A year ago there was a paper on the effect of diet on enzyme production, Diet and the evolution of human amylase gene copy number variation. That human evolution skeptic, P. Z. Myers, has just noticed the paper, and says:
This work by Perry and others went on to look for patterns in different human populations with different dietary historys, and discovered that there is a correlation: cultures with diets heavy in starch, agricultural populations such as Americans, Europeans, and Japanese, or hunter-gathers who live on many roots and tubers, have a higher average copy number than cultures that depend more on hunting and fishing.
Look at the distributions! Populations with little starch in their diets also have a relatively low copy number of 5.44 amylase genes per individual; we french fry eaters have a higher number of 6.72 amylase genes per individual. The difference is small, and the distributions also overlap significantly (note that some with high starch diets only have 2 copies, and some living on low starch diets have 13 copies), but the difference is measurable and significant. It implies that there may have been some selection for greater copy numbers in cultures with diets high in starchy plants.