RSC Adv. 2026 Aug 7. doi: 10.1039/d6ra03455a. Online ahead of print.
ABSTRACT
Welfare assessment of housed dairy cattle currently relies on subjective, snapshot-like observations that fail to capture the physiological impacts of the inhabited environment. Metabolomic profiling offers potential for objective welfare assessment, but its application to behavioural interventions remains unexplored. Here, we show that provision of outdoor access to housed dairy cows produces detectable changes in milk metabolite levels and associated pathway signatures, with effects impacted by exposure duration. Using untargeted metabolomics via direct infusion mass spectrometry on 371 samples across five timepoints, we applied univariate analysis and pathway analysis for the identification of fundamental physiological changes. Further analyses comprised stability explorations of machine learning methods alongside Cox-Battey selection, to aid the triangulation of high specificity detection of mass ions. Univariate analysis of the 14 day intervention period (S2-S4) identified 10 statistically significant mass ions (9 decreased, 1 increased; adjusted p < 0.05, FC > 2.0), while no significant ions emerged at early (S2-S3) or post-intervention (S4-S5) timepoints. Functional pathway analysis revealed impacted steroid hormone biosynthesis, pyruvate metabolism, and riboflavin metabolism during intervention, with glycerophospholipid metabolism and branched-chain amino acid pathways altered post-intervention. Machine learning identified discriminant features at all timepoints, with only m/z 621.2404 achieving both univariate significance and machine learning selection. These findings suggest that metabolomic profiling can detect welfare-relevant physiological adaptations to the environment provided to housed dairy cows, with convergent multi-method signals providing the most robust discriminant features for advancing dairy welfare assessment beyond subjective observation.
PMID:42569159 | PMC:PMC13448961 | DOI:10.1039/d6ra03455a
