PSLBII-11 - Predicting carcass chemical composition of crossbred bulls using Dual Energy X-Ray Absorptiometry (DXA) scanning of different carcass sections

Abstract: The nutrients required by cattle are influenced by body composition. Indirect methods, such as DXA scanning, have been used to predict lean, fat, and bone tissues without the need for carcass dissection. The rib section between the 9th and the 11th ribs (HH section, Hankins and Howe, 1946) is the most used section to estimate carcass chemical composition through DXA or other approaches. However, to facilitate application in commercial packing plants, alternative approaches need to be developed. We hypothesized that carcass chemical composition can be precisely estimated using DXA data from alternative sections that would be easier to accomplish in a commercial facility. Therefore, this study aimed to develop equations to predict the whole carcass chemical composition of crossbred bulls using DXA scanning a different carcass section. The carcasses from 24 Red Angus x Nellore growing bulls receiving maintenance (n = 4), and the ad libitum (n =20) diets with 20:80 (forage:concentrate) on a dry matter (DM) basis with corn silage as the forage. After 24 h of chilling, the left half-carcass of all animals were weighed, divided into 3 sections. The sections were obtained considering the carcass hanging by Achilles, as following: S1) longitudinal cut after the second rib, HH section, and the remaining section after the removal of S1 and HH section. The S1 and HH sections were scanned in a medical DXA unit (GE Healthcare, Lunar Prodigy Advance, USA). The small animal configuration mode of the GE Healthcare enCORE software, version 18, was used for DXA. The DXA scanning provided data on fat, lean, and bone mineral content (BMC). After scanning, each section was dissected into fat, muscle, and bone. DXA variables of carcass sections were used to predict carcass chemical composition (EE, CP, and ash) using general linear regression models in RStudio software. To develop prediction equations based on the input variables, a leave-one-out cross-validation method was utilized. The precision of the predictions was assessed based on the coefficient of determination (R²), root mean square error (RMSE), and Akaike’s information criterion (AIC). The predictive equations of carcass chemical composition based on DXA measurements of the different sections are reported in Table 1. Prediction equations for S1 for CP, EE, and Ash content had lower RMSE and AIC values and a greater R² when compared to the equations based on the HH section. In general, all equations had relatively high precision. However, S1 may be better for predicting carcass chemical composition because of improved goodness of fit and because this section can be more easily collected in a commercial packing plant. Nevertheless, further studies are needed to improve accuracy and evaluate the equations using an independent dataset.