Legumes play an increasingly important role in sustainable agriculture due to their ability to form a beneficial symbiotic interaction with nitrogen-fixing Rhizobium bacteria. Legumes are also a valuable source of protein for both feed and food, but are not grown as extensively as expected in Europe due to their high yield variability. Many genomic approaches are being developed to improve stress tolerance traits. However, to date, little attention has been paid to improving the interaction between symbiotic partners. The establishment of the symbiotic interaction is a complex evolutionary process in which the interests of both partners are not always aligned. No evidence was found in pea for co-selection of competitiveness for nodulation and nitrogen (N) fixation efficiency (Bourion et al., 2018). Furthermore, several data indicated that N fixation and plant growth could be suboptimal in fields where pea is exposed to populations of heterogenous rhizobial strains with contrasting effects on nodule, root and shoot development (Laguerre et al., 2007). We performed Genome-Wide Association Studies to decipher the genetic determinants and relationships between the complex trait of pea choice between rhizobial strains in mixture and plant architecture. A large panel of 340 pea accessions including very diverse cultivars, wild accessions and landraces, all inoculated with the same mixture of 28 diverse rhizobial strains, was grown in two successive experiments, on a high throughput non-destructive phenotyping platform. The proportion of each strain in the nodules of each pea at harvest was determined by DNA metabarcoding, and 20 variables of nodulated root architecture or plant growth traits were estimated by image analysis or measured. The results highlighted differential variation and largely uncoupled genetic bases between rhizobial partner choice and architectural or growth traits, along the pea domestication gradient.