to have significantly declining densities over my study period, the most precipitou s of which occurred for female P. nelsoni Since these congeners occupy the same hab itat within my study s ite they may compete for resources (Fuselier and Edds, 1994; Linde man, 2000) and be influenced by the same disturbances (Dodd et al., 2006). The density of these species may be connected to the availability of basking sites and frequency at which these were disturbed by recreational activities such as canoeing and fi shing. Fallen trees that extend gr eatly into the lagoon and spring run were removed to make the la goon more suitable for recreati on, which frequently disturbs turtles attempting to bask (Moore and Seigel, 2006). P. nelsoni which basks more frequently than P. peninsularis (Ernst et al., 1994), may be found in lower densities due to frequent disturbances while attempting to bask.
36 Several studies have estimated densities of tu rtles with which I can compare my results. In an adjacent area to my study site, Kram er (1995) estimated a population density of 24.1 turtles/ha for adult P. peninsularis (reported as P. floridana prior to classification change) and 78.6 turtles/ha for adult P. nelsoni at nearby Rock Springs Run, which connects to the Wekiwa Springs Run. Marchand (1942) found three species of Pseudemys at Rainbow Run (Marion County, FL): P. concinna P. floridana and P. nelsoni P. concinna was most abundant at the time, achieving a density of 170. 0 turtle/ha (Iverson, 1982), and P. nelsoni was consistently less abundant than either of its conge ners (Marchand, 1942). These trends in relative abundance at Rainbow Run continue, and from the year 2000 to 2003 P. concinna had a density of 3.3.8 turtles/ha, P. floridana had a density of 3.8.5 turtles/ha, and P. nelsoni was captured too infrequently to estimate density (Huestis and Meylan, 2004). Congdon et al. (1986) found a density of 7.0 turtles/ha for P. floridana in at Ellenton Bay, locat ed at the Savannah River Ecology Laboratory (SREL) near Ai ken, South Carolina. The findings of Dreslik et al. (2005) for P. concinna in southern Illinois pond also show a much lower overall density of 7.3 turtles/ha. Findings for T. scripta however, show that turtles of simila r sizes and life history characteristics are capable of reaching higher population densitie s, for example 353.0 turtles/ha at Capers Island, SC (Congdon et al., 1986) and 361.4 turtles/ ha in Alachua County, FL (Auth, 1975). Density estimates for the Sternotherus species varied widely due to lower recapture rates. S. minor minor had a geometric mean density of 327.2 (95% C.I. 153.4, 698.2) turtles/ha, and S. odoratus had a similar geometric mean density of 320. 5 (95% C.I. 202.7, 507.0) turtles/ha for the entire study period. While my study found both species of Sternotherus to be at nearly equal densities overall, this was not the case when consideri ng the lagoon and the spring run separately. S. minor minor was found at higher densities than S. odoratus in the spring run during
37 every sample period, and with th e exception of one sampling period S. odoratus was found at higher densities in the lagoon than S. minor minor Sampling only one of these habitats would have led to very different conc lusions as to abundance of eith er population in the general area. This differential habitat use may account for th e differences in relative abundance seen at Rainbow Run, as S. minor may favor these spring run habitats (Huestis and Meylan, 2004). As the geographic range of S. minor is so small compared to that of S. odoratus S. minor may be better specialized to thrive in southern spring environments (Iverson, 1977). The densities of Sternotherus species reported for this st udy fall into the mid-range of other similar studies elsewhere. S. minor has been estimated with densities as high as 2857 turtles/ha at Emerald Springs Boil (Bay Count y, FL) (Cox and Marion, 1979 from Onorato, 1996) which is nearly an order of magnitude hi gher than my estimates, and has also been estimated at 127 turtles/ha in Rainbow R un (Meylan et al., 1992). The density of S. odoratus Ellenton Bay was 7.5 turtles/ha, and 21.8 turtles/ ha at nearby Risher Pond (SREL) (Congdon et al., 1986). S. odoratus reached a density of 148.5 turtles/ ha in northern Alabama (Dodd, 1989), and 106 turtles/ha in Rainbow Run (Meylan et al., 1992). However, Iverson (1982) reported a density of 700.0 turtles/ha for a Florida pond, again considerably higher than my estimates. Comparisons of historic and current fi ndings at Rainbow R un illustrate that S. minor and S. odoratus do not always occur at similar de nsities. While Ma rchand (1942) found no S. minor at Rainbow Run during his study, 50 years later Huestis and Meylan (2004) found S. minor to be extremely abundant, more so than S. odoratus or any species of Pseudemys Biomass Turtles are important components of the energe tics of the communities in which they live. In part because of the lowe r food requirements for turtles compared to similarly sized endotherms, turtles can achieve higher population de nsities than endotherms in the same system
38 (Iverson, 1982; Pough, 1983). Turtle biomass has often either gone unmeasured or been estimated using average weights fr om closely related species rath er than actual weights found in the population (Iverson, 1982; Congdon et al., 1986). All turtles in my study were weighed and therefore I can produce more accurate estimates of biomass for the four species I measured in my Florida spring. The four species I measured at Wekiwa Sp rings are an important part of the spring ecosystem, with a geometric mean biomass of 1367.8 kg/ha (95% C.I. 1031.9, 1812.9) combined for sub-adult and adults of these four speci es over the entire study period. The largest contributors to this biomass estimate are P. peninsularis and P. nelsoni whose considerably larger body size overshadowed th e contributions of the sma ller although potentially more numerous Sternotherus species members. Adult and sub-adult P. peninsularis had a geometric mean biomass of 820.6 kg/ha (95% C.I. 633.3, 1063.4), while P. nelsoni had a smaller geometric mean biomass of 434.8 kg/ha (95% C.I. 279.6, 676. 0). The biomass estimates of both species of Pseudemys declined significantly throughout the study, a combination of significantly decreasing densities and decreasing mean masses (significant for P. peninsularis ). For both species approximately 70% of this biomass is attributab le to females, which are significantly heavier than males in both species. It is important to ha ve separate estimates of density and mass for each sex in order to account for this difference (Congdon et al., 1986). The only published estimate of freshwater tu rtle biomass that approaches that of P. peninsularis in my study is that of T. scripta for which Congdon et al. (1986) found a biomass of 877.3 kg/ha on Capers Island, South Carolina and for which biomass was calculated in the same way as in this study. Biomass estimates of closely related species at Rainbow Run (from Marchand, 1942) were within range for P. nelsoni in my study: 311.1 kg/ha for P. floridana and
39 384.2 kg/ha for P. concinna (Iverson, 1982). However, these estimates of average mass were taken from regressions based on another species ( P. rubriventris ) due to lack of actual mass measurements, and densities were not calculated se parately for each sex; therefore these biomass estimates may be biased low. No published estimates of biomass for P. nelsoni were found. However, using the data of Kramer (1995), I calc ulated the approximate biomass of this species in Rock Springs Run (masses were estimated usin g a regression of plastron length to mass for my population), to be 153.8 kg/ha, less than half the biomass of P. nelsoni estimated in my study. In my study adult and sub-adult S minor minor had a geometric mean biomass of 37.0 kg/ha (95% C.I. 16.4, 83.4), and S. odoratus had a smaller geometric mean biomass of 16.9 kg/ha (95% C.I. 10.3, 27.9). Iverson (1982) estimated S minor to have a biomass of 45.7 kg/ha in a Florida spring, based on Cox and Marion (1979) who found a density of 2857 turtle/ha at Emerald Springs Boil. Although Cox and Marions (1979) density estimate is considerably higher than the density I found for S. minor minor my biomass estimate is within the 95% confidence intervals for the biomass estimate from Emerald Springs Boil (Iverson, 1982). Meylan et al. (1992) worked in Rainbow Run and found a biomass of 12.5 kg/ha for S. minor significantly lower than found in my study, and 6.1 kg/ha for S. odoratus also significantly lower than my estimate for this species. S odoratus were estimated to have a biomass of 41.7 kg/ha in a Florida pond by Iverson (1982), which is significantly great er than my biomass estimate for the species. Conversely, Congdon et al. (1986) estimate a biomass of 1.2 kg/ha for S. odoratus in Ellenton Bay, and 1.4 kg/ha at Risher Pond, both substantially smaller than my population. Estimates of turtle biomass are important fo r calculations of energy and nutrient flow through turtle populations. The largely herbivorous Pseudemys species partially regulates the
40 aquatic plant system, while the omnivorous Sternotherus species preys upon insects and mollusks and scavenges (Ernst et al., 1994). Bioma ss is also necessary to estimate the prey base for turtle predators. Turtle eggs provide food for predators such as the raccoon ( Procyon lotor ) and imported red fire ants ( Solenopsis invicta ), and enrich the soil when clutches are unsuccessful (Marchand et al., 2002 ; Aresco, 2004). Hatchling and juvenile turtles provide food for American alligators (Bondavalli and Ulanowicz, 1999), armadillos ( Dasypus novemcinctus ), raccoons, red foxes ( Vulpes vulpes ), river otters ( Lutra canadensis ) (Brooks et al., 1991), Virginia opossums ( Didelphis virginiana ), and numerous fish and bird species (Frazer and Gibbons, 1990; Ernst et al., 1994; Jan zen et al., 2000). Adult turtles have high survival rates and suffer comparatively little pred ation from natural predators, although American alligator and river otters may still be threats. Human take is thought to be minimal or non-existent in this protected habitat. Survival Turtles, like other long-lived ve rtebrates with late sexual matu rity, are often characterized by high sub-adult and adult surviv al probabilities, which contri bute greatly to overall population stability (Heppell, 1998; Chaloupka and Limpus, 2002; Bjorndal et al., 2003). I determined the apparent annual survival probabilities of sub-ad ult and adult male and female turtles in each species. Apparent survival probabilities are differ ent from true survival probabilities in that mortality and emigration are confounded, and ther efore likely to be biased low (White and Burnham, 1999). Male and female geometric mean apparent annual survival pr obabilities were not significantly different for both P. peninsularis and P. nelsoni (Table 3-1). My estimates of apparent survival for sub-adult and adult P. peninsularis and P. nelsoni in this study are the first reported for the genus Pseudemys Although survival was not estimable in all time periods I am confident
41 that my results are robust. P. peninsularis had smaller 95% confiden ce intervals that did not cross 50%, as compared to its congener (Table 2). Survival decreas ed non-significantly for both male and female P. peninsularis Survival decreased significantly for P. nelsoni females, and non-significantly for males. Both species of Pseudemys had significantly declining densities throughout the study, which may be related to declining apparent survival probabilities, particularly for female P. nelsoni As I cannot separate mortality from permanent emigration using my estimates of apparent survival it is po ssible that the declining probabilities are due to higher rates of emigration. A greater percentage of P. nelsoni were captured in the spring run (15.46%) than P. peninsularis (3.93%), which could indicate a greater potential for emigration in P. nelsoni The estimated apparent annual survival probabilities for P. peninsularis and P. nelsoni are lower than many reported in the literature for Emydids. Some of the highest values are those reported by Litzgus (2006) of mi nimum annual survivorships for Clemmys guttata in Ontario, Canada, of 96.5% for females and 94.2% for males. Emydoidea blandingii in Michigans E. S. George Reserve were found to have an average minimum adult survival of 93.5% (SE 0.5%) (Congdon et al., 1993). Considerably closer to th e survival probabilities found in my study, Frazer et al. (1991a) found that Chrysemys picta in Sherriffs Marsh (Kalamazoo County, Michigan) had annual adult surv ival probabilities of 64% for males and 29% for females. Mitchell (1988) found overall surv ival probabilities of 77.0% fo r male and 79.5% for female C. picta over the course of 3 years in a Virginia pond. At Ellent on Bay, Frazer et al. (1991b) worked with T. scripta and found annual adult survival proba bilities of 77.4% (95% C.I. 76.3%, 78.6%) for females and 83.8% (95% C.I. 82.6% 85.1%) for males based on live recaptures. Estimates of annual survival probabilities from the same population using dead recoveries were
42 85.4% (95% C.I. 82.5%, 88.6%) for females and 85.0% (95% C.I. 83.5%, 86.4%) for males (Frazer and Gibbons, 1990). These differing esti mates of survival for the same population illustrate the importance of methodology in determining survival probabilities. Geometric mean apparent annual survival probabilities increased non-significantly for S. minor minor and showed no significa nt linear trends in S. odoratus (Table 2). The results of my survival analyses for the Sternotherus species show geometric means that are not significantly different from other Kino sternids, and those for S. minor minor are the first reported for the species. S. odoratus and S. minor minor utilized the spring run differently, which may relate to their ability or tendency to emigrate from the study site, which cannot be separated from mortality using apparent survival probabilities. Nearly 57% of all S. minor minor captured were found in the spring run, as compared to only 14.5% of all S. odoratus S. odoratus has been found to establish home ranges (Andres and Chambers, 2006) and while S. minor minor may do the same they would be more likely to move beyond the sampled area to locations farther down the spring run than their congener. Several studies have examined annual surviv al probabilities for Kinosternids. Mitchell (1988) studied S. odoratus in a Virginia pond for three y ears and found an overall annual survival probability of 78% for males, and 84.5% for females, an opposite trend than was observed in my study. The male survival probabili ty estimated by Mitchell (1988) falls within the lower 95% confidence interval for my estimate of male S. odoratus survival, and the female survival probabilities are also not significantly different. Frazer et al. (1991b) found Kinosternon subrubrum in Ellenton Bay to have an annual survival of 89.0% for male and 87.6% for females. Iverson (1991) found the annual survival pr obability of females in his population of K. flavescens to be 95%, within range of my 95% c onfidence intervals for both species of
43 Sternotherus Four populations of K. sonoriense had female annual survival probabilities ranging from 83% (Rosen, 1987, from Shine and Iverson, 1995). Various environmental factors can affect male and female survival differentially. Female turtles may experience high mortality when they atte mpt to nest, as they ar e in greater danger of being struck by cars (Steen and Gibbs, 2004; Ares co, 2005; Steen et al., 2006) and terrestrial predators (Tucker et al., 1999). Alternatively, male s of some species of turtles have been found to emigrate longer distances than females, thus simultaneously potentially increasing their risk of mortality (Tuberville et al., 1996) and lowering apparent survival probabilities. Size is also a factor in survival for turtles. Studies have repeatedly shown th at smaller individuals are more susceptible to predators and de siccation (Tucker et al., 1999; Bodie and Semlitsch, 2000). Age of turtles is not thought to affect adult survival probabilities (Congdon et al., 2001; Congdon et al., 2003). Realized Population Growth Rate Realized population growth rates ( ) show the observed growth rate between samples, as opposed to a projected value from projection matrices (Pradel, 1996; Nichols and Hines, 2002). Using Pradels reverse-time survival and models in Program MARK I was able to provide estimates of realized popu lation growth rate for sub-adu lts and adults in my turtle populations. This method of determining population gr owth rate has not yet been widely utilized for turtle populations. Both species of Pseudemys in my study had geometric mean realized population growth rates with point estimates below 1.000, although each upper 95% confidence interval reached or exceeded 1.000 (Table 3-1 ). These findings woul d normally lead us to conclude that the populations were stable during the study. However, this contrasts w ith the density estimates that
44 show a significant declining tre nd throughout the study period for both species. Inverse-variance weighted regression showed no si gnificant temporal trends in for either species of Pseudemys during the study period. This lends support to a tr uly declining value that I could not estimate satisfactorily given the 95% conf idence intervals surrounding the Pseudemys estimates. Even a realized population growth rate marginally be low 1.000 could manifest itself as significant declines in observed density. Further study is re quired to determine if these populations are in fact declining. Future work will focus on separatin g the factors that contribute to the estimated values and their relative importance (Nic hols and Hines, 2002; Converse et al., 2005). S. minor minor was the only species with realized population growth rate geometric means that were above 1.000 (Table 3-1). There were tw o apparent spikes in population density of S. minor minor that corresponds to high values with high variability. S. odoratus realized population growth rates had geometric means be low 1.000 with 95% confidence intervals that did not crossed 1.000, indicating a significant dec line in population growth rate throughout the study (Table 3-1). Regression analysis indicated a non-significant trend for increasing density in S. odoratus a finding that differs from the estimated realized population growth rate geometric means. As estimates of density and realized popul ation growth rate are so variable for these species, additional years of study will be necessary to determin e what trends actually exist in the population. Estimates of realized populati on rate have not been obtained for any of these four species previously. Realized population gr owth rates from Pradels revers e-time models have only been estimated for two species of turtles, both Emydids: Malaclemys terrapin (Mitro, 2003) and Terrapene ornata (Bowen et al., 2004; Converse et al., 2005). Mitro (2003) f ound that the most parsimonious Pradels survival and model had a constant value of 1.034 (95% C.I. 1.012,
45 1.056) for adult female M. terrapin over the course of a fi ve-year study. Abundance of M. terrapin increased over the course of the study, which is consistent with the estimated value, although simultaneously adult female apparent su rvival declined from 95.9% to 94.4%. Bowen et al. (2004) investigated the dynamics of a T. ornata population in Illinois surrounded by heavily human impacted habitats over the course of eight years. Alt hough the habitat was less than ideal Bowen et al. (2004) estimated appa rent annual survival at 97% (SE 6%), and was found to be 1.02 (SE 0.06). This value indicates that the popul ation should have marginally increased in size over the course of the study: however, as the error encompassed 1.000 it is more likely the population remained stable. Convers e et al. (2005) studied a remote population of T. ornata ornata in western Nebraska for 19 years and us ed model-averaging procedures as in my study. They found an average realized popula tion growth rate of 1.006 (SE 0.065) and further simulated their results to determine th e probability of populati on decline resulting from the temporal variance in over a 20-year period to be 58%. This study in particular illustrates the dangers of interpreting values close to 1.000 as definitive signs of population increase, and that the variability in factors strongly into actual population density changes (Converse et al., 2005). Sex Ratios Sex ratios are important in determination of the effective population size and may factor strongly in reproductive output and success for a population. Turt le populations throughout the United States have been found to be increasingly male biased over the past century (Gibbs and Steen, 2005). This trend has implications for turt le conservation because these sex ratio changes are often symptomatic of a higher mortality for adult female turtles, whose presence in the population is necessary for its stability and pe rsistence (Aresco, 2005; Gibbs and Steen, 2005;
46 Steen et al., 2006). Although signifi cant sex-specific differences in adult survival were not observed in my study, I determined sex ratios for each population because unequal sex ratios can result from other factors, such as differential emigration, immi gration, hatchling and juvenile survival rates. The sex ratios of P. peninsularis and P. nelsoni fluctuate with sampling period, and samples taken during different months of the sa me year can be significantly different (e.g., March and May 2003 for P. nelsoni Figure 3-7). The overall sex ratio for P. peninsularis was 1.09 males for every female, and 0.91 males/female for P. nelsoni Regression shows that during the study period the sex ratio tended to become increasingly female-biased for both species, though the finding was not significant. P. nelsoni was also the only species in my study to ever demonstrate a significantly female-biased sex ratio, which occurred twice (May 2003 and May 2004). These findings differ from those derived from Kramer (1995), who worked in nearby Rock Springs Run, where he found a sex rati o of 1.17 males/female (non-significant) for P. peninsularis and 1.42 males/female (non-significant) for P. nelsoni Even nearby populations, which may be connected by migration, can have very different sex ratios (Congdon et al., 1986; Gibbons, 1990), though the findings in this case were not signif icant. Recent findings at Rainbow Run showed a significan tly male-biased sex ratio for P. floridana (1.57 males/female, significant at P < 0.01) and closely related P. concinna at the same site ha d a sex ratio of 1.27 males/female (non-significant) (Huestis and Meylan, 2004). In Lake Jackson (Leon County, Florida) Aresco (2005) found P. floridana to be highly significantly ma le-biased, with a sex ratio of 3.96 males/female, whereas at Ellenton Bay the sex ratio for this species was 1.31 males/female (non-significant, Gibbons, 1990). More P. peninsularis and P. nelsoni were captured and with higher recapt ure rates in my study than in the aforementioned studies
47 (excluding Aresco, 2005), and my method of assigni ng sex at smaller sizes and potentially before sexual maturation differed from these studies and may have contributed to the different findings (Gibbons, 1990). Similar to the Pseudemys species in my study, the Sternotherus species had sex ratios that continually differed throughout the course of the study. S. odoratus had a significantly malebiased overall sex ratio of 1.34 males/female ( P < 0.05), and did not display any linear temporal trends in sex ratio, indicating that this male-bias is not an artifact. Many studies have reported sex ratios for this wide-ranging species, ranging from significantly female-biased to significantly male-biased (Gibbons, 1990). Gibbons (1990) found sex ratios of S. odoratus to differ from 1.08 males/female (non-significant) in Risher Pond to 2.59 males/female (significant, P < 0.01) in Par Pond, both located at SREL. Likewise, Smith an d Iverson (2002) found a male-biased sex ratio of 1.70 (significant, P < 0.001) at their long-t erm study site at Dewart Lake (Kosciusko County, Indiana). At Rainbow Run S. odoratus is also significantly male-biased, with a ratio of 2.63 males/female (Meylan et al., 1992). A recent stu dy by Swannack and Rose (2003) confirmed that these biased sex ratios for S. odoratus are not artifacts of sampli ng error but reflect the true population structure, finding a sex ratio of 1.81 (significant, P < 0.001) at their Texas site. Differing substantially, my study found that S. minor minor had an overall sex ratio of 0.89 males/female (non-significant), consistent with a report of 0.91 males/female for S. minor throughout their range (Iverson, 1977). Meylan et al (1992) established sex prior to sexual maturity, as in this study, and found a sex ra tio of 0.92 males/female (non-significant) for S. minor at Rainbow Run. Showing the opposite trend, Cox et al. (1991) found a sex ratio of 1.14 males/female (non-significant) for S. minor in northwestern Florida springs.
48 Habitat Alteration This protected Florida spring is still in good co ndition, and the ecosystem can still support many species of turtles at high densities. Howeve r, changes to the lagoon and spring run habitats during this study and those predicted to occur in the future are likely to influence the dynamics of these populations. One environmental factor that may be linked to some of the dynamics seen during this study was the invasion of the lagoon area by the aquatic plant Hydrilla verticillata in 2001. H verticillata was present in low quantities in Marc h 2001 but had choked off the entire lagoon except for a strip of fast flowing wate r in the center of the lagoon by March 2002. To combat this invasive plant, a turbidity barrier wa s installed that divided the lagoon in half shortly before March 2002. This barrier created stagnate water on the south side of the lagoon, which resulted in temperature fluctu ations, and a photosynthetic inhib itor was applied to kill the H. verticillata Once H. verticillata was largely removed from the area, the turbidity barrier was moved to enclose the north side of the lagoon and the same treatment was applied. This barrier remained in place until shortly before March 2003. The barrier had evenly spaced gaps that we re large enough for even the largest female P. peninsularis to pass through easily, and within-samp ling period recaptures indicated that the turtles often passed throu gh the barrier. Although this treatmen t caused no apparent harm to the turtles, H verticillata did provide abundant food and cover from predators such as American alligators, and made it far more difficult for turtle s to be hand captured. The significant drop in density for both Pseudemys species and concurrent drop in mean mass (significant for P. peninsularis ) may be connected to the removal of H. verticillata This disturbance may have lead to larger individuals emigrating from the area, and these individuals may then have stayed at these new locations rather than emigra ting back immediately (Dodd et al., 2006).
49 Models of water output from the Wekiwa spring based on data from 2003 and earlier indicate that flow will probably decrease by 8% between 1995 and 2005 (Saint Johns River Water Management District, 2006). This decrease is purported to be sustainable. However, it is only one in a series of decreases in water quantity output from the spring as a result of increased development, increased water usage, and decrea sed recharge area. Management is needed to ensure that flows remain at ecologically sustainable levels despite continued growth of the greater Orlando metropolitan area. Furthermore, water quality has decreased as a result of nitrate pollution. The water coming out the spring has spent roughly 20 years in the gr ound, and the nitrate is from lawn fertilizers and septic tanks in the 1980s and earlier (Toth and Fortich, 2002). Despite the fact that steps have been taken to educate the public in regards to fertilizer use, and ol d septic tanks are being removed and replaced with city water and sewa ge systems, the spring will continue to show increased nitrate concentrations for many more years. This eutrophication increases plant abundance but also increases the deposition of de ad plant matter and thus causes silting and a decrease in the overall lagoon depth and water quality. Wekiwa Springs State Park is a protected natural area with a thriving turtle community which should be protected. The data collected on these turtles can serve as a baseline for other less studied populations of the same species, or similar speci es (Heppell, 1998). Continued monitoring of these populations will document ho w decreased spring outflow and sustained high nitrate levels impact a natural freshwater turtle community. Shoul d it become apparent that the turtle community is declining, managers can take steps to improve the habitat for the turtles. Increasing basking areas in the lagoon and reduc ing impacts from recreation could both improve this habitat for turtles.
50APPENDIX A CORMACK-JOLLY-SEBER MODELS Appendix A. Cormack-Jolly-Seber models for each of the four most commonly caught specie s of turtles at Wekiwa Springs State Par k from March 2000 to November 2005. The models include parameters for apparent survival ( ) and recapture ( ) rate, which were allowed to vary over time (each primary period, t), with or without the addition of effects from sex (t+sex) and interactions with sex (t*sex). Models ar e compared via Akaikes Information Crit erion corrected for small sample size and overdispersion (QAICc). Both species of Psuedemys displayed over-dispersion in the data and were corrected via the value, whereas both Sternotherus species presented under-dispersi on and were therefore left at = 1.00. Models weights were used in model averaging proce dures. K represents the number of parameters estimated per model. Species (Model dispersion parameter) P. peninsularis ( = 1.05) P. nelsoni ( = 1.18) S. minor minor ( = 1.00) S. odoratus ( = 1.00) Model QAICc Weight K QAICc Weight K AICc Weight K AICc Weight K (t) (t) 2784.88 0.437 17 1182.12 0.005 16 457.66 0.141 13 421.64 0.307 14 (t) (t+sex) 2786.89 0.160 18 1184.10 0.002 17 455.49 0.417 13 423.82 0.103 15 (t) (t*sex) 2795.87 0.002 28 1197.51 0.000 27 457.36 0.164 18 425.12 0.054 20 (t+sex) (t) 2786.46 0.198 18 1176.76 0.074 16 457.23 0.175 15 422.63 0.188 15 (t+sex) (t+sex) 2787.40 0.124 19 1180.87 0.009 19 459.43 0.058 16 422.33 0.218 16 (t+sex) (t*sex) 2797.87 0.001 29 1189.23 0.000 28 460.72 0.030 20 424.83 0.062 20 (t*sex) (t) 2789.22 0.050 26 1172.97 0.492 23 462.96 0.010 18 427.55 0.016 19 (t*sex) (t+sex) 2790.71 0.024 27 1173.34 0.408 24 465.18 0.003 19 425.78 0.039 19 (t*sex) (t*sex) 2793.75 0.005 34 1180.74 0.010 31 465.58 0.003 23 427.90 0.013 22
51APPENDIX B PRADELS MODELS OF THREE SPECIES Appendix B. Pradels reverse-time survival and lambda models for P peninsularis S minor minor and S odoratus caught at Wekiwa Springs State Park from March 2000 to November 2005. The models include parameters for apparent survival ( ) and recapture ( ) rate and realized popul ation growth rate ( ). and were only allowed to vary over time (each primary period, t), with no eff ects of sex considered. was allowed to be constant (.), vary by sex only (sex), vary over time (t), vary over time with non-interac ting sex effects (t+sex), and vary over time w ith interaction of sex (t*sex). Models are compared via Akaikes Information Criter ion corrected for small sample size (AICc). Models weights were used in model averaging procedures. K represents the numbe r of parameters estimated per model. Species P. peninsularis S. minor minor S. odoratus Model AICc Weight K AICc Weight K AICc Weight K (t) (t) (.) 5419.77 0.001 23 1689.46 0.697 17 1659.29 0.716 19 (t) (t) (sex) 5420.78 0.001 24 1693.41 0.097 18 1661.56 0.230 20 (t) (t) (t) 5408.43 0.305 28 1694.79 0.048 28 1664.98 0.042 25 (t) (t) (t+sex) 5409.74 0.159 29 1692.43 0.158 27 1667.34 0.013 26 (t) (t) (t*sex) 5407.31 0.535 39 1703.88 0.001 33 1683.13 0.000 36
52 APPENDIX C PRADELS MODELS OF Pseudemys nelsoni Appendix C. Pradels reverse-time survival and lambda models for P nelsoni caught at Wekiwa Springs State Park from March 2000 to November 2005. The models include parameters for apparent survival ( ) and recapture ( ) rate and real ized population growth rate ( ). was allowed to vary over time ( each primary period, t), vary over time with non-interacting sex effects (t+sex), and vary over time with interaction of sex (t*sex). was held as time-specific. was allowed to be constant (.), vary by sex only (sex), and again (t), (t +sex), and (t*sex). Models are compared via Akaikes Information Criterion corrected for small sample size (AICc). Models weights were used in model averaging procedures. K represents the number of parameters estimated per model. Model AICc Weight K (t) (t) (.) 2912.72 0.003 20 (t) (t) (sex) 2913.22 0.002 21 (t) (t) (t) 2906.61 0.057 27 (t) (t) (t+sex) 2907.14 0.044 28 (t) (t) (t*sex) 2925.90 0.000 38 (t+sex) (t) (.) 2914.10 0.001 21 (t+sex) (t) (sex) 2912.17 0.004 22 (t+sex) (t) (t) 2905.27 0.112 27 (t+sex) (t) (t+sex) 2904.33 0.178 29 (t+sex) (t) (t*sex) 2919.57 0.000 38 (t*sex) (t) (.) 2907.77 0.032 27 (t*sex) (t) (sex) 2908.71 0.020 28 (t*sex) (t) (t) 2903.16 0.320 34 (t*sex) (t) (t+sex) 2903.90 0.222 35 (t*sex) (t) (t*sex) 2911.13 0.006 44
53 APPENDIX D REGRESSIONS OF ESTIMATES Appendix D. Regression line properties of seven estimates: annual recapture ( ) probability, density (turtles/ha), mean ma ss (kg), biomass (kg/ha), a nnual apparent survival ( ) probability, realized population growth rate ( ), and sex ratio (males/female). Estimates are from four species of turtle s captured at Wekiwa Springs State Park from March 2000 to November 2005. Regre ssion line properties include slope, the Pvalue of the slope, y-intercept, and R2 values followed by standard errors where applicable, all to 5 significant figures. -indicates failure of regression model convergence. Regression line parameters Species Estimate Sex Slope P y-intercept R2 M 0.0022 (0.0012) 0.1074 0.2081 (0.0375) 0.9666 Annual probability F 0.0022 (0.0012) 0.1054 0.2031 (0.0369) 0.9661 M -2.1705 (0.7439)0.0224 207.79 (26.678) 0.9557 Density F -2.1425 (0.7085)0.0193 200.45 (25.346) 0.9572 M -0.0061 (0.0017)0.0050 2.0064 (0.0755) 0.9959 Mean mass F -0.0082 (0.0037)0.0524 4.5579 (0.1609) 0.9965 M -4.7115 (1.5986)0.0215 388.56 (58.617) 0.9327 Biomass F -10.202 (3.3123)0.0178 881.13 (120.21) 0.9466 M -0.0016 (0.0046)0.7321 0.8206 (0.1374) 0.9452 Annual probability F -0.0017 (0.0040)0.6908 0.7224 (0.1209) 0.9633 M 0.0136 (0.0072) 0.1025 0.3545 (0.2241) 0.8352 F 0.0101 (0.0085) 0.2727 0.4549 (0.2746) 0.8132 P. peninsularis Sex ratio M/F -0.0062 (0.0038)0.1324 1.3443 (0.1676) 0.9467 M 0.0080 (0.0024) 0.0125 0.0660 (0.0704) 0.8828 Annual probability F 0.0074 (0.0024) 0.0168 0.0589 (0.0679) 0.8681 M -1.5788 (0.5797)0.0296 116.12 (22.833) 0.8995 Density F -2.6166 (0.5126)0.0014 180.87 (20.809) 0.9559 M -0.0020 (0.0026)0.4610 1.9559 (0.1093) 0.9890 Mean mass F -0.0085 (0.0064)0.2137 3.5123 (0.2843) 0.9801 M -3.1544 (1.2483)0.0394 223.01 (49.669) 0.8648 Biomass F -7.6937 (2.5586)0.0197 527.19 (102.09) 0.8872 M -0.0036 (0.0081)0.6793 0.5948 (0.2947) 0.7115 Annual probability F -0.0137 (0.0045)0.0291 1.1430 (0.1230) 0.9651 M -0.0025 (0.0166)0.8830 0.6906 (0.6172) 0.4740 F -0.0029 (0.0173)0.8710 0.7253 (0.6433) 0.4705 P. nelsoni Sex ratio M/F -0.0060 (0.0041)0.1780 1.1619 (0.1884) 0.9125
54 Appendix 4 cont. Regression line parameters Species Estimate Sex Slope Species Estimate Sex M -0.0003 (0.0027)0.9271 0.1522 (0.1047) 0.8392 Annual probability F 0.0006 (0.0016) 0.7195 0.0933 (0.0601) 0.8961 M 1.2358 (3.5725) 0.7435 77.249 (133.95) 0.5288 Density F 1.4057 (4.1825) 0.7504 79.830 (156.43) 0.4539 M -0.0001 (0.0003)0.7417 0.1147 (0.0156) 0.9755 Mean mass F 0.0004 (0.0003) 0.2364 0.0937 (0.0157) 0.9735 M 0.1653 (0.4067) 0.7013 7.3527 (15.092) 0.4916 Biomass F 0.0561 (0.5008) 0.9152 12.352 (19.600) 0.3940 M 0.0218 (0.0126) 0.1587 0.1245 (0.4461) 0.9558 Annual probability F 0.0194 (0.0139) 0.2348 0.0104 (0.4658) 0.8296 M 0.0017 (0.0092) 0.8583 0.9998 (0.3237) 0.9013 F 0.0017 (0.0092) 0.8583 0.9998 (0.3235) 0.9014 S. minor minor Sex ratio M/F 0.0007 (0.0079) 0.9291 1.0539 (0.3933) 0.8625 M 0.0008 (0.0009) 0.4259 0.0362 (0.0328) 0.7960 Annual probability F 0.0006 (0.0012) 0.6227 0.0627 (0.0397) 0.8228 M 2.4075 (2.8071) 0.4240 100.94 (97.991) 0.7709 Density F 1.8351 (1.6598) 0.3113 49.603 (54.096) 0.7609 M 0.0001 (0.0001) 0.5676 0.0434 (0.0042) 0.9832 Mean mass F ----M 0.1553 (0.1302) 0.2782 3.1383 (4.3936) 0.7616 Biomass F 0.1470 (0.1137) 0.2437 2.4447 (3.6167) 0.7371 M -0.0030 (0.0094)0.7646 0.8748 (0.3214) 0.8940 Annual probability F 0.0068 (0.0182) 0.7354 0.2801 (0.6670) 0.6951 M 0.0003 (0.0013) 0.8385 0.8161 (0.0441) 0.9946 F 0.0003 (0.0013) 0.8395 0.8164 (0.0442) 0.9946 S. odoratus Sex ratio M/F 0.0011 (0.0064) 0.8643 1.3602 (0.3018) 0.9183
55 LIST OF REFERENCES ANDRES, K. M., and R. M. CHAMBERS. 2006. A test of philopatr y by common musk turtles. American Midland Naturalist 156:45-51. ARESCO, M. J. 2004. Reproductive ecology of Pseudemys floridana and Trachemys scripta (Testudines: Emydidae) in northwestern Florida. Journal of Herpetology 38:249-256. ARESCO, M. J. 2005. The effect of sex-specific terrestrial moveme nts and roads on the sex ratio of freshwater turtles. Biol ogical Conservation 123:37-44. AUTH, D. L. 1975. Behavioral ecology of ba sking in the Yellow-bellied Turtle, Chrysemys scripta scripta (Schoepff). Bulletin of the Florida State Museum: Biological Science 20:145. BJORNDAL, K. A., A. B. BOLTEN, and M. Y. CHALOUPKA. 2003. Survival probability estimates for immature green turtles Chelonia mydas in the Bahamas. Marine Ecology-Progress Series 252:273-281. BJORNDAL, K. A., A. B. BOLTEN, and M. Y. CHALOUPKA. 2005. Evaluating trends in abundance of immature green turtles, Chelonia mydas in the Greater Caribbean. Ecological Applications 15:304-314. BODIE, J. R., and R. D. SEMLITSCH. 2000. Size-specific mortality and natural selection in freshwater turtles. Copeia 2000:732-739. BONDAVALLI, C., and R. E. ULANOWICZ. 1999. Unexpected effects of predators upon their prey: The case of the American alligator. Ecosystems 2:49-63. BOWEN, K. D., P. L. COLBERT, and F. J. JANZEN. 2004. Survival and recruitment in a humanimpacted population of ornate box turtles, Terrapene ornata with recommendations for conservation and management. Journal of Herpetology 38:562-568. BROOKS, R. J., G. P. BROWN, and D. A. GALBRAITH. 1991. Effects of a sudde n increase in natural mortality of adults on a populati on of the Common Snapping Turtle ( Chelydra serpentina ). Canadian Journal of Zoology 69:1314-1320. BURNHAM, K. P., and D. R. ANDERSON. 2002. Model Selection and Multi-Model Inference: A Practical Information-Theoretic Approach, 2nd Edition. Springer-Verlag, New York, USA. CAGLE, F. R. 1939. A system for marking turtles fo r future identificati on. Copeia 1939:170-173. CASWELL, H. 2001. Matrix Population Models: Constr uction, Analysis, an d Interpretation, 2nd Edition. Sinauer Associates, Sunderland, MA. CHALOUPKA, M. Y., and C. J. LIMPUS. 2002. Survival probability estimates for the endangered loggerhead sea turtle resident in southern Great Barrier Reef waters. Marine Biology 140:267-277.
56 CHOQUET, R., A. M. REBOULET, R. PRADEL, O. GIMENEZ and J. D. LEBRETON. 2003. User's manual for U-Care. CEFE/CNRS, Montpellier (f tp://ftp.cefe.cnrs-mop.fr/biom/Soft-CR/). CONGDON, J. D., A. E. DUNHAM, and R. C. V. SELS. 1993. Delayed sexual maturity and demographics of Blanding Turtles ( Emydoidea blandingii ) Implications for conservation and management of long lived organi sms. Conservation Biology 7:826-833. CONGDON, J. D., J. L. GREENE, and J. W. GIBBONS. 1986. Biomass of freshwater turtles a geographic comparison. American Midland Naturalist 115:165-173. CONGDON, J. D., R. D. NAGLE, O. M. KINNEY, and R. C. V. SELS. 2001. Hypotheses of aging in a long-lived vertebrate, Blanding's turtle ( Emydoidea blandingii ). Experimental Gerontology 36:813-827. CONGDON, J. D., R. D. NAGLE, O. M. KINNEY, and R. C. V. SELS. 2003. Testing hypotheses of aging in long-lived painted turtles ( Chrysemys picta ). Experimental Gerontology 38:765772. CONVERSE, S. J., J. B. IVERSON, and J. A. SAVIDGE. 2005. Demographics of an ornate box turtle population experiencing minimal human-induced disturbances. Ecological Applications 15:2171-2179. COX, W. A., and K. R. MARION. 1979. Population structure and su rvivorship in the musk turtle, Sternotherus minor in a north Florida spring run (R eptilia: Chelonia). Bulletin of the Association of Southeas tern Biologists 26:84. COX, W. A., J. B. HAZELRIG, M. E. TURNER, R. A. ANGUS, and K. R. MARION. 1991. A model for growth in the Musk Turtle, Sternotherus minor in a north Florida spring. Copeia 1991:954-968. CROUSE, D. T., L. B. CROWDER, and H. CASWELL. 1987. A stage-based population model for Loggerhead sea turtles and implications for conservation. Ecology 68:1412-1423. DODD, C. K. 1989. Population st ructure and biomass of Sternotherus odoratus (Testudines, Kinosternidae) in a northern Alab ama lake. Brimleyana 1989:47-56. DODD, C. K., A. OZGUL, and M. K. OLI. 2006. The influence of disturbance events on survival and dispersal rates of Flor ida box turtles. Ecological Applications 16:1936-1944. DRESLIK, M. J., A. R. KUHNS, and C. A. PHILLIPS. 2005. Structure and composition of a southern Illinois freshwater turtle assemblage Northeastern Naturalist 12:173-186. ERNST, C. H., R. W. BARBOUR and J. E. LOVICH. 1994. Turtles of the Un ited States and Canada. Smithsonian Institution Press, Washington D.C. FRAZER, N. B. and J. W. GIBBONS. 1990. Life tables of a slider turtle population, p. 183-200. In : Life History and Ecology of the Slider Turtle J.W. Gibbons (ed.). Smithsonian Institution Press, Washington D.C.
57 FRAZER, N. B., J. W. GIBBONS, and J. L. GREENE. 1991a. Growth, survivorship and longevity of Painted Turtles Chrysemys picta in a southwestern Michigan marsh. American Midland Naturalist 125:245-258. FRAZER, N. B., J. W. GIBBONS, and J. L. GREENE. 1991b. Life history a nd demography of the Common Mud Turtle Kinosternon subrubrum in South Carolina, USA. Ecology 72:22182231. FUSELIER, L., and D. EDDS. 1994. Habitat partitioning among 3 sy mpatric species of map turtles, genus Graptemys Journal of Herpetology 28:154-158. GARBER, S. D., and J. BURGER. 1995. A 20-yr study documenting th e relationship between turtle decline and human recreation. Ecol ogical Applications 5:1151-1162. GIBBONS, J. W. 1990. Sex ratios and their signif icance among turtle popul ations, p. 171-182. In : Life History and Ecology of the Slider Turtle J.W. Gibbons (ed.). Smithsonian Institution Press, Washington D.C. GIBBONS, J. W. 1997. Measuring declines and natura l variation in turtle populations: spatial lessons from long-term studies. Proceedi ngs of Conservation, Restoration and Management of Tortoises and Turtles An International Conference 1997:243-246. GIBBONS, J. W., D. E. SCOTT, T. J. RYAN, K. A. BUHLMANN, T. D. TUBERVILLE, B. S. METTS, J. L. GREENE, T. MILLS, Y. LEIDEN, S. POPPY, and C. T. WINNE. 2000. The global decline of reptiles, dj vu amphibians. Bioscience 50:653-666. HEPPELL, S. S. 1998. Application of life-history theo ry and population model analysis to turtle conservation. Copeia 1998:367-375. HUESTIS, D. L., and P. A. MEYLAN. 2004. The turtles of rainbow run (Marion County, Florida): Observations on the genus Pseudemys Southeastern Naturalist 3:595-612. IVERSON, J. B. 1977. Geographic variation in Musk Turtle, Sternotherus minor Copeia 1977:502-517. IVERSON, J. B. 1982. Biomass in turtle populations a neglected subject. Oecologia 55:69-76. IVERSON, J. B. 1991. Life history and dem ography of the Yellow Mud Turtle, Kinosternon flavescens Herpetologica 47:373-395. IVERSON, J. B., S. J. CONVERSE, G. R. SMITH, and J. M. VALIULIS. 2006. Long-term trends in the demography of the Allen Cays Rock Iguana ( Cyclura cychlura inornata ): Human disturbance and density-de pendent effects. Biological Conservation 132:300-310. JANZEN, F. J., J. K. TUCKER, and G. L. PAUKSTIS. 2000. Experimental analysis of an early lifehistory stage: Selection on size of hatchling turtles. Ecology 81:2290-2304.
58 KRAMER, M. 1995. Home range of the Florida red-bellied turtle ( Pseudemys nelsoni ) in a Florida spring run. Copeia 1995:883-890. LINDEMAN, P. V. 2000. Resource use of five sympatri c turtle species: eff ects of competition, phylogeny, and morphology. Canadian Journal of Zoology 78:992-1008. LITZGUS, J. D. 2006. Sex differences in longevity in the Spotted Turtle ( Clemmys guttata ). Copeia 2006:281-288. MARCHAND, M. N. 1942. A contribution to the knowle dge of the natural history of certain freshwater turtles. M.S. Thesis, Univ ersity of Florida, Gainesville, FL. MARCHAND, M. N., and J. A. LITVAITIS. 2004. Effects of habitat features and landscape composition on the population structure of a co mmon aquatic turtle in a region undergoing rapid development. Conservation Biology 18:758-767. MARCHAND, M. A., J. A. LITVAITIS, T. J. MAIER, and R. M. DEGRAAF. 2002. Use of artificial nests to investigate predation on freshwater tu rtle nests. Wildlife Society Bulletin 30:10921098. MCDONALD, T. L., and S. C. AMSTRUP. 2001. Estimation of populat ion size using open capturerecapture models. Journal of Agricultural Bi ological and Environmental Statistics 6:206220. MEYLAN, P. A., C. A. STEVENS, M. E. BARNWELL, and E. D. DOHM. 1992. Observations on the turtle community of Rainbow Run, Marion Co., Florida. Florid a Scientist 55:219-228. MITCHELL, J. C. 1988. Population ecology and life histories of the freshwater turtles Chrysemys picta and Sternotherus odoratus in an urban lake. Herpet ological Monographs 2:40-61. MITRO, M. G. 2003. Demography and viability anal yses of a diamondback terrapin population. Canadian Journal of Zoology 81:716-726. MOORE, M. J. C., and R. A. SEIGEL. 2006. No place to nest or bask: Effects of human disturbance on the nesting and basking habits of yellow-blotched map turtles ( Graptemys flauimaculata ). Biological Conservation 130:386-393. NICHOLS, J. D., and J. E. HINES. 2002. Approaches for the dir ect estimation of lambda, and demographic contributions to lambda, using capture-recapture data. Journal of Applied Statistics 29:539-568. ONORATO, D. 1996. The growth rate and age distribution of Sternotherus minor at Rainbow Run, Florida. Journal of Herpetology 30:301-306. OZGUL, A., L. L. GETZ, and M. K. OLI. 2004. Demography of fluctuating populations: temporal and phase-related changes in vital rates of Microtus ochr ogaster. Journal of Animal Ecology 73:201-215.
59 POLLOCK, K. H. 1982. A capture-recapture design r obust to unequal probability of capture. Journal of Wildlife Management 46:757-760. POUGH, F. H. 1983. Amphibians and reptiles as low-energy systems, p. 141-188. In : Behavioral Energetics. W. P. Aspey and S. I. Lustick (e ds.). Ohio State University Press, Columbus, OH. PRADEL, R. 1996. Utilization of capture-mark-recap ture for the study of recruitment and population growth rate. Biometrics 52:703-709. REID, J. M., E. M. BIGNAL, S. BIGNAL, D. I. MCCRACKEN, and P. MONAGHAN. 2004. Identifying the demographic determinants of population growth rate: a case study of red-billed choughs Pyrrhocorax pyrrhocorax Journal of Animal Ecology 73:777-788. ROSEN, P. C. 1987. Variation in female reproduct ion among populations of Sonoran mud turtles ( Kinosternon sonoriense ). M.S. Thesis, Arizona State University, Tempe, AZ. SAINT JOHNS RIVER WATER MANAGEMENT DISTRICT. 2006. Water supply assessment 2003. Technical Publication SJ2006-1 to the St. J ohns River Management District, Palatka, Florida. SHINE, R., and J. B. IVERSON. 1995. Patterns of survival, growth and maturation in turtles. Oikos 72:343-348. SMITH, G. R., and J. B. IVERSON. 2002. Sex ratio of common musk turtles ( Sternotherus odoratus ) in a north-central Indiana lake: a longterm study. American Midland Naturalist 148:185-189. STEEN, D. A., M. J. ARESCO, S. G. BEILKE, B. W. COMPTON, E. P. CONDON, C. K. DODD, H. FORRESTER, J. W. GIBBONS, J. L. GREENE, G. JOHNSON, T. A. LANGEN, M. J. OLDHAM, D. N. OXIER, R. A. SAUMURE, F. W. SCHUELER, J. M. SLEEMAN, L. L. SMITH, J. K. TUCKER, and J. P. GIBBS. 2006. Relative vulnerabili ty of female turtles to road mortality. Animal Conservation 9:269-273. STEEN, D. A., and J. P. GIBBS. 2004. Effect of roads on the st ructure of freshwater turtle populations. Conservati on Biology 18:1143-1148. SWANNACK, T. M., and F. L. ROSE. 2003. Seasonal and ontogenetic changes in the sex ratio of a population of stinkpots (Kinosternidae: Sternotherus odoratus ). Southwestern Naturalist 48:543-549. SYSTAT SOFTWARE INC. 2004. SigmaPlot 9.0. Richmond, California. TOTH, D. J. and C. FORTICH. 2002. Nitrate concentrations in the Wekiva groundwater basin with emphasis on Wekiwa Springs. Technical Pub lication SJ2002-2 to the St. Johns River Management District, Palatka, Florida.
60 TUBERVILLE, T. D., J. W. GIBBONS, and J. L. GREENE. 1996. Invasion of new aquatic habitats by male freshwater turtle s. Copeia 1996:713-715. TUCKER, J. K., N. I. FILORAMO, and F. J. JANZEN. 1999. Size-biased mortality due to predation in a nesting freshwater turtle, Trachemys scripta American Midland Naturalist 141:198-203. WHITE, G. C., and K. P. BURNHAM. 1999. Program MARK: survival estimation from populations of marked animals. Bird Study 46:S120-S139. WILLIAMS, B. K., J. D. NICHOLS, and M. J. CONROY. 2001. Analysis and Management of Animal Populations. Academic Press, San Diego, USA. YATES, F. 1934. Contingency table involving small numbers and the 2 test. Journal of the Royal Statistical Society Series A-G Supplement 1:217-235.
61 BIOGRAPHICAL SKETCH I began my work with the turtles of We kiwa Springs State Park (WSSP) as an undergraduate freshman. What would later become my masters proj ect began as the trial run of a field biology class at Penn State Hazleton campu s, initiated by Dr. J. Brian Hauge. Myself, Dr. Hauge and 17 other young students traveled in a van from eastern Pennsylvania to Apopka, Florida over 20 hours at the start of spring break. I had never been to Florida and was fascinated by the sheer quantity of flora and fauna, though I pa rticularly enjoyed working with the turtles. When we returned from the trip, Dr. Hauge spent a great deal of time teaching me the basics of population abundance estimation using our limited ma rk-recapture data. Since that time, I have learned many techniques for analyzing mark-recap ture data, all with th e help of this ongoing turtle population study. I comple ted my undergraduate honors thesis on the population biology of these turtle species, and graduated Distinguishe d with Honors from Penn State University in May, 2004. I came to the University of Florida a few short weeks after graduating, and I have been quite fortunate to work with my major a dvisor Dr. Karen A. Bjorndal, who encouraged me to continue studying the turtles at WSSP. The tu rtle study at WSSP will continue for as long as possible, and we are curren tly in our seventh year.