Estimation of Inbreeding and Substructure Levels in African-Derived Brazilian Quilombo Populations

Abstract

This article deals with the estimation of inbreeding and substructure levels in a set of 10 (later regrouped as eight) African-derived quilombo communities from the Ribeira River Valley in the southern portion of the state of São Paulo, Brazil. Inbreeding levels were assessed through F-values estimated from the direct analysis of genealogical data and from the statistical analysis of a large set of 30 molecular markers. The levels of population substructure found were modest, as was the degree of inbreeding: in the set of all communities considered together, F-values were 0.00136 and 0.00248 when using raw and corrected data from their complete genealogical structures, respectively, and 0.022 and 0.036 when using the information taken from the statistical analysis of all 30 loci and of 14 single-nucleotide polymorphic loci, respectively. The overall frequency of consanguineous marriages in the set of all communities considered together was ∼ 2%. Although modest, the values of the estimated parameters are much larger than those obtained for the overall Brazilian population and in general much smaller than the ones recorded for other Brazilian isolates. To circumvent problems related to heterogeneous sampling and virtual absence of reliable records of biological relationships, we had to develop or adapt several methods for making valid estimates of the prescribed parameters.

Inbreeding estimates in human populations: Applying new approaches to an admixed Brazilian isolate

“…The present study was performed in an admixed Brazilian isolate located in the Ribeira River Valley, in the southern part of the State of São Paulo, Brazil ( Fig 1 ). This isolate, known in Brazil as a quilombo , was founded around 1890 by runaway, abandoned and freed slaves (some of them being the admixed offspring of white farm owners and African female slaves) and a few pure or mixed native Americans, who created small rural settlements in isolated areas inside the Atlantic rainforest for several generations (other details of interest on the quilombo population structure and demography are described elsewhere [ 1 , 19 , 21 ]). The isolate aggregates twelve communities that were treated as a single one, since the degree of differentiation among its communities is very low, with F ST indices generally smaller than 0.05 [ 1 ].…”

“…In the long run, one expects that the estimates of f k thus obtained should be normally distributed around the average value of , with the limits of the usual 95% confidence interval being given approximately by , where var(f) is given by with x k as defined in formula (3) [ 1 ].…”

“…Measures of population inbreeding levels have been traditionally obtained from the direct genotyping of population samples followed by the estimation of heterozygous frequency deviations from the proportions expected under random-mating assumptions (HW expectations) or from the analysis of sets of individual or grouped genealogies (v.g., Lemes et al . [ 1 ]). The inbreeding coefficients F estimated from the two methods are however quantitatively and qualitatively different, since the first one (Wright's fixation index F IS ) estimates specifically systematic inbreeding or consanguinity, while the second (Wright's fixation index F IT ) measures the amount of total inbreeding, including the fraction due to small population effective number Ne .…”

“…The inbreeding coefficients F estimated from the two methods are however quantitatively and qualitatively different, since the first one (Wright's fixation index F IS ) estimates specifically systematic inbreeding or consanguinity, while the second (Wright's fixation index F IT ) measures the amount of total inbreeding, including the fraction due to small population effective number Ne . In humans, good examples of the usefulness of deep genealogies to measure inbreeding coefficients are the study of the isolated African-derived community of Valongo in Brazil [ 1 , 2 ] and the research on royal inbreeding [ 3 , 4 ]. Ceballos and Alvarez [ 4 ] study showed that it is possible to capture 95% of the actual inbreeding coefficient with a pedigree of at least 10 generations depth.…”

HLA imputation in an admixed population: An assessment of the 1000 Genomes data as a training set

“…As a consequence, there is a substantial degree of relatedness between individuals [18]. Using approximately 500.000 genomewide SNPs from the Affymetrix Axiom Human Origins Array platform, we estimated the kinship coefficient between all pairs of Quilombo samples based on the Moment Method implemented in SNPRelate R Package [19].…”