Published online by Cambridge University Press: 08 April 2016
The evolutionary bootstrap is a new approach to the analysis of patterns of taxonomic diversity. In general, the evolutionary bootstrap works by surveying the diversity history of a taxon, learning its dynamic properties, and then generating randomly large numbers of artificial diversity histories based upon what was learned. The distribution of artificial—or bootstrapped—diversity histories approximates the distribution of diversity histories that were possible for taxa with the dynamic properties of the real taxon, and serves as a paleontological null hypothesis for studying statistically the diversity history of the real taxon.
Two null hypotheses were established, the additive and the multiplicative. The additive null hypothesis assumes that the amount of diversity change that occurs in a higher taxon during an interval of time is independent of the number of member subtaxa present at the beginning of the interval. The multiplicative null hypothesis, in contrast, assumes that the amount of diversity change that occurs is dependent upon the number of member subtaxa present at the start. Thus the two null hypotheses represent end members of a diversity-independent/diversity-dependent continuum of possibilities.
Detailed analyses using the evolutionary bootstrap, in conjunction with the clade statistics of Gould et al. (1977), show that several of the 17 higher taxa studied have diversity histories that are statistically significantly different from the random expectation under one or both null hypotheses. Analyses of multiple taxa in aggregate also reveal several properties of diversity histories that are statistically significantly different from random. Real taxa tend to have higher uniformities and lower maximum diversities than expected under the multiplicative null hypothesis. They have lower uniformities, higher maximum diversities, and longer durations than expected under the additive null hypothesis. And, they have lower centers of gravity than expected under either null hypothesis. Overall, the results provide a possible statistical verification of the process of taxonomic (traditionally, adaptive) radiation and suggest the need to consider deterministic explanations for observed diversity patterns.