Stochastic and adaptive processes affect rapid evolution of two invasive plants in North America

Abstract

Evolution can occur rapidly on timescales of years to decades. In addition to adaptive responses to natural selection, stochastic processes such as gene flow and genetic drift may influence rapid evolution in many species. Invasive species are good candidates to study rapid evolution because they experience novel selective pressures in their introduced range, as well as founder effects, population bottlenecks, and long-distance dispersal. Consequently, I studied adaptive and stochastic evolution in Lythrum salicaria and Alliaria petiolata – two species that have rapidly expanded their introduced range in North America. Previous common garden and field studies in Lythrum salicaria showed that invasive populations have locally adapted to their environment, resulting in a latitudinal cline in flowering. To test whether latitudinal clines are replicated in other parts of North America, I analyzed phenological development using herbarium specimens in a ‘virtual common garden’ analysis that controlled for variation in sampling date and local growing conditions. After controlling for time of establishment, I found repeated latitudinal clines in phenology across North America. A non-linear least squares model of phenological development showed phenological curves shift over time, particularly in populations with short growing seasons. North American populations of Alliaria petiolata reproduce primarily through self-pollination, making it a good system to examine rapid evolution at the genome level. Molecular genetic markers are helpful for reconstructing the history of invasion to account for stochastic processes that may affect evolution during invasion (i.e. colonization, drift, gene flow). I analyzed restriction-site-associated DNA sequences (RAD-Seq) to identify single nucleotide polymorphisms (SNPs) in 95 inbred lines from eastern North America. Analysis of allele balance (i.e. proportion of each SNP at a heterozygote locus) showed that most individuals are likely hexaploid. I compared two variant callers, the diploid gstacks, and freebayes, which can call polyploid genotypes. I found gstacks was biased towards calling homozygote loci compared to freebayes. Lastly, I found significant population structure, despite the short history of this species in North America. Moreover, I found two distinct genetic clusters in eastern North American populations, which suggests at least two separate introductions.