THE EFFECT OF SPECIES-LEVEL EXTINCTION SELECTIVITY ON SUPRASPECIFIC EXTINCTION INTENSITY

The growth in the development of analytical methods for detecting extinction intensity since the establishment of taxonomic databases allowed quantification of extinction rates has not been matched by those methods for detecting extinction selectivity. Most of the latter approaches rely on correlating extinction intensity of taxa with traits, without considering data limitations as thoroughly as has been done for extinction intensity. For example, the ability to detect selectivity depends on the preservation of traits that caused selectivity, and that those traits will be observable in fossil samples. Additionally, while selectivity on a large number of traits will cause greater extinction selectivity, it is harder to detect than when only a few, easily measureable traits contributes to extinction selectivity. Thus, testing for correlations of traits with extinction probability does not allow paleontologists to say whether, for example, background extinctions are generally more or less selective than mass extinctions. A method based on the probability of a supraspecific taxon going extinct as a function of its number of component lower taxa allows for just such a test.

Paleontological databases used to generate diversity curves and for quantifying extinction intensity often consist of supraspecific taxa, typically genera or families. These taxa can serve as proxies for the underlying species richness and can be used to test relative changes in diversity while avoiding some of the taxonomic problems that might bias species-level datasets. However, for a given extinction intensity measured at supraspecific taxonomic levels, a range of species extinction intensities could exist if extinction selectivity of species differs. In other words, extinction intensity measured at supraspecific levels is a product of both intensity and selectivity of species extinctions. Therefore, without considering selectivity, it is not possible to say which extinction events measured at supraspecific levels were the most intense. Preliminary analyses show a correlation between measured genus extinction intensities and extinction selectivity, confirming that measured extinction intensities may be partly a product of extinction selectivity.