Criteria for identifying bone modification by termites in the fossil record

Three geographically dispersed Middle and Later Stone Age cave sites in South Africa, and a Middle Stone Age cave site in Ethiopia, share a similar taphonomic signature that includes destruction of bones associated with variable forms of star shaped features, clusters of microscopically visible sub-parallel striations, edge gnawing, pits, and etching of the bone surface. Similar traces preserved on Plio-Pleistocene fauna are interpreted by different authors as the possible work of termites or ants. Considering that ambiguity exists in the interpretation of these traces and that there are no modern examples available for comparison, we set out to create a reference collection of bones modified by southern African termites. Here we present the results of an actualistic experiment conducted with the harvester termite: Trinervitermes trinervoides (Sjostedt) (Isoptera: Termitidae) in the Sterkfontein Valley, South Africa. Results show that within six months all bones were approximately half covered with a dark surface residue, had an etched surface appearance and recorded boreholes and destruction, particularly of less dense elements and epiphyses. Star-shaped marks, edge gnawing, and clusters of sub-parallel striations on the periosteal surface were faint after six months, but became clearly visible and more plentiful after 12 months, a finding attributed to the termites being more active in austral autumn and spring. This result demonstrates the pertinence of insect modification studies to our understanding of regional palaeoenvironment, palaeoecology and palaeoclimate. This experiment has demonstrated that T. trinervoides can destroy bone in all stages of preservation, favouring fresh thin cortical and spongy bone with meat and marrow. The presence of appreciable quantities of calcium carbonate in termite mounds suggests that termites may be drawn to dolomitic cave sites and the calcium-rich bones they contain to satisfy their mineral requirements. They therefore have the potential to bias taxonomic and element representation, minimum number of individuals, and age profiles in a faunal assemblage, and may account in part for the patchy preservation of faunal remains, including hominins, in fossil deposits, and/or the lack of bone artefacts at some Middle Stone Age cave sites preserving long sequences of occupation. (C) 2012 Elsevier B.V. All rights reserved.