ESTIMATING EVOLUTIONARY VOLATILITY AND PREDICTING EXTINCTION USING A LIKELIHOOD-BASED MODEL

The emergence of large online databases enables methods of analyzing the fossil record that would previously have been too time-consuming or statistically weak. This study explores the use of a stochastic birth-death model to obtain maximum likelihood estimates of volatility (the combined rate of speciation and extinction).

Estimating evolutionary rates as model parameters allows for projections of future diversity. The use of a continuous-time model eliminates the need for binning and allows projections to be made at any time scale, from tens to millions of years. Projections in the near future can be used to evaluate extinction risk, and projections in geologic time can be used as a null model for events in the fossil record.

Simulations indicate the method can obtain precise rate estimates for fossil data sets of moderate to large size. Initial tests of the method on bivalve data confirm previous studies showing a correlation between speciation and extinction rates.