ISSR fingerprinting of Coffea arabica throughout Ethiopia reveals high variability in wild populations and distinguishes them from landraces

Forests of SW Ethiopia constitute the native habitat of Coffea arabica and also the place where domestication of Arabica coffee started. Selection from wild populations has led to numerous landraces (farmer’s varieties) and cultivars. Inter-simple sequence repeats (ISSRs) were generated from a representative set of forest coffee populations and landraces across Ethiopia. For the broad diversity assessment, nine di- and tri-nucleotide ISSR primers were applied, as chosen from a total of 102 primers tested initially. Tetranucleotide ISSR primers differed in amplifying fingerprints that could hardly be analysed due to excessive variation. Tree building analysis (NJ, UPGMA) of 84 polymorphic loci amplified for 125 C. arabica individuals provided evidence for several groups of related genotypes occurring in certain geographical areas of Ethiopia and underscored the existence of wild coffee distinct from landraces. Landraces seem to have originated in different geographical areas of Ethiopia in a stepwise domestication process. While the overall geographical signal in the dataset was weak, analysis in a Bayesian framework using the admixture model with geographical priors in STRUCTURE recovered some genetic clustering. Based on Shannon’s diversity index, populations from Yayu (0.47) and Bonga (0.46) showed highest diversity, followed by individuals from Berhane Kontir (0.41). A likely scenario for the differentiation of C. arabica after an allopolyploidization event is that the hierarchical-geographical patterning of wild Coffea genotypes expected from stepwise range extension was obscured by recent or ancient gene flow. The diversity and geographical distribution of autochthonous C. arabica genotypes indicates the need for a multi-site in situ conservation approach.