Purdue Pork Page Archive
Prediction of Daily Protein Accretion in Pigs from
Estimates of Fat-Free Lean Gain
A.P. Schinckel, P.V. Preckel and M.E. Einstein
Departments of Animal Sciences and Agricultural Economics
There has been interest in determining the maximum protein accretion rates which can be achieved under commercial pork production conditions. There are currently two economical methods for estimating on-farm protein accretion rates.
The first method is to obtain serial B-mode (real-time) ultrasonic backfat depth and loin muscle area measurements to predict body component mass from 50 to 260 lb live weight. Prediction equations from the 1991 lean growth trial have been developed to predict fat-free lean mass, total carcass fat, empty body protein and empty body lipid mass.
The suggested recommendation is to obtain data on 32 pigs of the same sex and genotype group. Pigs should be weighed every 3 weeks starting at weaning. Pigs should be measured with B-mode ultrasound every 3 weeks starting at approximately 50 lb, at two weeks prior to slaughter, and at slaughter. Live weight data are fit to a three-parameter, variable inflection, non-linear function utilizing a weighted least-squares analysis. The predicted body component mass data are fit to the augmented allometric function by using a log to log transformation. Use of this method will result in approximate mean absolute errors of 2.5% for empty body protein accretion at each live weight as derived by bootstrapping simulation procedures.
The second method to economically predict on-farm empty body protein accretion curves is from an estimate of the mean fat-free lean gain from 50 to 260 lb. At least 24 pigs of the same sex should be identified at 50 lb and days from 50 to 260 lb slaughter weight calculated. The fat-free percentage values can be obtained from a pork processor and multiplied by the warm carcass weight of each pig to estimate fat-free lean mass at slaughter.
The mean fat-free lean growth rate (g/d) can be used to estimate the daily protein accretion curve by using a generalized exponential form. At the request of the National Research Council modeling committee, research on the accuracy of the generalized function was conducted. Fat-free lean gain was calculated as the fat-free lean mass either measured or predicted at 250 to 260 lb minus the fat-free lean mass predicted by B-mode ultrasound at 45 to 55 lb. Daily fat-free lean gain was calculated as the fat-free lean gain divided by the days on test. Previous analyses have shown that a relationship which is both flexible and descriptive of biological responses has the form:
PA = A x exp(B x WT + C/WT + D x WT2)
where PA = daily protein accretion in g/d; WT = live weight in kg; and A, B, C and D are estimated parameters. A simple generalization of the above relationship between protein accretion and live weight, which allows changes in the coefficients as a function of the mean fat-free lean growth from 50 to 260 lb, is the following:
PA = (a + ãM x exp [(b + b1M) x WT + (c + c1M)/WT + (d + d1M) x WT2]
where M denotes mean fat-free lean gain from 50 to 260 lb; a, ã, b, b1, c, c1, d and d1 are parameters; and the other notation is as before. With this generalized form, each of the estimated parameters from the line-specific relationship has been expressed as a linear function of M (i.e., the parameter A is replaced by a + ãM, B is replaced by b + b1M, etc.).
The parameter values computed as the solutions to these curve-fitting problems are displayed in Table 1 for gilts and barrows. The predicted daily protein accretion rates for groups of barrows with different mean fat-free lean gains are shown in Figure 1. Regression of the genotype-environment specific protein accretion on the predicted protein accretion rates did not indicate any significant biases. The intercept values of 1.2 ± 1.2 and 1.1 ± 1.0 g/d for gilts and barrows, respectively, were not significantly (P>.10) different from zero. The regression coefficients of .992 ± .01 and .991 ± .01 were not significantly different from one.
The mean percentage absolute values of the errors from 45 to 260 lb live weight were 3.5% for gilts and 6.1% for barrows. The errors were largest between 240 and 260 lb live weight. From 45 to 240 lb live weight, the mean percentage errors averaged 2.7% for gilts and 4.8% for barrows.
Weight of fat-free lean mass at 50 lb live weight can be estimated from a prediction equation using live weight as the only independent variable (NPPC, 1991). This assumes that the variation in lean content of 45 to 55 lb pigs is very small even when different genotypes, sexes or management environments are represented. Data from the 1991 cooperative lean growth trial resulted in a mean fat-free lean mass of 19.4 lb at 57 lb live weight (unpublished data). The fourteen genotype-sex means, adjusted for live weight, had a standard deviation of 1.2 lb and a range of 16.2 to 22 lb, indicating that potential errors in estimating initial fat-free lean mass could occur by using only live weight as a predictor of pretest fat-free lean mass. Also, the accuracy and degree of bias associated with fat-free lean mass estimates at slaughter significantly influence estimates of fat-free lean gain. The majority of pork processors are utilizing either optical probe measurements or a single midline backfat thickness measurement with warm carcass weight to predict percent fat-free lean. Significant genotype and sex biases occur when predicting dissected or fat-free lean mass utilizing prediction equations including only these independent variables (Gu et al., 1992; Wagner, 1992; Wagner et al., 1993).
A method was developed for estimating daily protein accretion rates from fat-free lean growth data. Pork producers can utilize fat-free lean index data reported by pork processors and on-farm growth data to estimate daily protein accretion rates. One limitation of the method is potential biases in estimating fat-free lean at slaughter.
Gu, Y., A.P. Schinckel, T.G. Martin, J.C. Forrest, C.H. Kuei and L.E. Watkins. 1992. Genotype and treatment biases in lean estimation of swine carcass. J. Anim. Sci. 70:1708-1718.
NPPC. 1991. Procedures to evaluate market hogs. Third edition.
Wagner, J.R. 1992. Genotype and sex biases in the estimation of pork carcass composition. M.S. Thesis. Purdue Univ. West Lafayette, IN.
Wagner, J.R., A.P. Schinckel and J.C. Forrest. 1993. Genotype and sex biases in the estimation of pork carcass composition. Proc. National Swine Imp. Fed. St. Louis, MO. pp 47.
Table 1. Estimated parameters for the generalized function by sex.
|Sum of squares||2312||8619|
Purdue Pork Page Archive