The coevolutionary genetics of plant–microbe interactions


Joint annual meetings of the Society for the Study of Evolution (SSE), Society of Systematic Biologists (SSB), and American Society of Naturalists (ASN), Minneapolis, MN, USA, June 2008

The tightly coevolved interactions between plants and their microbial symbionts play key roles in the functioning of terrestrial communities and ecosystems. Plant–pathogen interactions, for example, influence the assembly of plant communities (Augspurger, 1988), and the evolution of mycorrhizal interactions has been linked with plant adaptation to life on land and subsequent ecological diversification (Pirozynski & Malloch, 1975; Selosse & Le Tacon, 1998). Evolution 2008, the joint annual meetings of the Society for the Study of Evolution (SSE), the Society of Systematic Biologists (SSB), and the American Society of Naturalists (ASN), took place recently at the University of Minnesota. Coevolution was one theme of this year's meeting, and many talks focused on the interactions between plants and their microbial symbionts. The authors highlighted a wide range of coevolutionary research, including the role of intergenomic epistasis in dynamic coevolution, community effects on the coevolutionary process, and the integration of genetic information into evolutionary research.

‘... stepping beyond the pairwise interaction and incorporating a community perspective can illuminate the forces shaping genetic variation and coevolutionary stability of a mutualism.’

Intergenomic epistasis and partner fitness

Genetic variation for the benefits exchanged in plant–microbe interactions has long been appreciated, for example for nitrogen fixation (Fred et al., 1932) and pathogen resistance (Flor, 1942). Of course, in microbial populations and communities, much about the very nature of genotypes, as well as how we define the individual, remains controversial (Gould, 1992). For example, Caroline Angelard (Université de Lausanne, Switzerland) presented evidence that different nuclei within even a single individual of an arbuscular mycorrhizal fungus (AMF) can differ dramatically in their effects on host plant fitness. Even in simpler genetic systems, however, the interactions between plant and symbiotic microbial genomes (i.e. intergenomic epistasis, or genotype (G) × genotype (G) interactions) can have important effects on the rate or even direction of coevolutionary selection (Wade, 2007). Lucie Salvaudon (Université Paris Sud, France) showed that the transmission success of the powdery mildew pathogen (Hyaloperonospora parasitica), and the effects on its Arabidopsis thaliana host plants, were dependent on the interaction between pathogen and host genotypes (G × G interactions). Similarly, G × G interactions also influence the fitness outcomes of interactions between the legume Medicago truncatula and its nitrogen (N2)-fixing mutualist Sinorhizobium meliloti (Katy Heath, University of Toronto, Canada) and can even shift the symbiosis from mutualism to parasitism.

Additionally, both talks (one focused on a parasitism and the other on a mutualism) highlighted the point that variation in overall ‘compatibility’ between partners may result in the lack of a negative correlation between host and symbiont fitnesses. Instead of consistent trade-offs between host and symbiont fitnesses, both studies found that increases in symbiont fitness do not necessarily result in decreased host fitness. Additionally, using a novel experimental coevolution approach, Maren Friesen (University of California, Davis, USA) presented data showing that N2-fixing rhizobium strains evolved increased competition for symbiosis in the laboratory, but that these more competitive strains had no detrimental effect on host plants – again suggesting that increased symbiont fitness does not necessarily come at a cost to host fitness. Because the trade-off between host and symbiont fitnesses is a fundamental assumption of many coevolutionary models, violations of this assumption have the potential to alter predictions about such coevolved traits as pathogen virulence and mutualism benefits.

Plant–microbe interactions in a community context

Species interactions are typically studied in a pairwise context (Stanton, 2003). However, the community context in which these interactions evolve can have profound effects on the ecological (e.g. Thompson & Cunningham, 2002) and evolutionary (e.g. Benkman, 1999) outcomes of species interactions. Natural legume communities at Bodega Marine Reserve in California include Lotus and Lupinus legume species that share rhizobium symbionts (Bradyrhizobium species). Martine Ehinger (University of California Berkeley, USA) presented evidence of strong differentiation at an N2-fixation locus (nifD) between rhizobia isolated from the two hosts, despite ample horizontal transfer across the rest of the genome. This result implies that selection imposed by sympatric host species may maintain allelic diversity within a single rhizobium population. Ellen Simms (University of California Berkeley, USA) also presented work showing that many rhizobia in the population are nonsymbiotic (at least with the hosts tested) but instead inhabit ‘rhizofilms’ on plant roots. Although these strains have an ecology and phylogeny that are distinct from those of their symbiotic brethren, they nevertheless can have important effects on the coevolution of the symbiosis by serving as a source of ‘cheater’ rhizobium strains that gain high fitness from the interaction while giving little or no benefits to their hosts. For example, one such Lotus-associated cheater, which forms nodules but fixes little nitrogen, appears to be the result of horizontal transfer of the nifD gene from Lupinus-associated rhizobia into a strain most closely related to rhizofilm rhizobia. These results have important implications for our understanding of mutualism stability (reviewed in Douglas, 2007), because rhizobium cheaters did not arise from within a symbiotic clade, but instead arose from a distinct clade of rhizobia that utilize a nonsymbiotic strategy. Moreover, this work provides an example of how stepping beyond the pairwise interaction and incorporating a community perspective can illuminate the forces shaping genetic variation and coevolutionary stability of a mutualism.

Coevolutionary genomics

Ecological genomics aims to integrate the fast-accumulating molecular genetic information into a meaningful ecological and evolutionary context. While most investigations to date have focused on the evolution of a single species in response to abiotic conditions (reviewed in Stinchcombe & Hoekstra, 2008), an advantage of model species interactions is the potential to understand how the coevolutionary process has affected specific loci in the genome of one or both partners. The flax–flax rust interaction, for example, enjoys a long history of study (Flor, 1942, Burdon & Thrall, 2000). Luke Barrett (University of Chicago, USA) presented evidence suggesting that genes contributing to virulence specificity (pathogen effectors AvrP123 and AvrP4) in the flax rust Melampsora lini have diversified in response to geographic variation in resistance in host flax (Linum marginale) from natural Australian populations.

Coevolution with multiple host species can also lead to diversification in pathogen avr sequences.  Joel Kniskern (University of Chicago, USA) showed that allelic variants of the effector gene AvrPphB in Pseudomonas syringae confer differential performance with Phaseolus vulgaris versus A. thaliana hosts – suggesting that these sequences have diversified in response to distantly related hosts. These studies contribute an understanding of how sequence variation at specific candidate loci contributes to reciprocal adaptation among species, and how that variation varies among host populations and species.


In this year's ASN presidential address, ‘The Coevolving Web of Life,’ John Thompson (University of California Santa Cruz, USA) highlighted the importance of geographic variation (or ‘mosaics’) in coevolutionary selection. In particular, G × G × environment (E) interactions, in which the fitness outcomes of particular host–symbiont combinations depend on the abiotic or biotic environment, define a selection mosaic (i.e. the geographic patterns of selection across a species range). Scott Nuismer (University of Idaho, USA) presented a methodological approach for evaluating the contributions of G × G, G × E, and G × G × E effects to local adaptation (Nuismer & Gandon, 2008). Although challenging for many plant–microbe interactions, this technique has the potential to shed light on whether reciprocal genetic changes between two coevolving species, versus the adaptation of each species to the local environment, generate the majority of geographic variation observed in species interactions.

Vijay Panjeti (University of Virginia, USA) presented a model suggesting that the effects of plant pathogens on the demography and genetics at the interior of a species’ range can actually facilitate adaptation to marginal habitats at the edge of the range. Few empiricists have addressed the topic of whether biotic versus abiotic factors determine distributional range limits. For example, do plant–microbe interactions facilitate the colonization of novel habitats (e.g. in invasive species) or serve to constrain range expansion, and are antagonistic or mutualistic interactions most important to range limit evolution? The integration of coevolutionary studies with theory on species range limits is likely to be a promising direction for future research.

We still know little about which specific loci contribute to symbiotic variation in nature, but gene expression analysis may prove useful for identifying candidate loci involved in coevolutionary interactions (Ranz & Machado, 2006). For example, Alexandre Colard (Université de Lausanne, Switzerland) used a model mycorrhizal mutualism to show that different genotypes of the AMF Glomus intraradices differentially affect plant gene expression at symbiosis-related loci. Indeed, one way to uncover the genetic targets of coevolutionary selection would be to screen plant and symbiont genomes for genes that differ in their expression in response to partner genotypes. Tools are readily available for many model organisms (e.g. Medicago truncatula, A. thaliana and Pseudomonas spp.), yet such work has not been attempted to date. Moreover, as tools such as Solexa (Warren et al., 2007) and 454 (Margulies et al., 2005) sequencing and gene expression analysis become increasingly available, genome-wide screening techniques will become a reality for nonmodel interactions. The majority of plant–microbe talks at this year's meeting focused on a handful of model organisms; however, future coevolutionary research will also benefit from the incorporation of a broader cross-section of plant and microbial diversity.


Thanks to the above-mentioned authors, as well as many others whose work I could not include, for their stimulating talks and thoughtful conversations during the meeting. Additional thanks to many authors listed for providing information via personal communication after the meeting, and to Jennifer Lau for comments on a draft manuscript.