Rapid genome-wide evolution in Brassica rapa populations following drought revealed by sequencing of ancestral and descendant gene pools

Abstract

There is increasing evidence that evolution can occur rapidly in response to selection. Recent advances in sequencing suggest the possibility of documenting genetic changes as they occur in populations, thus uncovering the genetic basis of evolution, particularly if samples are available from both before and after selection. Here, we had a unique opportunity to directly assess genetic changes in natural populations following an evolutionary response to a fluctuation in climate. We analysed genome-wide differences between ancestors and descendants of natural populations of Brassica rapa plants from two locations that rapidly evolved changes in multiple phenotypic traits, including flowering time, following a multiyear late-season drought in California. These ancestor-descendant comparisons revealed evolutionary shifts in allele frequencies in many genes. Some genes showing evolutionary shifts have functions related to drought stress and flowering time, consistent with an adaptive response to selection. Loci differentiated between ancestors and descendants (FST outliers) were generally different from those showing signatures of selection based on site frequency spectrum analysis (Tajima's D), indicating that the loci that evolved in response to the recent drought and those under historical selection were generally distinct. Very few genes showed similar evolutionary responses between two geographically distinct populations, suggesting independent genetic trajectories of evolution yielding parallel phenotypic changes. The results show that selection can result in rapid genome-wide evolutionary shifts in allele frequencies in natural populations, and highlight the usefulness of combining resurrection experiments in natural populations with genomics for studying the genetic basis of adaptive evolution.

Keywords: Brassica rapa; adaptation; climate change; contemporary evolution; natural selection; population genomics; rapid evolution.

Figure 1

Genetic differentiation (F _ST) throughout the genome between predrought ancestors (1997) and postdrought descendants (2004) of Brassica rapa in the (a) Arboretum and (b) Back Bay populations, and (c) within one flowering time gene in the Arb population. For (a) and (b), each point is a gene, and average F _ST was calculated for each gene using 100‐kb sliding windows. The green dashed line indicates the significance cut‐off (q‐value < 0.05), of F _ST > 0.15 for Arb and F _ST > 0.13 for BB. The number of significantly differentiated genes was 434 for Arb and 433 for BB. This shows evidence for rapid evolutionary shifts in these genes. A LOESS trend line is shown in black and grey. In (c), differentiation (F _ST) between ancestors and descendants and expected heterozygosity (H _e) from 1997 (red dashed line) and 2004 (blue dotted line) Arb populations are shown for 4 kb of Bra004928, the SOC1 paralog on chromosome 5. This region shows high F _ST and a decrease in H _e from 1997 to 2004, providing potential evidence of recent selection.

Figure 2

Boxplots of F _ST at different ranges of Tajima's D for (a) Arb and (b) BB descendants (2004). There is no apparent correlation between Tajima's D and F _ST, indicating that regions that were differentiated between 1997 and 2004 (high F _ST) are not the same as regions that were under longer‐term selection (low Tajima's D). Instead, genes with extreme Tajima's D values tended to have low F _ST values.

Figure 3

Few genes evolved in parallel in the Arb and BB populations. (a) Venn diagram of the outlier genes significantly differentiated (q‐value < 0.05) between all temporal and geographic populations. For example, the numbers in the ellipse labelled Arb are the number of outlier genes that were outliers between ancestors and descendants in the Arb population. The genes that overlap between the Arb and BB ellipses (11 genes, shown in bold) are the genes that were outliers in both the Arb and BB populations. (b) Comparison of changes in ancestor‐descendant allele frequencies in Arb and BB populations for the 11 genes with significant differentiation in both populations. Points are coloured by gene (see legend). The major allele for Arb in 2004 is always tracked for each locus. The central point is the mean, and bars represent standard errors. The lack of correlated change between the two populations indicates little evidence for parallel evolution.