Abstract
Human activities are driving rapid phenotypic change in many species, with harvesting considered to be a particularly potent evolutionary force. We hypothesized that faster evolutionary change in human-disturbed populations could be caused by a strengthening of phenotypic selection, for example, if human disturbances trigger maladaptation and/or increase the opportunity for selection. We tested this hypothesis by synthesizing 1,366 phenotypic selection coefficients from 37 species exposed to various anthropogenic disturbances, including harvest. We used a paired design that only included studies measuring selection on the same traits in both human-disturbed and control (not obviously human-disturbed "natural") populations. Surprisingly, this meta-analysis did not reveal stronger selection in human-disturbed environments; in fact, we even found some evidence that human disturbances might slightly reduce selection strength. The only clear exceptions were two fisheries showing very strong harvest selection. On closer inspection, we discovered that many disturbances weakened selection by increasing absolute fitness and by decreasing the opportunity for selection-thus explaining what initially seemed a counterintuitive result. We discuss how human disturbances can sometimes weaken rather than strengthen selection, and why measuring the total effect of disturbances on selection is exceedingly difficult. Despite these challenges, documenting human influences on selection can reveal disturbances with particularly strong effects (e.g., fishing), and thus better inform the management of populations exposed to these disturbances.