Trait development and genetic parameters of resilience indicators based on variability in milk consumption recorded by automated milk feeders in North American Holstein calves

The implementation of automated milk feeders (AMF) on precision dairy farms has enabled efficient management of large numbers of group-housed replacement calves with reduced labor requirements and improved calf welfare. In this study, we investigated the feasibility of deriving calf resilience indicators based on variability in milk consumption using data from 10,076 North American Holstein calves collected between 2015 and 2021. We modeled and evaluated deviations in observed and predicted daily milk consumption trajectories as indicators of resilience to environmental perturbations. We also analyzed average milk intake and the number of treatments for bovine respiratory disease (BRD) and their genetic correlations with the derived resilience parameters. Milk consumption was recorded using the Förster-Technik AMF. Deviations in cumulative milk intake were modeled using various methods, including quantile regression and the Gompertz function. Ten resilience indicators were derived to quantify the degree and duration of perturbations, including amplitude, perturbation time, recovery time, and deviation velocities. After data editing, genomic data from 9,273 calves and pedigree information from 10,076 calves with 321,388 phenotypic records were used to estimate genetic parameters for 12 traits, including 10 calf resilience indicators as well as average milk intake and treatments for BRD. Substantial phenotypic variability was observed for all calf resilience indicators derived and genetic parameters related to these novel resilience indicators were estimated. The heritability estimates for the resilience traits are as follows: amplitude of the deviation (in L) 0.047 (0.032, 0.064; HPD interval), perturbation time of deviation (in d) 0.011 (0.0056, 0.016), recovery time of the deviation (in d) 0.025 (0.016, 0.035), maximum velocity of perturbation (L/d) 0.039 (0.024, 0.053), average velocity of perturbation (L/d) 0.038 (0.022, 0.050), area between the curves (L × d) 0.039 (0.027, 0.054), recovery ratio 0.053 (0.036, 0.072), deviation variance 0.049 (0.32, 0.068), log-deviation variance 0.027 (0.016, 0.044), deviation autocorrelation 0.010 (0.0042, 0.017) and number of deviation occurrences 0.023 (0.0094, 0.036). Some of the highlighted genetic correlations observed with average milk consumption include amplitude: 0.569 (0.474, 0.666), perturbation time: -0.534 (-0.73, -0.342), and average velocity: 0.554 (0.432, 0.672). Similarly, the genetic correlations between the number of times treated for BRD with perturbation time was 0.494 (0.251, 0.723), -0.294 (-0.52, -0.095) with number of deviations, and 0.348 (0.131, 0.578) with deviation autocorrelation. This study highlights the genetic influence on various resilience traits in calves, including amplitude, perturbation time, recovery time, and velocity measures of the perturbation. Our findings suggest the need for prioritizing genetic selection based on traits such as recovery time, which exhibits higher heritability and a moderate genetic correlation with the number of times a calf is treated for BRD. The combination of AMF data, mathematical modeling, and genomic evaluation provides a comprehensive framework for assessing and breeding more resilient dairy calves in the face of environmental and health challenges.