Indicators of resilience in cattle based on test-day milk yield: comparison of within- and between-lactation and their relationship with milk production

Measuring variability in target phenotypes is a recognised strategy for assessing resilience, particularly in dairy cattle, where fluctuations in daily milk yield can be monitored using automated milking systems. However, these data are not available for all animals or populations, limiting the integration of current resilience indicators into routine selection indices. This underscores the need for alternative approaches based on more widely accessible data sources. The present study aimed to develop and evaluate novel resilience indicators derived from routinely collected test-day milk yield records in the Rendena breed, a local cattle population. Ten resilience indicators were proposed, categorised into two groups: WITHIN indicators, assessing milk yield variability within single and combined lactations (1 to 3), and BTW indicators, measuring fluctuations between lactations (1 to 2; 1 to 3; 1 to 4) based on standardised 305-day yields (BTWtot) and peak daily yields (BTWmax). The indicators were evaluated for heritability, genetic correlations, and associations with production, fitness, and milk composition traits. Structural equation models were also used to account for the confounding effects of milk production. All indicators showed heritable variation, with estimates ranging from 0.095 to 0.324. Strong genetic correlations (0.62–0.99) were observed among indicators, especially within the WITHIN group. Most indicators were positively correlated with milk production and negatively with fitness traits, including fertility and longevity. Milk solids showed weak or no associations, except for protein percentage, which was positively correlated with several indicators, particularly WITHIN types. Among all, vardmy2 (milk yield variance within the second lactation) stood out due to its moderate heritability (0.228), early availability, and strong negative genetic correlations with fitness traits. Structural equation model analyses confirmed these associations were not confounded by milk yield, supporting the biological relevance of milk variability as a resilience indicator. Specifically, cows with higher yield variability were genetically less fit, showing poorer fertility, reduced longevity, and higher somatic cell scores. In conclusion, resilience indicators based on test-day milk data, particularly those assessing within-lactation variability, are heritable and biologically meaningful. They can be integrated into breeding programmes to support the selection of more resilient and healthier cows, even in the absence of automated milking system data.