Eye pigmentation in humans varies along a spectrum of colors from dark brown, through lighter brown, hazel, and green, to light blue. These differences are caused by variation in the content of the dark pigment, eumelanin, in the layers of the iris. Several genes are involved in the variation in color, but most of the lighter colors require a change in the expression of a gene called OCA2.
The lighter eye colors are most common in Europe, and in northern Europe in particular. Much of the variation in eye pigmentation in this population is associated with one area of the genome, on chromosome 15 in the region of the genes HERC2 and OCA2. The strongest association is with a single site, 28365618 nucleotides from the beginning of chromosome 15 in the current draft of the human genome. At this site, some human sequences carry an A, and others have a G.
This kind of variation is called a single nucleotide polymorphism (SNP). The word polymorphism meaning "many forms", but in fact this SNP has only two different forms, or alleles in human populations.
There are millions of SNPs in the human genome. When they sequence many people, geneticists often find SNPs they have never noticed before, and enter them into a catalog called dbSNP. Each SNP gets a catalog number, beginning with the letters "rs". This one, associated with eye color in Europeans, is known as rs12913832 (dbSNP link).
We know that rs12913832 is associated with variation in eye color because it has been genotyped in thousands of people. Blue-eyed people are very likely to carry two G's here. Why this SNP is associated with eye color is not yet clear. OCA2 is essential to forming normal pigmentation, but rs12913832 does not change the amino acid sequence of this gene. In fact, it lies within another gene, HERC2. The SNP may change the regulation of OCA2, or it may be linked on the same chromosome sequence to another mutation that does. Or the activity of HERC2 may itself affect pigmentation. Right now, scientists simply don't know.
Finding a genetic association, like a correlation, is not the same as finding a cause. An association doesn't necessarily tell us that the genetic change caused a change in the body; it merely indicates that one form of the gene is common in people with a particular trait.
An association may give some hint about the history of a trait. In the case of eye color, blue eyes are most common in northern Europe, and occur more rarely across southern Europe, north Africa, and West Asia. The G allele of rs12913832 has roughly the same distribution:
The G allele is most common in northern Europe, and is rare or absent in most of Africa and East Asia. However, the indigenous people of South America actually have fairly high frequencies (up to 30-40%) for this allele. Those populations do not have blue eyes at any appreciable frequency. What can explain this discrepancy?
Again, a genetic association is not the same as a genetic cause. This SNP allele may be linked to blue eyes in Europe because of its history: Another mutation that causes blue eyes may have happened on a copy of chromosome 15 that carried this SNP allele. Meanwhile, a different copy of chromosome 15 carrying this SNP allele but unrelated to eye pigmentation was in the population that entered the New World some 15,000 years ago, and became common in the ancestors of South American populations.
Understanding the history of human movements helps us to uncover the genetic causes of traits. In this case, the SNP allele reflects two different histories in western Eurasia and in the New World.