Hybridization and low genetic diversity in the endangered Alabama red‐bellied turtle (Pseudemys alabamensis)

Abstract Pseudemys alabamensis is one of the most endangered freshwater turtle species in the United States due to its restricted geographic distribution in coastal Alabama and Mississippi. Populations of P. alabamensis are geographically isolated from one another by land and saltwater, which could act as barriers to gene flow. It is currently unknown how differentiated these populations are from one another and whether they have experienced reductions in population size. Previous work found morphological differences between Alabama and Mississippi populations, suggesting that they may be evolutionarily distinct. Other Pseudemys turtles such as P. concinna and P. floridana occur naturally within the same geographic area as P. alabamensis and are known to hybridize with each other. These more abundant species could threaten the unique genetic identity of P. alabamensis through introgression. In order to evaluate the endangered status of P. alabamensis and the level of hybridization with other species, we used mitochondrial and nuclear microsatellite markers to assess genetic variation within and among populations of P. alabamensis throughout its range and estimate admixture with co‐occurring Pseudemys species. In P. alabamensis, we found no variation in mitochondrial DNA and an excess of homozygosity in microsatellite data. Our results show genetic differentiation between Alabama and Mississippi populations of P. alabamensis, and low estimated breeding sizes and signs of inbreeding for two populations (Fowl River, Alabama and Biloxi, Mississippi). We also found evidence of admixture between P. alabamensis and P. concinna/P. floridana. Based on our results, P. alabamensis is highly endangered throughout its range and threatened by both low population sizes and hybridization. In order to improve the species’ chances of survival, focus should be placed on habitat preservation, maintenance of genetic diversity within both the Mississippi and Alabama populations, and routine population‐monitoring activities such as nest surveillance and estimates of recruitment.


| INTRODUC TI ON
The southeastern United States is a biodiversity hot-spot, harboring higher levels of endemic species than other areas of the country (Jenkins et al., 2015). Alabama, in particular, has a high concentration of regionally endemic species, especially freshwater turtles, and occurs within one of three global turtle priority areas for conservation (Buhlmann et al., 2009;Lydeard & Mayden, 1995). Freshwater turtles are a conservation concern worldwide, with >60% of species classified as threatened (Buhlmann et al., 2009). While some turtle species in the southeastern US are not currently imperiled, others have multiple risk factors for extinction such as low population size and restricted habitat range (IUCN, 2001;Mace et al., 2008;Purvis et al., 2000). The Alabama red-bellied turtle (Pseudemys alabamensis) is among the most atrisk turtle species in the US and is considered by some to be "the most endangered turtle on the continent" (Spinks et al., 2013).

Although it is classified as endangered by both the US Fish and
Wildlife Service (USFW, 1987) and the International Union for Conservation of Nature (IUCN) Red List, studies on this species across its entire distribution are lacking. This dearth of information prevents development of targeted management and conservation actions. Although P. alabamensis does occur within some protected areas (Heaton et al., 2021), there are currently no specific survey activities or targeted management actions to ensure monitoring and protection of this species (Figure 1).
Pseudemys alabamensis is threatened by habitat modification, including dredging, road-kill of adults and juveniles, and competition with other species (Nelson et al., 2009). Turtles may also be used for shooting practice (Alexander, 2018). This species has a very limited distribution and is found exclusively in coastal rivers along Mobile Bay in Alabama and the Mississippi Sound ( Figure 2) (Leary et al., 2008). An isolated population once existed further inland in southwestern Alabama not far from Little River State Park, but has since been extirpated (Mount, 1975). The freshwater bodies currently inhabited by P. alabamensis are separated by land and saltwater, which likely prevents substantial movement of individuals between river populations. In fact, although P. alabamensis shows some tolerance to brackish water, it does not occur in saltwater and is known to only disperse on land for nesting purposes at distances between 30 and 130 m from water bodies (Nelson et al., 2009). Some morphological differences have been previously noted between Alabama and Mississippi populations of P. alabamensis, such as the dorsal width of the cervical scute (Leary et al., 2003), supporting the existence of isolated populations within this species.
Despite the small range and fragmented populations of P. alabamensis, virtually nothing is known about key factors needed for developing a species survival plan such as population size, potential existence of genetically differentiated populations, and estimates of the level of admixture with closely related sympatric Pseudemys species. Many of these issues can be resolved with a range-wide study to assess population connectivity, genetic diversity, and levels of admixture among sympatric populations, and to establish appropriate conservation units for this species and consequently identify priority areas for monitoring and protection. To date, genetic data on P. alabamensis have been collected on a relatively small sample size to clarify the taxonomic status of this species (Jackson et al., 2012;Spinks et al., 2013) or to assess genetic diversity at a single locality (Hieb et al., 2014). These studies found complex relationships between species in the Pseudemys genus, possibly originating from hybridization and introgression, and low genetic diversity for the Mobile-Tensaw Delta population of P. alabamensis in Alabama.
Although hybridization has been observed within the Pseudemys genus, there are no documented cases of hybridization with P. alabamensis despite its co-occurrence with two other Pseudemys species, P. concinna and P. floridana, which are known to hybridize in the area (Mount, 1975). In addition to observed hybridization of other Pseudemys species, we have also made anecdotal observations of mixed shell morphologies within P. alabamensis (N. Moreno, personal observation). Introgression with native P. concinna and P. floridana, or non-native species that may have been introduced to the area, would have major conservation implications for P. alabamensis, as it would threaten the unique genetic identity of an already highly geographically restricted species with a likely low population size.
Here, we utilize mitochondrial DNA (mtDNA) and microsatellite markers to (1) identify the genetic structuring of populations of P.

Conservation genetics
F I G U R E 1 Pseudemys alabamensis in its natural environment. Photo credit: Nickolas Moreno alabamensis, (2) measure intraspecific genetic diversity, (3) investigate the possibility of recent reductions in population sizes, and (4) assess potential hybridization with sympatric species. Our results can serve as a necessary basis to further develop conservation and management activities in collaboration with local conservation organizations and authorities and to raise awareness of the current status of imperilment of this species. As climate change increasingly impacts coastal populations, understanding the current distribution and genetic diversity of P. alabamensis will be critical for determining its long-term survival potential, especially for small and isolated populations. F I G U R E 2 Projected range of P. alabamensis based on GIS-defined hydrologic unit compartments (HUCs) created around capture locations from this study along with data from Nelson (1994Nelson ( , 1995Nelson ( , 1996Nelson ( , 1997Nelson ( , 1998, Leary et al. (2003), and Jackson et al. (2012). Approximate locations of rivers sampled within the range are marked by numbers as follows (numbers as in Table 1 Note: Species identification was based on morphological assessment. Sampling effort is displayed as the number of "trap nights" where one trap is set for one night. Numbers next to sampled watersheds correspond to numbers on the map in Figure 2. Numbers of individuals for each sex are indicated in parentheses as (male, female, juvenile unsexed). Number of individuals per effort refers only to P. alabamensis captures. a This locality includes both roadkill and samples from live individuals collected in the water. The sampling effort for this locality refers only to samples obtained from live individuals collected in water. b Donated samples.

| Sample collection
Fieldwork was carried out from February to November in 2018-2019 throughout the range of P. alabamensis ( Figure 2 In addition to trapping turtles, samples were also collected from Due to admixture between individuals of the cooter complex in the area (P. concinna and P. floridana), many individuals captured in this study presented mixed morphological characteristics; therefore, individuals were identified to the most similar species following morphological descriptions of the species in Alabama as in Mount (1975) and Leary et al. (2008). Blood for DNA extractions was collected from the subcarapacial sinus of each turtle. The skin of animals at the site was treated with 70% isopropyl alcohol prior to drawing blood. A maximum of 0.5% of body weight (max 2 ml of blood per turtle) was collected from each animal using a 23-gauge needle and a 3-ml syringe. All animals were released at the point of capture after blood sampling was performed and after ensuring that the puncture site was not bleeding and the animal was well. One ml of sampled blood was stored in 2-ml microcentrifuge tubes with 1 ml of prepared blood preservative that consisted of 100 mM Tris-HCL, 100 mM EDTA, 10 mM NaCl, and 0.5% SDS. Samples were stored on ice until returned to the lab where they were then placed at −20°C for long-term storage until DNA extractions were performed. DNA extractions were carried out using the Qiagen DNeasy Blood and Tissue kit (Qiagen, Inc., Valencia, CA) following the manufacturer's instructions for nucleated blood.

| Mitochondrial DNA amplification and analysis
Fragments of the mitochondrial control region were amplified using the primers Des-1 and Des-2, which were originally developed by Starkey et al. (2003) for the painted turtle (Chrysemys picta). Twentyfive μl reactions were prepared using 12.5 μl GoTaq G2 Green Master Mix (Promega), 0.5 μl 10 mg/ml bovine serum albumin, 1.2 μl each of 10 μM forward and reverse primers, 6.8 μl H 2 O, and 2.8 μl DNA extract. PCR conditions were as follows: 95°C for 3 min, 35 cycles of 95°C for 1 min, 55°C for 30 s, 72°C for 1 min; and a final 10-min extension at 72°C. PCR products were checked on a 1% agarose gel to ensure proper amplification and then purified using ExoSAP-IT (Applied Biosystems) according to the manufacturer's instructions. Sequencing was carried out by the DNA Analysis Facility at Yale University. Sequences were checked and manually edited using FinchTV (Treves, 2010). Cleaned sequences were aligned and collapsed into haplotypes using UGENE (Okonechnikov et al., 2012).

Haplotypes were inputted into a BLAST (Basic Local Alignment
Search Tool) search against the National Center for Biotechnology Information (NCBI) database. DnaSP (Rozas et al., 2017) was used to estimate haplotype diversity of all three species based on morphological assignment for each population. In order to visualize haplotype sharing between species, a parsimony haplotype network was created in PopART v1.7 (Leigh & Bryant, 2015) using the TCS method (Clement et al., 2000).

| Microsatellite DNA amplification and analysis
Eight microsatellite loci were amplified in P. alabamensis, P. concinna, and P. floridana. These microsatellites were originally developed by King and Julian (2004) ) were assessed with the software Genetix v4.05 (Belkhir et al., 2004). The program Fstat v2.9.4 was used to generate a sample size corrected allelic diversity (Goudet, 2003). Private alleles were considered for each population within each species (Petit et al., 1998) and  (Cornuet & Luikart, 1996) was used under all three mutational models available to detect signatures of historic bottlenecks within populations. The program Ne ESTIMATOR was used to infer breeding population size estimates for each river population (Do et al., 2014).
The program STRUCTURE v2.3.4 (Pritchard et al., 2000) was used to identify patterns of genetic structure of P. alabamensis across the study area. The correlated allele frequency model with admixture was used to examine all Pseudemys captured as a whole, P. alabamensis alone, and P. concinna alone. Pseudemys floridana was not run independently of the other species due to only a few individuals being found outside of the Weeks Bay system ( Table 1).
STRUCTURE analysis consisted of 10 independent runs for each K value (1-10) with a burn-in period of 100,000 followed by an additional 100,000 repetitions. In order to determine the best value of K (number of clusters) for each species, the ΔK statistic (Evanno et al., 2005) was calculated using STRUCTURE HARVESTER (Earl, 2012). STRUCTURE was also used to calculate the estimated membership coefficients Q for each individual in each cluster. Q indicates whether each individual belongs to one or, if admixed, to several clusters. Finally, a principal component analysis (PCA) was used to further assess the level of introgression based on microsatellite data using the software Genetix v4.05 (Belkhir et al., 2004).

| RE SULTS
In total, 296 Pseudemys turtles were captured from water bodies known to be inhabited by P. alabamensis (Table 1)

| Microsatellite
The eight microsatellite loci analyzed were polymorphic in all species and populations, with the exception of one locus (D87) in one popu-

| Hybridization
Of the 96 samples morphologically identified as P. alabamensis, two individuals were considered to be potential hybrids based on mixed morphological characteristics and the presence of reduced P. alabamensis identifying characteristics. One of these individuals from Dog  One allele from locus D28 that was a private allele in the Biloxi P.
concinna population was also found in the neighboring Pascagoula population of P. alabamensis (frequency of the allele in P. alabamensis in Pascagoula = 0.027).
Hybridization appears to occur at a higher rate between P. concinna and P. floridana. F ST between the two species was 0.065, much lower than between P. alabamensis and either of these two species and even within P. alabamensis. The P. floridana population with >5 individuals possessed 6 of the alleles that were private alleles within P. concinna populations and 1 allele that was considered a private allele within a P. alabamensis population.
When examining species assignment and admixture by STRUCTURE, we found that three individuals (3%) that were morphologically identified as P. alabamensis were assigned to P. concinna Another individual from Pascagoula morphologically identified as P. alabamensis showed admixture with mixed assignment between P. alabamensis and P. concinna (Q = 0.62). Signs of hybridization with P. alabamensis were also found in individuals morphologically identified as P. concinna. Of the 102 individuals morphologically identified as P. concinna, 12 individuals (12%) were assigned to P.
alabamensis with Q ≥ 0.7, and another 7 (23%) showed mixed assignments (0.5 < Q < 0.7) between the two species. Across all the populations, the Biloxi river was the locality where many individuals (8 out of 38 with Q ≥ 0.7) morphologically identified as P. concinna were assigned to P. alabamensis on the basis of microsatellite data.
We also found that two individuals out of 30 (7%) that were mor-

| DISCUSS ION
In this study, we assessed genetic diversity, population structure, and potential hybridization of the endangered P. alabamensis and co-occurring congeneric species. While previous studies have also addressed some of these questions (Hieb et al., 2014;Jackson et al., 2012), the sample sizes, distribution range of sampled populations, and/or genetic markers were limited. In our study, we used both mitochondrial and microsatellite markers to analyze P. alabamensis from seven rivers throughout the entire narrow range of this species. Using mitochondrial DNA, we found no genetic differentiation within or among populations of P. alabamensis due to a complete lack of mtDNA variation. Low levels of mitochondrial diversity are not uncommon in turtles that are of conservation concern (Rosenbaum et al., 2007;Vargas-Ramírez et al., 2007). However, different than what has been observed in other endangered species, only one haplotype was found across 96 individuals from the entire distribution range of P. alabamensis. A comparable lack of mitochondrial diversity to P. alabamensis has been noted in a related species, Pseudemys gorzugi (Bailey et al., 2008). Pseudemys gorzugi also inhabits a restricted range, although larger than P. alabamensis, being found only in the Rio Microsatellite data also indicate low genetic diversity for P. alabamensis with overall lower allelic diversity than the other two sympatric congenerics and lower than expected heterozygosity. Signs of inbreeding were observed in two populations: Fowl River and Biloxi River. Biloxi showed signs of inbreeding also for P. concinna, most likely the result of low population sizes for both species at this site (the estimated breeding population for P. alabamensis at Biloxi was in fact low; see also hybridization discussion below). Despite the overall low genetic diversity observed in P. alabamensis, microsatellite data support genetic differentiation between Mississippi and Alabama populations of this species, in agreement with slight morphological differences previously observed between these areas (Leary et al., 2003). Although this genetic structure and morphological differentiation may be the result of genetic drift, little gene flow between Mississippi and Alabama populations may occur due to the large distance between the mouth of the Pascagoula River Delta and the Alabama populations. Land and saltwater can hinder gene flow for freshwater species that are distributed in riverine systems across the Gulf of Mexico (Soltis et al., 2006), including the Pascagoula River (e.g., Dugo et al., 2004;Ennen et al., 2010). level of tolerance to brackish water (Agha et al., 2018). This is further supported by the presence of barnacles on the shells of some individuals in our study indicating exposure to higher salinity waters (N. Moreno, personal observation).
We observed admixture among the three species. Individuals of P. concinna and P. floridana in the region can be difficult to tell apart due to hybridization between the two (Mount, 1975;Spinks et al., 2013). We found haplotype sharing and mixed assignments between species based on microsatellite data, even for individuals which could be confidently assigned to a species based on morphological characteristics. Specifically, based on microsatellite data, around 50% of the individuals that were morphologically identified as P. concinna or P. floridana were assigned to the other species based on genetics.
Haplotype sharing was also seen to a lesser degree ( where we found instances of haplotype sharing among species, we sampled an excess of female versus male P. alabamensis ( Table 1).
Hybridization of P. alabamensis with congeneric species across its distribution range may overall be driven by decreased opportunities to find mates of the same species ("desperation hypothesis", Hubbs, 1955). This may be the result of a potentially skewed sex ratio in Overall, based on our results, P. alabamensis is experiencing significant admixture with congeneric co-occurring species across its entire restricted distribution range. Hybridization is a well-known phenomenon in species of conservation concern with limited population sizes (e.g., see Chattopadhyay et al., 2019 and references therein), and it presents a challenge for conservation management (Allendorf et al., 2005;Mallet, 2005;Wayne & Shaffer, 2016).
Although historically hybridization has generally been seen as a threat to endangered species, hybridization can also be a component of evolutionary processes and the origin of new species (Draper et al., 2021;Haig & Allendorf, 2006;Willis, 2020). In the US, the Endangered Species Act (ESA) provides guidelines that can be interpreted by the US Fish and Wildlife Service, depending, for example, on the origin and ecological role of the hybrid species and whether or not hybrids can be used for recovery of endangered parental species (Haig & Allendorf, 2006;Willis, 2020). Based on this, protection of hybrids could be possible, although it is generally discouraged.
Mostly, removal of hybrids is suggested when hybridization presents a threat to an endangered species (Draper et al., 2021;Haig & Allendorf, 2006), as in the case of P. alabamensis. Our study represents the first genetic study supporting the endangered status of P. alabamensis throughout its range and provides evidence that the Mississippi and Alabama populations of P. alabamensis are genetically distinct.
The overall low amount of genetic diversity observed at mitochondrial and nuclear (microsatellite) levels in P. alabamensis, the severely limited geographic range of this species, and the occurrence of hybridization throughout its distribution require the urgent development of targeted conservation actions. It has been shown that low genetic diversity and inbreeding in combination with an endemic restricted distribution may make species more susceptible to diseases and to the risk of genetic swamping due to hybridization (Georges et al., 2018). If hybrids could be identified with confidence based on morphological characteristics, targeted removal of hybrids could help avoid hybrid swamping of P. alabamensis. Although two individuals were morphologically identified as hybrids of P. alabamensis and confirmed as such by genetic data, not all genetically identified hybrids were easily identified by morphological characteristics. As morphological identification of hybrids and closely related species may be challenging and sometimes misleading (Chiari & Claude, 2012), further efforts should be made to develop methods that could be applied in the field to identify hybrids with confidence in order to remove them.
Our results also identify several P. alabamensis populations of higher conservation concern due to their low population sizes and consequent inbreeding and hybridization: Bayou La Batre, Biloxi, Weeks Bay, and Fowl River. Furthermore, considering the observed genetic distinction of populations from Alabama and Mississippi, specific management actions should be developed to preserve their uniqueness. This includes searching for additional unknown branches of these main riverine systems where the species could occur. Finally, climate change is predicted to strongly affect coastal areas and wetlands in the Gulf of Mexico (Anderson et al., 2014;Mulholland et al., 1997;Scavia et al., 2002), influencing the salinity of coastal watersheds and consequently their vegetation, and potentially changing the connectivity of existing watersheds due to sealevel rise. These factors can greatly influence the geographic range of species (Garroway et al., 2010). Thus, the imperiled status of P.
alabamensis could further worsen due to changes in habitat salinity, effects on its vegetation food sources, and potentially increased hybridization. It is therefore imperative that measures to prevent the progressive decline of populations and mitigate current and future effects of climate change on P. alabamensis are considered and developed rapidly. There are currently no management and conservation initiatives being carried out throughout the species range or even for some populations, so the first step to ensure the survival of this species should be population and habitat monitoring. Headstart programs for this species have been proposed (D. Nelson, personal communication), but never funded. Local monitoring activities to ensure habitat protection of the few sites where the species occurs, the potential development of a head-start program for genetically pure P. alabamensis individuals, education of local citizens on the consequences of translocating and moving turtles to different water bodies (including species such as Trachemys scripta that can potentially compete with P. alabamensis for resources), maintenance of nesting sites, assessment of recruitment throughout the species' range, and monitoring of population sizes should therefore be developed for this species.

ACK N OWLED G M ENTS
We are thankful to Skyler Kerr, Dominika Houserova, Robin Loyd, and Nathan Katlein for help with field collections of turtles. We acknowledge the Central Mississippi Turtle Rescue and Dauphin Island Aquarium for providing waif samples of P. alabamensis. We are also grateful to Danielle Edwards and Guillermo Velo-Anton for feedback on some of the analyses performed in this study. We are thankful to the AE and two anonymous reviewers for their helpful comments on a previous version of this paper.

CO N FLI C T O F I NTE R E S T
None of the authors have competing interests.