1 INTRODUCTION

Sex-biased dispersal occurs when one sex is more likely to disperse or tends to disperse further than the other sex (Greenwood, 1980). The sex that disperses less is more likely to remain philopatric to its natal area (Greenwood, 1980). Sex-biased dispersal has only been widely studied in birds (reviewed in Greenwood, 1980; Pusey, 1987) and mammals (reviewed in Greenwood, 1980; Lawson Handley & Perrin, 2007) but has been detected in many other vertebrate taxa (amphibians: e.g., Liebgold et al., 2011; Palo et al., 2004; teleost fish: e.g., Bekkevold et al., 2004, squamates: e.g., Dubey et al., 2008; Stow et al., 2001; and turtles: e.g., Moore et al., 2020; Paquette et al., 2010; Sheridan et al., 2010). While inbreeding avoidance is considered an ultimate cause of sex-biased dispersal, it does not help predict which sex is more likely to disperse or disperse further (Prugnolle & De Meeus, 2002; Pusey, 1987). This is because dispersal of either sex alleviates the potential for inbreeding (Lebigre et al., 2010; Perrin & Mazalov, 2000). However, there are a number of ecologically-based hypotheses that have been proposed based on bird and mammal systems for why a particular sex is the dispersing sex (Goudet et al., 2002). Despite their formation based on data from birds and mammals, these hypotheses, along with relevant ecological information, can help us predict which sex is more likely to disperse in non-avian and non-mammalian taxa (e.g., Dubey et al., 2008; Liebgold et al., 2011).

Many bird species exhibit female-biased dispersal (Prugnolle & De Meeus, 2002; Pusey, 1987) as a result of resource competition (Greenwood, 1980). In birds, males are often involved in parental care, most typically feeding offspring, so males will remain philopatric to retain familiarity with their local environment, which is associated with foraging success and increased survival of offspring (Greenwood, 1980; Perrin & Mazalov, 1999). In contrast, in mammals, males are more likely to disperse to avoid kin competition (Greenwood, 1980), potentially due to male–male competition for mates and resulting avoidance of kin competition (Greenwood, 1980; Lawson Handley & Perrin, 2007). Additionally, female mammals provide most parental investment (Greenwood, 1980; Pusey, 1987) and may benefit more from philopatry (and familiarity with the local foraging environment) than males (Perrin & Mazalov, 1999).

There has been less generalization about potential causes and direction of sex-biased dispersal in non-avian reptiles despite a plethora of studies investigating and documenting its existence. All of the studies testing for sex-biased dispersal found male-biased dispersal (snakes: Dubey et al., 2008; Folt et al., 2019; Francois et al., 2021; Hofmann et al., 2012; Keogh et al., 2007; Rivera et al., 2006; Zhong et al., 2017; Zwahlen et al., 2021; lizards: Dubey & Shine, 2010; Johansson et al., 2008; Olsson & Shine, 2003; Stow et al., 2001; crocodiles: Tucker et al., 1998; turtles: Casale et al., 2002; Chaloupka & Limpus, 2002; FitzSimmons et al., 1997; Lee et al., 2007; Moore et al., 2020; Paquette et al., 2010; Sheridan et al., 2010) with the exception of one species of lizard (Chapple & Keogh, 2005). Our understanding of which sex is more likely to disperse during sex-biased dispersal in birds and mammals can provide insight into the causation underlying the general pattern of male-biased dispersal in non-avian reptiles. Similar to most mammals, many reptiles have intense male–male competition and polygynous mating systems, which may lead to male-biased dispersal (Rivera et al., 2006; Tucker et al., 1998). Non-avian reptiles frequently use male–male combat in territorial interactions or to compete for mates (reviewed in Vitt & Caldwell, 2014). In accordance with the local mate competition hypothesis (Greenwood, 1980), male non-avian reptiles may disperse to avoid kin competition (Dubey et al., 2008), as has been observed in many species of mammals (Lawson Handley & Perrin, 2007). Unlike avian reptiles (i.e., birds), males in non-avian reptile species tend to have little to no parental investment or parental care (reviewed in Vitt & Caldwell, 2014).

Few studies have tested for sex-biased dispersal in turtles (clade Testudines), however, and all such studies have been either in marine turtles (e.g., Casale et al., 2002; Lee et al., 2007; Sheridan et al., 2010) or fully-terrestrial turtles (Moore et al., 2020; Paquette et al., 2010), with no tests of sex-biased dispersal in freshwater turtle species to date. All turtles studied so far displayed male-biased dispersal. This is potentially because, similar to other reptiles such as squamates, turtles have little-to-no male parental investment and, in many species, male–male competition is common (Vitt & Caldwell, 2014). Therefore, turtles, like other non-avian reptiles, may be more likely to display male-biased dispersal due to avoidance of male–male competition with kin, consistent with the local mate competition hypothesis (sensu Greenwood, 1980). Alternatively, male-biased dispersal may be present because it is an ancestral trait of all turtles or all reptiles.

Like most species of turtles, and all freshwater turtles, nothing is known about sex-biased dispersal in spotted turtles (Clemmys guttata), and little is known about sex-biased dispersal in other turtles in the family Emydidae (but see Moore et al., 2020; Sheridan et al., 2010). Clemmys guttata is found in the eastern part of North America and occupies a variety of wetland habitats that are often semi-terrestrial although they aggregate in shallow ponds to breed (Ernst & Lovich, 2009). Clemmys guttata is listed on the IUCN Red List as endangered and is listed in Appendix II of CITES (Convention on International Trade in Endangered Species; CITES, 2013) as well as currently being reviewed by the U.S. Fish and Wildlife Service to determine whether it should be considered under the Endangered Species Act. Clemmys guttata has experienced 50% reduction in numbers of individuals in three generations (Van Dijk, 2011) and studies have shown that many populations of C. guttata have low densities, which has led to concerns about maintenance of genetic diversity and calls for the need for increased genetic conservation (Ernst & Lovich, 2009).

If a population is small, it has a higher chance of inbreeding and, once inbreeding starts to occur, the population size is likely to continue to get smaller each generation unless new individuals or alleles are introduced, which can be alleviated by sex-biased dispersal (Guillaume & Perrin, 2006; e.g., Folt et al., 2019). Maintaining genetic diversity, potentially in part via dispersal, is critical to endangered turtle species like C. guttata so they can avoid negative effects of inbreeding depression (Parker & Whiteman, 1993). Inbreeding can increase homozygosity within a population (Charlesworth & Charlesworth, 1987), potentially leading to decreased fitness (inbreeding depression) and smaller overall population size (Charlesworth & Charlesworth, 1999).

Few studies on genetic variation in pond turtle (Family Emydidae) populations have been done and very few on turtle species of conservation concern to understand the potential threats from inbreeding and the potential benefits of sex biases in dispersal. Blanding's turtle (Emydoidea blandingii), and bog turtles (Glyptemys muhlenbergii) turtles are both threatened or at risk throughout their range with populations of G. muhlenbergii having low genetic diversity, ostensibly due to habitat loss and small population sizes (Rosenbaum et al., 2007). In E. blandingii, however, genetic diversity is high because they seem to have sufficient habitats to allow gene flow (McCluskey et al., 2016, 2022; Reid et al., 2016) and are long-lived (McCluskey et al., 2016), which alleviates or delays the signature of genetic isolation (McCluskey et al., 2016) despite some potential ecological constraints to long-distance movements (McCluskey et al., 2022). The first genetic study on C. guttata in the northern extent of its range in Canada found population differentiation (Davy et al., 2014). In contrast, further toward the center of their range in Rhode Island, Buchanan et al. (2019) recently reported little genetic differentiation between populations and found similar genetic diversity to a common species, the Eastern Painted Turtle (Chrysemys picta).

We hypothesized that the C. guttata will exhibit male-biased dispersal, ostensibly to avoid inbreeding (Lebigre et al., 2010; Perrin & Mazalov, 2000), similar to other turtles and other species of reptiles. Male-biased dispersal was found in nearly every other species of reptile studied. While this could be due to fixation of male-biased dispersal from the ancestor of all reptiles, in most of the reptiles studied, male reptiles often display more intense intra-specific competition, potentially leading them to be more likely to have males disperse to avoid kin competition (e.g., Perrin & Mazalov, 2000).

2 METHODS

2.3 Tissue collection

During the months of January–July in 2016, each site was visited for four trapping periods and, from January to June in 2017, each site was visited for three trapping periods to mark-recapture C. gutatta. We used handmade D-hoop traps (87 × 50 cm) and collapsible mesh minnow/crawfish bait traps (61 × 30.5 cm; Promar) baited with sardines in oil. Once traps were set, they were checked every 24 h. When turtles were captured, we identified spatial coordinates of the trap using GPS units (Garmin), determined their sex (Figure 1) and notched their marginal scutes in a unique pattern to identify individuals (modified from Cagle, 1939). Each C. guttata adult was swabbed in their oral cavity to collect saliva and cheek cells for DNA analysis (sensu Poschadel & Möller, 2004). Swabs were then processed by air-drying for 15 min and freezing until DNA extraction.

2.4 Microsatellite genotyping

We extracted DNA from buccal swabs Figure (2) using a MasterAm Buccal Swab DNA Extraction Kit (Epicentre). We genotyped male and female adult spotted turtles at 12 repeat microsatellite loci (Table 2), 10 that were originally isolated from bog turtles (Glyptemys muhlenbergii), and two microsatellite loci originally isolated from Blanding's turtles (Emydoidea blandingii), all of which have been previously used to amplify C. guttata DNA (Davy et al., 2014; King & Julian, 2004; Osentoski et al., 2002). The loci were amplified using PCR protocols modified from King and Julian (2004) and McCluskey et al. (2016). PCR products were visualized using an ABI 3730xl Automated Sequencer (Applied Biosystems). Alleles were sized and scored using GENEMARKER (v2.6, Softgenetics). We tested for Hardy–Weinberg equilibrium and heterozygote deficiency at each locus at each site, and linkage disequilibrium for each pair of loci using GENEPOP (v4.0.10) at 1000 iterations (Raymond & Rousset, 1995; Rousset, 2008). We used MICROCHECKER (v2.2.3) to check each locus for potential null alleles and scoring errors (Oosterhout et al., 2004). We calculated observed and expected heterozygosities (H_O and H_E, respectively) and number of alleles per locus (Table 2) and H_O, H_E and the number of rarified alleles per locus per population (sensu Kalinowski, 2004, 2005; Table 3). We then used the linkage disequilibrium method (LDNe) implemented in N_eESTIMATOR (v2.01, Do et al., 2014) to estimate effective population size (N_e) for each site and compared N_e to mark-recapture estimates of population size (N_c; sensu Frankham, 1995; Nunney, 1995), calculated with the POPAN estimator in Program MARK (Cooch & White, 2005; White & Burnham, 1999) in Lamb (2018).

TABLE 2. Variation in the 12 microsatellites used: Range of alleles, number of alleles, observed heterozygosity (H_O) and expected heterozygosity (H_E) for each locus

Locus	Size range (base pairs)	Number of alleles	H O	H E
EB09	113–167	22	0.250	0.892
EB19	93–99	3	0.194	0.513
GmuA19	123–135	7	0.664	0.670
GmuB08	188–209	7	0.492	0.631
GmuD16	122–188	15	0.758	0.865
GmuD21	138–178	9	0.438	0.635
GmuD55	177–213	10	0.392	0.808
GmuD79	157–189	8	0.618	0.779
GmuD87	198–276	18	0.621	0.922
GmuD88	110–134	7	0.464	0.597
GmuD114	93–109	5	0.396	0.530
GmuD121	129–162	9	0.627	0.819

• Note: Loci with significant deviations from Hardy–Weinberg equilibrium (italicized) and significant heterozygote deficiency (bolded) at a majority of sites (after a Bonferroni correction, Table 2) are bolded and italicized.

TABLE 3. Variation in the 12 microsatellite loci used: Rarified number of alleles (R_a), observed heterozygosity (H_O) and expected heterozygosity (H_E) for each locus at each site

Locus	Site A N = 41			Site B N = 49			Site C N = 30			Site D N = 25			Site E N = 19
	R a	H o	H E	R a	H o	H E	R a	H o	H E	R a	H o	H E	R a	H o	H E
Eb09	6.2	0.227	0.762	7.9	0.179	0.912	7.7	0.313	0.859	8.8	0.467	0.936	6.3	0.176	0.706
Eb19	1.9	0.111	0.337	2.9	0.289	0.591	2.5	0.115	0.307	2.7	0.000	0.552	3.0	0.421	0.679
GmuA19	2.6	0.727	0.629	2.6	0.524	0.587	3.9	0.821	0.736	3.3	0.636	0.763	3.0	0.684	0.624
GmuB08	4.6	0.600	0.582	3.0	0.465	0.566	2.6	0.379	0.518	4.5	0.565	0.769	2.8	0.385	0.582
GmuD16	5.5	0.760	0.781	5.9	0.762	0.839	5.3	0.833	0.871	2.0	0.000	0.667	5.4	0.737	0.819
GmuD21	3.6	0.400	0.614	4.5	0.550	0.788	3.6	0.654	0.658	3.4	0.167	0.525	3.6	0.211	0.440
GmuD55	5.5	0.478	0.738	4.0	0.222	0.856	5.8	0.429	0.714	7.2	0.563	0.839	3.0	0.500	0.667
GmuD79	1.0	0.652	0.824	4.3	0.583	0.765	2.0	0.632	0.734	2.0	0.846	0.757	2.8	0.357	0.680
GmuD87	4.6	0.438	0.602	5.4	0.405	0.569	6.0	0.500	0.579	1.0	1.000	0.644	5.3	0.444	0.556
GmuD88	3.6	0.483	0.805	3.9	0.565	0.922	3.4	0.821	0.929	1.0	0.333	0.333	2.7	0.611	0.852
GmuD114	2.0	0.176	0.386	2.7	0.500	0.547	2.4	0.533	0.549	4.3	0.555	0.680	2.0	0.500	0.437
GmuD121	2.0	0.556	0.764	6.5	0.793	0.834	3.7	0.476	0.676	2.0	0.600	0.846	2.0	0.625	0.817
Averages	3.6	0.467	0.652	4.5	0.486	0.731	4.1	0.542	0.678	3.5	0.478	0.652	3.5	0.471	0.655

• Note: Loci with significant deviations from Hardy–Weinberg equilibrium are italicized and loci with significant heterozygote deficiency at a majority of sites (after a Bonferroni correction) are bolded.

3 RESULTS

4 DISCUSSION

We found that both male and female C. guttata, separate and combined, had significant fine-scale spatial genetic structure, with high relatedness at proximate locations <25 km and less than expected relatedness at distances beyond 25 km (Figures 3 & 5). We note that the largest distance between C. guttata to the maximum 25 km distance was 17 km and the smallest distance in the 25–50 km distance class was 30 km, so where in this 17–30 km range is too far for most C. guttata to travel is still to be determined. Little is known about dispersal distances in C. guttata, so knowledge that C. guttata rarely disperse to different sites beyond 17–30 km may be relevant to conservation of this species, as is determining if other landscape factors (e.g., large roads, rivers, and farmland) affect dispersal (Lamb, 2018).

Despite the detection of these strong patterns and restricted dispersal using genetic spatial autocorrelation (Peakall & Smouse, 2006; Smouse & Peakall, 1999), the sex-specific patterns paralleled each other closely. There was, therefore, no evidence for sex-biased dispersal (sensu Banks & Peakall, 2012). We had predicted we would detect male-biased dispersal in C. guttata as it has been detected in almost all reptile species studied to date including turtles (e.g. Moore et al., 2020; Sheridan et al., 2010; but see Chapple & Keogh, 2005). Our expectations were due to male–male competition for mates without paternal investment in offspring being common in reptiles (Vitt & Caldwell, 2014) similar to mammals (Greenwood, 1980; Lawson Handley & Perrin, 2007). However, despite finding significant and strong patterns of fine-scale genetic structure in our study in each sex, we detected no evidence for sex-biased dispersal in C. guttata (Figures 3 & 5; Tables 4 & 5).

Our study contrasts with the handful of previous studies in turtles, which all detected male-biased dispersal using population genetic methods. A single study inferred that a lack of female dispersal by nesting female E. blandingii may play a role in female philopatry by comparing nesting females to non-nesting females and males combined, but that study did not compare dispersal between the two sexes (Reid et al., 2016). With the exception of that study, the other studies, which found male-biased dispersal, were all conducted in saltwater turtles or fully terrestrial turtles, not freshwater turtles like C. guttata; two of these studies were in species in the same family as C. guttata, family Emydidae (Moore et al., 2020; Sheridan et al., 2010). However, in many reptile species, including turtles, males also have evolved, via sexual selection, larger size or more or brighter coloration compared to the females as a result of greater intensity of male–male interactions (reviewed in Brejcha & Kleisner, 2016). Unlike all the other reptile species tested for sex-biased dispersal, C. guttata is unique because it exhibits sexual dimorphism in which the female has brighter coloration – not males, while males and females are of similar size (Figure 1; Ernst & Lovich, 2009; Rowe et al., 2012). The reasons for this reversal in sexual dimorphism are unknown, although plausibly it could be due to some degree of sex-role reversal or female–female competition (sensu Blizzard & Pruett-Jones, 2017). If females are competing to some degree, and male–male mating competition is reduced relative to females, sex biases in dispersal patterns can be altered (Le Galliard et al., 2006; Pérez-González & Carranza, 2009). More research is needed to test these hypotheses in C. guttata. However, the potential lack of sex-biased dispersal in C. guttata stands in stark contrast to other turtle species where male-biased dispersal occurs (e.g., Moore et al., 2020; Sheridan et al., 2010, but see Chapple & Keogh, 2005). If there was a lack of male-biased dispersal in any species where the ancestral pattern is male-biased dispersal, it could be due to females having increased dispersal and/or males having reduced dispersal, both of which could change the ratio of male: female dispersal. The result of either of these behaviors would be dispersal patterns that are similar between the sexes (Figures 3 & 5) and not having the male-biased dispersal present in other reptiles. However, despite significant spatial genetic structure being present for each sex hinting that sample sizes are adequate for identifying overall genetic spatial structure, it is still possible that a larger sample size could lead to detection of subtle differences between the sexes within distance classes (bins). More research is needed to determine if larger sample sizes are needed in this species compared to other similar studies in reptiles or if sex biased dispersal is present in this species in different geographic areas or different time periods (sensu Liebgold et al., 2013).

An additional seemingly minor, but methodologically important, result was that we found similar patterns whether or not we included a large number of microsatellite loci, relative to the total number of loci used (n = 5 out of n = 12), that were not in Hardy–Weinberg equilibrium or had heterozygote deficiency in a majority of populations (Table 3), potentially due to null alleles, to the dataset of loci that were in Hardy–Weinberg equilibrium in most populations (n = 7). In contrast to the assumption that loci deviating from Hardy–Weinberg equilibrium can cause Type I statistical errors, in our analyses, inclusion of the five loci did not change the results but simply increased resolution to a small degree, in a manner consistent with the increased resolution expected by the increased number of loci despite the likelihood that these loci each possessed reduced power. That the general, statistically significant, spatial genetic structure patterns for all turtles combined (Figure 4 vs. Figure 6) did not change, or became more significant, when we included loci that deviated from Hardy–Weinberg equilibrium supports the hypothesis that some factors, such as, potentially, null alleles may have weak effects on genetic analyses of spatial genetic structure (Chapuis & Estoup, 2007), at least in this one study. On the other hand, we cannot rule out that a lack of statistically significant sex-biased dispersal in this study despite increased number of loci (Figure 3 vs. Figure 5) may stem from the additional loci (containing null alleles) providing weak additional power. As a result, it is possible that an overall number of loci studied that include loci with null alleles may provide a false sense of security that a sufficient number of loci have been utilized. Future research and modeling that compares the inclusion of loci that deviate from Hardy–Weinberg equilibrium is warranted to determine if using a test of Hardy–Weinberg equilibrium may or may not be appropriate litmus test to determine which loci should be included in certain types of population genetic analyses.

5 CONCLUSION

We did not detect differences between the sexes in fine-scale spatial genetic structure that would have indicated sex-biased dispersal occurs in C. guttata. However, we did detect significant overall fine-scale spatial genetic structure in each sex and combined (Figures 4 & 6), with high relatedness in proximate locations <25 km and less than expected relatedness at distances beyond 25 km (Tables 4 & 5), indicating that the overall genetic patch size for C. guttata in this part of the Atlantic Coastal Plain is <25 km. This is important from the standpoint of conservation as rapid declines in C. guttata, due to factors such as habitat loss and the pet trade, have reduced the number of populations (Van Dijk, 2011). In our area, populations are typically 30–40 km apart (Lamb, 2018). Despite dispersal typically occurring within 25 km in our study between our nearest populations, it is still unclear if populations are becoming more genetically isolated. Genetic diversity remains high in C. guttata populations (Buchanan et al., 2019; this study). Clemmys guttata are long-lived with a maximum lifespan between 51 and 110 years (Litzgus, 2006; Seburn, 2018), and most anthropogenic changes within a lifetime or two, so genetic tools may not be able to detect consequences of population isolation for at least several more generations (sensu McCluskey et al., 2016). Further studies should be conducted in this and other areas to identify if there are undiscovered proximate populations within 25 km of each other to determine if dispersal can alleviate the effects of potential future inbreeding depression (Buchanan et al., 2019) or if populations are currently too far apart for enough gene flow to alleviate inbreeding.

AUTHOR CONTRIBUTIONS

Eric B. Liebgold: Conceptualization (equal); data curation (equal); formal analysis (equal); investigation (equal); methodology (equal); project administration (equal); resources (equal); software (equal); supervision (equal); validation (equal); visualization (equal); writing – original draft (equal); writing – review and editing (equal). Myra J. Dickey: Conceptualization (equal); formal analysis (equal); investigation (equal); methodology (equal); writing – original draft (equal); writing – review and editing (equal). Stephanie M. Lamb: Data curation (equal); investigation (equal); methodology (equal); writing – review and editing (equal). Tami S. Ransom: Data curation (equal); investigation (equal); methodology (equal); project administration (equal); writing – review and editing (equal). Hunter Howell: Data curation (supporting); investigation (supporting); writing – review and editing (supporting).