Horses reared for meat production are generally fed a diet rich in starch with the aim of maximizing production performances. This study evaluated the effects of two feeding management systems on horse welfare by analysing the relative time spent engaged in different behavioural activities. Nineteen Bardigiano horses aged 14.3 ± 0.7 months were randomly assigned to one of two group pens: one group was fed high amounts of starch-rich concentrates (HCG; n = 10), the other was fed a fibre-based diet (HFG; n = 9). Behavioural activities performed by each horse were video-recorded over a 96-h period. A scan sampling process (n = 144 scans/horse/day; total n of scans sampled = 10,368) was used, and the scans were analysed according to a specific ethogram. The mean frequency (%/24 h) spent exhibiting each behavioural activity was calculated to obtain the time budget. After checking for normality (Shapiro–Wilk test), Student’s t tests (normally distributed data) and Mann–Whitney tests (not normally distributed data) were used to compare the time budgets of the two groups of horses (HCG vs. HFG). Principal Component Analysis (PCA) was applied to identify the components explaining the variability in behavioural activities between the two groups. K-means cluster analysis subsequently confirmed the PCA results. The behavioural activities associated with feeding horses a fibre-based diet correlated with better horse welfare compared with feeding horses a starch-based diet. Feeding horses a fibre-based diet resulted advantageous from both the welfare and economic perspective; it allowed horses to spend more time expressing feeding behaviours and reduced energy expenditure in the form of excitable, or “fizzy”, behaviours.

The application of hay feeding devices, such as the use of hay nets or slow feeders, can help with the management of weight in ponies; however, there is still a lack of knowledge regarding their effect on equine posture. Therefore, the aim of the study was to assess the effect of different feeding devices on the posture of ponies using morphometric analysis. Two different breed types, Shetland type (SH, n = 5) versus Welsh Cob type (WC, n = 4), were fed the same forage in four different ways: on the ground (G), using a fully filled haynet (HF), using a partially filled haynet (HL) and using a slow‐feeder hay box (HB). Video recordings were obtained and then geometric morphometric analysis was applied. Breed morphology was confirmed by body morphometric measurements. Data were analysed statistically using one‐way ANOVA, canonical variate analysis (CVA), principal component analysis (PCA), partial least‐squares (PLS) analysis and multivariate analysis of variance (MANOVA). Moreover, a mixed model was performed to study differences in mandibular angle. SH and WC ponies were shown to have significantly different body morphometric measurements. The geometric morphometric analysis results showed that ponies arch their back and modify their neck shape differently according to the feeding method and their morphological group. For the neck, the SH and WC ponies adapted similarly to the use of small‐holed hay nets, but their posture varied when feeding from the ground or hay box. The back postures consistently differed according to the breed type and feeding method. The mandibular angle for both breed types was reduced with all the feeding devices compared to feeding from the ground. Further studies are needed to evaluate the long‐term effects of slow‐feeding devices on posture and mandibular angle, taking into consideration animals with different morphologies.

Background: Understanding social contacts in livestock is essential for behavioural research, welfare assessment and modelling pathogen transmission. Proximity sensors based on received signal strength indicator (RSSI) offer a promising tool to quantify fine-scale social interactions in the context of precision livestock farming, but their performance depends strongly on environmental conditions and sensor configuration, requiring careful calibration and validation. In this study, we calibrated and validated commercially available RSSI-based proximity sensors under controlled indoor and outdoor conditions and through in vivo trials with sheep. Results: Calibration experiments conducted across distances ranging from 0.5 to 5 m showed a predictable decrease in RSSI with increasing distance, modulated by environmental setting and sensor orientation, with both factors significantly influencing signal attenuation. Probability-based modelling demonstrated that RSSI reliably predicts biologically relevant proximity thresholds (≤ 0.5, 1.0 and 1.5 m), supporting the derivation of context-specific RSSI cut-offs rather than universal thresholds. The in vivo validation, combining proximity sensor data with image-based distance measurements matched to each individual RSSI detection event, demonstrated an overall classification accuracy of 82%, with per-class F1-scores ranging from 0.76 for distances greater than 1.5 m to 0.86 for distances of 1 m or less. Conclusions: When appropriately calibrated, RSSI-based proximity sensors can reliably detect biologically meaningful social interactions in sheep across a range of inter-individual distances. These findings support the use of proximity sensing as a precision livestock farming tool for the study of social behaviour, welfare assessment, management and contact-based disease processes.

Early identification of sport potential in show jumping horses remains challenging because competition outcomes vary widely, and elite-level performance develops after considerable time. In this study, we evaluated the applicability of machine learning approaches to competition records from horses identified through FEI participation, complemented by national-level records from Sweden, Belgium and France, with the aim of identifying horses likely to reach elite performance. Performance data from Fédération Équestre Internationale (FEI) competitions were analysed using an Extreme Gradient Boosting (XGBoost) classifier trained on longitudinal competition histories recorded before horses reached ten years of age. Animals were classified as champions if they achieved at least one clear round at 160 cm, while non-champions were defined as those that completed their competitive career without reaching this level. Model performance was assessed using nested cross-validation and compared with two single-rule-based classifiers based on maximum obstacle height thresholds. XGBoost outperformed baseline approaches across all evaluation metrics, achieving an average area under the ROC curve of 83.5% and an average precision of 80.0%. Analysis of probability trajectories showed that predictive information accumulates progressively over a horse’s career, with increasing separation between champion and non-champion predictions as competition history expands. Model performance improved substantially from approximately seven years of age onward. Permutation importance and SHapley Additive exPlanations (SHAP) analyses revealed that obstacle difficulty, performance consistency and age-related progression were among the most influential predictors. Overall, these findings demonstrate that interpretable machine learning models can extract meaningful predictive signals from early competition trajectories and support evidence-based talent identification in show jumping, while maintaining attention to performance development and horse welfare.