Résumé
Germplasm from perennial ryegrass (Lolium perenne L.) natural populations
is useful for breeding because of its adaptation to a wide range of
climates. Climate-adaptive genes can be detected from associations between
genotype, phenotype and climate but an integrated framework for the
analysis of these three sources of information is lacking. We used two
approaches to identify adaptive loci in perennial ryegrass and their
effect on phenotypic traits. First, we combined Genome-Environment
Association (GEA) and GWAS analyses. Then, we implemented a new test based
on a Canonical Correlation Analysis (CANCOR) to detect adaptive loci.
Furthermore, we improved the previous perennial ryegrass gene set by de
novo gene prediction and functional annotation of 39,967 genes. GEA-GWAS
revealed eight outlier loci associated with both an environmental variable
and a phenotypic trait. CANCOR retrieved 633 outlier loci associated with
two climatic gradients, characterized by cold-dry vs mild-wet winter and
long rainy season vs long summer, and pointed out traits putatively
conferring adaptation at the extremes of these gradients. Our CANCOR test
also revealed the presence of both polygenic and oligogenic climatic
adaptations. Our gene annotation revealed that 374 of the CANCOR outlier
loci were positioned within or close to a gene. Co-association networks of
outlier loci revealed a potential utility of CANCOR for investigating the
interaction of genes involved in polygenic adaptations. The CANCOR test
provides an integrated framework to analyze adaptive genomic diversity and
phenotypic responses to environmental selection pressures that could be
used to facilitate the adaptation of plant species to climate change.