Positive phenotypic selection inferred from phylogenies

Rates of phenotypic evolution vary widely in nature and these rates may often reflect the intensity of natural selection.Here we outline an approach for detecting exceptional shifts in the rate of phenotypic evolution across phylogenies.We introduce a simple new branch-specific metric Δ V /Δ B that divides observed phenotypic change along a branch into two components: (1) that attributable to the background rate (Δ B ), and (2) that attributable to departures from the background rate (Δ V ).Where the amount of expected change derived from variation in the rate of morphological evolution doubles that explained by to the background rate (Δ V /Δ B > 2), we identify this as positive phenotypic selection.We apply our approach to six datasets, finding multiple instances of positive selection in each.Our results support the growing appreciation that the traditional gradual view of phenotypic evolution is rarely upheld, with a more episodic view taking its place.This moves focus away from viewing phenotypic evolution as a simple homogeneous process and facilitates reconciliation with macroevolutionary interpretations from a genetic perspective, paving the way to novel insights into the link between genotype and phenotype.The ability to detect positive selection when genetic data are unavailable or unobtainable represents an attractive prospect for extant species, but when applied to fossil data it can reveal patterns of natural selection in deep time that would otherwise be impossible.