EVALUATING TAPHONOMIC AND TAXONOMIC BIAS IN STUDIES OF PALEOCOMMUNITY DISSIMILARITY
First, paleocommunities represent only part of the original community. Are differences in paleocommunity composition representative of differences in whole faunas? I compare Jaccard distances between whole living communities against distances between the preservable portions of those communities using data from the California Shelf. These distances are positively correlated, and the preservable subset is as representative of the original community as a random subset of equal size. Thus, differences between paleocommunities reflect the behavior of both preserved and missing soft-bodied taxa.
Second, since fossil species are morphospecies and may suffer from lumping or splitting, some mismatch between the biological species composition of the original community and the morphospecies composition of the paleocommunity is likely. Does this bias studies of ecology in the fossil record? I simulate lumping and splitting in Carboniferous and Permian data from the Paleobiology Database and compare the original and simulated taxonomic distances between communities to the ecological distances between the guild compositions of those communities. Simulated splitting increases or decreases taxonomic distance, with greater effects at low ecological distance and a mean effect of 0. Lumping only decreases taxonomic distance, with larger effects at low ecological distance and a small mean effect. Taxonomic practice can bias measurements of community dissimilarity, especially when communities have similar guild structures, but the mean bias is weak overall.
This work will set bounds on the bias present in future work on ecological and taxonomic turnover in space and time in the Pennsylvanian and Permian of the Midcontinent, USA.