Abstract:
Proc. Assoc. Advmt. Anim. Breed. Genet. Voll3 USE OF FEEDER STEE R DESCRIPTOR S TO PREDIC T PERFORMANCE FEEDLO T AND CARCASS W. A. McKiernan' , S. A. Barwick * and D. J. Johnston' ' NSW Agriculture , Orange , NSW 2800 . *Animal Genetic s an d Breedin g Unit' , Universit y of New England , Armidale , NSW 235 I SUMMARY Feede r stee r descriptor s a t feedlo t entr y wer e examine d for thei r abilit y t o predic t determinant s of fina l valu e a s assesse d by feedlo t an d carcas s performance . Dat a analyse d wer e on 460 animal s from fou r group s of differin g breed , sex, managemen t histor y an d en d marke t weight . Variable s analysed wer e feedlo t dail y gain , fina l liveweight , carcas s weight , dressin g % , carcas s eye muscl e are a (EMA), carcas s P 8 fa t depth , retai l bee f yield % an d marblin g score . Stee r descriptor s wer e mos t usefu l for predictin g fina l weigh t an d carcas s weight , an d potentiall y usefu l in helpin g t o predic t carcas s P 8 fat dept h an d bee f yield % . Association s wit h feedlo t an d carcas s trait s ofte n differe d betwee n dat a sets, limitin g th e usefulnes s of descriptors . Stee r weigh t a t entr y wa s notabl e for th e constanc y of its associatio n wit h fina l weigh t (regressio n coefficien t approximatel y 1.0) an d carcas s weigh t over divers e dat a groups . Th e mos t usefu l stee r descriptor s wer e considere d t o b e liveweight , ag e (in months) , fatnes s (a sca n or score) , hi p height , an d muscl e thicknes s (a scor e or scan). Keywords X : INTRODUCTION Effectiv e marke t signal s ar e critica l t o bot h breeder s an d manager s of bee f cattle . On e wa y valuebase d tradin g a t feedlo t entr y migh t develo p is t o use , for valuatio n a t entry , prediction s of factors tha t decid e ultimat e valu e of th e finishe d slaughte r animal . Thes e factor s includ e thos e tha t specify marke t suitabilit y (age , weight , fa t depth , marblin g score , elc.), th e weigh t on whic h fina l paymen t is based , an d th e feed require d pe r da y in th e feedlot . Usefu l predictor s coul d als o b e utilise d in a store stee r descriptio n system . Researc h on live anima l descriptor s ha s concentrate d on descriptor s of the finishe d animal . Studie s on feede r or stor e animal s hav e bee n few (eg. Butt s et al. 1980). The objective s of thi s stud y wer e i) t o examin e a rang e of descriptor s of feede r steer s for thei r abilit y to predic t feedlo t an d carcas s performanc e trait s tha t affec t ultimat e valu e of th e animal , an d ii) to conside r th e transportabilit y of prediction s acros s dat a groups. MATERIALSAN D METHODS Fou r dat a set s wer e available , comprisin g 460 animal s fro m differen t breeds , sexes, vendor s and nutritiona l histories , an d whic h wer e manage d for differen t production-marke t syste m end-points. Animal s wer e fed for 102 t o 332 day s an d markete d a t averag e liveweight s of 5 19, 547, 603 an d 750 k g for lighte r an d heavie r Korean , an d shorter - an d longer-fe d Japanes e markets . Cattl e withi n a grou p wer e slaughtere d a t th e sam e time . Descriptor s evaluate d were : ag e (in months) , weight , height a t th e hip , a subjectiv e fa t scor e (a s pe r a marke t report) , ultrasoni c P 8 fa t depth , 12/13 ri b fa t depth I AGBU is a join institut of NSW Agricultur t e e and the Universit y of New England 333 Proc. Assoc. Advmt. Anim. Breed. Genet. Vol13 and eye muscle area (EMA), hip and stifle width, a visual muscle score (15point scale) and a maturity type score (9-point scale), where these were all taken at or near feedlot entry. Dependent variables considered were feedlot daily gain, final liveweight, carcass weight, dressing %, carcass EMA, carcass P8 fat depth, retail beef yield % and marbling score. Data from the four groups were initially analysed within groups, ignoring interactions, using the SAS REG Stepwise procedure (SAS Institute Inc. 1989). Models included management group and breed. Descriptors that accounted for less than one percent of variation in all data sets were excluded from further analyses. Data were then pooled for analysis, allowing assessment of changes to relationships between data groups and pre-entry managements. Pooled analyses used the SAS Mixed procedure (SAS Institute Inc. 1997). Analyses assessed the ability of descriptors to explain variation when animals were of known breed and pre-entry nutrition level (ie. `history known'), and whether relationships differed from one group to another. Account was taken of heterogeneous residuals where these occurred as a consequence of the pooling. Models included data set, prior nutrition level during pre-feedlot backgrounding (nested within data set), a random effect for breed-management group within data set, and all first-order interactions among fixed main effects and between these and the descriptors. Vendor (ie. herd of origin) was not included in the definition of management group since relationships in practice have to be applicable across vendors. Non significant (P<O. 1) terms were systematically dropped to yield the final model. The ability of descriptors to explain variation when the steer's history was completely also assessed. For this, all levels of effects for terms involving pre-feedlot variables from the above final models for pooled analyses. Data set was retained in these models. Results are described in terms of the percentage variation explained. successive models as 1 - (residual variance/ observed variance) x 100 . unknown was were dropped This was assessed for RESULTS AND DISCUSSION Selected results from the pooled analyses are presented in Table 1. Differences between data sets were a large contributor to variation in all traits. This is reflected in the percentage of variation explained by the base model when other history of the steer was unknown (Table 1). Known steer history included knowledge of breed and, in two of the four data sets, of the nutritional regime experienced prior to feedlot entry. When these aspects were known, models generally gave improved predictions, with the steer descriptors contributing relatively less. The steer descriptors considered most useful for prediction, for the range of traits analysed, were weight, age, hip height, a measure of fatness, and a measure of muscling. These accounted for an extra 2 to 20 percent of the total variation in feedlot and carcass performance traits (Table l), or 7 to 64 percent of the variation unexplained by base models. Use of a subjective fat score in place of a scanned fat depth, or of a muscle score in place of scanned eye muscle area, generally resulted in only a small loss of accuracy. 334 Proc. Assoc. Advmt. Anim. Breed. Genet. Vol13 Table 1. Percent variation in feedlot and carcass variables explained by base models and prediction models for feeder steers with histories either known or unknown. Numbers of steers in parentheses % variatison explained Steer history known Steer history unknown prediction prediction difference difference base model model model 61.0 (436) 92.7 (436) 48.4 (452) 93.0 (452) 90.6 (266) 75.2 (436) 28.4 (163) 61.3 4.0 10.2 3.8 9.4 1.8 14.5 16.3 4.6 40.4 (458) 79.2 (458) 34.6 (458) 81.5 (458) 81.0 (272) 53.5 (458) 10.3 (176) 58.4 47.8 (436) 92.5 (436) 42.7 (452) 92.7 (452) 90.4 (266) 73.1 (436) 27.2 (163) 62.2 7.4 13.3 8.1 11.2 9.4 19.6 16.9 3.8 base model Feedlot daily gain Feedlot final weight Carcass dressing % Carcass weight Carcass EMA Carcass P8 fat depth Retail beef yield % Marbling 57.0 (456) 82.5 (458) 44.6 (458) 83.6 (458) 88.8 (272) 60.7 (458) 12.1 (176) 56.7 Feedlot daily gain. Steer descriptors explained only 4 or 7 % extra variation in feedlot daily (Table l), emphasising the difficulty in predicting this trait without other knowledge. Comparison the base models showed knowledge of pre-entry nutrition level and breed was more useful than steer descriptors. Partial regression coefficients indicated younger steers and those with greater height had greater rates of gain. The association with entry P8 fat depth varied between data Leaner steers at entry had greater rates of gain in all but one data set of very diverse genotype. gain of the hip sets. Feedlot final weight. Feedlot entry descriptors accounted for an extra 10 or 13% of variation in final liveweight (58 or 64% of that unexplained by base models), this being mostly due to entry liveweight. Heavier, younger and taller steers had greater final weights. The association with entry P8 fat depth was generally negative, differing between data sets as for feedlot daily gain. The partial regression relating final weight to entry weight was consistently close to 1.O across data sets. Carcass dressing Partial regressions %. The association %. Steer descriptors accounted for onl:y 4 or 8 % of variation in dressing %. suggested steers with greater muscle score and taller steers had greater dressing with 12/l 3 rib fat depth differed between data sets, and was of inconsistent sign. with final weight, steer descriptors explained much of the remaining variation Heavier, taller and generally leaner steers at entry had greater carcass weights. weight of steers at intake was similar across diverse data groups. The P8 fat depth differed between data sets as for feedlot daily gain. Carcass weight. As in hot carcass weight. The association with association with entry 335 Proc. Assoc. Advmt. Anim. Breed. Genet. Vol I3 Carcass EMA. Steer descriptors explained an extra 9% of variation in steers of unknown history but only 2% when breed and pre-entry nutrition level were known. Steers with greater scanned EMAs, and which were leaner at entry, had larger carcass EMAs. Associations in each case differed between data sets, but were of consistent sign. Carcass PS fat depth. Table 1 suggests steer descriptors at entry can help predict carcass P8 fat depth. However, usefulness of the descriptors was limited by the fact that associations differed between data sets. Averaged over data sets, fatter and older steers at entry had fatter carcasses, and taller steers at entry had leaner carcasses. Retail beef yield %. Steer descriptors accounted for about an extra 19% of variation in beef yield %. The ability to predict yield, however, remained low (Table 1). Partial regressions suggested steers with greater EMA, and which were also leaner and lighter at entry, had a greater beef yield %. Marbling score. Differences in marbling occurred mostly between the data sets, these also reflecting post-entry management differences for different market end-points. The association of marbling with steer age became more positive as time on feed increased. CONCLUSIONS Results showed steer descriptors at feedlot entry are highly valuable for predicting final liveweight and carcass weight, potentially a help in predicting carcass P8 fatness and retail beef yield, and only broad indicators of feedlot daily gain, dressing %, carcass EMA and marbling. Associations with feedlot and carcass performance traits often differed between data sets, limiting the usefulness of descriptors. Separate predictions may be feasible for some differing circumstances. Partial regressions of foal weight and carcass weight on weight of steers at feedlot entry were notable for their constancy across diverse breeds and managements. Feeder steer descriptors can add to the accuracy of predicting feedlot and carcass performance traits that affect value. However, alone, they are of limited use. ACKNOWLEDGMENTS We are grateful to Meat and Livestock Australia for funding support, and to co-operating feedlots and Storelink groups, the Cattle and Beef Industry CRC for Meat Quality, and NSW Agriculture, Grafton and Glen Innes for access to supporting data. REFERENCES Butts, W.T., Backus, W.R., Lidvall, E.R., Corrick, J.A. and Montgomery, R.F. (1980) J. him. Sci. 51: 1297 SAS Institute Inc. (1989) 'SAS/STAT User's Guide, Version 6, Vol 2' 4'h ed. SAS Institute Inc., Gary, NC SAS Institute Inc. (1997) 'SAS/STAT Software, Changes and Enhancements through Release 6.12' SAS Institute Inc., Gary, NC 336