Non-invasive analysis of dairy cattle vocalizations offers a practical route to continuous assessment of stress and timely health interventions in precision livestock systems. We present a multi-modal AI framework that fuses standard acoustic features (e.g., frequency, duration, amplitude) with non-linguistic, transformer-based representations of call structure for behavior classification. The classification analysis represents the core contribution of this work, while the integration of the Whisper model serves as a complementary exploratory tool, highlighting its potential for future motif-based behavioral studies. Using contact calls recorded from a cohort of lactating Romanian Holsteins during a standardized, brief social-isolation paradigm, we developed an ontology distinguishing high-frequency calls (HFCs) associated with arousal from low-frequency calls (LFCs) associated with calmer states. Across cross-validated models, support vector machine and random-forest classifiers reliably separated call types, and fused acoustic + symbolic features consistently outperformed single-modality inputs. Feature-importance analyses highlighted frequency, loudness, and duration as dominant, interpretable predictors, aligning vocal patterns with established markers of arousal. From a clinical perspective, the system is designed to operate passively on barn audio to flag rising stress signatures in real time, enabling targeted checks, husbandry adjustments, and prioritization for veterinary examination. Integrated with existing sensor networks (e.g., milking robots, environmental monitors), these alerts can function as an early-warning layer that complements conventional surveillance for conditions where vocal changes may accompany pain, respiratory compromise, or maladaptive stress. While the present work validates behaviorally anchored discrimination, ongoing efforts will pair vocal alerts with physiological measures (e.g., cortisol, infrared thermography) and multi-site datasets to strengthen disease-specific inference and generalizability. This framework supports scalable, on-farm welfare surveillance and earlier intervention in emerging health and stress events.

The overall objective of this study was to evaluate the effects that milking temperament (MT) of water buffaloes has on milking traits, welfare indicators, and reproductive outputs. The study was performed on 60 multiparous buffalo cows (6.7 ± 0.6 lactations) at the beginning of their lactation (100 days in milk, DIM). Each buffalo cow was scored by two independent observers using a temperament scoring system (1: extremely calm, 2: calm, 3: alert, 4: reactive, and 5: aggressive), and then grouped as ‘calm’ (scores 1, 2, and 3; n = 42) or ‘nervous’ (scores 4 and 5; n = 18). Additionally, the milk yield at 100 DIM (MY), milking speed (MS), calving interval (CI), age at first calving (AFC), body condition score (BCS), animal-based welfare parameters, and infrared thermography data (IRT) were evaluated. The MT significantly influenced the MY (p = 0.0082), with calmer cows outperforming their nervous counterparts. The MS was significantly influenced by the MT (p = 0.0015), with calmer animals having a higher milk ejection rate. The MT of the cows had no influence on the CI, AFC, or BCS. The evidence from this study suggests that the responsiveness of buffalo cows during milking affects their milk yield and milking speed, with no associations being found for reproduction efficiency indicators or animal-based welfare indicators.

Enzootic Bovine Leukosis: Surveillance Measures and Control Program in the Northern Dobruja Area of Romania Between 2017 and 2020

Cattle breeding has traditionally focused on improving production traits; however, recent interest in positive animal welfare has shifted attention toward selecting for more robust animals that balance productivity with health and well-being. The aim of the current study was to assess whether behavioural responses during milking in dual-purpose cattle are associated with genetic markers, previously linked to temperament traits in dairy and beef breeds. We focused on 185 lactating cows belonging to the Simmental strain (Romanian Spotted, national name), which were evaluated for their milking behaviour. Genotyping was performed using an 88-SNP panel selected based on prior associations with dairy and beef cattle temperament. We identified five SNPs that were significantly associated with milking reactivity in the Romanian Spotted breed, located in genes previously linked to neural development, stress response and behavioural regulation (USH2A, ADAMTS7, TBC1D2B and ZMAT4). Our findings suggest that milking behaviour in dual-purpose Simmental cattle is influenced by genetics, supporting the potential for including behavioural traits in future selection strategies. This study contributes to a better understanding of the genetic mechanisms underlying stress-related behaviours in dual-purpose cattle breeds.

National greenhouse gas inventories commonly apply a single enteric methane emission factor to an entire dairy population, which can obscure meaningful differences between production systems and between feeding seasons. An IPCC Tier 2 assessment was performed for the Romanian dairy herd in the reference year 2024 (1,051,894 cows), stratified into extensive (72.77%), semi-intensive (23.10%), and intensive (4.12%) systems and, for the two grazing-based systems, into a winter indoor and a summer grazing period. Gross energy intake (GE), digestible energy (DE), and the methane conversion factor (Y_m) were derived from national feed composition references, with Y_m estimated from DE using a regression equation. Season-specific emission factors were combined into day-weighted annual values. GE ranged from 153.60 MJ head⁻¹ day⁻¹ to 387.78 MJ head⁻¹ day⁻¹ and Y_m from 5.48% (semi-intensive, grazing) to 6.44% (extensive, indoor). Day-weighted annual emission factors were 144.37 kg CH₄ head⁻¹ year⁻¹, 108.17 kg CH₄ head⁻¹ year⁻¹ and 81.92 kg CH₄ head⁻¹ year⁻¹ for the intensive, semi-intensive, and extensive systems, whereas methane intensity declined in the opposite direction, from 25.32 g CH₄ kg⁻¹ FPCM to 19.98 g CH₄ kg⁻¹ FPCM. National enteric emissions were estimated at 95,260 t CH₄ (2,572,014 t CO₂-eq), 65.8% of which originated from the extensive system. Sensitivity analysis showed that national emissions were more sensitive to gross energy intake (±10% change in emissions for a ±10% change in GE intake) than to digestible energy (±4.3–4.8% change for a ±2 percentage-point change in DE). These results suggest that the methane-related environmental ranking of Romanian dairy systems depends strongly on the functional unit considered and that seasonal variation in Y_m is comparable in magnitude to differences between production systems.