First occurrence of Duboisia (Bovidae, Artiodactyla, Mammalia) from Thailand

The first fossil record of Duboisia (Boselaphini, Bovidae) from Thailand confirms that this genus is no longer endemic to Java, Indonesia. The new fossil material is a calvarium with horn cores (older than the Middle Pleistocene) collected from a sandpit at Tha Chang, Nakhon Ratchasima Province, north-eastern Thailand. The present specimen is provisionally allocated to a species of Duboisia aff. D. santeng, which has weaker precornual ridges and anterior keels on the horn cores than D. santeng from Early and Middle Pleistocene deposits of Java, but these species share basic characteristics of horn cores as follows: the lower half inclined backwards; the upper half curved upwards; cross section rounded triangular, antero-posteriorly compressed, and with medial and lateral keels. Morphological similarities between D. aff. santeng and D. santeng support a strong faunal interchange between continental South East Asia and Java before the Middle Pleistocene, and suggest that the genus Duboisia diverged from the other genera of Boselaphini in the “Siva-Malayan” region.


Introduction
South East Asia is geographically divided from the Indian subcontinent by the Himalayan-Tibetan Plateau, the Indo-Burma Range, and the Ganges and Brahmaputra rivers. These barriers form a zoogeographical boundary for terrestrial mammals between the Indo-Chinese and Indian subregions (Corbet and Hill, 1992). The Neogene fauna of continental South East Asia is similar to that of the Siwaliks, in the Indo-Pakistan area (e.g. Stamp, 1922;Colbert, 1938), but recent studies based on mammalian fossils from the Irrawaddy beds (Middle Miocene to Early Pleistocene) of Myanmar have demonstrated a faunal difference between the Siwaliks and Irrawaddy beds, which had increased since the Late Miocene or Early Pliocene due to forming zoogeographical barriers (Nishioka et al., 2015(Nishioka et al., , 2018bTakai et al., 2016).
Mammalian endemism in South East Asia had increased during the Plio-Pleistocene (e.g. Van den Bergh et al., 2001;Van der Geer et al., 2010). The latest Pliocene and Early Pleistocene Satir fauna from Java, Indonesia, is impoverished and unbalanced and includes mammals of "Siva-Malayan" origin, such as Hexaprotodon sivajavanicus (Hippopotamidae, Artiodactyla) and Sinomastodon bumiajuensis (Proboscidea) (Sondaar, 1984). These mammals have been traditionally believed as endemic species to Java (Van der Maarel, 1932;Hooijer, 1950), but to what degree island species differ from the original continental species from the Siwaliks and the Irrawaddy beds is insufficiently studied (Van der Geer et al., 2010).
The "Stegodon-Homo erectus fauna" from the Early-Middle Pleistocene of Java (e.g. Trinil HK and Sangiran dome) is characterized by a relatively low degree of endemism and is composed of Siwalik species such as Pachycrocuta brevirostris (Carnivora) and Javanese species such as Stegodon trigonocephalus (Proboscidea) derived from mammals of the "Siva-Malayan" origin ( Van der Geer et al., 2010). Duboisia santeng is an endemic species of Boselaphini (Bovidae, Artiodactyla) in the Early-Middle Pleistocene fauna of Java, which represents a somewhat isolated environment (Hooijer, 1958;Aimi and Aziz, 1985;Van den Bergh et al., 2001;Rozzi et al., 2013). D. santeng is phylogenetically related to living Indian boselaphins, Boselaphus tragocamelus and Tetracerus quadricornis (Stremme, 1911;Pilgrim, 1939;Hooijer, 1958). However, when a putative ancestor of D. santeng colonized Java is still debatable (Rozzi et al., 2013), owing to the absence of certain fossil evidence on continental South East Asia to show a relationship between Javanese D. santeng and Indian Boselaphini.
Out of Java, Moigne et al. (2016) reported three isolated teeth belonging to the genus Duboisia from the Upper Pliocene Masol Formation of the Siwaliks, India, although they did not mention detail morphological comparisons on the basis of apomorphic characteristics of each taxon. The presence of Duboisia in continental South East Asia was mentioned by Hooijer (1962), based on some dental and postcranial remains from Quaternary cave deposits at Tambun, Peninsular Malaysia. However, we are not sure if this identification is valid because of the fragmentary nature of the remains. Furthermore, all specimens of D. santeng in the Hooijer collection are missing (Lim, 2013). A partial cranium of Duboisia was recently found from central Myanmar (but without geological context), which gives preliminarily evidence suggesting the existence of the genus on continental South East Asia (Nishioka et al., 2018b). In the present study, we describe the first specimen of Duboisia from Thailand. The specimen is a skull with complete horn cores that are useful for taxonomic comparisons of Duboisia remains between continental and island specimens from South East Asia.

Geological settings
The Tha Chang sandpits (15 • 05 N, 102 • 20 E) are located 22 km east of Nakhon Ratchasima (or Khorat) City, northeastern Thailand (Fig. 1). More than 10 sandpits have been mined since 1985 and have exposed Neogene and Quaternary fluvial deposits along the Mun River, a tributary of the Mekong River. One of the fossil localities, Tha Chang sandpit no. 8 (or Somsak sandpit), is well-known for its yield of Miocene orangutans, Khoratpithecus piriyai (Chaimanee et al., 2004). The sediments at sandpit no. 8 are an exposed outcrop of about 30 m depth from the ground to the bottom when the sandpit was first mined (Shinji Nagaoka, personal communication, 2002; see also Thasod et al., 2011).
Sediments at sandpit no. 8 have been divided into two units: a lower unit of organic greyish sand and gravel, and an upper unit of yellowish sand (Chaimanee et al., 2006(Chaimanee et al., , 2007. The basal part of the lower unit overlies on the Mahasarakham Formation including the evaporitic layer around 40 or 50 m depth (Chaimanee et al., 2006). The lower unit is formed by grey conglomeratic sandstone with some thin silt and clay layers. The sandstone shows clear cross-bedding, suggesting channel deposits, and contains abundant fossils of tree trunks, leaves and vertebrate animals. The upper part of the lower unit is characterized by the inclusion of peat (ca. 0.5 m thick) and gravel layers (ca. 0.5-1 m thick) around 10 m depth from the ground. Plant fossils have been found in all layers.
The upper oxidized unit starts from about 8 m depth from the ground, based on the stratigraphic column by Chaimanee et al. (2006). Some tektites have been collected from the lowermost part of the upper unit (ca. 8 m depth), which originate from the widespread tektite debris field at 0.8 Ma in northeastern Thailand (Howard et al., 2003). Haines et al. (2004) estimated that the sediments of the Tha Chang sandpits are dated to Quaternary, because those tektites were also collected from the lower unit.
Mammalian fossils have been collected primarily from the lower unit. An isolated molar of the proboscideans Stegolophodon cf. stegodontoides and a skull of the anthracothere Merycopotamus thachangensis were collected in situ from the silt (ca. 13.5 m depth) and sandstone (ca. 18 m depth) layers of sandpit no. 8, respectively (Hanta et al., 2008;Thasod et al., 2011). Geological information for other mammalian fossils from the Tha Chang sandpits is unavailable, but most of them were probably from the lower unit because the fossils are covered by greyish sandstone with pyrite (iron sulfide).

Material
The examined fossil specimen (PPN 01-000109) that is housed at the Sukhothai Elephant World Museum, Thai- land, was originally found in a private collection of Kamol Chaivanich in Bangkok. According to him, the specimen was collected from a sandpit in Tha Chang before 2007 when sandpit no. 8 was being mined. The fossil surface was originally covered by greyish sandstone with many pebble-or granule-sized gravel stones under 10 mm in diameter (Fig. 2). Gravel layers at Tha Chang sandpit no. 8 lie between 7 and 14 m depth from the ground, which includes both upper and lower units (Chaimanee et al., 2006). A combination of accompanying both greyish sandstone with pyrite and many pebbles on the fossil surface indicates that this fossil might have been retrieved from the lower unit, similar to mammalian fossils from the Miocene rather than the Pleistocene. PPN 01-000109 was compared with two skulls of Duboisia santeng from the Pleistocene of Java. One specimen (MGB.SA 290779) is a calvarium with right and left horn cores recovered from the Early-Middle Pleistocene deposits at Pucung, Sangiran Dome area (Aimi and Aziz, 1985). The other specimen (MGB.Dbs) on display at the same institute is a neurocranium with horn cores probably from Sangiran. Furthermore, we also referred to pictures of D. santeng skulls from Trinil, including the lectotype, which are housed at Naturalis Biodiversity Centre in Leiden, the Netherlands (Stremme, 1911;Hooijer, 1958;Tshen Tze Lim, personal communication, 2016   Locality: A sandpit at Tha Chang, Nakhon Ratchasima City, north-eastern Thailand.
Age: Younger than the Middle Miocene and older than the Middle Pleistocene.
Measurements: Antero-posterior diameter of the left horn core: 18.5 mm; medio-lateral diameter of the left horn core: 33.6 mm; length of brain case (between bregma and superior nuchal line): 65.1 mm; maximum width of brain case: 68.7 mm; height of the occipital without the foramen magnum: 38.1 mm (see also Table 1).
Description: The present specimen is a calvarium with complete right and left horn cores, including the frontal, parietal, temporal and occipital bones. The basioccipital part is broken. The dorsal surface on the fronto-parietal region is flat and faintly depressed. The frontal surface is rugose, with weak precornual ridges extending from the anterior ridge of the horn cores (Fig. 3g). Supraorbital foramina are not found on the preserved parts: that means they are situated further towards the front than the horn pedicels. The postero-dorsal rims of the orbits weakly project outwards. Temporal crests are moderately developed, slightly overhanging along the superior nuchal line on the occipital. The external occipital protuberance is projected posteriorly. The median nuchal line forms a sharp ridge and gently inclines below the centre. The brain case is squared in dorsal view (length/width = 0.95). The height of the occipital without the foramen magnum is low relative to the brain-case width (height/width = 0.55). The mastoid processes project laterally, with shallow squamosal shelves.
The horn cores insert caudolaterally from the orbits, strongly inclined backwards being parallel to the frontal surface at the base, and bend upwards at the middle part. In the antero-dorsal view (Fig. 3a), the horn cores diverge at an angle of 95 • at the base, and curve inwards after the middle part. The horn core is weakly twisted clockwise (right side). Right and left horn cores have a wide interval at the frontal surface (56.0 mm between the medial keels). The cross section of the horn cores has a rounded triangular outline antero-posteriorly compressed (Fig. 4b). There are medial and lateral keels from the base to the top. The anterior surface is convex but does not have a clear keel.
Remarks: Most antelope-sized species of Bovinae were traditionally included in the tribe Boselaphini (McKenna and Bell, 1997). However, it is currently accepted that the tribe encompasses only the following crown genera: Boselaphus, Tetracerus, and Duboisia (Bibi, 2009;Bibi et al., 2009). There are some bovines whose taxonomic positions are formally pending. Schlosser (1903) proposed Paraboselaphus ameghinoi, referring to some isolated cheek teeth from China, but his materials and description are insufficient to define an independent taxon based on apomorphies of the genus. Matsumoto (1915) described Proboselaphus from the Pleistocene of Sichuan, China, as a primitive form of living Boselaphus, but the holotype skull of P. watasei has neither cranial nor tooth characteristics of Bovidae, but rather those of Cervidae (Nishioka et al., 2018a). According to Pilgrim (1939), Sivaportax from the Neogene Irrawaddy beds of Myanmar is phylogenetically related to Boselaphus, but our recent studies indicated that the former shares horn core morphology with Miocene Tragoportacini or Stem Bovini rather than Boselaphini (Nishioka et al., 2018b). In this study, we accept that Duboisia is the only extinct genus included in Boselaphini.
Regarding cranial morphology, PPN 01-000109 has the precornual ridges extending from the anterior keels (or ridges) of the horn cores which are a synapomorphy of Boselaphini. Rugosity on the fronto-parietal surface is a symplesiomorphy of Tragoportacini (e.g. Protragocerus, Helicoportax, Tragoportax, and Miotragocerus) and Boselaphini, but that of PPN 01-000109 is finer than that of the tragoportacin genera. Moreover, PPN 01-000109 has relatively weaker temporal crests on the parietals and an antero-posteriorly shorter brain case than the tragoportacin genera listed above. According to previous studies (Stremme, 1911;Hooijer, 1958) and our observation, a combination of the following characteristics of the horn cores is diagnostic to distinguish the genus Duboisia from all of the other genera of Bovinae: the lower half inclined backwards; the upper half curved upwards; and the cross section rounded triangular, compressed antero-posteriorly, and with medial and lateral keels. Boselaphus namadicus is known as an extinct form of boselaphins from the older alluvium (Middle Pleistocene) of the Narmada basin, India (Rütimeyer, 1878;Pilgrim, 1939;Chauhan, 2008), and from the Plio-Pleistocene deposits of Sardhok, Pakistan (Siddiq et al., 2017). The horn core of the holotype (BMNH 36851) of B. namadicus has equilateral triangular cross section with a sharp anterior keel and is less compressed antero-posteriorly than that of PPN 01-000109 or the genus Duboisia. In measurements (Table 1), the brain case of PPN 01-000109 is antero-posteriorly shorter than that of Boselaphus tragocamelus and approximately as long as that of Boselaphus namadicus or Tetracerus quadricornis based on dividing length by width. In the occipital height divided by width, PPN 01-000109 is intermediate between Boselaphus and Tetracerus. Based on the qualitative and quantitative comparisons in horn core and brain-case morphology, as mentioned above, PPN 01-000109 is distinguished from Boselaphus and Tetracerus. The genus Duboisia formally contains two species i.e. D. santeng (Dubois, 1891) and D.? sartonoi Geraads, 1979 296 Y. Nishioka and C. Vidthayanon: First occurrence of Duboisia from Thailand The horn core/pedicel boundary Orbit an. me. an. me.

mm
The horn core/pedicel boundary (or D. saatensis Von Koenigswald, 1934), although the latter was established based on some isolated teeth. Van den Bergh (1988), in his unpublished master dissertation, suggested that Cervus problematicus Von Koenigswald, 1933, from Bumiayu, Java, should be combined with the genus Duboisia based on a large skull fragment. Both ?D. sartonoi and D. problematicus are likely larger than D. santeng, but these species are still debatable taxonomically (Rozzi et al., 2013). PPN 01-000109 is almost as large as the specimen MGB.Dbs. of D. santeng from Sangiran (Table 1). The horn cores of PPN 01-000109 are primarily similar to those of D. santeng, in having keels on the medial and lateral sides and cross section compressed antero-posteriorly, but there are some minor differences between PPN 01-000109 and D. santeng (MGB.SA 290709) from Pucung. Compared to the latter, the former has slender horn cores, curved prominently, and incomplete anterior and medial keels (Fig. 4). Moreover, precornual ridges of PPN 01-000109 are weaker than those of MGB.SA 290709. These differences of horn cores are possibly due to intraspecific variation. Although a species identification of PPN 01-000109 should be discussed in detail based on additional specimens from Thailand and comparing with a sufficient number of specimens of D. santeng from Java, PPN 01-000109 has a strong affinity with D. santeng.

Discussion
What Duboisia species were distributed on the continental region of southern Asia has been discussed in previous studies (Hooijer, 1962;Moigne et al., 2016), but these fossil records based only on isolated teeth are still debatable in taxonomy because cranial (and possibly dental) morphology should be diagnostic of the genus Duboisia (Stremme, 1911;Hooijer, 1958). Recently, we described an incomplete calvarium of Duboisia from an indeterminate horizon at Bagan, central Myanmar (Nishioka et al., 2018b). The present finding of D. aff. santeng from Thailand in conjunction with the fossil record from Myanmar confirmed the existence of Duboisia on continental South East Asia, and provides a certain fossil evidence indicating a morphological affinity of Duboisia species between continental and island South East Asia. Rozzi et al. (2013) discussed if D. santeng had been downsized and acquired unique characteristics in postcranial bones as a result of insular processes (see also Rozzi, 2018) and/or an adaptive shift to dense forest. Although we have not examined strict size and shape differences between Duboisia species from Thailand and Java, our results suggest that basic cranial characteristics of the genus Duboisia (e.g. a squared brain case, weak temporal crests and antero-posteriorly compressed horn cores with medial and lateral keels) had been already acquired before or at the beginning of the Early Pleistocene on continental South East Asia. The occurrence of the genus Duboisia from Thailand supports either of the two hypotheses. If PPN 01-000109 was collected from the upper unit that includes Early Pleistocene tektites (0.8 Ma by Howard et al., 2003), this age corresponds approximately with the D. santeng horizon on Java, or the Early-Middle Pleistocene ages of Sangiran and Trinil H.K. (ca. 0.9-0.5 Ma by Joordens et al., 2015;Falguères et al., 2016). In fact, the Pleistocene fauna share many mammalian species between continental and island regions in South East Asia (Tougard, 2001;Louys et al., 2007;Zin-Maung-Maung-Thein et al., 2010;Suraprasit et al., 2016). Morphological affinities between the Thai and Javanese forms of Duboisia are consistent with what existed contemporaneously in South East Asia. On the other hand, greyish sandstone and pyrite coated the surface of PPN 01-000109 (Fig. 2) are the same as those in the lower unit dated to the Miocene. If PPN 01-000109 was collected from the lower unit, as most mammalian fossils from Tha Chang sandpits were (Chaimanee et al., 2004(Chaimanee et al., , 2007Hanta et al., 2008;Thasod et al., 2011;Deng et al., 2013;Nishioka et al., 2014;Duangkrayom et al., 2017), undeniable morphological differences between D. aff. santeng and D. santeng should be found, which coherently explains geographical and chronological gaps of these species. At the present state, however, we have no more evidence from which horizon PPN 01-000109 was recovered.
The present study clarifies that the genus Duboisia is undoubtedly a member dispersed via the "Siva-Malayan" route and supports a low degree of endemism of the Early-Middle Pleistocene fauna of Java (e.g. Van der Geer et al., 2010). Bovines on the southern Asian continent were dominated by Tragoportacini (e.g. Protragocerus, Helicoportax, and Tragoportax) and Stem Bovini (Selenoportax and Pachyportax) during 9 and 6 Ma (Gentry et al., 2014), and after the latest Miocene the Irrawaddy fauna of central Myanmar began being dominated by Irrawaddy-endemic taxa by the formation of geographical barriers, such as the Indo-Burma Range, between South Asia and South East Asia (Nishioka et al., 2018b). There is an exceptionally incomprehensible fossil record of Duboisia from the Pliocene of India (Moigne et al., 2016), but based on certain cranial fossil records of Boselaphus namadicus (Pilgrim, 1939) and Duboisia species (Nishioka et al., 2018b;this study), geographical isolation at the Indo-Burma boundary could be a possible factor explaining the divergence of Boselaphins between the South Asian lineage (Boselaphus + Tetracerus) and the South East Asian lineage (Duboisia) (Fig. 5).

Conclusions
A species of Duboisia aff. D. santeng newly discovered from a sandpit at Tha Chang, north-eastern Thailand, supports a faunal interchange between continental South East Asia and Java before the Middle Pleistocene. The finding of Duboisia on the South East Asian continent suggests that the genus was diverged from the other Boselaphini (Boselaphus and Tetracerus) owing to a geographical isolation on the continent rather than insular processes of South East Asian islands.
Data availability. All material examined in this study is accessible in the listed institutions and all data are described in the text and table.
Author contributions. YN and CV designed this study and prepared the examined fossil material. YN prepared the manuscript with contributions from all co-authors.