replying to S. C. Engelhardt & M. Taborsky Nature Ecology & Evolution (2020)


While Engelhardt and Taborsky1 agree with the core argument of our recent article2—that enforcement is central to the evolution of cooperation—they go on to suggest that our use of the word ‘enforcement’ is too broad and that we neglect to distinguish between behavioural manipulations that occur via helping or harming. We find no merit in these points.

First, semantics. Engelhardt and Taborsky object that our use of ‘enforcement’ does not correspond exactly with their everyday use of the word. The fit between common parlance and the terminology of evolutionary biology is an old concern3, but one that can be mitigated by clearly defining terms4. Our definition of enforcement—an action that evolves, at least in part, to reduce selfish behaviour within a cooperative alliance—reflects the established usage for describing mechanisms such as partner choice, punishment and reward in cooperative systems across all scales of life (for example, see Fig. 2 in West et al.5). Moreover, it was precisely because of the subjectivity of language that we also provided an extensive theory supplement that defines enforcement mathematically (see also Box 2 in our original piece2). Instead, Engelhardt and Taborsky provided an ‘everyday-language’ definition of enforcement (“an action involving manipulation by force to the benefit of an actor at the expense of a receiver”) that is so vague, and indeed broad, that it would draw in examples of both predation and parasitism. This illustrates why everyday-language definitions are often insufficient for evolutionary biology.

The second key point of Engelhardt and Taborsky is that we do not sufficiently distinguish between enforcement through harm and that which occurs through the provision of incentives or help. Again, we disagree. Our review explicitly distinguished between harming and helping mechanisms of enforcement (see Table 1 in our original piece2) and gave numerous examples of each. We also reject their point that manipulations by harming versus helping are so different that distinguishing them is ‘essential’. Consider, for example, a host controlling its microbial symbionts either by differential harming or differential feeding, as is thought to occur in the mammalian microbiome via various antimicrobial peptides and glycosylated mucins, respectively6. To increase the relative abundance of a beneficial symbiont, the host can preferentially feed cooperative symbionts, preferentially intoxicate non-cooperative symbionts, or both. However, importantly, the different mechanisms can have exactly the same evolutionary effects on cooperation. Therefore, while we appreciate that harming and helping mechanisms of enforcement can have their differences2, they have a lot more in common to the extent that they will be sometimes indistinguishable in their effects on cooperative evolution. For this reason, it makes a lot of sense to study them together in one framework.

The utility of our framework2 becomes clear when one looks beyond Engelhardt and Taborsky’s focus on the animal behaviour literature. One of the major advances in evolutionary biology of recent decades is the recognition that the principles of social evolution apply equally to selfish genetic elements, microbes, mutualisms and societies7. Restricting focus to animal behaviour1 prevents one from recognizing the shared principles and establishing the common vocabulary needed to understand the evolution of cooperation across all biological systems.