Assortment of Flowering Time and Immunity Alleles in Natural Arabidopsis thaliana Populations Suggests Immunity and Vegetative Lifespan Strategies Coevolve

Abstract

The selective impact of pathogen epidemics on host defenses can be strong but remains transient. By contrast, life-history shifts can durably and continuously modify the balance between costs and benefits of immunity, which arbitrates the evolution of host defenses. Their impact on the evolutionary dynamics of host immunity, however, has seldom been documented. Optimal investment into immunity is expected to decrease with shortening lifespan, because a shorter life decreases the probability to encounter pathogens or enemies. Here, we document that in natural populations of Arabidopsis thaliana, the expression levels of immunity genes correlate positively with flowering time, which in annual species is a proxy for lifespan. Using a novel genetic strategy based on bulk-segregants, we partitioned flowering time-dependent from -independent immunity genes and could demonstrate that this positive covariation can be genetically separated. It is therefore not explained by the pleiotropic action of some major regulatory genes controlling both immunity and lifespan. Moreover, we find that immunity genes containing variants reported to impact fitness in natural field conditions are among the genes whose expression covaries most strongly with flowering time. Taken together, these analyses reveal that natural selection has likely assorted alleles promoting lower expression of immunity genes with alleles that decrease the duration of vegetative lifespan in A. thaliana and vice versa. This is the first study documenting a pattern of variation consistent with the impact that selection on flowering time is predicted to have on diversity in host immunity.

Fig. 1.

—Distribution of Spearman correlation coefficients between expression levels of each expressed Arabidopsis thaliana gene and flowering time. Gray: All expressed genes; Blue: Genes annotated as flowering time genes (FT genes); Red: Genes annotated as immunity genes; Pink: Flagellin-responsive (FlaRe) genes (Navarro et al. 2004). (A) For 138 Swedish genotypes; (B) Analysis restricted to 51 Swedish genotypes showing correlated flowering time at 10°C and 16°C; (C) Species-wide sample of 52 genotypes. Distribution for each group of genes was compared with the genome-wide distribution (black double-head arrow) with a Kolmogorov–Smirnov test. P values are given in the color corresponding to the gene class. Spearman correlation coefficients were computed between expression levels of each of 23,511 expressed A. thaliana genes, reported in Durbin et al. (2015) for ninth leaf seedlings, and flowering time measured in the same condition for 51 genotypes originating from natural populations in Sweden (Sasaki et al. 2015). ***P < 0.001.

Fig. 2.

—Distribution of Spearman correlation coefficients between gene expression level and flowering time. (A) Partition of genes controlled by flowering time (hatched boxes with blue border) versus independent from flowering time (uniform boxes with black border); (B) Partition of genes controlled by development (hatched boxes with orange border) versus independent from development (uniform boxes with black border). Inserts in the top of the figure illustrates how these gene classes were defined. Immunity genes that are not controlled by flowering time but controlled by development tend to have higher correlation coefficients of natural variation for expression with natural variation for flowering time. Gray: All expressed genes; Blue: Genes annotated as flowering time genes (FT genes); Red: Genes annotated as immunity genes; Pink: Flagellin-responsive (FlaRe) genes (Navarro et al. 2004). P values for Kolmogorov–Smirnov test comparing the distribution of genes within each category that are independent of or regulated by (A) flowering time or (B) age are shown when significant. Note that only 12 FlaRe genes are controlled by flowering time in our experiment. Spearman correlation coefficients were computed between expression levels of each of 23,511 expressed Arabidopsis thaliana genes, reported in Dubin et al. (2015) for ninth leaf seedlings, and flowering time measured in the same condition for 51 genotypes originating from natural populations in Sweden (Sasaki et al. 2015). *P < 0.05, ***P < 0.001.

Fig. 3.

—Distribution of Spearman correlation coefficients between gene expression level and flowering time. All expressed genes—uniform boxes with black border—versus genes with fitness-associated SNPs in Fournier-Level et al. (2011)—hatched boxes with purple border. Gray: All expressed genes; Blue: Genes annotated as flowering time genes (FT genes); Red: Genes annotated as immunity genes. Immunity genes that carry SNPs associating with fitness tend to have higher correlation coefficients of natural variation for expression with natural variation for flowering time. P values for Kolmogorov–Smirnov test comparing the distribution for genes within each category are shown when significant. Spearman correlation coefficients were computed between expression levels of each of 23,511 expressed Arabidopsis thaliana genes, reported in Durbin et al. (2015) for ninth leaf seedlings, and flowering time measured in the same condition for 51 genotypes originating from natural populations in Sweden (Sasaki et al. 2015). *P < 0.05.