RATIONALE Testing whether and how natural populations are adapted to their local environment and predicting their response to future habitat alterations is of key importance in the face of future climate change. This is particulary the case for coffee trees for which the pace of climate change could be too fast and drastic for population adaptations. Using the geographic distribution of wild populations with contrasted habitats, the aim of the present study was to identify single-nucleotide polymorphisms (SNPs) in candidate genes (CGs) identified as being involved in the adaptation of C. canephora populations to their local environment. By identifying environmental factors driving these processes we would predict the adaptedness of the populations to their future local climate. METHODS Based on the previous molecular studies (EMBRAPA/CIRAD/Nestle studies) and using whole coffee genome sequence annotation (Denoeud et al. 2014, Dereeper et al. 2015), a set of 324 CGs was selected, such as those coding for dehydrins, heat shock proteins, enzymes of sugar metabolism, as well as transcription factors like DREB/CBF (dehydration responsive element binding/cold-binding factor). Wild accessions of C. canephora from Uganda with recorded position (geo-localized samples) were used to assess the relationship between climate variation (www.worldclim.org/bioclirn) and CG nucleic diversity. We apply available statistical population genomic methods and model of allele distribution to detect CGSNPs correlated with climate parameters. The LFMM (Latent Factor Mixed Models) R package (Frichot et al. 2013) was used for screening sequences for signatures of environmental adaptation in coffee genomes. RESULTS The genotype-environment (GxE) association suggests regional adaptation with spatially varying environments. More specifically, we found selection signals tightly linked to several CGs involved in response to biotic and abiotic stress like MYB20 (coding a transcription factor involved