Testing for the Genomic Footprint of Conflict Between Life Stages in an Angiosperm and Moss Species

Abstract

The maintenance of genetic variation by balancing selection is of considerable interest to evolutionary biologists. An important but understudied potential driver of balancing selection is antagonistic pleiotropy between diploid and haploid stages of the plant life cycle. Despite sharing a common genome, sporophytes (2n) and gametophytes (n) may undergo differential or even opposing selection. Theoretical work suggests antagonistic pleiotropy between life stages can generate balancing selection and maintain genetic variation. Despite the potential for far-reaching consequences of gametophytic selection, empirical tests of its pleiotropic effects (neutral, synergistic, or antagonistic) on sporophytes are generally lacking. Here, we examined the population genomic signals of selection across life stages in the angiosperm Rumex hastatulus and the moss Ceratodon purpureus. We compared gene expression between life stages and sexes, combined with neutral diversity statistics and the analysis of the distribution of fitness effects. In contrast to what would be predicted under balancing selection due to antagonistic pleiotropy, we found that unbiased genes between life stages were under stronger purifying selection, likely explained by a predominance of synergistic pleiotropy between life stages and strong purifying selection on broadly expressed genes. In addition, we found that 30% of candidate genes under balancing selection in R. hastatulus were located within inversion polymorphisms. Our findings provide novel insights into the genome-wide characteristics and consequences of plant gametophytic selection.

Keywords: antagonistic pleiotropy; balancing selection; gametophytic selection; gene expression; intralocus conflict.

Fig. 1.

Effect of expression bias between life stages on nucleotide diversity in Rumex hastatulus (a–c) and Ceratodon purpureus (d–f). x axis: quantiles of log2FoldChange between gametophyte and sporophyte expression. Number of genes in each bin: 1,009 to 1,847 for R. hastatulus; 1,943 to 2,564 for C. purpureus. The weighted mean and 95% confidence intervals are based on 1000 bootstraps of the original dataset. Note that the scales on the y axis are different between species.

Fig. 2.

Weighted mean nucleotide diversity of gametophyte-specific, sporophyte-specific, unbiased genes in Rumex hastatulus (a–c) and Ceratodon purpureus (d–f). Number of genes in each category: 1,149 (s), 920 (g), 524 (unbiased) for R. hastatulus; 1,232 (s), 1,217 (g), 3,380 (unbiased) for C. purpureus. The weighted mean and 95% confidence intervals are based on 1,000 bootstraps of the original dataset. Note that the scales on the y axis are different between species.

Fig. 3.

Effect of expression level on weighted mean nucleotide diversity in Rumex hastatulus (a–c) and Ceratodon purpureus (d–f). Number of genes in each bin: 686 to 1,464 for R. hastatulus; 1,034 to 2,328 for C. purpureus. The weighted mean and 95% confidence intervals are based on 500 bootstraps of the original dataset. qt1: lowest expression, qt10: highest expression, g: gametophyte-biased genes, s: sporophyte-biased genes. Note that the scales on the y axis are different between species.

Fig. 4.

Effect of expression bias between life stages on Tajima's D in Rumex hastatulus (a, b) and Ceratodon purpureus (c, d). x axis: quantiles of log2FoldChange between gametophyte and sporophyte expression. Number of genes in each bin: 1,009 to 1,734 (D_s), 1,197 to 1,966 (D_n) for R. hastatulus; 2,138 to 2,534 (D_s), 2,097 to 2,579 (D_n) for C. purpureus. Error bars represent mean ± SEM across genes in each bin. Note that the scales on the y axis are different between species.

Fig. 5.

Genome-wide scan of balancing selection in Rumex hastatulus. y axis: genetic positions of sites being tested (a), composite likelihood ratio (b). $\hat{a}$: estimated dispersion parameter, a positive log₁₀$\hat{a}$ value suggests balancing selection.