Estimates predict that somewhere between 15 percent and 30 percent of the variability in human lifespan is due to genetics. But in a study published in Genetics today (November 6), researchers have shown that those are likely overestimates and that assortative mating—that is, people choosing partners with traits that resemble their own—can account for most of what looks like heritability.
“What [the authors] did that no one else has done is try to really evaluate this idea of assortative mating” and its contribution to lifespan, says Braxton Mitchell, an epidemiologist at the University of Maryland who did not participate in the study. “The novel part of this was how they were able to put together this massive dataset of 400 million records . . . and are coming up with some very interesting analyses and hypotheses. It really reminds us how much we can do with these bigger and bigger datasets that come available.”
According to coauthor Graham Ruby of Calico Life Sciences, an aging-focused research organization that is funded by Google’s parent company Alphabet, the broad question driving the study is how much of a role genetics plays in any phenotype, in this case, lifespan.
Traditionally, when researchers want to investigate how much a phenotype runs in families compared to the rest of the population, they make do with whatever family tree datasets are available, Ruby says. “Generally, those are relatively small from a statistical power perspective,” he says.
To overcome this limitation, Ruby and colleagues collaborated with scientists from Ancestry, the genealogy and consumer genomics company. The Ancestry team aggregated family trees containing more than 400 million individuals, mostly American customers of European descent and their relatives. They used an algorithm to eliminate duplicates and to evaluate the accuracy of the entries—specifically, the names, dates, and locations indicated in the trees—and then de-identified the data.
The researchers used the pedigrees to calculate how similar lifespan is within families, compared to the population as a whole. Their estimates were consistent with other groups’ previous findings: somewhere between 20 percent and 30 percent of longevity appeared to be due to genetics. But the team also found correlations between lifespan in spouses that were higher than that of opposite gender siblings.
The scientists hypothesized that the similarities in length of life between spouses could be due to their shared living situation in adulthood, assortative mating, or both. To tease apart these possibilities, they next looked for concordance in lifespan among in-laws, who don’t share an environment and are not genetically related.
The results revealed a greater-than-expected correlation between the longevity of siblings-in-law and that of cousins-in-law, and this similarity persisted among even more distant relationships, including brothers or sisters of a sibling’s spouse. These correlations over familial distance are consistent with assortative mating being involved, the researchers write in the study. Using statistical models that accounted for mate choice, they estimated the heritability of longevity at about 7 percent, well below previous estimates.
Without seeing the future, it’s difficult to choose a spouse based on lifespan. But, as the authors discuss in the text, one complex factor that has been previously shown to be transferred within families and to also influence lifespan is socioeconomic status, which encompasses a variety of aspects, including wealth and education level. Thus, when people choose a partner with a similar educational background or familial level of wealth, they could be contributing to the observed effect of assortative mating.
“They make a pretty convincing case that most of the previously published estimates of the heritability of longevity are probably inflated by” assortative mating, says Richard Kerber, who studies longevity at the University of Louisville and was not involved in the work.
“The thinking in studying the genetics of longevity has been shifting slowly over the last couple of decades,” Kerber says. “A more realistic picture of the very complex phenotype of longevity is obviously something that's needed and something we probably won’t make any real headway on until we’ve got a good handle on all the many pieces that are involved.”