Predicting the strength of urban-rural clines in a Mendelian polymorphism along a latitudinal gradient
Cities are emerging as models for addressing the fundamental question of whether populations evolve in parallel to similar
environments. Here, we examine the environmental factors that drive the evolution of parallel urban-rural clines in a Mendelian
trait—the cyanogenic antiherbivore defense of white clover (Trifolium repens). Previous work suggested urban-rural gradients in
frost and snow depth could drive the evolution of reduced hydrogen cyanide (HCN) frequencies in urban populations. Here, we
sampled over 700 urban and rural clover populations across 16 cities along a latitudinal transect in eastern North America. In each
population, we quantified changes in the frequency of genotypes that produce HCN, and in a subset of the cities we estimated
the frequency of the alleles at the two genes (CYP79D15 and Li) that epistatically interact to produce HCN. We then tested the
hypothesis that cold climatic conditions are necessary for the evolution of cyanogenesis clines by comparing the strength of clines
among cities located along a gradient of winter temperatures and frost exposure. Overall, half of the cities exhibited urban-rural
clines in the frequency of HCN, whereby urban populations evolved lower HCN frequencies. Clines did not evolve in cities with the
lowest temperatures and greatest snowfall, supporting the hypothesis that snow buffers plants against winter frost and constrains
the formation of clines. By contrast, the strongest clines occurred in the warmest cities where snow and frost are rare, suggesting
that alternative selective agents are maintaining clines in warmer cities. Some clines were driven by evolution at only CYP79D15,
consistent with stronger and more consistent selection on this locus than on Li. Together, our results demonstrate that urban
environments often select for similar phenotypes, but different selective agents and targets underlie the evolutionary response in
Plants were collected in the field and brought back to the lab for phenotyping.
Environmental data was collected from publically available datasets.
Phenotype and environmental data were processed in R v3.6.1. All scripts used for data processing and analyses are contained in the ZIP file associated with this repository.
Code and data are similarly available in this Github repository
See README files in atached ZIP folder.