Multivariate structure of human populations: Sedentary versus migrants

The rate change of gene frequency in a population subject to emigration obviously depends on differences in the effective reproductive size (and resulting random genetic drift effects) between emigrants and natives. An important additional force may be the different selection pressures of the original and the new environment into which the population penetrates. Discrete traits and monolocus systems have been studied in many natural populations of various species. However, knowledge about the migration influence on quantitative, i.e. polygenic, characters is very limited. The present study set out to answer the following biological questions: (1) Does migration induce changes in sets of phenotypic, genetic and environmental correlations? (2) If so, are these changes expressed in levels and/or structure of the correlation matrices? Data on 20 anthropometric traits in 305 Mexican families [129 families from Mexico (sedentary population) and 176 families living in Texas (migrant population)] were used for analysis. The curves of distribution and average values of phenotypic, genetic and environmental correlations remained unchanged between the two populations. However, qualitatively (i.e., as far as the agreement between matrix compositions is concerned), all three matrices changed significantly. The phenotypic correlations appear to be the most highly canalized, the correspondence between the two matrices being 62.1 %. The environmental matrices had the highest variation, and although 26.3 % of the correlations were in agreement, this was statistically nonsignificant. The most important finding in the present study was the relatively low correspondence between the two genetic matrices (35.6%). We suggest that these changes were provoked by preselection (i.e., by a nonrandom sample) of migraiting individuals.