PSII-2 Application of partial least squares regression to predict feed efficiency and intake using feeding behavior patterns in growing steers and heifers

Abstract

Abstract Objectives of this study were to evaluate the use of feeding behavior traits to predict individual-animal RFI and DMI of growing cattle fed high-grain finishing diets. Performance, DMI, and feeding behavior data were collected from 1 study utilizing 498 Angus-based composite steers (Study 1), and 2 studies utilizing 408 heifers (Study 2) and 321 steers (Study 3) composed of Brangus, Braford, Simbrah, and Angus breeds. DMI and feeding behavior traits were measured using a GrowSafe system, and RFI calculated within trial. Seventeen feeding behavior traits were evaluated: Frequency and duration of bunk visit (BV) and meal events, head-down duration (HDD), average meal length, maximum non-feeding interval, corresponding day-to-day variation (SD) of these traits, and ratios of HDD per BV duration, HDD per meal duration, and BV events per meal event. Partial least squares regression (PLSR) models for DMI and RFI were calibrated using data from Study 1 and 2, and independently validated using Study 3. Independent variables for the DMI models included mid-test BW0.75, ADG, frame size, and ultrasound traits, with and without feeding behavior traits, and for the RFI model included frame size, ultrasound, and feeding behavior traits. For prediction of DMI, validation R2 (R2v) of the base model (Mid-test BW0.75, ADG, frame size, and ultrasound) was 0.46. Inclusion of feeding behavior traits to the base model increased R2v to 0.66. For prediction of RFI, R2v was low (0.37), but the model classified 51% of calves into the correct RFI group (± 0.50 SD), with only 7% incorrectly classified across 2 RFI groups. Ongoing development of biosensor-based technologies to quantify feeding behavior patterns provides opportunities to predict DMI in support of precision nutrition, and reduce costs of identifying feed-efficient cattle. Further research is warranted to evaluate the robustness of PLSR-based models to predict RFI and DMI in cattle.