Abstract
Expected changes in climate warrant research on selection for a phenotypically stable cattle population that can perform consistently across diverse environmental conditions. In this study, we utilize a heteroscedastic random regression model to 1) characterize the additive genetic and other phenotypic components of dry matter intake (DMI) and respiration rate (RR) concerning the temperature humidity index (THI), 2) assess the presence of genotype-by-environment interactions (G × E) by determining whether the additive genetic reaction norm changes along the observed THI range and by evaluating the additive genetic correlations between DMI or RR at different THI values, and 3) evaluate model-derived accuracy of estimated breeding values (EBV) along a range of THI. Data consisted of repeated observations of DMI and RR on 788 and 569 steers, respectively, over a period of 70 d. A hierarchical model with subject-specific additive genetic and permanent environment effects was fitted to each trait using Bayesian inference. Estimated population slopes, expressed as posterior median and 95% highest posterior density (HPD) interval, were -0.046 (-0.053, -0.039) kg DMI per unit increase in THI per day and 0.027 (0.026, 0.029) breaths per 30s (BP30S) RR for each unit increase in THI on the logarithmic scale, thereby suggesting environmental sensitivity for both traits. Estimated correlations between the additive genetic intercept and slope were -0.78 (-0.86, -0.69) and -0.66 (-1.0, -0.20) for DMI and RR, respectively, indicating that selection for increased DMI and decreased RR at the onset of heat stress can be expected to associate positively with mean population environmental sensitivity to THI. Heritability estimates for DMI at the onset of heat stress (i.e., THI of 70) ranged from 0.30 (0.17, 0.44) to 0.37 (0.20, 0.48) across cohorts but decreased as THI increased. Heritability estimates for RR were low, with 95% HPD upper boundaries ranging from 0.03 to 0.08 across the range of THI evaluated. For DMI, the median additive genetic correlation between 70 and 85 THI and the Spearman correlations between EBV at 70 and 85 THI levels were 0.42 (0.26, 0.57) and 0.39 (0.26, 0.52), respectively, indicating substantial G × E. The median DMI EBV model-derived accuracy at a specific THI value increased from 70 THI to 0.65 at 82 THI, at which point it stabilized. This was likely because more data points were gathered at greater THI.