Contents lists available at ScienceDirect

Molecular Phylogenetics and Evolution

journal homepage: www.elsevier.com/locate/ympev

Genetic variation and admixture of red-eared sliders (Trachemys scriptaelegans) in the USAJames F. Parhama,b,⁎, Theodore J. Papenfussc, Anna B. Sellasb,d, Bryan L. Stuarte,W. Brian Simisonba Department of Geological Sciences, California State University, Fullerton, CA 92834, USAb Center for Comparative Genomics, California Academy of Sciences, 55 Music Concourse Drive, San Francisco, CA 94118, USAcMuseum of Vertebrate Zoology, 3101 Valley Life Sciences Building, University of California, Berkeley, CA 94720-3160, USAd Sana Biotechnology, 1 Tower Place, STE 500, South San Francisco, CA 94080, USAeNorth Carolina Museum of Natural Sciences, Raleigh, NC 27601, USA

A R T I C L E I N F O

Keywords:Red-eared sliderRESTSEHybridTurtleInvasiveGenetic pollutionTrachemys gaigeaeTrachemys scripta troostii

A B S T R A C T

The most ubiquitous, abundant, and invasive turtle on Earth, Trachemys scripta elegans (TSE, “red-eared slider”), isone of four taxa in a clade that is native to the USA and adjacent Mexico (three subspecies of Trachemys scripta plusTrachemys gaigeae). The present range-wide study of this clade is based on 173 known-locality mtDNA sequencescombined with ddRAD libraries for 43 samples emphasizing the western part of the range of TSE, its contact withthat of T. gaigeae, and anthropogenic hybrids between TSE and T. s. scripta. The data presented here are the first tosample the TSE × T. s. scripta intergrade zone or TSE × T. s. scripta crosses from introduced turtles. In the westernpart of its range (New Mexico and Texas), most samples of TSE from the Pecos River have mtDNA haplotypesmatching T. gaigeae. Structure analysis of SNPs from the ddRAD show evidence of genetic admixture between T.gaigeae and TSE in all included samples from the Rio Grande and Pecos River. These populations also exhibit T.gaigeae-like head stripes, i.e., a postorbital marking that does not reach the eye. The genetic and morphologicaldata are thereby reconciled, as both suggest that these TSE are intergrades. We recommend that these populationscontinue to be considered TSE, despite the admixture with T. gaigeae. In the Eastern United States, some samples ofthe morphologically intermediate subspecies T. s. troostii are not genetically distinct from TSE and some samplesshare morphological characters and genetic affinities with T. s. scripta. Based on these observations we concludethat the taxon T. s. troostii represents intergrades between TSE and T. s. scripta and should not be considered a validtaxon. Near the already established part of the intergrade zone between TSE and T. s. scripta, TSE mtDNA hap-lotypes have naturally introgressed into typical-looking samples of T. s. scripta in Georgia. Hybrids between in-troduced TSE and T. s. scripta are also confirmed deeper within the natural range of T. s. scripta in South Carolinaand Virginia. Given the examples of feral hybrids deep within its range shown here and elsewhere, the threat ofgenetic pollution of T. s. scripta by feral TSE is established.

1. Introduction

The most ubiquitous and abundant turtle on Earth, Trachemys scriptaelegans (TSE hereafter), is native to the USA (Fig. 1) where it is com-monly known as the red-eared slider. TSE are the most frequently keptpet turtle around the world and are farmed at industrial scales (Hughes,2000; Shi et al., 2008; Herrel and van der Meijden, 2014; Mali et al.,2014). Through escapes and intentional release, TSE has spread farbeyond its natural borders to every continent except Antarctica(Tracking Red-Eared Sliders, 2019; GISD, 2019). Wherever TSE areintroduced they impact ecosystems through their omnivorous diet, but

for native turtles the threat is potentially compounded by TSE as directcompetitors and vectors for disease (Somma et al., 2019). For all ofthese reasons, TSE is considered one of the 100 worst invasive species(GISD, 2019). The taxa most threatened by TSE are other taxa of Tra-chemys (some of which are endangered [Tortoise and Freshwater TurtleSpecialist Group, 1996; van Dijk and Flores-Villela, 2007]) throughgenetic pollution. Previous studies have demonstrated the ability forrelated testudinoid turtles to hybridize across lineages that have beenseparated for tens of millions of years (Parham et al., 2001; Buskirket al., 2005; Stuart and Parham, 2007). Not surprisingly, there aredocumented hybrids between introduced TSE and other taxa of

https://doi.org/10.1016/j.ympev.2019.106722Received 11 June 2019; Received in revised form 13 December 2019; Accepted 20 December 2019

⁎ Corresponding author.E-mail address: jparham@fullerton.edu (J.F. Parham).

Molecular Phylogenetics and Evolution 145 (2020) 106722

Available online 23 December 20191055-7903/ © 2020 Elsevier Inc. All rights reserved.

T

Trachemys (Forstner et al., 2014; Mitchell, 1994; Palmer and Braswell,1995; Parham et al., 2013; Seidel et al., 1999).

Trachemys is a clade of freshwater turtles that is endemic to Northand South America that includes 25 taxa (species and subspecies;Parham et al., 2015; TFTSG, 2017). The four taxa that occur in the USA(three subspecies of Trachemys scripta plus T. gaigeae) are considered amonophyletic group (Parham et al., 2015). Of these four, TSE had thelargest geographic range even before it became an invasive species,naturally occurring in drainages from the Rio Grande to Mississippi(Fig. 1). In the eastern part of its range, TSE has a broad range ofmorphological intergradation with Trachemys scripta scripta (yellow-bellied slider) in and around the Alabama River drainage (Carr, 1952;Mount, 1975). Pure T. s. scripta occur east of the Eastern ContinentalDivide, although introduced TSE can be found throughout its range(Somma et al., 2019) and hybridization has been reported based onobservations of intermediate specimens (Mitchell, 1994; Palmer andBraswell, 1995). In the western part of its range, TSE is parapatric withT. gaigeae (Big Bend slider), a vulnerable (van Dijk, 2011) species that isrestricted to the upper Rio Grande. Downstream from T. gaigeae, in thelower Rio Grande and Pecos River, TSE are morphologically distinctfrom other TSE and have occasionally been referred to as intergradeswith T. gaigeae (Hamilton, 1947; Degenhardt and Christiansen, 1974;Stuart, 2000). The intergrade hypothesis was refuted by a study usingallozymes (Seidel et al., 1999), but has never been followed up withother molecular methods. In addition to this unresolved contact zone,hybridization of T. gaigeae with introduced TSE was hypothesized basedon morphology and allozyme electrophoresis (Seidel et al., 1999) andlater confirmed with a combination of mitochondrial DNA and micro-satellites (Forstner et al., 2014). Aside from these studies, all othergenetic studies of TSE have been limited to their inclusion in broaderphylogenetic works (e.g., Jackson et al., 2008; Wiens et al., 2010; Fritzet al., 2011; Parham et al., 2013, 2015).

The present study is the first to include range-wide sampling of theclade that includes all four Trachemys in the USA. Understanding thegenetics of invasive species such as TSE is important for determiningsource populations as well as the routes and numbers of invasions(Holland, 2001; Handley et al., 2011). Both T. gaigeae and T. s. scriptahave purported examples of intergradation with native TSE and

hybridization with introduced TSE, and in some cases both types ofadmixture (natural and anthropogenic) are occurring in close proxi-mity. Our study tests the following questions: (1) What are the majorgenetic groupings within this widespread clade and how do they cor-relate with morphology-based taxa?; (2) Can more modern methodsreconcile the apparent conflict between morphological and genetic data(i.e., is there genetic evidence for genetic intergradation to match thewidespread intermediate morphologies)?; (3) What is the nature of thelineage boundaries between parapatric taxa that have different levels ofgenetic variation (e.g., between TSE and T. gaigeae vs. TSE and T. s.scripta)?; (4) Finally, given the purported threat of genetic pollutionfrom introduced TSE, is there any genetic evidence of admixturethroughout the range of Trachemys taxa in USA? We address thesequestions using a combination of genetic markers (mtDNA and ddRADSNPs).

2. Materials and methods

2.1. Sampling

Our study is based on 173 known-locality genetic samples withcorresponding museum voucher specimens (Appendix). Four of thesequenced samples are outgroups (Pseudemys gorzugi, Trachemys neb-ulosa, Trachemys venusta [n = 2]). Our ingroup sampling emphasizesthe western part of the range of TSE and its contact with that of T.gaigeae. We consider the Mexican species Trachemys hartwegi to be aseparate species from T. gaigeae following Parham et al. (2015). The169 samples of the ingroup include 17 T. gaigeae from seven differentlocalities in New Mexico and Texas, 22 TSE from 10 different localitiesin the Pecos River drainage, 16 TSE from nine different localities in theRio Grande, 62 other TSE from other, mostly western, parts of theirrange in USA (IL, 2; IN, 3; KS, 4; KY, 3; LA, 4; MS, 6; NM, 6; OK, 2; TX,28), four samples from introduced populations of TSE (Cayman Islands,3; FL, 1), 22 samples from the intergrade zone of TSE and T. s. scripta(AL, 22;), five suspected anthropogenic hybrids between introducedTSE and native T. s. scripta (SC, 1; VA, 4), 16 T. s. scripta (FL, 2; GA, 11;NC, 1; SC, 2), and 9 T. s. troostii (NC, 2; TN, 7). This study includessamples from the type locality of most valid Trachemys taxa native to

Fig. 1. Map showing major morphological and genetic groupings of Trachemys in the USA. Small squares refer to localities of samples in this study (see Appendix),with color shading of squares corresponding to mtDNA haplotype. Circles refer to samples in the ddRAD analyses. Admixed samples are represented by pie charts(larger circles) corresponding to the Structure analysis of SNPs from the ddRAD libraries. Samples that are mentioned in the text are numbered on the map.

J.F. Parham, et al. Molecular Phylogenetics and Evolution 145 (2020) 106722

2

the USA inasmuch as they are known (T. gaigeae, Boquillas [Rio GrandeVillage], Brewster Co., TX, samples 12–17; T. s. elegans: Fox River,White Co., IL, near New Harmony, IN, samples 113–114; T. s. scripta:Berkeley Co., SC, sample 158). For T. s. troostii, we do not have anysamples from the type locality (Cumberland River, TN) but do includeseven samples from adjacent Cumberland Plateau in TN (samples 69,103–107, 145). Taxonomic identification of ingroup samples was basedon morphological characteristics of the voucher specimens, i.e., colorand amount of head striping and plastral patterning. For all 173 sam-ples we sequenced a partial sequence of the control region to representthe mitochondrial genome. In order to characterize the genetic varia-tion at the contact of T. gaigeae and T. scripta, a subset of 43 samples ofTrachemys gaigeae and T. scripta were further analyzed using ddRAD(with an emphasis [n = 29] from the western part of the range).

2.2. Mitochondrial DNA sequencing and analyses

Genomic DNA was extracted from approximately 25 mg of livertissue using the Qiagen DNeasy tissue kit (Qiagen, Valencia, CA) fol-lowing the manufacturer's instructions. We ran standard PCR reactionsand sequenced a ~660 bp region of mtDNA control region from the 5′end using the primers DES-1 (5′-GCATTCATCTATTTTCCGTTAGCA-3′)and DES-2 (5′-GGATTTAGGGGTTTGACGAGAAT-3′) (Starkey et al.,2003). Sequencing was done on an ABI PRISM 3130xl Genetic Analyzer(Applied Biosystems). Contigs were quality trimmed and assembled inGeneious 11.1.4 (https://www.geneious.com), and aligned using theMAFFT 1.3.7 plugin (Katoh and Standley, 2013) for Geneious using the“Auto” settings. Uncorrected pairwise distances (p) were calculatedusing PAUP* 4.0a166 (Swofford, 2003). All mitochondrial sequenceswere deposited in GenBank (Appendix).

The phylogenetic analyses were performed using RAxML 8.2.10(Stamatakis, 2014). We used the “-f a” option which performs a rapidbootstrap analysis and searches for the best scoring maximum like-lihood tree. We used the AutoMRE option, which is an a posterioribootstopping analysis to determine when enough bootstrap analyseshave been performed to reach convergence using the majority-ruleconsensus tree criterion, which resulted in 650 bootstrap replicates.

2.3. ddRADlibrary prep

Samples were extracted using DNeasy Tissue kits (Qiagen) followingthe manufacturer’s protocol for animal tissues, and stored at −20 °C.We prepared ddRAD libraries for 43 individuals (T. gaigeae, n = 7; TSE,n = 29; invasive TSE × T. s. scripta hybrids, n = 4; T. s. scripta, n = 3)using the double digest protocol of Peterson et al. (2012). We followedthe Peterson et al. (2012) protocol closely with few modifications, in-cluding a 3-hour digest at 37 °C using the SphI and MluCl restrictionendonucleases (New England Biolabs; enzyme pair was chosen based onTable 1 of Peterson et al. [2012]). After cleaning using Dynabeads M-270 Streptavidin (Life Technologies), universal P2 adaptors were li-gated to each DNA fragment. Additionally, a uniquely barcoded P1adapter was ligated to each of sixteen individuals for three groups ofsixteen individuals. These groups were later pooled and a unique Illu-mina (Illumina) index was added to each group of sixteen and pooledtogether. Prior to pooling, DNA was cleaned using magnetic beads (asabove). Pooled DNA was size-selected using a Blue Pippin Prep (SageScience) with all fragments between 376 and 450 bp recovered. Theunique Illumina indices were incorporated onto the P2 adaptor end ofDNA fragments using a real-time library amplification kit (Kapa Bio-systems). The concentration of each pool was standardized and com-bined for HiSeq 2000 Illumina sequencing. Libraries were quantifiedusing a High Sensitivity DNA Kit on a 2100 Bioanalyzer (AgilentTechnologies) and sequenced at the Harvard Genomics Core facility onan Illumina HiSeq 2000 (100 bp paired-end reads). Sequencing runsresulted in ~200 million reads passing initial quality control at thesequencing facility.

2.4. ddRAD data analysis

Raw Illumina sequences were de-multiplexed using the Stacks v1.29(Catchen et al., 2011) ‘process_radtags’ script (resulting in 182 millionreads with the number varying between 43 thousand and 4.6 millionreads per individual. All ddRAD data sequences were deposited as anNCBI BioProject (Appendix). Loci were assembled using the Stacks v.2.3e (Catchen et al., 2013) denovo_map.pl pipeline and its componentprograms. To eliminate the possibility that mtDNA is driving the RADStructure pattern of asymmetrical introgression, mtDNA was filteredout of the raw illumina reads by mapping the raw reads to a Trachemysscriptamitochondrial genome (Genbank accession NC_011573.1) in CLCGenomic Workbench 7.0.3 (CLC Bio, Aarhus, Denmark) and runningthe remaining reads in the Stacks pipeline. Loci were generated by themerging of three or more similar “stacks.” Those “stacks” in which thenumber of reads were more than two standard deviations above themean were assumed to be repetitive elements and removed from theanalysis. Loci were identified by aligning the sequence reads from eachindividual to the assembled contigs resulting in 468,175 genotyped lociwith an effective per-sample coverage of 7.3x and a mean number ofsites per locus of 95.2. Using the “populations” component in stacks weset the minimum number of populations a locus must be present in totwo (-p 2) and the minimum percentage of individuals in a populationrequired to process a locus to 80% (-r 0.8). We used the ‘write_sin-gle_snp’ option to extract a single SNP from each locus. The resultingdataset consisted of 23,582 loci with 21,210 variant sites. From thisdataset, Structure 2.3.4 (Pritchard et al., 2000) input files were pro-duced using the structure option in “populations.” Using Structure, weevaluated varying numbers of clusters (K = 1–6) and six runs of100,000 generations for each K value with burn-in values of 10,000.The six Structure output files for each K value were all submitted toStructure Harvester (Earl and von Holdt, 2012), which uses the Evannomethod (Evanno et al., 2005) to estimate the K value that best fits thedata. Structure Harvester also outputs input files for CLUMPP(Jakobsson and Rosenberg, 2007), which helps correct for the organi-zational problem of ‘label switching’ (cluster values from each replicateare assigned different columns) and corrects for stochastic differencesof cluster solutions across replicates of the same K value. We usedDistruct 1.1 to create vector graphics of the Structure charts (Fig. 3B).We also used the populations script in Stacks to calculate FST, expectedand observed homozygosity, and mean nucleotide diversity (π).

We generated a maximum likelihood tree (Fig. 3A) from the ddRADdata using a phylip file generated by the Stacks2 ‘populations’ script.The ‘–phylip’ option produces a phylip formatted alignment file withnucleotides that are fixed-within, and variant among populations. Togenerate an alignment with all 43 individuals, we created a populationinput file with each individual assigned to its own population and ranthe ‘populations’ script with the ‘–phylip’ and ‘–write_single_snp’ op-tions. This produced an alignment with 43 OTUs and a length of 91,219positions. We used the same RAxML parameters used for the mtDNAcontrol region analysis, which resulted in the autoMRE stopping thebootstrap analysis at 50 replicates.

3. Results

3.1. Mitochondrial DNA results

Phylogenetic analyses of partial sequences of the control region(Fig. 2) identify three reciprocally monophyletic clades that correspondto the taxa T. gaigeae, TSE, and T. s. scripta. Uncorrected (p) distances ofcontrol region sequence within T. gaigae ranged from 0 to 0.16%, withinT. s. scriptaranged from 0 to 1.37%, and within TSE ranged from 0 to1.98%. TSE shows the highest mtDNA haplotype diversity in the wes-tern part of its range (KS, OK, NM, TX; Fig. 2). One mtDNA haplotype isbroadly distributed in the southeastern and midwestern USA (samples100–142) including most samples of T. s. troostii (samples 103–109).

J.F. Parham, et al. Molecular Phylogenetics and Evolution 145 (2020) 106722

3

The most notable discrepancy between mtDNA and accepted species-level taxonomy involves samples that are traditionally referred to TSEfrom the Pecos River in New Mexico that have mtDNA haplotypes

matching T. gaigeae. Of the 21 samples of TSE from the Pecos River, 17have T. gaigeaemtDNA haplotypes. Fifteen of the 17 TSE from the Pecosshare a single mtDNA haplotype. Four samples of TSE from the PecosRiver were found to have TSE mtDNA.

On the eastern side of the range of T. scripta, mtDNA haplotypes ofTSE occur throughout the natural intergrade zone with T. s. scriptasouth of the Appalachian Mountains. TSE mtDNA haplotypes penetrateeast into the range of typical looking T. s. scripta in Georgia. In contrastto the single T. gaigeae mtDNA haplotype recovered in TSE in the PecosRiver, there is higher variation among the TSE mtDNA haplotypes in-trogressed into T. s. scripta (n = 8, Fig. 2). Away from the intergradezone mtDNA haplotypes of TSE also appear deep within the range of T.s. scripta at areas where anthropogenic hybrids from feral populationsare identified based on the presence of intermediate-looking individuals(samples 142, 159,162–163). Finally, the taxon T. s. troostii cannot bedifferentiated from TSE with mtDNA.

3.2. ddRAD DNA results

The Delta K value for K= 2 (Fig. 3B) was 99.36 and all other DeltaK values were below 30. The two clusters shown in the Structure ana-lysis of SNPs from the ddRAD libraries correspond to T. gaigeae and T.scripta with FST estimated at 0.201. The subspecies of T. scripta are notdistinguished by the structure analysis of SNP data or by the any of thedivergence or fixation statistics, likely because of the low number ofpure T. s. scripta (n = 3) and T. s. troostii (n = 2) in the ddRAD analysis.Observed and expected heterozygosity for T. gaigaea (0.034 and 0.030)was lower than for T. scripta (0.123 and 0.217), with similar resultscalculated for mean nucleotide diversity, π, for T. gaigaea (0.037) and T.scripta, (0.240). The ddRAD data show evidence of genetic admixturebetween T. gaigeae and T. scripta in samples from the Rio Grande andPecos River (Fig. 3B).

Phylogenetic analyses of SNP data (Fig. 3A) place the admixed in-dividuals of T. gaigeae and TSE as a paraphyletic grade between the twospecies. Two samples of T. s. troostii do not form a clade and are nestedwithin TSE. The T. s. troostii sample from TN formed a clade withnearby midwestern TSE, whereas the T. s. troostii sample from NorthCarolina (sample 108) formed a clade with samples identified as T. s.scripta or T. s. scripta intergrades. The three samples of T. s. scripta forma clade exclusive of TSE, although like T. s. troostii, nested among TSE.Anthropogenic hybrids from South Carolina and Virginia with either T.s. scripta or TSE mtDNA haplotypes grouped with the pure T. s. scripta.

4. Discussion

Our study set out to ask the following questions: (1) What are themajor genetic groupings within this widespread clade and how do theycorrelate with morphology-based taxa?; (2) Can more modern methodsreconcile the apparent conflict between morphological and genetic data(i.e., is there genetic evidence for genetic intergradation to match thewidespread intermediate morphologies)?; (3) What is the nature of thelineage boundaries between parapatric taxa that have different levels ofgenetic variation (e.g., between TSE and T. gaigeae vs. TSE and T. s.scripta)?; (4) Finally, given the purported threat of genetic pollutionfrom introduced TSE, is there any genetic evidence of admixturethroughout the range of Trachemys taxa in USA? Using a combination ofgenetic markers (mtDNA and ddRAD SNPs) we answer the questionsabove by (1) Providing the first range-wide assessment of the USATrachemys clade revealing an uneven application of subspecific taxa; (2)Resolving the question of intergradation of T. gaigeae and TSE byshowing that populations in the lower Rio Grande and Pecos River areadmixed; (3) Providing the first genetic characterization of the inter-grade zone between TSE and T. s. scripta; and finally (4) Providing thefirst genetic confirmation of hybrids between introduced TSE and T. s.scripta.

The comparison of rapidly evolving, maternally-inherited, mtDNA

Fig. 2. Phylogenetic (ML) tree showing the three main mtDNA haplotype cladesof USA Trachemys (T. gaigeae, TSE, and T. s. scripta) with dashed lines.Outgroups are not shown. Scale bar represents substitutions per site. Eachsample includes a number (Appendix 1), a taxon assignment based on mor-phology (labelled with text, but also with white for T. gaigeae, red for TSE,yellow for T. s. scripta, and orange for morphological intergrades), and a locality(state, and country if not the USA) with drainage listed if it is Pecos or RioGrande. Circles are given for samples in the ddRAD analyses, and even largerpie charts are given for admixed samples with white corresponding to T. gaigeaeand red corresponding to T. scripta. Bootstrap values are shown for clades withbranch lengths ≥ 0.01. (For interpretation of the references to color in thisfigure legend, the reader is referred to the web version of this article.)

J.F. Parham, et al. Molecular Phylogenetics and Evolution 145 (2020) 106722

4

with more slowly evolving, biparentally-inherited, nuclear DNA, as wellas gross morphology and geographic distribution, provides a range-wide framework for understanding the evolutionary underpinnings ofTSE and other Trachemys native to the USA. Although it is not an em-phasis of this study, the range-wide assessment of TSE should be usefulfor determining source populations for this globally invasive species.

In the following sections we discuss our results by region, firstdiscussing the western part of the range, specifically the admixture ofTSE and T. gaigeae (4.1) and the taxonomic status of the intergrades(4.2). Next, we discuss our genetic results as they pertain to the re-lationship of TSE to the two eastern taxa such as the possibility of in-tergradation with and taxonomic status of T. s. troosti (4.3) and thenatural and anthropogenic admixture of TSE with T. s. scripta (4.4).

4.1. Admixture of TSE and T. gaigeae

The populations of TSE in the Rio Grande drainage downstreamfrom the range of T. gaigeae (including the Pecos River) exhibit T. gai-geae-like head stripes, i.e., a postorbital marking that does not reach theeye (Fig. 1; Stuart, 2000; Stuart and Ward, 2009). Based on allozymesand morphometrics, Seidel et al. (1999) argued against calling the TSEin the Rio Grande or Pecos River intergrades, highlighting the geneticand morphometric distinctiveness of T. gaigeae. Forstner et al. (2014)reported some hybridization from the easternmost edge of the range ofT. gaigeae in the Rio Grande, but the scope of their project did not ex-tend to the lower Rio Grande or Pecos River.

The data presented here show that TSE from the Rio Grande andPecos River are admixed with T. gaigeae. Seventeen of 21 Pecos Riversamples assigned to TSE based on morphology show T. gaigeae mtDNAhaplotypes (Fig. 2). A subsample of these Pecos River samples (n = 8)

Fig. 3. ddRAD analyses of USA Trachemys. A. Phylogenetic (ML) tree based on the SNP data. Bootstrap values for all clades are 100 unless otherwise indicated. Scalebar represents substitutions per site. Dashed lines show the different boundaries of T. gaigeae, T. gaigeae mtDNA, and T. gaigeae SNPs. Each sample includes a number(Appendix 1), a taxon assignment based on morphology (labelled with text, but also with white for T. gaigeae, red for TSE, yellow for T. s. scripta, and orange formorphological intergrades), a locality (state, and Country if not the USA) with drainage listed if it is Pecos or Rio Grande, circles for samples in the ddRAD analyses,and even larger pie charts for admixed samples. B. Structure analysis of SNPs from the ddRAD libraries, (K = 2, T. gaigeae in white, T. scripta in red) showing evidenceof genetic admixture in the Rio Grande and Pecos River. (For interpretation of the references to color in this figure legend, the reader is referred to the web version ofthis article.)

J.F. Parham, et al. Molecular Phylogenetics and Evolution 145 (2020) 106722

5

were further analyzed by ddRAD data, revealing a mixture of T. gaigeaeand TSE SNPs. The higher frequency of T. gaigeae mtDNA in these PecosRiver samples is similar to that noted for hybrids between TSE and T.gaigeae by Forstner et al. (2014) in the Rio Grande, who suggested thatdifferential receptiveness to the male courtship behavior of the otherspecies could explain this asymmetrical introgression. The ddRADanalyses also included four samples from the Rio Grande that showed T.gaigeae SNPS, although in lower frequencies than in the Pecos River. Incontrast to the Pecos River samples, all samples of TSE from the RioGrande have TSE mtDNA haplotypes. A putative hybrid sample (in-termediate morphology; sample 42) from within the range of T. gaigeaealso differs from the previously reported hybrids (Forstner et al., 2014)by having TSE mtDNA haplotype. Sample 42 could be a translocatedsample from the lower Rio Grande, where admixed samples with TSEmtDNA occur. In any case, sample 42 and the intergrades in the lowerRio Grande show that hybridization does not always involve T. gaigeaefemales.

The TSE from both the Rio Grande and Pecos River show evidenceof admixture and an intermediate morphology and so should be con-sidered intergrades. All sampled individuals show predominantly TSESNPs. In the Rio Grande, the frequency of T. gaigeae SNPs generallydeclines as samples get further from the range of T. gaigeae (Fig. 1).Similarly, in the Pecos River, the sample (80) with the least amount ofT. gaigeae SNPs is furthest from the current range of T. gaigeae. Thefrequency of T. gaigeae SNPs is greater and more consistently dis-tributed in the Pecos River than in the lower Rio Grande. The persis-tence of T. gaigeae SNPs in TSE from the Pecos River compared to theLower Rio Grande might reflect the fact that the Pecos River is moreisolated from other populations of TSE.

The admixture of TSE and T. gaigeae suggest an ancient contactbetween the two species. Until the Pleistocene (<2.6 Ma), the upperRio Grande was part of a system of internally draining (endorheic)basins terminating in what is now the Chihuahuan Desert (Gallowayet al., 2011). This ancient endorheic basin could explain the origin of T.gaigeae as a distinct lineage. Subsequently, the Rio Grande evolved intoits current configuration and began draining to coastal lowlands con-nected to the Mississippi River drainage. This Pleistocene drainageevolution facilitated the contact between T. gaigeae and TSE (Legler andVogt, 2014). A Pleistocene contact is recent enough to explain the lackof differentiation between the mtDNA haplotypes of TSE in the PecosRiver and T. gaigeae, and that timing is consistent with that hypothe-sized other freshwater vertebrates in the region (Echelle and Echelle,1978; Smith and Miller, 1986).

Unfortunately, human activities complicate our ability to fully re-construct the evolutionary history of Trachemys in this region. For ex-ample, human introductions make it difficult to discern between hy-brids and translocated intergrades (see above). Furthermore, theconstruction of dams along the Pecos and Rio Grande has altered thedistribution of freshwater vertebrates by changing the flow regime. Forexample, the spread of invasive fishes in dammed drainages in westernUSA is well documented, resulting in hybrid swarms (Moyle and Light,1996). In some cases, the replacement can be extremely rapid; the RioGrande silvery minnow (Hybognathus amarus) was replaced in the Pecosin less than 10 years by the plains minnow (Hybognathus placitus), anintroduced congener from the Mississippi drainage (Hoagstrom et al.,2010). Trachemys gaigeae is adapted to heterogeneous desert rivers, butdrainage impoundments create more lentic habitats favored by TSE(Stuart and Ward, 2009; Forstner et al., 2014). Because the range of T.gaigeae does not currently contact the Pecos River, the hypothesis ofancient contact is the favored explanation for the admixed individualsin the Pecos River and lower Rio Grande.

4.2. Taxonomic status of T. gaigeae × T. scripta intergrades

Previous studies cited the lack of an intergrade zone between TSEand T. gaigeae to argue for the species status of T. gaigeae (Seidel et al.,

1999; Stuart and Ward, 2009). Therefore, the recognition of an inter-grade zone here raises questions about the species status of T. gaigeae.Some other authors employ a strict Biological Species Concept and soconsider all Trachemys to be one species with all named taxa as sub-species T. scripta (e.g., Legler and Vogt, 2014). Other authors recognizethe distinctiveness of other Trachemys species despite intergrade zones(e.g., Parham et al., 2013; TFTSG, 2017). The taxonomic history ofTrachemys was recently reviewed by Seidel and Ernst (2016), whoidentified the variable use of species vs. subspecies as the major sourceof taxonomic instability in the genus. The most widely-used nomen-clature for Trachemys (Seidel, 2002; Parham et al., 2013, 2015; Seideland Ernst, 2016; TFTSG, 2017) employs a Phylogenetic Species Conceptthat applies the species rank for most morphologically and geneticallydistinct lineages, whereas subspecies are unevenly applied to some taxathat show intergradation (e.g., T. scripta sspp.), but not others (seeParham et al., 2013), or have unclear boundaries (e.g., T. decussatasspp.), or else are not readily diagnosed (e.g., T. stejnegeri sspp.).

This ad hoc system is imperfect, but since all current taxonomicarrangements of Trachemys employ subspecies, any proposed taxo-nomic scheme that eliminated this rank would necessitate many no-menclatural changes and so should not be undertaken lightly. Becausethe subspecies rank is problematic, as well as the sake of current sta-bility, we recommend that T. gaigeae continue to be recognized as adistinct species. Furthermore, at the current time, the most significantgenetic break between T. gaigeae and T. scripta still coincides with thecurrent species borders (Figs. 2, 3) and they are otherwise morpholo-gically, genetically, and ecologically distinct (Seidel et al., 1999; Stuartand Ward, 2009). We also recommend that T. scripta in the lower RioGrande and Pecos River continue to be considered T. scripta, despite theadmixture with T. gaigeae. This is because they are morphologically andgenetically most similar to T. scripta as well as the prediction that moredata will likely confirm a clinal decrease in T. gaigeae influence in thelower Rio Grande making any other taxonomic boundary somewhatarbitrary. Finally, we propose that future studies should include moredetailed analyses of Trachemys contact zones and that the uneven use ofsubspecies needs to be further evaluated.

4.3. Taxonomic status of T. s. troostii

Trachemys scripta troostii is native to the Cumberland and Tennesseerivers (Carr, 1937, 1952) and intergrades with TSE in Kentucky (Ernstand Jett, 1969). Unlike the other three taxa of Trachemys studied here,the nine samples of T. s. troostii did not form distinct clades in either thephylogenetic analyses of mtDNA or nuclear SNP data. Three differentmtDNA haplotypes were recovered, with seven of the samples sharingthe broadly distributed TSE mtDNA haplotype (see 3.1). In addition tonot being genetically distinct in the markers used here, of the four taxaof Trachemys native to the USA, T. s. troostii is the least morphologicallydistinct. In fact, Carr (1952) stated that from a morphological per-spective one could consider that “troostii is merely a population of in-tergrades, in that its characters are almost precisely intermediate be-tween those of scripta and elegans.”

In support of the hypothesis that T. s. troostii are intergrades, pre-viously undescribed samples (108–109) from a population of Trachemysin North Carolina referred to T. s. troostii by Beane et al. (2008) matchmost characters of T. s. troostii except that some individuals also showvertical postorbital blotches (Fig. 4; similar to that seen in T. s. scripta,Fig. 1). Despite having mtDNA matching TSE, in the phylogeneticanalysis of SNP data, a sample (108) of this North Carolina populationformed a clade with samples identified as T. s. scripta or T. s. scriptaintergrades. Because of the remote location of these samples, we submitthat the intermediate morphological and genetic characteristics arosenaturally (i.e., not the result of introduced T. s. scripta). The retention ofT. s. scripta characters in the NC samples likely reflect that they are far-removed from the range of TSE, similar to the pattern seen for thegenetic influence of TSE samples in the upper reaches of the Pecos River

J.F. Parham, et al. Molecular Phylogenetics and Evolution 145 (2020) 106722

6

and in the lower Rio Grande (see 4.2).The overall intermediate morphology of T. s. troostii and the ap-

parent lack of any genetic distinctiveness of T. s. troostii both argue thatthis taxon is an intergrade of TSE and T. s. scripta. Because of this, werecommend that T. s. troostii should not be considered a valid taxon.The range of “T. s. troostii” most likely represents a northern extensionof a well-established intergrade zone to the south (see 4.4, Fig. 1) albeitwith more genetic influence from TSE in the midwest. Alternatively, theintergrades of the western Appalachians may also indicate past contactof TSE and T. s. scripta across the Eastern Continetal Divide, whichwould also explain the persistence of T. s. scripta features in the NorthCarolina population of “T. s. troostii.”

4.4. Admixture of TSE and T. s. scripta

In the eastern USA, intergradation between TSE and T. s. scripta hasbeen long-recognized based on specimens with intermediate morphol-ogies from Alabama (Carr, 1952; Conant, 1958; Mount, 1975; Conantand Collins, 1998). The contact between TSE and T. s. scripta is part of asuture zone shared by other southeastern USA taxa that is generallyhypothesized to result from postglacial range expansions (Swensen andHoward, 2005). The data presented here provide the first genetic as-sessment of this natural intergrade zone for Trachemys scripta. All 22 ofour samples from within the range of intergradation had TSE mtDNAhaplotypes (Figs. 1, 2). TSE mtDNA haplotypes further penetrate east,extending the genetic signature of the intergradation zone well into therange of typical-looking T. s. scripta. In some sites, typical-looking T. s.scripta with either TSE mtDNA and T. s. scripta mtDNA can be found atthe same site (central Georgia, Fig. 1). Unfortunately, our ddRADanalysis (k = 2) could not distinguish between TSE and T. s. scripta,probably because of the low number of included T. s. scripta (n = 3;Fig. 2). Additional genomic sampling of T. s. scripta may help elucidatethe genetic structure and age of this contact zone.

Away from this intergrade zone, intermediate morphologies as wellas the presence of TSE mtDNA haplotypes identify examples of hybridswith feral TSE (Fig. 1). A sample (141) with intermediate morphologyfrom South Carolina, well within the range of typical-looking T. s.scripta, has a TSE mtDNA haplotype. The northern part of the range ofT. s. scripta intergrades into invasive TSE in northern Virginia andMaryland (Mitchell, 1994; Conant and Collins, 1998; Somma et al.,2019). Four samples (142, 159, 162–163) with intermediate mor-phology from a reported “hybrid swarm” in Virginia (Mitchell, 1994)show a mixture of TSE (n = 1) and T. s. scripta (n = 3) mtDNA

haplotypes. This is the first confirmation of genetic pollution of T. s.scripta with feral TSE. Given the examples of feral hybrids deep withinits range shown here and elsewhere, the threat of genetic pollution of T.s. scripta by feral TSE is established.

CRediT authorship contribution statement

James F. Parham: Conceptualization, Methodology, Data curation,Writing - original draft, Visualization, Supervision, Project adminis-tration, Funding acquisition. Theodore J. Papenfuss:Conceptualization, Resources, Writing - review & editing. Anna B.Sellas: Investigation, Data curation. Bryan L. Stuart: Resources,Writing - review & editing. W. Brian Simison: Conceptualization,Methodology, Formal analysis, Resources, Data curation, Writing - re-view & editing, Funding acquisition.

Acknowledgments

This project started many years ago in collaboration with Andrew H.Price (Texas Parks and Wildlife Department) and is dedicated to him.Many people and agencies provided assistance with collections. Thelate Jim Lee (Hokes Bluff) is thanked for his generosity and assistancewith fieldwork in Alabama and the late Charlie Painter is thanked forcollecting some of the key samples in New Mexico. Kory Heiken assistedwith fieldwork in Texas. Jeffrey C. Beane, L. Todd Pusser, Robert A.Davis, C. A. d’Orgeix, and E. N. Sismour collected samples deposited atNCSM. Jeffrey C. Beane is also thank for his photograph used in Fig. 4.For permits we thank Alabama Division of Wildlife and FreshwaterFisheries, Cayman Islands Department of Environment, Florida Fish andWildlife Conservation Commission, Georgia Department of NaturalResources, Illinois Department of Natural Resources, Indiana Depart-ment of Natural Resources, Louisiana Department of Wildlife andFisheries, Kansas Department of Wildlife, Parks, and Tourism, KentuckyDepartment of Fish and Wildlife, Mississippi Museum of Natural Sci-ence, New Mexico Department of Game and Fish, North CarolinaWildlife Resources Commission, Oklahoma Department of WildlifeConservation, South Carolina Department of Natural Resources, Ten-nessee Wildlife Resources Agency, and the Texas Parks and WildlifeDepartment. We also thank the following collections managers in al-phabetical order according to the institutional abbreviations given inthe appendix: CAS, Lauren Scheinberg and Jens Vindum, Lauren is alsothanked for help with the map figure; MPM RA, Julia Colby (Alan Re-setar of the Field Museum of Natural History, Chicago, IL, also helpedwith this specimen); MSB: Tom Giermakowski; MVZ: Carol Spencer;NCSM, Jeffrey C. Beane; TCWC, Toby Hibbits; TNHC, Travis J. LaDuc.Funding for this project was provided by a grant from the CaliforniaState University Program for Education and Research in Biotechnology(CSUPERB; Parham, Simison) as well as a Hagey Award and a ResearchExcellence Grant from the California Academy of Sciences, CaliforniaAcademy of Sciences (Simison and Parham).

Appendix A

Vouchers for genetic sequences (Mitochondrial DNA sequences,GenBank accessions MN561855-MN562027; ddRAD data, BioProjectPRJNA576258) of emydid turtles of the genera Trachemys andPseudemys used in this study. Museum abbreviations: CAS, CaliforniaAcademy of Sciences Herpetology Collection, San Francisco, California,USA; MPM RA, Milwaukee Public Museum, Milwaukee, WI, USA; MSB,Museum of Southwestern Biology, University of New Mexico,Albuquerque, NM, USA; MVZ, Museum of Vertebrate Zoology,University of California, Berkeley, USA; NCSM, North Carolina Museumof Natural Sciences, Raleigh, NC, USA; TCWC, Texas CooperativeWildlife Collection, Texas A&M University, College Station, TX, USA;TNHC, Texas Natural History Collections, University of Texas, Austin,TX, USA. Vouchers are whole-body preserved specimens. 1. T. gaigeae

Fig. 4. Sample referred to T. s. troostii with mtDNA matching TSE showing headmarking similar to T. s. scripta (Sample 108, NCSM 74447, Madison Co., NC).Phylogenetic analysis of SNP data placed this sample with those identified as T.s. scripta or T. s. scripta intergrades (Photo by J. Beane).

J.F. Parham, et al. Molecular Phylogenetics and Evolution 145 (2020) 106722

7

(Brewster Co., TX; MVZ 250723); 2. T. gaigeae (Presidio Co., TX; MVZ265710); 3–7. T. gaigeae (Socorro Co., NM; MSB 50519-23); 8–11. T.gaigeae (Sierra Co., NM; MVZ 250711, 265664-6); 12. T. gaigeae(Presidio Co., TX; MVZ 265711); 13–17. T. gaigeae (Brewster Co., TX;TCWC 68545-7, TNHC 50874-5); 18–25. T. s. elegans (Eddy Co., NM;250715-6, 265724, 265726-9, 265731); 26–30. T. s. elegans (Eddy Co.,NM; MVZ 265676, 265678-80, 265725); 31–32. T. s. elegans (De BacaCo., NM; MVZ 265736, 265733); 33–34. T. s. elegans (Reeves Co., TX;MVZ 265683, 265685); 35–36. T. s. elegans (Rusk Co., TX; TNHC 50876;TCWC 68555); 37. T. s. elegans (Uvalde Co., TX; MVZ 265719); 38. T. s.elegans (Travis Co., TX; TCWC 68637); 39. T. s. elegans (Calhoun Co.,TNHC 50361); 40. T. s. elegans (Travis Co., TNHC 50865); 41. T. s.elegans (Bexar Co., TX; TNHC 50872); 42–45. T. s. elegans (Brewster Co.,TX; MVZ 250717-9, TCWC 68571); 46. T. s. elegans (Uvalde Co., TX;MVZ 265718); 47. T. s. elegans (Bexar Co., TX; TCWC 68549); 48–49. T.s. elegans (Hidalgo Co., TX; TCWC 68629, TNHC 50864); 50–54. T. s.elegans (Quay Co., NM; MVZ 265669-70, 265673-74, 265737); 55. T. s.elegans (Hemphill Co., TX; TCWC 68567); 56. T. s. elegans (Kenedy Co.,TX; TCWC 68567); 57. T. s. elegans (Hemphill Co., TX; TCWC 68631);58. T. s. elegans (Brooks Co., TX; TCWC 65331); 59. T. s. elegans(Blanco/Gillespie Co., TX; TCWC 66060); 60. T. s. elegans (Lynn Co., TX;TCWC 68569); 61. T. s. elegans (Dawson Co., TX; TCWC 68553); 62. T. s.elegans (Shackleford Co., TX; TCWC 65334); 63–64. T. s. elegans(Runnels Co., TX; MVZ 250689-90); 65. T. s. elegans (Eddy Co., NM;MVZ 265732); 66–67. T. s. elegans (Menard Co., TX; MVZ 250677,250692); 68. T. s. elegans (Hardeman Co., TX; TCWC 68564); 69. T. s.troostii (Sullivan Co., TN; MVZ 265789); 70. T. s. elegans (Zapata Co.,TX; MVZ 265715); 71. T. s. elegans (Webb Co., TX; MVZ 265712);72–74. T. s. elegans (Val Verde Co., TX; MVZ 265723, TCWC 68550,68561); 75. T. s. elegans (Webb Co., TX; 265713); 76–77. T. s. elegans(Kinney Co., TX; MVZ 265720-1); 78. T. s. elegans (Maverick Co., TX;MVZ 265722); 79. T. s. elegans (Zavala Co., TX; MVZ 265717); 80. T. s.elegans (Guadalupe Co., NM; MVZ 265675); 81. T. s. elegans (HarrisonCo., TX; TCWC 68554); 82. T. s. elegans (Eddy Co., NM; MVZ 265730);83. T. s. elegans (Cayman Islands; MVZ 265799); 84. T. s. elegans (QuayCo., NM; MVZ 265734); 85. T. s. elegans (Kearney Co. KS; MVZ265743); 86–87. T. s. elegans (Seward Co., MVZ 265745-6); 88–89. T. s.elegans (Texas Co., OK; MVZ 265738, 265740); 90–91. T. s. ele-gans× scripta intergrades (Bibb Co., AL; CAS 255161, 255163); 92. T. s.elegans (Kearney Co., KS; MVZ 265744); 93. T. s. elegans (Eddy Co., NM;MVZ 265677); 94. T. s. elegans (Burleson Co. TX; TCWC 68628); 95–96.T. s. elegans (Calhoun Co., TX; TCWC 68565, 68572); 97. T. s. elegans(Burleson Co., TX; TCWC 68634); 98. T. s. elegans (Puerto Rico; MVZ265691); 99. T. s. elegans × scripta intergrade (Monroe Co., AL; CAS255156); 100–102. T. s. elegans (Madison Co., KY; MVZ 265786-8); 103.T. s. troostii (Loudon Co., TN; MVZ 265785); 104–105. T. s. troostii(Grainger Co., TN; MVZ 265792-3); 106–107. T. s. troostii (JeffersonCo., TN; MVZ 265795, 265797); 108–109. T. s. troostii (Madison Co.,NC; NCSM 74447-8); 110. T. s. elegans (Sierra Co., NM; MVZ 265667);111–112. T. s. elegans (Caymans Islands; MVZ 265798, 265800);113–114. T. s. elegans (White Co., IL; CAS 252980-1); 115–116. T. s.elegans (Posey Co., IN; CAS 252978-9); 117. T. s. elegans (White Co., IL;MPM RA 34011); 118–119. T. s. elegans (St. Tammany Par., LA; MVZ250674, 238121). 120–125. T. s. elegans (Lowndes Co., MS; CAS255165-70); 126–129. T. s. elegans × scripta intergrades (CovingtonCo., AL; CAS 255141-4); 130–134. T. s. elegans × scripta intergrades(Etowah Co., AL: CAS 255145-9); 135–136. T. s. elegans × scripta in-tergrades (Clark Co., AL; CAS 255151, 255157); 137. T. s. ele-gans × scripta intergrade (Monroe Co., AL; CAS 255158); 138. T. s.elegans × scripta intergrade (Tuscaloosa Co., AL; CAS 255160);139–140. T. s. elegans× scripta intergrades (Bibb Co., AL CAS 255162,255164); 141. T. s. elegans × scripta hybrid (Charleston Co., SC; MVZ250679); 142. T. s. elegans × scripta hybrid (King and Queen Co., VA;NCSM 74673); 143–144. T. s. elegans (Allen Par., LA; MVZ 250675-6);145. T. s. troostii (Sullivan Co., TN; MVZ 265791); 146. T. s. elegans(Leon Co., FL; MVZ 241524); 147. T. s. elegans × scripta intergrade

(Monroe Co., AL; CAS 255154); 148. T. s. elegans × scripta intergrade(Tuscaloosa Co., AL; CAS 255159); 149. T. s. scripta (Lowndes Co., GA;CAS 255138); 150. T. s. elegans× scripta intergrade (Clark Co., AL; CAS255152); 151. T. s. elegans × scripta intergrade (Monroe Co., AL; CAS255155); 152. T. s. scripta (Bleckley Co., GA; CAS 255129); 153. T. s.scripta (Jefferson Davis Co., GA; CAS 255128); 154. T. s. scripta(Bleckley Co., GA; CAS 255130); 155–156. T. s. scripta (Macon Co., GA;CAS 255133-4); 157. T. s. scripta (Charleston Co., SC; MVZ 250680);158. T. s. scripta (Berkeley Co., SC; MVZ 250681); 159. T. s. ele-gans × scripta hybrid (King and Queen Co., VA; NCSM 74672); 160. T.s. scripta (McDuffie Co., GA; CAS 255124); 161. T. s. scripta (Dare Co.,NC; MVZ 250678); 162–163. T. s. elegans × scripta hybrids (King andQueen Co., VA; NCSM 74674-5); 164. T. s. scripta (Evans Co., GA; CAS255125); 165. T. s. scripta (Leon Co., FL; MVZ 241523); 166. T. s. scripta(Jefferson Davis Co., GA; CAS 255127); 167–168. T. s. scripta (LowndesCo., GA; CAS 255137, 255139); 169. T. s. scripta (Baker Co., FL; MVZ241522). 170. Pseudemys concinna (Jefferson Davis Co., GA; CAS255126); 171. T. venusta (Costa Rica; MVZ 233245); 172. T. nebulosa(Mexico; MVZ 137443); 173. T. venusta (Guatemala; MVZ 264176).

Appendix B. Supplementary material

Supplementary data to this article can be found online at https://doi.org/10.1016/j.ympev.2019.106722.

Referen ces

Beane, J.C., Davis, R.A., Pusser, L.T., 2008. Distribution record: Trachemys scripta troostii.Herp. Rev. 39, 482.

Buskirk, J.R., Parham, J.F., Feldman, C.R., 2005. On the hybridisation of two distantlyrelated Asian turtles (Sacalia x Mauremys). Salamandra 41, 21–26.

Carr Jr., A.F., 1937. The status of Pseudemys scripta and Pseudemys troostii. Hepetologica 1,75–77.

Carr, A., 1952. Handbook of Turtles. Comstock, New York.Catchen, J., Amores, A., Hohenlohe, P., Cresko, W., Postlethwait, J., 2011. Stacks:

building and genotyping loci de novo from short-read sequences. G3: Genes.Genomes, Genetics 1, 171–182.

Catchen, J., Hohenlohe, P., Bassham, S., Amores, A., Cresko, W., 2013. Stacks: an analysistool set for population genomics. Molec. Ecol. 22, 3124–3140.

Conant, R., 1958. A Field Guide to Reptiles and Amphibians of Central and Eastern NorthAmerica. Houghton Mifflin Company, Boston.

Conant, R., Collins, J.T., 1998. A Field Guide to Reptiles and Amphibians: Eastern andCentral North America, 3rd ed. Houghton Mifflin, Boston.

Degenhardt, W.G., Christiansen, J.L., 1974. Distribution and habitats of turtles in NewMexico. Southw. Nat. 19, 21–46.

Earl, D.A., von Holdt, B.M., 2012. STRUCTURE HARVESTER: a website and program forvisualizing STRUCTURE output and implementing the Evanno method. Cons. Genet.Res. 4, 359.

Echelle, A.A., Echelle, A.F., 1978. The Pecos River pupfish, Cyprinodon pecosensis n. sp.(Cyprinodontidae), with comments on its evolutionary origin. Am. Soc. Ichthyol.Herpetol. 4, 569–582.

Ernst, C.H., Jett, B.G., 1969. An Intergrade Population of Pseudemys scripta elegans xPseudemys scripta troosti in Kentucky. J. Herp 3, 103.

Evanno, G., Regnaut, S., Goudet, J., 2005. Detecting the number of clusters of individualsusing the software STRUCTURE: a simulation study. Mol. Ecol. 14, 2611–2620.

Forstner, M.R.J., Dixon, J.R., Guerra, T.M., Winters, J.M., Stuart, J.N., Davis, S.K., 2014.Status of U.S. populations of the Big Bend slider (Trachemys gaigeae). In: Hoyt, C.A.,Karges, J. (Eds.), Proceedings of the Sixth Symposium on the Natural Resources of theChihuahuan Desert Region. October 14-17. Chihuahuan Desert Research Institute,Fort Davis, pp. 335–367.

Fritz, U., Stuckas, H., Vargas-Ramírez, M., Hundsdörfer, A.K., Maran, J., Päckert, M.,2011. Molecular phylogeny of Central and South American slider turtles: implicationsfor biogeography and systematics (Testudines: Emydidae: Trachemys). J. Zool. Syst.Evol. Res. 50, 125–136.

Galloway, W.E., Whiteaker, T.L., Ganey-Curry, P., 2011. History of Cenozoic NorthAmerican drainage basin evolution, sediment yield, and accumulation in the Gulf ofMexico basin. Geosph. 7, 938–973.

GISD (Global Invasive Species Database), 2019. Species profile: Trachemys scripta elegans.<http://www.iucngisd.org/gisd/species.php?sc=71>.

Hamilton, R.D., 1947. The range of Pseudemys scripta gaigeae. Copeia 1947, 65–66.Handley, R.-J., Estoup, A., Evans, D.M., Thomas, C.E., Lombaert, E., Facon, B., Aebi, A.,

Roy, H.E., 2011. Ecological genetics of invasive alien species. Biocontrol 56,409–428.

Herrel, A., van der Meijden, A., 2014. An analysis of the live reptile and amphibian tradein the USA compared to the global trade in endangered species. Herpetol. J. 24,2013–2110.

Hoagstrom, C.W., Zymonas, N.D., Davenport, S.R., Propst, D.L., Brooks, J.E., 2010. Rapid

J.F. Parham, et al. Molecular Phylogenetics and Evolution 145 (2020) 106722

8

species replacements between fishes of the North American plains: a case historyfrom the Pecos River. Aquat. Invas. 5, 141–153.

Holland, E.L., 2001. Evolutionary genetics of invasive species. TREE 17 (8), 386–391.Hughes, D.W., 2000. The contribution of the pet turtle industry to the Louisiana economy.

Louisiana Rural Economist 62, 4–5.Jackson, J.T., Starkey, D.E., Guthrie, R.W., Forstner, M.R.J., 2008. A mitochondrial DNA

phylogeny of extant species of the genus Trachemys with resulting taxonomic im-plications. Chel. Cons. Biol. 7, 131–135.

Jakobsson, M., Rosenberg, N.A., 2007. CLUMPP: a cluster matching and permutationprogram for dealing with label switching and multimodality in analysis of populationstructure. Bioinf. 23, 1801–1806.

Katoh, K., Standley, D.M., 2013. MAFFT multiple sequence alignment software version 7:improvements in performance and usability. Mol. Biol. Evol. 30, 772–780.

Legler, J., Vogt, R.C., 2014. The Turtles of Mexico. University of California Press,Berkeley.

Mali, I., Vandewege, M.W., Davis, S.K., Forstner, M.R.J., 2014. Magnitude of the fresh-water turtle exports from the US: long term trends and early effects of newly im-plemented harvest management regimes. PLoS One 9, e86478.

Mitchell, J., 1994. The Reptiles of Virginia. Smithsonian Institution Press,Washington D.C.

Moyle, P.B., Light, T., 1996. Biological invasions of fresh water: Empirical rules and as-sembly theory. Biol. Cons. 78, 149–161.

Mount, R.H., 1975. The Reptiles and Amphibians. Auburn University, Auburn.Palmer, W.M., Braswell, A.B., 1995. Reptiles of North Carolina. The University of North

Carolina Press, Chapel Hill.Parham, J.F., Papenfuss, T.J., Buskirk, J.R., Parra-Olea, G., Chen, J.Y., Simison, W.B.,

2015. Trachemys ornata or not ornata: reassessment of a taxonomic revision forMexican Trachemys. Proc. Calif. Acad. Sci. 62, 359–367.

Parham, J.F., Papenfuss, T.J., van Dijk, P.P., Wilson, B.S., Marte, C., Rodriguez Schettino,L., Simison, W.B., 2013. Genetic introgression and hybridization in Antillean fresh-water turtles (Trachemys) revealed by coalescent analyses of mitochondrial andcloned nuclear markers. Mol. Phyl. Evol. 67, 176–187.

Parham, J.F., Simison, W.B., Kozak, K.H., Feldman, C.R., Shi, H., 2001. New Chineseturtles: Endangered or invalid? A reassessment of two species using mitochondrialDNA, allozyme electrophoresis, and known locality specimens. Anim. Cons. 4,357–367.

Peterson, B.K., Weber, J.N., Kay, E.H., Fisher, H.S., Hoekstra, H.E., 2012. Double digestRADseq: an Inexpensive Method for De Novo SNP Discovery and genotyping in modeland non-model species. PLoS ONE 7 (5), e37135. https://doi.org/10.1371/journal.pone.0037135.

Pritchard, J.K., Stephens, M., Donnelly, P., 2000. Inference of population structure usingmultilocus genotype data. Genetics 155, 945–959.

Seidel, M.E., 2002. Taxonomic observations on extant species and subspecies of sliderturtles, genus Trachemys. J. Herp. 36, 285–292.

Seidel, M.E., Ernst, C.H., 2016. A systematic review of the turtle family Emydidae. Vert.Zool. 67, 1–122.

Seidel, M.E., Stuart, J.N., Degenhardt, W.G., 1999. Variation and species status of sliderturtles (genus Trachemys) in the southwestern United States and adjacent Mexico.Herpet. 55, 470–487.

Shi, H., Parham, J.F., Fan, Z., Hong, M., Yin, F., 2008. Evidence for the massive scale ofturtle farming in China. Oryx- Internat. J. Cons. 42, 147–150.

Somma, L.A., Foster, A., Fuller, P., 2019. Trachemys scripta elegans (Weid-Neuwied, 1838):U.S. Geological Survey, Nonindigenous Aquatic Species Database.

Smith, M.L., Miller, R.R., 1986. The evolution of the Rio Grande Basin as inferred from itsfish fauna. In: Hocuu, C.H., Wiley, E.O. (Eds.), Zoogeography of North AmericanFreshwater Fishes. John Wiley and Sons, New York, pp. 457–485.

Stamatakis, A., 2014. RAxML Version 8: A tool for phylogenetic analysis and post-analysisof large phylogenies. Bioinf. 30, 1312–1313.

Starkey, D.E., Shaffer, H.B., Burke, R.L., Forstner, M.R.J., Iverson, J.B., Janzen, F.J.,Rhodin, A.G.J., Ultsch, G.R., 2003. Molecular systematics, phylogeography, and theeffects of Pleistocene glaciation in the painted turtle (Chrysemys picta) complex. Evol.57, 119–128.

Stuart, B.L., Parham, J.F., 2007. Recent hybrid origin of three rare Chinese turtles. Cons.Gen. 8, 169–175.

Stuart, J.N., 2000. Additional notes on native and non-native turtles of the Rio Grandedrainage basin, New Mexico. Bull. Chic. Herp. Soc. 35, 229–235.

Stuart, J.N., Ward, J.P., 2009. Trachemys gaigeae (Hartweg, 1939) - Big Bend slider,Mexican Plateau slider, jicotea de la Meseta Mexicana. Chel. Res. Mon. 5032.1–032.12.

Swensen, N.G., Howard, D.J., 2005. Clustering of contact zones, hybrid zones, and phy-logeographic breaks in North America. Amer. Nat. 166, 581–591.

Swofford, D.L. 2003. PAUP*: Phylogenetic Analysis Using Parsimony *(and OtherMethods). Version 4.0a166. Sinauer Associates, Sunderland, MA.

TFTSG (Tortoise and Freshwater Turtle Specialist Group), 1996. Trachemys adiutrix(errataversion published in 2016). The IUCN Red List of Threatened Species 1996: e.T40762A97298795.

TFTSG (Tortoise and Freshwater Turtle Specialist Group), 2017. Turtles of the World:Annotated Checklist and Atlas of Taxonomy, Synonymy, Distribution, andConservation Status (8th Ed.). Chel Res. Mon. 7, 1–292.

Tracking Red-Eared Sliders, 2019. iNaturalist Project. <https://www.inaturalist.org/projects/tracking-invasive-red-eared-sliders>.

van Dijk, P.P., 2011. Trachemys gaigeae(errata version published in 2016). The IUCN RedList of Threatened Species 2011: e.T22024A97429519.

van Dijk, P.P., Flores-Villela, O., 2007. Trachemys taylori(errata version published in2016). The IUCN Red List of Threatened Species 2007: e.T63662A97431374.

Wiens, J.J., Kuczynski, C.A., Stephens, P.R., 2010. Discordant mitochondrial and nucleargene phylogenies in emydid turtles: implications for speciation and conservation.Biol. J. Linn. Soc. 99, 445–461.

J.F. Parham, et al. Molecular Phylogenetics and Evolution 145 (2020) 106722

9