Published December 11, 2014 Assessment of Lamb Carcass Composition from Live Animal Measurement of Bioelectrical Impedance or Ultrasonic Tissue Depths 1,2,3 E. P. Berg*, M. K. Neary*,4, J. C. Forrest*, D. L. Thomas, and R. G. Kauffman *Purdue University, West Lafayette, IN 47907-1151 and University of Wisconsin, Madison 53706 ABSTRACT: Market weight lambs, average weight 52.5 kg ( ±6.1), were used to evaluate nontraditional live animal measurements as predictors of carcass composition. The sample population ( n = 106) represented U.S. market lambs and transcended geographic location, breed, carcass weight, yield grade, and production system. Realtime ultrasonic (RU) measurements and bioelectrical impedance analysis (BIA) were used for development and evaluation of prediction equations for % boneless, closely trimmed primal cuts (BCTPC), weight or % of dissected lean tissue (TDL), and chemically derived weight or % fatfree lean (FFL). Longitudinal ultrasonic images were obtained parallel to the longissimus thoracis et lumborum (LTL), positioning the last costae in the center of the transducer head. Images were saved and fat and LTL depths were derived from printed images of the ultrasonic scans. Bioelectrical impedance analysis was administered via a four-terminal impedance plethysmograph operating at 800 ma at 50 khz. Impedance measurements of whole-body resistance and reactance were recorded. Prediction equations including common linear measurements of live weight, heart girth, hindsaddle length, and shoulder height were also evaluated. All measurements were taken just before slaughter. Bioelectrical impedance measurements (as compared to RU and linear measurements) provided equations for %BCTPC, TDL, %TDL, FFL and %FFL with the highest R 2 and lowest root mean square error. Even though BIA provided the best equations of the three methodologies tested, prediction of proportional yield (%BCTPC, %TDL, and %FFL) was marginal (R 2 =.296,.551, and.551, respectively). Equations combining BIA, RU, and linear measurements greatly improved equations for prediction of proportional lean yield. Key Words: Lambs, Carcass Composition, Impedance, Ultrasound J. Anim. Sci. 1996. 74:2672 2678 Introduction A survey including questions addressing marketing and merchandizing problems facing the U.S. lamb industry, as perceived at each level of the market chain, was conducted by the Texas Agricultural Marketing Research Center s (TAMRC) Lamb Study Team (TAMRC, 1991). Lamb feeders reported the number one problem was over-finished slaughter lambs. Packers replied that live weight variability was a primary concern, and 68% of non-breaking wholesalers responded that the combination of quality/yield 1 Journal Paper No. 14915, Purdue Agric. Exp. Sta., West Lafayette, IN. 2 This project was funded by a grant from the American Sheep Industry Association, Englewood, CO. 3 Authors wish to acknowledge Gary Waters, Mike Booth, David Schoon, Ted Fisher, Joanne Kuske, and Jim Stouffer for technical assistance on this project. 4 To whom correspondence should be addressed. Received February 12, 1996. Accepted June 21, 1996. variability, unacceptable quality lamb, fat trim issues, and live weight variation were the largest problems affecting the marketability of U.S. slaughter lambs. As the lamb industry becomes more progressive in the sale of "case ready" products these issues will become a greater concern, because excess fat production and product variation will result in lost revenue (Magagna, 1991). Carcass characteristics are highly heritable (Lasley, 1987). Identification and development of electronic technology to predict carcass composition on live lambs would be valuable for improving the sheep industry. Additionally, the identification of replacement seedstock possessing desirable carcass composition traits should increase the rate of genetic improvement for a consumer-preferred product. Umberger (1994) described a system of valuebased marketing that established carcass weight, quality grade, and yield grade as a means of lamb procurement. This system would reward timely delivery of slaughter lambs to market and reduce quality and yield problems within the packing industry via established price incentives. As the industry moves 2672
PREDICTION OF LIVE LAMB COMPOSITION 2673 Table 1. Number of market lambs in each weight group and yield grade (YG) classification (n = 106) Live weight YG 1 YG 2 YG 3 YG 4 YG 5 Percentage 63.5 68.0 kg 1 1 0 0 0 2% 59.0 63.4 kg 0 7 4 5 0 15% 54.5 58.9 kg 6 11 5 2 2 25% 50.0 54.4 kg 6 11 7 6 0 28% 45.4 49.9 kg 8 4 4 0 0 15% <45.3 kg 10 4 1 1 0 15% Percentage 29 36 20 13 2 closer to value-based marketing, technology will be needed to assist producers in orderly marketing of lambs at optimum yield of carcass lean. The objective of this study was to determine the value of bioelectrical impedance ( BIA) and realtime ultrasound ( RU) measurements recorded on live lambs for predicting carcass composition. Materials and Methods Market lambs (n = 106) were selected to represent a cross-section of slaughter lambs in the United States. Lambs were acquired from the lower mountain region of Montana, the Ohio Valley, Michigan, Wisconsin, Indiana, and Texas. The experimental group comprised 28 ram, 33 wether, and 44 ewe lambs representing Suffolk, Dorset, Columbia, North Country Cheviot, Montadale, Finnsheep, Hampshire, Romanov, Targhee, and Texel breeding. Before lamb purchase, a selection criteria was designed. Selected lambs were representative of market weights and USDA yield grades of lambs marketed in 1993 according to USDA reports. The sample population included 29% yield grade ( YG) 1 carcasses, 36% YG 2, 20% YG 3, 13% YG 4, and 2% YG 5 (Table 1). Data were collected on lambs from late August to early December. Lambs were purchased at the various buying stations throughout the country at or near the desired market weight, delivered to the Purdue University Animal Sciences Sheep Research Center, and allowed time to recover from transportation stress. After weight lost during transportation was recovered, lambs were delivered to the Purdue University Animal Sciences Meat Laboratory for live measurements, slaughter, and carcass measurements. Realtime ultrasound measurement of longissimus thoracis et lumborum ( LTL) and subcutaneous fat depth was conducted using a methodology recommended by Stouffer (personal communication). A longitudinal ultrasonic scan was taken placing the last costae in the center of the transducer head, parallel with the LTL, and lateral to the spinous process. Lambs were closely shorn along the dorsal plane and light vegetable oil was applied to allow maximum contact of the transducer head. An Aloka 210 realtime ultrasound device (Corometrics Medical Systems, Wallingford, CT) with a 5-MHz transducer was used for all ultrasonic measurements. Realtime images were saved and printed on a Sony Video Graphic Printer UP-811 for manual analysis of tissue depths. Depending on the image resolution, three to five tissue depths were manually identified along the ultrasonic image. An average of several muscle ( ULD) and fat ( UFD) tissue depths along the LTL should be superior to a single data point obtained from ultrasonic measurement of loin muscle area and overlying fat depth. Bioelectrical impedance analysis was conducted on the live lambs using a modified methodology of Berg and Marchello (1994). The same anatomical reference points were used for cranial and caudal placement of the black transmitter electrodes. Eight centimeters was the distance between transmitter and detector electrodes in this study. A four-terminal impedance plethysmograph (RJL Systems, Detroit, MI) introduced a deep, homogeneous, alternating current through the subject at 800 ma at 50 khz. The transdermal tetrapolar electrode configuration consisted of two signal-generating electrodes and two receiving electrodes. Twenty-one-gauge Vacutainer needles (Beckton Dickinson, Rutherford, NJ) served as electrodes to ensure transmission of the electric signal and to eliminate skin electrode interaction common with the use of conductive adhesive tape (Hall et al., 1989). Impedance measurements of resistance ( Rs), reactance ( Xc), and length between detector electrodes ( LENG) were collected. The plethysmograph transmits an alternating current between the outer two electrodes and the voltage drop between the inner two detector electrodes was measured. Lean tissue is a highly conductive substance composed mostly of water containing electrolytes, whereas fat tissue serves as an insulator that impedes the flow of an applied electrical current (Swatland, 1984). The hypothesis was that a fatter lamb should impede the transmission of an electrical current to a larger extent. Electrical impedance is dependent on the conductor length, cross-sectional area, and signal frequency. Assuming a similar geometric shape of
2674 BERG lambs, and using a constant electrical signal, the components of impedance (Rs and Xc) should be derived relative to body composition. Electrode placement was modified slightly from the procedure of Berg and Marchello (1994). Electrodes were inserted approximately 1 cm from the dorsal midline (avoiding contact with the spinal process) 10 and 18 cm caudal from the neck fold (last cervical vertebra) and approximately 5 and 13 cm cranial from the first coccygeal vertebra. All electronic measurements were collected just prior to slaughter. Easily obtained linear measurements were also collected. Live weight, hindsaddle length, shoulder height, and heart girth measurements are often used as selection criteria for breeding stock. These data were collected and evaluated for prediction of carcass characteristics. Hindsaddle length was measured as the distance from a position immediately caudal to the last costae to the first coccygeal vertebra; shoulder height was measured as standing height from the floor to the dorsal point of the scapula; and heart girth was measured as the circumference of the thoracic cavity, immediately caudal to the humerus at approximately the 4th costae. Lambs were slaughtered according to normal industry procedures. Kidney and pelvic fat were removed and weighed at time of slaughter. Carcasses were chilled at 2 C for 24 h. Chilled carcasses were ribbed between the 12th and 13th thoracic costae, separated into fore- and hindsaddle and then cut medially. The right sides were cut into wholesale parts according to specifications outlined by the NAMP (1992) for the square-cut shoulder (NAMP #207), rack (NAMP #204), trimmed loin (NAMP #232), and leg (lower shank removed; NAMP #233A). ET AL. The major primal cuts were trimmed to 3.1 mm external fat depth. A boneless, closely trimmed primal cut weight was then obtained. Both major and minor primal cuts (breast, plate, flank, and fore- and hindshank) were dissected into separable lean, external fat, internal fat, and bone. Each tissue component was weighed and recorded for each primal cut. The dissected lean of the major primal cuts and the pooled minor primal cuts were ground twice through a 3.1-mm plate and sampled for chemical analysis of crude fat (ether extraction; AOAC, 1990). Percentage of chemical fat within the sample was then multiplied by dissected lean mass to derive FFL (lean wt [lean wt %fat]). Statistical Analysis. The data were analyzed by SAS (1991) linear regression procedures. The data collected via electronic means served as the independent variables to predict total and % dissected carcass lean ( TDL), weight and % of fat-free dissected total carcass lean ( FFL), and percentage of boneless, closely trimmed primal cuts ( BCTPC; 3.1-mm trimmed leg, loin, rack and shoulder minus dissected bone). All independent variables are reported for individual equations developed from BIA, RU, and linear measurements with corresponding partial R 2 and actual P-value. The partial R 2 gives an indication of the percentage of sample variation explained by a specific variable. The partial R 2 can be compared to the regression equations overall R 2 to show which independent variables explain the majority of the variation within the sample population. Final prediction equations were selected for maximum R 2 and minimum root mean square error ( RMSE). Forward STEPWISE regression (SAS, 1991) was used to determine significant variables in prediction equations that incorporated all independent variables from BIA, Table 2. Means, standard deviation, and ranges for live and carcass traits of market lambs Variable n Mean SD Min Max Live wt, kg 101 52.5 6.1 38.6 64.0 Warm carcass wt, kg 101 29.4 4.3 19.5 37.6 Dressing percentage 101 55.6 4.4 43.2 68.4 USDA yield grade 101 2.6 1.1 1.0 5.4 Heart girth, cm 101 97.7 7.7 83.8 119.4 Hindsaddle length, cm 100 39.2 3.9 30.4 63.5 Shoulder height, cm 100 62.9 5.2 43.2 73.7 Live resistance, ohms 101 33.5 8.0 17.0 52.0 Live reactance, ohms 101 4.7 1.4 2.0 10.0 Live length, cm a 101 36.4 5.4 22.9 48.3 Avg ultrasonic fat depth, cm 82.5.2.1 1.3 Avg ultrasonic loin depth, cm 82 2.6.4 1.8 4.3 Boneless, closely trimmed primal cuts, kg 101 13.9 2.2 9.2 18.0 % Boneless, closely trimmed primal cuts 101 47.7 4.1 29.2 56.7 Total dissected lean, kg 101 14.2 2.0 9.8 19.1 % Total dissected lean 101 48.7 4.7 38.1 59.8 Fat-free lean, kg 96 13.3 2.0 9.1 17.9 % Fat-free lean 96 45.2 5.0 33.4 57.8 a Measured between BIA (red) detector terminals.
PREDICTION OF LIVE LAMB COMPOSITION 2675 Table 3. Coefficients between live measurements and measures of carcass composition LWT RS XC LENG HG HSDL SH UFD ULD Variable ab (n = 101) (n = 101) (n = 101) (n = 101) (n = 100) (n = 100) (n = 100) (n = 79) (n = 79) % BCTPC.01.44**.04.17.06.13*.22*.32**.04 TDL.77**.03.25*.22*.52**.32**.33**.12.40** % TDL.14.63**.06.18.32**.09.08.50**.09 FFL.74.10.23*.17.46**.32**.42**.04.36** % FFL.08.63**.04.17.29**.13.16.50**.07 a b BCTPC = boneless, closely trimmed primal cuts; TDL = total dissected carcass lean weight; FFL = fat-free lean. LWT = live wt; RS = BIA resistance (ohms); XC = BIA reactance (ohms); LENG = length between BIA red detector terminals (cm); HG = heart girth (cm); HSDL = hindsaddle length (cm); SH = shoulder height (cm); UFD = average ultrasonic fat depth; ULD = average ultrasonic loin depth (cm). *P <.05. **P <.01. RU, and linear measurements. Sex was regressed on residual outputs from prediction equations developed from linear sheep measurements, BIA, and ultrasonic tissue depths as a test to determine whether these equations were biased relative to sex. Results and Discussion Table 2 includes means, standard deviations, and ranges for the data collected on 106 market weight lambs. Live weight, warm carcass weight, and dressing percentage were consistent with previous studies testing BIA (Berg and Marchello, 1994; Slanger et al., 1994) and slightly greater than those reported by Edwards et al. (1989) for ultrasound measurement. Table 3 reports simple coefficients of correlation between the live animal measurements and various carcass traits. The number of observations used in development of prediction equations differed for some of the methodologies. Ultrasonic measurements of LTL tissue depths were obtained on 82 market lambs (79 lambs were used for development of prediction equations); four data entries were mislabeled for BIA (n = 101) and five for linear body measurements (n = 100). Live weight showed the highest correlations to TDL (r =.77) and FFL (r =.74), bioelectrical Rs presented a significant ( P <.01) correlation to %TDL and %FFL (r =.63), and ULD had a correlation of.40 and.36 for TDL and FFL, respectively. Linear body measurements had poor correlations to carcass traits except for HG, which showed the highest correlation to TDL at.42. Tables 4, 5, and 6 report coefficients for predicting lamb carcass composition, the partial R 2 for each independent variable, and their actual P-values. Bioelectrical impedance measurements do not predict proportional carcass yield as accurately as carcass lean mass. Equations predicting percentage of boneless, closely trimmed primal cuts ( %BCTPC) showed an R 2 of.296 and RMSE of 2.53% (Table 4). Prediction of FFL with bioelectrical impedance measurements resulted in a R 2 (.674), which is lower than equations reported by Berg and Marchello (1994) for weight of fat-free mass ( FFM = LWT [LWT %Fat]; R 2 =.776), yet this study reports a lower RMSE (1.18 vs 1.97 kg; FFL vs FFM, respectively). Berg and Marchello (1994) did not report levels of significance of the independent variables within the prediction equations for FFM. Our study revealed LWT, Rs, and Xc to be significant at P <.001 within the prediction model. Table 4. Coefficients for predicting lamb carcass composition a from bioelectrical impedance measurement of live market weight lambs % BCTPC TDL, kg % TDL FFL, kg % FFL Item (n = 101) (n = 101) (n = 101) (n = 96) (n = 96) Intercept 49.8.224 57.05.186 52.47 Live wt, kg.018 (.0012,.682) b.254 (.5963,.001).100 (.0152,.075).241 (.5492,.001).061 (.0059,.279) Resistance, ohms.255 (.1934,.001).095 (.0217,.001).588 (.392,.001).108 (.0394,.001).613 (.3927,.001) Reactance, ohms.687 (.088,.001).403 (.0594,.001) 1.08 (.0723,.001).419 (.0723,.001) 1.16 (.0915,.001) Length, cm c.072 (.013,.243).053 (.0112,.067).312 (.0619,.001).055 (.0131,.059).312 (.0612,.001) R 2, RMSE.296, 2.53.689, 1.19.551, 3.27.674, 1.18.551, 3.45 b Partial R 2 and P-value of test that true coefficient value is zero, reported respectively, in parentheses. Length is the length between BIA (red) detector terminals.
2676 BERG ET AL. Table 5. Coefficients for predicting lamb carcass composition a from ultrasonic measurement of live market weight lambs % BCTPC TDL, kg % TDL FFL, kg % FFL Item (n = 79) (n = 79) (n = 79) (n = 78) (n = 78) Intercept 51.56.694 58.22.422 53.23 Live wt, kg.036 (.0036,.544) b.213 (.47,.001).095 (.0009,.305).207 (.4164,.001).054 (.2481,.589) US fat depth, cm 4.28 (.1021,.008).789 (.0074,.275) 11.1 (.2456,.001) 1.24 (.0199,.094) 12.01 (.0026,.001) US loin depth, cm.276 (.0016,.713) 1.12 (.0608,.002).349 (.0019,.765) 1.05 (.0535,.034).0341 (.0008,.781) R 2, RMSE.107, 2.86.538, 1.31.256, 4.46.490, 1.34.252, 4.67 b Partial R 2 and P-value of test that true coefficient value is zero, reported respectively, in parentheses. Live lamb BIA prediction of total carcass dissected lean has not been previously reported. Live weight, Rs, Xc, and LENG accounted for 68.9% of the variation in TDL, possessing a RMSE of 1.19 kg, and the same independent variables predict 55.1% of the variability of %TDL (RMSE = 3.27%) and %FFL (RMSE = 3.45%). Houghton and Turlington (1992) published a review of ultrasonic techniques for evaluation of carcass characteristics from live animal measurements taken on sheep, swine, and beef. Longitudinal ultrasonic scans of fat and loin muscle tissue depths have been documented for live swine (Gresham et al., 1994) but not for lambs. Table 5 shows the prediction equations derived from ultrasonic measurements of fat and loin muscle tissue depths obtained on live lambs. Edwards (1989) reported significant ( P <.01) coefficients of correlation between 12th rib ultrasonic fat depth and various primal and retail fabrication end points to range from.45 to.53. Similar correlations were observed in this study (Table 2) between the various carcass end points and UFD ranging from.32 to.50 ( P <.01). In this study, a correlation coefficient of.616 ( P <.001) was found between UFD and actual carcass fat depth measured adjacent the 12th costae. The review by Houghton and Turlington (1992) reported ultrasonic measurement of swine fat depth was much higher (r =.47.92). The variation among external fat depth is much larger in swine than in sheep. Fat depth measured adjacent the 10th costae of pork carcasses may span.5 to 5 cm, whereas the range of lamb carcass fat depth may only span.1 to 1 cm. A smaller prediction error associated to ultrasonic fat depth measurement may result in a lower correlation between actual and RU estimated fat depth. The final ultrasonic equations (Table 5) were similar to BIA equations whereby weights of carcass lean were more accurately predicted than percentages of carcass lean. The independent variables LWT and average ultrasonic loin depth (ULD) were significant ( P <.01) in equations predicting TDL and FFL, but UFD was not. The reverse was true for equations predicting carcass percentiles (%BCTPC, %TDL, and %FFL); UFD possessed the highest partial R 2 and was significant ( P <.001) within these equations. The relationship of external fat and LWT measurements to carcass yield was consistent with Edwards et al. (1989), who found LWT to be nonsignificant ( P >.05), whereas ultrasonic fat depth and visually estimated fat depth were significant ( P <.05) for prediction of proportional carcass yield. It is convenient that a measurement of fat depth is a significant predictor of carcass percentages, since the current USDA yield grade (USDA, 1992) is derived via a single fat-depth measurement adjacent the 12th Table 6. Coefficients for predicting lamb carcass composition a from linear measurements of live market weight lambs % BCTPC TDL, kg % TDL FFL, kg % FFL Item (n = 100) (n = 100) (n = 100) (n = 95) (n = 95) Intercept 46.36 1.5 58.56 1.5 55.40 Live wt, kg.014 (.0003,.848) b.271 (.5958,.001).020 (.0002,.857).262 (.549,.001).062 (.0026,.598) Hindsaddle, cm.084 (.0089,.340).031 (.0016,.442).046 (.0018,.736).089 (.01,.124).250 (.0124,.206) Shoulder ht, cm.146 (.0213,.029).020 (.0016,.512).196 (.041,.057).019 (.0015,.582).344 (.0833,.005) Heart girth, cm.050 (.0476,.324).010 (.0007,.686).258 (.0961,.002).002 (.0001,.930).260 (.0771,.002) R 2, RMSE.078, 2.91.600, 1.35.139, 4.52.561, 1.37.175, 4.68 b Partial R 2 and P-value of test that true coefficient value is zero, reported respectively, in parentheses.
PREDICTION OF LIVE LAMB COMPOSITION 2677 Table 7. Coefficients for predicting lamb carcass composition a from best fit combinations of electronic and linear measurements obtained on live market weight lambs (n = 78) Item % BCTPC TDL, kg % TDL FFL, kg % FFL Intercept 48.03.003 54.65 1.92 50.64 Live wt, kg.114 (.0229, 093) b.201 (.4502,.001).233 (.0256,.009).163 (.4164,.001).213 (.0193,.020) Resistance, ohms.26 (.2121,.001).112 (.038,.001).588 (.387,.001).127 (.0656,.001).62 (.4016,.001) Reactance, ohms.706 (.0755,.001).356 (.0619,.001) 1.07 (.0818, 001).401 (.0796,.001) 1.21 (.0995,.001) Length, cm.126 (.0214,.044).083 (.0359,.008).356 (.1001,.001).086 (.0355,.005).342 (.084,.001) US loin depth, cm 1.10 (.0638,.001) 1.77 (.0197,.036) 1.08 (.0536,.001) 1.72 (.0171,.047) Hindsaddle, cm.195 (.0556,.077) Shoulder ht, cm.233 (.0521,.001) 341 (.0419,.001).051 (.0133,.098).37 (.0523,.001) Heart girth, cm.183 (.0413,.003).193 (.0344,.003) R 2, RMSE.440, 2.31.650, 1.15.697, 2.91.664, 1.10.708, 3.0 b Partial R 2 and P-value of test that true coefficient value is zero, reported respectively, in parentheses. costae. Although UFD was significant within equations predicting proportional lean yield, the R 2 and RMSE remained less than desirable. Anatomical linear measurements are a simple means for monitoring animal growth (Kempster et al., 1982). Linear measurements were essentially ineffective predictors of proportional carcass yield (Table 5). Equations incorporating linear measurements for the prediction of TDL and FFL had R 2 of.60 and.561, respectively. In these equations, LWT acts as the only significant variable in the equation and has a partial R 2 of.5958 for TDL and.549 for FFL. A combination of these measurements may not be practical for a typical sheep production system; however, equations that combine BIA, RU, and linear measurements (Table 7) dramatically improve the predictive capacity for proportional carcass yield, especially %TDL (R 2 =.697; RMSE = 2.9%) and %FFL (R 2 =.708; RMSE = 3.0%). All independent variables used in the equations in Table 7 were significant at P <.1 according to STEPWISE regression. Although UFD was the only RU variable to be significant in ultrasonic equations (Table 5), STEP- WISE regression did not allow entry of UFD as a significant variable within the combination equations. Sex was regressed on the residual outputs for each prediction equation to determine its significance relative to the model error. The prediction equations developed in this study were not significantly biased by sex ( P >.05). Implications More uniform, lean breeding stock need to be selected to address the concerns within the lamb industry. Seasonal fluctuations in market prices (Ward, 1995) for lambs encourage the feeding of lambs past physiological maturity and appropriate market weight. Furthermore, the common procurement of lambs based on higher dressing percentage inadvertently gives incentive for over-fattening of the live animal. Harris et al. (1990) reported that lamb carcass fat had the largest influence on saleable product. Individual electronic methodologies tested in this study were moderate predictors of proportional carcass lean. Combination equations produced more accurate predictions of proportionate yield yet may not be practical. Modifications must be made to improve estimation techniques or new systems of evaluation must be identified. Literature Cited AOAC. 1990. Official Methods of Analysis (15th Ed.). Association of Official Analytical Chemists, Washington, DC. Berg, E. P., and M. J. Marchello. 1994. Bioelectrical impedance analysis for the prediction of fat-free mass in lambs and lamb carcasses. J. Anim. Sci. 72:322. Edwards, J. W., R. C. Cannell, R. P. Garrett, J. W. Savell, H. R. Cross, and M. T. Longnecker. 1989. Using ultrasound, linear measurements and live fat thickness estimates to determine the carcass composition of market lambs. J. Anim. Sci. 67:3322. Gresham, J. D., S. R. McPeake, J. K. Bernard, M. J. Riemann, R. W. Wyatt, and H. H. Henderson. 1994. Prediction of live and carcass characteristics of market hogs by use of a single longitudinal ultrasonic scan. J. Anim. Sci. 72:1409. Hall, C. B., H. C. Lukaski, and M. J. Marchello. 1989. Determination of rat body composition using bioelectrical impedance analysis. Nutr. Rep. Int. 39:627. Harris, J. J., J. W. Savell, R. K. Miller, D. S. Hale, D. B. Griffin, L. C. Beasley, and H. R. Cross. 1990. A national market basket survey for lamb. J. Food Qual. 13:453. Houghton, P. L., and L. M. Turlington. 1992. Application of ultrasound for feeding and finishing animals: A review. J. Anim. Sci. 70:930. Kempster, A. J., A. Cuthbertson, and G. Harrington. 1982. Carcass Evaluation in Livestock Breeding, Production and Marketing. Westview Press, Inc. Boulder, CO. Lasley, J. F. 1987. Genetics of Livestock Improvement (4th Ed.). Prentice-Hall, Englewood Cliffs, NJ. Magagna, J. 1991. Value-based marketing A panel discussion. Proc. Recip. Meat Conf. 44:133. NAMP. 1992. The Meat Buyers Guide. National Association of Meat Purveyors, Reston, VA. SAS. 1991. SAS Systems for Linear Models (3rd Ed.) SAS Inst. Inc., Cary, NC.
2678 BERG ET AL. Slanger, W. D., M. J. Marchello, J. R. Busboom, H. H. Meyer, L. A. Mitchell, W. F. Hendrix, R. R. Mills, and W. D. Warnock. 1994. Predicting total weight of retail-ready lamb cuts from bioelectrical impedance measurements taken at the processing plant. J. Anim. Sci. 72:1467. Swatland, H. J. 1984. Diagnostic electrical properties of meat. In: Structure and Development of Meat Animals. Prentice-Hall, Englewood Cliffs, NJ. TAMRC. 1991. Assessment of marketing strategies to enhance returns to lamb producers. TAMRC Commodity Market Research Report No. CM-1-91. Texas A & M University, College Station, TX. Umberger, S. H. 1994. Management and marketing practices altered through a value-based marketing program for slaughter lambs. Sheep Goat Res. J. 10:148. USDA. 1992. Standards for grades of lamb, yearling mutton and mutton carcasses and standards for grades of slaughter lambs, yearlings and sheep. Fed. Reg. 57 (97):21338. Ward, C. E. 1995. Seasonality in budgeting lamb feeding returns. Sheep Goat Res. J. 11:45.