Accurate estrus detection in grazing crossbred dairy cows (Zebu × Taurine) remains challenging in Brazil, and automated activity monitoring (AAM) systems may be a viable alternative. This study evaluated the effects of management and environmental factors on AAM performance for estrus detection in grazing Girolando cows. Nonpregnant 5/8 Gir × Holstein cows ( n = 25) were monitored during both dry and rainy seasons. During the dry season, cows were managed in semiconfinement with corn silage supplementation. In the rainy season, they grazed Panicum maximum cv. Mombaça pastures with or without eucalyptus trees. Each cow wore an AAM collar (CowMed, Brazil), and after individual baseline calibration, cows followed a timed artificial insemination (TAI) protocol. A statistical model included season, location, parity, and detection method (AAM, visual observation VO, adhesive patches APs) as fixed effects. Cow and cow × season interaction were random effects. Multiparous cows had higher estrus detection probabilities (97%) compared to primiparous cows (85%), regardless of method, location, or season. AAM detection probability was higher during the rainy season (98%) than during the dry season (73%). For AP, the detection probability was 80% in the rainy season and 97% in the dry season. VO detection remained stable at 92% in the dry season and 95% in the rainy season. Based on corpus luteum (CL) presence, AAM sensitivity and accuracy were 85% and 75%, respectively, while specificity was 33%. Parity and seasonal management significantly influenced AAM performance. The generally low specificity across detection methods highlights the need for improved algorithms and integrated estrus detection strategies in grazing systems.
Oliveira et al. (Thu,) studied this question.