The present study was aimed to evaluate the reduction in fermentative losses and the improvement of aerobic stability of maize silage treated with Lactobacillus buchneri bacteria, antifungal natamycin and a combination of L. buchneri and natamycin. The study was completely randomized using...

Additives and inoculants are the focus of silage research, as new additives and their combinations can improve fermentation and silage quality. We performed two trials evaluating two doses (4 and 8 g/t, wet weight) of natamycin as an additive to maize silage compared to a control without...

Virtual fencing (VF) is a promising technology for pasture management, but its successful implementation depends on understanding the factors that influence the animal learning. The aim of the present study was to evaluate the significance of individual animal factors that affect the behavioural and performance-based learning process in cattle during a VF training period. We analysed collar-derived performance data from 171 mixed-age Angus and Salers breed cattle over a 12-day training trial. Learning performance was primarily measured by daily success ratio (SR) and confidence ratio (CR). Learning trajectories and the influence of animal factors were analysed using linear mixed-effects models. A significant increase was observed in the SR over the 12-day period (p < 0.0001). A significant interaction between training day and age class (p < 0.001) revealed distinct learning strategies: younger animals showed rapid initial adaptation but lower consistency, whereas adult animals exhibited a more stable final performance. Our findings suggest that breed may also influence the learning rate to virtual fencing (p < 0.001). Individual behaviour accounted for a substantial portion of performance variation (p < 0.0001). Recognising these different individual and social learning pathways is essential for developing more effective and improved animal welfare VF training protocols.

Automated cattle posture-classification systems frequently report near-perfect accuracy, yet their robustness under realistic deployment conditions remains largely unknown. In particular, it is unclear whether multimodal sensor fusion improves generalisation or leads models to rely on context-specific signals that fail under distribution shift. Here, we evaluate the robustness of automated posture classification (lying versus standing) using collar accelerometers, rumen-bolus sensors, and environmental measurements collected from a pasture-based beef cattle herd across two consecutive years (2024-2025). XGBoost served as the primary model, with Logistic Regression, Random Forest, and Long Short-Term Memory (LSTM) networks evaluated as comparative baselines. Model robustness was assessed under progressively more stringent evaluation protocols, ranging from conventional random train–test splits to leave-one-animal-out validation and cross-year evaluation on an independent cohort of previously unseen animals recorded one year later. While multimodal models achieved strong within-year performance (macro-F1 = 0.94), the performance declined substantially under cross-year evaluation (macro-F1 = 0.49). Unexpectedly, the collar-only model outperformed all multimodal configurations under temporal shift (macro-F1 = 0.54), indicating that additional physiological and environmental information did not improve robustness. Explainability analysis revealed persistent reliance on rumen-bolus activity and environmental variables even when predictive performance deteriorated, a pattern consistent with shortcut learning in which models exploit context-dependent signals that do not remain reliable under covariate shift. Distribution-shift diagnostics further confirmed substantial differences in the feature distributions between recording years. Our findings demonstrate that commonly used evaluation protocols can substantially overestimate real-world performance and that multimodal sensor fusion may reduce, rather than improve, robustness under temporal distribution shift. More broadly, the results highlight that benchmark accuracy alone is insufficient to assess deployment readiness and underscore the need for robustness-centred evaluation in livestock-monitoring research.Together, our results highlight the importance of evaluating livestock-monitoring systems under realistic deployment conditions and demonstrate how explainability analysis can reveal hidden model dependencies and provide insight into model behaviour.