The evolutionary ecology of rhizobia: multiple facets of competition before, during, and after symbiosis with legumes.

Rhizobial bacteria have complex lifestyles that involve growth and survival in bulk soil, plant rhizospheres and rhizoplanes, legume infection threads, and mature and senescing legume nodules. In nature, rhizobia coexist and compete with many other rhizobial strains and species to form host associations. We review recent work defining competitive interactions across these environments. We highlight the use of sophisticated measurement tools and sequencing technologies to examine competition mechanisms in planta, and highlight environments (e.g. soil and senescing nodules) where we still know exceedingly little. We argue that moving toward an explicitly ecological framework (types of competition, resources, and genetic differentiation) will clarify the evolutionary ecology of these foundational organisms and open doors for engineering sustainable, beneficial associations with hosts.

Combining GWAS and population genomic analyses to characterize coevolution in a legume-rhizobia symbiosis.

The mutualism between legumes and rhizobia is clearly the product of past coevolution. However, the nature of ongoing evolution between these partners is less clear. To characterize the nature of recent coevolution between legumes and rhizobia, we used population genomic analysis to characterize selection on functionally annotated symbiosis genes as well as on symbiosis gene candidates identified through a two-species association analysis. For the association analysis, we inoculated each of 202 accessions of the legume host Medicago truncatula with a community of 88 Sinorhizobia (Ensifer) meliloti strains. Multistrain inoculation, which better reflects the ecological reality of rhizobial selection in nature than single-strain inoculation, allows strains to compete for nodulation opportunities and host resources and for hosts to preferentially form nodules and provide resources to some strains. We found extensive host by symbiont, that is, genotype-by-genotype, effects on rhizobial fitness and some annotated rhizobial genes bear signatures of recent positive selection. However, neither genes responsible for this variation nor annotated host symbiosis genes are enriched for signatures of either positive or balancing selection. This result suggests that stabilizing selection dominates selection acting on symbiotic traits and that variation in these traits is under mutation-selection balance. Consistent with the lack of positive selection acting on host genes, we found that among-host variation in growth was similar whether plants were grown with rhizobia or N-fertilizer, suggesting that the symbiosis may not be a major driver of variation in plant growth in multistrain contexts.

Host-Associated Rhizobial Fitness: Dependence on Nitrogen, Density, Community Complexity, and Legume Genotype.

The environmental context of the nitrogen-fixing mutualism between leguminous plants and rhizobial bacteria varies over space and time. Variation in resource availability, population density, and composition likely affect the ecology and evolution of rhizobia and their symbiotic interactions with hosts. We examined how host genotype, nitrogen addition, rhizobial density, and community complexity affected selection on 68 rhizobial strains in the Sinorhizobium meliloti-Medicago truncatula mutualism. As expected, host genotype had a substantial effect on the size, number, and strain composition of root nodules (the symbiotic organ). The understudied environmental variable of rhizobial density had a stronger effect on nodule strain frequency than the addition of low nitrogen levels. Higher inoculum density resulted in a nodule community that was less diverse and more beneficial but only in the context of the more selective host genotype. Higher density resulted in more diverse and less beneficial nodule communities with the less selective host. Density effects on strain composition deserve additional scrutiny as they can create feedback between ecological and evolutionary processes. Finally, we found that relative strain rankings were stable across increasing community complexity (2, 3, 8, or 68 strains). This unexpected result suggests that higher-order interactions between strains are rare in the context of nodule formation and development. Our work highlights the importance of examining mechanisms of density-dependent strain fitness and developing theoretical predictions that incorporate density dependence. Furthermore, our results have translational relevance for overcoming establishment barriers in bioinoculants and motivating breeding programs that maintain beneficial plant-microbe interactions across diverse agroecological contexts. IMPORTANCE Legume crops establish beneficial associations with rhizobial bacteria that perform biological nitrogen fixation, providing nitrogen to plants without the economic and greenhouse gas emission costs of chemical nitrogen inputs. Here, we examine the influence of three environmental factors that vary in agricultural fields on strain relative fitness in nodules. In addition to manipulating nitrogen, we also use two biotic variables that have rarely been examined: the rhizobial community's density and complexity. Taken together, our results suggest that (i) breeding legume varieties that select beneficial strains despite environmental variation is possible, (ii) changes in rhizobial population densities that occur routinely in agricultural fields could drive evolutionary changes in rhizobial populations, and (iii) the lack of higher-order interactions between strains will allow the high-throughput assessments of rhizobia winners and losers during plant interactions.

Phenotypic and genomic signatures of interspecies cooperation and conflict in naturally occurring isolates of a model plant symbiont.

Given the need to predict the outcomes of (co)evolution in host-associated microbiomes, whether microbial and host fitnesses tend to trade-off, generating conflict, remains a pressing question. Examining the relationships between host and microbe fitness proxies at both the phenotypic and genomic levels can illuminate the mechanisms underlying interspecies cooperation and conflict. We examined naturally occurring genetic variation in 191 strains of the model microbial symbiont Sinorhizobium meliloti, paired with each of two host Medicago truncatula genotypes in single- or multi-strain experiments to determine how multiple proxies of microbial and host fitness were related to one another and test key predictions about mutualism evolution at the genomic scale, while also addressing the challenge of measuring microbial fitness. We found little evidence for interspecies fitness conflict; loci tended to have concordant effects on both microbe and host fitnesses, even in environments with multiple co-occurring strains. Our results emphasize the importance of quantifying microbial relative fitness for understanding microbiome evolution and thus harnessing microbiomes to improve host fitness. Additionally, we find that mutualistic coevolution between hosts and microbes acts to maintain, rather than erode, genetic diversity, potentially explaining why variation in mutualism traits persists in nature.

Evolving together, evolving apart: measuring the fitness of rhizobial bacteria in and out of symbiosis with leguminous plants.

Most plant-microbe interactions are facultative, with microbes experiencing temporally and spatially variable selection. How this variation affects microbial evolution is poorly understood. Given its tractability and ecological and agricultural importance, the legume-rhizobia nitrogen-fixing symbiosis is a powerful model for identifying traits and genes underlying bacterial fitness. New technologies allow high-throughput measurement of the relative fitness of bacterial mutants, strains and species in mixed inocula in the host, rhizosphere and soil environments. I consider how host genetic variation (G × G), other environmental factors (G × E), and host life-cycle variation may contribute to the maintenance of genetic variation and adaptive trajectories of rhizobia - and, potentially, other facultative symbionts. Lastly, I place these findings in the context of developing beneficial inoculants in a changing climate.

A Select and Resequence Approach Reveals Strain-Specific Effects of Medicago Nodule-Specific PLAT-Domain Genes.

Genetic studies of legume symbiosis with nitrogen-fixing rhizobial bacteria have traditionally focused on nodule and nitrogen-fixation phenotypes when hosts are inoculated with a single rhizobial strain. These approaches overlook the potential effect of host genes on rhizobial fitness (i.e. how many rhizobia are released from host nodules) and strain-specific effects of host genes (i.e. genome × genome interactions). Using Medicago truncatula mutants in the recently described nodule-specific PLAT domain (NPD) gene family, we show how inoculating plants with a mixed inoculum of 68 rhizobial strains (Ensifer meliloti) via a select-and-resequence approach can be used to efficiently assay host mutants for strain-specific effects of late-acting host genes on interacting bacteria. The deletion of a single NPD gene (npd2) or all five members of the NPD gene family (npd1-5) differentially altered the frequency of rhizobial strains in nodules even though npd2 mutants had no visible nodule morphology or N-fixation phenotype. Also, npd1-5 nodules were less diverse and had larger populations of colony-forming rhizobia despite their smaller size. Lastly, NPD mutations disrupt a positive correlation between strain fitness and wild-type host biomass. These changes indicate that the effects of NPD proteins are strain dependent and that NPD family members are not redundant with regard to their effects on rhizobial strains. Association analyses of the rhizobial strains in the mixed inoculation indicate that rhizobial genes involved in chromosome segregation, cell division, GABA metabolism, efflux systems, and stress tolerance play an important role in the strain-specific effects of NPD genes.

Legacy of prior host and soil selection on rhizobial fitness in planta.

Measuring selection acting on microbial populations in natural or even seminatural environments is challenging because many microbial populations experience variable selection. The majority of rhizobial bacteria are found in the soil. However, they also live symbiotically inside nodules of legume hosts and each nodule can release thousands of daughter cells back into the soil. We tested how past selection (i.e., legacies) by two plant genotypes and by the soil alone affected selection and genetic diversity within a population of 101 strains of Ensifer meliloti. We also identified allelic variants most strongly associated with soil- and host-dependent fitness. In addition to imposing direct selection on rhizobia populations, soil and host environments had lasting effects across host generations. Host presence and genotype during the legacy period explained 22% and 12% of the variance in the strain composition of nodule communities in the second cohort, respectively. Although strains with high host fitness in the legacy cohort tended to be enriched in the second cohort, the diversity of the strain community was greater when the second cohort was preceded by host rather than soil legacies. Our results indicate the potential importance of soil selection driving the evolution of these plant-associated microbes.

Filling in the gaps in survival analysis: using field data to infer plant responses to environmental stressors.

Elucidating how organismal survival depends on the environment is a core component of ecological and evolutionary research. To reconcile high-frequency covariates with lower-frequency demographic censuses, many statistical tools involve aggregating environmental conditions over long periods, potentially obscuring the importance of fluctuating conditions in driving mortality. Here, we introduce a flexible model designed to infer how survival probabilities depend on changing environmental covariates. Specifically, the model (1) quantifies effects of environmental covariates at a higher frequency than the census intervals, and (2) allows partitioning of environmental drivers of individual survival into acute (short-term) and chronic (accumulated) effects. By applying our method to a long-term observational data set of eight annual plant species, we show we can accurately infer daily survival probabilities as temperature and moisture levels change. Next, we show that a species' water use efficiency, known to mediate annual plant population dynamics, is positively correlated with the importance of "chronic stress" inferred by the model. This suggests that model parameters can reflect underlying physiological mechanisms. This method is also applicable to other binary responses (hatching, phenology) or systems (insects, nestlings). Once known, environmental sensitivities can be used for ecological forecasting even when the frequency or variability of environments are changing.

Genome-Wide Association Analyses in the Model Rhizobium Ensifer meliloti.

Genome-wide association studies (GWAS) can identify genetic variants responsible for naturally occurring and quantitative phenotypic variation. Association studies therefore provide a powerful complement to approaches that rely on de novo mutations for characterizing gene function. Although bacteria should be amenable to GWAS, few GWAS have been conducted on bacteria, and the extent to which nonindependence among genomic variants (e.g., linkage disequilibrium [LD]) and the genetic architecture of phenotypic traits will affect GWAS performance is unclear. We apply association analyses to identify candidate genes underlying variation in 20 biochemical, growth, and symbiotic phenotypes among 153 strains of Ensifer meliloti For 11 traits, we find genotype-phenotype associations that are stronger than expected by chance, with the candidates in relatively small linkage groups, indicating that LD does not preclude resolving association candidates to relatively small genomic regions. The significant candidates show an enrichment for nucleotide polymorphisms (SNPs) over gene presence-absence variation (PAV), and for five traits, candidates are enriched in large linkage groups, a possible signature of epistasis. Many of the variants most strongly associated with symbiosis phenotypes were in genes previously identified as being involved in nitrogen fixation or nodulation. For other traits, apparently strong associations were not stronger than the range of associations detected in permuted data. In sum, our data show that GWAS in bacteria may be a powerful tool for characterizing genetic architecture and identifying genes responsible for phenotypic variation. However, careful evaluation of candidates is necessary to avoid false signals of association.IMPORTANCE Genome-wide association analyses are a powerful approach for identifying gene function. These analyses are becoming commonplace in studies of humans, domesticated animals, and crop plants but have rarely been conducted in bacteria. We applied association analyses to 20 traits measured in Ensifer meliloti, an agriculturally and ecologically important bacterium because it fixes nitrogen when in symbiosis with leguminous plants. We identified candidate alleles and gene presence-absence variants underlying variation in symbiosis traits, antibiotic resistance, and use of various carbon sources; some of these candidates are in genes previously known to affect these traits whereas others were in genes that have not been well characterized. Our results point to the potential power of association analyses in bacteria, but also to the need to carefully evaluate the potential for false associations.

Select and resequence reveals relative fitness of bacteria in symbiotic and free-living environments.

Assays to accurately estimate relative fitness of bacteria growing in multistrain communities can advance our understanding of how selection shapes diversity within a lineage. Here, we present a variant of the "evolve and resequence" approach both to estimate relative fitness and to identify genetic variants responsible for fitness variation of symbiotic bacteria in free-living and host environments. We demonstrate the utility of this approach by characterizing selection by two plant hosts and in two free-living environments (sterilized soil and liquid media) acting on synthetic communities of the facultatively symbiotic bacterium Ensifer meliloti We find (i) selection that hosts exert on rhizobial communities depends on competition among strains, (ii) selection is stronger inside hosts than in either free-living environment, and (iii) a positive host-dependent relationship between relative strain fitness in multistrain communities and host benefits provided by strains in single-strain experiments. The greatest changes in allele frequencies in response to plant hosts are in genes associated with motility, regulation of nitrogen fixation, and host/rhizobia signaling. The approach we present provides a powerful complement to experimental evolution and forward genetic screens for characterizing selection in bacterial populations, identifying gene function, and surveying the functional importance of naturally occurring genomic variation.

Canalization of Seasonal Phenology in the Presence of Developmental Variation: Seed Dormancy Cycling in an Annual Weed.

Variation in the developmental timing in one life stage may ramify within and across generations to disrupt optimal phenology of other life stages. By focusing on a common mechanism of developmental arrest in plants-seed dormancy-we investigated how variation in flowering time influenced seed germination behavior and identified potential processes that can lead to canalized germination behavior despite variation in reproductive timing. We quantified effects of reproductive timing on dormancy cycling by experimentally manipulating the temperature during seed maturation and the seasonal timing of seed dispersal/burial, and by assessing temperature-dependent germination of un-earthed seeds over a seasonal cycle. We found that reproductive timing, via both seed-maturation temperature and the timing of dispersal, strongly influenced germination behavior in the weeks immediately following seed burial. However, buried seeds subsequently canalized their germination behavior, after losing primary dormancy and experiencing natural temperature and moisture conditions in the field. After the complete loss of primary dormancy, germination behavior was similar across seed-maturation and dispersal treatments, even when secondary dormancy was induced. Maternal effects themselves may contribute to the canalization of germination: first, by inducing stronger dormancy in autumn-matured seeds, and second by modifying the responses of those seeds to their ambient environment. Genotypes differed in dormancy cycling, with functional alleles of known dormancy genes necessary for the suppression of germination at warm temperatures in autumn through spring across multiple years. Loss of function of dormancy genes abolished almost all dormancy cycling. In summary, effects of reproductive phenology on dormancy cycling of buried seeds were apparent only as long as seeds retained primary dormancy, and a combination of genetically imposed seed dormancy, maternally induced seed dormancy, and secondary dormancy can mitigate variation in germination behavior imposed by variation in reproductive phenology.

Transcriptomic basis of genome by genome variation in a legume-rhizobia mutualism.

In the legume-rhizobia mutualism, the benefit each partner derives from the other depends on the genetic identity of both host and rhizobial symbiont. To gain insight into the extent of genome × genome interactions on hosts at the molecular level and to identify potential mechanisms responsible for the variation, we examined host gene expression within nodules (the plant organ where the symbiosis occurs) of four genotypes of Medicago truncatula grown with either Ensifer meliloti or E. medicae symbionts. These host × symbiont combinations show significant variation in nodule and biomass phenotypes. Likewise, combinations differ in their transcriptomes: host, symbiont and host × symbiont affected the expression of 70%, 27% and 21%, respectively, of the approximately 27,000 host genes expressed in nodules. Genes with the highest levels of expression often varied between hosts and/or symbiont strain and include leghemoglobins that modulate oxygen availability and hundreds of Nodule Cysteine-Rich (NCR) peptides involved in symbiont differentiation and viability in nodules. Genes with host × symbiont-dependent expression were enriched for functions related to resource exchange between partners (sulphate/iron/amino acid transport and dicarboxylate/amino acid synthesis). These enrichments suggest mechanisms for host control of the currencies of the mutualism. The transcriptome of M. truncatula accession HM101 (A17), the reference genome used for most molecular research, was less affected by symbiont identity than the other hosts. These findings underscore the importance of assessing the molecular basis of variation in ecologically important traits, particularly those involved in biotic interactions, in multiple genetic contexts.

Natural variation for gene expression responses to abiotic stress in maize.

Plants respond to abiotic stress through a variety of physiological, biochemical, and transcriptional mechanisms. Many genes exhibit altered levels of expression in response to abiotic stress, which requires concerted action of both cis- and trans-regulatory features. In order to study the variability in transcriptome response to abiotic stress, RNA sequencing was performed using 14-day-old maize seedlings of inbreds B73, Mo17, Oh43, PH207 and B37 under control, cold and heat conditions. Large numbers of genes that responded differentially to stress between parental inbred lines were identified. RNA sequencing was also performed on similar tissues of the F1 hybrids produced by crossing B73 and each of the three other inbred lines. By evaluating allele-specific transcript abundance in the F1 hybrids, we were able to measure the abundance of cis- and trans-regulatory variation between genotypes for both steady-state and stress-responsive expression differences. Although examples of trans-regulatory variation were observed, cis-regulatory variation was more common for both steady-state and stress-responsive expression differences. The genes with cis-allelic variation for response to cold or heat stress provided an opportunity to study the basis for regulatory diversity.

A Guide to Genome-Wide Association Mapping in Plants.

Genome-wide association studies (GWAS) have developed into a valuable approach for identifying the genetic basis of phenotypic variation. In this article, we provide an overview of the design, analysis, and interpretation of GWAS. First, we present results from simulations that explore key elements of experimental design as well as considerations for collecting the relevant genomic and phenotypic data. Next, we outline current statistical methods and tools used for GWA analyses and discuss the inclusion of covariates to account for population structure and the interpretation of results. Given that many false positive associations will occur in any GWA analysis, we highlight strategies for prioritizing GWA candidates for further statistical and empirical validation. While focused on plants, the material we cover is also applicable to other systems. © 2017 by John Wiley & Sons, Inc.

A cline in seed dormancy helps conserve the environment experienced during reproduction across the range of Arabidopsis thaliana.

PREMISE OF THE STUDY: Understanding the factors shaping range limits is critical given current changes in climate as well as human-mediated introduction of species into novel environments. Phenological responses to climate influence range limits by allowing plants to avoid conditions that decrease population growth rates. Studying these processes is a challenge due to the joint contributions of both genetic and environmental variation to phenology. METHODS: Using a previously developed model that predicts phenology of three dormancy "genotypes" in four locations spanning the European range of Arabidopsis thaliana, we examined how variation in seed dormancy influences the environmental conditions experienced by reproductive individuals and how those conditions influence reproductive potential. We calculated two metrics: temperature experienced during reproduction and the length of thermal window available for reproduction. KEY RESULTS: Seed dormancy levels determine whether a spring-flowering life cycle is expressed and thus determine the reproductive environment. A genetic cline in seed dormancy across the range reduces differences in reproductive environment and increases the thermal opportunity for reproduction before conditions become unfavorable for survival. Counter-intuitively, these putatively local genotypes are predicted to reproduce in slightly cooler conditions in the south than in the north, suggesting that maternal environmental effects on average could induce deeper dormancy in southern seeds reinforcing the observed genetic cline. However, within a location, we found large individual level differences. CONCLUSIONS: Phenological adjustments of early life stages can contribute to the maintenance of consistent reproductive environments experienced by individual plants across ranges despite variable environmental conditions over time and space.

Fluctuating, warm temperatures decrease the effect of a key floral repressor on flowering time in Arabidopsis thaliana.

The genetic basis of growth and development is often studied in constant laboratory environments; however, the environmental conditions that organisms experience in nature are often much more dynamic. We examined how daily temperature fluctuations, average temperature, day length and vernalization influence the flowering time of 59 genotypes of Arabidopsis thaliana with allelic perturbations known to affect flowering time. For a subset of genotypes, we also assessed treatment effects on morphology and growth. We identified 17 genotypes, many of which have high levels of the floral repressor FLOWERING LOCUS C (FLC), that bolted dramatically earlier in fluctuating - as opposed to constant - warm temperatures (mean = 22°C). This acceleration was not caused by transient VERNALIZATION INSENSITIVE 3-mediated vernalization, differential growth rates or exposure to high temperatures, and was not apparent when the average temperature was cool (mean = 12°C). Further, in constant temperatures, contrary to physiological expectations, these genotypes flowered more rapidly in cool than in warm environments. Fluctuating temperatures often reversed these responses, restoring faster bolting in warm conditions. Independently of bolting time, warm fluctuating temperature profiles also caused morphological changes associated with shade avoidance or 'high-temperature' phenotypes. Our results suggest that previous studies have overestimated the effect of the floral repressor FLC on flowering time by using constant temperature laboratory conditions.

Multiple paths to similar germination behavior in Arabidopsis thaliana.

Germination timing influences plant fitness, and its sensitivity to temperature may cause it to change as climate shifts. These changes are likely to be complex because temperatures that occur during seed maturation and temperatures that occur post-dispersal interact to define germination timing. We used the model organism Arabidopsis thaliana to determine how flowering time (which defines seed-maturation temperature) and post-dispersal temperature influence germination and the expression of genetic variation for germination. Germination responses to temperature (germination envelopes) changed as seeds aged, or after-ripened, and these germination trajectories depended on seed-maturation temperature and genotype. Different combinations of genotype, seed-maturation temperature, and after-ripening produced similar germination envelopes. Likewise, different genotypes and seed-maturation temperatures combined to produce similar germination trajectories. Differences between genotypes were most likely to be observed at high and low germination temperatures. The germination behavior of some genotypes responds weakly to maternal temperature but others are highly plastic. We hypothesize that weak dormancy induction could synchronize germination of seeds dispersed at different times. By contrast, we hypothesize that strongly responsive genotypes may spread offspring germination over several possible germination windows. Considering germination responses to temperature is important for predicting phenology expression and evolution in future climates.

Modeling the influence of genetic and environmental variation on the expression of plant life cycles across landscapes.

Organisms develop through multiple life stages that differ in environmental tolerances. The seasonal timing, or phenology, of life-stage transitions determines the environmental conditions to which each life stage is exposed and the length of time required to complete a generation. Both environmental and genetic factors contribute to phenological variation, yet predicting their combined effect on life cycles across a geographic range remains a challenge. We linked submodels of the plasticity of individual life stages to create an integrated model that predicts life-cycle phenology in complex environments. We parameterized the model for Arabidopsis thaliana and simulated life cycles in four locations. We compared multiple "genotypes" by varying two parameters associated with natural genetic variation in phenology: seed dormancy and floral repression. The model predicted variation in life cycles across locations that qualitatively matches observed natural phenology. Seed dormancy had larger effects on life-cycle length than floral repression, and results suggest that a genetic cline in dormancy maintains a life-cycle length of 1 year across the geographic range of this species. By integrating across life stages, this approach demonstrates how genetic variation in one transition can influence subsequent transitions and the geographic distribution of life cycles more generally.

Applying developmental threshold models to evolutionary ecology.

Process-based models of development predict developmental rates and phenology as a function of physiological responses to multiple dynamic environmental factors. These models can be adapted to analyze diverse processes in evolutionary ecology. By linking models across life stages, they can predict life cycles and generation times. By incorporating fitness, they can identify environmental and physiological factors that limit species distributions. By incorporating population variance, they can investigate mechanisms of intraspecific variation or synchronization. By incorporating genetics, they can predict genotype-specific phenology under diverse climatic scenarios and examine causes and consequences of pleiotropy across life stages. With further development, they have the potential to predict genotype-specific ranges and identify key genes involved in determining phenology and fitness in variable and changing environments.

The community-level effect of light on germination timing in relation to seed mass: a source of regeneration niche differentiation.

Within a community, species may germinate at different times so as to mitigate competition and to take advantage of different aspects of the seasonal environment (temporal niche differentiation). We illustrated a hypothesis of the combined effects of abiotic and biotic competitive factors on germination timing and the subsequent upscale effects on community assembly. We estimated the germination timing (GT) for 476 angiosperm species of the eastern Tibetan Plateau grasslands under two light treatments in the field: high (i.e. natural) light and low light. We also measured the shift in germination timing (SGT) across treatments for all species. Furthermore, we used phylogenetic comparative methods to test if GT and SGT were associated with seed mass, an important factor in competitive interactions. We found a significant positive correlation between GT and seed mass in both light treatments. Additionally, small seeds (early germinating seeds) tended to germinate later and large seeds (late germinating seeds) tended to germinate earlier under low light vs high light conditions. Low light availability can reduce temporal niche differentiation by increasing the overlap in germination time between small and large seeds. In turn, reduced temporal niche differentiation may increase competition in the process of community assembly.