Plants can gain substantial growth benefits from microbial symbionts, but these benefits are threatened by ineffective symbionts that infect plants without providing a service. To minimize this threat, plants can preferentially associate with effective symbionts and avoid or punish ineffective symbionts. Although these ‘host control traits’ are central to our understanding of mutualism evolution, we know little about how much they genetically vary within plant species or how they perform in different environments. Here, I investigated variation in host control traits using the California native legume Acmispon strigosus and wild strains of its nitrogen-fixing symbiont, Bradyrhizobium. In the first chapter, I tested for genetic variation in host control traits among six population sources of A. strigosus by inoculating plants with pure cultures of effective and ineffective Bradyrhizobium. In all hosts, the strain content of root nodules was biased toward the most effective Bradyrhizobium strain, but hosts varied genetically in mean nodule size as well as growth benefits. These patterns did not change under experimental nitrogen fertilization. In the second chapter, I examined host genetic variation in nodule size in a new experimental setting, inoculating plants with soil slurries rather than pure Bradyrhizobium cultures. I found similar variation in nodule size and host benefits as I observed in chapter 1, indicating that host control trait variation was robust to the biotic complexity of inocula. Furthermore, plant growth benefits from soils were more strongly driven by plant genotype than soil source, further highlighting the importance of plant genotype. In the third chapter, I investigated host and symbiont contributions to strain content of nodules by inoculating four plant genotypes with nine combinations of effective and ineffective rhizobia strains. I found significant variation among ineffective strains in the relative fitness they achieved in nodules. However, the dominant force shaping strain relative abundance was the nitrogen fixation phenotype, consistent with hosts being in control of this trait. Overall, my dissertation research provides evidence of limited genetic variation in host control traits of A. strigosus in a variety of experimental settings.

In the hundred-plus years since the invention of the Haber-Bosch process, industrial production of nitrogen (N) fertilizers has increased agronomic production and, consequently, the human population. Synthetic N fertilizers also promote the emission of reactive nitrogen forms, which can significantly damage the environment and human health. One alternative to N fertilizer is the use of beneficial nitrogen fixing soil bacteria (rhizobia) that form tumor-like structures called nodules on the roots of legumes. The most effective rhizobia for nitrogen fixation have been used to develop bioinoculants for field application, which have the potential to decrease synthetic N fertilizer use. But importantly, inoculant strains are often outcompeted by indigenous strains that fix little nitrogen. Because of this, bioinoculants are often only marginally effective, and plant growth suffers. To resolve this rhizobial competition problem and improve the N fixing benefits to plants, we must gain a better understanding of i) competitive interactions among rhizobia, ii) the epidemiological patterns of strains that dominate natural populations and spread among them, and iii) gain insight of N fixing efficiency and developmental variation of bacteroids inside the nodule. This dissertation research uses native Californian legumes in the genus Acmispon, which naturally host Bradyrhizobium strains. The three parallel aims are to: i) characterize rhizobial competition and model plant performance based on nodule occupancy of competing strains, ii) investigate the epidemiology of rhizobial genotypes across multiple hosts in a large transect (>800 km) of California, and iii) examine the nodule-bacteroid morphology and nitrogen fixation efficiency of rhizobia that vary in N fixation. The main findings from this dissertation reveals a competitive-dominance hierarchy of rhizobial strains which leads to reduced host-benefit, the epidemic spread of certain dominant Bradyrhizobium strains in natural populations, and ultrastructural variation in nodule bacteroid morphology during different stages of nodule development. This dissertation research holds significant importance in improving crop yields, reducing dependency on synthetic nitrogen fertilizers, and directly benefiting agricultural research aimed at boosting bioproduct yields from legumes without the use of synthetic nitrogen fertilizers.

Plant-associated microbes can provide substantial benefits to crops, improving yields and reducing necessary nutrient inputs. The benefits of plant-microbial symbioses are highly variable and rely on several contextual drivers, including host symbiosis traits, makeup of available soil microbes, and environmental conditions such as soil nutrient levels and field management practices. Legume species receive fixed nitrogen from symbiosis with rhizobia, soil bacteria which can infect roots. However, many legume species still require nitrogen fertilization, suggesting breakdowns in the forces directing this symbiosis. To examine these forces, we turned to cowpeas (Vigna unguiculata)—a legume crop grown across Africa, Asia, and the Americas—and tested the effects of host genotype, microbial community, and soil conditions on plant growth benefits. In Chapter 1, we performed a full factorial soil inoculation experiment on twenty wild and domesticated cowpea genotypes to test the effects of cowpea domestication, host genotype, and soil microbial community on cowpea growth benefits in a greenhouse setting. In Chapter 2, we tested whether cowpea-associating rhizobial communities are structured spatially or by host genotype in a field that had a long history of cowpea growth and was demonstrated to have a highly beneficial soil microbial community. In Chapter 3, we planted a different subset of nineteen cowpea genotypes in the same field and an adjacent field with no history of cowpea growth to test whether seed coat inoculants could improve growth in either field. We found strong evidence that cowpea host benefits are primarily shaped by soils, which drove most of the observed variation in both nodulation and host growth. We found that cowpea host genotype has very little effect on structuring variation in host benefits from inoculation, with no significant variation in nodule rhizobial communities across host genotype. UCR Field 11 is enriched with Bradyrhizobium and dominated by a small handful of strains, with soil that offers high benefits to plants under greenhouse conditions. However, neither soil nor strains from this field were effective inoculants under field conditions, and the commercial inoculant tested also did not perform under field conditions.

Legume-rhizobium symbiosis has become a model system for studying beneficial symbiosis between eukaryotic hosts and bacterial symbionts. Hosts are predicted to exert control over beneficial bacterial symbionts to prevent the evolution and spread of exploitative genotypes. Since nitrogen is exchanged in legume-rhizobium symbiosis and legumes can acquire nitrogen from soils, there can context dependent effects from nitrogen-enriched soils that can dramatically alter legume-rhizobium symbiosis. Anthropogenic nitrogen deposition has been enriching terrestrial ecosystems for more than a century and is accelerating in some regions. One prediction is that nitrogen enrichment can relax selection to maintain host control traits, leading to evolution and spread of exploitative rhizobia and/or an breakdown in symbiosis if legumes cease to gain benefit from rhizobial infection.

Lotus strigosus is an annual legume that experiences variable nitrogen deposition and mineral nitrogen contexts across California. To examine the potential for adaptation to mineral nitrogen enrichment in L. strigosus, inbred lines of L. strigosus from populations that have experienced either little nitrogen deposition or more than seven decades of intense nitrogen deposition were exposed to a simulated nitrogen deposition gradient and Bradyrhizobium that vary in growth benefit provided to L. strigosus. To examine the effects of mineral nitrogen saturation on a host control in a legume that has historically experienced nitrogen-poor soils, L. strigosus from a pristine site were exposed to growth saturation mineral nitrogen and infected with either combinations of Bradyrhizobium strains that vary in benefit to L. strigosus and individual strains that vary in benefit.

Lotus exhibited little evidence of adaptation to use mineral nitrogen more efficiently in response to simulated nitrogen deposition. Symbisois with Bradyrhizobium was reduced at very high fertilizer levels, but this was most likely because of nitrogen toxicity rather than a loss of benefit from symbiosis.

Host control traits in L. strigosus from a pristine site over pairs of Bradyrhizobium strains were resilient to growth saturating nitrogen, contrary to expectations if such traits are costly. When inoculated with single strains of Bradyrhizobium in different mineral nitrogen contexts, L. strigosus exhibited fine tuned investment in Bradyrhizobium to prevent exploitation.

Microbial communities indigenous to crop fields- particularly root-nodulating rhizobia – enhance growth and yield of legume crops, but the magnitude of microbial effects and their drivers are poorly understood. We sampled cultivated soils across a 1000km transect in California focusing on ten sites with recent cowpea cultivation. Grower practices and crop histories were analyzed, and soils were characterized for resident microbial communities and physiochemical features. Live soil inocula were created from each of the field soils to test the capacity to induce root nodulation and growth effects to six legume cultivars, including cowpea, soybean, and lima- selected because they associate with root-nodulating Bradyrhizobium. Soil inoculations influenced host growth response, which was largely positive to cowpeas on average, but varied depending on soil source and cowpea genotype. Soybean and lima showed little compatibility with root-nodulating rhizobia resident to cowpea fields. Soil source, plant genotype, and soil-plant interactions all impacted variation in plant growth, nodule counts and nodule mass. Crop history and plant genotype were primary factors influencing plant-soil interactions, suggesting a primacy of host plants in reshaping resident soil microbial communities. Links between soil physiochemical features, resident microbial communities, and plant performance offer novel routes to manipulate soils towards improved crop yields.

Understanding the drivers of variation in symbiont quality is a fundamental objective in the study of mutualisms. Eukaryotic hosts express control traits that can selectively favor beneficial symbionts over ineffective genotypes, but bacterial symbionts range widely in beneficial quality. Evolutionary instability in symbiotic function and/or context dependency in the expression of symbiotic traits are predicted to contribute to this variation.

The Acmispon-Bradyrhizobium mutualism is a model system for studying variation in symbiotic traits. Acmispon hosts sanction ineffective symbionts, yet Bradyrhizobium naturally vary in symbiotic function. However, the incidence, distribution, and evolution of symbiotic quality in Bradyrhizobium from natural habitats remains unclear. I investigated the evolution and spatial distribution of Bradyrhizobium symbiotic effectiveness across a metapopulation of A. strigosus hosts. Symbiotic quality was evolutionarily unstable, consistent with the repeated evolution of non-nitrogen fixing Bradyrhizobium, and suggests that the loss of nitrogen fixation may be a common process.

To examine if symbiotic ineffectiveness was expressed as a maladapted, context dependent outcome, I quantified fitness and fitness-effects of divers Bradyrhizobium isolates on sympatric and allopatric Acmispon hosts. Several isolates were found to be symbiotically ineffective and rhizobial fitness proxies uncovered evidence of rhizobial exploitation. This data suggests that host exploitation also maintains ineffective rhizobia, and thus overall variation in symbiont quality in natural populations.

Symbiotic quality can also vary due to biotic environment, and although rhizobia are best known for symbiotic function, the majority of rhizobia are non-symbiotic. To test if non-symbiotic conspecifics effect the Acmispon-Bradyrhizobium mutualism, I coinoculated hosts with mixtures of symbiotic and non-symbiotic Bradyrhizobium. In most cases, the presence of non-symbiotic Bradyrhizobium reduced host and symbiont performance and data suggests this occurs via competitive interactions at the root-soil interface.

Microbes can dramatically alter the fitness of host organisms, ranging in effect from mutualistic to antagonistic. Across this spectrum of fitness effects the microbial symbiont is predicted to optimize its own fitness benefits either through positive feedback (i.e., mutualism) or exploitation (i.e., antagonism). The legume-rhizobium symbiosis has emerged as a powerful system to study the control mechanisms of hosts and the corresponding subversion of control by symbionts. Rhizobial bacteria are housed in legume root nodules where they provide the costly services of nitrogen fixation in return for host derived carbon. However, variation in symbiont quality and lifestyle strategies can result in fitness conflict between host and symbiont. I used the symbiosis between Lotus japonicus and Mesorhizobium loti to investigate host control over symbionts across the mutualism-antagonism spectrum. For chapter 1 I investigated the effects of host control on symbiont fitness when symbionts vary in their fitness effects. I found that the L. japonicus host can adaptively alter the fitness of symbionts dependent on symbiotic nitrogen fixation. For chapter 2 I examined how varying host investment into symbiosis affects the fitness of hosts and symbionts. I found host control was maintained when host investment into symbiosis was altered. However, at high levels of host investment in symbiosis with a high-quality symbiont I uncovered evidence that host fitness is optimized, while symbiont fitness continues to increase. For chapter 3 I tested the effects of host control and varying levels of host investment on the evolution of symbiont services to hosts. After experimentally evolving rhizobial symbionts of mediocre quality in multiple experiments, the phenotypic data was most consistent with a shift towards antagonistic phenotypes. Together, these results highlight the intense conflict over resources that can lead to the destabilization or breakdown of the mutualism between legumes and rhizobia.