Voi 5 (1996): 17-23. Predicting body weight from body measurements of pre-pubertal Ayrshire heifers Päivi Mäntysaari Agricultural Research Centre ofFinland, Institute ofAnimalProduction, FIN-31600 Jokioinen, Finland The relationship between heart girth, wither height, body length and body weight in 3- to 9.5-month- old pre-pubertal Finnish Ayrshire heifers gaining 600-650 g/d was analysed (experiment I). Regres- sion analysis showed that heart girth was the trait most highly correlated to body weight (R 2 = 0.969). Including body length or wither height as a second term in the regression, increased R 2 values only slightly. When the relationship between heart girth and body weight was used to predict the body weight of heifers reared at two feeding levels (experiment II), the precision of prediction was affect- ed by the plane of nutrition. Actual body weight for a given heart girth was slightly higher on the high than on the low feeding level. It is, nevertheless, concluded that the equations presented in the paper can be used to estimate accurately the body weight of pre-pubertal (95-140-cm heart girth) Ayrshire heifers gaining 550-700 g/d. Key words: heart girth, wither height, body length, feeding level, replacement heifer ntroduction Body weight (BW) measurements of replace- ment dairy heifers are important for monitoring the growth rate, optimizing nutrient allowance and determining a suitable size for breeding. Growth rate needs to be monitored, because the feeding level before puberty can affect mamma- ry development (Sejrsen et al., 1982; Harrison et al., 1983; Niezen et al., 1992; Mäntysaari et al., 1995) and hence the subsequent milk-pro- ducing ability of heifers (Little and Kay, 1979; Foldager and Sejrsen, 1991). Since nutrient re- commendations are based on BW and daily gain. it is necessary to know the live weight to opti- mize the nutrient intake of heifers. Since most farmers do not have animal scales for measuring BW directly, an alternative method of estimating BW is needed. Among those used have been measurements such as heart girth, wither height, body length and hip width (Jo- hansson and Hildeman, 1954;Kenttämies et al., 1974; Kenttämies and Vehmaan-Kreula, 1978; Verma and Hussain, 1985; Sprensen and Fol- dager, 1991; Heinrichs et al. 1992). In many cases, heart girth has been the trait most highly correlated to BW (Johansson and Hildeman, 1954;Nelson, et al., 1985; Heinrichs et al. 1992). However, the relationship between heart girth © Agricultural and Food Science in Finland Manuscript received February 1996 17 AGRICULTURAL AND FOOD SCIENCE IN FINLAND https://www.c-info.fi/en/info/?token=wNPH9zkWe9LizSJK.5pM7Zf5xx4gblm0AHYiyMA.g43utiGJ_FI6YZkiRV345Nr7U9RhI1XHTkZZFoEyLEe8P6Jse6Sn3NvsOKhaujBKqhq-a2_D-DT3RUgUH3zuF_4hhYerMQjbAnMuAU8vfagRl_A0nWfocm2UaFEti0-DALossN_ezZDtyo60hnJ3wp7XXNYMYo6MGRPbo_RChwYR62yIxgwWmjKtQbch8j3ym8IKr8IIQeHfUbiL2rFSgbkR6eRPG0NSVAJzgH5jLLgDHuLZp2NTk3LoKopqaP5YNN9DrL1kwAZ7JS_HvPF8aHA Mäntysaari, P.: Predicting body weightfrom body measurements and BW may differ between breeds (Kenttämies and Vehmaan-Kreula, 1978; Sorensen and Fol- dager, 1991), animals of different age (Johansson and Hildeman, 1954) and animals on different feeding levels (Hvidsten, 1940; Johansson and Hildeman, 1954; Sorensen and Foldager, 1991). In Finland, the relationship used to estimate BW from the heart girth of heifers is based on investigations by Kenttämies and Vehmaan- Kreula (1978). In their study they excluded heif- ers that were younger than 6 months. However, the critical rearing period, when feeding inten- sity may affect mammary development, begins at 3 months of age (Sejrsen and Foldager, 1991). The purpose of this study was to investigate the value of heart girth, wither height and body length in predicting the BW ofpre-pubertal Finn- ish Ayrshire heifers fed restricted diets. Material and methods Animals, diets and measurements Experiment I A group of 51 Finnish Ayrshire heifers were reared to gain 650 g/d from 3 months of age and 98 kg BW to 9.5 months of age and 225 kg BW. All heifers were fed the same diet, which includ- ed silage, hay, barley (when BW < 200 kg) and rapeseed meal (when BW < 130 kg). Average feed, energy and protein intakes are given in Table 1. The body weight, heart girth, wither height and body length (from the point of the shoul- ders to the pinbone) of the heifers were meas- ured every four weeks. All measurements were carried out by the same person. The total number of observations for each measurement was 408. Experiment II A 2x2 factorial experiment was conducted on 24 Finnish Ayrshire heifers. The factors were two feeding levels and two protein sources. The av- erage age and weight of the heifers at the begin- ning were 3 months and 87 kg and at the end of the experiment 9 months and 221 kg. The aver- age gain of the heifers on the low feeding level was 668 g/d and on the high feeding level 848 g/d. The heifers were fed hay and barley supple- mented with rapeseed meal or urea. Feed, ener- gy and protein intakes by feeding level are giv- en in Table 1. During the experiment the body weight, heart girth and wither height were measured by the same person every four weeks. More detailed information about diets, feed intake, growth, slaughter weight and carcass quality of the heif- ers is given by Mäntysaari (1993). The mamma- ry development of the heifers is presented by Mäntysaari et al. (1995). Statistical analysis The best prediction equation for BW from the body measurements was determined from the data of experiment I. After the fit of different functions had been tested, the best fit equation BW = elbi M i (1) where BW is body weight, M. body meas- urement and b the corresponding parameter, was chosen to estimate BW from body measure- ments. Equation 1 was transformed to a linear statistical model ln(BW) = IbM. (2) Regressions of the ln(BW) on heart girth, wither height and body length using individual observations were performed. Models were fit- ted using the GLM procedure of SAS (1987). Linear and quadratic effects of the independent variables were considered. The base model was: ln(BW)jk = b 0 + b,M.k +b2 M jk 2 +Sj + ejk (3) where BWjk is body weight, b 0 the intercept, Mjk the kth body measurement (heart girth, wither height or body length) of the j,h heifer, S- the ef- fect of the heifer, b, and b 2 are regression coeffi- cients, and e. is the residual associated with thejk k ,h measurement of the jth heifer. ln(BW) was also regressed on multiple inde- pendent variables. Since the R 2 values from the 18 AGRICULTURAL AND FOOD SCIENCE IN FINLAND Vol. 5 (1996): 16-23. Table 1.Age, body weight, growth and feed intake of the experimental heifers. Experiment I Experiment II Low High x std x std x std N 51 12 12 Initial age, d 92 11.6 88 9.3 87 6.5 Final age, d 288 11.6 293 21.1 249 20.3 Initial BW, kg 98 12.6 86 10.7 87 10.5 Final BW, kg 225 16.2 220 3.9 221 3.9 Daily gain, g 647 52.6 668 45.7 848 71.3 Intake, kg/d Dry matter 3.59 0.190 3.73 0.088 3.91 0.306 Hay 0.56 0.124 2.33 0.093 1.64 0.343 Silage 2.13 0.178 - - - - Concentrate 0.89 0.122 1.39 0.024 2.27 0.060 Nutrient NE,fu/d 3.28 0.157 2.73 0.055 3.34 0.150 CP,g/d 479 14.7 492 9.0 561 28.8 AATg/d 321 13.7 327 14.8 353 35.6 BW, body weight; NE, net energy in fattening feed units; CP, crude protein; AAT, amino acids absorbed from the small intestine (Madsen, 1985). base regression showed that heart girth was the trait most highly correlated to ln(BW), heart girth was chosen as the first independent variable. A linear term for the second and third independent variables was added to the model. Thus, the models were: ln(BW)jk = b 0 + b,(HG)jk + b 2(HG)jk 2 + b3(BL or WH) jk +S j + e jk (4) or: ln(BW). =b 0 + b,(HG)jk + b2 (HG) jk 2 + b,(BL)jk + b4(WH)jk + S- + ejk (5) where(BW)jk is body weight, b 0 the intercept, (HG). k heart girth, (BL). k body length, (WH)jk wither height, Sj heifer, b,, b 2, b 3 and b 4 are re- gression coefficients, and ejk is the residual. The equation to predict BW, which was de- termined from the data of experiment I, was test- ed with the dataof experiment 11. The difference between predicted and actual BW was calculated and presented graphically. The goodness of fit of the equation between treatments was evaluated by testing the difference (predicted - actual BW) using the MIXED procedure of SAS (1992). The model was: ym = +Fi+ Pj + (F*P) y + Bijk8ijk +T, + c ijkl (6) where y is the prediction error, F, the i ,h feed- ing level, P. the j* protein source, S ijk the kth heifer on feeding level i and protein source j, T, the Ith1 th measuring time, and e jjk| the residual associated with observation of the kth heifer on the i ,h feed- ing level and jth protein source at time 1. Results and discussion Regression of body weight on different body measurements The increase with age in the BW, heart girth, wither height and body length of the heifers in experiment I is presented in Figure 1. Correla- tions between BW and body measurements were high (r > 0.93). In agreement withprevious stud- ies (Johansson and Hildeman, 1954; Nelson et al., 1985; Heinrichs et al., 1992) heart girth was the trait most highly correlated to BW (r = 0.97). 19 AGRICULTURAL AND FOOD SCIENCE IN FINLAND Mäntysaari, P.: Predicting body weightfrom body measurements 80 100 120 140 160 180 200 220 240 260 280 300 age, d Body weight, kg Body length, cm Figure 1. Development of body weight, heart girth, experiment I. Body weight is estimated from body meas- urements using the exponential function (Johans- son and Hildeman, 1954; Kenttämies et al. 1974; Kenttämies and Vehmaan-Kreula, 1978, SO- - and Foldager, 1991), although Verma and Hussain (1985) predicted the BW of calves us- ing the linear relationship of heart girth to BW. Since the purpose of the present study was to establish the relationship of body measurements to the BW of pre-pubertal heifers, the ranges in BW and body measurements were relatively small. The linerregression of hearth girth, with- er height or body length, therefore, estimated the BW fairly accurately (R 2 > 0.914). The fit was, however, improved when equation 1 was used to predict BW from body measurements. The R 2 values for the linear regression of heart girth, wither height and body length to ln(BW) were 0.967, 0.929 and 0.945, respectively (Table 2). The additional quadratic terms were highly sig- nificant for heart girth and wither height although the increases in R 2 values were small. Including them in the regression would, however, improve the prediction. Johansson and Hildeman (1954) concluded —I- Heart girth, cm —s- Wither height, cm body length and wither height of the heifers in that BW can be estimated with about the same accuracy from heart girth only as with a combina- tion of two or several measurements. Likewise Heinrichs et al. (1992) found that an additional linear term gave only little predictive benefit. Here, too, the addition of the linear effect of body length or wither height to the regression includ- ing the linearand quadratic effects of heart girth improved the fit only slightly (Table 3). The in- clusion of all three independent variables in the equation increased the R 2 value by 0.014. If a higher accuracy of the BW estimation is Table 2. Regressions of ln(BW) on various body measure- ments. Measurement Intercept Linear Quadratic R 2 Heart girth -2.1183 0.02307"' - 0.967 1.0453 0.04196'” -0,00007713"'0.969 Body length 2.2654 0.02821' 1 0.945 2.2512 0.02851'" -0.00000154 0.945 Wither height 1.3685 0.03711'" - 0.9291.36850.03711•" - 0.929 -0.5716 0.07688'" -0.00020253" 0.931 *** P< 0.001 **P