Phylogeography and rhizobia-legume genotypic interactions in common bean (Phaseolus vulgarisL.) and chickpea (Cicer arietinumL.) in Ethiopia

Grain legumes are major components of cropping systems in the tropics that provide important products such as food and feed and that enhance soil fertility through biological nitrogen fixation (BNF) when being in symbiosis with rhizobia. As is the case with other crops, the productivity of these grain legumes in the tropics, and particularly in developing countries in sub-Saharan Africa, is very poor, due to low soil fertility, low input supply and poor agronomic practices. It has been shown that the application of rhizobial inoculants can improve the nitrogen fixation efficiency of the legumes and their grain yield; but it has also been shown that legumes like beans and chickpea poorly fix nitrogen and respond erratically to rhizobium application (inoculation). Among factors affecting the efficiency of nitrogen fixation, the identity of legume genotypes (GL) and of rhizobial strains (GR) and the interaction between them are important. The main objective of this study is to explore the genetic diversity and phylogeography of rhizobia nodulating beans and chickpea in Ethiopia (Chapters 2 and 3) and to evaluate the interaction (Chapter 4 and 5) between the host genotypes (chickpea and common bean) and their respective rhizobial strains.The study was conducted in greenhouses and laboratories at Hawassa University in Ethiopia and at Wageningen University & Research in the Netherlands. Rhizobia nodulating common bean were trapped from soils in southern Ethiopia using the locally released cultivars Hawassa Dume, Nasir and Ebadu. They were characterized by a multilocus sequence analysis of 16s rRNA, gyrB, glnII, recA and rpoB genes and symbiotic genes nodC and nifH. Previously-published common-bean rhizobia from Ethiopia were also included; their phylogeny was reconstructed, and their genetic diversity and geographical distribution were evaluated. Chickpea-nodulating rhizobia were retrieved from previous collections maintained at the Laboratory of Soil Microbiology of the Hawassa University and similarly characterized, except that glnII was replaced with atpD; they were analysed along with previously published collections from Ethiopia. The factors influencing the phylogeographic distribution of these strains were studied using principal coordinates analysis and Mantel and partial Mantel tests. For the study of GL x GR interaction, common bean and chickpea genotypes were selected, each one belonging to a specific genetic cluster. Similarly, representative rhizobial reference strains for chickpea and common bean were selected on the basis of their genetics and geographical origin. Nodulation and biomass production were evaluated for factorial combinations of legume and rhizobium genotypes in modified Leonard jars for primary screening; subsequently, reduced combinations were evaluated in pot experiments.I detected seven genospecies of rhizobia nodulating common bean (Chapter 2) and four genospecies of mesorhizobia nodulating chickpea (Chapter 3). The isolates were genetically less diverse compared to reference species known to nodulate the host legumes, which reflects the selective nature of these hosts. Some genospecies from local collections seem to be specific to Ethiopia (R. aethiopicum, M. abyssinicae). The symbiotic gene nodC was found to be diverse among bean rhizobia while it is commonly shared among our current collection of chickpea rhizobia. This could mean that a relatively well-conserved symbiotic island is shared by different mesorhizobial species. The species R. etli and R. phaseoli are the dominant symbionts of common bean in Ethiopia, while most of the chickpea- nodulating strains belong to M. hawassense and/or M. plurifarium that were originally isolated from tree legumes. This suggests that symbiotic genes were transferred from M. ciceri or M. mediterraneum to the mesorhizobia of tree legumes, equipping them to nodulate chickpea. The geographical structure among the rhizobia of common bean was weak, with most genospecies spread evenly in space. There was no evidence of genetic differences being dependent on distance. In contrast, chickpea rhizobia showed a non-random geographical distribution of genospecies; differences were found at the locus and nucleotide level. In both cases, we found evidence of non-spatial biases in taxonomic composition between different strain collections. We found no evidence for a relationship between the genetic distance and differences in symbiotic performance of mesorhizobia, suggesting that genetic information does not help to identify elite strains.The analysis of GL x GR interaction (Chapters 4 and 5) revealed strong patterns of interaction both for beans and chickpea in terms of nodulation and symbiotic effectiveness in jar experiments. These patterns were not detected in the pot experiments, which suggests that the genetic interaction depends on the growth environment. The presence of weak or the absence of any GL x GR interaction in the pot experiment, suggests that identifying universally superior strains is possible. Well-known commercial type strains R. tropici CIAT 899T and CFN 42T were found to be stable and superior performers with bean genotypes, while the reference strain M. ciceri LMG 14989T was consistently effective across chickpea genotypes.