Testing archaeofaunal collections for differential fragmentation

Body dimensions of animal remains provide evidence for subsistence technologies, seasonality of hunting and fishing, and generalized resource stress. Studies that rely on reconstructed body dimensions of fishes rarely address the possibility of differential preservation or recovery, which may result in inaccurate representations of size classes. We devise a method to test for differential fragmentation of archaeofauna1 specimens based on specimen size and location within the stratigraphic column. We also compare two methods for reconstructing animal size classes: one includes only complete specimens that can be measured accurately, and the other includes both fragmented and complete specimens. We use a collection of mullet (Mugil spp.) atlases from the McQueen Shell Ring, St. Catherines Island, Georgia (USA), many of which were included in a previous study of fish body size as evidence for fishing technologies. All atlases are sorted by size class and categorized as having complete or fragmented osteometric landmarks. There is a positive correlation (ρ = 0.45, P = 0.07) between atlas size class and proportion of specimens that are fragmented: smaller atlases are more likely to be recovered, identified, and measured without fragmentation compared to larger atlases. Size-based differential fragmentation may bias against the representation of larger-bodied individuals in this context. Frequency of fragmentation is not monotonically correlated with burial depth (ρ = −0.19, P = 0.39), and is higher in the upper-most and bottom-most levels of the stratigraphic column compared to intermediate depths. Depth-based differential fragmentation may bias results when stratigraphic sequences are used to evaluate change through time in animal body sizes.