Mutualistic hosts often engage in interactions with multiple symbionts simultaneously. Most studies of multiple mutualists have focused on the ecological effects of symbionts on each other, or dual effects on the host. The evolutionary effects of multiple mutualists are much less known, though the way in which host populations evolve in response to selective pressure from multiple interacting partners is important for coevolution and mutualism stability. In the tripartite mutualism between legumes, nitrogen-fixing rhizobium bacteria, and arbuscular mycorrhizal fungi (AMF), host symbiosis traits such as signaling and control over infection density are well-known from physiological and molecular genetic work to possess a shared genetic basis. This unusual a priori evidence for pleiotropy led us to hypothesize that these shared genes would generate genetic correlations between rhizobium and AMF colonization in the host. If host allocation to AMF and rhizobia are inextricably linked by shared symbiosis genes, then the evolution of these traits might be constrained in the face of multiple mutualists. This is important because, given the different but complementary benefits provided by rhizobia and AMF (nitrogen versus phosphorus, respectively), selection on plant allocation to one symbiont versus the other is expected to shift across nutrient environments.
Using a common garden experiment with 75 maternal families of Chamaecrista fasciculata grown in two phosphorus environments in the field, we found mostly positive phenotypic correlations among plant size, colonization by rhizobia (nodule number, nodule size), and colonization by AMF (arbuscule number, hyphal colonization), but little evidence for genetic correlations between colonization by the two symbionts. Nevertheless abundant genetic variation for rhizobium and AMF colonization, paired with natural selection in the field, suggests that these traits should evolve in nature. As expected with phosphorous addition, rhizobium colonization increased while AMF colonization decreased. Moreover genetic variance-covariance matrices differed between environments, indicating that phosphorus addition might affect the evolutionary trajectories of these plant traits. Our key results: 1) suggest that host evolution in response to one symbiont will not be constrained by the other symbiont, and 2) implicate genes other than those known through mutant screens in determining ecologically-relevant variation in these traits. These findings highlight the importance of studying the mechanistic basis of naturally-occurring genetic variation, since these ecological genomic studies will often resolve new genes important for symbiosis.