Using machine learning on new feature sets extracted from three-dimensional models of broken animal bones to classify fragments according to break agent

Distinguishing agents of bone modification at paleoanthropological sites is an important means of understanding early hominin evolution. Fracture pattern analysis is used to help determine site formation processes, including whether hominins were hunting or scavenging for animal food resources. Determination of how these behaviors manifested in ancient human sites has major implications for our biological and behavioral evolution, including social and cognitive abilities, dietary impacts of having access to in-bone nutrients like marrow, and cultural variation in butchering and food processing practices. Nevertheless, previous analyses remain inconclusive, often suffering from lack of replicability, misuse of mathematical methods, and/or failure to overcome equifinality. In this paper, we present a new approach aimed at distinguishing bone fragments resulting from hominin and carnivore breakage. Our analysis is founded on a large collection of scanned three-dimensional models of fragmentary bone broken by known agents, to which we apply state of the art machine learning algorithms. Our classification of fragments achieves an average mean accuracy of 77% across tests, thus demonstrating notable, but not overwhelming, success for distinguishing the agent of breakage. We note that, while previous research applying such algorithms has claimed higher success rates, fundamental errors in the application of machine learning protocols suggest that the reported accuracies are unjustified and unreliable. The systematic, fully documented, and proper application of machine learning algorithms leads to an inherent reproducibility of our study, and therefore our methods hold great potential for deciphering when and where hominins first began exploiting marrow and meat, and clarifying their importance and influence on human evolution.

Life and death at Dmanisi, Georgia: Taphonomic signals from the fossil mammals

There are many hypotheses regarding influences on the early hominin biogeographic spread into Eurasia; among them is increased meat-eating. Dmanisi in Georgia is one of the rare Early Pleistocene sites in Eurasia, and here we present primary information and analysis of the medium and large mammal taphonomy, contributing information about site formation and the hominins' interaction with the fauna. Nearly 85% of the specimens come from the B1 stratum. Relative abundances of mammal families demonstrate some bias toward carnivores, especially Canis borjgali, and diverse Felidae species. Bones display little weathering. Post-depositional surface modifications and matrix obscure many bone surfaces, but carnivore tooth marking is the most common bone surface modification from the nutritive taphonomic phase. Tooth pits are large, in the size range of those made by modern Crocuta crocuta and Panthera leo. Breakage variables indicate most breaks occurred while the bones were still fresh, many by carnivore consumption. Fairly even limb bone representation of herbivores suggests carcasses were introduced to the site nearly whole. Hominin tool marks are present in low frequencies, but they suggest a variety of behaviors. These marks are found on Equus, Palaeotragus, Bison, large cervids, Pseudodama, Canis, and Mammuthus. Some were made by filleting proximal limb segments, and so are likely indicative of early access to carcasses, while other marks suggest scavenging. The Homo taphonomic variables resemble the rest of the taphonomic signatures from the site with little weathering, a slightly higher percentage of their bones are whole, but only a few have probable carnivore damage. The assemblage characteristics are compared to modern actualistic and experimental assemblages, and it is concluded that Dmanisi presents a palimpsest of hyena denning, felid activity, hominin meat-eating and likely natural deaths.

Early Pleistocene stratigraphy, sedimentary environments, and formation contexts at Dmanisi in the Georgian Caucasus

The Early Pleistocene site of Dmanisi is now well known for its large number of fossils of early Homo erectus as well as associated artifacts and faunal remains, recovered mainly in pipe-related geologic features. Testing in the M5 unit 100 m to the west of the main excavations revealed a thick stratigraphy with no evidence of pipes or gullies, indicating that the geologic record at Dmanisi included spatially distinct sedimentary environments that needed further investigation. Here we report the results of a geoarchaeological program to collect data bearing on contexts and formation processes over a large area of the promontory. That work has defined over 40,000 m² of in situ deposits with artifacts and faunas. Stratum A ashes bury the uppermost Mashavera Basalt, which we have dated to 1.8 Ma in the M5 block. The Stratum A deposits contain stratified occupations that accumulated quickly and offer good potential for recovery of in situ materials. Stratum B1 deposits above the A/B unconformity include all of the pipe and gully facies at Dmanisi, reflecting a brief but very intense phase of geomorphic change. Those deposits contain the majority of faunas and all of the hominin fossils. B1 slope facies offer excellent formation contexts away from the piped area, and all B1 deposits are sealed by Stratum B2 over the whole promontory. Strata B2 to B5 register a return to slope facies, with no further evidence of pipes or gullies. Those deposits also present excellent contexts for recovery of in situ occupations. Overall, Dmanisi's geologic history preserves an exceptional record of the activities and environmental context of occupations during the first colonization of Eurasia.

Spatial patterning of the archaeological and paleontological assemblage at Dmanisi, Georgia: An analysis of site formation and carnivore-hominin interaction in Block 2

This study addresses the roles of biotic agents in site formation in the B1 strata of Block 2 at Dmanisi, Georgia, using theoretical and analogous frameworks for the interpretation of spatial behaviors of carnivores and hominins. For this study, stone material, faunal remains, and coprolites are analyzed to determine if any spatially distinct behaviors can be identified, located, and attributed to either hominins or carnivores. Faunal, stone, and coprolite assemblages are compared with each other, and lithic, taxonomic, and taphonomic subassemblages are compared with the overall distribution of their parent material. The spatial and taphonomic signatures suggest that hominin activity was only a small part of the contributing factors to site formation, whereas carnivores played a major role in the accumulation of bone.

Early Pleistocene enamel proteome from Dmanisi resolves Stephanorhinus phylogeny

Ancient DNA (aDNA) sequencing has enabled reconstruction of speciation, migration, and admixture events for extinct taxa¹. Outside the permafrost, however, irreversible aDNA post-mortem degradation² has so far limited aDNA recovery to the past ~0.5 million years (Ma)³. Contrarily, tandem mass spectrometry (MS) allowed sequencing ~1.5 million year (Ma) old collagen type I (COL1)⁴ and suggested the presence of protein residues in Cretaceous fossil remains⁵, although with limited phylogenetic use⁶. In the absence of molecular evidence, the speciation of several Early and Middle Pleistocene extinct species remain contentious. In this study, we address the phylogenetic relationships of the Eurasian Pleistocene Rhinocerotidae^7–9 using a ~1.77 Ma old dental enamel proteome of a Stephanorhinus specimen from the Dmanisi archaeological site in Georgia (South Caucasus)¹⁰. Molecular phylogenetic analyses place the Dmanisi Stephanorhinus as a sister group to the woolly (Coelodonta antiquitatis) and Merck’s rhinoceros (S. kirchbergensis) clade. We show that Coelodonta evolved from an early Stephanorhinus lineage and that the latter includes at least two distinct evolutionary lines. As such, the genus Stephanorhinus is currently paraphyletic and its systematic revision is therefore needed. We demonstrate that Early Pleistocene dental enamel proteome sequencing overcomes the limits of ancient collagen- and aDNA-based phylogenetic inference. It also provides additional information about the sex and taxonomic assignment of the specimens analysed. Dental enamel, the hardest tissue in vertebrates¹¹, is highly abundant in the fossil record. Our findings reveal that palaeoproteomic investigation of this material can push biomolecular investigation further back into the Early Pleistocene.

Postcranial evidence from early Homo from Dmanisi, Georgia

The Plio-Pleistocene site of Dmanisi, Georgia, has yielded a rich fossil and archaeological record documenting an early presence of the genus Homo outside Africa. Although the craniomandibular morphology of early Homo is well known as a result of finds from Dmanisi and African localities, data about its postcranial morphology are still relatively scarce. Here we describe newly excavated postcranial material from Dmanisi comprising a partial skeleton of an adolescent individual, associated with skull D2700/D2735, and the remains from three adult individuals. This material shows that the postcranial anatomy of the Dmanisi hominins has a surprising mosaic of primitive and derived features. The primitive features include a small body size, a low encephalization quotient and absence of humeral torsion; the derived features include modern-human-like body proportions and lower limb morphology indicative of the capability for long-distance travel. Thus, the earliest known hominins to have lived outside of Africa in the temperate zones of Eurasia did not yet display the full set of derived skeletal features.

A fourth hominin skull from Dmanisi, Georgia

Newly discovered Homo remains, stone artifacts, and animal fossils from Dmanisi, Republic of Georgia, provide a basis for better understanding patterns of hominin evolution and behavior in Eurasia ca. 1.77 million years ago. Here we describe a fourth skull that is nearly complete, lacking all but one of its teeth at the time of death. Both the maxillae and the mandible exhibit extensive bone loss due to resorption. This individual is similar to others from the site but supplies information about variation in brain size and craniofacial anatomy within the Dmanisi paleodeme. Although this assemblage presents numerous primitive characters, the Dmanisi skulls are best accommodated within the species H. erectus. On anatomical grounds, it is argued that the relatively small-brained and lightly built Dmanisi hominins may be ancestral to African and Far Eastern branches of H. erectus showing more derived morphology.

Anthropology: the earliest toothless hominin skull

The site of Dmanisi in the Eurasian republic of Georgia has yielded striking hominin, faunal and archaeological material as evidence for the presence of early Homo outside Africa 1.77 million years ago, documenting an important episode in human evolution. Here we describe a beautifully preserved skull and jawbone from a Dmanisi hominin of this period who had lost all but one tooth several years before death. This specimen not only represents the earliest case of severe masticatory impairment in the hominin fossil record to be discovered so far, but also raises questions about alternative subsistence strategies in early Homo.

A new skull of early Homo from Dmanisi, Georgia

Another hominid skull has been recovered at Dmanisi (Republic of Georgia) from the same strata in which hominid remains have been reported previously. The Dmanisi site dated to approximately 1.75 million years ago has now produced craniofacial portions of several hominid individuals, along with many well-preserved animal fossils and quantities of stone artifacts. Although there are certain anatomical differences among the Dmanisi specimens, the hominids do not clearly represent more than one taxon. We assign the new skull provisionally to Homo erectus (=ergaster). The Dmanisi specimens are the most primitive and small-brained fossils to be grouped with this species or any taxon linked unequivocally with genus Homo and also the ones most similar to the presumed habilis-like stem. We suggest that the ancestors of the Dmanisi population dispersed from Africa before the emergence of humans identified broadly with the H. erectus grade.

Recognizing hominoid-modified bones: the taphonomy of colobus bones partially digested by free-ranging chimpanzees in the Kibale Forest, Uganda

We present a taphonomic study of bones that have passed though the digestive tracts of free-ranging chimpanzees from the Kibale Forest in Uganda. The bone assemblage can be characterized as having a very low species diversity; low number of identifiable specimens (NISP) per scat; bones extremely broken up (very small size range); skeletal part frequencies similar in some ways to those resulting from carnivore partial digestion; and sometimes articulated specimens. Modifications to the bones include corrosion, tiny tooth scores and pits, cracking, and fraying of bone edges. Together, these characteristics suggest that hominoid bone digestion may be recognizable, despite some similarities with leopard-, canid-, and eagle-modified bone. Chimpanzees are well-documented hunters of medium-sized vertebrates such as monkeys. This is significant in the study of human evolution if, as it seems, the last common ancestor of chimpanzees and humans was chimpanzee-like. It suggests there was a pre-stone-tool-using hunting phase in human evolution, perhaps by australopiths or the last common ancestor. Taphonomically, pre-stone tool meat eating has been very difficult to detect in the fossil record. However, if chimpanzees leave a recognizable taphonomic signature on the bones of their prey, we will be able to look for analogous signatures in fossil bones associated with fossil hominoids and hominids.