The model that Greg is dancing around suggests (1) that there is variation in mutation rate dependent on temperature or something correlated with temperature, (2) higher mutation rates cause a higher genetic burden in human populations, (3) leading to IQ reduction and other minor dings. One interesting possible ding is premature birth, which affects 10 to 12 percent of pregnancies among Whites and Hispanics in America and about 18 percent of pregnancies among US Blacks. The trio of prematurity, low birth weight, and infant mortality is an especially conspicuous burden for the Black population.
It is easy these days to pull relevant data off the web, bearing in mind that data quality varies a lot in unknown ways. The scatter plot below shows average temperature by state for the US, on the x axis, and rates of preterm birth on the y axis, for non-Hispanic Whites.
The simple correlation between temperature and prematurity is 0.55, not bad for social science. There is no real correlation at all for the US Black population and only an insignificant hint of one for the US Hispanic population.
This number should not be regarded as an ordinary correlation because there is spatial autocorrelation among the states. For example, should Mississippabama be treated as one state or two? How long has the DNA of people in a state been in the state? For example one suspects that much of the White population of Florida has only arrived within the last one or two generations. There are a lot of other ways that one could pick at these data.
The overall pattern is roughly mirrored by state IQ estimates or their proxies like NAEP scores, by infant mortality rates, low birth weight rates, and so on. In color coded maps I often notice a kind of V of lower IQ and worse health in US Whites, the base of the V in Mississippi with the right arm extending up the Appalachians to West Virginia and the left arm, less clearly defined, reaching to Colorado. A nice source of such interactive maps is at this link.
PS: The jpeg of the scatter plot is not very good: a direct link to a pdf is here.