I found this astonishing article detailing the decline in glucose
intolerance amongst an isolated island population over a 12 year period.
Although I do not think it is well presented, and some of the control data
can be confusing, I think the conclusion is clear. Basically it says that,
all else being equal, rampant diabetes(type 2) and glucose intolerance
amongst this population induced such pressures on it in time, that there
was a decline the incidence of glucose intolerance and diabetes (type2).

"the decline in incidence of glucose intolerance is probably due to the
intensity of the (diabetic)epidemic in Nauru, which has already removed a
high proportion of the genetically susceptible individuals from the pool
with normal glucose tolerance. Coupled with the observations that mortality
is higher and fertility lower in diabetic Nauruans across the age range,
the observed decline in the incidence and prevalence of
non-insulin-dependent diabetes mellitus and impaired glucose tolerance may
presage a fall in the population frequency of the diabetic genotype, at
least in its more severe form, as might be predicted on the basis of Neel's
"thrifty genotype" hypothesis. "

[Decline in incidence of epidemic glucose intolerance in Nauruans:
implications for the "thrifty genotype" [see comments]
Dowse GK; Zimmet PZ; Finch CF; Collins VR
Am J Epidemiol, 1991 Jun 1, 133:11, 1093-104 ]

The time period is extremely short. We are hypothesizing that 10000 years
is not long enough to elicit change. I have another complete paper by John
Allen called "The non-thrifty Genotype" in which he expands on Neel's and
others, and talks about the existence of the non-thrifty genotype and its
correlation to lactose tolerance and low incidence of diabetes. As soon as
I figure how to scan it in and with John Allen's permission, perhaps I
could post it to someones web page?

Gary