Evaluation of sources of variation affecting ultrasonic predictors of carcass merit in beef cattle



Journal Title

Journal ISSN

Volume Title



Feedlot steers (n=180) representing 11 different sirebreed groups were serially measured ultrasonically for estimates of fat thickness (FTU) and longissimus muscle area (LMU). The first measurements were taken 30 dafter being placed on feed and at 60 d intervals until initiation of slaughter procedures. cattle were sacrificed in 4 groups (n=45/group) at 21 d intervals, with all remaining animals being measured at each time point. Weights were collected on each corresponding date and ages were derived from actual birthdates. Following overall analysis, animals were assigned to one of four biological types (BT) based on percent retail product at 0 mm trim level. Analysis of variance was performed to allow determination of sources of variation in ultrasonic measures. Significant weight (WT) effects (P < .001) were detected for all FTU measures and many of the LMU measures. Additionally, age (AGE) was a source of variation (P < .05) for FTU and LMU over the first five measurements. Sirebreed (SLINE) effects (P < .001) were present for all FTU estimates and early LMU predictions. Some dam line (DLINE) and SLINE*DLINE interactions were detected (P < .1). Regression techniques were utilized to describe change in ultrasonic measures over the course of the feeding period. Age effects were described quadratically (P <.0001) when FTU, LMU, FTU per 45 kg of live weight (FTUKG) and LMU per 45 kg of live weight (LMUKG) were regressed on age (R2 = 47, 68, 8 and 33%, respectively). Additionally, FTU and LMU were regressed on WT and described quadratically (P < .0001) with R2 values of 46 and 72%, respectively. When analyzed by BT, AGE effects on ultrasonic variables were described quadratically (P < .01) except FTUKG in one BT. Furthermore, all variables were significantly affected (P < .0001) by WT effects for all BT. Exponential functions were also used to describe relative change resulting in similar R2 values and standard errors. The prediction of carcass retail yield was evaluated at two trim levels, 8 mm and 0 mm and were used to calculate % trimmable fat (FAT8P and FATOP) and % retail product (RPD8P and RPDOP). When FAT8P, FATOP, RPD8P and RPDOP were regressed on YG, R2 values ranged from 75 to 76 % (P < .0001). Equations using final FTU as the independent variable explained 58 to 64% (P < .0001) of variation in FAT8P , FATOP, RPD8P and RPDOP. Retail yield percentages were then adjusted to age, weight or quality grade constant basis for further analysis. Using final FTU as the independent variable, 38 to 52% of the variation was accounted for in adjusted retail yield percentages. Quality grade prediction was done using computer video image analysis techniques where the traced LMU was made up of pixels, each assigned to 1 of 64 grey shades. Grey shade percentages were then analyzed using multivariate discriminant analysis to estimate USDA quality grades was achieved in 69% of the animals. These results indicate that standardization procedures area needed for valid comparisons of ultrasonic images between animals. Furthermore, these adjustment procedures need to be developed for differing biological types or breed groups. Additionally, the use of ultrasound for estimation of carcass attributes looks promising.



Beef cattle -- Composition, Beef cattle -- Carcasses -- Composition -- Measurement, Ultrasonic waves -- Industrial applications -- Case studies