Garbage in, garbage out, in modeling symbiosis
Every year or two, someone publishes a computer model showing how easy it would be for cooperation between species to evolve, if only the world were fundamentally different than it actually is. In particular, many models incorrectly assume that there is only one symbiont genotype per host individual.
The latest paper to make this incorrect assumption was published in Proceedings of the National Academy of Science, and titled "Economic contract theory tests models of mutualism." The paper admits that:
"the fact that legumes are often infected by multiple strains of rhizobia [symbiotic bacteria that provide plants with nitrogen] remains a problem, because our model assumes only one type of agent."Similarly, each individual plant typically interacts with several different mycorrhizal fungi and many different pollinators. How much difference does this make? Enough difference to completely reverse their conclusions that "positive feedback between host fitness and symbiont fitness is sufficient to prevent cheating."
We showed in 2002 that symbionts that benefit their hosts more would out-compete those that benefit their hosts less (even in the absence of "host sanctions", discussed below), if there were only one symbiont genotype per host (above right). That is the situation incorrectly assumed by Weyl et al. (2010). In the real world, however, with multiple symbionts per host, strains that divert more host resources to their own reproduction would win (above left). For rhizobia, if nitrogen is readily available in the soil (s=0.5), then as few as two strains per host is enough to undermine cooperation, unless hosts impose sanctions on cheaters. Source: West et al. (2002) "Sanctions and mutualism stability: why do rhizobia fix nitrogen?" Proc. Roy. Soc. B 269:685-694.
Weyl et al. (2010) is one of the 100+ papers to cite our 2002 paper, but the critical importance of multiple symbionts per host doesn't seem to have sunk in. If you're looking for a model that represents an actual advance on our 2002 paper, rather than an embarassing retreat, I recommend Maren Friesen's "Mixed infections may promote diversification of mutualistic symbionts: why are there ineffective rhizobia?", J. Evol. Biol. 23:323.
West et al. (2002) also showed that then-hypothetical "host sanctions" imposed by legume plants against symbiotic rhizobia that fail to provide them with nitrogen can prevent the evolutionary breakdown in rhizobial cooperation shown above. Toby Kiers subsequently showed that soybean plants, at least, do actually impose such sanctions on individual root nodules that fail to provide the host with nitrogen ("Host sanctions and the legume-rhizobium mutualism", Nature 425:78). But, Friesen and Mathias asked, what if some nodules contain more than one genotype of rhizobia? Depending on the frequency of mixed nodules, they showed that rhizobial "cheaters" can coexist with more-beneficial strains.