Structural equation modeling as a tool to interpret genotype-environment interaction and genome-wide association (GWAS) for growth and reproductive traits in Nellore heifers using whole-genome sequence data

Abstract

Practices that result in improved reproductive and growth efficiencies have been recognized as an important factor affecting the economic sustainability of beef cattle herds by reducing production costs and generation intervals. However, in the tropical beef cattle production systems, Nellore heifers are raised mainly on pasture under a wide range of topographic and climatic regions with seasonal variations in forage production and nutritional levels leading to genotype-environment (GxE) interaction. In this context, a better understanding of the genetic background for animal adaptation can aid in designing breeding programs in harsh conditions. Hence, applying whole-genome sequencing in genome-wide association studies (GWAS) combined with Structural Equation Models (SEM) represents a great opportunity to clarify the main mechanism and causal links between growth and reproductive traits in harsh environmental conditions (EC). Hence, this research project aims to include whole-genome sequencing to unravel the genomic sensitivity in Nellore heifers raised in different EC and apply the Bayesian network to infer the interrelationships among growth and reproductive traits in Nellore heifers. In addition, we will combine the GWAS with graphical modeling (SEM) to represent the interrelationships among SNP markers and traits in the form of equations and path diagrams in harsh EC. For this, phenotypic information for traits related to reproduction and growth of approximately 550,000 Nellore heifers belonging to commercial breeding programs, widely distributed in the Midwest, Southeast, and Northeast of Brazil, will be used. The traits related to growth traits will be: bodyweight at yearling (BWY), average daily gain from weaning to yearling (ADGWY), and reproductive performance in heifers will be: age at first calving (AFC), heifer early pregnancy (HP), heifer rebreeding (HR) and days to second calving (DSC). Genotypic data from 5,550 heifers, 6,745 dams, and 35,500 sires genotyped with BovineHD BeadChip (770k) will be imputed to the whole-genomic sequencing using FImpute. The imputation will be carried out using the DNA sequence of 152 bulls with high genetic connectivity with the population genotyped at 770k. A two-step reaction norm model will be used to assess the animal's sensitivity to environmental changes. In the first step, the environmental conditions will be defined according to the contemporary group (CG) solution for BWY and ADGWY traits and then condensed using the principal component (PC) analyses to summarize into a common EC descriptor. Finally, this EC descriptor will be standardized to show a mean value of 0 and a standard deviation (sd) of 1. In the second step, a reaction norm model will be used to estimate the animals' sensitivity to EC for growth and reproduction traits. Thus, the genomic breeding values (GEBV) will be estimated for three EC named Low (-3.0 sd), Medium (0.0 sd), and High (3.0 sd). The GEBV estimates in three different EC levels will be used to detect the causal networks among traits and EC levels using a Bayesian network approach. These networks for traits related to growth and reproduction in three EC levels will be used with prior information to perform the SEM-GWAS considering the whole-genome sequencing. The insights acquired with the present research will provide new knowledge into animal adaptation's genetic basis, leading to a better understanding of the processes and relationships that affect animal adaptability for traits of economic importance. In addition, these results could be useful for herd management activities designed to produce specific outcomes and desired results within the limits of environmental changes.