The use of remote monitoring technologies for early detection of lameness in beef bulls during the breeding season

Lameness during the breeding season can impair bulls' reproductive performance, but early detection may enable timely treatments or replacements to mitigate pregnancy losses. This study used remote monitoring technologies to investigate whether changes in beef bulls' behavior could be used for early identification of lameness during the breeding season. Twenty-five Angus bulls (12 in year 1; 13 in year 2) considered satisfactory at the breeding soundness exam were enrolled. Collars with GPS and accelerometers were placed on bulls at the start of the breeding season. Bulls were monitored on pastures by video recording 3 times a week, on alternate days, during 2 breeding seasons to identify signs of lameness. Blinded raters analyzed videos, assigning a locomotion score (LS; 0-3 scale). Bulls with LS 2 or 3 were considered lame. The day lameness was identified was D0, and the xth days prior were D-x (e.g., D-1). Machine learning algorithms were used to estimate behaviors from the accelerometer data, including Activity (AC), Fighting (FI), Grazing (GR), Resting (RE), Ruminating (RU), and Walking (WA). State-space models were used to estimate the GPS trajectory to calculate the distance traveled (DIST), the average velocity (VEL), and the home range (HR). Statistical analyses were performed using generalized mixed-effect models. Behaviors of lame and non-lame bulls within the same period and breeding pasture were compared. The baseline of each bull was calculated for all behaviors using the average of D-14 to D-4. All behaviors from D-3 to D-1 of lame bulls were compared to their baseline for each day. Behaviors of non-lame bulls were compared to their baseline during the same period that lame bulls were observed. Lame bulls presented lower AC (D-1, -4.2 percentage points, p.p., P = 0.02), WA (D-2, -2.22 p.p., P = 0.03), DIST (D-1, -1,253 m, P = 0.02), VEL (D-2, -0.015 m/s, P = 0.01; D-1, -0.022 m/s, P < 0.001), and HR (D-1, -158,145 m2 P < 0.001), and higher RU (D-1, +4.4 p.p., P = 0.03) than non-lame bulls. Lame bulls presented a reduction in FI (D-1, -0.66 p.p., P = 0.03), WA (D-3, -2.09 p.p., P = 0.02; D-1, -3.02 p.p., P < 0.001), DIST (D-2, -1,051 m, P = 0.02; D-1, -1,562 m P < 0.001), VEL (D-2, -0.01 m/s, P = 0.04; D-1, -0.02 m/s, P = 0.002), and HR (D-1, -150,189 m2 P = 0.01) prior to lameness. Non-lame bulls presented a reduction in FI (D-3, -1.01 p.p., P < 0.001; D-1, -1.32 p.p., P < 0.001) and an increase in GR (D-3, +3.6 p.p., P < 0.001; D-1, 3 p.p., P = 0.005) within the period lame bulls were identified. Reductions in movement-related metrics (WA, DIST, and VEL) up to 3 days before lameness suggest these metrics may enable earlier lameness detection than visual observation. These results provide insights into key behaviors for early lameness detection in breeding beef bulls and inform future development of monitoring technologies.