WHAT, IF ANYTHING, IS A GENUS? TESTING THE ANALYTICAL UNITS OF PALEOBIOLOGY AGAINST MOLECULAR DATA

Genera are the taxonomic rank of choice for most large-scale paleontological and biogeographical analyses. However, morphologically-defined genera (“morphogenera”) have repeatedly been called into question, not only in terms of the comparability of the rank among higher taxa, but more fundamentally in terms of species composition: the explosion of molecular phylogenies has indicated that morphology is not always a reliable indicator of genetic relationships. For an array of paleobiologically important clades (bivalves, gastropods, corals, forams, echinoderms, brachiopods, and non-volant mammals), we statistically compared the species composition of morphogenera to the composition of those genera when defined from molecular phylogenetic data (for a database of 440 genera and 2576 species, from 181 molecular analyses). Monophyletic morphogenera were far more pervasive than expected by chance, the distribution of polyphyly was not an artifact of sample size, and polyphyletic morphogenera were concentrated in a few groups long known to be problematic (e.g. corals, nonmarine mollusks). Less than 30% of morphogenera in mammals and marine mollusks, two groups heavily used by paleontologists, are polyphyletic in molecular phylogenies. To evaluate the implications of that 30% figure, we compiled data on two key macroevolutionary and macroecological variables, (a) species-level and total geographic range and (b) mean and variance in body size, and found that all of these values for morphogenera are tightly correlated with (and often identical to) the values for molecularly defined genera. In some ways these results represent a worst-case estimate, in that (1) many molecular studies focus on “difficult” groups, where morphogenera have been exceptionally unstable, and (2) we assumed no phylogenetic error in the molecular phylogenies, despite numerous inconsistencies among molecular approaches. These results suggest that morphogenera, while imperfect, are robust analytical units for many purposes.