Introduction
Middle to late Pleistocene vertebrate cave faunas dominated by isolated teeth are common in southwestern China and the Indochinese peninsula (e.g., Pei, Reference Pei1935; Olsen and Ciochon, Reference Olsen and Ciochon1990; Ciochon and Olsen, Reference Ciochon and Olsen1991; Long et al., Reference Long, de Vos and Ciochon1996; Bacon et al., Reference Bacon, Demeter, Schuster, Long, Thuy, Antoine, Sen, Nga and Huong2004, Reference Bacon, Demeter, Rousse, Long, Duringer, Antoine and Thuy2006, Reference Bacon, Duringer, Antoine, Demeter, Shackelford, Sayavongkhamdy and Sichanthongtip2011, Reference Bacon, Westaway, Antoine, Duringer, Blin, Demeter and Ponche2015; Lenoble et al., Reference Lenoble, Zeitoun, Laudet, Seveau, Doy Asa, Pautreau, Coupey, Zeitoun and Rambault2006 Fan et al., Reference Fan, Shao, Bacon, Liao and Wang2022; Yao et al., Reference Yao, Fan, Bae, Tian, Liang, Chen and Zhang2023). These fossils typically occur in iron and quartz-rich breccias on the walls, floors, and ceilings of cave complexes (Ciochon and Olsen, Reference Ciochon and Olsen1991; Lenoble et al., Reference Lenoble, Zeitoun, Laudet, Seveau, Doy Asa, Pautreau, Coupey, Zeitoun and Rambault2006; Yao et al., Reference Yao, Fan, Bae, Tian, Liang, Chen and Zhang2023).
The accumulation of vertebrate fossils in caves has been attributed to various taphonomic agents, including fluvial floods entering cave mouths (e.g., Duringer et al., Reference Duringer, Bacon, Sayavongkhamdy and Thuy2012), the actions of cave-dwelling predators carrying prey back to their caves (e.g., de Ruiter and Berger, Reference de Ruiter and Berger2000), inadvertent falls into a natural death trap (Val et al., Reference Val, Dirks, Backwell, d’Errico and Berger2015), the action of scavenging porcupines (e.g., Alexander, Reference Alexander1956; Maguire, Reference Maguire1976; Brain, Reference Brain, Behrensmeyer and Hill1980, Reference Brain1981; Duthis and Skinner, Reference Duthis and Skinner1986; O’Regan et al., Reference O’Regan, Kuman and Clarke2011; Horwitz et al., Reference Horwitz, Cohen, Wieckowski, Mienis, Baker and Jastrzebska2012) or a combination of multiple factors (Ciochon and Olsen, Reference Ciochon and Olsen1991; Fan et al., Reference Fan, Shao, Bacon, Liao and Wang2022).
This contribution provides a taphonomic assessment of the Làng Tráng cave complex in northern Vietnam. Taphonomic analyses are essential for better understanding the fossil record preserved within these caves. Porcupine scavenging, in particular that attributable to the large porcupine Hystrix cf. H. kiangsenensis, is identified as the primary taphonomic agent responsible for introducing fossil vertebrate material into the caves, as well as restricting the nature of the material preserved. Porcupine scavenging, particularly that attributed to Hystrix spp., has been identified as a significant contributor to faunal accumulations in several caves in the Vietnam region (Ciochon and Olsen, Reference Ciochon and Olsen1991; Ciochon et al., Reference Ciochon, Olsen and James1991; Bacon et al., Reference Bacon, Demeter, Schuster, Long, Thuy, Antoine, Sen, Nga and Huong2004, Reference Bacon, Demeter, Rousse, Long, Duringer, Antoine and Thuy2006, Reference Bacon, Westaway, Antoine, Duringer, Blin, Demeter and Ponche2015; Fan et al., Reference Fan, Shao, Bacon, Liao and Wang2022). The level of porcupine behavior in each of these caves and cave complexes varies, from accumulations that originated through physical transport and were modified by porcupines, to assemblages that resulted solely or dominantly from the activities of porcupines.
This study is intended to better understand the implications of porcupines as taphonomic agents that contribute to and alter the fossil record. Our goal is to clearly identify the diagnostic attributes of porcupine-generated, cave-hosted vertebrate faunas; isolate the attributes that indicate severe porcupine-generated biogenic modification of teeth and bone; discuss the significance of porcupines in framing the Southeast Asian Pleistocene fossil record; and delineate how porcupine-generated assemblages can be differentiated from other types of cave faunas.
Terminology
This contribution focusses on bone modification traces (gnaw and bite marks) preserved on fossil teeth and bones from both an ichnological and taphonomic perspective. Some ichnologists differentiate the terms “mark” and “trace” to separate out structures produced by physical and biological means (e.g., Ekdale et al., Reference Ekdale, Bromley and Pemberton1984; Vallon et al., Reference Vallon, Rindsberg and Martin2015). We avoid this restriction here, as this invokes a priori assumptions in terms of the origin of the features and is contrary to convention used in the greater ichnological, biological, zoological, anthropological, archaeological, medical, and forensic literature (Zonneveld et al., Reference Zonneveld, Fiorillo, Hasiotis and Gingras2022).
We differentiate gnaw marks (or scrape marks) from bite marks herein, in that the former are emplaced with the force directed primarily perpendicular to the axis of the bone resulting in punctures and crushing and the latter are emplaced when the force of the tooth moves parallel or subparallel to the surface of the bone, resulting in grooves.
Trace fossils attributed to vertebrates on bone include both traces attributed to biting as well as those attributed to tooth scraping/biting. Ichnogenera interpreted as bite marks include Mandaodonites (now considered nomen erratum), Heterodontichnites, Nihilichnus, and Brutalichnus (Cruickshank, Reference Cruickshank1986; Mikulás et al., Reference Mikulás, Kadlecová, Fejfar and Dvorák2006; Rinehart et al., Reference Rinehart, Lucas and Spielmann2006; Zonneveld, Reference Zonneveld2022), although Brutalichnus was considered nomen dubium by Wisshak et al. (Reference Wisshak, Knaust and Bertling2019). These traces comprise solitary or grouped jagged to smooth holes in bone (Cruickshank, Reference Cruickshank1986; Mikulás et al., Reference Mikulás, Kadlecová, Fejfar and Dvorák2006; Rinehart et al., Reference Rinehart, Lucas and Spielmann2006). Trace fossils attributed to teeth gnawing or scraping bone include Linichnus, Knethichnus, and Machichnus (Mikulás et al., Reference Mikulás, Kadlecová, Fejfar and Dvorák2006; Jacobsen and Bromley, Reference Jacobsen and Bromley2009).
The biogenic modification features discussed occur on teeth and bones of a diverse suite of mammal fossils. These fossils are discussed using conventional vertebrate paleontology terminology. “Left” and “right” are from the perspective of the organism being discussed. Tooth abbreviations include “I/i” (incisor), “C/c” (canine), “P/p” (premolar), and “M/m” (molar). Uppercase abbreviations denote teeth from the maxilla and premaxilla and lowercase abbreviations denote teeth from the dentary.
We introduce two new terms herein. “Faceting” denotes alteration of bone and tooth root surfaces through mechanical bioerosion (most typically gnawing) to produce planar to gently curved facets on the bone. The term “wedging” is introduced for instances in which the roots of teeth are gnawed to the point that the roots are reduced to pyramidal wedges.
Study area and fossil occurrence
The Làng Tráng cave complex occurs approximately 100 km southwest of the center of Hanoi and 35 km northeast of the border with Laos (Fig. 1A). The caves lie ∼250 m east and southeast of a bend in the Sông Mã (Mã River) and 2 km southwest of the village of Cành Nàng on the eastern side of National Road 217 (Fig. 1B). The Sông Mã has an elevation of ∼58 m above sea level in the vicinity of the caves, with the adjacent highlands rising ∼150 m above the river. The Làng Tráng cave complex consists of four main caves with several smaller associated openings. Làng Tráng I, II, and III occur in close proximity, with Làng Tráng IV occurring 30 m to the south of the north cluster (Fig. 1C).
Location of the Làng Tráng cave complex, northern Vietnam. (A) Map of Vietnam and surrounding area showing the locations of Pleistocene cave-hosted vertebrate faunas on the Indochinese peninsula. (B) Map showing the position of the Làng Tráng cave complex east of the Sông Mã and southwest of the village of Cành Nàng, at the base of the heavily forested Lâm Xa ridge. (C) Detail map showing the location of the Làng Tráng caves on the east side of National Road 217 and the proximity of the cave complex to the Sông Mã. Scale bar is 200 m long.
Upstream of the study area, the Sông Mã is sourced in the passes through a structurally complex region that includes Lower Paleozoic sedimentary strata and a variety of intrusive igneous bodies. The Làng Tráng cave complex occurs in the Ba Thuoc tower karst district, near the base of an uplifted and tilted ridge of Devonian limestone that terminates at the Sông Mã (Fig. 1B).
All four caves occur at the same approximate elevation, approximately 20 m above the surface of the modern Sông Mã. Làng Tráng I and III are relatively shallow (more overhangs than caves), whereas Làng Tráng II and IV (Fig. 2) are deeper and more complex. Làng Tráng II was modified for use as a bomb shelter and weapons depot by the North Vietnamese Army during the war and thus has a brick wall with a metal door extending across its front (Figs. 2A and B and 3A).
Sketches of Làng Tráng caves. After Ciochon and Olsen (Reference Ciochon and Olsen1991). (A) Longitudinal profile through Làng Tráng II showing the wall built by the North Vietnamese Army (NVA) and both the cave front and center chamber. (B) Plan view sketch of Làng Tráng II showing the front region, separated from the center chamber by the NVA wall and the deep rear chamber. (C) Longitudinal profile through Làng Tráng IV showing the location of the upper and lower chambers, the disturbed Hoabinhian burial site, and the excavation pit on the cave floor. (D) Cross-sectional profile of Làng Tráng IV showing the upper and lower chambers
Fossiliferous, yellow-orange to brick-red breccia occurs on the floors, walls, and ceilings of the caves (Figs. 3D and E and 4A–C). The matrix of the breccia consists of iron oxide (>80%) and angular to subangular quartz silt and sand (<20%). The breccia composition in Làng Tráng I–III and the lower chamber of Làng Tráng IV is identical, whereas the breccia in the upper chamber of Làng Tráng IV is enriched in iron oxide, has less quartz, and includes clay (Ciochon and Olsen, Reference Ciochon and Olsen1991; Ciochon et al., Reference Ciochon, Olsen and James1991). The breccia deposits are massive (unbedded) and preserve no evidence of physical sedimentary structures. They are characterized by numerous calcite veins. Lithoclasts in the breccia consist of angular limestone clasts, fragments of flowstone, and breccia intraclasts (Fig. 4B and C). Fossils in the breccia occur in random orientations and are sporadically distributed throughout the breccia. No evidence of heavy mineral, bone, or teeth concentration lags or lenses was observed.
Photographs of Làng Tráng caves. (A) Photograph of the North Vietnamese Army (NVA) wall built across the front of Làng Tráng II. (B) View to the south from the front of Làng Tráng II. (C) View of the Ba Thuoc tower karst with Làng Tráng IV near the base (arrow). (D) View from the Làng Tráng IV inner chamber toward the door in the NVA wall (rectangular light area). E. View of a small branch of the cave on the wall of the inner chamber of Làng Tráng II showing the breccia (base) and the Devonian limestone (top).
Photographs of Làng Tráng caves. (A) Blind side tunnel in Làng Tráng II showing red-brown cave breccia and gray limestone walls. (B) Photograph of breccia in Làng Tráng II ceiling showing abundant lithoclasts, common teeth, and scattered flowstone intraclasts. (C) Breccia in the wall of Làng Tráng IV illustrating the complex nature of the cave fill. The breccia rests upon an irregular surface with numerous, small, breccia-filled pipes in the basal part of the photograph (arrows). The breccia in the upper two-thirds of the photograph consists of numerous breccia intraclasts within a breccia matrix.
Fossiliferous breccia at Làng Tráng have been dated as middle to late Pleistocene (Ciochon and Olsen, Reference Ciochon and Olsen1991; Fig. 5). The breccias were dated using electron spin resonance (ESR) resulting in preliminary dates of 480,000 ± 40,000 BP for Làng Tráng I, 285,000 ± 24,000 BP for Làng Tráng II and 146,000 ± 2000 BP for the lower breccia in Làng Tráng IV (Ciochon and Olsen, Reference Ciochon and Olsen1991; Fig. 5).
Age of Pleistocene Indochinese cave faunas. Ages obtained from a variety of resources (Ciochon and Olsen, Reference Ciochon and Olsen1991; Schwartz et al., Reference Schwartz, Long, Cuong, Kha and Tattersall1995; Ciochon et al., Reference Ciochon, Long, Larick, González, Grün, de Vos, Yonge, Taylor, Yoshida and Reagan1996; Tougard et al., Reference Tougard, Jaeger, Chaimanee, Suteethorn and Triamwichanon1998; Bacon et al., Reference Bacon, Demeter, Schuster, Long, Thuy, Antoine, Sen, Nga and Huong2004)
Material and methods
This article is focused on fossils that were collected in the Làng Tráng caves on several expeditions in the late 1980s and early 1990s (Ciochon and Olsen, Reference Ciochon and Olsen1991; Ciochon et al., Reference Ciochon, Olsen and James1991; Long et al., Reference Long, de Vos and Ciochon1996). Blocks of fossiliferous breccia were removed from the cave using a portable rock saw, and the fossils were removed from the matrix using a hammer and chisel. Blocks of matrix were reduced down to granule-sized sediment to minimize chances of overlooking smaller teeth. The original collectors focused on collecting fossils that could be identified as an individual element or to a taxon on the family level. Unidentifiable bone fragments were not collected on the original Làng Tráng expeditions. These fossils are archived at the Vietnam Institute of Archaeology (VIA) in Hanoi, Vietnam.
The Làng Tráng collection includes very rare postcranial elements, including several unidentifiable long bone shafts, a single ungulate humerus, and several vertebral centra. Several mandibles and skull fragments, and one complete ursid skull, also occur. During a visit to the Làng Tráng cave complex in 2020 we identified numerous teeth (primarily suid and cervid) in the matrix in the walls of the cave but observed no other bone fragments.
With the exception of purported Gigantopithecus teeth reported by Lopatin et al. (Reference Lopatin, Golovachev, Serdyuk, Maschenko, Vislobokova, Dac, Phuong, Parkhaev and Syromyatnikova2022a), all taxa discussed herein are extant, with well-known body size distributions. “Body size” refers to categories established by Bunn (Reference Bunn1986) and Bunn et al. (Reference Bunn, Bartram and Kroll1988) for African prey animals, emended by Sahnouni et al. (Reference Sahnouni, Rosell, van der Made, Vergés, Ollé, Kandi, Harichane, Derradji and Medig2013) to include all mammals. We emend this framework herein to include mammals between 5 and 100 kg as small and restrict very small mammals to those below 5 kg. Thus, the size classifications used herein include size class 1: very small size (0–5 kg); size class 2: small (∼5–100 kg), further broken down into 2a (5–20 kg) and 2b (20–100 kg); size class 3: medium (∼100–300 kg); size class 4: large (∼300–1000 kg); and size class 5: very large size (>1000 kg).
The Làng Tráng collection is overwhelmingly dominated by isolated teeth. The present assessment focused on teeth with complete crowns, although fragmentary material was also common. More recent field collections from the Làng Tráng caves were made in 2020 and are archived in the collections of the Paleontological Institute (PIRAS) at the Russian Academy of Sciences (Lopatin, Reference Lopatin2020a, Reference Lopatin2020b, Reference Lopatin2020c; Lopatin et al., Reference Lopatin, Maschenko, Vislobokova, Serdyuk and Dac2021, Reference Lopatin, Golovachev, Serdyuk, Maschenko, Vislobokova, Dac, Phuong, Parkhaev and Syromyatnikova2022a, Reference Lopatin, Maschenko and Dac2022b). These collections are used in discussion of species diversity but were not assessed taphonomically and are not included in discussions of bone modification. In total, vertebrate fossils recovered from the Làng Tráng cave complex include 37 identified mammalian species within 32 genera, 1 chelonian genus, and at least 1 crocodilian genus (Ciochon and Olsen, Reference Ciochon and Olsen1991; Long et al., Reference Long, de Vos and Ciochon1996; Lopatin, Reference Lopatin2020a, Reference Lopatin2020b, Reference Lopatin2020c; Lopatin et al., Reference Lopatin, Maschenko, Vislobokova, Serdyuk and Dac2021, Reference Lopatin2022; Table 1).
Mammalian fauna of the Làng Tráng cave complex, northern Vietnam. Numbers in bold are totals for each mammalian order.
a Taxa with percentages indicated are archived at the Vietnam Institute of Archeology (VIA) in Hanoi, Vietnam, and formed the basis of this study. Taxa denoted by asterisks are reported by a variety of other sources and are archived in Moscow (Lopatin Reference Lopatin2020a, Reference Lopatin2020b, Reference Lopatin2020c, Reference Lopatin2022; Lopatin et al., Reference Lopatin, Maschenko and Dac2022b).
Mammals from Làng Tráng Cave complex, northern Vietnam, organized by size, illustrating the percentage of clearly gnawed, uncertain, and ungnawed teeth (with complete crowns) of each taxon.
a Ab (= relative abundance): R (rare) indicates taxa represented by less than 10 individual teeth, P (present) indicates taxa present represented by 10–100 specimens, C (common) indicates taxa represented by 101–250 specimens, and VC (very common) indicates taxa represented by more than 251 individual teeth.
b Gnawed teeth illustrate clear Machichnus, faceting, or wedging. Ungnawed teeth retained complete, unaltered roots and teeth labeled uncertain exhibited a lack of roots, but no clear gnaw marks. Taxa indicated in bold all exhibit extensive bone modification attributable to porcupines.
All available fossils were assessed for evidence of perimortem and postmortem bone modification, including rounding or breakage due to physical transport and crushing, gnawing, or biting due to predation or scavenging. Bones and teeth were examined under low-angle polarized LED illumination at the VIA. Selected samples were examined under a binocular microscope to confirm macroscopic observations. Following Behrensmeyer (Reference Behrensmeyer1978), all bone was assessed for evidence of weathering in the form of cracks, flaking, and splintering.
Isolated teeth, which comprise most of the fossils collected, were graded according to degree of biogenic modification. The five bins used were: (1) unmodified (roots present, not gnawed), (2) roots broken (no evidence of gnawing); (3) roots partially gnawed; (4) roots absent (completely gnawed); and (5) roots absent (unclear). This was simplified into three categories “gnawed,” “uncertain,” and “ungnawed” (Table 2), with the “gnawed” category lumping bins 3 and 4, the “uncertain” category including bins 2 and 5, and the “ungnawed category” including bin 1. Only teeth with complete crowns were included in this assessment, as many partial teeth, although perhaps diagnostic as to taxon, were broken during the collection process and could not be accurately assessed for the presence or absence of biogenic modification. Although teeth attributed to Homo sapiens occur within the collection of teeth from Làng Tráng IV, these were not included in analyses in this study, as they are likely associated with a much later (Hoabinhian) occupation of the cave (Ciochon, Reference Ciochon, Fleagle, Shea, Grine, Baden and Leakey2010).
Fossils were photographed using an Olympus OM-D E-M1 Mark II digital camera with 20.4 megapixel resolution using either an Olympus 12–40 mm zoom lens or an Olympus 60 mm macro lens. The built-in bracketed focus-stacking option was used to ensure that the entire surface of three-dimensional fossils was in focus. Each image consists of nine stacked images that were automatically amalgamated by onboard software into a single image in real time. Direct, low-to-moderate angle incident lighting, using 1150 Lux LED flood lamps set at ∼0.5 m, was used in all fossil photographs. Samples with clear evidence of vertebrate modification were analyzed and imaged under a scanning electron microscope (SEM) using a Zeiss Sigma 300 VP-FESEM.
Results
Fossil bones and teeth were fully disarticulated and occurred in random orientations. Isolated teeth are abundant (>99.9% of fossils collected), whereas postcrania are sparse. The very few postcranial bones collected preserve strong evidence of postmortem modification by scavengers (Fig. 6). This modification includes discrete, narrow (0.2–0.5 mm) U-shaped grooves consistent with the ichnotaxon Machichnus bohemicus, which were observed on a single long bone shaft (Fig. 6B and C). In contrast, clustered, overlapping, parallel to subparallel, broad (2–7 mm) U-shaped grooves consistent with M. multilineatus (Fig. 6B, D, and E) are much more common. Machichnus multilineatus were identified on all postcrania preserved. Bones with extensive M. multilineatus commonly exhibit a faceted appearance, formed where rounded bone was peneplaned by persistent gnawing. Long bone shaft fragments also occur in the Làng Tráng breccia, some of which exhibit isolated to clustered M. bohemicus.
Rusa unicolor left femur from Làng Tráng II. (B) Close-up of the proximal end showing Machichnus multilineatus and M. bohemicus. (C) Close-up of a section of the shaft with dense M. bohemicus. (D) Close-up of a portion of the proximal end of the femur showing dense M. multilineatus. (E) Close-up of the femoral showing faceting on the head and proximal shaft.
Mandibles and skulls are also rare at Làng Tráng. The few that do occur are either representative of the very small size category 1a taxa (e.g., Rhizomys sp., Leopoldamys sabanus, Chodsigoa caovansunga, C. hoffmanni, and Ia io; specimens herein and illustrated in Lopatin [Reference Lopatin2020a, Reference Lopatin2022]) or are extensively modified small- to medium-sized mammals (size categories 2b–3) mammals, most typically with Machichnus multilineatus (e.g., Figs. 7A and C and 8A and B). The gnawing commonly occurs primarily on one side of the mandible with the other side largely untouched (Figs. 7A–C and 8A and B. The sole skull preserved in the study material is assigned to Ursus thibetanus. It is approximately half complete, preserving most of the right side of the skull, with the exception of the zygomatic, the premaxilla, and the anterior portion of the maxilla. The skull retains considerable matrix on it and could not be confidently assessed for evidence of biogenic bone modification.
As discussed earlier, the bulk of Làng Tráng collections are overrepresented by isolated teeth. The VIA collection analyzed, not including Homo sapiens, consists of a total of 2862 teeth representing 29 species of mammals within 6 orders, collected from the 4 caves (Table 1). Of these, 6 species (Sus scrofa, Muntiacus muntjak, Rusa unicolor, Hystrix cf. H. kiangsenensis, Macaca sp., and Pongo pygmaeus) comprise over 87% of the teeth collected and 11 species are represented by 10 or fewer specimens. The PIRAS collection (not studied herein) did not report exact numbers of fossils but does add several species to the Làng Tráng fauna (Lopatin, Reference Lopatin2020a, Reference Lopatin2020b, Reference Lopatin2020c, Reference Lopatin2022; Lopatin et al., Reference Lopatin, Maschenko, Vislobokova, Serdyuk and Dac2021, Reference Lopatin, Golovachev, Serdyuk, Maschenko, Vislobokova, Dac, Phuong, Parkhaev and Syromyatnikova2022a, Reference Lopatin, Maschenko and Dac2022b). Mammal teeth at Làng Tráng represent all size categories of mammals with particularly strong representation of small taxa and significant underrepresentation of very small mammals (Tables 1 and 2).
Bony material (postcrania, mandibles, and skulls) all exhibit some light cracking parallel to the fiber structure of each element. In the Rusa unicolor femur, the cracking occurred parallel to the long axis of the bone shaft (Fig. 6). In dentary bones this occurs parallel to the long axis of the mandible (Fig. 7). In the Ursus thibetanus skull, this occurred parallel to the long axis of the skull. This level of weathering is consistent with Behrensmeyer weathering stage 1 (Behrensmeyer, Reference Behrensmeyer1978). Similarly, the teeth preserved at Làng Tráng exhibit low levels of weathering, restricted primarily to longitudinal fractures in tooth roots and faint fractures in the tooth enamel (Figs. 7–12).
Rusa unicolor right mandible from Làng Tráng showing strong biogenic modification. (A) Labial side with minimal evidence of biogenic modification. (B) Lingual side, with extensive biogenic modification in the form of continuous Machichnus multilineatus along the entire mandible. (C) Close-up of a small portion of the biogenically modified portion of the mandible.
Mandibles showing evidence of biogenic alteration. (A) Two-tooth mandibular fragment of Pongo pygmaeus with /p3 and /p4. The labial surface has been biogenically modified such that the roots are exposed (arrows). (B) Two-tooth mandibular fragment from Rusa unicolor. Note the grooves on the lingual surface (arrows)
Lower right molar of Panthera tigris. Specimen LT93-C2B5-005. (A) Labial view. (B) Lingual view. (C) Close-up of the base of the tooth roots showing tooth faceting resulting from intensive biogenic modification.
Làng Tráng ursid teeth. (A) Lower right canine showing root faceting. (B) Upper left canine showing root faceting. (C) Lower left canine with abundant Machichnus multilineatus on the roots. (D) Upper M3 showing pronounced tooth wedging. (E) Lower /m3 showing pronounced root wedging.
Although represented by 10 species, carnivores are not particularly common at Làng Tráng, comprising 3.79% of the teeth collected. Evidence of postmortem bone modification was not observed on small forms, such as mustelids (Lutrogale perspicillata, Arctonyx collaris), viverrids (Paradoxurus hermaphroditus), canids (Cuon alpinus), or small felids (Catopuma temminckii). In contrast, the teeth of larger felids (Panthera tigris and P. pardus) and ursids (Ailuropoda melanoleuca, Ursus thibetanus, and Helarctos malayanus) are all partially to fully gnawed (Figs. 9 and 10). This ranges from roots with minor faceting of the root tips (Fig. 9) to moderate reduction of root mass (Fig. 10A–C) and reduction of the root to a small, pyramidal wedge of bone (Fig. 10D and E).
Perissodactyls at Làng Tráng are represented by tapirs (Tapirus indicus) and rhinoceros (Dicerorhinus sumatrensis), both large mammals (Table 2). All T. indicus tooth roots are moderately to severely modified by gnawing (Fig. 11). Most teeth exhibit wedged roots (Fig. 11A); however, one tooth preserved partial roots that were characterized by wide, paired, U-shaped, striated grooves attributable to M. multilineatus (Fig. 11B). Upper teeth of Dicerorhinus collected in this study were strongly biased toward deciduous premolars. The roots are invariably reduced to a thin (2–3 mm) margin below the crown. Where not covered in matrix, they commonly exhibit evidence that the absence was due to rodent gnawing. Lower teeth include permanent molars. As with upper teeth, roots are typically reduced to a thin margin of bone. Approximately 35% of Dicerorhinus teeth exhibit definitive evidence of rodent gnawing in the form of striated longitudinal grooves consistent with Machichnus multilineatus in a thin rim of root adjacent. The remaining Dicerorhinus teeth preserve too little tooth root to definitively assess biogenic modification, but all are devoid of roots.
Làng Tráng tapir teeth. (A) Lower /m1 in anterior perspective (i), labial perspective (ii), and lingual perspective (iii) showing separate wedging of the anterior and posterior roots. (B) Lower M3 in lingual view (ii), labial view (ii), and oblique-basal-anterior view (iii) showing faceted roots and pronounced, striated U-shaped grooves consistent with Machichnus multilineatus.
Artiodactyls, represented by five species, are the most common forms preserved at Làng Tráng, comprising approximately 60.8% of the vertebrate. Sus scrofa, a medium-sized artiodactyl, is the most abundant taxon in the study area. Roots were preserved on a few of the smaller premolars but are absent in all larger premolars, molars, and canines. Molars are the most common tooth collected and invariably exhibit abundant, overprinted M. multilineatus, with the roots reduced to low-relief, multifaceted, pyramidal wedges beneath the enamel (Fig. 12). The teeth of Muntiacus muntjak, a small artiodactyl, exhibit a similar pattern to those of S. scrofa, wherein larger teeth are devoid of roots and commonly exhibited significant evidence of biogenic modification and smaller teeth retained partial to complete roots. Roots were preserved on none of the small artiodactyl Capricornis sumatraensis or the medium-sized Rusa unicolor teeth analyzed, and most exhibited evidence of biogenic modification, including wedging (Fig. 12). SEM imaging of wedged tooth roots reveals closely spaced parallel striae with older striae overprinted by younger striae (Fig. 13). The teeth of Bubalus bubalus, a large mammal and the sole bovid that occurs at Làng Tráng, were invariably limited to crowns that were completely devoid of any root material. Evidence for biogenic bone modification was classified as uncertain (Table 2), as the lack of bony material associated with the crowns precluded preservation of traces.
Artiodactyl teeth from Làng Tráng. (A) Sus scrofa /m3 in labial perspective (i), lingual perspective (ii), and basal perspective (iii) showing pronounced tooth wedging. (B) Sus scrofa /M2 in basal perspective (i), anterior perspective (ii), labial perspective (iii), and lingual perspective showing wedged roots. (C) Sus scrofa M1/ in labial perspective (i) and lingual perspective (ii) showing root faceting. Image iii illustrates the left root from ii in a slightly different orientation showing the planar surface. (D) Rusa unicolor /m1 in posterior perspective (i), labial perspective (ii), and lingual perspective (iii) exhibiting severely faceted roots. (E) Rusa unicolor M2/ in lingual perspective (i), labial perspective (ii), and basal perspective (iii) exhibiting roots gnawed down until they comprise a low wedge beneath the enamel.
Scanning electron microscope (SEM) scans of Rusa equina tooth roots from Làng Tráng. Arrows indicate areas were striae sets meet and where one set terminates and is overprinted by another set. (A) Wavy striae on the roots of an R. equina molar. (B) Linear striae on the roots of an R. equina molar. (C) Two sets of linear striae with an oblique interface on the roots of an R. equina molar.
Primates comprise approximately 18.5% of the Làng Tráng vertebrate fauna. Four genera (excluding Homo), all of which fit in the small size category, occur. The least common are Hylobates, known from only 5 teeth, and Presbytis, known from 40 teeth. Many of these teeth retain complete to partial roots. Those with partial to absent roots appear rounded, with no evidence of biogenic modification. Macaca are known from several hundred teeth. Roots are present on many teeth and are broken or worn on many others. Only a few of the larger teeth (molars) exhibited clear evidence of biogenic modification. The largest primate in the Làng Tráng collection is Pongo pygmaeus (Tables 1 and 2). Nearly 200 teeth were collected, representing most of the dentition of P. pygmaeus. Approximately 80% of P. pygmaeus teeth exhibited significant biogenic modification, which consisted of partial to complete root removal (Fig. 14). Root wedging is common, occurring both on individual roots (Fig. 14A) as well as on the whole root (Fig. 14B, C, E, G, and I). One partial mandible was collected (Fig. 8). The bone on one side was gnawed down to the roots (Fig. 8A), whereas the other side was largely unaltered (Fig. 8B).
Rodent-gnawed Pongo pygmaeus teeth from Làng Tráng. (A) Lower molar (/m2) with roots gnawed down to sharp triangular wedges. (B) Lower molar (/m1) with roots completely gnawed away. (C) Upper molar (M1/) with planar facets on greatly reduced tooth roots. (D) Lower molar (/m1 or /m2) with wedged tooth roots. (E) Lower molar (/m3) with wedged tooth roots. (F) Lower molar (/m1) with tooth roots completely removed by gnawing. (G) Lower molar (/m1) with tooth roots reduced to a rounded wedge. (H) Upper incisor with reduced root. (I) Lower premolar with wedged roots. (J) Lower incisor with partial root. (K) Lower molar (/m1 or /m2) with reduced roots and grooves assigned to Machichnus multilineatus. (L) Lower molar (/m1) with wedged roots. (M) Upper premolar (P4/) with wedged roots.
Proboscidean taxa at Làng Tráng are represented by 10 molar fragments attributed to cf. Palaeoloxodon aff. P. namadicus and 11 fragments, a deciduous molar, and a lower molar of Stegodon orientalis. None of this material has roots; however, clear evidence of biogenic modification was not observed.
Six rodent species have been identified at Làng Tráng (Tables 1 and 2). Four of these, (Rhizomys sp., Rattus sabanus, Leopoldamys neilli, and L. sabanus) are very small species and exhibit no evidence of biogenic modification (Table 2). Two species of porcupine were collected. Atherurus macrourus (the Asiatic brush-tailed porcupine) is a very small species (size class 1b) represented by eight teeth. Fossils of this taxon exhibit no evidence of biogenic modification. The largest porcupine in the Làng Tráng collection is Hystrix cf. H. kiangsenensis (size class 2a), which is represented by 367 teeth as well as several mandibular (Fig. 15) and maxillary fragments. Large porcupine fossils from Làng Tráng were initially assigned to Hystrix brachyura (Long et al., Reference Long, de Vos and Ciochon1996; Pei Reference Pei1965). Van Weers (Reference Van Weers2005) reassessed the material and assigned it to H. indica, based in part on the large size of the Làng Tráng fossils. Van Weers (Reference Van Weers2005) noted a similarity in size to H. kiangsenensis but, in the absence of preserved skulls at Làng Tráng, chose a conservative approach and placed the Làng Tráng fossils in H. indica. Subsequently, Lopatin (Reference Lopatin2020c) reassigned the Làng Tráng fossils to H. kiangsenensis based on the premolar to molar length ratio. We choose a conservative approach. It is clear that H. brachyura, H. indica, and H. kiangsenensis are closely allied species. Further work is needed to definitively differentiate these taxa, and thus we assign the Làng Tráng material to Hystrix cf. H. kiangsenensis pending completion of a full systematic review of the genus.
Hystrix cf. H. kiangsenensis mandible from Làng Tráng. (A) Labial surface of mandible. (B) Lingual surface of mandible. (C) Occlusal surface of mandible.
The premolars and molars of Hystrix cf. H. kiangsenensis at Làng Tráng are high crowned and devoid of roots (Fig. 15). Grooves in the remaining bone indicate that this is likely due to biogenic modification. A number of Hystrix cf. H. kiangsenensis incisors, which range in width from 5 to 7.5 mm were collected, including several in mandibular and maxillary fragments (Fig. 15).
Discussion
Composition of the Làng Tráng fauna
Preliminary analyses of the fauna represented by Làng Tráng fossils and their significance for palaeoecological reconstruction were discussed in several papers in which 34 species within 31 genera were identified (Ciochon and Olsen, Reference Ciochon and Olsen1991; Long et al., Reference Long, de Vos and Ciochon1996; Fig. 16). Recent work by a Russian team has resulted in additions to the Làng Tráng fauna, including a bat (Ia io), several artiodactyls (Hydropotes inermis, Axis porcinus, Tragulus kanchil, and Sus barbatus), primates (Trachypithecus sp.), rodents (Leopoldamys neilli and Rattus rattus), a soricid (Chodsigoa hoffmanni), and a carnivore (Neofelis nebulosa) (Lopatin, Reference Lopatin2020a, Reference Lopatin2020b, Reference Lopatin2022; Lopatin et al., Reference Lopatin, Maschenko, Vislobokova, Serdyuk and Dac2021). This material was not included in taphonomic analyses in the present paper; however, it does add to the diversity represented by the Làng Tráng fauna (Fig. 16).
Size distribution of Làng Tráng mammals. Dots show the median size of the mammal taxa, with the bars showing the size range exhibited by each taxon (scale on left). The bars at the base illustrate the absolute abundances of each taxon in the Làng Tráng fauna (scale on right). Size categories (shown at left) are based on Bunn (Reference Bunn1986), Bunn et al. (Reference Bunn, Bartram and Kroll1988), and Sahnouni et al. (Reference Sahnouni, Rosell, van der Made, Vergés, Ollé, Kandi, Harichane, Derradji and Medig2013).
In total, 37 species, within 32 genera have been identified from Làng Tráng (Ciochon and Olsen, Reference Ciochon and Olsen1991; Long et al., Reference Long, de Vos and Ciochon1996; Lopatin, Reference Lopatin2020a, Reference Lopatin2020b, Reference Lopatin2020c, Reference Lopatin2022; Lopatin et al., Reference Lopatin, Maschenko, Vislobokova, Serdyuk and Dac2021). These numbers exclude Homo sapiens, which is possibly associated with a much later (Hoabinhian) occupation of the cave, as well as a purported Gigantopithecus blacki that was recently reported from Làng Tráng (Lopatin et al., Reference Lopatin, Golovachev, Serdyuk, Maschenko, Vislobokova, Dac, Phuong, Parkhaev and Syromyatnikova2022a, Reference Lopatin, Maschenko and Dac2022b). We believe this report to be erroneous. The teeth illustrated in these papers are consistent in size and identical in morphology with Pongo pygmaeus teeth from Làng Tráng and elsewhere and do not likely represent a new occurrence of G. blacki.
Although the Làng Tráng cave assemblage is moderately diverse, it should not be considered representative of the complete local fauna. Modern Vietnam, which retains many of the species represented in the Làng Tráng system, has approximately 250 non-marine mammal species, with a high proportion of very small (<5 kg) mammals (Đặng et al., Reference Đặng, Endo, Son, Oshida, Canh, Phuong, Lund, Kawada, Hayashida and Sasaki2008; Abramov et al., Reference Abramov, Đặng, Bui and Nguyen2013). For instance, 53 species of rodent are known from Vietnam, many of which are common and widespread (Đặng et al., Reference Đặng, Endo, Son, Oshida, Canh, Phuong, Lund, Kawada, Hayashida and Sasaki2008). In contrast, the Làng Tráng Pleistocene assemblage is characterized by six rodent species, of which only Hystrix cf. H. kiangsenensis occurs in abundance. Similarly, only three insectivore specimens (representing two species within the genus Chodsigoa) occur within the Làng Tráng Pleistocene fauna (Lopatin, Reference Lopatin2022), whereas dozens of insectivore taxa occur in modern Vietnam (Đặng et al., Reference Đặng, Endo, Son, Oshida, Canh, Phuong, Lund, Kawada, Hayashida and Sasaki2008; Abramov et al., Reference Abramov, Đặng, Bui and Nguyen2013; Balakirev et al., Reference Balakirev, Bui and Rozhnov2025). Very small mammals are clearly severely underrepresented in the Làng Tráng fauna.
It is also worth noting that of the 42 taxa identified to date, 6 of these (Macaca sp., Hystrix kiangsenensis, Pongo pygmaeus, Muntiacus muntjak, Sus scrofa, and Rusa unicolor) represent approximately 87.5% of the fossils collected. Sus scrofa alone comprises 24% of the fossils recovered. The Làng Tráng Pleistocene cave accumulation illustrates a strong bias toward small- and medium-sized mammals in size classes 2a, 2b, and 3 and significant deficiencies in other size classes, particularly very small mammals, but also in medium- and large-sized mammals.
Weathering and transport of the Làng Tráng assemblage
The fossil bone and tooth roots at Làng Tráng exhibit minimal evidence of weathering. All bones and teeth rank very low (∼1) on the Behrensmeyer (Reference Behrensmeyer1978) scale. Although we acknowledge that Làng Tráng occurred in a very different setting than the Amboseli Basin, where Behrensmeyer conducted her analyses, restriction of bone weathering to light surface cracking suggests that, while the bone material at Làng Tráng may have lain exposed on the forest floor for a short period of time, it was transported into the Làng Tráng cave complex before long-term exposure could have occurred. Behrensmeyer (Reference Behrensmeyer1975) also used the tooth to vertebra ratios to assess postmortem transport of vertebrate remains, with ratios ranging from 0.89 in swamps to 1.44 in lake beds (Behrensmeyer and Dechant Boaz, Reference Behrensmeyer, Deschant Boaz, Behrensmeyer and Hill1980). With thousands of teeth and only a few vertebral centra in the Làng Tráng material, the Làng Tráng greatly exceeds 500. This clearly indicates that an unusual fossil concentration mode was in play at Làng Tráng that was not accounted for in Behrensmeyer’s work.
During the Pleistocene, the Sông Mã had not likely eroded down to its present level, and the caves may have experienced fluvial input during seasonal floods (Ciochon et al., Reference Ciochon, Olsen and James1991). Sedimentological evidence of this is sparse, but the occurrence of isolated crocodilian teeth and gastropods similar to gastropods that occupy the modern Sông Mã may support this hypothesis (Ciochon et al., Reference Ciochon, Olsen and James1991). However, the bulk of the fossils occur in random orientations in the cave breccia, which exhibits no physical sedimentary structures consistent with fluvial transport. The fossils occur in random orientations with no evidence of concentration layers/lags. Voorhies Group and similar analyses are commonly used to assess fluvial transport of vertebrate material (e.g., Voorhies, Reference Voorhies1969; Boaz and Behrensmeyer, Reference Boaz and Behrensmeyer1976; Lyman, Reference Lyman1994; Aslan and Behrensmeyer, Reference Aslan and Behrensmeyer1996; Carpenter, Reference Carpenter2020). Although cave accumulations could be affected by fluvial processes, the sparsity of postcrania, mandibles, and skulls and the overwhelming dominance of isolated teeth in the Làng Tráng material render their application problematic. It is clear that cave-hosted faunas are subjected to different taphonomic influences than savannah, forest, or alluvial plain accumulations and, consequently, should not be assessed by the same taphonomic criteria.
Porcupines as originators of the Làng Tráng assemblage
As discussed earlier, the Làng Tráng fauna is dominated by isolated teeth but also includes a few partial mandibles, maxillae, skulls, and sparsely distributed postcrania, which commonly exhibit evidence of biogenic modification. Rare specimens (<0.1%) preserve narrow (0.2–0.5 mm) U-shaped grooves, most typically on the shafts of long bones (Fig. 6B and C), but were also observed on a fragment of a Rusa unicolor mandible (Fig. 7B). These traces, included within the ichnotaxon Machichnus bohemicus, were most likely emplaced by sharp, narrow teeth. Of the carnivores preserved in the Làng Tráng fauna, leopards (Panthera pardus) are the only carnivore known to bring prey into caves for storage (Brain, Reference Brain and Brain1993; de Ruiter and Berger, Reference de Ruiter and Berger2000; Sauqué et al., Reference Sauqué, Rabal-Garcés, Sola-Almagro and Cuenca-Bescós2014). Panthera pardus are certainly large enough to have carried large artiodactyl bones and have teeth appropriate in size and width to have made the Machichnus bohemicus observed on the Rusa unicolor femur.
The most common trace observed (>99.9%) comprises elongate, approximately linear, broad, gently U-shaped grooves with internal striae (Figs. 6B, D, and E and 7B and C). These traces, included within the ichnotaxon Machichnus multilineatus, are indicative of an animal with closely spaced, paired, approximately flat-tipped teeth. The broad (2–5 mm), gently curved, flat-bottomed grooves could only have been made by rodent incisors. Of the animals identified at Làng Tráng, the wider, more abundant traces are consistent in morphology and size only with Hystrix cf. H. kiangsenensis incisors, whereas the smaller, less common traces are likely attributable to Atherurus cf. A. macrourus.
South African porcupines have been observed to gnaw by holding the bone in position with the upper incisors and scraping with the lower incisors (Goodwin, Reference Goodwin1956). This is consistent with the patterns observed on several partially gnawed mandibles in the study material, wherein one side of the mandible is strongly gnawed and the other side is largely unmodified (Figs. 7 and 8). Porcupine incisors have been observed to emplace similar broad, shallow scrape marks in bone (Maguire et al., Reference Maguire, Pemberton and Collet1980; Kibii, Reference Kibii2009; Kaur et al., Reference Kaur, Patnaik, Singh and Krishan2019).
Like all rodent groups, porcupines are obligate gnawers. Rodent incisors are open-rooted and ever-growing, requiring constant attrition to maintain an appropriate length (Klippel and Synstelien, Reference Klippel and Synstelien2007; Pokines, Reference Pokines2015; Pokines et al., Reference Pokines, Santana, Hellar, Bian, Downs, Wells and Price2016, Reference Pokines, Sussman, Gough, Ralston, McLeod, Brun, Kearns and Moore2017). Limestone and dolomite, the most common rock types in the study area, have hardnesses of 3 and 4 on the Mohs scale, which is one to two orders of magnitude softer than tooth enamel, rendering them ineffective gnawing media. Other lithology types, such as granite and syenite, are quartz-rich (Mohs 6–7), rendering them much harder than tooth enamel and would risk too-rapid attrition and possible breakage. In contrast tooth enamel ranks 5 on the Mohs scale, similar to (albeit slightly harder than) bone, making bone an ideal medium on which rodents can hone their incisors.
Both Atherurus and Hystrix are well-known bone gnawers (e.g., Brain, Reference Brain, Behrensmeyer and Hill1980, Reference Brain1981; Jori et al., Reference Jori, Lopez-Bejar and Houben1998; Bacon et al., Reference Bacon, Demeter, Duringer, Helm, Bano, Long and Thuy2008, Reference Bacon, Westaway, Antoine, Duringer, Blin, Demeter and Ponche2015; Coppola et al., Reference Coppola, Guerrieri, Simoncini, Varuzza, Vecchio and Felicioli2020). Hystrix, in particular, has long been recognized as a primary agent responsible for many African bone accumulations (Shortridge, Reference Shortridge1934; Hughes, Reference Hughes1954; Dart, Reference Dart1957; Brain, Reference Brain, Behrensmeyer and Hill1980, Reference Brain1981). Hystrix has also been a major taphonomic agent in southern China and peninsular Southeast Asia since shortly after its appearance in the Miocene (Pei, Reference Pei1965; Wang and Qiu, Reference Wang and Qiu2002; Wang and Qi, Reference Wang and Qi2005; Takai et al., Reference Takai, Zhang, Kono and Jin2014; Avalos, Reference Avalos2017).
In addition to simple attrition, Hystrix may also obtain some nutrients from bone (Laudet and Fosse, Reference Laudet and Fosse2001; Bacon et al., Reference Bacon, Westaway, Antoine, Duringer, Blin, Demeter and Ponche2015). It has been postulated that Hystrix gnaws bone in order to obtain nutrients essential in quill production (Singer, Reference Singer1959; Roze et al., Reference Roze, Locke and Vatakis1990). Porcupine quills consist of keratin, with only trace amounts of phosphorous and calcium (Roze et al., Reference Roze, Locke and Vatakis1990; Inayah et al., Reference Inayah, Farida and Purwaningsih2020). Minor amounts of collagen that remain in desiccated bone is broken down into amino acids during digestion and may aid in the production of quills. A multiyear study of Hystrix cristata in central Italy illustrated that gnawed bones are more common near porcupine setts during birthing intervals suggesting that bone may be preferentially gathered and gnawed by lactating females (Mori et al., Reference Mori, Lovari and Mazza2018). In addition, some hystricids, such as Hystrix africaeaustralis and Atherurus macrourus, will actually eat carrion flesh (Roth, Reference Roth, Mohr and Rohrs1964; Kingdon, Reference Kingdon and Kingdon1974; Jori et al., Reference Jori, Lopez-Bejar and Houben1998; Coppola et al., Reference Coppola, Guerrieri, Simoncini, Varuzza, Vecchio and Felicioli2020). It appears, however, that one of the main impetuses driving porcupines to continuously gather and gnaw bone is a behavioral adaptation to hone and trim their ever-growing incisors (Kibii, Reference Kibii2009).
Brain (Reference Brain, Behrensmeyer and Hill1980, Reference Brain1981) observed that these Hystrix may favor bones with extensive compact tissue surfaces, such as the diaphyses of long bones, over articular surfaces. In contrast, Maguire et al. (Reference Maguire, Pemberton and Collet1980) noted that Hystrix may also target epiphyses, destroying them completely, reducing skeletal elements to tubular diaphyses. They postulated that this type of bone modification could be misinterpreted as the hollowing of diaphyses caused by carnivores, which gnaw on epiphyses to access bone marrow. These hypotheses can be neither verified nor discounted in the Làng Tráng material, as postcranial material is quite rare. This is interpreted to be a function of extreme porcupine-induced bone modification.
Brain (Reference Brain, Behrensmeyer and Hill1980) noted that the taxonomic abundance of bones in recent porcupine lairs was a reasonable proxy for the taxonomic abundance of mammalian taxa in the Kalahari region (based on the Nossob porcupine lair). We do not assume the same is true in the Làng Tráng mammal fauna. The Nossob assemblage was characterized by abundant postcrania, including abundant transported, but non-gnawed bone (Brain, Reference Brain, Behrensmeyer and Hill1980; O’Regan et al., Reference O’Regan, Kuman and Clarke2011). This contrasts significantly with the Làng Tráng fauna, which is overwhelmingly dominated by isolated tooth crowns. Arguably, the extreme level to which the Làng Tráng fossil assemblage was reduced to elements too hard to reduce further may indicate a resource stress in the Làng Tráng forest system in which insufficient skeletal material of an appropriate size was available for gnawing by the resident porcupine population.
It is worth noting that this resource stress was not a short-term situation. Although differing in age, all four of the Làng Tráng caves exhibit similar taphonomic attributes. As discussed earlier, three of the caves were dated using ESR, which indicated that the cave deposits formed from ∼480,000 until approximately 145,000 BP (Ciochon and Olsen, Reference Ciochon and Olsen1991; Fig. 5). Identical taphonomic attributes between all caves and all material collected, regardless of level, supports the hypothesis that porcupines were the primary bone accumulator through this entire interval and that whatever stressor resulted in extreme bone modification persisted throughout the entire interval of bone accumulation.
Contrasting porcupine-generated bone accumulations with carnivore accumulations
Faunal accumulations assembled by porcupines differ significantly from those accumulated by carnivores such as lions (e.g., Arriaza et al., Reference Arriaza, Domínguez-Rodrigo, Yravedra and Baquendano2016), leopards (e.g., Watson, Reference Watson1993; de Ruiter and Berger, Reference de Ruiter and Berger2000; Sauqué et al., Reference Sauqué, Rabal-Garcés, Sola-Almagro and Cuenca-Bescós2014), and hyenas (Cruz-Uribe, Reference Cruz-Uribe1991; Boaz et al., Reference Boaz, Ciochon, Xu and Liu2000; Domínguez-Rodrigo and Pickering, Reference Domínguez-Rodrigo and Pickering2010; Kuhn et al., Reference Kuhn, Berger and Skinner2010). Hyena-derived bone accumulations commonly consist of abundant long bones, long bone ends, and bone fragments, with scattered mandibles and skulls (e.g., Skinner et al., Reference Skinner, Davis and Ilani1980; Cruz-Uribe, Reference Cruz-Uribe1991; Boaz et al., Reference Boaz, Ciochon, Xu and Liu2000; Domínquez-Rodrigo et al., Reference Domínguez-Rodrigo, Gidna, Yravedra and Musiba2012). Of these, partial limbs and skulls are most abundant (Lansing et al., Reference Lansing, Cooper, Boydston and Holekamp2009). Many bone shafts lack epiphyseal ends (Cruz-Uribe, Reference Cruz-Uribe1991). Most bones belong to mid-sized to large ungulates (size classes 3 and 4), with few small or very small (size classes 1 and 2) animals represented, although carnivore bones may also be abundant (Skinner et al., Reference Skinner, Davis and Ilani1980; Lacruz and Maude, Reference Lacruz and Maude2005; Kuhn et al., Reference Kuhn, Berger and Skinner2010). Hyenas rarely carry whole carcasses back into their dens, typically dismembering prey or scavenged carcasses near the kill site and transporting body parts instead (Skinner et al., Reference Skinner, Davis and Ilani1980). Biogenic modification features, in the form of pits, puncture marks, and straight to gently curved, U-shaped grooves, are common on many bones (Cruz-Uribe, Reference Cruz-Uribe1991; Lansing et al., Reference Lansing, Cooper, Boydston and Holekamp2009).
Although less commonly than hyenas, leopards have also been shown to bring bones into caves (de Ruiter and Berger, Reference de Ruiter and Berger2000; Pickering et al., Reference Pickering, Domínguez-Rodrigo, Egeland and Brain2004; Sauqué et al., Reference Sauqué, Rabal-Garcés, Sola-Almagro and Cuenca-Bescós2014). In contrast to hyenas, leopards commonly carry entire prey back to their lairs. Similar to hyenas, biogenic modification features, particularly puncture marks and U-shaped grooves, are common on many bones (de Ruiter and Berger, Reference de Ruiter and Berger2000; Pickering et al., Reference Pickering, Domínguez-Rodrigo, Egeland and Brain2004; Sauqué et al., Reference Sauqué, Rabal-Garcés, Sola-Almagro and Cuenca-Bescós2014). As discussed earlier, evidence for biogenic modification by leopards was observed on rare bones in the Làng Tráng assemblage.
Porcupine-generated accumulations have some characters in common with carnivore-generated accumulations but have significant differences as well. Porcupines gather disarticulated, dry bone and thus primarily transport individual elements (Egeland et al., Reference Egeland, Egeland and Bunn2008). They gnaw many/most of the bones they drag (Egeland et al., Reference Egeland, Egeland and Bunn2008), resulting in broad, flat-bottomed gnaw marks consistent with the ichnotaxon Machichnus multilineatus. These traces contrast sharply with the grooves with U-shaped to arcuate profiles assigned to Machichnus regularis and M. bohemicus that are produced by carnivore gnawing. An overwhelming proportion of bone and tooth roots in the Làng Tráng assemblage exhibit the ichnotaxon M. multilineatus, commonly in dense, monotypic, overprinted associations.
Root-faceting and root-wedging—signs of severe rodent modification
Isolated teeth with either faceted root tips or wedged roots dominate the Làng Tráng assemblage. As mentioned earlier, where sufficient bone remains, distinct Machichnus multilineatus are preserved (Figs. 8 and 9C). Tooth roots with minor to moderate levels of Hystrix-produced modification exhibit multiple, obliquely oriented planar facets (Figs. 8, 9A and B, and 10B). These facets are typically covered in parallel to subparallel, linear to gently curved microscopic striae.
Tooth roots with major levels of Hystrix-produced modification have either a pyramidal or a wedge-shaped profile (Figs. 8–12 and 17). Similar to faceted roots, wedged roots are often covered in parallel to subparallel, linear to gently curved microscopic striae (Fig. 13). In many cases, these occur within distinct Machichnus multilineatus traces emanating outward from near the edge of the enamel to the tip of the wedge (Figs. 8, 9, 10, 11, 12 and 17). Tooth roots with extreme biogenic modification commonly retain too little bone to preserve ichnotaxonomically identifiable traces.
Sketches of Làng Tráng teeth showing root-faceting and root-wedging. Dashed lines show the approximate outline of the original roots. (A) Two Tapirus indicus molars, with the one on left illustrating faceting and partial wedging and the one on right showing full wedging. (B) Ursus thibetanus canines showing root-faceting. (C) Ursus thibetanus M3/ illustrating root-wedging. (D) Ursus thibetanus /m2 illustrating root-wedging. (E) Sus scrofa /m3 illustrating root wedging.
The roots of the teeth of most small, medium, and large mammals at Làng Tráng (size classes 2–4) are characterized by planar facets, wedged roots, or complete root removal (Figs. 7–12 and 17). Root-faceting and root-wedging are herein considered to be diagnostic attributes of rodent-generated and severely rodent-modified cave assemblages. The occurrence of tooth-dominated fossil assemblages with extensive root-faceting and root-wedging may indicate a lack of balance between resident porcupine populations and sufficient material of the right hardness for gnawing.
Size distribution of the Làng Tráng fauna
The Làng Tráng Pleistocene vertebrate fauna is represented by 37 mammalian species, within 32 genera and represents a wide range of body sizes. It likely records accumulation in a mixed forest with heterogeneous tree cover (Louys and Meijaard, Reference Louys and Meijaard2010).
Despite the overall diversity, most taxa are known only from a few specimens. The fauna comprises a taphocoenosis dominated by medium and moderately large-bodied species (Fig. 16). As discussed earlier, the Làng Tráng fauna includes 11 species represented by 10 or fewer specimens. With few exceptions (six species represented by only a few specimens each), the fauna is devoid of very small (<1 kg) species. Similarly, large mammals (average body mass >250 kg) are represented by five taxa, comprising <5% of the fossils collected. Most teeth collected from Làng Tráng represent small-, medium-, and large-sized mammals (size classes 2–4, between 7 and 250 kg average body mass). Of these, six taxa (Sus scrofa, Muntiacus muntjak, Rusa equina, Hystrix kiangsenensis, Macaca sp., and Pongo pygmaeus) comprise more than 87% of the teeth collected.
The Làng Tráng mammalian fauna is dominated by the teeth of animals with skulls and postcrania that are within the transport capabilities of the porcupines represented in the Làng Tráng collection. Although, undoubtedly, collector bias likely contributed to the disproportionately diminutive collection of very small taxa (e.g., collectors are more likely to overlook smaller fossils), we believe that this paucity was primarily caused by porcupine prejudice. Very small animals fell below the size range of material sought by porcupines. Consequently, these taxa are disproportionately rare in the Làng Tráng fauna and their teeth are rarely biogenically altered. The scarcity of teeth from large and very large mammals (>250 kg average body mass) in the fossil material in the Làng Tráng collection likely reflects the inability of porcupines to collect this material. The skulls and mandibles of large and very large mammals are typically common at primary predator kill sites but are uncommon in porcupine lairs, as large skulls are beyond the transportive ability of most porcupines (Peterhans and Singer, Reference Peterhans and Singer2006).
The importance of porcupines in Pleistocene bone accumulations in southern China and peninsular Southeast Asia
The soils on tropical forests floors in southeastern Asia are invariably acidic, with pH values as low as 3.9 (van Schaik and Mirmanto, Reference van Schaik and Mirmanto1985; Motavalli et al., Reference Motavalli, Palm, Parton, Elliott and Frey1995; Fujii et al., Reference Fujii, Hartono, Funakawa, Uemura and Kosaki2010; Fujii, Reference Fujii2014), and pose particular challenges to bone preservation. Bones on tropical forest floors are exposed to myriad scavenging insects and vertebrates, as well as microbial degradation and destruction by plant roots (Tappen, Reference Tappen1994). Although burial can protect bone from animal and insect activity, bone buried in tropical soil can fully degrade in as little as 6 years due to soil acidity (Ross and Cunningham, Reference Ross and Cunningham2011). Karst caves and fissures, formed through the dissolution of limestone, provide a pH-neutral environment wherein bones and teeth brought into this environment are protected by the buffering effect of abundant calcium carbonate (Avalos, Reference Avalos2017). In these environments, porcupines, particularly Hystrix spp., are the dominant taphonomic agent.
Almost all Pleistocene fossil vertebrate localities in southern China and peninsular Southeast Asia occur in caves (e.g., Pei, Reference Pei1935, Reference Pei1965; Ciochon and Olsen, Reference Ciochon and Olsen1991; Bacon et al., Reference Bacon, Demeter, Schuster, Long, Thuy, Antoine, Sen, Nga and Huong2004, Reference Bacon, Demeter, Rousse, Long, Duringer, Antoine and Thuy2006, Reference Bacon, Demeter, Duringer, Helm, Bano, Long and Thuy2008, Reference Bacon, Duringer, Antoine, Demeter, Shackelford, Sayavongkhamdy and Sichanthongtip2011, Reference Bacon, Westaway, Antoine, Duringer, Blin, Demeter and Ponche2015; Lenoble et al., Reference Lenoble, Zeitoun, Laudet, Seveau, Doy Asa, Pautreau, Coupey, Zeitoun and Rambault2006; Wang et al., Reference Wang, Liao, Li and Tian2014; Fan et al., Reference Fan, Shao, Bacon, Liao and Wang2022). Numerous studies have highlighted the importance of porcupines in contributing to cave-hosted vertebrate fossil assemblages both in Southeast Asia and elsewhere (e.g., Brain, Reference Brain, Behrensmeyer and Hill1980, Reference Brain1981; Bacon et al., Reference Bacon, Demeter, Duringer, Helm, Bano, Long and Thuy2008, Reference Bacon, Westaway, Antoine, Duringer, Blin, Demeter and Ponche2015; Fan et al., Reference Fan, Shao, Bacon, Liao and Wang2022).
In most cases, taphonomic influences contributing to bone accumulation and faunal modification include fluvial transport, cave falls, predator caching, and the action of scavengers (Bacon et al., Reference Bacon, Demeter, Schuster, Long, Thuy, Antoine, Sen, Nga and Huong2004, Reference Bacon, Demeter, Rousse, Long, Duringer, Antoine and Thuy2006, Reference Bacon, Demeter, Duringer, Helm, Bano, Long and Thuy2008, Reference Bacon, Duringer, Antoine, Demeter, Shackelford, Sayavongkhamdy and Sichanthongtip2011, Reference Bacon, Westaway, Antoine, Duringer, Blin, Demeter and Ponche2015; Lenoble et al., Reference Lenoble, Zeitoun, Laudet, Seveau, Doy Asa, Pautreau, Coupey, Zeitoun and Rambault2006; Wang et al., Reference Wang, Liao, Li and Tian2014; Fan et al., Reference Fan, Shao, Bacon, Liao and Wang2022). Some cases, such as at Tam Hang, Nam Lot, Punung, and Làng Tráng, comprise the end-member scenario, in which porcupines were not only the dominant agent that contributed to the accumulation of the cave-hosted fauna, but they were likely the main taphonomic factor contributing to the destruction of the bone as well.
Conclusions
The Làng Tráng cave complex in western Vietnam, near the Laos–Vietnam border, preserves a diverse and abundant accumulation of middle Pleistocene mammals. The fauna is devoid of articulated material and is dominated by isolated teeth, with only a few long bones, mandibles, and skulls. Although moderately diverse, the Làng Tráng fauna preserves relatively few fossils of exceptionally large (>200 kg) or small (<7 kg) animals. Six taxa (Macaca sp., Hystrix cf. H. kiangsensis, Muntiacus muntjak, Pongo pygmaeus, Sus scrofa, and Rusa unicolor) dominate the fauna.
Most fossils exhibit abundant evidence of postmortem biogenic alteration. Rare long bones and mandibles preserve clear examples of trace fossils referable to Machichnus bohemicus and to M. multilineatus. The narrow, U-shaped cross-sectional profile of M. bohemicus is consistent with gouges produced by carnivore teeth, and the large feline Panthera pardus is implicated in their emplacement. They are rare (only identified on a single long bone and a single dentary) but provide evidence that cave-dwelling carnivores may have been contributed to the genesis of the Làng Tráng cave assemblages. The broad, shallow grooves attributed to M. multilineatus are pervasive in the Làng Tráng material (e.g., they comprise >99.9% of identifiable traces observed and occur on all bones and many of the teeth observed). The only organisms preserved in the Làng Tráng fauna with teeth of the appropriate size and shape to have emplaced most of these traces is the Malayan porcupine Hystrix kiangsenensis. Some of the smaller examples may have been emplaced by the Asiatic brushy-tailed porcupine Atherurus macrourus.
Teeth with fully preserved roots are rare and are limited primarily to tiny teeth (molars from very small mammals and premolars or incisors from small- to medium-sized mammals). A large proportion of the teeth of small-, medium-, and large-sized mammals exhibit faceted and wedged roots. Faceted roots preserve moderate biogenic modification in the form of multiple, approximately planar surfaces characterized by fine parallel to subparallel striae. Wedged roots are reduced to pyramidal wedges below the crown and again exhibit parallel to subparallel striae, most commonly oriented away from the crown (parallel to the original axis of the root). It is proposed herein that root faceting and root wedging are attributes that can only be emplaced through rodent-generated biogenic modification.
Although a variety of carnivores are known to transport predated and scavenged prey back to cave dens, only a few bones exhibit tooth marks consistent with carnivores. In contrast, an overwhelming proportion of the vertebrate fossils at Làng Tráng provide definitive evidence of the action of rodents, most likely porcupines. Although porcupines are common contributors to bone accumulation in many Pleistocene caves, at Làng Tráng they are interpreted to have been the main taphonomic agent involved in bone transport and modification for several hundred thousand years. It is ironic that the same taphonomic agent responsible for the destruction of skeletal and cranial bone is also responsible for bringing the fossils into a pH-neutral cave environment that protected the residual teeth and bone from destructive plant acids and degradation on the tropical forest floor.
Acknowledgments
We are grateful to Vu The Long and the Vietnam Institute of Archeology for facilitating our work in Hanoi and at Làng Tráng. We are grateful to the local villagers in the vicinity of the Làng Tráng cave complex for welcoming us to their community and providing refreshment. We thank Luis Gonzalez, Charles Young, and Roy Larick for their work on the geology of the Làng Tráng cave complex and John DeVos for his work on the taxonomy of the Làng Tráng mammalian fauna. Figure 3 is based on mapping by Katerina Semendeferi.
Open access
