Chrysemys, Gray, 1844

Jasinski, Steven E., 2023, A new species of Chrysemys (Emydidae: Deirochelyinae) from the latest Miocene-Early Pliocene of Tennessee, USA and its implications for the evolution of painted turtles, Zoological Journal of the Linnean Society 198, pp. 149-183 : 167-175

PHYLOGENETIC RELATIONSHIPS OF MODERN NON-CHRYSEMYS DEIROCHELYINES

All modern non- Chrysemys deirochelyines in the current study were included in the phylogenetic analysis from Jasinski (2018a). Most of these relationships were also discussed by Jasinski (2018a) and will not be restated here. However, better resolution (i.e. fewer polytomies) is achieved in several parts of the tree. Within Trachemys, Jasinski (2018a) found Trachemys decorata Barbour & Carr, 1940 , Trachemys stejnegeri stejnegeri Schmidt, 1928 and Trachemys terrapen Bonnaterre, 1789 to form a polytomy in their strict consensus tree (Jasinski, 2018: fig. 12) but found T. decorata + ( T. stejnegeri stejnegeri + T. terrapen ) in their 50% majority rule consensus tree (see Jasinski, 2018: fig. S74). The present study finds T. decorata + ( T. stejnegeri stejnegeri + T. terrapen ) in both the strict consensus ( Fig. 8 ) and 50% majority rule consensus ( Fig. 9 ) trees. Both Jasinski (2018a) and the present study find a sister relationship between Trachemys gaigeae gaigeae Hartweg, 1939 and Trachemys scripta elegans Wied-Neuwied, 1839 in the 50% majority rule consensus trees, while they are part of a polytomy with a clade of fossil Trachemys and modern southern Trachemys species in the strict consensus trees. Jasinski (2018a) recovered Malaclemys terrapin terrapin Schoepff, 1793 + Graptemys barbouri Carr & Marchand, 1942 + ( Graptemys geographica LeSueur, 1817 + Graptemys pseudogeographica pseudogeographica Gray, 1831 ) for a Malaclemys and Graptemys clade. The present study recovers these three modern Graptemys species in a polytomy, with Malaclemys sister to this clade in the strict consensus tree. The 50% majority rule consensus tree recovers the same relationships of the three Graptemys species as Jasinski (2018a), with G. barbouri as sister to G. geographica and G. p. pseudogeographica , and M. terrapin terrapin as sister to the Graptemys clade. Pseudemys species maintain the same intrageneric relationships ( Pseudemys concinna concinna Le Conte, 1830 + ( Pseudemys nelson Carr, 1938 + Pseudemys rubriƲentris Le Conte, 1830 )) as in Jasinski (2018a). Intergeneric relationships among non- Chrysemys deirochelyines also agree between the two studies and in both the strict consensus and 50% majority rule consensus trees, with Deirochelys sister to other nonChrysemys members of the subfamily ( Pseudemys + (( Malaclemys + Graptemys ) + Trachemys )).

PHYLOGENETIC RELATIONSHIPS OF FOSSIL NON-CHRYSEMYS DEIROCHELYINES

All fossil non- Chrysemys deirochelyines in the current study were also included in the phylogenetic analysis from Jasinski (2018a) except for two fossil species currently inferred to belong to Chrysemys ( C. antiqua and C. corniculata ). Among the fossil Trachemys species, the only difference between Jasinski (2018a) and the present study is a sister relationship between T. haugrudi + Trachemys inflata Weaver & Robertson, 1967 , with Trachemys platymarginata Weaver & Robertson, 1967 as sister to this clade. The south-eastern and mid-western fossil Trachemys species clades recovered by Jasinski (2018a) are again recovered. Jasinski (2018a) found Graptemys kerneri Ehret & Bourque, 2011 as part of a polytomy with Malaclemys terrapin terrapin and modern Graptemys species in their strict consensus tree and sister to Malaclemys + modern Graptemys in their 50% majority rule consensus tree. The current analysis recovers G. kerneri in an unresolved polytomy with modern Graptemys species in the strict consensus tree and sister to G. barbouri in a clade in the 50% majority rule consensus tree. This agrees with the original hypothesis of Ehret & Bourque (2011), who inferred G. kerneri was most closely related to G. barbouri , and with the recent hypothesis by Vlachos (2018). Also similar to Jasinski (2018a), Deirochelys reticularia reticularia Latreille in Sonnini & Latreille, 1801 groups with three fossil species, including two previously referred to Deirochelys (‘ C. ’ carri and Deirochelys floridana Hay, 1908 ), and one previously referred to ‘ Pseudemys ’ (‘ P.’ caelata ). While the strict consensus tree recovers a polytomy between these four species, the 50% majority rule consensus tree finds a clade of ‘ P.’ caelata + ( D. r. reticularia + (‘ C.’ carri + D. floridana )). In both their strict consensus and 50% majority rule consensus trees, Jasinski (2018a) recovered a clade of D. floridana + (‘ P.’ caelata + (‘ C.’ carri + D. r. reticularia )). ‘ Chrysemys ’ carri is consistently found to be a derived member of this clade, while ‘ P.’ caelata is consistently found as a basal member, either as sister to the other OTUs or, at least, sister to a derived subclade within this clade. Regardless of the interrelationships of the clade, all four OTUs are consistently recovered together. This is here inferred to mean all four taxa probably represent species of Deirochelys , including ‘ P.’ caelata . This would be different from its original assignment in Pseudemys by Hay (1908) and Seidel & Smith (1986) and in Chrysemys by Jackson (1976). Jackson (1976) specifically investigated the relationships of ‘ P.’ caelata , and Jackson (1978) studied fossils of Deirochelys , but neither study thoroughly investigated how ‘ P.’ caelata compared to D. reticularia , instead focusing on comparing the species with Pseudemys and Chrysemys . While its basal position in Deirochelys may be due to less material available of the OTU, its inclusion in Deirochelys is convincing. These relationships will be further explored in a larger scale study on emydid systematics.

PHYLOGENETIC RELATIONSHIPS OF MODERN CHRYSEMYS

The four modern Chrysemys taxa were included in the phylogenetic analysis. As noted above the taxonomic level of C. dorsalis has been suggested as a separate species by some authors (e.g. Starkey et al., 2003; Jensen et al., 2014, 2015), while others have maintained it as a subspecies of C. picta (C. p. dorsalis ; e.g. Ernst & Lovich, 2009). Some, including the Turtle Taxonomy Working Group (2010, 2011, 2014, 2017, 2021) have listed it as a ‘species or subspecies’, waiting on future studies to more clearly determine its taxonomic level and relationships. The present phylogenetic analysis found the four Chrysemys taxa (or the C. picta complex) to form an unresolved polytomy in the strict consensus tree. In the 50% majority rule consensus tree, the clade was recovered as C. picta marginata + (C. p. bellii + (C. p. picta + C. p. dorsalis )). There has been mention that subspecies of Chrysemys are unique morphologically (e.g. Vamberger et al., 2020), although this suggests that, at least morphologically, C. p. dorsalis should still be considered a subspecies within C. picta , as the morphological differences are not more extreme than between the other subspecies. Both trees recover C. picta (or the C. picta complex) as sister to other modern deirochelyine genera. Jasinski (2018a), while only including C. picta picta among modern Chrysemys taxa, also found Chrysemys to be basal among deirochelyines. However, he found C. picta picta to be part of a polytomy with Deirochelys at the base of Deirochelyinae in their strict consensus tree and sister to all modern deirochelyines other than Deirochelys reticularia reticularia in their 50% majority rule consensus tree. As the current analysis focuses more on Chrysemys , its position as sister to all other modern deirochelyines seems more probable. The relatively more derived nature of Deirochelys reticularia has also been mentioned before (e.g. Jackson, 1978), and this makes the basal position of Chrysemys in relation to it not wholly unexpected. This basal position of Chrysemys , as sister to other modern deirochelyines, has been found in other studies, including Stephens & Wiens (2003, fig. 1) and Spinks et al. (2009) when using cytochrome b, and Spinks et al. (2009) for a seven-locus nuclear DNA data set.

PHYLOGENETIC RELATIONSHIPS OF FOSSIL CHRYSEMYS

‘ Chrysemys ’ aeilliamsi is found in nearly the same position in both the strict consensus ( Fig. 8 ) and 50% majority rule ( Fig. 9 ) consensus trees, sister to all deirochelyines except Chrysemys in the latter, and part of a polytomy at the base of the non- Chrysemys deirochelyine clade in the former. This is distinct from its placement by Vlachos (2018), who found it to lie in a polytomy with modern Pseudemys species. Its current phylogenetic position, unique from recognized deirochelyine genera, was also reported by Jasinski (2018a), although he found it to lie in a polytomy at the base of Deirochelyinae . Regardless, its phylogenetic position suggests it represents a unique genus of deirochelyine; however, naming a new genus for this species is currently being withheld until a more complete phylogenetic analysis of emydids, including both modern and fossil OTUs, is conducted. In the strict consensus tree ( Fig. 8 ), C. timida , C. corniculata and C. antiqua are all found to be part of an unresolved polytomy, with a clade of modern Chrysemys , at the base of the tree. However, in the 50% majority rule consensus tree ( Fig. 9 ) these three taxa form a clade that is also part of a polytomy with a modern Chrysemys clade at the base of the tree. In this latter tree, C. antiqua is sister to C. timida + C. corniculata . Although C. antiqua is by far the oldest fossil in the present study (approximately 35 Mya), its position either in an unresolved polytomy at the base of Deirochelyinae (in the strict consensus tree) or in a clade with two other fossil taxa previously referred to Chrysemys (in the 50% majority rule consensus tree) is not completely expected. This is mainly in respect to a potentially long ghost lineage from the Late Chadronian NALMA to the occurrences of C. corniculata in the Late Hemphillian-earliest Blancan NALMA and C. timida in the Irvingtonian-Rancholabrean NALMA.

BASAL FOSSIL DEIROCHELYINES

This basal deirochelyine clade, including Chrysemys antiqua , C. corniculata and C. timida , may be due to a more ‘unspecialized’ morphology. This basal morphology can make placing taxa near the root less conclusive. This is noted in the placement of C. timida as a basal emydine by Vlachos (2018), although he noted the fragmentary nature of the type specimen [YPM (PU) 10853] makes its current placement more ambiguous and more information is needed. Based on the dataset, and Clemmys guttata as a basal member of the sister Emydinae , it is not unexpected that basal OTUs in the present phylogenetic analysis are located basally near its root. Chrysemys also has a rather unspecialized (or more generalized) morphology, particularly in comparison to deirochelyines such as Trachemys and Graptemys . Taxa that tend to possess more ‘extreme’ morphologies, such as Malaclemys , Graptemys and Trachemys , tend to be recovered as monophyletic in more derived clades. While several fossil taxa previously referred to Chrysemys are also recovered at the base of the deirochelyine tree, these fossil taxa do not form a clade with modern Chrysemys and their inclusion in Chrysemys would currently make the genus polyphyletic or paraphyletic, depending on which consensus tree was utilized. If the 50% majority rule consensus tree is correct, C. antiqua , C. corniculata and C. timida are part of an extinct lineage of basal deirochelyines that is sister to the rest of the family ( Fig. 9 ). However, the monophyly of this group is lost in the strict consensus tree ( Fig. 8 ). This would suggest basal taxa with more generalized or ‘simpler’ morphologies group together at the base of Deirochelyinae even though they are not, or may not be, part of a single genus or lineage. Although they have been considered members of Chrysemys before, it may be more likely that they represent unique fossil lineages with basal morphologies among deirochelyines. They have been considered members of Chrysemys as they possess shared characteristics of the genus (see diagnosis above for referral of C. corniculata to Chrysemys ), but this may be due to the basal morphology of modern Chrysemys among deirochelyines, causing these taxa to converge near or at the base of Deirochelyinae . This may also have implications for our understanding of not only the base of the subfamily, but the base of the family Emydidae as well. While this suggests that these taxa should potentially be moved to a new genus, characteristics discussed in the diagnosis show multiple similarities with the current understanding of Chrysemys . Furthermore, the addition of other fossil deirochelyines could affect the morphological relationships among OTUs. Therefore, these taxa are still considered Chrysemys until other fossil taxa and more data are included. Regardless, there is still a significant ghost lineage present after C. antiqua of approximately 25 Myr, and even that is to a fossil species of ‘ Chrysemys ’ (‘ C.’ carri ) from the latest Clarendonian. Fragmentary fossils have been found in this gap [e.g. C. isoni and Trachemys (indeterminate species), Weems & George, 2013], but nothing more complete has been recovered to help provide us with more understanding of what is happening at this time. More fossils of deirochelyines within this gap, particularly those that are more complete, are vital for understanding the evolution of the subfamily, particularly the evolution of more basal forms, like C. antiqua , to more derived forms.

CHRYSEMYS THROUGH TIME

As the genus is currently understood, Chrysemys represents the oldest deirochelyine genus, originating in at least the Middle Chadronian NALMA (35.7–34.7 Mya) with C. antiqua . In this I consider C. antiqua , C. timida and C. corniculata to represent Chrysemys with the knowledge that these may eventually be found to be part of a different genus or genera. However, more data may also provide more resolution and show these taxa to be part of a monophyletic clade with modern Chrysemys . Regardless, they agree with the characteristics considered diagnostic of Chrysemys , and will remain in Chrysemys until further study potentially determines whether these species lie elsewhere. Our earliest evidence of Chrysemys , therefore, is from the Chadronian, as the climate was cooling from Eocene greenhouse conditions (e.g. Graham, 1999; Woodburne, 2004; Katz et al., 2008; Zachos et al., 2008; Hansen et al., 2013). Current fossil evidence shows that Chrysemys and deirochelyines show up as the climate was changing and after a significant turnover from an Eocene turtle fauna represented by different taxa such as trionychids to others such as testudinoids (e.g. Hutchison, 1992, 1996). While testudinoids diversify significantly in North America during this time (e.g. Lichtig et al., 2021), emydids do not appear (based on current fossil evidence) to diversify as much until closer to the Hemphillian NALMA, after the Mid-Miocene Climatic Optimum (MMCO). During this time, while some taxa maintain a more basal deirochelyine morphology (e.g. C. antiqua ), others show a more derived morphology (e.g. Pseudograptemys Hutchison, 1996 ), with more extreme and pronounced notching along the carapacial rim. It is also noted that although Pseudograptemys has previously been considered one of the earliest emydids (e.g. Hutchison, 1996), Joyce et al. (2013) more recently found it to lie outside of the Emydidae phylogenetically, and Vlachos (2018) considered it a geoemydid synonymous with Echmatemys latiƲertebralis Cope, 1877. However, Pseudograptemys is distinct from E. latiƲertebralis based on several features, including: the inguinal buttress suture confined to the posterior of costal 5 (near the costal 5–6 suture) while it spans the costal 5–6 suture in Echmatemys . The geoemydid E. latiƲertebralis is currently undergoing further study and this potential synonymization will be re-evaluated there (see also Lichtig et al., 2021). Chrysemys first shows up near the western portion of its modern range ( Fig. 10I ) in Pennington County, South Dakota, USA ( Clark, 1937; Hutchison, 1996). It is also noted that several fossil specimens have been referred to C. antiqua from the Orellan and Whitneyan of South Dakota ( Hutchison, 1996), suggesting this taxon persisted for several million years. The next fossil taxon referred to Chrysemys is C. isoni from the Hemingfordian of Virginia, when grasslands became more prevalent in North America. Chrysemys corniculata shows up in the Late Hemphillian-Early Blancan NALMA when fossil emydids become more common and temperatures continue to cool after the MMCO. Chrysemys timida , from the Pleistocene of Nebraska, is also closer to the western portion of the modern range of Chrysemys and closer to the fossil localities of C. antiqua . Global temperatures would have been fluctuating during this time and temperatures often would have been cooler in this region compared to today with multiple glacial cycles (e.g. Augustin et al., 2004).

While the fossil record is incomplete, based on our current data the range of C. picta has expanded over the past 16 Myr. Fossils currently referred to the modern species C. picta have been recovered from the Barstovian NALMA through the Saintaugustinean NALMA ( Fig. 10 A-G, see Supporting Information (Data S1 -S4) and references therein). Early C. picta fossils have thus far been found in the mid-western United States, near the southern portion of the current range of C. p. bellii, namely Nebraska, with a southern expansion to Kansas by the Clarendonian NALMA ( Fig. 10A ). The Hemphillian NALMA sees an eastern expansion to Indiana ( Fig. 10B ), followed by a significant southern expansion of the genus in the Blancan NALMA ( Fig. 10C ). Although C. picta does extend into western Texas near the Rio Grande and Pecos rivers as C. p. bellii and the eastern portion as C. p. dorsalis , it currently does not reach nearer the interior of the state. Chrysemys picta is also not currently native to Florida. Nevertheless, the genus has been identified from the Blancan NALMA of both Scurry County, Texas ( Rogers, 1976) and Alachua County, Florida ( Bourque et al., 2007). The Irvingtonian NALMA has fossils recovered from Alleghany County, Maryland, showing a smaller expansion east towards its modern distribution ( Fig. 10D ). The Rancholabrean NALMA sees a significant increase in the number of C. picta fossils identified and their geographic range ( Fig. 10E ). Other North American turtle groups (e.g. Chelydridae , Trionychidae ) have also been found to have increased their geographic range during the Rancholabrean (e.g. Moscato & Jasinski, 2016) or potentially recolonized areas as temperatures became warmer (e.g. Jasinski, 2013b). These Chrysemys fossils show some slight increases in their previous range, particularly to the south in places like Bartow County, Georgia ( Holman, 1967) and Colbert County, Alabama ( Holman & Andrews, 1994) and farther east in Smyth/ Washington County, Virginia ( Holman & Andrews, 1994). Fossils from Oklahoma ( Smith & Cifelli, 2000) suggest a reduction in their range after the Blancan NALMA, although not yet reduced to the extent of their modern range ( Fig. 10 E-H). The Santarosean NALMA also has numerous C. picta fossils and most are from within the current range, except for farther range extension to the south-east in South Carolina (see Bentley & Knight, 1998), and farther north into Michigan (e.g. Holman, 1992) ( Fig. 10F ). There also appears to be a significant range extension at this time into Nova Scotia (see Holman & Clouthier, 1995). Additionally, there is further reduction in their range in the southern mid-west portion of their range, showing more reduction towards their current range. Specimens of C. picta from the Saintaugustinean NALMA show further expansion north into southern Ontario ( Fig. 10G ). Without more fossils we currently are uncertain when Chrysemys expanded to several other areas of its current range, including those regions to the north-west, such as Washington, the north such as Saskatchewan and Manitoba, and farther south, such as Louisiana ( Fig. 10H ). While some of this range expansion through time may change as more fossils are found and identified, it is also plausible that the modern species C. picta has not been static for 16 Myr, and is likely that some of these fossils, particularly older ones, will be re-identified to different species in time. Nevertheless, current data shows the species making larger range expansions during the Hemphillian and Santarosean, although the Santarosean expansion may be largely due to more material being available (i.e. Pull of the Recent).

Chrysemys picta is currently the most widely distributed turtle in North America (e.g. Ernst & Lovich, 2009) suggesting it can deal with a wide range of conditions physiologically. Even so, Chrysemys does not range significantly far south (approximately 28°N), and particularly tends to avoid more humid climates like those in Florida. It is more often found toward the mid-west and north-east of the United States and is known to prefer slow-moving shallow freshwater habitats with soft bottoms, plentiful aquatic vegetation and basking sites (e.g. Ernst & Lovich, 2009). These habitats often come in the form of lakes, ponds, swamps, marshes and slow-moving rivers. Females normally begin to ovulate in the spring and often lay multiple clutches of eggs per year, and average temperatures during incubation, particularly during the middle third, are especially important as they influence the sex of the offspring. Temperature-dependent sex determination is well documented in turtles and other vertebrates, such as some fish and other reptiles (e.g. Bull & Vogt, 1979; Ewert et al., 1994; Valenzuela & Lance, 2004; Mitchell & Janzen, 2010). Typically, temperatures of 23–27 °C produce male offspring in Chrysemys , whereas temperatures outside this range result in females (e.g. Ernst & Lovich, 2009). Therefore, locations without the proper temperatures can result in monotypic sexes of C. picta offspring, and nonviable populations moving forward. Fossil occurrences of C. picta can be used to track temperatures and general habitats through time in North America, at least for particular times of the year. Fossil localities with C. picta can be inferred to have been slow-moving, shallow freshwater habitats with optimum temperatures at least lying near those that could produce male offspring, with variation resulting in female offspring. Although, again, C. picta can deal with more variable temperatures, it could result in smaller windows for reproduction and in the possibility of fewer clutches annually (e.g. Moll, 1973; Nussbaum et al., 1983; Ernst & Lovich, 2009). Additionally, Chrysemys has not been found farther south in locations like Florida where temperatures are too warm to produce consistent male offspring. However, its occurrences farther south during the Blancan NALMA suggests cooler temperatures during this time, and cooling during this time period has been found in benthic δ 18 O records (e.g. Lisiecki & Raymo, 2005).

Although Chrysemys is not found in lower latitudes (approximately 28°N) and warmer temperatures today, testudinoids, and particularly emydids, appear to have undergone geographic and adaptive radiations at particularly warmer periods in the past, such as during the Early Eocene and Middle Miocene. Cooler temperatures and changing climates following these periods would lead to range reduction in these turtles. Even though Chrysemys does not deal with higher temperatures as efficiently as some turtle taxa, their reproductive strategies are still contingent on temperatures not becoming too cool, implying their range would also be reduced due to cooler temperatures. Range reduction could also lead to less habitable ecosystems, effectively leading to isolation of allopatric populations. This vicariance would then lead to speciation and the diversification seen in the fossil record of emydids. Biogeographical vicariance provides a possible explanation to the origination of the Emydidae and a time of high speciation in the Late Miocene into the Pliocene. More fossils and a better understanding of fossil taxa already described are needed to further evaluate this hypothesis.

PALAEOBIOLOGY AND PALAEOENVIRONMENTAL INTERPRETATIONS

Unsurprisingly, C. corniculata is interpreted as living in a similar environment to that preferred by modern Chrysemys and with a similar natural history ( Fig. 11 ). As the GFS is interpreted as a pond deposit with multiple sediment-filled sinkholes (e.g. Shunk et al., 2006; J. Whitelaw et al., 2008; M. Whitelaw et al., 2011; Zobaa et al., 2011)with a woodland or woodland savannah environment in the vicinity, and discrete wet and dry seasons ( Ochoa et al., 2012, 2016), it agrees with some of the normally preferred environments of modern Chrysemys (e.g. Ernst & Lovich, 2009). The lower triturating surface of C. corniculata is slightly wider than in C. picta , but not as wide as those in some other emydids such as Pseudemys . As the lower jaws of turtles can be indicative of diet (e.g. Jasinski et al., 2018), a wider triturating surface could indicate durophagy or more surface area for grinding foods such as plant material. The diets of Chrysemys also often vary throughout ontogeny, so comparing the jaws and triturating surfaces of different ontogenetically-aged individuals would be useful for determining changes in diet as C. corniculata age (i.e. grow older), particularly as modern Chrysemys is known to alter its diet through ontogeny (e.g. Ernst & Lovich, 2009). More cranial material, of varying sizes, is needed before this can be determined, but for now it is likely that C. corniculata had a similar diet to C. picta , potentially with a slightly higher percentage of plant material. Freshwater ecosystems frequently support multiple turtle species living together (e.g. Holman, 1992; Smith et al., 2006; Jasinski et al., 2011; Sullivan et al., 2013; Anthonysamy et al., 2014; Ferri et al., 2020), often with habitat partitioning. Chrysemys picta is usually the most abundant turtle in the suitable habitats within its range (e.g. Ernst & Lovich, 2009). However, it has been noted that when C. picta and Trachemys scripta Thunberg in Schoepff, 1792 occur in the same localities, C. picta occurs in definitively smaller numbers (e.g. Bodie et al., 2000; Lindeman & Scott, 2001; Dreslik & Phillips, 2005; Dreslik et al., 2005; Jaeger & Cobb, 2012). This also seems to be the case at the GFS, where Trachemys specimens (i.e. T. haugrudi ) currently far outnumber Chrysemys specimens (i.e. C. corniculata ). Additionally, when Chrysemys and Trachemys co-occur, there appears to be some habitat partitioning between the genera, with Chrysemys tending to prefer shallower habitats closer to shore while Trachemys select more open water areas (e.g. Ernst & Lovich, 2009; Jaeger & Cobb, 2012). It is likely this was also occurring at the GFS and other fossil localities where these taxa co-occurred. Further work is being done on the spatial relationships of fossils at the GFS, and the concentrations of particular taxa in different locations may have additional implications on hypothesizes for habitat partitioning between different species.

The raised anteromedial portion of the carapace, including the nuchal, is distinct between C. corniculata and C. picta . This, combined with the more extreme morphology of the anterior of the carapace in the former ( Fig. 7 ), are potentially important for sexual display and dimorphism. Although modern Chrysemys are not known to use head movements in courtship (e.g. Ernst & Lovich, 2009), this behaviour has been documented in other turtles (e.g. Liu et al., 2013). Differences in courtship behaviour between C. corniculata and C. picta would have been one way to avoid interspecific crossbreeding. The more extreme nature of the anterior carapace of C. corniculata may have also helped the species avoid crossbreeding. Moldowan et al. (2020b) discussed the anteromedial projections of the carapace (weaponized anterior shell) of C. picta with older males using coercion for breeding purposes. They mentioned this also changed through ontogeny, although C. corniculata possesses nuchal horns regardless of size. It is more likely that these were sexual display or species recognition features rather than as features for coercion in ontogenetically older individuals. This is also more likely as they are present in all individuals of C. corniculata . Additionally, as the anteromedial projections of the nuchal seem to come in different sizes, it is possible that, while their presence is not dimorphic, their size may indicate sexual dimorphism, with males having larger and more extreme nuchal horns ( Fig. 7A–D ).

Figures

Figure 7. Chrysemys carapace comparisons, focusing on the morphology of the anterior edge. A-D, Chrysemys corniculata sp. nov. carapace in dorsal view: A, C. corniculata sp. nov., holotype carapace (ETMNH–12491); B, C. corniculata sp. nov. paratype carapace (ETMNH–20609); C, C. corniculata sp. nov., paratype carapace (ETMNH-20544); D, C. corniculata sp. nov., paratype carapace (ETMNH-3559), with arrow pointing to nuchal horns that occupy the anteromedial portion of marginal 1. E-I, C. picta anterior carapace morphology: E, female C. picta nuchal (UF 48670) in dorsal view; F, male C. picta nuchal (UF 49963) in dorsal view; G, male C. picta anterior carapace (ETVP 9683) in dorsal view with arrow pointing to the marginals 1 portion of the nuchal where the nuchal horns are present as an inconspicuous anterior projection; H, male C. picta anterior carapace with an arrow pointing to the lighter-coloured anterior projections of marginal 1 that would lie above the nuchal horns; I, female C. picta anterior carapace showing no anterior projections. Scale bars: A-D, 5 cm; E-G, 1 cm; H-I, not to scale.

Figure 8. Strict consensus tree of the phylogenetic relationships of deirochelyine emydids supported by this study based on morphological data constrained by a molecular backbone. See text and Supporting Information (Data S1–S4) for details. Intergeneric relationships of modern taxa constrained based on molecular phylogenetic analysis presented by Spinks et al. (2016: fig. 5). Analysis includes the emydine Clemmys guttata as the outgroup. The tree presented represents the strict consensus tree of deirochelyine relationships found in the phylogenetic analysis of 246 characters within 35 deirochelyines (ingroup taxa). Thickened bars indicate age ranges for fossil taxa. † refers to a fossil or extinct taxon.Tree length equals 881 steps, consistency index equals 0.413, retention index equals 0.515. Numbers beside branches provide the Bremer supports (first) and bootstrap values (second). Numbers are shown for branches with bootstrap values greater than 50.

Figure 9. Fifty percent majority rule consensus tree of the phylogenetic relationships of deirochelyine emydids supported by this study based on morphological data constrained by a molecular backbone. See text and Supporting Information (Data S1–S4) for details. Intergeneric relationships of modern taxa constrained based on molecular phylogenetic analysis presented by Spinks et al. (2016: fig. 5). Analysis includes the emydine Clemmys guttata as the outgroup. The tree presented represents the 50% majority rule consensus tree of deirochelyine relationships found in the phylogenetic analysis of 246 characters within 35 deirochelyines (ingroup taxa). Thickened bars indicate age ranges for fossil taxa. † refers to a fossil or extinct taxon. Tree length equals 881 steps, consistency index equals 0.413, retention index equals 0.515. Numbers beside branches provide the Bremer supports (first) and the bootstrap values (second). Numbers are shown for branches with bootstrap values greater than 50.

Figure 10. Fossil occurrencesof C.picta through time, with the modern range shown indarker grey(see/compare with Figure 1). A, movement of C. picta based on fossil occurrences from the Barstovian to the Clarendonian NALMAs. B, movement of C. picta based on fossil occurrences from the Clarendonian to the Hemphillian NALMAs. C, movement of C. picta based on fossil occurrences from the Hemphillian to the Blancan NALMAs. D, movement of C. picta based on fossil occurrences from the Blancan to the Irvingtonian NALMAs. E, movement of C. picta based on fossil occurrences from the Irvingtonian to the Rancholabrean NALMAs. F, movement of C. picta based on fossil occurrences from the Rancholabrean to the Santarosean NALMAs. G, movement of C. picta based on fossil occurrences from the Santarosean to the Saintaugustinean NALMAs. H, extensions to the range of C. picta currently temporally unknown based on lack of fossil evidence. I, range extensions of the genus Chrysemys based on fossil species unique from C. picta, with numbered arrows showing these extensions temporally as beginning with 1 followed by 2 and finally 3. Note: Alachua County, Florida, is shown in (I) representing the fossil localities of ‘C.’ aeilliamsi and a potential new Chrysemys species, although it is not considered part of the expansion as ‘C.’ aeilliamsi probably does not represent Chrysemys (see text) and the potential novel species of Chrysemys has not been thoroughly studied (see Bourque et al., 2007).

Figure 11. Life reconstruction of Chrysemys corniculata sp. nov. at the Gray Fossil Site in north-eastern Tennessee during the latest Miocene-earliest Pliocene by Sergey Krasovskiy.