Miroslav Kapš

Genome-wide mapping and estimation of inbreeding depression of semen quality traits in a cattle population

Inbreeding depression is known to affect quantitative traits such as male fertility and sperm quality, but the genetic basis for these associations is poorly understood. Most studies have been limited to examining how pedigree- or marker-derived genome-wide autozygosity is associated with quantitative phenotypes. In this study, we analyzed possible associations of genetic features of inbreeding depression with percentage of live spermatozoa and total number of spermatozoa in 19,720 ejaculates obtained from 554 Austrian Fleckvieh bulls during routine artificial insemination programs. Genome-wide inbreeding depression was estimated and genomic regions contributing to inbreeding depression were mapped. Inbreeding depression did affect total number of spermatozoa, and such depression was predicted by pedigree-based inbreeding levels and genome-wide inbreeding levels based on runs of homozygosity (ROH). Genome-wide inbreeding depression did not seem to affect percentage of live spermatozoa. A model incorporating genetic effects of the bull, environmental factors, and additive genetic and ROH status effects of individual single-nucleotide polymorphisms revealed genomic regions significantly associated with ROH status for total number of spermatozoa (4 regions) or percentage of live spermatozoa (5 regions). All but one region contains genes related to spermatogenesis and sperm morphology. These genomic regions contain genes affecting sperm morphogenesis and efficacy. The results highlight that next-generation sequencing may help explain some of the genetic factors contributing to inbreeding depression of sperm quality traits in Fleckvieh bulls.

Modeling variance structure of body shape traits of Lipizzan horses

Heterogeneity of variance of growth traits over age is a common issue in estimating genetic parameters and is addressed in this study by selecting appropriate variance structure models for additive genetic and environmental variances. Modeling and partitioning those variances connected with analyzing small data sets were demonstrated on Lipizzan horses. The following traits were analyzed: withers height, chest girth, and cannon bone circumference. The measurements were taken at birth, and at approximately 6, 12, 24, and 36 mo of age of 660 Lipizzan horses born in Croatia between 1948 and 2000. The corresponding pedigree file consisted of 1,458 horses. Sex, age of dam, and stud-year-season interaction were considered fixed effects; additive genetic and permanent environment effects were defined as random. Linear adjustments of age at measuring were done within measuring groups. Maternal effects were included only for measurements taken at birth and at 6 mo. Additive genetic variance structures were modeled by using uniform structures or structures based on polynomial random regression. Environmental variance structures were modeled by using one of the following models: unstructured, exponential, Gaussian, or combinations of identity or diagonal with structures based on polynomial random regression. The parameters were estimated by using REML. Comparison and fits of the models were assessed by using Akaike and Bayesian information criteria, and by checking graphically the adequacy of the shape of the overall (phenotypic) and component (additive genetic and environmental) variance functions. The best overall fit was obtained from models with unstructured error variance. Compared with the model with uniform additive genetic variance, models with structures based on random regression only slightly improved overall fit. Exponential and Gaussian models were generally not suitable because they do not accommodate adequately heterogeneity of variance. Using the unstructured error variance model, the heritability estimates ranged from 0.17 to 0.33 for withers height, 0.07 to 0.27 for chest girth, and 0.14 to 0.30 for cannon bone circumference. This study demonstrated the necessity of accounting for heterogeneity of variances and covariances for body shape traits in Lipizzan horses, and possible difficulties in estimating variance and covariance components when applying more complicated structure models on a small data set. The choice of models depends not only on overall fit but also on the fit of genetic and environmental components.