Publication Cover
Historical Biology
An International Journal of Paleobiology
Volume 34, 2022 - Issue 8
403
Views
4
CrossRef citations to date
0
Altmetric
Articles

The first Asian record of the mouse-deer Afrotragulus (Ruminantia, Tragulidae) reassess its evolutionary history and offers insights on the influence of body size on Afrotragulus diversification

ORCID Icon, , ORCID Icon, , & ORCID Icon
Pages 1544-1559 | Received 22 Jan 2022, Accepted 03 Mar 2022, Published online: 21 Mar 2022

ABSTRACT

New fossils of the Miocene crown-tragulid Afrotragulus from Chinji and Dhok Pathan Formations of the Pakistan Siwaliks Group represent its first record out of Africa. This material from Babri Wala (ca. 12.6 Ma), Hasnot 6 (ca. 6.5 Ma) and Barnum Brown’s B 51 classic locality (ca. 13.7 Ma) constitutes three new species, Afrotragulus akhtari, A. moralesi and A. megalomilos. We reassess Afrotragulus ingroup phylogeny recovering two clades with African and Asian representatives. Our results reject the existence of a strictly African lineage in the genus. Body-size estimates show three tiny Afrotragulus with a size corresponding to the lower spectrum of extant Tragulus. However, both Afrotragulus lineages produced species larger than 10 kg. Previously considered very small tragulids, these new forms demonstrate that size range of Afrotragulus equals that of all living tragulids. The smallest forms could be frugivorous/browsers but A. megalomilos and A. moralesi could be opportunistic feeders, specially accounting for their highly derived dentition. These new Asian Afrotragulus extend the biochronological range of the genus from the lower Miocene to the late upper Miocene. Afrotragulus is surprisingly uncovered here as one of the longest-lived and most successful members of the Tragulidae, existing during ca. 13.5 million years (20–6.5 Ma).

Introduction

The genus Afrotragulus (Sánchez et al. Citation2010) comprises small early Miocene crown tragulids (chevrotains and mouse deer; ) originally included within the genus Dorcatherium (as D. moruorotensis Pickford Citation2001 and D. parvum Whitworth Citation1958), a genus now recognised as a paraphyletic assemblage of stem and crown-tragulids requiring extensive revision (Rössner Citation2007; Sánchez et al. Citation2010, Citation2015a, Citation2018). Tragulids are the most basal members of the extant Ruminantia and include 10 currently described living species (Meijaard Citation2011) within three genera: Moschiola (India and Sri Lanka), Tragulus (Southeast Asia and the Philippines) and Hyemoschus (tropical Africa from Sierra Leone to Uganda; see Nowak Citation1999; Meijaard and Gooves Citation2004; Rössner Citation2007; Meijaard Citation2011 and references therein). The oldest record of tragulids dates back to the Late Eocene (Métais et al. Citation2001) and they are the only non-pecoran ruminant lineage that survived the Oligo-Miocene boundary (Rössner Citation2007; Sánchez et al. Citation2010 and references therein). Yet the group’s Palaeogene history remains very partially understood (Mennecart et al. Citation2021). Tragulids experienced an extraordinary diversification during the final part of the lower Miocene, enjoying great evolutionary success during the Miocene (see e.g. Whitworth Citation1958; Hamilton Citation1973; Mein Citation1989; Demment and Ginsburg Citation1997; Demment et al. Citation1999; Ginsburg et al. Citation2001; Pickford Citation2001, Citation2002; Rössner Citation2007; Quiralte et al. Citation2008; Sánchez et al. Citation2010, Citation2015a, Citation2018; Guzmán-Sandoval and Rössner Citation2019; Mennecart et al. Citation2021). The diversity of tragulid lineages and morphologies described at the end of the lower Miocene, together with some putative Oligocene tragulid forms (Nalameryx and Iberomeryx), suggests that a pre-Miocene radiation event of tragulids that produced the Miocene and living lineages took place either in the Oligocene–Miocene transition or before that in the early-late Oligocene (Sánchez et al. Citation2010, Citation2015a, Citation2018; Mennecart et al. Citation2021).

Figure 1. Afrotragulus moruorotensis (lower Miocene, Kenya) is one of the smallest ruminants known to science. Here, an adult male forages on the forest floor looking for fruits, sprouts, seeds and other high-energy vegetal food, without realising the danger nearby in the form of a Bitis adder. Art by IMS.

Figure 1. Afrotragulus moruorotensis (lower Miocene, Kenya) is one of the smallest ruminants known to science. Here, an adult male forages on the forest floor looking for fruits, sprouts, seeds and other high-energy vegetal food, without realising the danger nearby in the form of a Bitis adder. Art by IMS.

The fossil record of Afrotragulus consists mostly of lower dentition, some upper molars and extremely scarce postcranial remains. Afrotragulus lower molars are extraordinarily derived and easily recognisable by their narrow and elongated crowns with developed cristids, extremely reduced M-structure and characteristically separated mesial and distal lobes, which produce a unique structure called the interlobular bridge (Sánchez et al. Citation2010). The two currently described species of Afrotragulus are diminutive in size. Afrotragulus parvus is approximately the size of the extant smallest Tragulus, while A. moruorotensis is one of the smallest ruminants ever described. In previous studies, Afrotragulus was recovered as the sister group of the clade formed by the extant Tragulus and Moschiola within a ‘Selenodont clade’ of tragulids (Sánchez et al. Citation2015a). Diminutive, recognisable and remarkable as they are, Afrotragulus fossils were not identified out of Africa. Afrotragulus moruorotensis (East Africa, ca. 17.5–16.8 Ma) and A. parvus (East and Southern Africa, ca. 20–17.5 Ma) appeared in the fossil record as the only representatives of a genus that was not diverse and that was relatively short-lived. However, since the original description of Afrotragulus (Sánchez et al. Citation2010), some tragulid remains were discovered and/or identified in museum collections that matched the anatomical features of this genus. All of these fossils were Asian, coming from the Miocene deposits of the Siwaliks Group in Northern Pakistan ().

Figure 2. Geographic location of B 51, Babri Wala and Hasnot 6 Afrotragulus-yielding fossil sites in the Chinji-Kothera and HASNOT areas in Northern Pakistan. The towns of Chinji, Kothera and Hasnot are also marked on the map.

Figure 2. Geographic location of B 51, Babri Wala and Hasnot 6 Afrotragulus-yielding fossil sites in the Chinji-Kothera and HASNOT areas in Northern Pakistan. The towns of Chinji, Kothera and Hasnot are also marked on the map.

In 2013, the palaeontological team of the University of the Punjab including SGA, MAB and MAK surveyed the uppermost conglomerate levels of the locality of the site of Hasnot 6. Among the fossils discovered is a single m3 of a relatively large Afrotragulus-like tragulid (PUPC 13/323). Hasnot 6 is located ca. 2.5 km northwest of Hasnot village and belongs to the upper Dhok Pathan Formation (Middle Siwaliks). During 2019 more fieldwork carried out by the same palaeontological team found several Afrotragulus fossils in the fossil site of Babri Wala, part of the Chinji Formation (Lower Siwaliks Subgroup) stratotype located nearby. Babri Wala is about 2.82 km southeast of Kotehra, 3.7 km southeast of Chinji town and 3.9 km Northeast of Chinji Rest House (now Thati Rest House). All the tragulid remains were recovered from the mudstones that dominate the site, and the whole sample is from a single Tragulus-sized form.

In addition to the new findings, the Barnum Brown expeditions to the Siwaliks (1921–1925) collected a large number of tragulid fossils curated by the American Museum of Natural History (New York, USA). This collection includes a left hemimandible (AMNH 19310) assigned by Colbert (Citation1935) to Dorcatherium minus (probably due to its small size relative to other ‘Dorcatherium’), describing its site of origin and horizon as ‘Lower Siwaliks, 1600 feet above the level of Chinji Rest House, one and one half miles northeast of Chinji Rest House’ (Colbert Citation1935; page 309). AMNH 19310 was marked with the inscription ‘B 051’ which unambiguously refers to the locality B 51 (estimated coordinates, Lat. 32°40ʹ43.62” N, Long. 72°23ʹ32.36”E and altitude 636.11 m), with an estimated distance (using Google Earth) from the famous building of Chinji Rest House of 2.334 km (1.45 miles). This is almost the same distance calculated by Colbert (Citation1935). During our own revision of the AMNH tragulid material from the Siwaliks we realised that the molars of AMNH 19310 do not have the morphological traits of ‘Dorcatherium’ minus (see e.g. Guzmán-Sandoval and Rössner Citation2019) showing instead the full array of derived features associated with Afrotragulus.

The discovery of these three Asian Afrotragulus species completely changes the evolutionary paradigm of this genus. Instead, Afrotragulus has a complex and much longer evolutionary history, substantial morphological and body-size variation, and a broader palaeogeographical distribution spanning Africa and Asia. This last in particular questions the existence of an African lineage as proposed by Pickford (Citation2002).

Here, we describe the new findings of Afrotragulus with the following aims: revise the hypodigm, diagnosis and definition of Afrotragulus; and test the phylogenetic affinities of the new taxa, examine the biogeographic history of Afrotragulus and the possible existence of independent African and Asian lineages within the genus.

Material and methods

Locality, geological setting and age

Both Babri Wala and B 51 sites () belong to the Chinji Formation of the Lower Siwaliks Group (14.070–11.146 Ma, late Middle Miocene; see Barry et al. Citation2013 for chron designation), which has been classically divided into lower, middle and upper sections. A recent calibration (Barry et al. Citation2013) of the Siwaliks palaeomagnetic data (Sheikh Citation1984; Johnson et al. 1085; Johnson et al. Citation1988; McRae Citation1990; Kappelman et al. Citation1991) with the Magnetic Polarity Time Scale of Ogg and Smith (Citation2004) have resulted in a calculated age for the three Chinij Formation sections of 13.608–14.070 Ma for the Lower Chinji Fm (chrons C5AB-C5ACn), 12.735–13.363 Ma for the middle Chinji Fm (chrons C5Ar-C5AA) and 11.146–12.174 Ma for the upper Chinji Fm (chrons C5r-C5An; see Barry et al. Citation2013).

The area (locally known as Kund) of locality B 51 lies within the administrative boundaries of Chinji village. This area also represents the type section of Chinji Formation mentioned by Shah (Citation1977, 2009). McRae (Citation1990) described the lithological details of B 51 in his KR-4 (Kundal Rainbow Sandstone-4) section (later redrawn by Barry et al. Citation2013, fig. 15.3). Located 25 m above the base level, B 51 is composed of conglomerates, sandstone, siltstone, mudstone (predominant) and paleosols (McRae Citation1990). Paleomagnetically, B 51 fall in chron C5ACn, dating 14.070–13.739 Ma (Barry, pers. comm. 2019; Barry et al. Citation2013, fig. 15.3), which includes it in the lower Chinji Fm.

Babri Wala site is located ca. 1.6 km northeast of B 51 within the administrative boundaries of Kohtehra (Lat. 32° 42’ 57.25” N, Long. 72° 23’ 35.95” E and altitude 694.33 m), a small village about 2.41 km northeast of Chinji town (Lat. 32° 42’ 33.95” N, Long. 72° 22’ 09.32” E and altitude 698.3 m; ). The site is heavily eroded and with the upper part missing, with dimensions of about 4.5 m high, 2.74 m long and 1.5 m wide, and showing prevalent red mudstones with iron concretions and less abundant grey sandstones (). Like Chinji town, Kohtehra is located on the sandstones of the Nagri Formation. However, outcrops of the Chinji Formation (part of the Chinji Fm stratotype) are well exposed about 2.41 km south of Kohtehra village. According to Shah (Citation1977, 2009), the type section of the Chinji Fm lies in the administrative boundaries of Kohtehra. Johnson et al. (Citation1985), (Citation1988), Willis (Citation1993) and Badgley et al. (Citation1998) designated a 12 km long area as the Chinji Fm stratotype, ranging from Chinji Rest House and Chinji (locality 430 and 41) up to the area of Bhilomar (locality 495). The studied area of the Kohtehra village outcrops belonging to the Chinji Fm range between the K2-K5 sections of Johnson et al. (Citation1988) and part of a long section that includes the localities Y-GSP 693 and Y-GSP 732 (Badgley et al. Citation1998). Babri Wala site does not currently appear in the database of Yale-Geological Survey of Pakistan (Barry, pers. comm. 2019), constituting a new addition to the list of fossil productive sites of the Chinji stratotype, lying between the two unnumbered sites of Badgley et al. (Citation1998). Lithological comparison with the section KR-2 of McRae (Citation1990) reveals that Babri Wala correlates well with the upper half of this section located ca. 130 metres above the base level. The lithological features of the site indicate that it belongs to the middle Chinji Formation, with an approximate age of 12.6 Ma. (Johnson et al. Citation1988; McRae Citation1990). The associated fauna of the site and adjacent area include bovids (Miotragocerus gluten and Gazella sp.), stem-giraffids (Giraffokeryx punjabiensis), rhinoceroses (Gaindatherium browni) and rodents (Rodentia indet.).

Figure 3. a, lithological column of the site of Babri Wala showing the two Afrotragulus-yielding mudstone levels; b, lithological column of the site of Hasnot-6, showing the upper conglomerate level from which PUPC 13/323 was recovered.

Figure 3. a, lithological column of the site of Babri Wala showing the two Afrotragulus-yielding mudstone levels; b, lithological column of the site of Hasnot-6, showing the upper conglomerate level from which PUPC 13/323 was recovered.

The locality of Hasnot 6 (; Lat. 32° 50’ 42.8” N and Long. 73° 17’ 51.5” E, elevation 359.89 m) belongs to the upper Dhok Pathan Formation of the Middle Siwaliks Subgroup, located over the locality B117 (Colbert Citation1935). Locality B117 has a palaeomagnetic age of 7.360 to 7.083 Ma (Barry et al. Citation2002) and it lies slightly below the fossiliferous levels of Hasnot 6, which is in turn situated about fifteen metres below the Tatrot Formation caps yielding Anancus sivalensis fossils (Abbas Citation2018). Hasnot 6 was once part of a larger locality that was cut away for the construction of the Padhri to Hasnot road. Lithostratigraphically, the topmost layer of Hasnot 6 is composed of a conglomerate fossiliferous bed. Below this uppermost level, there is a thick stratum of light grey coloured fine-grained non-fossiliferous sandstone, and below it there is a layer of loose conglomerate with light pink coloured clay and dark back coloured small sized pebbles. This layer is the most fossiliferous one and the level from which PUPC 13/323 was collected. The locality has yielded remains of Hippohyus cf. sivalensis, Propotamochoerus hysudricus, Gazella lydekkeri and Tragoportax punjabicus (Babar Citation2017). According to Pickford (Citation1988), Hippohyus sivalensis lived from ca. 5–2 Ma, which would correspond to latest Miocene to early Pleistocene. However, Barry et al. (Citation2002) consider the oldest occurrence of H. sivalensis around 6 Ma. According to Badgley et al. (Citation2008), Gazella lydekkeri ranged from 10.2 to 6.1 Ma, Tragoportax punjabicus ranged from 8.2 to 6.2 Ma and Propotamochoerus hysudricus ranged from 10.2 to 6.5 Ma. Hence, the estimated age for Hasnot 6 is roughly 6.5 Ma.

Material

The tragulid fossils from the Siwaliks described in this paper are curated by the AMNH (AMNH 19310, lower Chinji Fm) and the Dr. Abu Bakr Fossil Display and Research Centre, Department of Zoology, University of the Punjab, Lahore (PUPC 19/58, PUPC 19/59, PUPC 19/60, PUPC 19/61, middle Chinji Fm; and PUPC 13/323, Dhok Pathan Fm).

Data from the taxa included in the phylogenetic analysis come from Dorcabune anthracotherioides (originals AMNH); Dorcabune welcommi (cast MNCN, also publication); Dorcatherium crassum (Sansan originals MNHN and MNCN casts, photographs from Guzmán-Sandoval Citation2018, Mennecart et al. Citation2018); D. namaquensis (originals GSN); D. pigotti (originals NHM, casts MNCN, NMK); D. iririensis (original UM); D. guntianum (photographs SMNS); D. naui (original NHM, original from Los Valles de Fuentidueña, MNCN); Hyemoschus aquaticus (AMNH, MNCN); Siamotragulus sanyathanai (casts MNCN); S. bugtiensis (casts MNCN); S. songhorensis (originals NHM, UM); Yunnanotherium simplex (original publication Han 1986); Tragulus javanicus (AMNH, MNCN, MAUV); Moschiola memmina (AMNH, MNCN). All Afrotragulus studies were based on originals, with the exception of the mandible KNM-SO 1345 of A. parvus (Geraads Citation2010) from Songhor (photographs).

Tragulid fossil material assigned to ‘D’. parvum and ‘D’. moruorotensis different from that described from Rusinga (including that of Whitworth Citation1958) and Moruorot (Pickford Citation2001) when the genus Afrotragulus was first erected (Sánchez et al. Citation2010) was cited from other East African sites (Pickford Citation2001, Citation2002; Geraads Citation2010), potentially expanding the chronostratigraphic range of the genus Afrotragulus in Africa (as effectively proposed by Geraads Citation2010). This material mostly consists of isolated postcranial fragments/bones (e.g. astragali) that preclude their unambiguous attribution to Afrotragulus, except, apparently, by size. However, size alone is not a trustworthy tool on tragulid systematics (see Sánchez et al. Citation2010 for the discussion on this topic), as evident by the larger-bodied Asian Afrotragulus described here. Hence, until part or all of these fossils are properly described and unambiguously assigned to Afrotragulus, we choose to be conservative and limit our sample to the African Afrotragulus material originally described in Sánchez et al. (Citation2010) plus all additional materials that can be unambiguously assigned to Afrotragulus, like the hemimandible KNM-SO 1345 from Songhor (Kenya, 20 Ma) included by Geraads (Citation2010) in D. parvum. The tragulid ankle KNM-RU 15136 from Rusinga (Kenya) was assigned by Geraads (Citation2010) to Dorcatherium cf. parvum. It has a free malleolar bone and the cubonavicular is fused with the ectomesocuneiform (Geraads Citation2010). Also the astragalus length appears similar (from the scale in the picture) to that of A. parvus from Rusinga (see Sánchez et al. Citation2010). However, until direct examination of this specimen is possible, we prefer to exclude it from the morphological/cladistic analysis.

Measurements

We follow the set of measurements proposed by Sánchez and Morales (Citation2008). All measurements of the new dental material were taken with digital calipers. The diminutive lower dentition of the Afrotragulus moruorotensis Mor1ʹ2000 holotype (see the original odontometrics in Pickford Citation2001) was measured again using a Nikon Measuroscope 5 10x microscope with an incorporated micrometre of 0.025 mm accuracy. The detailed odontometrics are presented as supplementary information.

Nomenclature and anatomical definitions

We use the general terminology of Azanza (Citation2000) for nomenclature of the ruminant dentition (English version in Sánchez and Morales Citation2008). The Dorcatherium-fold is the fold that occurs on the linguo-distal side of the metaconid (). The Tragulus-fold is the fold situated on the distal side of the protoconid, usually linked to the post-protocristid (sometimes to the conid itself; ). Combined, both the Dorcatherium-fold and the Tragulus-fold form the M-structure (; see also e.g. Janis Citation1987; Geraads et al. Citation1987), a unique trait of the lower dentition that diagnoses the Tragulidae (see e.g. Métais et al. Citation2001; Sánchez et al. Citation2015a; Guzmán-Sandoval and Rössner Citation2019). The inter-lobular bridge is the rectilinear bridge of enamel that connects the anterior and posterior lobes in an Afrotragulus-type lower molar (; Sánchez et al. Citation2010). The Zhailimeryx-fold is a fold of enamel that originates from the anterior part of the entoconid, the development of which is highly variable amongst tragulids (from well-developed to absent). The Dorcatherium-platform is the mesial semicircular structure of the lower molars formed by a hyper-developed pre-protocristid that turns lingually to contact a very small pre-metacristid (; see also Sánchez et al. Citation2010; Morales et al. Citation2012).

Figure 4. Occlusal anatomical elements of tragulid lower molars. a, Afrotragulus moruorotensis, m2 of the holotype CMK Mor 1ʹ2000; b, Dorcatherium naui, m2, right hemimandible of the specimen BMNH M40432 from the type locality Eppelsheim; c, Moschiola meminna, left m2 (private collection Jan van der Made, Madrid); d, ‘Dorcatheriumcrassum, left m2 of neotype hemimandible Sa 9950 from Sansan (Morales et al. Citation2012), mirrored for comparison purposes. Taken from Sánchez et al. (Citation2015a); e, schematic depiction of the three types of mesial morphology in tragulid molars. From left to right: very large pre-protocristid which turns lingually and contacts a very small pre-metacristid, forming the Dorcatherium-platform; large rectilinear pre-protocristid that meets a smaller pre-metacristid, with no Dorcatherium-platform; straight pre-protocristid and pre-metacristid which are subequal in length, meeting parasagitallly and forming a triangular mesial outline of the teeth. All images by IMS.

Figure 4. Occlusal anatomical elements of tragulid lower molars. a, Afrotragulus moruorotensis, m2 of the holotype CMK Mor 1ʹ2000; b, Dorcatherium naui, m2, right hemimandible of the specimen BMNH M40432 from the type locality Eppelsheim; c, Moschiola meminna, left m2 (private collection Jan van der Made, Madrid); d, ‘Dorcatherium’ crassum, left m2 of neotype hemimandible Sa 9950 from Sansan (Morales et al. Citation2012), mirrored for comparison purposes. Taken from Sánchez et al. (Citation2015a); e, schematic depiction of the three types of mesial morphology in tragulid molars. From left to right: very large pre-protocristid which turns lingually and contacts a very small pre-metacristid, forming the Dorcatherium-platform; large rectilinear pre-protocristid that meets a smaller pre-metacristid, with no Dorcatherium-platform; straight pre-protocristid and pre-metacristid which are subequal in length, meeting parasagitallly and forming a triangular mesial outline of the teeth. All images by IMS.

Phylogenetic analysis

We performed a cladistic analysis at the species-level to explore the phylogenetic relationships of the new studied Siwaliks tragulids using the TNT software (Goloboff and Catalano Citation2008). We chose Zhailimeryx jingweni as the outgroup (TNT only allows a single outgroup) since this taxon was previously used successfully for root tragulids (Métais et al. Citation2001; Sánchez et al. Citation2015a, Citation2018). As the ingroup, we chose the same terminals as in Sánchez et al. (Citation2018) with the addition of Siamotragulus bugthiensis, S. songhorensis, Yunnanotherium simplex and the three new Asian Afrotragulus. The morphological data matrix consists of an updated and extended version of the matrix presented in Sánchez et al. (Citation2015a), including 73 characters picked from the skull (11), upper dentition (6), lower dentition (32) and postcranial skeleton (24). We performed a run using a traditional search with 1000 replicates with TBR (Tree Bisection Reconnection) that recovered one most parsimonious tree (MPT) of 162 steps. The character/state distribution for the discussed nodes and terminals is presented in .

Table 1. Character/state distribution for the internal nodes and terminals discussed in the text. Ambiguous apomorphies marked in italics.

Body size estimates

We predict body mass (in kg) – as a measurement of body size – of fossil Afrotragulus spp. and Yunnanotherium simplex from the (occlusal) length measure of their molars (), as the correlation between this variable and the body size in a wide variety of living ruminants (including tragulids) has been successfully tested (Janis Citation1990a) and allowed accurate estimates of body mass for extinct forms (Janis Citation1990b; DeMiguel et al. Citation2012; DeMiguel Citation2016). We included Yunnantoherium into the calculations since it is part of the sister clade to Afrotragulus together with the extant Tragulus and Moschiola. The selection of craniodental remains for body mass estimation is due to the fact that these are the only (and most diagnostic) elements available in our fossil sample (except for a couple of astragali), though skeletal parts (and in particular those that play a role in body support – such as major limb bones) are much weight bearing (Gingerich Citation1990; Janis Citation1990a). We calculated the mean body mass for each specimen following regression equations by Janis Citation1990a (log(Y) 0 a + b log (X); Y = body mass in kg; a-b constants for each group; X = m1 m2 length in cm) and then for each species by averaging mean across individuals (). Only lower molars and adult specimens (in which the length growth is complete) were considered. When several specimens (i.e. different tooth positions) from the same individual were available, only the first molar was selected for analysis because m1 length results in better weight estimates.

Table 2. Summary of body mass estimates for Afrotragulus spp. and Yunnanotherium.

Systematic palaeontology

Cetartiodactyla (Montgelard et al. Citation1997)

Ruminantia (Scopoli Citation1777)

Tragulidae (Milne-Edwards Citation1864)

Afrotragulus (Sánchez et al. Citation2010)

Emended diagnosis. Tragulids with narrow and selenodont lower molars; extremely reduced M-structure lacking a Tragulus-fold; well-developed cristids and relatively flat lingual cuspids; interlobular bridge in m1-m2; short interlobular bridge in m3; enlarged and wide central valley in the lower molars.

Type species. Afrotragulus moruorotensis (Pickford Citation2001)

Afrotragulus moruorotensis (Pickford Citation2001)

2001 Dorcatherium moruorotensis Pickford: 437

Emended diagnosis. Afrotragulus with long interlobular bridge in the m3; aligned cuspids forming an almost rectilinear lingual wall. Also, very elongated and narrow lower molars; very high cristids that extend ‘web-like’ between the cuspids; absent posterior cingulid; and flattish entoconid.

Afrotragulus parvus (Whitworth Citation1958)

1958 Dorcatherium parvum Whitworth: 11–14

2002 Dorcatherium parvum Pickford: 88

Emended diagnosis. Afrotragulus with short and straight post-hypocristid that does not reach the lingual wall and does not extend between the third lobe and the post-entocristid in the m3; almost aligned entoconid and hypoconid.

Afrotragulus akhtari sp. nov.

)

Figure 5. Afrotragulus akhtari sp. nov. type series. a–c, PUPC 19/61, holotype, left hemimandibular fragment with m2-m3. a, buccal, b, lingual and c, occlusal views; d–f, PUPC 19/60, paratype, right m3. d, buccal, e, lingual and f, occlusal views; g–i, PUPC 19/59, paratype, left m2. g, buccal, h, lingual and i, occlusal views; j–l, PUPC 19/58, paratype, left M2. j, occlusal, k, buccal and l, lingual views. Afrotragulus moralesi sp. nov. type series. m–o, PUPC 13/323, holotype, left m3. m, buccal, n, lingual and o, occlusal views.

Figure 5. Afrotragulus akhtari sp. nov. type series. a–c, PUPC 19/61, holotype, left hemimandibular fragment with m2-m3. a, buccal, b, lingual and c, occlusal views; d–f, PUPC 19/60, paratype, right m3. d, buccal, e, lingual and f, occlusal views; g–i, PUPC 19/59, paratype, left m2. g, buccal, h, lingual and i, occlusal views; j–l, PUPC 19/58, paratype, left M2. j, occlusal, k, buccal and l, lingual views. Afrotragulus moralesi sp. nov. type series. m–o, PUPC 13/323, holotype, left m3. m, buccal, n, lingual and o, occlusal views.

LSID. – zoobank – 7FC6D555–D653–4632–A8BE–FCB4A88A8482

Etymology of name. In honour of Prof. Dr. Muhammad Akhtar, who revived palaeontological science in Pakistan during his extensive professional career.

Holotype. PUPC 19/61, left hemimandibular fragment with m2-m3 ().

Paratypes. The remaining referred material from Babri Wala, Kohtehra (see below).

Locality and age. Babri Wala, Kohtehra, middle Chinji Formation (~12.6 Ma), late Middle Miocene.

Material. PUPC 19/58, left M2; PUPC 19/59, left m2; PUPC 19/60, right m3; PUPC 19/61, left hemimandibular fragment with m2-m3.

Diagnosis. Afrotragulus with closed third lobe of the m3; long and curved post-hypocristid that turns mesially to contact the entoconid; strong posterior cingulid; almost aligned entoconid and hypoconid; bifurcated post-hypocristid in the m3, with small branches.

Differential diagnosis. Afrotragulus akhtari differing from its sister taxon A. moralesi in: closed third lobe of the m3; long and curved post-hypocristid that turns mesially to contact the entoconid; and strong posterior cingulid. Differing from A. moruorotensis in: less vertically developed and broader cuspids; less extensively developed cristids; almost aligned entoconid and hypoconid; closed third lobe of the m3; long and curved post-hypocristid that turns mesially to contact the entoconid and that becomes bifurcated in the m3. Differing from A. parvus in: closed third lobe of the m3; long and curved post-hypocristid that turns mesially to contact the entoconid and that becomes bifurcated with small branches in the m3, and strong posterior cingulid. Differing from A. megalomilos sp. nov. in: less aligned lingual cuspids; strong posterior cingulid; almost aligned entoconid and hypoconid; closed third lobe of the m3; and long and curved post-hypocristid that turns mesially to contact the entoconid and that becomes bifurcated in the m3. Differing from all other non-Afrotragulus tragulids in the presence of interlobular bridge together with high cristids and extremely reduced M-structure with no Tragulus-fold and very reduced Dorcatherium-fold.

Description

Upper permanent dentition. The upper dentition is represented in the sample by a single left M2. The anterior lobe is anteroposteriorly larger than the posterior lobe. The protoconal cingulum is well developed, a typical Afrotragulus feature (see Sánchez et al. Citation2010). The molar is markedly selenodont with well-developed cristae. The lingual ribs are strong and elongated. The post-protocrista is short but the pre-metaconulecrista is well developed, almost reaching the inner buccal wall.

Lower permanent dentition. The lower molars are narrow, although they do not have the elongated look of the molars of A. moruorotensis and A. megalomilos. Similarly, the cuspids are more bunoid than in clade B (see Discussion) Afrotragulus. Also, the cristids are high but not so horizontally extended as in A. moruorotensis and A. megalomilos. This condition is especially visible in the pre-hypocristid. The distal lobe of m1-m2 is distinctly larger than the mesial lobe. The third lobe of the m3 appears in a central position (character 18:1). The interlobular bridge is short in all three molars and has a relatively primitive aspect, with the distal end of some of the cristids still visible (). The post-hypocristid is long and curved, turning mesially to contact the entoconid, it is bifurcated with two small branches in the m3. The mesial cingulid is present, and the posterior cingulid is strong. The ectostylid is short although sometimes with a relatively broad base.

Afrotragulus moralesi sp. nov.

)

LSID. – zoobank – 7FC6D555–D653–4632–A8BE–FCB4A88A8482

Etymology of name. Honouring Prof. Jorge Morales, for a life devoted to the study of ruminants.

Holotype. PUPC 13/323, left m3 ().

Paratypes. None.

Locality and age. Locality Hasnot 6, Dhok Pathan Formation (ca. 6.5 Ma), Upper Miocene.

Material. The same as the holotype, PUPC 13/323, left m3.

Diagnosis. Afrotragulus with prehypoconulid-cristid developed as a buccal wall that connects with the post-hypocristid in the upper half of the crown; posthypocristid bifurcated in the m3 with two small branches; almost aligned entoconid and hypoconid.

Differential diagnosis. Afrotragulus moralesi differing from its sister taxon A. akhtari in: prehypoconulid-cristid developed as a buccal wall that connects with the post-hypocristid in the upper half of the crown. Differing from A. parvus in: prehypoconulid-cristid developed as a buccal wall that connects with the post-hypocristid in the upper half of the crown; and posthypocristid bifurcated in the m3 with presence of two small branches. Differing from A. moruorotensis in: posthypocristid bifurcated in the m3 with presence of two small branches; non-aligned lingual cuspids; strong posterior cingulid; cristids developed but without the horizontal extension (webbing-like morphology) of clade B Afrotragulus. Differing from A. megalomilos in: posthypocristid bifurcated in the m3 with presence of two small branches; non-aligned lingual cuspids; strong posterior cingulid; cristids developed but without the horizontal extension (webbing-like morphology) of clade B Afrotragulus; and central position of the third lobe of the m3.

Description

Lower permanent dentition. The m3 of A. moralesi is shorter and more compact than that of A. moruorotensis, with less horizontally extended cristids, especially the pre-hypocristid. Nevertheless, the cristids are distinctively higher than those of A. akhtari. Typical of Afrotragulus, the anterior and posterior lobes are sub-equal in size, with very similar width. The M-structure is very reduced, and the interlobular bridge is short. The post-hypocristid is long and bifurcated, showing two small branches. The third lobe is located in a central position. The prehypoconulid-cristid is developed as a buccal wall that connects with the post-hypocristid in the upper half of the crown. This constitutes a parallelism with the A. megalomilos-A. moruorotensis clade and, together with the high cristids, easily differentiates A. moralesi from A. akhtari. The entoconid and hypoconid are aligned. There is a mesial cingulid, and the ectostylid is small.

Afrotragulus megalomilos sp. nov.

)

Figure 6. Afrotragulus megalomilos sp. nov. holotype, with the holotype of its sister taxon A. moruorotensis (new CT SCAN images) added for comparison purposes. a–c, AMNH 19310, holotype, left hemimandible with m1-m3. a, buccal, b, lingual and c, occlusal views; d–f, Afrotragulus moruorotensis holotype CMK Mor 1ʹ2000, left hemimandible with m1-m3. d, buccal, e, lingual and f, occlusal views.

Figure 6. Afrotragulus megalomilos sp. nov. holotype, with the holotype of its sister taxon A. moruorotensis (new CT SCAN images) added for comparison purposes. a–c, AMNH 19310, holotype, left hemimandible with m1-m3. a, buccal, b, lingual and c, occlusal views; d–f, Afrotragulus moruorotensis holotype CMK Mor 1ʹ2000, left hemimandible with m1-m3. d, buccal, e, lingual and f, occlusal views.

1935 Dorcatherium minus Colbert: 309.

2019 Dorcatherium dehmi Guzmán-Sandoval and Rössner: 13, 20,

LSID. – zoobank – 7FC6D555–D653–4632–A8BE–FCB4A88A8482

Etymology of name. Greek for ‘Big Apple’, the universally used nickname for New York City (USA), and thus referred to the City and the AMNH where the fossil has been curated during almost a century.

Holotype. AMNH 19310, left hemimandible with m1-m3 (Whitworth Citation1958) ()).

Paratypes. None.

Locality and age. Locality B 51, Lower Chinji Formation (ca. 13.8–13.7 Ma), Middle Miocene.

Material. The same as the holotype, AMNH 19310, left hemimandible with m1-m3.

Diagnosis. Afrotragulus with the third lobe of the m3 in buccal position (character 18:0). Also, high web-like cristids and aligned cuspids that form a rectilinear lingual wall; no posterior cingulid; prehypoconulid-cristid developed as a buccal wall that connects with the post-hypocristid in the upper half of the crown.

Differential diagnosis. Afrotragulus megalomilos differing from its sister taxon A. moruorotensis in: presence of short interlobular bridge in the m3; more rounded (less flattish) metaconid; lingually broader entoconid; relatively more marked Dorcatherium-fold; weak cingulid in the protoconid; and relatively lower lingual cuspids that are also less rectilinear in m1-m2. Differing from A. parvus in: aligned lingual cuspids; no posterior cingulid; entoconid clearly in mesial position relative to the hypoconid; long and straight post-hypocristid that reaches the lingual wall, extending in the m3 between the third lobe and the post-entocristid; prehypoconulid-cristid developed as a buccal wall that connects with the post-hypocristid in the upper half of the crown. Differing from A. akhtari sp. nov. in: presence of aligned lingual cuspids; absent posterior cingulid; entoconid clearly in mesial position relative to the hypoconid; long and straight post-hypocristid that reaches the lingual wall, extending in the m3 between the third lobe and the post-entocristid; open third lobe of the m3; prehypoconulid-cristid developed as a buccal wall that connects with the post-hypocristid in the upper half of the crown. Differing from A. moralesi in: presence of aligned lingual cuspids; absent posterior cingulid; entoconid clearly in mesial position relative to the hypoconid; long and straight post-hypocristid that reaches the lingual wall, extending in the m3 between the third lobe and the post-entocristid; open third lobe of the m3. Differing from all other non-Afrotragulus tragulids in the presence of interlobular bridge together with high cristids and extremely reduced M-structure with no Tragulus-fold and very reduced Dorcatherium-fold.

Description

Lower permanent dentition. AMNH 19310 represents an adult Afrotragulus with medium to high (in the m1) molar wear. At first sight, A. megalomilos appears like a very large version of A. moruorotensis, with the same kind of elongated molars with horizontally extended, web-like, cristids. However, contrary to A. moruorotensis, the interlobular bridge in the m3 is short, and also the entoconid is broader, the metaconid is less flattish, and lacks a strong cingulid in the protoconid. The lingual cuspids are aligned, although not as much as in A. moruorotensis. The m2 is distinctively larger than the m1. In the m1, the distal lobe is broader than the anterior lobe, as usual in Afrotragulus. However, in the m2-m3 both lobes are very similar in size. The M-structure is very reduced, with a very faint expression of the Dorcatherium-fold in the m3. The third lobe of the m3 is tilted buccally. The post-hypocristid is long and straight, and in the m3 is simple (non-bifurcated) and extends almost between the third lobe and the post-entocristid. The pre-hypoconulidcristid is developed as a buccal wall that joins the post-hypocristid in the upper half of the crown. A mesial cingulid is present and is especially visible in m2-m3, while the entoconid is very small in all three molars.

Phylogenetic relationships of Afrotragulus

We tested the ingroup phylogenetic relationships of Afrotragulus using a Maximum Parsimony search, obtaining a single MPT () of 162 steps (CI – 0.593; RI – 0.692) in which the genus Afrotragulus is recovered as the sister group of a clade containing Yunnanotherium simplex plus the Asian extant forms Moschiola and Tragulus. Two lineages of Afrotragulus, each containing Asian and African forms, are recovered. The first lineage (C) clusters the African A. parvus with the Asian A. akhtari sp. nov. plus A. moralesi sp. nov., whereas the second lineage (B) groups the African A. moruorotensis with the Asian A. megalomilos.

Figure 7. MPT showing the ingroup phylogeny of Afrotragulus. The new species A. akhtari sp. nov., A. megalomilos sp. nov. and A. moralesi sp. nov. are marked in bold type. Clades A, B, C and D are discussed in the text. Numbers below the branches represent the absolute Bremer support values, and numbers above the branches are the bootstrap values (only those >50 are represented).

Figure 7. MPT showing the ingroup phylogeny of Afrotragulus. The new species A. akhtari sp. nov., A. megalomilos sp. nov. and A. moralesi sp. nov. are marked in bold type. Clades A, B, C and D are discussed in the text. Numbers below the branches represent the absolute Bremer support values, and numbers above the branches are the bootstrap values (only those >50 are represented).

The present work is not intended as a full phylogeny of Tragulidae, instead focusing on Afrotragulus. Some of the more interesting results (e.g. the radiation of stem-tragulids, the possible paraphyly of Siamotragulus and the position of the extant Hyemoschus) need to be further confirmed with more terminals and more characters, including DNA sequences from the extant species (Sánchez et al. in prep.).

Body mass estimates

Body mass in Afrotragulus spp. () ranges from approximately 1.5 kg in A. moruorotensis to 23 kg in A. moralesi (though this latter inference is based on a single m3, which provides higher size estimates). Afrotragulus parvus and A. akhtari are small species (2.3–3 kg) and A. megalomilos is an intermediate-sized species (13 kg). Body mass for Yunnanotherium simplex was determined as approximately 5.5 kg ().

Comparisons and discussion

We define Afrotragulus as the least inclusive clade of tragulids that contains Afrotragulus moruorotensis and Afrotragulus moralesi. The diagnosis of the clade Afrotragulus (clade A in ) of Sánchez et al. (Citation2010) is modified as follows: selenodont tragulids with interlobular bridge in m1-m2; short interlobular bridge in the m3; enlarged and wide central valley; and extremely reduced M-structure with no Tragulus-fold. Afrotragulus appears as the sister-group of the clade that contains Yunnanotherium and the extant Asian forms Tragulus and Moschiola as in Sánchez et al. (Citation2015a), (Citation2018). This relationship is supported by the presence of well developed (Tragulus-like) cristids, anteriorly triangular lower molars due to a well developed and straight pre-metacristid and pre-protocristid that contact mesially following the long central axis of the lower molars (see ), and poorly developed, almost non-existent Dorcatherium-fold. Clearly, the entire lineage of the living Asian tragulids is very ancient, with an evolutionary history that covers at least the last 19 million years ().

Figure 8. Calibrated phylogeny of Afrotragulus and its sister clade – Yunnanotherium + (Tragulus + Moschiola) –, with a representation of the chronological/geographical record of Afrotragulus in Africa and Asia. The interrogation mark refers to the Vihowa Formation ‘Dorcatherium’ cf. parvum cited by Antoine et al. (Citation2013). Image by IMS.

Figure 8. Calibrated phylogeny of Afrotragulus and its sister clade – Yunnanotherium + (Tragulus + Moschiola) –, with a representation of the chronological/geographical record of Afrotragulus in Africa and Asia. The interrogation mark refers to the Vihowa Formation ‘Dorcatherium’ cf. parvum cited by Antoine et al. (Citation2013). Image by IMS.

Recently, the holotype of A. megalomilos (AMNH 19310) was included by Guzmán-Sandoval and Rössner (Citation2019) in the hypodigm of Dorcatherium dehmi (Guzmán-Sandoval and Rössner Citation2019). However, when comparing AMNH 19310 with the rest of D. dehmi fossils figured by the authors (Guzmán-Sandoval and Rössner Citation2019; ) it is very clear that this taxon is very different from AMNH 19310, since it has fully developed M-structure, the post-protocrista is short and it lacks interlobular bridge, with the mesial and distal lobes fully contacting at the centre of the teeth.

The remains of Afrotragulus moralesi are very scarce. However, even if the species is currently represented by a single m3, the morphological, phylogenetic position, size and chronostratigraphical differences of A. moralesi with the remaining representatives in the genus Afrotragulus are clear, making it easily recognisable and well diagnosed. Accounting for the rarity of Afrotragulus findings and the previously mentioned differences, we find it safe to name this species, with the hope that future excavations at Hasnot 6 will produce more A. moralesi fossils to add to the hypodigm of the species.

There are two main lineages of Afrotragulus, one containing A. moruorotensis (Clade B, ) and the other containing A. parvus (Clade C, ). Clade C is diagnosed by the presence of almost aligned entoconid and hypoconid, and clusters Afrotragulus with lower molar dentition morphologically more primitive than that of Clade B. Hence, Clade C Afrotragulus possesses shorter molars that lack the elongated morphology and the horizontally extended cristids so characteristic of Clade B. Also, they have more rounded and less aligned cuspids than A. moruorotensis and A. megalomilos. However, A. moralesi has cristids more developed in height than A. parvus and A. akhtari, albeit maintaining the ‘short molar-type’ morphology and the lack of horizontal extension of the cristids. Also, this species developed a parallelism with clade B Afrotragulus, the presence in the m3 of a characteristically developed pre-hypoconulidcristid that forms a sort of buccal wall that joins to the post-hypocristid at mid-crown. Afrotragulus akhtari sp. nov. appears to be the most autapomorphic of all Afrotragulus, showing a closed third lobe of the m3, a parallelism shared with Yunnanotherium simplex and Siamotragulus bugtiensis within the selenodont clade of tragulids; a long and curved post-hypocristid that turns to contact the entoconid; and a strong posterior cingulid. On the other hand, clade B is diagnosed by possessing aligned and flattish lingual cuspids, no posterior cingulid and the aforementioned presence of an extended pre-hypoconulidcristid. These are derived forms with characteristically elongated molars in which the Afrotragulus morphology is more exaggerated, particularly in A. moruorotensis with its long m3 interlobular bridge and the flat lingual wall of the entoconid. In morphofunctional terms (cuspid height and morphology, cristid expansion, alignment of lingual wall and how all these parts are related), the construction of the lower molars of these forms is not far from that of certain pecorans such as e.g. basal bovids like Namacerus.

Due to the rarity of Afrotragulus fossils in the fossil record and the consequent lack of data, some of the branch supports within Afrotragulus are not high. The majority of the coded morphological data come from a single data subset, the lower molars, with the upper molar subset being complete in three species (A. moruorotensis, A. parvus and A. akhtari) and totally unknown in the remaining two (A. megalomilos and A. moralesi). The postcranial skeleton of Afrotragulus is very poorly known. Some putative Afrotragulus astragali have been described from the Namibian Sperrgebiet deposits in Southern Africa and from Rusinga Island in Kenya, with more citations of possible isolated postcranial remains of Afrotragulus coming from East Africa (Pickford Citation2001, Citation2002; Quiralte et al. Citation2008; Geraads Citation2010; Sánchez et al. Citation2010). However, this material poses some problems (see Materials section and discussion below). Thus, despite being totally resolved, the Afrotragulus ingroup tree will certainly improve if better fossil material including postcranial is found in the future.

We reject the hypothesis of an independent African lineage of Afrotragulus (as the ‘D’. parvum - ‘D’. moruorotensis African line proposed by Pickford (Citation2002))– since A. parvus and A. moruorotensis belong to different basal lineages. Very probably, Afrotragulus appeared in Asia around the middle early Miocene. Very shortly after, representatives of both basal lineages expanded their biogeographic ranges into the African continent – as part of the faunal interchanges that occurred during the early and middle Miocene (see e.g. Rögl Citation1999; Grossman et al. Citation2019 and references therein) – spanning from East to (probably) South Africa and thriving there until the final part of the early Miocene (). Tragulids were not the only ruminants to enter Africa during the early Miocene faunal interchanges, since both giraffomorphs (e.g. Propalaeoryx) and stem-bovidomorphs (e.g. Sperrgebietomeryx) are recorded at the same time as some of the oldest African tragulids (Morales et al. Citation1999, Citation2008; Sánchez et al. Citation2015b). We do not rule out the possibility of African Afrotragulus surviving into the middle Miocene (see Material and methods Pickford Citation2001, Citation2002; Geraads Citation2010), but we were unable to determine this in the current study.

In Asia, both basal branches persisted to the middle Miocene. However, only the most primitive Clade C persisted to the late Miocene. The Asian branches of Afrotragulus have considerably younger records than their African sister taxa, and this translates into two ghost lineages of more than 3 million years (). The basal lineages of Afrotragulus were established by 20 Ma (). Therefore, it is clear that a more ancient large radiation of tragulids must have occurred, probably during the Oligocene (Sánchez et al. Citation2010, Citation2015a, Citation2018). This early radiation produced the early–middle Miocene taxonomic diversity of tragulids evident in the Miocene fossil record. Although some hints of this early radiation were recently published (Mennecart et al. Citation2021), more discoveries and analyses are required to elucidate the particulars of this foundational event in tragulid evolution.

The maximum number of coexisting species of Afrotragulus is two species, first in the early Miocene (Africa) and then in the middle Miocene (Asia). If the middle Miocene African records of Afrotragulus are confirmed, then this would be the time of maximum species diversity of the genus. Future new findings will hopefully help to refine these data. Interestingly, Antoine et al. (Citation2013) cite Dorcatherium sp. cf. parvum in the latest early Miocene beds of the Vihowa Formation (Bugti Hills, Sulaiman Province, Pakistan; ca. 16–17 Ma) that would be the oldest Asian record of Afrotragulus. However, they do not describe or figure the fossils. Another tiny Siwalik tragulid described in the upper Chinji Fm. is ‘Dorcatherium’ minimus (West Citation1980), known from a single astragalus and two upper molars (H-GSP 1983 and H-GSP 2300) that are equivalent in size to those of A. parvus. ‘Dorcatheriumminimus lacks lingual cingulum – as already noted by Geraads (Citation2010) – and possesses fully developed post-protocrista. These characters distinguish ‘D’. minimus from Afrotragulus and suggest that this tiny tragulid is probably closer to the Tragulus lineage.

This discovery of new Asian Afrotragulus profoundly changes the evolutionary paradigm of the genus, unveiling an unexpected and surprisingly high specific diversity and a remarkable palaeogeographic extension, greatly extending the biochronologic range of Afrotragulus at least 13.5 million years from the early Miocene almost to the end of the Miocene (ca 13.5 million years), thus turning Afrotragulus into one of the most successful tragulid genera. Despite this, Afrotragulus fossils remain rare, and a great part of its anatomy remains unknown.

Palaeobiological insights: the influence of body size in driving diversification and diet evolution of Afrotragulus

Other than its highly derived lower molars, what first caught the eye about African Afrotragulus was the very small size of both A. parvus and A. moruorotensis, a feature remarked by Pickford (Citation2001) and Whitworth (Citation1958) in the first description of ‘Dorcatherium’ moruorotensis and ‘D’. parvum respectively. The new Asian Afrotragulus provides interesting new data about body size in Afrotragulus, and hence offers an opportunity for the analysis of this feature, its evolutionary consequences, and its palaeobiological implications related to diet and habitat preferences. Body size (calculated as body mass) profoundly affects life history of living organisms, as it is directly related to morphology, energy needs, physiology and ecology (with clear implications on both food choice, and resource and habitat partitioning; Demment and Van Soest Citation1985; Clauss et al. Citation2013; DeMiguel et al. Citation2014 and references therein). Body size estimates are essential for inferring the palaeobiology of extinct organisms (Campione and Evans Citation2012) and also underlie the diversity of feeding niches and digestive capacity in the Ruminantia (Demment and Van Soest Citation1985; Clauss et al. Citation2013; DeMiguel et al. Citation2014). In general, larger ruminants are able to meet their energetic requirements by eating more abundant foods of lower quality such as twigs or stems because they can process those in large quantities. By contrast, smaller ruminants require higher-quality parts (e.g. forbs, leaves, seeds or fruits) that are easier and quicker to digest in order to meet their energetic requirements (Clauss et al. Citation2013).

Extant tragulids (ranging 1.5–16 kg; Meijaard Citation2011) are all selective feeders that pick leaves, shoots, flowers and fruits from the forest floor. Unlike pigs and some ruminants (e.g. moschids and certain bovids), tragulids cannot rise on their hind limbs to reach food at a certain height level, restricting them to look for fallen fruits and leaves and to nip low-growing shoots (Meijaard Citation2011). Extant tragulids are mostly frugivorous (with additional browsing behaviour) that prefer forested habitats in which fruit is available more or less year round (Meijaard Citation2011; Clauss and Rössner Citation2014). Living tragulid density correlates with fruit abundance (Heydon and Bolloh Citation1997). For example, the lesser mouse deer Tragulus kanchil, one of the smallest living forms (1.5–2.5 kg; ) is mostly frugivorous, consuming seeds, shoots and mushrooms (Meijaard Citation2011). The African chevrotain Hyemoschus supplements its mostly frugivorous diet (approximately 70% of its food intake) with a small amount of animal protein, hunting for ants and small crabs, and eating carrion and fish. Animal protein appears to play no role in the Asian living species (Meijaard Citation2011). It is possible that the more primitive dentition (with more pointed cuspids) of the African chevrotain is better suited for processing small occasional animal prey than the fully selenodont molars of its Asian relatives. Some researchers argued that fossil tragulids exploited a wider trophic spectrum than extant forms, ranging from pure browsing to mixed feeding (Kaiser and Rössner Citation2007; Ungar et al. Citation2012). One hypothesis posits that after flourishing in the first half of the Miocene, in the middle Miocene tragulids could not compete against the rapidly radiating pecorans due to their different digestive physiology and reproductive life history (Clauss and Rössner Citation2014). As a result, tragulids only survived in tropical forested relic areas in which a mostly frugivorous diet is possible year round, places where the advantages of pecorans were not so decisive. This hypothesis proposes that the mostly frugivorous feeding habits of living tragulids do not represent the ancestral condition but a secondary adaptive constraint (Ungar et al. Citation2012; Clauss and Rössner Citation2014).

Our body size estimates show two different classes of Afrotragulus species. Although originally considered very small forms, Afrotragulus body size range is larger than the entire size spectrum of living tragulids (). Afrotragulus moruorotensis, A. parvus and A. akhtari were tiny – A. moruorotensis is smaller than the smallest living ruminant, the lesser mouse deer (and is instead similar to smallest species of macropods such as the nabarlek Peradorcas concinna). Afrotragulus parvus and A. akhtari are similar in size to the three smallest Tragulus (T. javanicus, T. kanchil and T. versicolor) and to Moschiola chevrotains (; size values from Meijaard Citation2011). At the other end of the range, Afrotragulus megalomilos and A. moralesi are larger forms similar in size to Hyemoschus aquaticus and the bovid Cephalophus dorsalis. Additionally, Yunnanotherium simplex may be considered as a very small species, similar to the bovid Cephalophus monticola and slightly larger than Yunnanotherium sister group, the living Asian tragulids (). Contrary to Afrotragulus, the clade containing the extant Asian forms apparently developed a small body size constraint early on at the root of the lineage, never producing larger forms. Other than Yunnanotherium and the two living genera, additional Miocene forms potentially belonging to this clade, namely the previously mentioned ‘D’. minimus and at least one undescribed form from the Siwaliks (AMNH collection, IMS pers. obs.), are also diminutive. If they indeed belong to the Yunnanotherium + ‘extant Asian’ clade, that adds support to the hypothesis of a synapomorphic basal small size constraint in the entire lineage of the living Asian tragulids.

Figure 9. Body size (in terms of body mass) distribution in Afrotragulus spp. and its sister group, the extant Asian lineage (including its extinct member Yunnanotherium simplex). The African chevrotain Hyemoschus aquaticus is included here for comparison purposes between Afrotragulus spp. and all the living tragulids. The body size data of the extant forms come from Meijaard (Citation2011). The dashed boxes represent body mass ranges. Three living taxa have unknown body size and are not included here: Moschiola kathygre, Tragulus williamsoni and T. nigricans (Meijaard Citation2011). It is worth noting that the body size of A. moralesi was calculated using the only known m3 (PUPC 13/323), and this dental piece offers higher size estimations than the m1 does. Hence, a more realistic figure for the body mass of this species would fall within the 15–20 kg segment. Hyemoschus and Tragulus art by Marco Ansón. Composed image by IMS.

Figure 9. Body size (in terms of body mass) distribution in Afrotragulus spp. and its sister group, the extant Asian lineage (including its extinct member Yunnanotherium simplex). The African chevrotain Hyemoschus aquaticus is included here for comparison purposes between Afrotragulus spp. and all the living tragulids. The body size data of the extant forms come from Meijaard (Citation2011). The dashed boxes represent body mass ranges. Three living taxa have unknown body size and are not included here: Moschiola kathygre, Tragulus williamsoni and T. nigricans (Meijaard Citation2011). It is worth noting that the body size of A. moralesi was calculated using the only known m3 (PUPC 13/323), and this dental piece offers higher size estimations than the m1 does. Hence, a more realistic figure for the body mass of this species would fall within the 15–20 kg segment. Hyemoschus and Tragulus art by Marco Ansón. Composed image by IMS.

Among extant ruminants, species with a similar body mass to that of Afrotragulus spp. and Yunnanotherium simplex are all browsers. Such small size might have acted as a serious physiological and energetic constraint on the smaller Afrotragulus, and may have limited this species’ ability to expand dietary niches (i.e. preventing transition from browsing to intermediate feeding), and subsequently live outside of habitats capable of delivering the necessary amount of fruit year round. Hence, at least the smallest forms (i.e. 1.5–3 kg) A. parvus, A. akhtari and (especially) A. moruorotensis were probably very limited by the time and energy requirements of finding scantier and unevenly distributed high-quality (easy-to-digest) foods required by a low digestive capacity, as observed in small animals (Demment and Van Soest Citation1985). The smaller species of Afrotragulus were probably highly frugivorous selective feeders consuming mostly small fallen fruits and seeds rather than typical browsers. Larger body sizes (>10 kg) were attained twice in parallel in both lineages of Afrotragulus by A. megalomilos (middle Miocene) and A. moralesi (late Miocene). This probably enabled larger Afrotragulus to escape the energetic constraints associated with very small body sizes, permitting these species to potentially exploit a broader (more mixed-feeder) diet, taking advantage of their derived molar dentition, which morphofunctionally mimicked that of pecorans. Increased plasticity in diet possibly facilitated these larger forms to adopt new lifestyles and enter new niche spaces to survive the climate-induced environmental changes of their respective epochs, A. megalomilos to the mid-Miocene cooling and A. moralesi to the drying and increased seasonality of the late Miocene (Quade and Cerling Citation1995), as occurred with some large-bodied cetartiodactyls in less productive habitats of North America (Pound et al. Citation2012). A previous analysis of dental texture of early Miocene small tragulid species from Kenya (included A. parvus; Ungar et al. Citation2012) found that they were mixed feeders apparently contrasting with our interpretation that the bulk of Afrotragulus spp. (A. moruorotensis, A. parvus and A. akhtari) had mainly frugivorous diets, but instead agree with the more opportunistic diets proposed for A. megalomilos and A. moralesi. Unfortunately, that study does not provide data for extant tragulid species, nor for that matter are individual modern species demonstrated in the study at all, making comparisons with individual modern species impossible. Also, the study does not include A. moruorotensis. While more research is needed to examine dietary adaptations in other Afrotragulus and other extinct tragulids, we agree with Ungar et al. (Citation2012) that extinct African tragulids exploited a broader spectrum of foods during the early Miocene, and our new discoveries demonstrate the same for extinct Asian Afrotragulus.

Conclusion

We describe three new species, Afrotragulus akhtari, A. megalomilos and A. moralesi, from the upper Middle Miocene sites of Babri Wala (middle Chinji Formation) and B 51 (lower Chinji Formation), and the Upper Miocene site of Hasnot 6 in the Siwaliks Group (Pakistan).

Using a phylogenetic analysis of Maximum Parsimony we recover two clades within the genus Afrotragulus, each one with Asian and African representatives, rejecting the hypothesis of independent African and Asian lineages of Afrotragulus and revealing a complex evolutionary history. One of these clades includes A. parvus, A. akhtari sp. nov. + A. moralesi sp. nov, whereas the other one includes A. moruorotensis and A. megalomilos sp. nov. Afrotragulus is the sister-group of a clade containing the extant Asian Moschiola and Tragulus, implying that the lineage of the living Asian tragulids is very ancient, beginning at least 19 million years ago and turning Moschiola and Tragulus into important fossil relics that require strong conservation efforts. Soon after the genus Afrotragulus appeared in Asia during the middle part of the early Miocene, representatives of both Afrotragulus lineages entered Africa as part of the faunal interchanges that occurred between Asia and Africa during the early to middle Miocene.

Our body size estimates show that the majority of the species of Afrotragulus were small ruminants within the lower range of the extant Tragulus spp. As with living small Tragulus, these Afrotragulus (A. moruorotensis, A. parvus and A. akhtari) were highly limited in energetic terms by their tiny size and were probably selective floor-level feeders, mainly frugivorous with additional, more or less accentuated, browsing behaviour. Hence, these forms probably inhabited habitats in which the access to fallen fruit was more or less constant year round. However, both Afrotragulus lineages include representatives larger than 10 kg (A. megalomilos and A. moralesi). It is possible that these larger Afrotragulus species were capable to have a broader trophic spectrum than that of the smaller species, and consequently invade different habitats, which is supported by their derived dentition (e.g. in morphofunctional terms the molars of A. megalomilos were not much different from those of a basal bovid). Such adaptations may have allowed these larger species to adopt new lifestyles and enter new niche spaces to survive the climate-induced environmental changes of their respective epochs. Further research is needed (including meso- and microwear and if possible isotopic analyses), but this is in line with previously proposed hypotheses that Miocene tragulids had a wider trophic spectrum than the living forms, not only in the African early Miocene ecosystems but also in Miocene Asia.

The description of these new Asian Afrotragulus dramatically changes the original status of the genus as enigmatic small tragulids with very limited specific diversity and relatively short chronostratigraphic range, instead uncovering a relatively high specific diversity, a broad palaeogeographic distribution, a body size diversity that encompasses that of all living tragulids, and an extended geochronological range that stretches from the middle early Miocene (ca. 20–19 Ma) to the final part of the Miocene (ca 6.5 million years) during 13.5 million years. Hence, Afrotragulus is unveiled here as one of the most successful representatives of the Tragulidae.

Abbreviations

AMNH American Museum of Natural History, New York

BMNH and NHM Natural History Museum, London

GSN Geological Survey of Namibia, Windhoek, Namibia

MNCN-CSIC Museo Nacional de Ciencias Naturales-CSIC, Madrid

MAUV Museo Anatómico de la Universidad de Valladolid, Valladolid

MNHN Muséum National d’Histoire Naturelle, Paris

NMK National Museums of Kenya

PUPC Punjab University Palaeontological Collection

SMNS Staatliche Museum für Naturkunde Stuttgart, Stuttgart; UM, Uganda Museum, Kampala, Uganda

Supplemental material

SA_NCHEZ_ET_AL_METRICS.doc

Download MS Word (50 KB)SA_NCHEZ_ET_AL_METRICS.doc

SA_NCHEZ_ET_AL_ASIAN_AFROTRAGULUS_CHARACTER_LIST.docx

Download MS Word (503.5 KB)SA_NCHEZ_ET_AL_ASIAN_AFROTRAGULUS_CHARACTER_LIST.docx

SA_NCHEZ_ET_AL_ASIAN_AFROTRAGULUS_73CHAR.nex

Download (10.3 KB)SA_NCHEZ_ET_AL_ASIAN_AFROTRAGULUS_73CHAR.nex

Acknowledgments

IMS thanks Job Kibii (National Museums of Kenya) for granting us permission to photograph the tragulid material (‘D’. pigotti) from Maboko, and specially acknowledges Marta Pina (Manchester University, UK) for taking the actual pictures that allowed us to code the morphological characters of that material. IMS and VQ want to thank the AMNH curators of the Departments of Palaeontology and Mammalogy for accessing us grant over the years to study the collections in their care, including the Siwaliks Group and extant tragulid collections. We also thank Helke Mocke (Geological Survey of Namibia, Windhoek), Sarah Musalizi (Uganda Museum, Kampala), Pip Brewer (NHM, London) and Christine Argot (MNHN, Paris) for providing access to the fossil collections of their respective institutions. Thanks to Denis Geraads (CNRS, France) for the pictures of the Afrotragulus hemimandible KNM SO 1345 (Songhor, Kenya). We also acknowledge Nikos Solounias (College of Osteopathic Medicine, New York Institute of Technology, USA) for his help with the Greek language and the naming of A. megalomilos sp. nov. Thanks to Pablo Peláez-Campomanes (MNCN-CSIC) for his help with the stratigraphy, to Martin Pickford (CNRS) for providing us of many Miocene African tragulids to study and make us part of the Namibia Palaeontological Expedition, and to Marco Ansón for giving us permission to use his illustrations of Tragulus and Hyemoschus. We thank John C. Barry (Harvard University, USA) for providing the palaeomagnetic data of site B 51, and Sergio Almécija and Chris Gilbert (both AMNH) for their invaluable help with the new photographs of AMNH 19310 that have been used in the present work. SGA is grateful to Mr. Allah Yaar Khan and his family for their hospitality during fieldwork, and also to Mr. Muhammad Khan for his guidance on the field in the Kohtehra area. Finally, we want to thank the reviewers of this work (Ari Grossman and two anonymous reviewers) for their comments that greatly improved the quality of the manuscript.

This work pays homage to the career of Prof. Jorge Morales (MNCN-CSIC). IMS, VQ and DDM want to thank Jorge for so many years of mentorship, friendship and, of course, fossils. Thanks for supervising our doctoral theses and in the process instilling in us the love for the evolution of the fascinating and challenging ruminants. Thanks for so many excavations and discussions. It has been—and always shall be— an honour to work with you.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

This work and the nomenclatural acts it contains have been registered in Zoobank: 7FC6D555–D653–4632–A8BE–FCB4A88A8482

Supplementary material

Supplemental data for this article can be accessed here.

Additional information

Funding

This work was supported by the Spanish Government Agencia Estatal de Investigación–European Regional Development Fund of the European Union (projects MINECO/FEDER-UE; CGL2016-76431-P and CGL2017-82654-P, AEI/FEDER EU), the Generalitat de Catalunya (CERCA Programme), the Spanish Research Projects I+D+i PID2020-117289GBI00 and I+D+i PID2020-117289GBI00 funded by MCIN/AEI/10.13039/501100011033, and the Research Groups CSIC 64 1538 and CAM-UCM 910607;Generalitat de Catalunya CERCA programme;MCIN/AEI/10.13039/501100011033/;

References

Reprints and Corporate Permissions

Please note: Selecting permissions does not provide access to the full text of the article, please see our help page How do I view content?

To request a reprint or corporate permissions for this article, please click on the relevant link below:

Academic Permissions

Please note: Selecting permissions does not provide access to the full text of the article, please see our help page How do I view content?

Obtain permissions instantly via Rightslink by clicking on the button below:

If you are unable to obtain permissions via Rightslink, please complete and submit this Permissions form. For more information, please visit our Permissions help page.