Whenever I give a talk on evolution for the public, I always get this question: “Are we still evolving?” People want to know if our species is changing. Are we getting smarter or better-looking? Or are we degenerating as medicine keeps alive many of us who would have been ruthlessly culled on the savanna?
My answer usually goes like this: “Yes, many of the genes that would have been eliminated now persist. I am myopic, and wouldn’t have been a good hunter. Genes for nearsightedness are undoubtedly accumulating in our species. So in some respects we are genetically degenerating. But so long as there are sources of mortality — and the genetic variation to resist them — we will continue to evolve. In Africa, for instance, thousands of children die yearly from infectious diseases. It’s highly likely that genes for resistance to those diseases are becoming more numerous.”
Now this answer is speculative, based on evolutionary theory. There’s not a lot of evidence about whether and how fast the human species is changing. But a new paper in The Proceedings of the National Academy of Sciences goes one step further: it not only demonstrates genetic variation for sources of mortality, but shows that that variation is correlated with reproductive output. If both of these hold, then we must be evolving. The paper goes on to predict how much genetic change for several traits that Homo sapiens will experience over the next ten generations.
The paper is by Byers et al. (see reference below), and should be online this week. (Thanks to Steve Stearns for letting me see a galley proof.) There has already been a flurry of publicity — one example is here — and there will be more.
The sample population included about 5,000 women from the famous “Framingham heart study,” a survey begun in 1948, with the individuals sampled every four years for a variety of physical and traits (weight, height, age at first reproduction, blood pressure, serum cholesterol, etc.). Since the individuals measured are now on the senior side, the original sample population (now past reproduction) has produced their allotment of children.
This gives us a basis for detecting natural selection. The underlying rationale is that we look for a correlation between an individual’s traits and the number of children she produced. If we find one, and there is genetic variation for the trait, we can reasonably assume that natural selection is acting on it, and we can see what type of natural selection by seeing how variation in the trait is associated with variation in offspring number. If, for example, higher cholesterol is associated with lower reproductive output (as it might be if cholesterol is associated with heart disease that strikes before reproduction is finished), then we can assume that there is natural selection against higher cholesterol. If we know how much variation in people’s cholesterol is based on variation in their genes (a proportion called heritability, which ranges between zero and one), we can then calculate how we expect the trait to change in the future. That is, we can see if we’re evolving, and predict how much.
The underlying method, developed by quantitative geneticists such as my old mate Russell Lande, involves looking at several traits, any or all of which may be subject to selection. These traits may have some “genetic correlation,” that is, genes affecting one trait might affect another. (For example, genes for higher weight might raise blood pressure as a byproduct). These correlations can be measured using data on the observed variation in the traits and the degree to which this variation is passed from one generation to the next (the heritability of the traits).
If you calculate all these correlations and then multiply them by the amount of selection that appears to act on each trait (this has to be discounted using the correlations), you can estimate the degree of genetically based change you expect each generation: that is, the amount that each trait will change over one generation by natural selection.
What they found.
Several traits did indeed appear to be undergoing selection. From the amount of this selection, we can predict the percentage change in the trait that we expect to see after ten more generation of reproduction (roughly 300 years from now).
Total cholesterol: going down. Projected to drop 3.6% in ten generations
Weight: going up a tad, projected to increase 1.4% in ten generations
Height: we’re getting shorter projecting a drop of 1.3% (2.1 cm) in ten generations.
Systolic blood pressure: Going down, as predicted. Projected to drop 1.9% in ten generations.
Age at menopause: Going up; projected to rise 1.6% (0.8 years) in ten generations.
Age at first reproduction: Going down. Projected to drop 1.7% (from 26.18 to 25.74 years).
So women, at least, are getting shorter, stouter, and reproducing earlier and over a longer period of time. This is evolutionary change. Based on this study, we can tentatively say, with more assurance than I used to, that yes, our species is still evolving.
But there are two important caveats to this study, both of which were recognized by the authors.
1. The “inheritance” of the trait includes not just genetic inheritance, but cultural inheritance. Humans pass not only their genes to their offspring, but aspects of culture that may mimic a genetic inheritance. For example, parents who eat a lot may induce their kids to eat a lot, and some of the correlation of weight between parents and their kids may be due not to shared genes, but to shared food. Parents who for cultural reasons have their kids early may induce their own kids to produce grandchildren early. As the authors say, “We are not able to differentiate the effects of genes and culture with these data.”
This is a bit of a problem, because the evolutionary projections are based on assuming that all of inheritance is genetic. It’s hard to get around, since distinguishing genetic from cultural inheritance involves difficult work using data from adopted children or twins raised together versus apart. Nevertheless, we can probably assume that some selection is acting on these traits, discounted by the degree to which parent/offspring resemblance reflects cultural similarity.
2. The predictions may be hard to verify, because they assume that the environments of our species — that is, the environments that are relevant to each trait — will remain constant over the next few generations. Cholesterol is predicted to drop from 224 to 216 mg/100 ml blood over the next ten generations, but this assumes that diet is constant. If people stop eating fatty foods, the drop may be even steeper because of this dietary change. Conversely, if people flock even more frequently to fast-food joints, the predicted drop may be negated by an increased intake of burgers and fries.
These problems mean that one has to be a bit careful about not only predicting the degree of selection, but also testing those predictions in the future. Yesterday Stearns was on NPR’s Science Friday, discussing how the authors deal with the conflation of culture/environment and genes, and why the prediction of increasingly plump women may not be what it seems.
But regardless of the caveats, the study of Byars et al. is based on a good idea, and makes a reasonable case that selection is indeed acting in our species. One would, of course, like confirmation from other datasets, but since the Framingham study is unique, this may be hard to do.
h/t: Steve Stearns (who reviewed this post for accuracy).
Byars, S. G., D. Ewbank, D. R. Govindaraju, and S. C. Stearns. 2009. Natural selection in a contemporary human population. Proc. Nat. Acad. Sci. USA: in press.