Cooperative behaviour is susceptible to cheats — individuals who reap the benefits of collective behaviour but do not contribute their fair share will do better than altruists. It has been assumed that cooperation persists despite this because individuals tend to interact with their relatives, and reducing the fitness of relatives by cheating indirectly reduces an individual's own fitness; however, this has been difficult to demonstrate. Research on the social amoeba Dictyostelium discoideum has now shown experimentally that cheats do not prosper because of the high levels of relatedness in natural populations.

When food is scarce, solitary D. discoideum cells aggregate into a fruiting body that distributes spores. However, only 75% of the aggregating cells become spores — the other 25% form the body's stalk and altruistically die. Mutants that are less likely to contribute to the stalk are at a selective advantage, and one such mutant is fbxA −.

The authors showed that fbxA− cells had a fitness advantage when introduced into a wild-type population at any frequency, but also that the spore production of the whole fruiting body declined sharply with an increase in fbxA− frequency. From this data they showed that, in fruiting bodies in which more than 25% of the cells were cheats, fbxA− cells are less fit than wild-type cells are in the absence of cheats. This means that if cells are more likely to make fruiting bodies with their relatives — in other words, if the level of relatedness in the population is high — different fruiting bodies will have different fitnesses depending on whether they contain a clone of cheats. Cheats will be selected against because the bodies in which they participate will do less well than predominantly wild-type ones.

So, what are the levels of relatedness in natural populations? The authors showed that most fruiting bodies consisted of just one clone. The average relatedness within fruiting bodies was more than 0.86, which from their experiments above is sufficient to put the cheats at a disadvantage. As expected, therefore, they did not find any fbxA− cheats in the natural populations that they studied, despite the huge selective advantage of such mutants in low-relatedness populations.

It remains unclear why relatedness is so high in D. discoideum populations, higher even than in eusocial insects, but it does explain how cooperative behaviour can be maintained. This has been difficult to show in other multicellular species, for which relatedness is easily measured but in which no cheater mutations are available for study.