Taxonomic revision of Chinemys pani (Testudines: Geoemydidae) from the Pleistocene of Taiwan and its implications of conservation paleobiology

Proper taxonomic identification is critical to our understanding of biodiversity and the underlying evolutionary history. Here we re‐examine the cast of the holotype of Chinemys pani, a geoemydid turtle from the Pleistocene of Taiwan; the actual specimen was not curated appropriately after the original publication and was most likely lost. Our results provide substantial evidence to show that Chinemys pani should be identified as Mauremys reevesii. The replica, though not ideal, preserves various morphological features that allow reliable taxonomic identification of Mauremys reevesii, including the presence of three longitudinal keels on the carapace, the second to sixth neural bones anteriorly short‐sided, and the lack of a movable plastral hinge. In addition, we also confirm that the original diagnostic features that established the new taxon: Chinemys pani – are polymorphic characters of Mauremys reevesii. Our taxonomic revision of a Pleistocene geoemydid turtle from Taiwan confirms the existence of the Pleistocene Mauremys reevesii. More importantly, this study offers new insights into the origin of modern biodiversity in Taiwan and gives a straightforward example of how fossils can be applied to conservation policies.


| INTRODUCTION
Chinemys pani (Geoemydidae, Testudines) is the first described and named fossil turtle from Taiwan (Tao, 1985). Interestingly, without re-examining the actual specimen or giving any morphological evidence, numerous publications have designated Chinemys pani a junior synonym of Mauremys reevesii (e.g., Iverson, 2022;Turtle Taxonomy Working Group, 2021;Zhao & Adler, 1993). Further complicating this taxonomic issue is the loss of the actual fossil (holotype of Chinemys pani). Our intense search for the holotype in the private (originally owned by Mr. Chang-Wu Pan, a renowned but deceased fossil collector in Taiwan) and public museum collections only discovered the plaster casts of the holotype (both in the collections of Chang Jung Senior High School in Tainan and the National Taiwan Museum in Taipei). The plaster cast of the holotype then provides the exclusive opportunity to resolve the taxonomic conundrum.
Mauremys reevesii is a moderate-sized geoemydid with broad distribution across east Asia but is considered an introduced species into southeast Asia (Turtle Taxonomy Working Group et al., 2021). On the other hand, Mauremys reevesii had long been regarded as a native species in Taiwan, but this view has been challenged by molecular studies (Fong & Chen, 2010;Lee et al., 2019;Suzuki et al., 2011). Thus, the arrival of Mauremys reevesii in the natural history of Taiwan requires further examination based on the fossil record. Our re-examination of the taxonomic identity of this turtle fossil from the Pleistocene of Taiwan (originally identified as an endemic species but later assigned to be extant Mauremys reevesii without any solid arguments) then has profound implications for conservation paleobiology and represents a straightforward example of how fossils can be applied to conservation policy-decision.

| Size and ontogenetic age
No fontanelles on the shell indicate the well-ossified condition. In addition, given the relatively small size (maximum preserved anteroposterior length: 112 mm length; in comparison with an extant Mauremys reevesii: 221 mm, NTUM-VP 2110121), the fossil individual likely belongs to a sub-adult. The original publication suggested that the fossil represents a male individual due to the presence of a shallow concavity. Yet, we consider it problematic as the specimen may have experienced postmortem deformation, and the reliability of this feature to identify the gender of Mauremys reevesii remains uncertain.

| Locality and geological setting
The fossil was originally claimed to be found in the Chiting Formation of Zuojhen (also known as Chochen or; Figure 1), Tainan (Tao, 1985). However, Hu and Tao (1993) corrected the locality, confirming that the actual locality is the sea bottom of the Taiwan Strait (between Taiwan and the Penghu Islands, Figure 1). Abundant vertebrate fossils have been dredged up by fishermen, but the stratigraphic data or its geological age remains uncertain. The sea bottom of the Taiwan Strait has yielded marine and terrestrial fossil vertebrates, reflecting the sea-level change during the Pleistocene, indicating this area includes terrestrial and marine deposits (e.g., Gao, 1982

| Morphological description and taxonomic remarks
The general morphology has been described in the original publication (Tao, 1985). To avoid redundancy, we describe and mark the key morphological features that allow further understanding and interpretation of this specimen. The relative length between the anterior margin of the neural II and the posterior end of the neural VII and the maximum body width is 0.75. This ratio was considered to be a differential trait from other geoemydid species by Tao (1985), but the ratio is variable and in the wide range of intraspecific variation (0.71 in NTUM-VP 2110121 and 0.64 in another M. reevesii in the 1985 article). Similarly, the ratio of the maximum body width and the length from the plastron to the anterior margin of the plastron fork is 1.64, also included in the range of extant Mauremys reevesii. Besides, the shape of bony plates (i.e., wide neurals IV and VI and narrow neural VII) is variable, precluding the use of differentiating species. Asymmetry of the neurals I and II, which was not mentioned by Tao (1985), likely results from the posterior margin of the neural I extending more posteriorly and the missing of the anterior margin of the neural II. This asymmetry is similar to the heptagonal neural VI and pentagonal neural V observed in NTUM-VP 2110121. The asymmetry of neural bones in geoemydid turtles appears to be a common feature but still needs to be assessed thoroughly.
Most importantly, the original publication (Tao, 1985) recognized the morphological differences of Chinemys pani from other species of extant Mauremys, mainly based on the relative lengths of the abdominal, femoral, and anal scutes on the plastron (abdominal scute shorter than femoral one and the femoral longer than the anal scute). Updated studies (e.g., Lovich & Ernst, 1989) show the polymorphic states exist in Mauremys reevesii, rejecting the recognition of Chinemys pani. Our re-examination of the casts (the actual fossil is lost) corroborates its Mauremys affinity: a joint bony bridge between the carapace and plastron and the anteriorly short-sided margins on the second to sixth neural bones (Garbin et al., 2018). In addition, the joint bony bridge instead of a plastral hinge differs from the geoemydid Cuora, Cyclemys, and Notochelys; short-sided margins on the second to sixth neural bones anteriorly distinguish from the geoemydid Heosemys, Pangshura, Melanochelys, Geoemyda, and Rhinoclemmys; and lack of serration on the posterior peripherals differentiates from the geoemydid Heosemys and Geoemyda. Further, the presence of three longitudinal keels on the carapace is a critical feature shared by Mauremys reevesii, M. nigricans, and M. annamensis but more prominent in M. reevesii. Given the overall morphological combination, its taxonomic assignment should belong to Mauremys reevesii. Our detailed re-examination of the plaster casts (after our intense search in the museum collections, the whereabouts of the actual fossil remain unknown and likely lost) shows that the so-called Chinemys pani is taxonomically invalid and indeed represents a Pleistocene fossil record of extant Mauremys reevesii. Key diagnostic features of Mauremys reevesii include (a) a bony bridge connecting the carapace and plastron, (b) the second to sixth neural bones being anteriorly short-sided, and (c) three longitudinal keels on the carapace (Figure 2). Our re-identification of a Pleistocene Mauremys reevesii represents the oldest record from Taiwan and, in turn, a straightforward example of how fossils can be applied to conservation policies, that is, conservation paleobiology. The emerging field of conservation paleobiology provides novel perspectives on conserving modern biodiversity and responding to the probable sixth mass extinction (e.g., Barnosky et al., 2017;Dietl et al., 2015). Our taxonomic revision of Chinemys pani to Mauremys reevesii from the Pleistocene of Taiwan demonstrates the potential of developing vertebrate paleontology, a poorly-represented research field in Taiwan (see Tsai & Mayr, 2021), for applications of conservation paleobiology (see Tsai et al., 2014 for discovering a paleo-breeding ground of gray whales in Taiwan that also offers a decent example).
Whether the extant Mauremys reevesii is a native or human-introduced turtle in Taiwan remains controversial (Lee et al., 2019). This species in Taiwan was first documented by Horikawa (1934). The Conservation Act of Taiwan, first established in the 1980s, considered it a threatened species; the population of extant M. reevesii and its existence in Taiwan started to attract some research attention. In 2008, after some follow-up studies (Chen & Lue, 2010), the updated Wildlife Conservation Act in Taiwan re-evaluated the status of M. reevesii from conservation-deserving wildlife (Class III) to critically endangered (Class I, the highest level in Taiwan). Interestingly, the degree of conservation was degraded to rare and valuable (Class II) in 2019 owing to the concern that extant M. reevesii in Taiwan may be a result of human introduction rather than a native population (Fong & Chen, 2010;Lee et al., 2019). A similar issue also occurs with the extant population of M. reevesii in Japan; the molecular approach that estimated the genetic divergence among populations in East Asia (China, Japan, Korea, and Taiwan) suggested both populations of Japan and Taiwan should be recent introduction, that is, originating from human release to the wild environment. Besides, no fossils but only archeological remains of Mauremys reevesii (Takahashi, 2015;Takahashi et al., 2019) further support the conclusion that M. reevesii in Japan should be considered a humanintroduced species. Here, our study, for the first time, confirms the Pleistocene existence of extant M. reevesii in Taiwan (the fossil occurrence has been suggested by a 1988 article (Tao, 1988), but the material is too incomplete to confirm its taxonomic assignment to the species level), suggesting that M. reevesii belonged to a native species in Taiwan and providing solid evidence to support the ancient arrival of this moderate-sized freshwater geoemydid turtle in Taiwan.
Our results then offer a definitive example for conservation paleobiology -Mauremys reevesii in Taiwan should be considered a native species. Although the current population of M. reevesii in Taiwan seems to be human-introduced turtles from the pet trade or other human activities (also see Bu et al., 2021 for the issues that Mauremys reevesii faces in China, including habitat destruction, poaching to sell as Chinese medicine, pets, and food, etc.), our discovery offers a novel perspective to consider whether it is likely for such a secondary-invaded population to rebuild its natural habitat that restores the vanished ecological role of Pleistocene M. reevesii in Taiwan. Our pioneering practice of conservation paleobiology in Taiwan then sheds new light on the decisionmaking of conservation policy. With our paleontological study and the molecular results, the current population of Mauremys reevesii may not be the native lineage in Taiwan, and the Pleistocene population went extinct at some geological or archeological timings. If our interpretation is correct, future conservation directions should include more in-depth research on the paleontological or archeological remains to pin down the extinction event and timing and on their ecological role in the paleontological and modern ecosystems to further discuss the need of conserving the current population. In addition, the evolutionary history of M. reevesii in East Asia remains poorly explored; the fossil record is scarce, including a partial shell and limb bones from the Early Pleistocene of Anhwei (also known as Anhui), China (Chow, 1955) and a partial carapace and an almost complete plastron from the Middle Pleistocene of Choukoutien, China (Bien, 1934). Our Pleistocene M. reevesii from Taiwan with implications for conservation paleobiology should then draw more research and attention to revealing the poorly-understood lineage of geoemydid turtles.