Masters Theses
Date of Award
8-2025
Degree Type
Thesis
Degree Name
Master of Science
Major
Animal Science
Major Professor
J Lannett Edwards
Committee Members
J Lannett Edwards, Neal F Shrick, Hao Gan, Yang Zhao
Abstract
Estrual cattle exhibit varying levels of estrous active behaviors and higher estrous-associated temperatures (HEAT), both influential in preovulatory follicle progression and contents important for fertility outcomes. Previous analyses utilizing descriptive animal variables left 52 to 95% of HEAT variation unexplained. Increased walking alone elevates body temperature, leading to the hypothesis that estrous active behaviors (e.g. mounting or standing to be mounted) significantly contribute to HEAT variability. The objective of these studies were to: 1) characterize estrous active behaviors in beef cows and heifers, 2) determine to what extent estrous active behaviors are associated with HEAT, 3) determine the ability of estrous active behaviors to predict vaginal temperature (VTp) during HEAT using machine learning models, and 4) determine which variables were more predictive of HEAT than others. Estrus was induced using gonadotropin-releasing hormone, a seven-day progesterone-infused controlled internal drug release (CIDR) application, and prostaglandin F2α (PGF2α) administration. HEAT onset was defined when VTp (monitored via iButton attached to a blank CIDR) increased 0.1°C above baseline. Estrous behaviors, including mounting and standing to be mounted, among others, were recorded and summarized every 15 minutes alongside the number of sexually active animals (SAGsum). Additional activity measures were captured using collar-based triaxial accelerometers, and ambient temperature-humidity index (THI) was recorded hourly. Estrus was observed in 93.3% of cows (14/15) and 83.3% of heifers (20/24). Average HEAT duration was 21.3 hours (cows) and 16.3 hours (heifers). Hierarchical linear regression indicated behaviors such as mounting others and standing to be mounted significantly associated with VTp increases (0.03 to 0.05°C per event), explaining up to 54% of HEAT variation. An ensemble machine learning model incorporating 66 predictors (behavioral, accelerometer-derived, ambient, and descriptive animal variables) achieved improved predictive accuracy. Using Shapley value analysis, the dataset was reduced to 12 highly predictive variables, enhancing computational efficiency and accuracy, explaining 66.3% of HEAT variability. Animal weight and SAGsum emerged as most predictive, followed by accelerometer-derived metrics and observed estrous behaviors. This research underscores the predictive value of estrous behaviors on HEAT, offering potential for optimized fertility management through integrated machine learning approaches on-farm.
Recommended Citation
Batey, Ian E., "Estrous Active Behaviors and Other Factors Related to Higher Estrous Associated Temperatures in Beef Cattle. " Master's Thesis, University of Tennessee, 2025.
https://trace.tennessee.edu/utk_gradthes/14532