Genomic reaction norms for tick resistance in Hereford and Braford beef cattle

• Ano de defesa: 2015
• Autor(a) principal: Mota, Rodrigo Reis
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Universidade Federal de Viçosa
• Palavras-chave em Português: Bovino de corte; Interação genótipo-ambiente; Variação (Genética); Carrapato; Genética e Melhoramento dos Animais Domésticos
• Link de acesso: http://www.locus.ufv.br/handle/123456789/6847

The cattle tick is a parasite that adversely affects livestock performance in tropical areas. Although countries such as Australia and Brazil have provided genetic evaluations for tick resistance, these evaluations have not typically considered genotype by environment interaction (G*E); hence genetic gains could be adversely affected as breedstock comparisons are environmentally- dependent in the presence of G*E, particularly if residual variability is also heterogeneous across environments. The objective of this study was to investigate the existence of G*E based on various models with different assumptions on genetic and residual variability. Data were collected by the Delta G Connection improvement program including 10,673 tick count phenotypes on 4,363 animals. Nine models including two traditional animal models (AM) and seven different hierarchical Bayesian reaction norm models (HBRNM) were investigated. One-step and two-step modeling approaches were used to infer upon G*E. Model choice was based on the deviance criterion information (DIC). The best-fitting model specified heterogeneous residual variances across 10 subclasses as delimited by every decile of the contemporary group estimates of tick count effects. One-step models generally had the highest estimated genetic variances. Estimates of heritabilities were generally higher for HBRNM than AM. Furthermore, one- step models based on heterogeneous residual variances also generally lead to higher heritability estimates, especially in harsh environments. Estimates of repeatability varied along the environmental gradient (range 0.18-0.45) implying that the relative importance of additive and permanent environment effects for tick resistance is environmentally influenced. The posterior means of the genetic correlations across environmental tick infestation surface plot demonstrated a large plateau above 0.80. HBRNM represent powerful tools to infer G*E and account for their effects for genetic evaluations of tick resistance. Additional increases in accuracies on estimated breeding values are also expected based on HBRNM analyses that additionally consider heterogeneity of residual variances across environments. In a second study, we incorporated marker information to compare a conventional genomic- based single step BLUP model with its one-step genomic reaction norm model extension on tick infestation phenotypes and to compare the performance of genomic estimates breeding values (GEBV) predictions obtained from using only phenotypes and phenotypes plus marker information. Four different models were tested: two conventional animal models, and two one-step reaction norm model with and without genomics. The non reaction norm models seem to be poorer fitting in comparison with its one-step extensions. The reaction norm model including marker information presented lower intercept and slope genetic variance estimates in comparison with the models that included the pedigree-based relationship matrix. Heritability and repeatability estimates were, in general, similar for both models and ranged over the environmental gradient (EG) from 0.07 to 0.46 and from 0.20 to 0.60, respectively. Genetic correlations were remarkably low between extreme EG, indicating the presence of G*E for tick resistance. Cross validation estimates were in average 0.66±0.02, 0.67±0.02, 0.67±0.02 and 0.66±0.02 for BLUP, GBLUP, GLRNM and LRNM, respectively, based on K-means partitioning, whereas GLRNM was 0.71±0.01 and tend to better than BLUP (0.67±0.01), GBLUP (0.70±0.01) and LRNM (0.70±0.01) based on random partitioning. However, no statistical significance was reported between GLRNM and LRNM. Our results also suggest that marker information do not lead for higher prediction accuracies which decreased as the tick infestation level increased and as the relationship between animals in training and validation datasets decreased. In third and last study, was aimed to perform genome-enabled predictions for tick resistance in Hereford and Braford cattle by using single step genomic BLUP methodology (ssGBLUP), to estimate marker effects from reaction norms associated with tick resistance as well as to identify candidate genes derived from the most relevant SNP markers. A one-step reaction norm model was fitted to estimate the (co)variance components and genetic parameters. To study SNP effects across different tick infestation (TI) levels, we identified the top 1% of SNPs in each TI and pointed out to the similarity between these markers across the levels. The additive genetic and permanent environment effects showed significant slope confirming the presence of G*E. Correlations between intercept and slope were positive with high (0.52±0.18) and moderate (0.26±0.15) magnitude for genetic and permanent environment effects, respectively. From the top 1% SNPs (410), 75 were consistently relevant across TI and indicated SNP by environment interaction. The most relevant SNPs were located on chromosomes 1, 2, 6, 7, 9, 11, 14, 21 and 23 and the annotated genes closest these markers showed functions related to energy metabolism, retinal pigment epithelium, maintenance and integrity of the photoreceptor cells, and cell differentiation.