The two tree-building approaches delivered similar results (Figs 4, 5). In agreement with earlier studies (Fritz et al. 2012, 2023; Vargas-Ramírez et al. 2017), the previously studied taxa correspond to five deeply divergent and well-supported clades within Trachemys. These are, in Figure 5 from top to bottom, one clade each for the two Mexican and Central American species (1) T. venusta, a species having also two subspecies in northern South America (T. v. callirostris, T. v. chichiriviche), and (2) T. grayi, (3) a further clade comprising the South American species T. dorbigni + T. medemi, (4) another clade consisting of the North American species T. scripta + T. gaigeae, and (5) yet another clade containing all West Indian species (T. decorata, T. decussata, T. stejnegeri, T. terrapen). However, there are two additional clades, one comprised of sequences of T. yaquia and T. ornata and the other clade, of the two subspecies of T. nebulosa (see below).

Figure 4. 

Mitochondrial phylogeny of Trachemys species and related taxa as inferred by RAxML 8.0.0, rooted with Deirochelys reticularia and based on 3221 bp of mtDNA (12S, ND4L, ND4, cyt b plus adjacent tRNA-Thr, 133 specimens). Codes preceding taxon names are voucher or ENA accession numbers (see also Table S1). Numbers at nodes are bootstrap values. Note the placement of Trachemys nebulosa (red) outside Trachemys as sister lineage of Malaclemys terrapin and the well-supported monophyly of the remaining Trachemys taxa. Inset picture: T. n. hiltoni (photo: P. C. Rosen).

Figure 5. 

Mitochondrial phylogeny of Trachemys species and related taxa as inferred by MrBayes 3.2.6, rooted with Deirochelys reticularia and based on 3221 bp of mtDNA (12S, ND4L, ND4, cyt b plus adjacent tRNA-Thr, 133 specimens). Clades collapsed to cartoons. Codes preceding taxon names are voucher or ENA accession numbers (see also Table S1). Numbers at nodes are posterior probabilities. Note the placement of Trachemys nebulosa (red) outside Trachemys as sister lineage of Malaclemys terrapin and the well-supported monophyly of the remaining Trachemys taxa. Grey rectangle top left shows details for right grey rectangle. Inset pictures, top and bottom: T. v. venusta (photo: U. Fritz) and M. terrapin (photo: A. T. Coleman).

Within T. grayi, the three currently recognized subspecies T. g. grayi, T. g. emolli and T. g. panamensis and within T. dorbigni, the two subspecies T. d. dorbigni and T. d. adiutrix represent reciprocally monophyletic clades. However, this is not the case with respect to the subspecies of T. venusta. Also, the placement of some previously unstudied taxa was unexpected and two taxa are not distinct. Our only representative of T. v. iversoni clusters within T. venusta and shares with some T. v. venusta and T. v. uhrigi the same mitochondrial lineage. Trachemys taylori clusters within T. venusta as well. In contrast, sequences of T. yaquia and T. ornata are distinct and reciprocally monophyletic. They represent together a well-supported and deeply divergent clade which is, with high support, sister to T. venusta.

The clade (Trachemys yaquia + T. ornata) + T. venusta is with weak support sister to T. grayi. These four mainly Mexican and Central American species represent together a well-supported clade that also contains the South American taxa T. dorbigni + T. medemi. This more inclusive clade comprised of Mexican, Central and South American taxa occurs in an unresolved but well-supported clade that also contains the two clades of North American and West Indian Trachemys. Notably, T. nebulosa is excluded from this Trachemys clade and appears unexpectedly with weak support as sister taxon of the diamondback terrapin Malaclemys terrapin. This latter Malaclemys + T. nebulosa clade and another clade corresponding to Graptemys occur together with the monophyletic Trachemys exclusive T. nebulosa in an unresolved polytomy in another well-supported clade; the sister group of this clade is Chrysemys + Pseudemys.

Our only sequence of the nominotypical subspecies of T. nebulosa from the Baja California Peninsula is not clearly differentiated from seven T. n. hiltoni from Sinaloa.

Mitochondrial divergence within Trachemys commenced 6.1 million years ago (mya; Fig. S7). The mitochondrial divergence between T. gaigeae and T. scripta was dated to 2.2 mya; and that between T. nebulosa and M. terrapin, to 7.0 mya. All obtained estimates were younger than those presented in Fritz et al. (2012) for a similar taxon sampling (Table S10). However, the calculations in Fritz et al. (2012) included in addition to the same mitochondrial genes also the five nuclear loci of the present study.

Using our five nuclear loci Cmos, ODC, R35, Rag1, and Rag2, the relationships of the studied taxa are only incompletely resolved. However, several firm conclusions can be deduced.

Our SplitsTree analysis using a phased dataset with a maximum of 5% missing sequence data (Fig. 6) clearly places T. nebulosa into the same cluster as all other Mexican and Central American taxa; Graptemys and Malaclemys are closely related and highly divergent. There is no evidence for a close relationship of Malaclemys and T. nebulosa. Trachemys gaigeae, the Antillean taxa T. decussata angusta and T. stejnegeri vicina, Pseudemys and Chrysemys represent further deeply divergent branches or clusters. Within the cluster comprised of Central and South American Trachemys, most species were returned as distinct subclusters or distinct branches. Conversely, most subspecies could not be resolved, even though T. v. venusta, T. v. cataspila and T. v. uhrigi appear as largely distinct subclusters. However, their distinctness is not supported in the phylogenetic analyses (Figs S8, S9, see below). The two South American subspecies T. v. callirostris and T. v. chichiriviche represent together another weakly distinct subcluster in the SplitsTree analy­sis (Fig. 6), with an intermediate position between the subcluster of T. v. uhrigi and the South American species T. medemi and T. dorbigni. Though, in the phylogenetic analyses, the placement of T. v. callirostris and T. v. chichiriviche is unresolved and without statistical support (Figs S8, S9). The single specimen of T. n. nebulosa was included in the same SplitsTree subcluster as seven T. n. hiltoni (Fig. 6). With regard to species allocation, two T. grayi samples were misplaced. One T. g. panamensis (MTD D 42599, allele numbers 77 and 78 in Fig. 6) represents a distinct cluster and one T. g. emolli (SMF 71417, numbers 27 and 28 in Fig. 6) was placed among T. v. uhrigi. The alleles of one T. v. uhrigi clustered among T. v. cataspila (FMNH 283808, numbers 125 nd 126 in Fig. 6).

Figure 6. 

SplitsTree for phased and concatenated nuclear DNA sequences of Trachemys species and related taxa (Cmos, ODC, R35, Rag1, Rag2, 3396 bp, 82 specimens; sequences with less than 5% missing data). Numbers at branch tips refer to alleles, see Table S1 for explanation. Note the similarity of Graptemys and Malaclemys and the placement of Trachemys nebulosa (red) next to the geographically neighboring Trachemys taxa (T. gaigeae, T. ornata, T. yaquia). Conflicting samples highlighted with solid blue circles. Inset picture: T. ornata (photo: P. C. Rosen).

RAxML and MrBayes analyses using all phased sequences (Figs S8, S9), i.e., also those with more than 5% missing data, confirm the general patterns. Notably, the two algorithms place T. nebulosa into a well-supported clade together with the other Mexican, Central and South American Trachemys species, i.e., T. dorbigni, T. grayi, T. medemi, T. ornata, T. venusta (including T. taylori), and T. yaquia. Furthermore, Graptemys and Malaclemys are deeply divergent and well-supported sister taxa. The two museum specimens of T. taylori for which nuclear DNA sequences could be obtained, are placed within T. venusta and not distinct from sequences from T. v. cataspila. These two specimens were not included in the SplitsTree calculation due to missing data. The sequences of the two subspecies of T. nebulosa were slightly distinct in the trees, in contrast to the SplitsTree analysis. The unexpected position of the two above-mentioned T. grayi samples (MTD D 42599, SMF 71417) and the T. v. uhrigi (FMNH 283808) is also reflected in the phylogenetic trees.

Even though the trees are generally not well resolved, the following additional observations are noteworthy (i) the South American species T. dorbigni (with the subspecies T. d. dorbigni and T. d. adiutrix) and T. medemi are well-supported sister taxa within the Mexican, Central and South American clade; (ii) sequences of the two T. nebulosa subspecies are another well-supported subclade within the Mexican, Central and South American clade and the only representative of T. n. nebulosa is distinct from T. n. hiltoni; (iii) T. grayi, T. taylori and T. venusta are not reciprocally monophyletic; (iv) sequences of T. taylori cluster with sequences of T. v. cataspila; (v) several sequences of T. v. uhrigi represent a weakly supported subclade that corresponds to the subcluster for T. v. uhrigi in the SplitsTree; this subclade includes 11 (MrBayes) or 13 (RAxML) sequences of T. v. uhrigi from Guatemala, Honduras, Nicaragua and the two alleles of one T. g. emolli (SMF 71417) from Costa Rica, (vi) the remaining 13 or 11 of the 24 sequences of T. v. uhrigi (which were mostly not used in the SplitsTree because of missing data) appear in remote positions across the Mexican, Central and South American clade, either in unresolved polytomies or they cluster with weak support with alleles of T. v. venusta, T. v. cataspila, T. v. chichiriviche, or T. g. panamensis; (vii) T. ornata and T. yaquia constitute distinct subclades within the Mexican, Central and South American clade; (viii) the only representative of T. gaigeae is deeply divergent from the Mexican, Central and South American clade and clusters within an unresolved polytomy that also contains the other North American taxa, i.e., T. scripta, a well-supported clade comprised of Malaclemys with Graptemys as its the well-supported sister, Chrysemys and Pseudemys, and the West Indian Trachemys taxa; (ix) within this polytomy, the West Indian species represent a well-supported monophylum; (x) some sequences of the North American T. scripta are sister to the West Indian taxa, while others cluster with weak support with Pseudemys.

The results of our analyses based on mitochondrial and nuclear DNA are not in complete agreement. When the mitochondrial and nuclear topologies are compared (Figs 4, 5, 6, S8, S9), the most notable difference is the conflicting placement of Trachemys nebulosa. In the mitochondrial phylogenies (Figs 4, 5), T. nebulosa is with weak support sister to Malaclemys terrapin, a highly distinctive species confined to coastal zones with brackish water. It is distributed farther east along a narrow strip following the southern and eastern US coasts from Texas to Cape Cod and represents a monotypic genus (Ernst and ­Barbour 1989; Seidel and Ernst 2017; TTWG 2021). In sharp contrast, our analyses based on nuclear DNA (Figs 6, S8, S9) reveal that T. nebulosa belongs to the Mexican and Central American Trachemys clade, in agreement with its traditional taxonomic assignment (e.g., Moll and Legler 1971; Ernst and Barbour 1989; Legler 1990; Seidel and Ernst 2017; TTWG 2021). According to the nuclear data, Malaclemys is phylogenetically deeply divergent from Trachemys and sister to the genus Graptemys, the morphologically distinctive map and sawback turtles from southeastern North America. This topology is in line with previous investigations (see the reviews in Seidel and Ernst 2017 and TTGW 2021) including a study based on 15 nuclear genes (Thomson et al. 2021). However, the sister group relationship of T. nebulosa + (T. dorbigni + T. taylori) shown in the tree in Thomson et al. (2021) is questionable because samples misidentified on the species level seem to have been used for Trachemys (Fritz et al. 2023). Further insecurity arises from the conflicting placement of T. nebulosa in the tree presented in the supplementary information of Thomson et al. (2021), where T. nebulosa is sister to a clade comprised of all other Mexican, Central and South American Trachemys taxa.

Already the results of the pioneering study by Wiens et al. (2010) on the relationships of emydid turtles resembled our mitochondrial topologies in that a single sample of T. nebulosa was with weak support sister to Malaclemys in their tree based on mitochondrial cyt b and ND4 sequences, while it was the well-supported sister taxon of a single sample of T. venusta in an analysis using six nuclear loci. Seidel and Ernst (2017) suggested that the mitochondrial results of Wiens et al. (2010) were erroneous, which seemed reasonable in the light of the phylogeny presented by Parham et al. (2015). The latter authors included in their analysis of Central American and Mexican Trachemys both T. n. nebulosa and T. n. hiltoni, which were sister taxa and occurred on a long branch within a monophyletic Trachemys. Parham et al. (2015) used a combined dataset of mitochondrial (ND4) and nuclear (R35) DNA sequences. Therefore, the placement of the two T. nebulosa subspecies was due to the signal of the nuclear locus. A comparison of the mtDNA sequences from Wiens et al. (2010) and Parham et al. (2015) with ours shows a complete match.

Mitochondrial phylogenies of Trachemys and related emydids can be easily confounded by the unintended inclusion of numts (non-coding nuclear mitochondrial DNA insertions) obtained with standard PCR primers (see Fritz et al. 2012). However, this is unlikely for T. nebulosa. We used for all of our Sanger-sequenced samples the approach described in Fritz et al. (2012) to minimize the risk of amplifying numts. This approach involves long-range PCR and tailored primers. Thus, we are confident that our mitochondrial DNA sequences are authentic. However, how can the conflicting evidence for nuclear and mitochondrial DNA be explained?

Mitochondrial introgression and mitochondrial capture are known to have occurred in multiple turtle clades and sometimes across deeply divergent taxa (Chelidae: Chelodina, Emydura, Myuchelys – Hodges 2015; Kehlmaier et al. 2019, 2024; Emydidae: Actinemys, Graptemys – Spinks and Shaffer 2009; Praschag et al. 2017; Geoemydidae: Cuora, Cyclemys, Malayemys, Rhinoclem­mys – Spinks and Shaffer 2007; Fritz et al. 2008; Vargas-Ramírez et al. 2013; Ihlow et al. 2016; Vamberger et al. 2017; Kinosternidae: Kinosternon – Hurtado-Gómez et al. 2024; Trionychidae: Pelodiscus – Gong et al. 2018). Fritz et al. (2023) also considered mitochondrial capture for T. gaigeae, whose mitogenome is the sister lineage of North American T. scripta, even though T. gaigeae is highly distinct according to nuclear information (see also our Figs 6, S8, S9) and resembles Mexican, Central and South American Trachemys taxa in morphology, sexual dimorphism and the head bobbing behavior during courtship (Legler and Vogt 2013).

This situation suggests that the deeply divergent mitochondrial lineage in T. nebulosa represents another case of mitochondrial capture, either from the ancestor of the extant Malaclemys terrapin or its extinct sister taxon. According to our exploratory molecular clock calculations, the mitochondrial lineages of M. terrapin and T. nebulosa diverged 7.0 mya (95% HPD: 5.2–9.6 mya; Fig. S7 and Table S10), and this estimate might reflect the approximate time of mitochondrial capture. It is remarkable that this estimate predates that for the divergence of the mitochondrial lineages within Trachemys, even though the 95% HPD intervals widely overlap (5.2–9.6 mya and 4.9–7.7 mya; Table S10). However, there are several caveats. In particular, the inferred date could correspond to the divergence of the ancestor of M. terrapin and its extinct sister taxon, and not to the date of the mitochondrial capture, i.e., T. nebulosa could have captured the mitogenome later. Furthermore, the calibration points used may be misleading because the divergence history of mitogenomes is not necessarily congruent with the diversification of the ‘host’ organisms. Mitochondrial genes behave like a single locus, and a molecular clock should be ideally applied to a species tree or a multilocus dataset, not a single locus. Also, the mitochondrial sister group relationship of T. nebulosa and M. terrapin is only weakly supported and the foreign mitogenome of T. nebulosa could originate from another extinct emydid lineage. In any case, the divergence time estimate and the deep divergence of the mtDNA of T. nebulosa suggest that the mitochondrial capture occurred very early during the diversification of Trachemys, perhaps when the diversification of the genus began. In contrast, T. gaigeae captured its mitogenome from the ancestor of T. scripta much later, although our estimate of 2.2 mya (Table S10) should be treated with the same reservations as for T. nebulosa.

Our nuclear dataset of five loci obviously does not completely resolve the phylogeny of Trachemys. However, the placement of T. nebulosa is consistent in Splits­Tree, Bayesian and Maximum Likelihood analyses (Figs 6, S8, S9) and well supported. In addition, and in agreement with the calculations presented in Fritz et al. (2023), the analyses of our nuclear data allow further insights. Mexican, Central and South American Trachemys taxa appear to be closely related, while T. scripta, the West Indian taxa and T. gaigeae are distinct. The placement of some taxa and sequences in our SplitsTree analysis makes geographic sense, suggestive of past or current gene flow. This is true for the clusters of the South American taxa T. d. dorbigni, T. d. adiutrix and T. medemi which are neighbors of the geographically close South American subspecies of T. venusta (T. v. callirostris, T. v. chichiriviche; Fig. 6). Furthermore, five individuals of the southern Central American subspecies T. v. uhrigi are neighbor to the South American taxa. Also, the geographically close T. ornata and T. yaquia are neighbors in the SplitsTree. However, neither T. v. venusta and T. v. cataspila nor the three subspecies of T. grayi (T. g. grayi, T. g. emolli, T. g. panamensis) are clearly distinct.

Based on a cladistic analysis of morphological traits, Seidel (2002) recognized the allopatric taxa T. v. callirostris and T. v. chichiriviche as subspecies of a distinct species, T. callirostris. However, his results are questionable in the light of current genetic evidence. They included many untenable findings, such as the deeply divergent non-sister placement of two populations of T. d. dorbigni (‘brasiliensis’ and ‘dorbigni’), or the non-monophyly of the Antillean taxa and of T. grayi, or the paraphyly of T. venusta with respect to T. g. grayi (Seidel 2002: fig. 2). According to gross morphology, T. v. callirostris and T. v. chichiriviche differ little from other T. venusta subspecies. With respect to mtDNA, T. v. callirostris and T. v. chichiriviche are embedded in the remaining mitochondrial lineages of T. venusta (Figs 4, 5), and our nuclear markers do not provide unambiguous evidence for any classification (Figs 6, S8, S9). We therefore continue to treat T. v. callirostris and T. v. chichiriviche as subspecies of T. venusta, as proposed in Fritz et al. (2012).

Some sequences in the SplitsTree analysis are at first glance misplaced (highlighted with blue circles in Fig. 6). This could reflect ancestral polymorphism or hybridization. The latter option seems likely for a specimen (SMF 71417) morphologically identified as T. g. emolli (in agreement with its mitochondrial identity; see Figs 4, 5), which clusters within T. v. uhrigi. It originates from Costa Rica, where a contact zone of T. g. emolli and T. v. uhrigi along the Caribbean zone is expected (see Fritz et al. 2023). Regarding another misplaced specimen (MTD D 42599, T. g. panamensis) from Panama, Fritz et al. (2023) speculated about an artifact because all mutations separating this specimen from other T. g. panamensis occur in only one nuclear locus. However, hybridization and subsequent recombination could also explain this. A third misplaced specimen is a T. v. uhrigi from Honduras (FMNH 283808), which clusters among T. v. cataspila, reflecting the incomplete differentiation among T. venusta subspecies, also revealed by the bifurcating trees (Figs S8, S9).

Some sequences of North American Trachemys taxa cluster in the phylogenetic analyses either with the West Indian Trachemys species or with Pseudemys. It is speculative whether the latter finding reflects ancestral polymorphism or past hybridization. Pseudemys is widely sympatric with Trachemys in the southeastern USA (compare the maps in TTWG 2021) and it is well known that even very distantly related chelonians are capable of successful hybridization (e.g., Graptemys x Trachemys; Gooley et al. 2016). It is noteworthy that males of North American and West Indian Trachemys share the same sexually dimorphic traits (elongated foreclaws) and the innate courtship behavior consisting of claw vibrations (“titillation”) in front of the female’s head. In contrast, this courtship behavior is unknown in Central and South American Trachemys, in which males do not have greatly elongated foreclaws. However, in some Central and South American taxa males have prominent elongated and upturned snouts, emphasizing head bobbing movements during courtship (Fritz 1990; Seidel and Fritz 1997).

The presence of the titillation behavior is a plesio­morphic character state since it also occurs in other genera (in particular in Chrysemys, Graptemys, and Pseudemys), while its loss is an autapomorphy of Central and South American Trachemys which still sporadically display claw titillation in another behavioral context (aggressive male-male encounters; Fritz 1990, 1991; Seidel and Fritz 1997; Seidel and Ernst 2017).

Compared to our nuclear DNA dataset, phylogenetic analyses of faster evolving mtDNA sequences delivered more information, which albeit reflects only matrilinear evolution confounded by mitochondrial introgression or capture. Except for the above-mentioned unexpected placement of T. nebulosa, our mitochondrial trees (Figs 4, 5) contain Trachemys as a well-supported monophylum with T. ornata and T. yaquia together as a deeply divergent mitochondrial lineage. Trachemys ornata and T. yaquia are reciprocally monophyletic; their genetic divergence resembles those between the three subspecies of T. grayi or within T. venusta. In contrast to the weak mitochondrial divergence of T. ornata and T. yaquia, the two taxa appear well separated in our analyses of nuclear DNA (Figs 6, S8, S9), supporting their recognition as distinct species.

In the mitochondrial trees, not all subspecies of T. venusta are reciprocally monophyletic, in contrast to the subspecies of T. dorbigni and T. grayi (Figs 4, 5). The sequences of the subspecies of T. venusta are scattered across a polytomy in which also T. taylori is embedded. Both the mitochondrial (Figs 4, 5) and the nuclear analyses (Figs S8, S9) fail to identify T. taylori as a lineage distinct from T. venusta. This conflicts with the current status of T. taylori as a distinct species. Using ND4 and R35 sequences, Parham et al. (2015) could not resolve the phylogenetic position of T. taylori, and the placement of sequences labeled as T. taylori in the nuclear phylogeny of Thomson et al. (2021) is questionable, as misidentified samples appear to have been used (Fritz et al. 2023). Our present data strongly suggest that T. taylori does not represent a distinct species and either is a disjunct population of T. venusta, perhaps of T. v. cataspila, a morphologically similar and geographically close taxon, or a subspecies that recently diverged from T. venusta. Notably, based on general morphology, Legler and Vogt (2013: 259) concluded that T. taylori, T. v. cataspila and T. v. venusta are most closely related. Further research is needed to clarify this situation in detail, but in the light of the current data we propose to treat T. taylori as a subspecies of T. venusta. Using SNP data, Espindola et al. (2022) revealed T. taylori as sister to a clade comprised of T. v. venusta and T. v. cataspila, without discussing taxonomic implications. The branch lengths in their phylogram do not contradict our classification.

Trachemys venusta taylori nov. comb. is endemic to the endorheic Cuatro Ciénegas Basin of Coahuila, Mexico, from where two further endemic turtle taxa have been described, Apalone spinifera atra, the black spiny softshell turtle, and Terrapene coahuila, the Coahuilan box turtle. While A. s. atra was originally described as a distinct species (Webb and Legler 1960), it later turned out that it is genetically much less differentiated than expected (­McGaugh and Janzen 2008; McGaugh et al. 2008; McGaugh 2012), leading to its recognition as a subspecies of A. spinifera (TTWG 2021). Also, the genetic divergence of the morphologically and ecologically divergent aquatic Coahuilan box turtle (cf. Howeth and Brown 2011; Legler and Vogt 2013) is unexpectedly weak; phylogenetically it is nested within the widely distributed and allopatric Terrapene carolina (Martin et al. 2021; Thomson et al. 2021). This suggests that all endemic turtle taxa in the Cuatro Ciénegas Basin are phylogenetically very young.

Within the mitochondrial clade of Trachemys venusta, some taxa correspond to distinct subclades (Figs 4, 5: T. v. callirostris, T. v. chichiriviche, T. v. taylori), while others do not (T. v. venusta, T. v. cataspila, T. v. iversoni, T. v. uhrigi). In contrast to the latter taxa having contiguous parapatric distributions, T. v. callirostris, T. v. chichiriviche, and T. v. taylori are allopatric (compare maps in TTWG 2021 and our Fig. 1). It is remarkable that some of the mitochondrially not clearly distinct taxa appear relatively well-separated in our SplitsTree analysis of nuclear DNA sequences with less than 5% missing data (Fig. 6). However, when the SplitsTree analysis is compared to the remaining phylogenetic analyses of nuclear data, the distinction is no longer so clear (Figs S8, S9). Nevertheless, that some samples appear distinct could indicate an incipient differentiation which is not reflected by mtDNA. This is unexpected considering the much slower pace of nuclear DNA evolution. Possibilities that could have contributed to this confusing pattern are mitochondrial introgression, selective sweeps, or the translocation and subsequent hybridization of turtles.

Several of our samples identified as T. v. venusta originate from the region of Acapulco de Juárez, a tourist destination where multiple introductions of T. venusta have been inferred (Parham et al. 2015; Fritz et al. 2023). This is also underlined by the photos of three live turtles from Acapulco published in Parham et al. (2015: fig. 3) matching specimens from Tamaulipas (T. v. cataspila) in having an irregular carapacial coloration with or without small ocelli and broken postorbital stripes. Our museum specimens from Acapulco (MTD D 39071, 39077, 42598) have in contrast complete well-developed carapacial ocelli and wide continuous postorbital stripes, matching the nominotypical subspecies. This supports that anthropogenic admixture plays a role. However, introductions alone cannot explain our genetic results because we also sequenced museum specimens of T. v. cataspila and T. v. venusta that have been collected between 1936 and 1955 (Table S1), presumably long before human-mediated long-distance translocations of slider turtles happened. These specimens were genetically also not distinct. Further research, preferably using more informative nuclear genomic markers and more samples, is needed to clarify this intricate situation.

An inspection of the external morphology of our sequenced museum specimens and published known-locality photographs revealed that some individuals display unexpected traits. While most of our specimens of T. v. uhrigi (identified according to their collection sites) show the diagnostic characters highlighted in the original description (narrow postorbital stripes, large dark pattern covering most of the plastron; McCord et al. 2010), two have wide postorbital stripes and a narrow dark plastral pattern (MTD D 41609, Honduras; SMF 77494, Nicaragua). This is in line with the observations by McCranie (2018) who studied the morphology of many putative T. v. uhrigi from Honduras and concluded “the diagnostic characters given for that nominal form [T. v. uhrigi] … conceal the fact that much more variation occurs in Honduran specimens.”

It is clear that further research is needed to examine whether the coloration and pattern traits used by McCord et al. (2010) to tell apart subspecies of Central American sliders represent more than population-specific or even individual variation which has been overestimated by cherry-picking morphologically matching turtles. That these coloration and pattern traits are taxonomically unreliable is supported by photos of two live sliders from Dziuché, Quintana Roo, Mexico, and two preserved specimens from Puerto Morelos in the same Mexican state published in Legler and Vogt (2013: figs 40.5 and 40.6). According to the collection sites, these turtles represent T. v. iversoni. However, they show narrow to extremely narrow postorbital stripes as described by McCord et al. (2010) for T. v. uhrigi. The plastra of the two preserved sliders have a narrow dark figure, whereas McCord et al. (2010) characterize T. v. iversoni by a “greatly expanded plastral pattern.” Another two live T. v. iversoni figured in TTWG (2021: 169) from Cobá, Quintana Roo, and Muná, Yucatán, Mexico, resemble the specimens depicted in Legler and Vogt (2013). Our only studied and sequenced museum specimen of T. v. iversoni (SMF 70537) has a nearly uniform yellow plastron. It only yielded mtDNA sequences, and these were not differentiated from those of nine T. v. venusta and T. v. uhrigi.

Pictures of another two sliders from Alvarado, Veracruz, Mexico, in Legler and Vogt (2013: figs 40.1 and 40.2, according to collection site T. v. venusta), show that in the same population individuals with narrow and wide postorbital stripes and very different carapacial ocelli may occur. Also, the lacking or weak genetic differentiation of the subspecies of another slider species, T. scripta, and among the species and subspecies of two closely related turtle genera, Graptemys and Pseudemys, argue for caution (Spinks et al. 2013; Praschag et al. 2017; Vamberger et al. 2020). This situation underlines the concerns voiced in Fritz et al. (2023) that the conspicuous and elaborate color pattern of sliders and their kin resulted in the recognition of taxa that merely reflect population-level variation, a phenomenon of taxonomic inflation also known from other biota with a complex external morphology (e.g., beetles, butterflies, mollusks; Páll-Gergely et al. 2019).

Our analyses of mitochondrial and nuclear DNA sequences do not unambiguously support the distinctness of many subspecies of T. venusta (including T. v. taylori comb. nov.), while the three currently recognized subspecies of T. grayi are distinct in mtDNA. Remarkably, both T. v. uhrigi and T. v. iversoni were ignored in the monograph on the freshwater turtles and tortoises of Mexico by Legler and Vogt (2013) and the respective populations were treated under the nominotypical subspecies T. v. venusta.

Although based on subtle differences only, the color patterns of the allopatric South American subspecies of T. venusta (T. v. callirostris, T. v. chichiriviche) and the Mexican subspecies T. v. cataspila are easily recognizable, and this is also true for two subspecies of T. grayi, T. g. grayi and T. g. emolli (compare Pritchard and Trebbau 1984; Legler and Vogt 2013; the third subspecies of T. grayi, T. g. panamensis, is not sufficiently known but may be distinctive as well, see Fritz et al. 2023). However, the putatively diagnostic coloration and pattern traits of the remaining subspecies need to be re-examined and seem to be unreliable to tell apart T. v. venusta, T. v. iversoni and T. v. uhrigi; the same could be true for T. v. cataspila and T. v. taylori nov. comb. Thus, in the face of our genetic results, one option could be to synonymize T. v. iversoni and T. v. uhrigi under T. v. venusta and T. v. taylori under T. v. cataspila.

Instead, we call for further research using larger sample sizes and preferably genome-wide nuclear markers such as SNPs or low-coverage genome sequencing. In times of large-scale biodiversity loss, the continued use of subspecies names for allopatric and parapatric populations will help prevent inadvertent admixture and erosion of biodiversity when confiscated turtles are released or during conservation measures (ex-situ breeding, population reinforcements) until a better scientific foundation allows for solid evidence-based conservation decisions. Indeed, analyses of SNP data for T. v. venusta, T. v. cataspila and T. v. taylori (Espindola et al. 2022) support that these taxa are distinct.

In a similar vein, more research is also needed to examine whether the two currently recognized subspecies of T. nebulosa are distinct. Their recognition is largely based on their allopatric distribution ranges (Legler and Vogt 2013) with the mainland subspecies T. n. hiltoni in the Río Fuerte drainage while the nominotypical subspecies occurs on the Baja California Peninsula (Fig. 1). The putatively diagnostic traits (plastral coloration, pygal bone shape; Seidel 2010) are subtle and seem to have never been systematically examined. Our genetic data allow no firm conclusions. Parham et al. (2015) found one individual of T. n. nebulosa slightly different from four specimens of T. n. hiltoni using combined analyses of the mitochondrial ND4 gene and the nuclear R35 intron. However, even though this matches our results, larger sample sizes are needed both for examining genetic and morphological differentiation before conclusions about the validity of the taxa can be drawn. Legler and Vogt (2013: 299) discussed that T. n. nebulosa and T. n. hiltoni either represent natural relicts of a formerly more widely distributed taxon around the entire Gulf of California or that the Baja California population (T. n. nebulosa) originates from natural or human-mediated dispersal across the Gulf. However, since the range of T. yaquia would separate a continuous range along the Gulf coast, a subsequent range disruption cannot explain the current distribution pattern (compare Fig. 1). An alternative reverse scenario would be T. nebulosa originated on the Baja Cali­fornia Peninsula and crossed from there the Gulf to the mainland.