Legume-imposed selection for more-efficient symbiotic rhizobia

Agriculture could reduce the economic and environmental costs of nitrogen (N) fertilizers by relying more on atmospheric N taken up (fixed) by rhizobia bacteria in legume root nodules. Just scaling up current N fixation might not help, however, because many nodules are occupied by rhizobia with high carbon (C) costs relative to N fixed. Given this low efficiency, greater plant investment in nodules can decrease yields (1). Even if we inoculate with more-efficient rhizobia, evolution can favor mutants that divert more resources from N fixation to their own reproduction (2). Could we instead breed legumes that preferentially support the reproduction and release into soil of only the most-efficient rhizobia in their nodules? If so, how much would this benefit future crops grown in the same soil? In PNAS, Westhoek et al. (3) present modeling and convincing results that expand our understanding of host-imposed selection among rhizobia differing in efficiency. Their methods could help plant breeders “select for legumes which are better at discriminating among strains” (3). Although their results are also relevant to broader questions about the evolution of cooperation, this commentary will focus on issues related to that suggested application.

Previous modeling showed that, without host-imposed selection, rhizobia strains that invest anything in N fixation would have lower fitness than competing strains in the same plants that allocate resources only to their own reproduction. Fixing strains would only have greater fitness if hosts impose fitness-reducing “sanctions” on less-beneficial strains (2). All legumes tested appear to reduce allocation to nonfixing nodules (4–6), although decreased allocation may not always limit rhizobia reproduction (7). Most nodule-forming rhizobia therefore fix at least some N.

Mediocre rhizobia appear to be common, however. Westhoek et al. (3) therefore asked how much a pea cultivar discriminates among N-fixing nodules differing in efficiency, taking a more comprehensive approach than most previous work. Their model assessed plant-optimal allocation of resources among shoots, roots, and nodules, assuming plant growth is maximized when equally limited by N and C. The model shows that if a plant hosts a high-efficiency strain then C allocation to a low-efficiency strain should be less—ideally, zero—than if the low-efficiency strain is alone.

We know more about sanctions on nonfixing rhizobia than on mediocre rhizobia. This is partly due to the lack of reliable data on even relative efficiency of fixing strains. Comparing growth of plants, each inoculated with a different single strain, is relatively easy. Growth of singly inoculated plants may, however, depend more on rapid or profuse nodulation than on efficiency (8). Mixing different ratios of strains and regressing plant growth on each strain’s nodule occupancy might better predict real-world benefits (9, 10). Direct efficiency measures include the ratio of plant N to nodule mass or the ratio of N-fixation rate to nodule respiration (11). However, for high-throughput screening of legume genotypes, strains that differ only in N fixation (not in signals, etc.) may be most useful. So, Westhoek et al. (3) made and validated isogenic nonfixing and “intermediate-fixing” versions of a standard strain. Inserting a Ω Sp^r cassette into the nifH gene prevented N fixation, whereas insertion between fixX and nifA genes reduced N fixation rate, presumably due to decreased nifA expression and synthesis of nif and fix proteins. The latter strain will be referred to here as “low-fixation,” as its fixation rate was only about 25% of wild type.

Given a plant hosting isogenic strains of rhizobia differing only in N fixation, how should sanctions be measured? Nodule size may be a good proxy for treatment-related differences in rhizobia reproduction by a single strain, but not always for comparisons among strains (7, 12). Westhoek et al. (3) therefore used an impressive range of methods for assessing sanctions and their consequences. They measured nodule size but also viable rhizobia per nodule. They assessed relative N-fixation rates using the acetylene-reduction assay, with subsaturating acetylene. Low acetylene concentration avoids the “acetylene-induced decline” (13). Like slower growth of less-beneficial nodules (14) a shut-down by plants of nodules that are processing only acetylene rather than atmospheric N gave early evidence for sanctions. Westhoek et al. also assessed nodule redness from leghemoglobin, which is correlated with N fixation (15) because of its “facilitation of O₂ flux” (16). To measure C allocation to nodules, they linked lux genes to promoters for C transporters, making nodule luminescence depend on C supply (3).

All these methods led to the same conclusion: Low-fixation rhizobia received more C and reproduced more when alone or with the nonfixing strain, relative to when they shared a plant with the high-fixing strain. This discrimination was less extreme than the theoretical optimum predicted by the model. Starving mediocre nodules while waiting for more-efficient nodules to grow may not enhance plant fitness. The main evolutionary and agricultural implication of this research is that the pea cultivar used would impose selection against low-fixing as well as nonfixing strains, if high-fixing strains are available.

The luminescence assay for nodule C supply could complement use of green fluorescent protein fluorescence to estimate N fixation rate per nodule (17), also published in PNAS and also with Philip Poole as a senior author. Such optical measures of nodule-level processes could be even more powerful if used nondestructively to track nodules over time. For example, optical tracking of leghemoglobin oxygenation during sanctions showed that decreased oxygen supply may precede decreased C supply (4). Similarly, repeated imaging of individual nodules revealed rapid and reversible decreases in nodule growth rate with nitrate exposure (18). Differences in growth rate among nodules with different isogenic strains could reveal differences in sanctions among legume genotypes. Using artificial intelligence to identify and measure nodules (19) could aid high-throughput approaches. Plants grown in growth pouches are ideal for such nondestructive assays (Fig. 1). For example, this approach revealed an osmoelectrical mechanism for regulation of nodule oxygen permeability (20). Growing plants in soil (3, 17) may, however, increase field credibility.

Fig. 1.

Roots in growth pouches expose nodules for optical assays. Red shows nodules identified and measured by RootPainter (19); false positives at the top are removed in postprocessing.

In PNAS, Westhoek et al. present modeling and convincing results that expand our understanding of host-imposed selection among rhizobia differing in efficiency.

If these methods eventually lead to legume crop genotypes that impose stricter sanctions against mediocre rhizobia, will increases in the relative abundance of the most-efficient local rhizobia last long enough to greatly benefit the next crop? Sequencing-based approaches may help track rhizobia evolution as well as helping to explain differences in efficiency (21, 22). Rhizobia released from nodules may persist for years in the soil (23), but resources diverted from N fixation by rhizobia could enhance their subsequent soil survival. This trade-off would be an even greater problem if there were no sanctions (2), a point relevant to developing cereal crops that host rhizobia.

Legume cultivars with stricter sanctions may be the best near-term approach to improving the efficiency of N fixation and decreasing agricultural use of N fertilizers. Targeting sanctions could be difficult, however, if many nodules contain multiple strains. Sanctions against less-beneficial strains within mixed nodules have been reported (24) and could be useful for crops like soybean and bean, which host rhizobia descended from the N-fixing, bacteroid form that reproduced in nodules of previous crops. Pea, however, hosts “eusocial” rhizobia: Bacteroids are terminally differentiated and nonreproductive. To favor more-beneficial strains, sanctions in mixed nodules hosting eusocial rhizobia would have to be targeted against the still-reproductive, undifferentiated clonemates of low-fixing bacteroids (25), which might not be possible. If mixed nodules are common in pea (17), improving sanctions may be difficult. It is possible, therefore, that the methods described here could be applied more quickly to crops like soybean.