Hosts select symbionts for greater mutual benefit, an evolutionary experiment shows

The roots of a barrel medick plant, with pink nodules housing rhizobia. (Wikimedia Commons: Ninjatacoshell)

Who’s in charge of a symbiotic mutualism? You might think the host organism, whose body is the venue for an exchange of nutrients or services with a microbial symbiont, is running the show, able to evict or punish symbionts that don’t play nice. However, there are many examples of hosts making do with symbionts that aren’t particularly good partners, and some evolutionary theory has suggested that competing symbionts can gain the upper hand. Results from an evolutionary experiment recently reported in the journal Science lend support to the host-in-the-driver-seat view, though — bacterial symbionts selected by five generations of hosts evolved to be better mutualists.

The core of the paper, led by Rebecca Batstone, is a careful greenhouse experiment with a model for a well studied symbiotic mutualism between plants in the legume family and nitrogen fixing bacteria. The details vary among species, but generally the legume host allows free-living soil bacteria to infect its roots, forming specialized nodules of root tissue to house the infection, where the bacteria are fed with sugar while they “fix” atmospheric nitrogen, converting it to a chemical form the host plant can use as fertilizer. Eventually (often when the host plant dies) the root nodules senesce, releasing a sugar-fed population of bacteria into the soil, available to help other hosts. If hosts preferentially interact with more productive bacteria, the rewards of the symbiosis can potentially create selection favoring bacteria that fix more nitrogen — making the free-living population in the soil better mutualists, on average.

Batstone and her coauthors planted Medicago truncatula, or barrel medick, a small greenhouse-friendly legume, with two strains of its preferred nitrogen-fixing symbiont, Ensifer meliloti, that differed in their nitrogen fixation capacity. Barrel medic is self-pollinating, and there’s a library of inbred, homozygous lines of the plant that produce seeds that are essentially clones of their parent. This let Batstone et al. evolve their bacterial inoculations with multiple generations of host plants having the same genotype — they selected five genotypes that varied in their ability to selectively interact with more beneficial bacteria. When a plant had completed fruit production, they gently pulled it out of the soil to measure it and sample its nodules, then put the roots back in the soil so the remaining nodules could senesce and release their bacterial tenants. Then, finally, they planted a new seed of the same inbred line in the soil to start a new “generation.” This let the bacteria in the soil evolve while interacting with a succession of genetically individual plants. In greenhouse conditions barrel medick can go from seed to fruit in five or six weeks; the team got through five host plant generations in about a year.

The coauthors used quantitative PCR to track the frequency of the original two strains of bacteria in the soil substrate and within sampled root nodules, over time. They also tested how well “derived” bacterial strains isolated from the final generation of hosts’ nodules performed compared to samples of the “ancestral” strains used to start the experiment, taking measurements from host plants inoculated with a single ancestral or derived strain of bacteria. And, finally, they sequenced the genomes of derived isolate strains and the ancestral strains, and tested for associations between single-nucleotide polymorphism (SNP) and indel (insertion or deletion) genotypes and the fitness outcomes of the mutualism for both host plants and bacterial.

Frequency of Em1022 (the bacterial strain that fixed nitrogen more effectively) in the soil over five host plant generations, as determined by qPCR (Batstone et al. 2020, Fig 1.)

The results very clearly show the bacteria evolving to be more beneficial to the host — and to derive more benefit from the host. The frequency of the more beneficial bacterial strain in the soil rapidly increased, getting to fixation within three generations on some of the host lines Batstone et al. used. Evolution with four out of five of the host lines led to increased benefits of the symbiosis, for both the host plant and the rhizobia. Derived bacteria generally provided more growth benefit to the host plant line than the ancestral strain, and plants hosting derived bacteria produced more nodules than they did for the ancestor.

The genomic data reflected this alignment of host and symbiont interests, and demonstrated that the improvement in mutualistic outcomes was fueled in part by new mutations over the course of the experiment, not just variants present in the original inoculations. SNPs and indels that were significantly associated with increased nodule formation (greater benefit to the bacteria) also tended to be significantly associated with increased host plant biomass (greater benefit to the host). There was only one variant showing significant association to increased bacterial benefit and reduced host plant benefit — and it was an ancestral variant.

Strength of associations (beta) between bacterial SNP or indel genotypes and either the number of nodules their host plants formed (x axis) or the shoot biomass of the host. Each point is an individual SNP or indel; larger points have associations above a statistical significance threshold, with square points indicating variants present in the ancestral bacteria (standing variation) and round points indicating variants that appeared as new mutations in the derived bacteria. Panels present results for all plant lines (top left) or each of the five individual lines. Gray areas indicate a “cheating zone” in which points indicate increased benefit to the bacteria but reduced benefit to the host plant. (Batstone et al. 2020, Fig 3A.)

This experiment isn’t really capturing host-symbiont coevolution; the design means that the host was prevented from evolving in response to the symbiont. (And the logistics of an experiment that would let the hosts meaningfully evolve would be challenging, to say the least.) That’s significant because it means that the evolution of more mutualistic bacteria wasn’t driven by the fitness benefits the hosts received — one mechanism that has been proposed that could make mutualists more efficient over evolutionary time. Instead, it looks like the conditions of interaction with an individual host plant favor more beneficial strains, possibly because plants allocate more resources to nodule formation when existing nodules are productive. (Legumes have also been shown to allocate different resources to individual nodules based on bacterial performance, but the data here can’t assess that.) That’s consistent with results from the best precedent for this experiment, in which a much more diverse set of bacterial strains were selected by just a single generation of barrel medick plants — and the bacteria’s success at nodule formation was correlated with the benefit they provided the plants.

Also, as Batstone et al. point out, it’s significant that their largest-scale result, the rapid increase in frequency of the more beneficial starting strain, is so consistent across the five plant lines they tested. It means that even less-choosy plant lines were able to select for more beneficial bacteria. If that’s broadly true, it may mean that the secret to the success of the legume-rhizobium mutualism is plant populations cultivating their own locally adapted community of soil bacteria — domesticating the local pool of symbionts, in a manner of speaking.


Akçay E, and EL Simms. 2011. Negotiation, sanctions, and context dependency in the legume-rhizobium mutualism. The American Naturalist 178, no. 1 (2011): 1-14. doi: 10.1086/659997.

Axelrod R. and WD Hamilton. 1981. The evolution of cooperation. Science211(4489): 1390-1396. doi: 10.1126/science.7466396.

Batstone RT, AM O’Brien, TL Harrison, and ME Frederickson. 2020. Experimental evolution makes microbes more cooperative with their local host genotype. Science 37: 476-478. doi: 10.1126/science.abb7222.

Burghardt LT, B Epstein, J Guhlin, MS Nelson, MR Taylor, ND Young, MJ Sadowsky, and P Tiffin. 2018. Select and resequence reveals relative fitness of bacteria in symbiotic and free-living environments. Proceedings of the National Academy of Sciences115(10): 2425-2430. doi: 10.1073/pnas.1714246115.

About Jeremy Yoder

Jeremy B. Yoder is an Assistant Professor of Biology at California State University Northridge, studying the evolution and coevolution of interacting species, especially mutualists. He is a collaborator with the Joshua Tree Genome Project and the Queer in STEM study of LGBTQ experiences in scientific careers. He has written for the website of Scientific American, the LA Review of Books, the Chronicle of Higher Education, The Awl, and Slate.
This entry was posted in adaptation, Coevolution, evolution, microbiology and tagged , , . Bookmark the permalink.