Characterization of Coffea chloroplast microsatellites and evidence for the recent divergence of C. arabica and C. eugenioides chloroplast genomes

Abstract

Comparative sequencing of >7 kb of highly variable chloroplast genome regions (atpB–rbcL, trnS–trnG, rpl22–rps19, and rps19–rpl2 spacers; introns in atpF, trnG, trnK, and rpl16) with microsatellites known from other angiosperms was carried out in Coffea . Samples comprised 8 diploid species of Coffea, 5 individuals of tetraploid C. arabica representing geographically distant wild populations from Ethiopia, 2 commercial cultivars of C. arabica, and Psilanthus leroyi and Ixora coccinea as outgroups. Phylogeny reconstruction using maximum parsimony and Bayesian inference resulted in congruent topologies with high support for C. arabica and C. eugenioides being sisters. Partitioned analyses showed that all regions except the atpB–rbcL spacer resolved this sister-group, although this was often unsupported. The large sequence data set further shows that chloroplast genomes of C. arabica and C. eugenioides each possess apomorphies, indicating that not C. eugenioides but an ancestor or close relative of C. eugenioides is the maternal parent of C. arabica. Seven variable chloroplast microsatellites were characterized in Coffea. Most microsatellites are poly(A/T) stretches, whereas one in the trnS–trnG spacer has an (AT)n motif. Most strikingly, all individuals of C. arabica possess identical sequences, suggesting a single chloroplast haplotype. This can be explained by a recent origin of C. arabica in a unique allopolyploidization event, or by severe bottleneck effects in the evolutionary history of the species. Reconstruction of the evolution of microstructural mutations shows much higher levels of homoplasy in microsatellite loci than in other parts of spacers and introns. Microsatellites are inferred to evolve by insertion and deletion of 1 to 3 motif copies in one step.