Body size and supraspecific extinction risk in terrestrial mammals: A paleontological perspective

Background/Question/Methods

The need for proactive approaches to mammalian conservation has precipitated efforts to identify biological correlates of extinction risk. Studies frequently point to life-history traits and ecological properties that are significantly correlated with body size as reliable predictors of species’ vulnerability to environmental and exploitative pressures. Our understanding of differential extinction risk, however, is largely based on population statuses of living species, recent extinctions on islands, and late-Quaternary collapses of megafaunas. Relatively little is known about biological patterns of extinction above the level of species, on continents, and in the absence of human influence. The rich fossil record of North American mammals provides a valuable data set to address this issue. I investigated the relationship between body size and temporal durations (as a proxy for extinction probability) of 221 genera of extinct terrestrial mammals that are known from 30 to 1 million years ago. After adjusting observed temporal ranges for body-size bias in fossil sampling probability, generalized least-squares regression analyses were conducted to (1) assess the strength of phylogenetic signal in durations and (2) test for correlation between body weights and durations across all the genera and within their subsets grouped by the timing of extinction or trophic position.

Results/Conclusions

The maximum-likelihood estimates for the strength of phylogenetic signal were generally low (Pagel’s λ < 0.3) for the ecological and temporal subsets of 221 genera as well as for the full data set spanning 7 orders of magnitude along the body-weight spectrum. In particular, λ = 0 was obtained for carnivorans and secondary consumers as a whole. The correlation between body weights and durations was non-significant in all analyses except for secondary consumers (df = 71, R² = 0.105, P = 0.003) and, more equivocally, lipotyphlans, depending on the simulated set of branch lengths. Thus, so far as these North American terrestrial mammals from 30 to 1 million years ago are concerned, extinction risk at the genus level exhibits little phylogenetic structure, and body size appears to be a generally poor predictor of extinction probability. In view of the theoretical expectations and empirical observations of elevated extinction risk for large mammalian species, the findings reported here suggest that the process of extinction is distinct at different levels of phylogenetic hierarchy, cautioning against extrapolation of predictions for individual species to more inclusive higher taxa.