Unveiling the genetic structure of Costa Rican bovines through genetic admixture models

Introduction. The cattle population in Costa Rica is highly diverse due to the variety of production systems, ongoing crossbreeding, and the increase in genetic imports. Under these circumstances, admixture models provide a more suitable approach to describe the genetic composition of this population than phylogenetic trees based on predefined populations. Objective. To model the genetic structure of the Costa Rican cattle population by means of supervised and unsupervised genetic admixture models. Materials and methods. Hair samples from 1412 randomly selected bovines from 744 herds across 8 regions of Costa Rica were collected in 2015 and genotyped for 18 microsatellite markers. Two approaches that make use of admixture genetic models were compared: an unsupervised scenario, based exclusively on genotype data; and a supervised scenario, which relied on genetic data assisted by prior information on phenotype and production purpose. Results. Analysis of genetic data under both scenarios provided similar results when the number of clusters (K) was lower than five, although estimates from the supervised model were more homogeneous with lower standard deviations. A priori defined subpopulations were distributed consistently among clusters in both scenarios. The most probable subpopulation clustering was observed at K = 3, which mainly separated Bos indicus breeds, Jersey and other Bos taurus breeds. Breed types clustered concordantly with breed clustering, shedding light on the genetic structure of the population. Conclusions. The combination of admixture genetic models under a supervised approach yielded the most consistent results, which reveals the importance of considering genetic-environmental interrelationships to achieve a more precise description of the genetic structure of the Costa Rican cattle population.