Résumé
Uncovering genomic regions involved in adaption is a major goal in
evolutionary biology. High-throughput sequencing now makes it possible to
tackle this challenge in nonmodel species. Yet, despite the increasing
number of methods targeted to specifically detect genomic footprints of
selection, the complex demography of natural populations often causes high
rates of false positive in gene discoveries. The aim of this study was to
identify climate adaptations in wild pearl millet populations, Cenchrus
americanus ssp. monodii. We focused on two climate gradients, one in Mali
and one in Niger. We used a two-step strategy to limit false-positive
outliers. First, we considered gradients as biological replicates and
performed RNA sequencing of four populations at the extremities. We
combined four methods—three based on differentiation among populations and
one based on diversity patterns within populations—to identify outlier
SNPs from a set of 87 218 high-quality SNPs. Among 11 155 contigs of pearl
millet reference transcriptome, 540 exhibited selection signals as
evidenced by at least one of the four methods. In a second step, we
genotyped 762 samples in 11 additional populations distributed along the
gradients using SNPs from the detected contigs and random SNPs as control.
We further assessed selection on this large data set using a
differentiation-based method and a method based on correlations with
environmental variables based. Four contigs displayed consistent
signatures between the four extreme and 11 additional populations, two of
which were linked to abiotic and biotic stress responses.