155. Dry Matter Intake Prediction of Cattle Fed High-grain Diets Improves When Using Predictors Describing Degree of Finish.

Abstract As cattle accumulate body fat, physiological systems signal satiety, thereby reducing feed intake. Consequently, incorporating the degree of fatness into feed intake predictions is a logical and meaningful adjustment. The concept of equivalent shrunk body weight (EQSBW), adopted by NASEM (1996), adjusts cattle body weight to a standardized degree of fatness. The standard reference body weight (BW) was set at 478 kg BW for a steer with small00 (USDA Low Choice) marbling and yield grade 3. Guiroy et al. (2001) later proposed a similar adjustment of final body weight (AFBW) using empty body fat (EBF) estimated from carcass traits. We hypothesized that using EQSBW or AFBW would improve DMI prediction across interim feeding periods because these variables reflect both intake and tissue composition. Therefore, the objective of this meta-analytical regression study was to provide proof of concept that EQSBW and AFBW can offer better predictive accuracy for intake than metabolic BW (MBS), initial BW, or dietary NEm (NASEM, 2016; Eq. 10-1 and 10-4). Data from interim feeding periods (∼28 days each) across 18 finishing trials conducted at South Dakota State University were compiled (1,588 observations). Of the 128 blocks represented, 100 were used to develop equations which were later validated on the remaining 28. Intake was expressed as: absolute (kg/d), relative to body weight (% BW), multiples of maintenance (NEm, X), and relative to MBS (g/kg MBW). Proportion of AFBW (PAFBW) for each period was calculated as interim period BW / AFBW, while EQSBW was calculated as interim period BW × (478 / AFBW). The proportion of dry grain in the diet relative to high-moisture grain (GRAIN) was included as a continuous variable. Intake prediction equations were generated using the NLMIXED procedure of SAS, with study as random effect. When PAFBW was used as the predictor, the resulting equations were quadratic (P < 0.05), whereas those based on EQSBW were linear, but could be defined by two segmented lines breaking at 410 ± 15 kg. Effects of grain moisture content were significant (P < 0.05) and retained in all models. Observed DMI was regressed over predicted DMI. Observedtoexpected ratios and tests of intercepts crossing at zero with slopes of one were used to validate all equations. New equations predicted 36.4 to 38.4% of observations within 5% of expected values, whereas the NASEM equations predicted 28.9 to 31.2%. Using the GLIMMIX procedure, predicted intakes were also regressed against observed intakes. Intercepts differed (P < 0.05) from zero and slopes differed from one when using NASEM but not (P > 0.05) newly developed equations. This work demonstrates the value of incorporating body-composition related predictors when modeling feed intake of finishing cattle.