Population genomic analyses reveal a history of range expansion and trait evolution across the native and invaded range of yellow starthistle (Centaurea solstitialis)

Abstract

Identifying sources of genetic variation and reconstructing invasion routes for non-native introduced species is central to understanding the circumstances under which they may evolve increased invasiveness. In this study, we used genome-wide single nucleotide polymorphisms to study the colonization history of Centaurea solstitialis in its native range in Eurasia and invasions into the Americas. We leveraged this information to pinpoint key evolutionary shifts in plant size, a focal trait associated with invasiveness in this species. Our analyses revealed clear population genomic structure of potential source populations in Eurasia, including deep differentiation of a lineage found in the southern Apennine and Balkan Peninsulas and divergence among populations in Asia, eastern Europe and western Europe. We found strongest support for an evolutionary scenario in which western European populations were derived from an ancient admixture event between populations from eastern Europe and Asia, and subsequently served as the main genetic 'bridgehead' for introductions to the Americas. Introductions to California appear to be from a single source region, and multiple, independent introductions of divergent genotypes likely occurred into the Pacific Northwest. Plant size has evolved significantly at three points during range expansion, including a large size increase in the lineage responsible for the aggressive invasion of the California interior. These results reveal a long history of colonization, admixture and trait evolution in C. solstitialis, and suggest routes for improving evidence-based management decisions for one of the most ecologically and economically damaging invasive species in the western United States.

Keywords: admixture; biological invasion; invasion routes; phylogeography; rapid evolution; restriction site-associated sequencing.

Fig. 1

Evolutionary scenarios tested for two successive steps of ABC analyses of range expansions of the core lineage of YST. The first analysis (a) tested among four evolutionary scenarios for populations in Eurasia. The second analysis (b) incorporated the most highly supported scenario from (a) to test among seven competing evolutionary scenarios for the invasion of California. Scenarios are shown together with their respective tree topologies and relative times of divergence (t₁ represents the oldest split and 0 represents the present day) and admixture (t_Adm1 and t_Adm2). Narrow lines in topologies of (b) indicate population bottlenecks at times DB1 and DB2. Times are not to scale. The best-fit scenario for each analysis is indicated with a bold topology. Additional details regarding evolutionary scenarios and prior distributions for their demographic parameters are described in Appendix S1 and Table S2 (Supporting information) respectively.

Fig. 2

Discriminant analysis of principal components (DAPC) based on single nucleotide polymorphisms, using regions as prior clusters. Results are shown for (a) the analysis that included YST individuals from across the entire sampled range (i.e. the “FULL” data set) and (b) the analysis that included only individuals belonging to the core lineage (i.e. the “CORE” data set). Ovals are 95% inertia ellipses. Individual genotypes are depicted with closed squares (Eurasia), closed triangles (South America), dots (western US), and open triangles (southern Apennine and Balkan Peninsulas), and are colored according to the geographic region from which they were sampled.

Fig. 3

Individual assignments from STRUCTURE analyses and a BAPS admixture analysis based on 1013 SNP loci of 550 individuals belonging to the core lineage of YST, which excludes populations from the southern Apennine and Balkan Peninsulas (i.e. the “CORE” data set). (a) STRUCTURE bar plots of assignment probabilities averaged across ten runs (calculated by CLUMPP) are shown for K = 2 to K = 5, where K is the number of genetic clusters. CLUMPP H’ values indicated low heterogeneity at K = 2 (0.998), K = 4 (0.940), and K = 5 (0.923; Fig. S5, Supporting information). (b) BAPS bar plots are shown for the best estimate of K (4), where K is the number of genetic clusters. Each vertical bar shows the proportional representation of the estimated cluster membership for a single individual.

Fig. 4

Pie charts depicting the average assignment probabilities of individuals of the core lineage of YST at each sampling site in Eurasia (a), the western US (b), and South America (c) to a genetic cluster inferred by STRUCTURE analyses at K = 4 (the average of ten replicates), where K is the number of genetic clusters. Black leader lines indicate the geographic location of nearby sampling sites. Black triangles in Eurasia indicate sampling sites for populations of the divergent Apennine-Balkan lineage of YST.

Fig. 5

Illustration of the most highly supported evolutionary scenario (Scenario 4.1) in the ABC analysis of YST in Eurasia and the Americas, in relation to variation in plant size among regions. (a) Within Europe, eastern European populations diverged from those in Asia, and subsequently these two lineages admixed to form western European populations. The western European lineage then served as the primary source of the YST introduction to Chile, followed by introduction of the Chilean lineage to California. (b) Least Squares Means (+/− standard error of the mean) of a linear size index of biomass. Letters indicate significantly different regions based on Tukey’s HSD post-hoc tests.