Do humans form genetically similar social groups independent of kinship?

Proponents of kin selection as the most parsimonious explanation of how cooperation evolves face a problem when it comes to humans: counter to the predictions of kin selection theory, humans aim a fair amount of altruism at non-kin. While we do not aim our helping behaviors solely at our relatives, we also do not randomly distribute our kindness among strangers. Generally, our altruism is aimed at people we know well and spend a lot of time around, our “friends”.

We often choose as our friends people who are quite similar to us. We may find friends among others who share our hobbies, or our passion for particular foods, or our sense of humor (or even lack thereof). We take this “assortative affiliation” for granted, as some set of shared traits seems to form the foundation of almost all friendships. But does this nature of friendship have any evolutionary significance? Are we choosing our friends based on their similar behaviors and physical characteristics? And if we are, could our ability to choose friends based on similar phenotypes allow us to forge social alliances with those who are more genetically-similar to us than the general population?

Evidence presented in a recent PNAS article with the bold title “Friendship and natural selection” suggests that our preference in friends may be an evolved characteristic. Using data from a study that measured both social relationships and genomic variation, researchers Nicholas Christakis and James Fowler present compelling evidence that we choose friends that are more genetically similar than the general population. Not only is it possible that humans mate assortatively; we may also exercise a kind of “assortative affiliation” when we choose our friends. Somehow we are able to use phenotypic characteristics — which could include widely-different traits such as scent, physical appearance, and behavior — to create friend-groups that are equivalent to a group of distant cousins. These researchers go so far as to label our friends as “functional kin”, a term that reminded me a lot of the bonds of “fictive kinship” that have been well-studied by sociologists.

Does this all mean that we are a bunch of green-beards, replacing “cooperation with genetic kin” with “cooperation with genetically-similar individuals”? If so, does that mean that the kin selection paradigm explains human cooperation well, albeit via the more indirect process of phenotype matching? This is a tempting conclusion to reach, especially if you favor kin selection as the primary explanation of how cooperation evolves. I would suggest that you should resist this temptation to take the idea of “functional kin” too seriously. The problem is that these “functional kin” are not actually very closely related. In this study, the genetic similarity of friends was analogous to that of fourth cousins. While this similarity was significantly greater than non-friends (and that is the big finding here!), friendship appears to lead to social groups that still maintain rather low levels of genetic similarity (probabilities of shared genotypes of about r = 0.001). Understood strictly from a kin selection perspective, this level of relatedness-by-assortative-affiliation is too low to promote significant altruism. Just pumping this number into Hamilton’s Rule, you can see that helping behaviors have to maintain a 1000-to-1 benefit-to-cost ratio in order to be evolutionarily stable. That kind of evolutionary gradient would only favor a tiny subset of helping behaviors, and although it is tough to accurately assess benefit-to-cost ratios for human helping behaviors, I cannot imagine that many result in a 1000-to-1 ratio.

What I find most provocative about these findings does not pertain to the particular benefits that may accrue to an individual who chooses friends with slightly-more-similar genotypes. Instead, I see these findings in a social context: they suggest that larger networks of friends may form groups that maintain higher levels of genetic similarity than the population at large. This suggests that there may be competition between genetically-dissimilar groups within our society. Yes, I am connecting these findings to group selection! What is fascinating about social networks created among friends is that the “groups” involved do not share any sort of geographical territory. Instead, social boundaries maintain distinct groups. It would be fascinating to study social networks from the perspective of genetic similarity.

The other thing to keep in perspective here is just how genetically similar we all are. The following diagram — Figure 1A from this paper — captures this reality nicely:

Do we choose friends that are significantly more genetically similar to ourselves? Apparently so, but this similarity is a shade greater than the overall similarity maintained among members of the entire population.

Another really fascinating finding in this study is that we choose as our friends those who are less genetically-similar at critical immunological loci. This finding parallels the finding that mate-choice is also driven by selection for genetic dissimilarity at immunological loci. While the choice of a mate with different a different immunological genotype can be explained strictly based on the assumed fitness benefits conferred to offspring, explaining why you would want to hang out with friends who do not share your immune profile is a bit more difficult. These authors suggest that our social groups also represent epidemiological groups, and that by choosing individuals with different immunological strengths and weaknesses we minimize our risk of disease. This is a hypothesis that needs to be tested more rigorously, but if it proved true would bolster the perspective that much human evolution is influenced by characteristics at the level of the social group.

If you want a quick introduction to this research, you should check out the radio segment “Do We Choose Our Friends Because They Share Our Genes?” on NPR’s All Things Considered.