Population Structure of the Florida Softshell Turtle, Apalone ferox, in a Protected Ecosystem, Wekiwa Springs State Park, Florida

The Florida softshell turtle, Apalone ferox (Schneider 1783) is the largest North American representative of the family Trionychidae and the second-largest freshwater turtle species in North America (Webb 1962, 1973), yet most of what we know about its natural history is anecdotal. This is surprising, as it is one of Florida's most “highly visible” turtles (Meylan and Moler 2006). Recently a comprehensive analysis regarding knowledge of turtles in the United States and Canada revealed that the genus Apalone was in the bottom third of studied genera and that the Florida softshell turtle ranked in the bottom 40% and is the least studied of North American softshell turtles (Lovich and Ennen 2013). Florida softshell turtles occur in many freshwater habitats throughout the state including lakes, rivers, large streams, and calcareous freshwater spring habitats (Meylan and Moler 2006). Despite being commercially and individually collected for food, little information has been published regarding habitat use or season of collection for this species. Bancroft et al. (1983) documented a strong seasonal effect on observations, with more animals being observed during the summer months than any other time of year. Although no data have been presented, it has been suggested that fishing for Florida softshell turtles with trot lines is most effective in the hottest months (Meylan and Moler 2006), which could be interpreted as support for a seasonal bias in catchability with increased activity during the summer months. Thus, there is some evidence for seasonal shifts in capture or observations of Florida softshell turtles, but study is warranted.

Also lacking is information on the population structure of this species, as capture techniques used for most other freshwater turtles (e.g., baited hoop nets and basking traps) are not as effective in capturing large numbers of softshell turtles (Moler and Berish 1995; Meylan and Moler 2006). Capture methods for Florida softshell turtles have included dip-netting from a boat (Bancroft et al. 1983), drift fence arrays set by a drying lake (Aresco 2003), trotlines (Iverson and Moler 1997), hoop nets (Aresco 2009; Johnston et al. 2011), and hand capture via snorkeling (Marchand 1942 in Carr 1952; Johnston et al. 2011; Weber et al. 2011). Multiple capture techniques may be required to properly sample all age classes within the population due to ontogenetic habitat or activity shifts.

The Florida softshell turtle is deemed common throughout much of its range (Meylan and Moler 2006). However, the species population structure is poorly understood despite the fact that this species was once the most-important commercially harvested turtle in Florida (Meylan and Moler 2006). Prior to 2000, Florida softshell turtles suffered unlimited harvest due to a lack of regulations. In 2000, the Florida Fish and Wildlife Conservation Commission (FWC) enacted a harvest closure on all softshell turtles and their eggs during the prime breeding and nesting season (1 May until 31 July; Interim FWC Rule 68A-25.002). In 2008, the FWC enacted an interim rule limiting the daily allotted possession of Florida softshell turtles to 5 turtles per person per day. During this interim period, licensed commercial fishermen could still remove up to 20 softshell turtles per day (Interim FWC Rule 68A-25.002). Commercial harvest of freshwater turtles in Florida ended in 2009 and, currently, the legal limit of take is 1 turtle per person per day, with some exceptions. Furthermore, no softshell turtles (Apalone spp.) may be taken from 1 May to 31 July, which reflects protection during the nesting season. Select permits that allowed harvest for aquaculture use remained in effect until 2012. These particular take permits have not been renewed (Interim FWC Rule 68A-25.002); thus, all commercial harvest of Florida softshell turtles has effectively ceased. Because few studies have been published on the life history of the Florida softshell turtle (Meylan and Moler 2006; Lovich and Ennen 2013), very little information is available on population levels or on how populations were affected by harvest and the subsequent ban. Only two studies have attempted to analyze and collect population data for this species (Bancroft et al. 1983; Aresco 2003). Neither study provided estimates of population size or survival; the methodologies employed have also been questioned as to their efficacy for sampling the populations (Meylan and Moler 2006).

We sampled a population of Florida softshell turtles during 2007–2012 as part of a multi-species monitoring program for turtles in a protected, spring-run complex. The objective of this study was to gain a better understanding of the Florida softshell turtle population at a protected site by describing sex ratios and estimating population size, survival, biomass, and density.

METHODS

Field-Site Description

The study took place at Wekiwa Springs State Park (WSSP), Orange and Seminole counties, Florida (lat 28°42′N, long 81°27′W). The park (WSSP) was established in 1970 and is now considered one of two wild and scenic rivers in the state. The study area includes the public swimming area, main lagoon, and spring run habitat (Fig. 1). The entire study area consists of approximately 2.67 ha of protected water habitats. Large, low-lying bottomland hardwood forested wetlands, dominated by black gum (Nyssa sylvatica) and bald cypress (Taxodium distichum) trees, surround the spring, along with sandhill uplands maintained by frequent prescribed fires encircling the spring area (Hrycyshyn 2006). The spring expels approximately 164 million l of water each day with a constant temperature of 22.0°C ± 2°C (Hubbs 1995; Scott et al. 2004). The area that surrounds the spring boil has been modified with concrete walls and steps to facilitate public recreational use. This small public swimming area (0.20 ha) opens into a much-larger natural lagoon (1.67 ha) which flows into the spring run. The lagoon is ~ 6.5 m deep at the deepest spot, and much of the bottom is covered with a thick (up to 1 m) layer of silt and detritus composed primarily of decaying vegetation from many years of herbicide application to control hydrilla (Hydrilla verticillata). The run is approximately 1.1 km long and ranges in width from 15 to 20 m (0.80 ha). The run eventually joins Rock Springs Run, which occurs within Rock Springs Run State Preserve, with over 14,000 acres of protected habitat that abuts WSSP to the west and north. The 2 runs merge to become the headwaters of the Wekiva River (Fig. 1).

View Figure

• Download as Powerpoint
• Download Figure

Surveys

We conducted 12 surveys from 17 March 2007 through 26 July 2012 to assess the turtle assemblage in the spring at WSSP. All species of turtles were targeted; however, here we present only Florida softshell turtle captures. Each year of the study we conducted 2 sampling sessions that averaged 5 d (range 4–10) depending on schedules and available personnel. The first sampling period was in mid- to late March, hereafter called Spring sampling, and the second sampling period took place from late-July to mid-August, hereafter called Summer sampling. For each sampling session, we set out three baited, double-throated hoop nets (1.9 m in diameter, 5.7 m long). Bait used to lure turtles into traps included fried chicken, sardines, or watermelon. While the hoop net traps were in place, a variable number of snorkelers, typically 10–20, snorkeled and hand-captured turtles for approximately 3–4 hrs each day, weather dependent. All captured turtles were placed in canoes and brought to a central location for processing.

Measurements were recorded to the nearest mm for all captured turtles and included maximum straight-line carapace length (CL), plastron length (PL), carapace width (CW), and shell height (SH). Body mass (BM) of turtles was measured to the nearest 1 g or 10 g using either Pesola spring scales (Pesola AG, Baar, Switzerland) or Ohaus top-loading scales (Ohaus Corp., Parsippany, NJ), depending on the size of turtle and scale used. Turtles were sexed based on size and relative tail length. Female Florida softshell turtles have much-shorter tails that barely extend past the carapace rim, whereas males have long, thick tails and their vents extend well past the carapace rim (Moler and Berish 1995; Ernst and Lovich 2009). If sex could not be determined, turtles were assigned to the juvenile class.

Florida softshell turtles were marked using a handheld, battery-powered tattoo wand (EZ TATT; Woody's Wabbits, Astoria, OR; Weber et al. 2011) and, since 2009, also with passive integrated transponder (PIT) tags injected into the muscle and connective tissue between the pelvis and the plastron, just lateral to the midline of the turtle. All turtles greater than 65 mm CL were implanted with PIT tags. Runyan and Meylan (2005) suggested that it was safe to PIT tag turtles larger than 55 mm CL; however, we chose 65 mm due to the size of our tagging equipment. Turtles were released back into the lagoon or spring run, depending on capture location.

We defined a capture as the first time a novel animal was caught. A recapture was defined as a previously identified animal captured at a subsequent sampling session, whereas a resample was a previously captured animal caught within the same sampling session. Capture and handling protocols were approved by The Florida Department of Environmental Protection (District III, Orlando) and the Institutional Animal Care and Use Committees (IACUC) at Pennsylvania State University and the University of North Florida and conformed to the American Society of Ichthyologists (ASIH), The Herpetologists League (HL), and the Society for the Study of Amphibians and Reptiles (SSAR) animal use guidelines (ASIH/HL/SSAR 2001).

Data Analyses

We compared body size between males and females using two-sample t-tests. Data were normally distributed for all measurements and variances were equal between samples (p  =  0.19–0.46) except for mass (p  =  0.003); thus, we used a t-test for unequal variances to compare mass.

Chi-square (χ²) tests were used to determine if the observed sex ratio differed from 1∶1 and to determine if captures of female, male, and juvenile turtles were different between Spring and Summer sampling sessions. We then used a sexual dimorphism index (SDI; Lovich and Gibbons 1992) to calculate the degree of difference in size to compare with other populations. The SDI was calculated as (CL ♀♀ / CL ♂♂) − 1. To compare sexual size dimorphism from our study population to that of other populations of Florida softshell turtles, we calculated SDI values from data in Bancroft et al. (1983) and Moler and Berish (1995).

We used POPAN parameterization of Jolly-Seber models (Schwarz and Arnason 1996) to calculate population abundance for adults and juveniles in Program MARK (White and Burnham 1999). We then used population estimates for each sex multiplied by the average mass of each sex to calculate biomass and calculated density by dividing by the area of the spring. To calculate total biomass, the biomass of adults was added to the juvenile mean mass multiplied by the juvenile population estimate. Biomass per hectare was calculated by dividing the total biomass by the area of the spring.

Recapture probability and survivorship were calculated using Cormack-Jolly-Seber (CJS) models (Cormack 1964; Jolly 1965; Seber 1965) in Program MARK (White and Burnham 1999). Recapture probability across the 12 sampling periods was calculated using CJS models in Program MARK. Apparent survival (Φ) and recapture rates (p) were calculated using open-population CJS models (Lebreton et al. 1992). Models were generated to test whether Φ or p values were best explained by group (male, female, juvenile) or were independent of group-level influences. Model selection was based on corrected Akaike Information Criterion (AIC_c) values, with lower values denoting greater parsimony (Burnham and Anderson 2002).

Traditionally, encounter rates are used to calculate the probability that an individual will leave a population. If the encounter rates are reversed, then the probability of an individual entering the population can be estimated (Pradel 1996). In doing so, λ can be estimated where λ  =  rate of individuals entering a population or cohort. This differs from a traditional calculation of λ in which the population growth rate is derived as a dominant eigenvalue from a projection matrix model. The λ estimated using Pradel models only estimates the realized growth rates of the age class from which the encounter rates were generated but is not necessarily equivalent to the growth rate of the population. Regardless, it still provides an important metric of the life-history characteristics of a population. Pradel's λ was estimated by Program MARK in conjunction with the CJS model described above.

RESULTS

We captured 36 individual adult Florida softshell turtles, with 10 males and 26 females, a male∶female sex ratio of 1∶2.6 that was significantly female-biased (χ²₁  =  7.11, p  =  0.008). Juvenile turtles (n  =  20) represented 36% of all turtles captured. One female had missing data on mass and was only used in comparisons of shell measurements but not weight. Three juvenile turtles were too small to permanently mark with the two marking methods used and so were not used in any of the population analyses. Female Florida softshell turtles at Wekiwa Springs were significantly larger than males, resulting in an SDI value for the population of 0.439 based on CL length (Table 1, Fig. 2).

Table 1. Mean (± SE) morphometric variables for Apalone ferox collected at Wekiwa Springs State Park, Florida 2007–2012.

View Fullscreen

View Figure

• Download as Powerpoint
• Download Figure

We had a total of 101 captures including recaptures and resamples; however, only 16 (12 individuals) of these were captured in hoop nets whereas the remainder were all captured by hand. No Florida softshell turtles < 120 mm CL were captured in the hoop nets. The largest female had a CL of 566 mm and weighed 18.9 kg, and the largest male had a CL of 421 mm and weighed 6.4 kg; both were captured in hoop nets. Adult turtles captured in hoop nets tended to be slightly larger than those captured by hand (Table 2).

Table 2. Florida softshell turtle (Apalone ferox) carapace lengths (mean ± SE) separated by method of capture from a population in Wekiwa Springs State Park, Orange and Seminole counties, Florida from 2007 to 2012.

View Fullscreen

For all years combined, more turtles were captured during Spring than during Summer (Table 3). However, the ratio of male∶female∶juvenile turtles captured was not significantly different between Spring and Summer sampling sessions (χ²₂  =  1.42, p  =  0.49; Table 3).

Table 3. Numbers of Apalone ferox captured by season, year, and sex. Numbers in parentheses are the total numbers captured (including turtles resampled within the same sampling session). Sex ratios are based on first captures only.

View Fullscreen

The population estimate for male Florida softshell turtles at Wekiwa Springs was considerably lower than for females and suggested a 1∶5.1 (♂∶♀) sex ratio (Table 4). Population densities based on estimates were 34.5/ha for adults and 18.4/ha for juveniles. Biomass was estimated at 290.4 kg/ha for adults and 7.6 kg/ha for juveniles, resulting in a total biomass estimate of 795.7 kg, or 298 kg/ha.

Table 4. Apparent survivorship (Φ), recapture probability (p), lambda (λ), and population size (n) (each ± 1 SE) for male, female, and juvenile Florida softshell turtles (Apalone ferox) at Wekiwa Springs State Park, Florida.

View Fullscreen

Apparent survivorship was similar for males and females, but slightly lower for juveniles, although λ was the same across all three classes (Table 4). Recapture probabilities were much lower for females than for males or juveniles (Table 4). The model that best explained the encounter histories was Φ(.)p(group), in which survival was independent of group while recapture probabilities varied by group (Table 5).

Table 5. Cormack-Jolly-Seber model set analyzing the effect of group (male, female, juvenile) on apparent survivorship and recapture rates of Apalone ferox at Wekiwa Springs State Park, Florida.

View Fullscreen

DISCUSSION

Few studies have attempted a comprehensive population survey for the Florida softshell turtle; specifically, ones looking at capture methodology, sex ratios, SDI, population estimates, and survivorship. Two previously published studies provide some information on Florida softshell turtle populations in lake systems; however, a potential bias exists in their survey methods. Bancroft et al. (1983) sampled by capturing softshells with dip nets from the side of a boat. This method may be effective for capturing smaller individuals but may underrepresent larger turtles due to their size (i.e., they may not fit into a dip net), wariness, and ability to swim rapidly to escape capture (Bancroft et al. 1983). Aresco (2003) sampled a drying lake using a drift fence. This method also has its limitations, as it assumes that all individuals exited the lake in the same direction (the lake was not completely fenced), or did not burrow into the mud, or both. Additionally, smaller individuals may have been more prone to predation or being overcome by heat stress on their way to the drift fence (Meylan and Moler 2006).

The vast majority of captures in the present study were the result of hand-capture while snorkeling, in which every turtle observed is targeted. However, some turtles are able to escape capture, particularly the larger, faster animals, thus some bias likely remains. Double-throated hoop-net traps proved rather inefficient at capturing Florida softshell turtles in this study. Although our hoop traps did not capture high numbers of Florida softshell turtles, the individuals captured tended to be large. In comparison, in a study that compared sampling method success between snorkeling and trapping, all softshell turtles, including 14 spiny softshell turtles (Apalone spinifera) and 1 Florida softshell turtle, were captured in hoop nets (Sterrett et al. 2010). The authors observed spiny softshell turtles on several occasions while snorkeling, but all eluded hand-capture due to rapid swimming. Our results were quite different; the majority of our animals were captured by hand, which may reflect a greater ability to capture softshells due to training and experience or that the size and shape of the aquatic system at Wekiwa Springs is more conducive to capturing softshell turtles.

Differences in seasonal activity may be a potential explanation for biased sex ratios; for example, because of more females being captured during the nesting season. However, we detected no difference in the number of captures between seasons, suggesting that there was no seasonal variation in capture rates. With few exceptions, we captured more females than males regardless of season or year. There are a few possible explanations for the strong skew toward females in our population estimates. Our sampling methods may be biased towards habitats that females prefer, those along the lagoon edges that offer structure (fallen trees and root systems as well as deep, open water). Males may prefer shallower, more-vegetated habitats that are largely inaccessible due to the sampling methods we employ. Another possibility is that females, being larger than males, are easier to see and give chase. In contrast to our results, the sex ratio reported by Dalrymple (1977) for Florida softshell turtles was approximately 1∶1. However, Plummer (1977) similarly observed female-biased captures in a population of smooth softshell turtles (Apalone mutica). A potential explanation for biased sex ratios could be that males and females differ in habitat preferences (Webb 1962), a topic that should be the study of future research on Florida softshell turtles.

Although little is known about Florida softshell turtles, in Lake Texoma, Texas and Oklahoma, Webb (1962) found that males of both spiny and smooth softshell turtles preferred open-water habitats, whereas reproductive females occupied shallower waters prior to egg deposition. Our sampling methods focused on the edges of the lagoon habitat, where the water is shallow and there are more structures and basking sites than in the center of the lagoon. Many of the Florida softshell turtles we captured were first observed at the edge of the lagoon and subsequently captured in deeper water after significant chases. The significant bias in population estimates reported in Table 3 may be a result of a habitat sampling bias.

Turtles captured during our study were larger than previously reported in other studies. The largest turtle captured during our study was considerably larger (566 mm CL) than reported in other studies. Dalrymple (1977) reported sizes from 150 to 400 mm CL near White City, Florida and 140 to 520 mm CL from Lake Okeechobee, Florida. Similar results were found in a survey of Lake Conway, Orange County, Florida, where the largest reported capture had a CL of 450 mm and a PL of 380 mm (Bancroft et al. 1983). Finally, a maximum PL of 340 mm was recorded from Lake Jackson, Leon County, Florida (Aresco 2003). There are several reasons why our turtles may be larger than what has previously been reported. The habitat at Wekiwa Springs is relatively pristine, having a constant temperature of 22.0° ± 2°C and a healthy, diverse food base (Hubbs 1995; Wetland Solutions, Inc 2007). Turtles captured in south and central Florida may have a longer growing season in comparison to northern populations. Another potential reason for capturing larger turtles could be that our study site is within a protected state park that has not recently been subjected to turtle harvest.

Few previously published works presented, or allowed us to calculate, the SDI for the Florida softshell turtle. We calculated SDI values from data presented on 11 Florida softshell turtles from a drying pond in San Felasco Hammock Preserve State Park, Gainesville, Florida, resulting in a value of 0.44 (Moler and Berish 1995), identical to our population (SDI  =  0.439). A much-lower SDI (0.19) was calculated from Lake Conway in central Florida (Bancroft et al. 1983). These differences may simply be a result of how turtles were captured. Dipnetting, the technique used by Bancroft et al. (1983), possibly limited capture of larger (female) turtles due to the size of the net being used and the speed of the animal targeted. Alternatively, it is unknown if harvest has impacted Lake Conway, but it could possibly explain why turtles in our protected system and the study site of Moler and Berish (1995) tend to be larger. Lake Conway, the location of the study by Bancroft et al. (1983), is in the area that Moler and Berish (1995) described as having heavy harvest of Florida softshell turtles. These harvests tend to target larger animals, usually female, which could explain the low SDI values of their study. Although we have no evidence for geographic variation of Florida softshell turtles, we cannot discount this as a contributing factor to the size differences between the compared studies. Regardless, our SDI value can serve as a baseline for future comparisons from a population that has not recently been subjected to harvest. Our recapture rates and subsequent population estimates may be low, as this species is extremely fast and because of the cryptic coloration that has them well camouflaged by the detritus layer on the bottom of the lagoon. We have lost numerous potential captures over the years due to turtles diving down into the detritus layer. If these individuals have been able to consistently evade capture, the population levels may be underestimated. On the other hand, if these individuals have already been marked and are occasionally recaptured, our estimates may not be affected to the same extent. In either case, continued and more-rigorous sampling will provide more-accurate population estimates.

We calculated biomass and density estimates based on adult and juvenile populations as well as the entire population within the community. For instance, approximately 91% of the biomass is attributable to females due to the extreme sexual size dimorphism observed in this species. There are currently no estimates for density or biomass provided for the Florida softshell turtle with which to compare our results, but estimates for congeners and emydids are instructive. Density and biomass estimates were 154/ha and 311.1 kg/ha, respectively, for a population of Florida cooters (Pseudemys floridana) in Florida and 42/ha and 19.4 kg/ha for smooth softshell turtles in Kansas (Iverson 1982). Density and biomass estimates from a lake system in Missouri included red-eared sliders (Trachemys scripta elegans) at 205.8/ha and 178.5 kg/ha and spiny softshell turtles at 1.9/ha and 7.1 kg/ha (Glorioso et al. 2010). Density estimates for two species of softshells in a large pond ecosystem in southern Illinois were much lower, 0.11/ha for smooth and 0.33/ha for spiny softshell turtles (Dreslik et al. 2005). Biomass for Florida softshell turtles in our population was thus much greater than that reported for other species of softshells and similar to that reported for populations of emydids. This was somewhat surprising as we expected low density and biomass of Florida softshell turtles, which are top predators with a primarily carnivorous diet. This result suggests that Florida softshell turtles may exert a large effect on the biota and community structure at WSSP.

Regardless of the seeming abundance of this species, common species have become uncommon in relatively short time spans (Gibbons et al. 2000). Before the 2009 Florida freshwater turtle harvest ban, this species was intensively harvested, with thousands of pounds of turtles being taken out of freshwater systems (Enge 1993; Meylan and Moler 2006). Our study site has been under protection since 1970, long before the 2009 turtle harvest ban, which suggests that this population of Florida softshell turtles may be in relatively good health compared to other unprotected locations. In the past, commercial harvest may have resulted in regional declines and even localized extirpations. Species with delayed sexual maturity and long lifespans are often very slow to replace the loss of breeding individuals (Brooks et al. 1991). A chronic decrease in adult survival within a population would require populations to have exaggerated high levels of juvenile survivorship in order to replace the loss of these adults, something that is seldom observed and highly unlikely to occur in turtle populations (Brooks et al. 1991; Congdon et al. 1993). Population recovery would be very slow wherever large numbers of breeding individuals are removed by harvest. Harvest, however, is not the only threat facing turtles. Habitat loss and degradation have been identified as major factors contributing to widespread declines of turtle species (Buhlmann et al. 2009). Although these factors may not be as extreme in a protected spring ecosystem, they still occur (i.e., anthropogenic disturbance, pollution, water quantity and quality) in the surrounding landscape. Despite this habitat degradation, our estimates of survivorship and population density are good benchmarks for examining the future of this population at WSSP and for comparison to other populations.