The anthropological utility of genetic data in small -scale populations: Migration rates and patterns among the Yanomamo.

Anthropological questions are difficult to address using conventional population genetic approaches. In this thesis, I illustrate this problem using F statistics to estimate migration rates and patterns among the Yanomamo, an indigenous South America population. I show that Fst fails to produce meaningful estimates of migration parameters because this statistic fails to consider the complex ways that individual and group-level human behaviors affect genetic diversity. Given this problem, I utilize three alternative non-conventional methods to estimate migration parameters from genetic and related data. The first two, one pedigree-based and the other multilocus genetic, are relatively insensitive to the effects of human behavior. They produce migration rate estimates between 3--29% per generation and indicate that both males and females move frequently between villages. However, both methods still ignore important aspects of behavior and produce only short-term estimates of migration. I therefore develop a third, computer simulation method that is capable of incorporating unlimited behavioral complexity at any time depth. I implement this approach by simulating, over multiple generations, the effects of fixed demographic and varying reproductive behavior on the estimation of male and female migration rates. I show that, for ethnographically documented demographic conditions, and for the observed pattern of genetic variation, migration rate estimates increase as a function of increased variance in reproductive success for both males and females. Depending on the degree of reproductive variance, I estimate migration rates between 0.3--37% per generation. I show that it is impossible to determine whether male or female migration is higher unless the degree of reproductive variance is known. This thesis thus demonstrates that human behaviors affect genetic variation in important ways. Because it incorporates realistic behavioral complexity over any time depth, I conclude that the simulation approach is optimal for measuring migration rates and patterns from genetic data. Though different behaviors may be important at other levels of population structure, I argue that this conclusion holds for any parameter measured from genetic data at any level of population structure.Ph.D.Physical anthropologySocial SciencesUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttp://deepblue.lib.umich.edu/bitstream/2027.42/132089/2/3057969.pd