Genetic diversity of Italian goat breeds assessed with a medium-density SNP chip

Abstract

Background: Among the European countries, Italy counts the largest number of local goat breeds. Thanks to the recent availability of a medium-density SNP (single nucleotide polymorphism) chip for goat, the genetic diversity of Italian goat populations was characterized by genotyping samples from 14 Italian goat breeds that originate from different geographical areas with more than 50 000 SNPs evenly distributed on the genome.

Results: Analysis of the genotyping data revealed high levels of genetic polymorphism and an underlying North-south geographic pattern of genetic diversity that was highlighted by both the first dimension of the multi-dimensional scaling plot and the Neighbour network reconstruction. We observed a moderate and weak population structure in Northern and Central-Southern breeds, respectively, with pairwise FST values between breeds ranging from 0.013 to 0.164 and 7.49 % of the total variance assigned to the between-breed level. Only 2.11 % of the variance explained the clustering of breeds into geographical groups (Northern, Central and Southern Italy and Islands).

Conclusions: Our results indicate that the present-day genetic diversity of Italian goat populations was shaped by the combined effects of drift, presence or lack of gene flow and, to some extent, by the consequences of traditional management systems and recent demographic history. Our findings may constitute the starting point for the development of marker-assisted approaches, to better address future breeding and management policies in a species that is particularly relevant for the medium- and long-term sustainability of marginal regions.

Fig. 1

Geographic origin of the analyzed Italian goat breeds

Fig. 2

Neighbor-network based on pairwise F_ST genetic distances between breeds

Fig. 3

Multidimensional-scaling plot. Multidimensional-scaling plot of distances based on a genomic kinship matrix. The axes corresponding to first (abscissa, variance explained: 6.29 %) vs. second dimension (ordinate, variance explained: 3.88 %) are shown

Fig. 4

Bayesian clustering performed with ADMIXTURE software on goat genotyping data. Assignment of single individuals (thin vertical bars) to the different clusters when K = 2 and K = 11 hypothetical populations are assumed. Different colours identify different clusters. The reconstruction at K = 11 had the smallest cross-validation error [See Additional file 4 Figure S2]