Gender and population history: sex bias revealed by studying genetic admixture of Ngazidja population (Comoro Archipelago).

The peopling of Comoro Archipelago is defined by successive waves of migration from three main areas: the East African Coast (Bantu-speaking populations), the Persia and Arabian Peninsula, and Southeast Asia (especially Indonesia). It follows an apparent classic trihybrid admixture model. To better understand the Comorian population admixture dynamics, we analyzed the contributions of these three historical parental components to its genetic pool. To enhance accuracy and reliability, we used both classical and molecular markers. Samples consist of published data: blood group frequencies, 14 KIR genes, 19 mitochondrial DNA SNPs (to highlight female migrations), 14 Y chromosome SNPs (male migrations). We revealed distinct admixture patterns for autosomal and uniparental markers. KIR gene frequencies had never been used to estimate admixture rates, this being a first assessment of their informative power in admixture studies. To avoid major methodological and statistical bias, we determined admixture coefficients through nine well-tried estimators and their associated software programs (ADMIX95, ADMIX, admix 2.0, LEA, LEADMIX, and Mistura). Results from mtDNA and Y chromosome markers point to an important sex-bias in the admixture event. The original Bantu gene pool received a predominant male-mediated contribution from the Arabian Peninsula and Persia, and a female-mediated contribution from Southeast Asia. Admixture rates estimated from autosomal KIR gene markers point also to an unexpected elevated Austronesian contribution.