Zooarchaeology by Mass Spectrometry (ZooMS) is a rapid, low-cost, collagen-based method for the taxonomic identification of animal tissues. It is now increasingly applied to bone fragments from archeological contexts, creating large taxonomic datasets. How to integrate these ZooMS identifications within general zooarchaeological theoretical frameworks, such as estimates of species abundance and taxonomic richness, remains problematic. Past large-scale ZooMS analyses of Eurasian Paleolithic faunal assemblages have shown a general trend towards an increased representation of large ungulates (mainly Bos/Bison) in the ZooMS fraction, often coupled with a decrease in medium-sized taxa (e.g., reindeer). Here we propose several hypotheses to explain these identification discrepancies, involving identification biases and differential fragmentation patterns across various taxa, and test them using the case study of the Paleolithic site of Cassenade.

At the Châtelperronian site of Cassenade (France), nearly all bone fragments larger than 20mm (n=1,119) have been identified to taxa, either through comparative morphology (n=364) or ZooMS (n=755). Each of these fragments was weighed and measured, creating a unique database to explore the relation between fragmentation and identification. Analysis shows that fragment size and mass distributions are distinct across taxa if only bones identified by morphology are considered, but, somehow counter-intuitively, extremely similar across taxa of various body sizes when all their bones are integrated. In particular, the bones of larger ungulates tend to be broken into a higher number of fragments, inducing an over-representation of larger taxa in ZooMS-NISP. Our dataset also shows that sorting long-bone shaft fragments by body size classes should be seen by zooarchaeologists as a process that, in addition to being prone to risks of misidentification, provides highly biased information of little use for estimating species abundance. To overcome this issue, we propose the calculation of an adjusted equivalent ZooMS NISP (ZooMS-eNISP) by dividing, for each taxon, the total ZooMS mass of identified bones (g) by the mean mass of morphologically identified bones for that taxon (g/NISP). The advantage of this method is that it considers site-specific characteristics of the faunal assemblage, notably bone preservation, which is especially important in Paleolithic contexts. Finally, we propose that ZooMS-eNISP, despite its limits, can facilitate the integration of both identification methods to produce a more refined picture of patterns of species representation, site formation, and human behavior at an archaeological site.