Several recent papers give me the impression that there is regional variation in mutational load. One can slice this a number of ways. A recent Science article looked at mutations that knocked out genes – loss-of-function or LOF mutations. Mutational load is the sum of all deleterious mutations – LOF mutations are a clear-cut subset of total mutational load.
Some of the LOF mutations they are found are common, and are presumably neutral, maybe even beneficial, but most are rare and likely deleterious. The kicker is that they found significantly more LOF mutations in their African population sample than in their European and East Asian samples – 25% higher. That was unexpected. Population history (and mutation rate) determine the variation you expect to find in neutral genes, but significantly deleterious mutations should be in mutation-selection balance. A neutral variant might easily be a million years old, but a deleterious variant will last, on average, 1/s generations, when s is the decrease in fitness caused by that variant. A mutation that decreases fitness by 1% should disappear in 100 generations or so, about 2500 years. Ancient bottlenecks should not influence the frequency of such noticeably deleterious mutations.
Another related paper looked at a large set of coding genes, sequencing many times (average depth of 111x) for high accuracy. As in the LOF paper, they found that the average person carries many probably-deleterious mutations, mutations which are individually rare. Each person carried, on average, mutations expected to change function (almost always for the worse, although usually only a little for the worse) in 313 genes (out of the 15,585 they studied.
They looked at African-Americans and Americans of European descent, about a thousand of each. They saw what what was seen in the LOF paper did: there were significantly more probably-deleterious mutations in the 80%-African population. When they used a loose definition of functional variation, about 20% more : with a more conservative definition, which should have a higher fraction of truly deleterious genes, about 29% more.
This African excess is similar in every category they examined: in synonymous changes, in missense changes, in mutations that affect splicing or that turn a protein into hash. Synonymous changes are neutral (or close enough), and more neutral variants in Africans could be a consequence of population history (a generally larger population size over most of the last few tens of thousands of year). But that kind of ancient population history can’t matter much when we’re talking mutations that drop fitness by 1% or more, so what’s going on? The only simple explanation (that I can think of) is a higher mutation rate.