Hosts winnow symbionts with multiple layers of absolute and conditional discrimination mechanisms

In mutualism, hosts select symbionts via partner choice and preferentially direct more resources to symbionts that provide greater benefits via sanctions. At the initiation of symbiosis, prior to resource exchange, it is not known how the presence of multiple symbiont options (i.e., the symbiont social environment) impacts partner choice outcomes. Furthermore, little research addresses whether hosts primarily discriminate among symbionts via sanctions, partner choice or a combination. We inoculated the legume, Acmispon wrangelianus, with 28 pairs of fluorescently labeled Mesorhizobium strains that vary continuously in quality as nitrogen-fixing symbionts. We find that hosts exert robust partner choice, which enhances their fitness. This partner choice is conditional such that a strain's success in initiating nodules is impacted by other strains in the social environment. This social genetic effect is as important as a strain's own genotype in determining nodulation and has both transitive (consistent) and intransitive (idiosyncratic) effects on the probability that a symbiont will form a nodule. Furthermore, both absolute and conditional partner choice act in concert with sanctions, among and within nodules. Thus, multiple forms of host discrimination act as a series of sieves that optimize host benefits and select for costly symbiont cooperation in mixed symbiont populations.

Methods
We transformed eight Mesorhizobium strains to express fluorescent markers. We grew Mesorhizobium in symbiosis with a single genotype of the native Acmispon wrangelianus host plant as both 1-strain inoculations (8 treatments) and all possible 2-strain combinations (28 treatments), replicated over 8 blocks. Rhizobia and host plant fitness data were measured in the greenhouse experiment (symbiosis_phenotype_data.csv).  We also measured reproductive fitness of rhizobia in nodules (cfu_data.csv) and measured each strains intrinsic axenic growth rate (growth_data.csv).