Signatures of local adaptation in lowland and highland teosintes from whole‐genome sequencing of pooled samples

Abstract

Spatially varying selection triggers differential adaptation of local populations. Here, we mined the determinants of local adaptation at the genomewide scale in the two closest maize wild relatives, the teosintes Zea mays ssp parviglumis and ssp. mexicana. We sequenced 120 individuals from six populations: two lowland, two intermediate and two highland populations sampled along two altitudinal gradients. We detected 8 479 581 single nucleotide polymorphisms (SNPs) covered in the six populations with an average sequencing depth per site per population ranging from 17.0× to 32.2×. Population diversity varied from 0.10 to 0.15, and linkage disequilibrium decayed very rapidly. We combined two differentiation-based methods, and correlation of allele frequencies with environmental variables to detect outlier SNPs. Outlier SNPs displayed significant clustering. From clusters, we identified 47 candidate regions. We further modified a haplotype-based method to incorporate genotype uncertainties in haplotype calling, and applied it to candidate regions. We retrieved evidence for selection at the haplotype level in 53% of our candidate regions, and in 70% of the cases the same haplotype was selected in the two lowland or the two highland populations. We recovered a candidate region located within a previously characterized inversion on chromosome 1. We found evidence of a soft sweep at a locus involved in leaf macrohair variation. Finally, our results revealed frequent colocalization between our candidate regions and loci involved in the variation of traits associated with plant-soil interactions such as root morphology, aluminium and low phosphorus tolerance. Soil therefore appears to be a major driver of local adaptation in teosintes.

Phylogeographic and population genetic analyses ofCucurbita moschatareveal divergence of two mitochondrial lineages linked to an elevational gradient

“…The BAPS analysis also supported two main clusters: rather than a subdivision among geographic areas, these two clusters highlighted a clear altitudinal cline (Appendix S3). As previously mentioned, similar patterns of genetic structure between highland and lowland landraces have been noted in other domesticated taxa (e.g., maize and barley; Tanto et al, 2010;Van Heerwaarden et al, 2011); and genomic analyses have uncovered signals of local adaptation to these altitudinal gradients (Aguirre-Liguori et al, 2017;Fustier et al, 2017). Divergence in highland and lowland, together with human dispersal over large areas (Besnard et al, 2013), could explain the current wide geographic-altitudinal tolerance of C. moschata (Lira, 1995).…”

Genomic footprints of local adaptation along elevation gradients associate with present phenotypic variation in teosintes

“…The number of SNPs associated per trait varied from 3 to 61, with no association for leaf and grain coloration (S5 Table). Our results therefore point to a high proportion of associated SNPs, and since LD is very limited in teosinte populations (decaying within 490 <100bp, [55]) this indicates that phenotypic variation is driven by numerous independent loci. Consistently, Weber et al [83] found that individual SNPs account for small proportions of the phenotypic variance.…”

“…A substantial proportion of our set of outlier SNPs was chosen due to their significant correlation among six populations between variation of allele frequencies and the projection on the first environmental principal component [55]. We showed here that these correlations hold on our broad sample of 28 populations, with allele frequency at 115 (53%) SNPs correlating with environmental variation.…”

“…We successfully genotyped 218 (~81%) out of the 270 outlier SNPs, of which 141 were previously detected in candidate regions for local adaptation [55]. Candidate regions were identified from re-sequencing data of six teosintes populations (S1 Table) following an approach that included high differentiation between highlands and lowlands, environmental 235 correlation, and in some cases their intersection with genomic regions involved in quantitative trait variation in maize.…”

“…More modest structural changes include megabase inversions that harbor clusters of SNPs whose frequencies are associated with environmental variation [52,53] . Also, differentiation-and correlation-based genome scans in six teosinte populations succeeded in finding outlier SNPs for local adaptation [54,55] .Yet, a link with phenotypic variation has 130 yet to be established.…”