HIERARCHICAL DECOMPOSITION OF GEOGRAPHIC RANGE AND ITS IMPLICATIONS FOR UNDERSTANDING EXTINCTION RESISTANCE IN MARINE ANIMAL GENERA

Breadth of geographic range is known to be one of the most consistent predictors of extinction resistance in marine animal species and genera. What is less clear is the importance of how the ranges of genera and other supraspecific taxa are structured. For example, does a composite genus range consisting of many restricted species have different implications for survival compared to one consisting of fewer widespread species?

Using data from the Paleobiology Database, we measure and decompose genus geographic ranges in a way that is analogous to the analysis of variance. The total dispersion of a genus is equal to the sum of squared great-circle distances of its occurrences from its geographic centroid. The total within-species dispersion is equal to the sum of squared distances of species occurrences from the respective species centroids. And the total among-species dispersion is equal to the sum of squared distances of species centroids from the genus centroid, weighted by the number of occurrences of each species. The mean genus dispersion, mean within-species dispersion, and mean among-species dispersion are equal to the respective sums of squares divided by the corresponding degrees of freedom.

We then use logistic regression to model the survival of genera as a function of their mean total dispersion, mean within-species dispersion, and mean among-species dispersion. The geographic dispersion of genera is a strong predictor of survival. Once the genus range is specified, however, the ranges of its constituent species and the dispersion among them provide no additional predictive power. Thus, the emergent properties of the genus matter most, a result inconsistent with the notion that genus-level patterns are epiphenomenal. This result agrees with prior work on survival of mollusks through the end-Cretaceous extinction event, but it is more general.

We also find that species richness is a significant predictor of genus survival, even beyond its contribution to genus geographic range. This result could imply that ecological breadth of supraspecific taxa reduces extinction risk, if congeneric species are ecologically differentiated. Or it could simply be that the species is the fundamental unit of extinction, so that genera with more species are buffered through "strength in numbers."