HLA polymorphism and evaluation of European, African, and Amerindian contribution to the white and mulatto populations from Parana, Brazil

Human Biology, Aug 2000 by Probst, C M, Bompeixe, E P, Pereira, N F, de O Dalalio, M M, Et al

where f^sub alix^ and f^sub aliy^ are the frequencies of allele i in populations x and y, and m is the number of alleles analyzed. This measure of distance is useful to summarize information about variation of gene frequencies among populations (Meyer 1995). The populations considered were Sao Paulo city white (Rosales et al. 1992); white Brazilian, Portuguese, Italian, Spanish, German, Polish, Ukrainian, French, British, North African, West African, South Af rican, Senegalese, Zimbabwean, Zairian, African American, African Brazilian, Brazilian Amerindian (Imanishi et al. 1992a); Dutch (Schipper et al. 1996); Shona (the main tribe of Zimbabwe), Xhosa, and Zulu (the main South African tribes) (Hammond et al. 1997); Guarani and Kaingang (Petzl-Erler et al. 1993).

The computer program PHYLIP (Felsenstein 1989) was used for phylogenetic analysis by the UPGMA method. The program TREEVIEW (Page 1996) was used for construction of a dendrogram.

The degree of admixture was calculated by Chakraborty's method of gene identity (Chakraborty 1975) using the ADMIX program. HLA-A, -B and -C allele frequencies of European, African, and Amerindian populations were obtained from the literature. Since Parana's population is heterogeneous regarding parental populations in each main racial group, selection of individual populations considered history, surnames of individuals sampled, and genetic distance between parental populations. The European populations selected were Portuguese, Italian, German, and Polish (Imanishi et al. 1992a). The African populations were from South Africa (Hammond et al. 1997), Zimbabwe, Zaire, and West Africa (Imanishi et al. 1992a), all but the last being Bantu-speaking. There is no HLA study on the black populations that were most important for Parana's colonization (see the discussion at the beginning of this paper), and the populations utilized are from geographic areas close to Angola and Mozambique. Amerindian populations were the Kaingang and Guarani (Petzl-Erler et al. 1993).

To estimate the degree of genetic admixture, the relative contribution of each of the several founding populations from the 3 continents (Europe, Africa, and America) should be taken into account, because this contribution was not equal (again, see the introductory discussion). However, the relative contribution of each parental population is not exactly known. We developed an approach to minimize the error resulting from this uncertainty. Not just one but many estimates of admixture were generated. Several putative parental (European, African, and Amerindian) groups were produced, using a series of weighted means of the allele frequencies of the founding populations in each one of the 3 groups. The ranges of contribution (Table 1) were estimated on the basis of historical population size, as a first parameter, and surnames of individuals sampled, as a second parameter. This resulted in 6 European, 4 African, and 3 Amerindian different virtual parental groups. Chakraborty's method was applied to all 72 possible combinations of these groups, providing a distribution of admixture rates. The distribution average was then assumed to represent the degree of admixture. This procedure gives a better picture of admixture, providing a view of tendencies, if one or the other population contributed more to the resulting admixed population. Also, the standard deviation of the admixture distribution shows the degree of genetic heterogeneity inside each of the 3 groups of parental populations. This method reveals the groups for which changes of the proportions of contributing populations would have a more noticeable impact on the admixture estimate.