SFRUTTARE L'ANALISI DEL LATTE MEDIANTE SPETTROSCOPIA NEL MEDIO INFRAROSSO PER LA FENOTIPIZZAZIONE SU LARGA SCALA DEI CARATTERI LEGATI AL BENESSERE ANIMALE E ALLE EMISSIONI DI METANO PER VACCHE DA LATTE ALLEVATE IN AZIENDE AFFERENTI AL CONSORZIO PARMIGIANO REGGIANO

Milk quality has long been evaluated using mid-infrared (MIR) technology, a well-established tool in milk quality programs. Mid-infrared can capture variations in milk composition through its interaction with chemical bonds. Because milk is an exceptionally sensitive and dynamic matrix, influenced by both individual animal and environmental factors, its chemical composition reflects these influences. Consequently, each MIR spectrum serves as a unique fingerprint. Since MIR-based milk evaluation has been routinely conducted as part of milk quality assessments, the use of (raw) MIR spectral data offers an ideal opportunity to generate large-scale phenotypic information. In this context, the aim of this thesis is to explore the use of MIR technology for generating predicted phenotypes and identifying their sources of variation, particularly those related to animal welfare and sustainability. To achieve these goals, a substantial amount of longitudinal data was provided by the Breeders Association of the Emilia Romagna Region, and the Italian Breeders Association through the Livestock Environment Open Data project, and the Parmigiano Reggiano Cheese Consortium. The database covered a period of nearly two years and included information at both the farm and individual levels. At the farm level, the data comprised farm production records and bulk tank milk quality measurements, totaling 22,010 records from 1,508 farms. In addition, 20,363 MIR bulk milk spectra were available from 1,485 farms. The farm-level information included animal welfare evaluations conducted by the Italian National Reference Center for Animal Welfare, along with key farm characteristics from 983 farms, including feeding system, geographical location, breed, and housing system. At the individual level, the dataset comprised 2,175,148 test-day production and milk quality records from 250,071 cows, as well as 1,869,687 individual MIR spectra. Additional individual-level information included age at calving, days in milk, breed, and parity order. This thesis is structured into three main chapters. The first step (chapter I) was to cluster farms based on shared characteristics and management practices to identify distinct dairy systems. The aim was to better characterize the study population and analyze how these dairy systems influence milk quality, production, and animal welfare. At the farm level, we then evaluated the ability of the MIR spectrum, applied to bulk milk, to distinguish among the identified dairy systems and to predict farm-level animal welfare indicators, as well as individual farm practices and characteristics. In chapter II, milk spectra were used to estimate the concentrations of blood biomarkers and minerals related to energy balance and body reserve mobilization. The objective was to explore patterns of variation in these predicted biomarker concentrations according to various sources of variation, particularly the interaction between breed and dairy system, parity, and lactation stage. Finally, in chapter III, the aim was to evaluate the sources of variation in estimated enteric methane emissions according to different phenotypes, calculated from MIR predictions of milk fatty acids. This chapter analyzed the impact of the identified dairy systems on methane emissions and the interaction between breed and system. Additionally, this chapter examined the effect of farm-level animal welfare on methane emissions, exploring the relationship between farm-level animal welfare evaluation and individual methane phenotypes. This thesis illustrates the potential of MIR as a predictive tool to support multiple areas of the dairy industry. This is particularly important in the context of Italian dairy production, where certifications are a fundamental part of the food chain. The findings of this thesis are expected to contribute to herd sustainability and management, as well as to support the future implementation of management and genetic programs.