Genetic variation in wild and cultivated Arabica coffee (Coffea arabica L.): Evolutionary origin, global distribution, and its effect on fungal disease incidence in Southwest Ethiopia

Crop biodiversity, which comprises both the diversity within and between crop species and their wild relatives, is highly valued because of its potential to adapt crops to changing environmental conditions and consumer’s demands. Crop diversity is currently threatened land use conversion and agricultural intensification. The characterisation of the genetic diversity and make-up of the wild and cultivated gene pool of a crop is key to make conservation decisions and for crop improvement. Yet, such a characterisation can be challenging due to the hybrid and/or polyploid background of many crops. Arabica coffee (Coffea arabica) is an example of an allotetraploid, self-compatible crop (2n = 4x = 44) of which the gene pool is insufficiently characterised to optimally conserve and use its diversity. The species was presumably founded after a natural hybridization event between two wild Coffea species, but the identity of its putative progenitors, the timing of the hybridization event, and the evolution of self-compatibility in the species require further research. Contemporary wild C. arabica populations are mainly present in the Afromontane rainforest in Southwest Ethiopia, from which the crop was reportedly distributed to multiple tropical and subtropical regions across over fifty countries around the world. This historical distribution led to the worldwide spread of two lineages of coffee varieties known as Typica and Bourbon, and likely coincided with a strong genetic bottleneck in the globally cultivated Arabica coffee gene pool. As the genetic relationships between Arabica coffee varieties from different regions were not sufficiently resolved, the make-up of the cultivated Arabica coffee gene pool is currently to a large extent unclear.

Furthermore, Arabica coffee yields have been declining due to multiple fungal diseases, most notably coffee leaf rust (caused by Hemileia vastatrix) and coffee berry disease (caused by Colletotrichum kahawae). Both diseases also occur in Arabica coffee sites in Southwest Ethiopia along with less common fungal diseases such as coffee wilt disease (caused by Fusarium/Giberella xylarioides) and Armillaria root rot (caused by Armillaria spp.). Most Arabica coffee sites in Southwest Ethiopia are forest coffee sites with little human disturbance or semi-forest coffee sites owned by smallholder farmers with a varying management intensity. To reduce yield losses caused by coffee berry disease in Southwest Ethiopia, locally bred resistant coffee cultivars were introduced to the region. However, wild C. arabica populations in Ethiopia may harbour additional genetic variation that can be of potential interest in breeding programs for a lower fungal disease susceptibility in cultivated Arabica coffee. For instance, when populations with a higher genetic diversity or a certain genetic composition might have a lower incidence of fungal diseases, this may suggest the presence of a genetic component that lowers fungal disease incidence. Nonetheless, it is unknown if fungal disease incidence correlates to genetic diversity and genetic composition in wild C. arabica populations. The genetic variation in wild C. arabica populations might also be at risk due to the nearby presence of introduced coffee berry disease resistant coffee cultivars. Gene flow from these coffee cultivars to their wild relatives can permanently transfer genetic material from cultivars into the wild gene pool (i.e. crop-wild introgression), which may erode the wild coffee gene pool if crop-wild hybrids outcompete wild individuals. Although disease resistant cultivars are likely present across many coffee sites with wild C. arabica individuals, a thorough estimate of their abundance in these sites is lacking. Such an estimate would provide crucial information for a risk assessment on crop-wild introgression in Ethiopian Arabica coffee.

The characterisation of the C. arabica gene pool was addressed in this thesis at three levels: (i) the genetic relationships between C. arabica and congeneric species, (ii) the genetic diversity and composition in wild C. arabica populations in Southwest Ethiopia, and (iii) the genetic diversity in globally cultivated C. arabica varieties. Genetic variation was assessed as Single Nucleotide Polymorphisms in high-throughput sequencing data from genotyping-by-sequencing libraries, supplemented with data from whole genome sequencing and HiPlex amplicon sequencing libraries to quantify variation in globally cultivated coffee cultivars.

First, we inferred genetic relationships between C. arabica and other Coffea species based on Jaccard genetic distance estimates and a multilabeled phylogenetic tree to clarify the evolutionary origin of C. arabica. Coffea eugenioides and C. canephora were identified as the putative progenitor species of C. arabica. We estimated the hybridization event at the origin of C. arabica to have occurred between 1.08 million and 543 thousand years ago using a Bayesian molecular clock analysis, which situates the origin of the species in a period with environmental fluctuations. Self-compatibility was suggested to have evolved independently in C. arabica based on a maximum likelihood reconstruction of the evolution of this trait in the coffee genus.

Then we characterised the genetic relatedness between living plants and historical herbarium specimens of Arabica coffee varieties to reconstruct a phylogenetic tree reflecting the spread of Arabica coffee cultivation to different tropical and subtropical regions. We confirmed several previously reported introductions of coffee varieties in different countries and the distinct genetic origin for the two globally cultivated lineages of Arabica coffee (i.e. Typica and Bourbon). The Bourbon lineage assumedly emerged from C. arabica var. Mokka. In addition, thirty-three plant accessions were identified as putative hybrids, of which twenty-seven accessions were not a priori labelled as hybrids. The high number of possibly unintentional hybrids originated from different regions, suggesting that spontaneous hybridization events between coffee varieties may compromise their genetic integrity and authenticity in multiple regions around the world.

Next, we assessed the genetic diversity and genetic composition in pooled leaf samples of sixty Arabica coffee sites in the Jimma region in Southwest Ethiopia. Each pooled sample consisted of sixteen C. arabica individuals. The Arabica coffee sites varied in management intensity and in the incidence of four fungal diseases: coffee leaf rust, coffee berry disease, coffee wilt disease, and Armillaria root rot. We found a correlation between the differences in genetic composition (quantified as principal components of the allele frequency spectra) of Arabica coffee sites and the variation in the incidence of each fungal disease. These correlations did not consistently involve the same principal components, suggesting that different parts of the genetic composition of the Arabica coffee sites can be related to the incidence of a fungal disease. The genetic diversity (measured as the mean expected heterozygosity) did not correlate to the incidences of the four fungal diseases. Hence, the presence of certain genotypes may reduce the incidence of a fungal disease in an Arabica coffee site, but a higher genetic variation may not lower fungal disease incidences.

Finally, we estimated the relative abundance of coffee berry disease resistant cultivars in the same sixty Ethiopian Arabica coffee sites using a combination of field observations, a farmer’s survey, and a molecular fingerprinting approach. The disease resistant cultivars were present in the majority of the Arabica coffee sites. Their relative abundance was spatially structured across the study area, and strongly correlated to the incidence of coffee berry disease. Although the resistant cultivars seem to successfully decrease the incidence of coffee berry disease in Arabica coffee sites, their widespread occurrence may pose an imminent risk to the genetic integrity of the wild coffee gene pool in Ethiopia.

The findings presented in this thesis contribute to the characterisation of the wild and cultivated gene pool of Arabica coffee. They are of potential value for the conservation of wild coffee genetic resources and the improvement of coffee cultivation. We demonstrated that the wild genetic basis of Arabica coffee is relatively narrow, which can likely be attributed to the recent single hybridization event at the origin of the species and its self-compatible nature. The genetic make-up of globally cultivated Arabica coffee was shaped by the historical spread of coffee varieties, advocating for the enrichment of the cultivated gene pool to increase the adaptation potential of Arabica coffee cultivation to a changing environment and changing consumer’s preferences. The conservation of coffee genetic resources seems to be of major interest for coffee breeding, as illustrated by the correlation between the genetic composition of Arabica coffee sites and the incidence of fungal diseases. These conservation efforts are urgent, given the ongoing coffee management intensification in Southwest Ethiopia and the high abundance of coffee berry disease resistant cultivars across the natural distribution area of C. arabica.