Selective extinction by definition causes some groups to become extinct at higher rates than others and is often assumed to be the cause of functional diversity changes following mass extinctions. However, significant changes in functional diversity can occur even when extinction is nonselective (random). Counterintuitively, lack of functional diversity change can also be evidence of selective extinction. This can occur because (1) the number of taxa in each functional group varies and (2) the abundance of each functional group varies. Consequently, random removal of taxa or individuals disproportionately affects rare functional groups. We simulate random extinction with a range of initial rank abundance distributions, varying the number of taxa in each functional group and the evenness of abundances across taxa. We then measure change in functional diversity before and after random extinction using PERMANOVA. This generates a null distribution of how functional diversity would be expected to change under random extinction for each initial rank abundance distribution. Change in functional diversity outside the null distribution is then positive evidence of nonrandom extinction, nonrandom origination, or nonrandom changes in abundance.

We apply this approach to examine changes in the bryozoan fauna of Stevns Klint, Denmark following the Cretaceous-Paleogene mass extinction, using a dataset of 4,453 specimens from 13 pre-extinction and 7 post-extinction samples. We find that functional diversity changes significantly across the K/Pg boundary, but functional change is largely consistent with random extinction, followed by an increase in the abundances of the functional groups with the fewest taxa. Our approach thus allows us to differentiate between selection extinction and other mechanisms of functional diversity change, which can be applied to other extinction events.