Resequencing 93 accessions of coffee unveils independent and parallel selection during Coffea species divergence.

Coffea arabica, C. canephora and C. excelsa, with differentiated morphological traits and distinct agro-climatic conditions, compose the majority of the global coffee plantation. To comprehensively understand their genetic diversity and divergence for future genetic improvement requires high-density markers. Here, we sequenced 93 accessions encompassing these three Coffea species, uncovering 15,367,960 single-nucleotide polymorphisms (SNPs). These SNPs are unequally distributed across different genomic regions and gene families, with two disease-resistant gene families showing the highest SNP density, suggesting strong balancing selection. Meanwhile, the allotetraploid C. arabica exhibits greater nucleotide diversity, followed by C. canephora and C. excelsa. Population divergence (FST), population stratification and phylogeny all support strong divergence among species, with C. arabica and its parental species C. canephora being closer genetically. Scanning of genomic islands with elevated FST and structure-disruptive SNPs contributing to species divergence revealed that most of the selected genes in each lineage are independent, with a few being selected in parallel for two or three species, such as genes in root hair cell development, flavonols accumulation and disease-resistant genes. Moreover, some of the SNPs associated with coffee lipids exhibit significantly biased allele frequency among species, being valuable for interspecific breeding. Overall, our study not only uncovers the key population genomic patterns among species but also contributes a substantial genomic resource for coffee breeding.