Precision Livestock Farming (PLF) applications to the dairy sector

In the modern livestock sector, several applications of Precision Livestock Farming (PLF) have been designed for improving efficiency, productivity, sustainability, and welfare of different farming systems. Precision dairy farming systems, in particular, enable to manage larger herds in a more time-efficient manner, through automated monitoring of individual cow health and welfare. Dairy farmers can be assisted in the identification of unexpected behaviors or in the early detection of pathologies, with the possibility to decide whether and how to act. Furthermore, the analysis of big data from PLF technologies can help dairy farmers in phenotyping the animals for complex traits such as resilience, longevity, or productive life span, bringing to optimized breeding, treatment, and culling decisions. The aim of this thesis was to present different PLF applications to the dairy sector, that can serve as support tools for the optimization of farm management strategies and cows’ welfare. A general introductive chapter (Chapter 1) focused on an overview about PLF systems, objectives, and limitations, specifically addressed to the dairy sector. Afterwards, a new statistical model within the animal science field (i.e., joint model for longitudinal and time-to-event data) was tested to predict cow’s survival using first-parity sensor data as input (Chapter 2). The algorithm had good repeatability across farms with modest performances. However, joint models offer such interesting opportunities in terms of applicability and flexibility to justify further research for improving the overall predictive accuracy in the dairy sector. Further research investigated heat wave effects on dairy cows’ behaviors registered with sensors (Chapter 3). The output revealed that ‘heat-sensitive’ subjects were more active and spent more time chewing during a heat wave challenge compared to 'heat-tolerant' ones, as an attempt to better dissipate heat load. This suggested that the information provided by high-frequency sensor data can assist farmers in the early identification of cows for which personalized interventions to alleviate heat stress are needed. In a dedicated study, three different mathematical methods to estimate dairy cows’ expected production, milk losses, and perturbations of the lactation curve were analyzed and compared (Chapter 4). The output of this study can help dairy practitioners in choosing the method that best fits their management strategies, to understand, for example, how the animals cope with challenges, or to optimize their production system. Finally, a pilot study addressed to early detect cows at risk of metabolic disorders was conducted, using milk fatty acids analysis obtained with FTIR spectroscopy (Chapter 5). Preliminary reference intervals for de novo, mixed, and preformed fatty acids were calculated for healthy cows’ during early and mid-lactation. These reference ranges could help farmers to screen cows at risk of specific health disorders (e.g., subclinical ketosis) even before clinical signs are visible. In conclusion (Chapter 6), this thesis highlighted the potential of PLF in assisting dairy farmers to make better choices about the sustainability and the efficiency of their production system, by providing more objective information about health and productivity of the animals.