Bidirectional Selection for Body Weight on Standing Genetic Variation in a Chicken Model

Abstract

Experimental populations of model organisms provide valuable opportunities to unravel the genomic impact of selection in a controlled system. The Virginia body weight chicken lines represent a unique resource to investigate signatures of selection in a system where long-term, single-trait, bidirectional selection has been carried out for more than 60 generations. At 55 generations of divergent selection, earlier analyses of pooled genome resequencing data from these lines revealed that 14.2% of the genome showed extreme differentiation between the selected lines, contained within 395 genomic regions. Here, we report more detailed analyses of these data exploring the regions displaying within- and between-line genomic signatures of the bidirectional selection applied in these lines. Despite the strict selection regime for opposite extremes in body weight, this did not result in opposite genomic signatures between the lines. The lines often displayed a duality of the sweep signatures, where an extended region of homozygosity in one line, in contrast to mosaic pattern of heterozygosity in the other line. These haplotype mosaics consisted of short, distinct haploblocks of variable between-line divergence, likely the results of a complex demographic history involving bottlenecks, introgressions and moderate inbreeding. We demonstrate this using the example of complex haplotype mosaicism in the growth1 QTL. These mosaics represent the standing genetic variation available at the onset of selection in the founder population. Selection on standing genetic variation can thus result in different signatures depending on the intensity and direction of selection.

Keywords: Chicken; White Plymouth Rock; body weight; quantitative trait; selective sweeps.

Figure 1

Representation of the formation of genomic signatures in the Virginia body weight lines. Ancestral haplotypes that contributed to the formation of the White Plymouth Rock breed would have recombined over time. Haplotype variation was likely constrained in the seven partially inbred lines, which were crossed to form the founder population of the Virginia body weight lines. The bidirectional change in eight-week body weight in the Virginia body weight lines is presented in the graph from the parental (P0) across the selected generations (S1-S60) for the high weight selected (HWS; blue unbroken line) and low weight selected (LWS; orange unbroken line) body weight lines. Body weight within relaxed sublines are also shown for the high weight relaxed line (HWR; blue broken line) and low weight relaxed line (LWR; orange broken line) at corresponding selected generation. Haplotype frequency and ancestral haplotype origin are depicted at two hypothetical loci to the right. At locus 1, HWS continues to segregate with multiple haplotypes, whereas LWS is fixed for an extended blue haplotype. At locus 2, HWS is fixed for an extended red haplotype, whereas LWS continues to segregate with multiple haplotypes.

Figure 2

Heterozygosity across chromosomes of divergently selected Virginia body weight chicken lines. A. Heterozygosity in generation 55 for high weight selected (HWS; black line above x-axis) and low weight selected (LWS; gray line below x-axis) presented across the chicken chromosomes within the differentiated regions shaded using color code: gray where neither line is fixed (); green where only HWS is fixed (); blue where only LWS is fixed within the differentiated region; red where both lines are fixed. B. F_ST and heterozygosity patterns across chromosome 2:58-63 Mb for selected generations 55 and 40 and relaxed generation 9. C. F_ST and heterozygosity patterns across chromosome 2:108-113 Mb for selected generations 55 and 40 and relaxed generation 9. Panels within B and C insets: Panel 1: Detail of heterozygosity trace from chromosome map. Panel 2: Mean F_ST within 1 kb windows between HWS and LWS at generation 55 indicated with gray points; region of differentiation indicated with the orange line above F_ST plot; allele frequencies of SNP markers with association to body weight indicated with blue diamonds. Panel 3-8: Mean adjusted allele frequency (adjAF) of 5kb windows in HWS generation 55 (HWS55) / HWS generation 40 (HWS40) / high weight relaxed generation 9 (HWR9) / LWS generation 55 (HLWS55) / LWS generation 40 (LWS40) / low weight relaxed generation 9 (LWR9). Shaded colors within these plots have been used to highlight the runs of adjusted allele frequencies that contribute to different haploblocks.

Figure 3

Mean heterozygosity in HWS (above the x-axis; gray) and LWS (below the x-axis; black) at generation 55 within differentiated regions greater than 0.5 Mb in length. Differentiated regions overlapping known associations with 8-week body weight (Zan et al. 2017) are indicated in orange. Regions presented are sorted in order of increasing cumulative heterozygosity across both lines with increased heterozygosity in HWS.

Figure 4

A. F_ST and heterozygosity patterns across chromosome 1:169-174 Mb for selected generations 55 and 40 and relaxed generation 9 of the Virginia body weight chicken lines. B. F_ST and heterozygosity patterns across chromosome 1:169.2-170.3 Mb for selected generations 55 and 40 and relaxed generation 9 with gene positions. Panel 1: Heterozygosity. Panel 2: Mean F_ST within 1 kb windows (gray points) in selected generation 55; defined region of differentiation indicated with orange line above F_ST plot; blue diamonds indicate allele frequencies of SNP markers with significant association to body weight from Zan et al. (2017) . Panel 3-8: Mean adjusted allele frequency of 5kb windows in HWS55 (panel 3), HWS40 (panel 4), HWR09 (panel 5), LWS55 (panel 6), LWS40 (panel 7), LWR09 (panel 8). Colors within the plots are used to highlight the runs of adjusted allele frequencies that contribute to different haploblocks.