Nicholas Wade gives some recent highlights of research into ongoing selection in humans.
We are at the center of this research
According to Wade’s account, the initial skepticism of many geneticists to this idea seems to have mostly evaporated. I think that much of the caution was reasonable conservatism – few people expected to see such widespread effects of selection. Only those of us who were thinking of the changes in the Neolithic and later were really prepared to interpret the evidence. But now, the sheer accumulation of studies has shown that our initial estimates may have been too conservative.
About 21 genome-wide scans for natural selection had been completed by last year, providing evidence that 4,243 genes 23 percent of the human total were under natural selection. This is a surprisingly high proportion, since the scans often miss various genes that are known for other reasons to be under selection. Also, the scans can see only recent episodes of selection probably just those that occurred within the last 5,000 to 25,000 years or so. The reason is that after a favored version of a gene has swept through the population, mutations start building up in its DNA, eroding the uniformity that is evidence of a sweep.
Unfortunately, as Joshua M. Akey of the University of Washington in Seattle, pointed out last year in the journal Genome Research, most of the regions identified as under selection were found in only one scan and ignored by the 20 others. The lack of agreement is sobering, as Dr. Akey put it, not least because most of the scans are based on the same Hap Map data.
From this drunken riot of claims, however, Dr. Akey believes that it is reasonable to assume that any region identified in two or more scans is probably under natural selection. By this criterion, 2,465 genes, or 13 percent, have been actively shaped by recent evolution. The genes are involved in many different biological processes, like diet, skin color and the sense of smell.
That’s 13 percent with statistical evidence in two or more studies. Keep in mind that our present sample size is small enough that we can’t reject the hypothesis of genetic drift on things that have frequencies lower than ten percent in a given population. So probably the variants we know about are the tip of a larger iceberg of rare selected variants, which originated within the last few thousand years and haven’t had time to increase to higher frequencies. Some may have stalled out at lower frequencies, because of epistases or changes in the environment.
The proportion of affected genes should approach some asymptote, as lower-frequency variants will be likely to hit the same gene categories again and again. Diet, skin color, smell, disease, brain, all systems that have been under strong selection pressure in recent human evolution. That may provide a promising way to uncover functional relationships among genes. Wade’s description of Anna Di Rienzo’s work seems to be along those lines.
Many workers seem to realize now that humans don’t live in hunter-gatherer environments. But a disappointment for me is that the article doesn’t discuss the role of demography in generating this unique evolutionary pattern. Demography provides an important filter on the results of genome-wide analyses, also. The power of statistical methods is not uniform across different ages of adaptive alleles. Some methods miss older events while all methods miss very recent ones.
Statistical power is an important reason why some studies find more evidence of selection in Europe and East Asia compared to Africa. The demography of those regions means that Africa has a broader distribution of ages of positively selected mutations: more older events, fewer events corresponding to the peak population growth of early agriculturalists.
There is some stuff in the article about “soft sweeps” – the hypothesis that much recent phenotypic change may result from selection on standing genetic variation in ancient populations. An allele that already existed neutrally in the population can come under new selection, and that kind of selection won’t trigger the criteria for genome-wide selection scans.
I have some thoughts about this phenomenon that I’ll write up and share. We know that there were some big phenotypic changes in the Late Pleistocene and early Holocene, and initially these changes should mostly have involved standing genetic variation. New adaptive mutations were coming into these populations at a relatively slow rate. When a new mutation is still rare, it doesn’t have much impact on the average phenotype in the population. So if we see a fast change to the average phenotype, we know that new mutations aren’t responsible, at least not initially.
But it doesn’t take very many genes to cause phenotypic changes. And if small populations have few new adaptive mutations, they also have relatively little standing variation. So the importance of soft sweeps to our evolution may be great, even if their numbers are ultimately small.