| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
ANIMAL GROWTH, PHYSIOLOGY, AND REPRODUCTION |
* Department of Animal Science, Cornell University, 134 Morrison Hall, Ithaca, NY 14853;
and
Department of Animal Science, Texas A&M University, College Station 77843; and
and
Department of Animal Science, Pennsylvania State University, University Park 16802
 |
Abstract |
|---|
 Top Abstract INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONIMPLICATIONSLITERATURE CITED |
|---|
Key Words: beef bull • body composition • carcass trait • dry matter required • ultrasound
|
INTRODUCTION |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONIMPLICATIONSLITERATURE CITED |
|---|
; Herd et al., 2003), and feed consumption accounts for the majority of this cost. Therefore, feed cost per unit of beef produced is directly related to profitability. However, it is difficult to determine individual consumption because of the cost of individually feeding large numbers of cattle. A mathematical model has been developed (Perry and Fox, 1997; Guiroy et al., 2001; Tedeschi et al., 2004) to predict DM required (DMR) by individuals fed in groups based on the growth rate and mature size of the animal, and using readily available carcass data obtained at harvest to predict energy content of the ADG. Carcass and live animal, ultrasound traits in steers and heifers have been used to develop equations to successfully predict carcass composition (Perry and Fox, 1997; Guiroy et al., 2001).
The accuracy of live animal, ultrasound prediction of carcass traits varies between studies, but has been shown to be moderately accurate (Perry and Fox, 1997; Hassen et al., 1998; Charagu et al., 2000). However, equations have not been developed and tested for the use of live animal, ultrasound measurements to predict body composition and DMR for the observed performance of yearling bulls fed in groups. Therefore, animal breeders lack adequate information regarding the prediction of efficiency of gain in young sires and replacement females fed in groups simply because carcass measurements are not available. Additionally, measurement of rump fat is routinely made with ultrasound on live animals, but there is a lack of information on whether its use would increase the accuracy in predicting body composition.
The objectives of this study were to 1) develop an equation based on live animal, ultrasound measurements to predict carcass fat (CF) and EBF in yearling bulls fed in groups, and 2) evaluate this equation using data from an independent database of individually fed bull calves.
|
MATERIALS AND METHODS |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONIMPLICATIONSLITERATURE CITED |
|---|
Development of Equations To Predict Empty Body Fat (Exp. 1)
One-hundred eighteen spring-born purebred and crossbred bull calves were purchased from farms in New York, Vermont, and West Virginia. Purebred bull calf breeds included Angus (n = 25), Hereford (n = 36), Limousin (n = 2), and Simmental (n = 10). The remaining bulls included crosses of British x British (n = 23) and British x Continental (n = 20) breed types. During a 4-wk adaptation period, all bulls were adjusted to a corn-based diet (Table 1). Based on the visual observation of 2 trained evaluators, the BW at which each bull would be expected to reach USDA low Choice Quality grade (QG) was estimated. These individual projections were entered into a dynamic growth model (Tedeschi et al., 2004) to project days to reach USDA low Choice QG for individual animals. Using these projected values, at the end of the adaptation period the bulls (BW = 288 ± 4.3 kg) were blocked into 3 market groups (4 pens/group) based on the predicted days to finish and assigned to 12 slatted-floor pens (9 to 10 head/pen). The bulls were fed once per day, orts were measured daily, and the bulls were weighed every 28 d.
|
When 80% of the individuals in a market block pen were estimated by visual appraisal to have reached the USDA low Choice QG, the pen was slaughtered at a commercial packing facility. Carcasses were chilled for 24 to 48 h. Carcass data including HCW, backfat over the 12 to 13th rib (BF), marbling score (MRB), and LM area (LMA) were collected on the left half of the chilled carcass.
The 9th to 11th-rib section of the left side of each carcass was removed and dissected into soft tissue and bone using the procedures of Hankins and Howe (1946). Chemical fat was determined by ether extract of the soft tissue and used to predict CF (Eq. 1; Hankins and Howe, 1946) and EBF (Eq. 2; Garrett and Hinman, 1969), as follows:
 | [1] |
 | [2] |
where pCF is predicted carcass fat, expressed as percentage of the carcass; chemFAT is chemical fat from the 9th to 11th rib, expressed as percentage; and pEBF is predicted EBF, expressed as percentage of empty body weight (EBW).
As recommended by the NRC (2000), full BW was converted to shrunk BW (SBW) using a factor of 96%, and SBW was converted to EBW using a factor of 89.1%. Then, predicted HCW (pHCW) was computed from EBW using Eq. 3 (Garrett and Hinman, 1969):
 | [3] |
where pHCW is predicted HCW (kg).
The mixed model procedure of SAS (Version 9.1, SAS Inst. Inc., Cary, NC) was used to generate multiple regressions to predict EBF from live animal, ultrasound measurements and carcass traits. Residual analyses were performed to evaluate the adequacy and assumptions of the multiple regressions. A preliminary analysis indicated no outliers. We utilized Monte Carlo simulation to perform a sensitivity analysis of each variable of the regression to predict EBF. The mean and SD of each independent variable were used, and a normal distribution was assumed. The regression, standard slope coefficient was estimated using @Risk (Palisade Corp., Newfield, NY) to assess the impact on the dependent variable per unit of change of the SD of the independent variable.
Evaluation of the Equations Developed To Predict Empty Body Fat (Exp. 2)
In Exp. 2, 40 spring-born crossbred bull calves were fed in individual pens. The bull calves were obtained from the Cornell Teaching and Research Center beef herd. The majority (n = 26) of the calves were sired by Angus bulls bred to English crossbred dams; the remainder were sired by Angus bulls bred to Continental crossbred cows (n = 6) or Gelbvieh sires bred to English crossbred cows (n = 8).
Thirty days after weaning, all bull calves (BW = 287 ± 5.8 kg) were adjusted to a corn-based diet (Table 1
). The bulls were fed in individual slatted-floor pens once daily. The remaining procedures were similar to the those used in Exp. 1; orts were measured daily, the calves were weighed every 28 d, and live animal, ultrasound measurements of uBF, uLMA, uRmpFt, and IMF were taken at about 365 d of age (125 d on feed). As in Exp. 1, the bulls were harvested when 80% of the bulls were visually estimated to have reached USDA low Choice QG. Carcass data were collected after a 48-h chill, and the 9th to 11th-rib section was removed from the left side of the carcass for prediction of CF (Hankins and Howe, 1946) and EBF (Garrett and Hinman, 1969).
Evaluation of the equations was performed according to the techniques described by Tedeschi (2006), including linear regression analysis (Neter et al., 1996) of EBF predicted from rib fat (Garrett and Hinman, 1969) and EBF predicted from carcass and ultrasound measurements, and included the concordance correlation coefficient (Lin, 1989), mean bias, and mean square error of prediction (Bibby and Toutenberg, 1977).
Evaluation of Equations To Predict Dry Matter Required
The equations developed in Exp. 1 to predict EBF along with animal and diet information were used in the growth model of Tedeschi et al. (2004), as implemented in the Cornell value discovery system, to predict DMR, which was evaluated with the individual feed intake data measured in Exp. 2. The DMR is calculated when animal BW and ADG are known. The model-predicted DMI is iterated using a spline interpolation algorithm by automatically changing the relative DMI factor until the mean predicted ADG matches the mean observed ADG. Then this DMI renamed to DMR because this is the amount of feed required to obtain the observed mean ADG. The growth model depends on a user-input for expected BW at the target body composition (adjusted final BW, AFBW; Eq. 4) to compute equivalent SBW and energy requirement for gain:
 | [4] |
where AFBW is adjusted SBW (kg); EBW is empty BW (kg); and EBF is empty body fat (%EBW).
In this study, we compared DMR predicted with AFBW at 28% EBF computed from EBF calculated with Eq. 1 and 2 or from live animal, ultrasound-estimated carcass variables and Eq. 3.
|
RESULTS AND DISCUSSION |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONIMPLICATIONSLITERATURE CITED |
|---|
. Of the 118 bulls that started the experiment, 5 died, 1 carcass was retained at the packing plant, and chemical composition data on 1 carcass was not reported. The DMR consumed by these bulls was estimated using the procedures of Tedeschi et al. (2004) and removed from the pen total consumption. The animal performance and carcass composition of the bulls are shown in Table 2. The ADG was 1.5 kg/d and average DMI was 8.69 kg/d. The bulls averaged near the target of a small degree of marbling. However only 23% of the bulls graded USDA Choice because of a high incidence of dark cutters, and 43.6% of the bulls were B maturity. Cattle with B-maturity carcasses must have modest or greater marbling to grade USDA Choice.
|
(n = 80). Carcass data were removed from 30 animals because they were not available at the packing plant for collecting the rib sections. The data from another bull were removed because it was considered an outlier based on the Studentized residuals. The relationship between carcass measurements and pEBF is shown in Eq. 5, which had an r2 of 0.72, and a root mean squared prediction error (RMSPE) of 2.07:
|
 | [5] |
where pEBF is EBF predicted from 9th to 11th-rib composition (% of EBW); BF is back fat (cm); HCW is hot carcass weight (kg); MRB is marbling, a visual determination of intramuscular fat based on the USDA photographic reference standards of the degrees of marbling of the LD between the 12th and 13th ribs that takes into consideration amount, texture, and distribution; and LMA is LM area (cm2).
The equations developed by Hankins and Howe (1946) have undergone several evaluations. Nour and Thonney (1994) used measured carcass chemical composition to evaluate the equations of Hankins and Howe (1946) to predict carcass composition and concluded these equations could be used with confidence. In comparing 6 methods of predicting carcass composition, Powell and Huffman (1968) reported that Hankins and Howe (1946) predicted carcass fat most accurately. In Powell and Huffman (1968), the carcass weights of the cattle were relatively constant, yet differed widely in yield grade and carcass fat. This would indicate that the cattle represented a range in biological type. Finally, Crouse and Dikeman (1974) evaluated Hankins and Howe (1946) comparing measured and predicted carcass composition using calves sired by Hereford, Simmental, and Limousin bulls. Within this group of cattle varying widely in biological type, these researchers reported the Hankins and Howe (1946) equation to be accurate with no breed group interactions for carcass fat.
All coefficients in Eq. 5 were different from zero (P < 0.01). These variables were the same as those used by Guiroy et al. (2001) to develop an equation to predict EBF for steers fed high concentrate diets. However, the r2 in our equation was greater than that reported by Guiroy et al. (2001; 0.72 vs. 0.61, respectively), and RMSPE was less (2.07 vs. 3.52%, respectively), suggesting improved accuracy in using these variables to predict EBF for bulls. This is likely due to the similarity in breed type and background of the bulls from these studies compared with the steers used by Guiroy et al. (2001).
However, carcass measurements are not available on breeding bulls, and the usefulness of Eq. 5 for practical application is limited. To evaluate the adequacy of using HCW predicted using Equation 3, pHCW was substituted for observed HCW. The relationship between carcass measurements, pHCW, and pEBF is shown in Eq. 6, which had an r2 of 0.72 and an RMSPE of 2.14:
 | [6] |
where pEBF is EBF predicted from 9th to 11th-rib composition (% of EBW); BF is back fat (cm); pHCW is predicted hot carcass weight (kg) from Garrett and Hinman (1969), pHCW = (EBW –30.26)/1.362; MRB is marbling; and LMA is LM area (cm2).
Based on similar r2 and RMSPE, ultrasound measurements were regressed on predicted EBF from 9th to 11th-rib section fat. Ultrasound measured rump fat, back fat, LMA, and intramuscular fat were included in the regression (Eq. 7):
 | [7] |
where pEBF is EBF predicted from 9th to 11th-rib composition (% of EBW); uBF is ultrasound back fat (cm); uRmpFt is ultrasound rump fat (cm); uLMA is ultrasound LM area (cm2); and pHCW is predicted HCW (kg); and IMF is ultrasound intramuscular fat.
Because the regression coefficient for uRmpFt was not significantly different (P = 0.14) from zero, we removed it from the final equation. These results differ from Bergen et al. (2005), who reported a reduction in prediction error when using ultrasound rump fat to predict carcass lean of yearling bulls. Equation 8 shows the coefficients for uBF, uLMA, and pHCW to compute pEBF; this equation had an r2 of 0.62, and an RMSPE of 2.49:
 | [8] |
where pEBF is EBF predicted from 9th to 11th-rib composition, (% of EBW); uBF is ultrasound back fat (cm); uLMA is ultrasound LM area (cm2); and pHCW is predicted HCW (kg). The partial sum of squares for each component was 74.2, 149.2, 56.4, and 43.0 for uBF, uLMA, pHCW, and the interaction between uBF and pHCW, respectively.
The number of days from ultrasound measurement to harvest was 15, 29, and 71 d for the 3 harvest groups. There was no effect (P = 0.63) of time from ultrasound to harvest on the prediction of EBF. This is in agreement with Charagu et al. (2000), who reported no effect of time period from scanning to slaughter on measures of bias and deviation between ultrasound and carcass measurements for periods up to 15 d. Hassen et al. (1999) reported that ultrasound measures taken at 365 d and used in end product prediction models had similar or better R2 and RMSPE as ultrasound measures taken at slaughter.
The P value for all coefficients in the equation developed from this experiment using ultrasound (Eq. 8) were all less than 0.01. The variables BF and LMA had the greatest impact in the prediction of EBF. In Eq. 6, BF and LMA had a standard slope coefficient of 0.839 and –0.441, respectively, indicating that EBF would change by 0.839 or –0.441 SD units per unit of SD change in BF and LMA, respectively. Marbling and HCW had standard slope coefficients of 0.278 and 0.175, respectively. Similarly, in Eq. 8 uBF and uLMA would affect the predictions of EBF by 0.869 and –0.444 SD units, respectively, and 1 SD unit in the calculated HCW would affect EBF by only 0.157 SD units. Therefore, the effects of BF and LMA were similar and consistent with uBF and uLMA in affecting the prediction of EBF. Additionally there was an interaction between BF and pHCW. Although the time between ultrasound and harvest did not affect the prediction of EBF, the ultrasound measured values for BF and LMA were smaller than those measured in the carcass.
Unlike the equation reported by Guiroy et al. (2001) in which QG was significant in predicting EBF, IMF was not significant (P = 0.57) in predicting EBF for bulls (Eq. 5). In the current study the range in ultrasound measured IMF was 2.2 to 5.5, reflecting a range in QG of low Standard to low Choice. Although the correlation of carcass and ultrasound traits was in the range of published data (Hassen et al., 1998; Charagu et al., 2000), the correlation between marbling and IMF in Exp. 1 was low (Table 3). This would limit its significance in the regression. Additionally, in the data set used by Guiroy et al. (2001), the range in QG was Standard to Prime+. Therefore, the smaller range in IMF in our bull database might explain the lack of impact of IMF in predicting EBF.
|
equation resulted in a r2 of 0.71, a mean bias of –16.4% and a mean predicted EBF of 28.2%. Thus, the equation developed by Guiroy et al. (2001) predicted greater EBF in yearling bulls. Because both incorporate the same equation to predict EBF from CF, this is due to the differences in the effect of the input variables (BF, HCW, LMA, QG) on the prediction of CF. We expected HCW to have less influence in bulls because they have less fat content at the same BW compared with steers (Owens et al., 1995).
Evaluation of the Equations Developed To Predict Empty Body Fat (Exp. 2)
The measured composition of the diet fed to the 37 individually fed bulls in the evaluation experiment (DM = 64.9%, ME = 2.9 Mcal/kg of DM, CP = 13.2% DM) was similar to that fed to the pen-fed bulls (Table 1). Three bulls were removed from the study: one for lameness, a second with excessive trim at the harvest facility, and a third that escaped loading. Additionally, based on the residual analysis, 1 animal determined to be an outlier was removed from the study.
Table 2 summarizes the performance and composition of the individually fed bulls. They were similar to the pen-fed bulls in initial BW, ADG, and DMI. However they were heavier at harvest, on feed longer and slightly more efficient. The bulls from both experiments were similar in BF and YG, but the bulls from Exp. 2 had heavier HCW and larger LMA and averaged a higher percentage of carcass fat. As in Exp. 1 the animals averaged the target of small degree of marbling; the low numbers of bulls grading USDA Choice was due to the high percentage (56.7%) of B-maturity carcasses.
The mean chemical composition of the 9th to 11th-rib section of bulls from Exp. 2 (Table 4) indicated their carcasses had a greater proportion of chemical fat with less variation as shown by a lower standard deviation and range in minimum to maximum values than bulls from Exp. 1. This occurred even though all the bulls from Exp. 2 were harvested at a constant time end point. These bulls were all from the same herd, thus being more genetically similar compared with the bulls from Exp. 1, which were from 20 different farms and breeding programs.
Comparisons of prediction of EBF using the bull carcass equation (Eq. 5), bull carcass equation using pHCW (Eq. 6) bull ultrasound equation (Eq. 8), and the steer carcass equation (Guiroy et al., 2001) are shown in Table 5. The bull carcass equation clearly provided the most accurate and precise prediction of EBF as shown by the less RMSPE (2.07) and greater r2 (0.76). Even though the steer carcass equation (Guiroy et al., 2001) had good precision (r2 of 0.74), accuracy was low (RMSPE of 3.89) and mean bias (–12%) was high, indicating an overprediction of EBF.
|
, Guiroy et al. (2001), and Tedeschi et al. (2004). The predicted EBF either from ultrasound measurements or carcass traits is used to adjust BW to a common body composition, assuming a fixed change in BW per unit of EBF 14.26 kg/% EBF). Then an equivalent BW (BW at the same composition as the standard reference animal) is computed, and prediction of energy required for growth is predicted with the equations recommended by the NRC (2000). Usually, an AFBW is computed for 28% EBF, which is close to the low Choice USDA grade (Guiroy et al., 2001). This AFBW accounts for differences in frame size and carcass composition (NRC, 2000).
Evaluation of the Prediction of DMR
The dynamic growth model developed by Tedeschi et al. (2004) as implemented in CVDS was used to predict DMR based on the observed ADG of the bulls from Exp. 2. Table 5 shows that the accuracy and precision of the carcass and ultrasound equations (Eq. 5 and 8) in predicting DMR was similar. These data suggested that ultrasound is as effective in predicting DMR as are equations that require actual carcass weight and measurements.
Using DMR and ADG, efficiency of gain was predicted using each method. Our analysis (Table 5) indicated similar r2 and RMSPE for predicting efficiency of gain when using carcass traits or ultrasound measurements from Eq. 5 and 8, respectively. We did not evaluate predicted DMR with observed ADG and BW adjusted for composition of gain. Tedeschi et al. (2006) found that using these variables alone to evaluate differences in efficiency of gain was inadequate because of differences in body composition at the same BW.
Brethour (2004) used ultrasound measurements to compute deposition of energy and protein of steers. The author indicated no correlation between ultrasound BF and feed conversion or efficiency of gain, suggesting that more complex models are required to account for more of the variation in energy content of the ADG.
Nutrition models can play an important role in identifying efficient animals for genetic selection purposes (Tedeschi et al., 2006). The rank of bulls is more valuable for evaluation and selection for genetic merit than actual efficiency of gain. Therefore, the question becomes are the most efficient and least efficient bulls identified using the equations developed in this study? The model was able to correctly identify 3 of the 4 most and least efficient bulls, using the ultrasound equation and pHCW to estimate AFBW. The use of the bull carcass equation (Eq. 5) and equation of Guiroy et al., (2001) and actual HCW to predict AFBW resulted in correctly identifying 2 of the 4 most and least efficient bulls. Expanding the range to include the top and bottom 25% of the bulls resulted in all 3 equations being able to correctly identify 6 of 9 least efficient bulls and 7 of 9 most efficient bulls.
|
IMPLICATIONS |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONIMPLICATIONSLITERATURE CITED |
|---|
|
Footnotes |
|---|
2 Corresponding author: mjb28{at}cornell.edu
Received for publication January 5, 2006. Accepted for publication May 27, 2006.
|
LITERATURE CITED |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONIMPLICATIONSLITERATURE CITED |
|---|
This article has been cited by other articles:
|
|
 |
|
  P. A. Lancaster, G. E. Carstens, F. R. B. Ribeiro, M. E. Davis, J. G. Lyons, and T. H. Welsh Jr. Effects of divergent selection for serum insulin-like growth factor-I concentration on performance, feed efficiency, and ultrasound measures of carcass composition traits in Angus bulls and heifers J Anim Sci, November 1, 2008; 86(11): 2862 - 2871. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
