Molecular adaptation in flowering and symbiotic recognition pathways: insights from patterns of polymorphism in the legume Medicago truncatula

Abstract

Background: We studied patterns of molecular adaptation in the wild Mediterranean legume Medicago truncatula. We focused on two phenotypic traits that are not functionally linked: flowering time and perception of symbiotic microbes. Phenology is an important fitness component, especially for annual plants, and many instances of molecular adaptation have been reported for genes involved in flowering pathways. While perception of symbiotic microbes is also integral to adaptation in many plant species, very few reports of molecular adaptation exist for symbiotic genes. Here we used data from 57 individuals and 53 gene fragments to quantify the overall strength of both positive and purifying selection in M. truncatula and asked if footprints of positive selection can be detected at key genes of rhizobia recognition pathways.

Results: We examined nucleotide variation among 57 accessions from natural populations in 53 gene fragments: 5 genes involved in nitrogen-fixing bacteria recognition, 11 genes involved in flowering, and 37 genes used as control loci. We detected 1757 polymorphic sites yielding an average nucleotide diversity (pi) of 0.003 per site. Non-synonymous variation is under sizable purifying selection with 90% of amino-acid changing mutations being strongly selected against. Accessions were structured in two groups consistent with geographical origins. Each of these two groups harboured an excess of rare alleles, relative to expectations of a constant-sized population, suggesting recent population expansion. Using coalescent simulations and an approximate Bayesian computation framework we detected several instances of genes departing from selective neutrality within each group and showed that the polymorphism of two nodulation and four flowering genes has probably been shaped by recent positive selection.

Conclusion: We quantify the intensity of purifying selection in the M. truncatula genome and show that putative footprints of natural selection can be detected at different time scales in both flowering and symbiotic pathways.

Figure 1

Geographic origin of the accessions. Colour-coding refers to the assignation in two groups as inferred using Instruct for K = 2. Group 1 (red) comprises 25 accessions and is broadly distributed geographically. Group 2 (blue) consists of n = 32 accessions.

Figure 2

Box plots summarizing patterns of nucleotide variation in Medicago truncatula. Boxplots (shaded gray) depict the empirical distribution obtained for control fragments. Dots represent individual flowering candidate genes (orange) and symbiotic genes (green). A: Distribution of the scaled mutation rate (as estimated with Watterson's θ) per bp for each fragment. B: Pairwise nucleotide diversity (π). C: Tajima's D statistic for each fragment. Z: standardized Fay and Wu's statistic.

Figure 3

Tests of selective neutrality of polymorphism within each group. The joint distributions of (D, Z) tests statistics expected under neutrality in the Eastern group (group 1, panel 3a) and the Western group (group 2, panel 3b) are plotted using a blue shading for the probability density. Symbiotic genes are plotted as green dots and flowering genes as orange dots using each candidate gene abbreviation. Control loci are represented as black dots. The joint distribution for (D, Z) within each group was obtained through 10⁵ coalescent simulations from models parametrized to fit patterns of polymorphisms in the set of controlled fragments (See Methods for further details). Note that in order to generate a unique graphical representation of the neutral joint distribution, simulations for each of the 53 loci were pooled and the resulting (D, Z) distribution was binned using 10⁴ categories (rigorous p-values computed using null distributions tailored for each locus length and polymorphism are available in Additional file 7, Table S5).

Figure 4

Sliding window analysis of the gene DMI1. Polymorphism was analysed in 500 nucleotide-long windows with 50 nucleotide steps along the alignments. Displayed statistics are: S, the number of polymorphic sites per window, non-synonymous polymorphism given by Watterson's θ per non-synonymous site, F_ST given by Weir and Cockerham's estimator [59] and Tajima's D computed in both groups. The grey frames denote the positions of exons and the arrowheads mark the position of the sites found with high posterior probability (> 0.95) to be targets of positive selection in the Medicago genus.