A summary of Friendship and natural selection by Nicholas A. Christakis…
A summary of
Friendship and natural selection
by Nicholas A. Christakis and James H. Fowler
Humans often seek friends in quantifiable ways whether it be genetically similar friends, called homophily, or genetically dissimilar friends, called heterophily. Christakis and Fowler. 2014.
Humans are unique species in that they form bonds with individuals that are not children or potential mates, something that is fundamental to the human experience since the hunter-gatherer days. Christakis and Fowler. 2014.
There are four reasons why homophily may be exhibited. The first is that people who have lived in the same place for a while might be interacting with people who share the same distant ancestry. The second is that people may actively foster relationships with similar genotypes and avoid those with dissimilar genotypes, often chosen through how those genes manifest themselves phenotypically. The third is that the places in which people live contain more genetically similar people, meaning that their phenotypes are influenced by specific and correlated phenotypes. The fourth is that people may be put into groups or events that contain similar people to make contact with, such as schools. Christakis and Fowler. 2014.
There are reasons why heterophily may be exhibited as well, but not as many. The first is that environments may encourage interactions between people with dissimilar traits. The second is that people may actively attempt to befriend others that are different than them, and has even been proposed in more detail within reproductive relationships Christakis and Fowler. 2014.
These actions are likely all working with or off of each other simultaneously, and evolve depending on fitness for given traits. Homophily seems to evolve under a much wider umbrella of traits than heterophily, however. This is likely because it is less costly to find similar people than dissimilar people. Christakis and Fowler. 2014.
Studies and Results
Single Nucleotide Polymorphisms were analyzed in unique subjects in friendship pairs. The kinship coefficient was studied for homophily because is reflects relatedness in genome wide complex trait analysis. Heterophily was measured by calculating the probability to have dissimilar genotypes at given SNPs These studies showed positive correlation in homophily and negative correlation in heterophily when compared to strangers. Christakis and Fowler. 2014.
Population stratification was studied closely in order to not skew results, and strict controls were applied because of this. Less friendship pairs were studied in this second study, but with negative kinship to reduce the possibility of interacting with a distant relative, using strangers as a control yet again. These results still showed significant positive and negative correlations in each of the areas expected for homophily and heterophily. The T statistics and P-value were even further scrutinized to support this. Christakis and Fowler. 2014.
Since these studies were done with SNPs and the focus of this study is on genome-wide analysis, a process to determine a "friendship score" based on measures of correlation was developed to determine whether or not people are likely to be friends based on how similar or dissimilarly their genotypes may be. Christakis and Fowler. 2014.
There are likely many hundreds of genetic pathways that form these levels of correlation, and the samples studied can allow for the discovery of some of these. The top 1% of homophilic and heterophilic genes were examined with the top 25% of polygenic traits in order to analyze this correlation, but more studies need to be carried out in order to resolve whether or not a few biological systems are highly correlated or many smaller systems are weakly correlated. Christakis and Fowler. 2014.
These pathways also likely play an important role in recent human evolution with regards to immune response, reproduction, and sensory perception. SNPs were analyzed once again to analyze potential positive selection and proliferation, with the top 20% of the most homophilic SNPs having higher Composite of Multiple Signals scores than those of strangers. Christakis and Fowler. 2014.
Implications and Analysis
These studies show that genetic structure is not entirely dependent on reproduction, and that friendship unions can also influence genotypes as shown above. This feeds into supporting the gene-environment correlation, both in a social and physical sense, wherein social environments that affect fitness are chosen. Christakis and Fowler. 2014.
This data also suggests that humans may be metagenomic with other humans in a similar way as they are to microbes within them. It was also found that friends are, on average, as genetically similar to each other as fourth cousins, meaning that friendship is important in biological processes as much as it is in social processes. Christakis and Fowler. 2014.
This may also reflect the idea of a proposed kinship detector in humans, wherein friends present functional relatedness to potential friends. This implies friends are more of a "functional kin" since they provide increases in fitness. The most intriguing is the relationship these genes have to olfactory processes, since this biological system is closely related to emotional cognition and recognition. Christakis and Fowler. 2014.
Implications of these studies include complimentary advantages rather than synergy when it comes to immune response due to adaptation, the foundation being laid for further in-depth kinship detection studies, and evidence that the human evolutionary environment is not just the physical environment, but the social environment as well. These all play a significant role in modern human evolution, which may imply the acceleration at which humans are evolving due to the increased gains from friendship driven synergy. Christakis and Fowler. 2014.
Christakis NA, Fowler JH. Friendship and natural selection. Proceedings of the National Academy of Sciences of the United States of America. 2014;111. doi:10.1073/pnas.1400825111