Comparison of two geo-evolutionary analysis methods using local and cross-border bovine breeds

Abstract

New clustering algorithms have emerged in population genetics and molecular ecology areas as important tools to infer the genetic structure of a given population. Currently, methodologies based on Bayesian models are used for this purpose. These methodologies consider the geographical area, whereas other studies explore the use of geographic coordinates based in Voronoi mosaics. The aim of this study was to use a multi-locus genotype database of bovine populations to compare the populations with two programmes that based their analysis on Bayesian models with different focuses. A database of 1053 animals from 33 bovine populations was analysed (native, Creole and cross-border breeds) using 27 microsatellite markers. The STRUCTURE® version 2.2.3 software was selected for the identification and calculation of the proportion of the mixture of individuals within each population as recommended by the software using different K calculations (n = 2, 3, …, 33). Additionally, TESS®, which uses the Bayesian model with clustering algorithms, was used for the analysis. We concluded that both models showed different results; the discrepancies probably occurred because some of the studied populations might have shown the panmixia principle, thereby altering the final result. This study draws attention to the selection of the most appropriate analysis software for different populations with different selection and mating models based on basic genetic principles, such as the Hardy–Weinberg and linkage disequilibrium, because the results can be misinterpreted, leading to incorrect conclusions. Population behaviour, including reproduction and genetic origins, should be taken into account before using a given quantitative analysis method.