Stephen Merritt

Previous zooarchaeological analysis at Koobi Fora indicates that Okote Member hominins were the primary agents of bone assemblage formation, gained early access to large and small mammal flesh, and consumed both high-and low-ranked carcass parts. The discovery of additional butchered specimens prompted the re-analysis presented here of three large and well-preserved zooarchaeological assemblages from the Okote member, GaJi14, FwJj14N and FwJj14S, to revisit paleoecological hypotheses about tool-assisted carnivory. Cow and goat limb butchery documenting the skeletal location of cut marks created by skinning, defleshing, and disarticulation was used to build an actualistic model to infer hominin consumption of distinct carcass resources. Archaeological specimens were assigned to early (defleshing limbs), middle (defleshing ribs, viscera, vertebrae, and head) and late (metapodial tendon removal, element disarticulation, long bone fragmentation) carcass consumption stages, and the incidence of these butchery behaviors was examined for specimens and minimum number of element and individual aggregates. Elbow specimens, where traces of defleshing, disarticulation, and percussion co-occur, offer a sequential view of carcass consumption behaviors that is free from fragmentation bias. Classification trees populated with actualistic data were used to identify defleshing and disarticulation cut mark clusters on archaeological elbow portions by their location, cut mark count, median length, and median cross-sectional width. Actualistically-informed configurational analysis offers high-resolution behavioral reconstruction of the butchered sub-assemblage and should be integrated with assemblage-scale zooarchaeological methods. These experiments highlight the bias for detecting butchery traces of early carcass access, because defleshing cut marks are abundant and introduced to dense midshaft portions, whereas disarticulation cut marks are rare and occur on epiphyseal portions, which are often deleted by density-mediated destruction. Butchery trace interpretation across multiple analytical scales confirms a flexible carnivorous paleoecological role for Okote hominins that included primary and secondary access to carcass resources from large and small mammals.

Zooarchaeological assemblages in a variety of geographic and temporal contexts are dominated by fragmentary long bone specimens, and precise identification of side, skeletal element, and bone portion underlie archaeological interpretations, including specimen counts for skeletal part profiles, minimum number of element (MNE), and individual (MNI) estimates. Actualistic hammerstone and anvil breakage of domestic goat limb bones was used to document how fragmentation impacts precise identification of skeletal specimens, analysis of assemblage composition, and reconstructions of butchery behavior. Specimens greater than 2-cm in size were assigned to categories that describe the precision with which side, element, upper, intermediate and lower limb segment, and long bone portion could be identified. Results suggest that specimen size is positively related to identifiability, and more identifiable specimens tend to include epiphyses and relatively complete shaft circumferences. Most elements produced a similar number of fragments, including highly identifiable ends that yield accurate skeletal part profiles, MNE, and MNI estimates. However, if density-mediated destruction removes these specimens, analysis of less-identifiable shaft fragments significantly underrepresents element and individual abundance. The number of identified limb specimens (NISP), MNE, and epiphysis-to-shaft ratios in fragmentary archaeological butchery assemblages suggest limb end underrepresentation deflates measures of assemblage abundance and reduces the behavioral resolution of butchery interpretations. However, zooarchaeological analyses can productively incorporate fragmentary, less-identifiable specimens when they define hypotheses that match the scale of archaeological data.