Research Note Predicting Metabolizable Energy of Normal Corn from its Chemical Composition in Adult Pekin Ducks F. Zhao, 1 H. F. Zhang, S. S. Hou, and Z. Y. Zhang The State Key Laboratory of Animal Nutrition, Institute of Animal Sciences, Chinese Academy of Agricultural Sciences, Beijing 100094, China ABSTRACT Two experiments were conducted to establish crude fiber (r = 0.905), ADF (r = 0.915), and NDF (r = an ME content prediction model for normal corn for ducks based on the grain s chemical composition. In Experiment 1, observed linear relationships between the determined ME content of 30 corn calibration samples and proximate nutrients, acid detergent fiber (ADF), and neutral detergent fiber (NDF) were used to develop an ME prediction model. In Experiment 2, 6 samples of corn selected at random from the primary corn-growing regions of China were used for testing the accuracy of ME prediction models. The results indicated that the AME, AME n, TME, and TME n were negatively correlated with 0.95) contents, and moderately correlated with gross energy (GE; r = 0.55) content in corn calibration samples. In contrast, no significant correlations were found for CP, ether extract, and ash contents. According to the stepwise regression analysis, both NDF and GE were found to be useful for the ME prediction models. Because the maximum absolute difference between the in vivo ME determinations and the predicted ME values was 61 kcal/kg, it was concluded that, for White Pekin ducks, the latter could be used to predict the ME content of corn with acceptable accuracy. Key words: duck, corn, metabolizable energy, prediction model 2008 Poultry Science 87:1603 1608 doi:10.3382/ps.2007-00494 INTRODUCTION In 2005, the total yield of Chinese corn was >130,000,000 t. Corn is the principal energy source for ducks, comprising >40% by weight of the duck diet in China. To produce an accurate evaluation of the ME content of corn for poultry diet formulation, a considerable number of studies have been conducted to predict the ME content of corn based on its physical characteristics or chemical composition. Using the adult rooster as a test animal, many researchers have shown that the ME content of corn was correlated with its bulk density or chemical composition (Conner et al., 1976; Leeson et al., 1977; Mollah and Annison, 1981; Dale, 1994; NRC, 1994; Lessire et al., 2003). These results also indicated that the ME value of corn could be predicted, but few studies have been reported with ducks. In general, models that are based on chemical composition and used to predict ingredient ME value are more accurate than models based upon physical characteristics of the test ingredient. However, there has been no uniform model predicting the ME content of corn for birds based 2008 Poultry Science Association Inc. Received December 4, 2007. Accepted April 9, 2008. 1 Corresponding author: zsummit@163.com on the chemical composition. Several factors can affect the accuracy of ME prediction models, which subsequently influences their successful use. One such factor is the sample size for regression analysis; another is the representativeness of samples for the feedstuff as a whole. In some experiments aimed at predicting the ME content of raw materials, more than 25 samples were included in the regression analysis (Dale, 1994; Lessire et al., 2003). However, in other studies, the number of samples was less than 15 (Mollah and Annison, 1981). Prediction models from smaller sample sizes may have greater R 2 and less residual standard deviation (RSD), but may not be as accurate as other models developed with a greater number of samples (Carré, 1990). On the other hand, the range of ME and chemical composition contents of samples obviously affect the accuracy of the prediction model. For example, low variation in the ME and chemical composition contents of calibration sample set might provide an incorrect prediction model (Carré, 1990). Because the ME from corn contributes more than 40% of the total dietary ME content in diets typically fed to ducks in China, research to establish a model for predicting the AME and TME contents of corn for ducks should improve the accuracy of calculating ME in diet formulations for ducks. Therefore, this study utilized a series of calibration samples comprising corn and corn plus corn gluten meal, corn hulls, corn germ, and corn starch to establish an ME prediction model for White Pekin ducks. 1603
1604 ZHAO ET AL. Table 1. Composition and nutrient content of the corn-soybean mealbased diet fed during the wash-out period between ME determinations Item % Ingredient Corn 70.87 Soybean meal 23.37 Soybean oil 1.42 Sodium chloride 0.30 Limestone 1.23 Calcium phosphate 1.70 DL-Methionine 0.07 Lysine HCl 0.04 Vitamin-mineral premix 1 1.00 Analyzed nutrient content DM 89.68 CP 16.79 Crude fiber 3.07 Ether extract 4.12 Ash 5.66 Calculated nutrient content ME, 2 kcal/kg 2,950 Lys 0.82 Met 0.32 Calcium 0.90 Total phosphorus 0.60 1 Supplied per kilogram of diet: vitamin A (retinyl acetate), 2,500 IU; vitamin D 3, 400 IU; vitamin E (DL-α-tocopheryl acetate), 10 IU; vitamin K 3, 0.50 mg; thiamin, 1.80 mg; riboflavin, 4 mg; pyridoxine HCl, 3 mg; vitamin B 12 (cobalamin), 7 g; D-Ca pantothenate, 11 mg; nicotinic acid, 55 mg; folate, 0.50 mg; D-biotin, 0.12 mg; choline chloride, 750 mg; copper (CuSO 4 5H 2 O), 8 mg; iron (FeSO 4 7H 2 O), 80 mg; zinc (ZnSO 4 ), 40 mg; manganese (MnSO 4 H 2 O), 60 mg; selenium (Na 2 SeO 3 ), 0.15 mg; iodine (KI), 0.35 mg. 2 Calculated value according to the AME of roosters. MATERIALS AND METHODS Duck ME Assay All procedures were approved by the animal care and welfare committee of Institute of Animal Science, Chinese Academy of Agricultural Sciences, Beijing. The method of ME determination was similar to the TME bioassay described by Sibbald (1976) and partly modified to account for the difference in digestive physiology between rooster and duck as shown by studies in our lab (Fan, 2003). The modifications to Sibbald s bioassay include feed withdrawal of all birds for 36 h before feeding test samples, use of 60 g of feedstuff for force feeding, and a 36-h period of excreta collection. In a 14-d wash-out period between ME trials, water and a corn-soybean mealbased diet (Table 1) were available for ad libitum consumption. Endogenous energy losses were determined using 4 replicates of 3 ducks per replicate during each ME trial. Four kilograms of each sample were made and ground through a 2-mm screen before pelleting. Pellets, 4 mm in diameter and 6 mm long, were prepared by regulating the ratio of water to feedstuff with a laboratory nonsteam press pellet mill, and were then air-dried until the water content was <14% before force feeding. A stainless steel funnel with a narrow stem (40 cm long and 1.0 cm inner diameter) was used for force feeding. The collection method of excreta was in accordance with that described by Adeola et al. (1997). In each ME trial, ducks were placed in individual cages (0.45 m 0.38 m 0.51 m) in a temperature-controlled room (25 C) and provided with 12 h of light daily. Experimental Design Experiment 1. The objective of this experiment was to determine the relationship between ME and the chemical composition of 30 corn calibration samples to develop a prediction model for ME that could be utilized for the formulation of diets for White Pekin ducks. The corn calibration samples were made by combining different percentages of corn, corn gluten meal, corn hulls, corn germ, and corn starch (Table 2) to provide a wide distribution of proximate nutrient compositions that spanned the range of values previously observed for 427 samples of Chinese normal corn, excluding high-oil corn. Separate ME trials were conducted under similar conditions from October 2005 to January 2006 to measure the ME contents of the each of the 30 corn calibration samples. One hundred and twenty 18-wk-old White Pekin drakes of similar weight (3.8 to 4.0 kg) provided by the Waterfowl Research Center of Chinese Academy of Agricultural Sciences (Beijing) were selected and randomly divided into 10 groups of 12 birds each. Each group contained 4 replicates of 3 ducks per replicate. One of the 10 groups was used for the determination of endogenous losses and each of the 9 remaining groups was used to determine the ME content of 1 calibration sample. After the ME determinations of the first 9 samples (numbers 1 to 9 in Table 2) were conducted, there was a 14-d wash-out period in which ducks were provided with free access to water and a corn-soybean meal-based diet (Table 1) formulated to meet the National Research Council (1994) requirements. Then, the same 120 ducks were randomly reassigned into 10 groups of 12 birds (4 replicates of 3 ducks) to determine endogenous losses and the ME of samples 10 to 18 (Table 2), followed by a 14-d wash-out period. Subsequently, the same 120 ducks were again randomly reassigned into 10 groups of 12 birds (4 replicates of 3 ducks) to determine endogenous losses and the ME of samples 19 to 27 (Table 2), followed by a 14-d wash-out period. Finally, 48 of the same 120 ducks were randomly selected and assigned into 4 groups of 12 birds (4 replicates of 3 ducks) to determine endogenous losses and the ME of samples 28 to 30 (Table 2). Experiment 2. The objective of this experiment was to test the suitability of various models to predict the ME content of a corn sample based on its chemical composition. Six corn cultivars were randomly selected from the main growing areas of China (Table 3) to test the accuracy of ME prediction model established in Experiment 1. The ME contents of each of the 6 corn cultivars were determined in vivo using 4 replicates of three 18-wk-old White Pekin ducks (3.8 to 4.0 kg) each per sample. Chemical Analysis After completion of excreta collection in each ME trial, all excreta samples were dried at 65 C for 48 h, then re-
RESEARCH NOTE 1605 Table 2. Composition of calibration samples, Experiment 1 Ingredients, % Analyzed nutrients on DM basis 1 Calibration Corn Corn Corn Corn Ether sample Corn gluten meal hull germ starch CF CP extract Ash ADF NDF GE AME AME n TME TME n % kcal/kg 1 42.24 8.01 9.78 39.97 1.9 10.8 2.3 0.8 2.9 11.1 4,493 3,610 3,535 3,873 3,711 2 45.08 6.86 20.15 27.91 3.1 11.3 3.0 0.9 5.1 16.9 4,571 3,419 3,342 3,679 3,515 3 54.08 6.00 20.80 19.12 3.5 11.5 3.4 1.0 5.4 18.4 4,621 3,316 3,241 3,575 3,414 4 66.19 0.26 15.44 18.11 2.9 8.5 3.1 1.1 4.1 15.4 4,488 3,417 3,365 3,683 3,543 5 68.35 3.33 0.27 28.05 1.1 8.9 2.4 1.1 1.8 6.0 4,458 3,851 3,779 4,111 3,952 6 71.08 3.52 8.15 17.25 2.2 10 3.0 1.1 3.5 11.8 4,518 3,560 3,492 3,820 3,665 7 74.42 0.28 24.41 0.89 3.5 9.9 4.5 1.1 6.8 21.8 4,642 3,214 3,162 3,477 3,338 8 74.61 3.15 11.87 10.37 2.3 10.6 3.2 1.1 4.0 13.9 4,558 3,509 3,447 3,773 3,623 9 75.65 10.87 1.16 12.32 2.4 8.7 3.7 1.1 3.8 13.6 4,517 3,482 3,431 3,743 3,605 10 76.72 0.09 22.67 0.52 3.4 9.8 4.2 1.2 6.4 21.1 4,590 3,253 3,193 3,515 3,368 11 76.91 2.20 15.13 5.76 2.8 10.7 3.6 1.1 4.5 15.8 4,582 3,473 3,395 3,739 3,572 12 77.94 20.77 0.21 1.08 3.7 9.7 4.3 1.2 6.2 20.4 4,599 3,274 3,222 3,534 3,396 13 78.21 4.92 16.17 0.70 3.5 13 4.6 1.2 5.6 16.6 4,664 3,415 3,327 3,679 3,504 14 79.24 2.02 12.82 5.92 2.8 10.5 3.5 1.2 4.3 15.3 4,584 3,468 3,395 3,729 3,569 15 81.00 0.87 17.76 0.37 3.5 10.2 4.3 1.4 5.7 18.3 4,623 3,448 3,386 3,708 3,559 16 82.25 4.38 11.40 1.97 2.5 12.6 3.8 1.2 4.2 14.5 4,598 3,480 3,405 3,739 3,578 17 83.02 9.30 2.64 5.04 2.5 9.2 4.8 1.5 4.2 14.7 4,608 3,570 3,508 3,828 3,680 18 83.03 0.92 15.17 0.88 3.4 10.4 4.2 1.1 5.5 17.5 4,603 3,412 3,346 3,681 3,525 19 83.51 3.28 12.54 0.67 2.9 11.9 3.9 1.3 4.4 15.1 4,643 3,500 3,424 3,761 3,598 20 83.52 1.27 13.68 1.53 2.9 10.8 4.4 1.1 4.9 16.6 4,610 3,436 3,368 3,702 3,545 21 84.90 0.18 14.86 0.06 3.3 10 3.8 1.1 5.2 16.5 4,587 3,451 3,384 3,718 3,562 22 85.15 6.68 1.49 6.68 2.2 9 4.3 1.3 3.3 12.4 4,541 3,556 3,503 3,818 3,678 23 85.15 1.11 9.00 4.74 2.4 9.5 3.2 1.3 3.9 13.2 4,489 3,462 3,405 3,723 3,579 24 85.36 2.01 9.58 3.05 2.8 10.8 5.3 1.2 4.6 14.5 4,654 3,506 3,460 3,767 3,635 25 85.53 0.50 12.77 1.20 2.7 10.2 4.3 1.3 4.3 15.0 4,582 3,429 3,365 3,689 3,538 26 86.63 9.20 0.35 3.82 2.4 9.2 3.8 1.4 4.1 14.0 4,506 3,514 3,455 3,776 3,630 27 86.74 1.58 11.60 0.08 2.6 10.9 3.3 1.2 4.2 14.7 4,595 3,425 3,376 3,686 3,550 28 88.50 7.41 2.85 1.24 2.3 9.1 5.0 1.4 3.8 12.6 4,579 3,561 3,507 3,820 3,680 29 91.62 2.09 4.19 2.10 2.0 10.5 4.8 1.3 3.4 11.9 4,592 3,669 3,598 3,932 3,774 30 100.00 1.6 9.3 3.6 1.3 2.5 9.4 4,489 3,712 3,639 3,977 3,816 Mean 2.7 10.3 3.9 1.2 4.4 15.0 4,573 3,480 3,415 3,742 3,590 SD 0.6 1.1 0.7 0.2 1.1 3.3 55 130 128 130 128 1 Mean of 3 determinations per sample. CF = crude fiber; ADF = acid detergent fiber; NDF = neutral detergent fiber; GE = gross energy. equilibrated with air for 24 h and ground through a 0.5- mm screen before analysis. The feedstuff and excreta were analyzed for DM (method 934.01), CP (method 954.01), crude fiber (CF; method 962.09), ether extract (method 920.39), ash (method 942.05), and nitrogen (method 955.04) using procedures of the AOAC (1990). Energy contents of feedstuff and excreta were determined by using a Parr 1281 automatic adiabatic calorimeter (Parr Instrument Co., Moline, IL). The ADF and NDF contents of feedstuffs were determined according to the procedure described by Van Soest (1963) and Van Soest et al. (1991), respectively. The AME, AME n, TME, and TME n of the samples were calculated according to the procedure described by Adeola et al. (1997). Statistical Analysis Possible relationships between chemical composition and ME content were analyzed with correlation and step- Table 3. Growth location, variety, and nutrients of test corn samples, Experiment 2 Test corn samples Item 1 2 3 4 5 6 Growth location Heilongjiang Shanxi Hebei Shandong Shaanxi Xingjiang Variety Baidan 9 Nongda 108 Haideng 3 Yedan 981 Zhengdan 958 Dika 656 Color White Yellow Yellow White Yellow Yellow Nutrient, 1 DM basis GE, kcal/kg 4,491 4,546 4,541 4,524 4,508 4,520 CP, g/kg 85 85 94 102 86 81 Ether extract, g/kg 39 45 41 37 39 36 CF, g/kg 16 15 13 15 13 11 Ash, g/kg 22 12 14 16 13 12 NDF, g/kg 94 85 65 79 63 61 ADF, g/kg 33 27 25 25 23 22 1 Mean of 3 determinations per sample. CF = crude fiber; ADF = acid detergent fiber; NDF = neutral detergent fiber; GE = gross energy.
1606 ZHAO ET AL. Table 4. Correlation coefficients 1 between chemical composition 2 and ME of calibration samples, Experiment 1 Ether CF CP extract Ash ADF NDF GE AME AME n TME CP 0.35 Ether extract 0.38 0.05 Ash 0.08 0.26 0.60 ADF 0.96 0.31 0.43 0.05 NDF 0.95 0.26 0.38 0.07 0.98 GE 0.71 0.62 0.69 0.16 0.72 0.67 AME 0.90 0.23 0.26 0.15 0.91 0.95 0.54 AME n 0.91 0.29 0.24 0.18 0.92 0.95 0.56 0.99 TME 0.90 0.23 0.26 0.15 0.91 0.95 0.54 0.99 0.99 TME n 0.91 0.29 0.24 0.17 0.92 0.95 0.55 0.99 0.99 0.99 1 Correlation whose absolute value is more than 0.38 is significantly different from zero at P < 0.05. 2 CF = crude fiber; ADF = acid detergent fiber; NDF = neutral detergent fiber; GE = gross energy. wise regression analysis using the CORR and REG procedures of SAS (SAS Institute, 1990). The correlation and regression coefficients were considered different from zero at P < 0.05. Residual standard deviation was used for measuring the goodness-of-fit of linear models. The smaller was RSD, the better fitting was the model. This was done by the procedure described by Kaps and Lamberson (2004). RESULTS AND DISCUSSION As planned, the chemical composition of calibration samples varied to a great extent in Experiment 1 (Table 1). The mean gross energy (GE) content was 4,573 kcal/ kg of DM, and ranged from 4,458 to 4,664 kcal/kg of DM, which was similar to that in a previous study by Lessire et al. (2003). The mean CP content was 103 g/kg of DM with a range from 85 to 130 g/kg of DM. The mean CF, ADF, and NDF contents were 27, 44, and 150 g/kg of DM with ranges from 11 to 37, 18 to 68, and 60 to 218 g/ kg of DM, respectively. The mean ether extract content was 39 g/kg of DM and ranged from 23 to 53 g/kg of DM. The mean ash was 12 g/kg of DM and ranged from 8 to 15 g/kg of DM. The values of these nutrients from corn published in Chinese Feed Database (CFD, 2005) were in the range obtained in our experiment. The mean AME n was 3,415 kcal/kg of DM, which was similar to the value of 3,503 kcal/kg of DM in Pekin ducks reported by King et al. (1997). The CFD (2005) and NRC (1994) AME n values for corn in cockerel were 3,740 and 3,764 kcal/kg of DM, respectively, which is in the range of 3,162 to 3,779 kcal/kg obtained for calibration samples in our experiment. However, the mean AME n of 30 sam- ples in our study was less than that of 37 corn samples in adult cockerels reported by Lessire et al. (2003), which might be due in part to the high-oil corn used. In our study, the variation of GE in calibration samples was less than that of the 4 ME measures, which suggested that the difference in chemical composition could affect the availability of energy of corn in ducks. This phenomenon was in accordance with a study with cockerels reported by Dale (1994). The range of chemical composition and ME contents in 30 calibration samples was greater than that of 37 corn samples observed by Lessire et al. (2003), Leeson et al. (1993), Dale (1994), and Dale and Jackson (1994). This result was advantageous for establishing a ME prediction model according to the results of Carré (1990). The correlation coefficients between AME and AME n, TME, and TME n of calibration samples were significantly high (0.99; P < 0.05; Table 4). These results were in accordance with the findings of Sibbald (1982), Lessire et al. (2003), and Francesch et al. (2002). Our study also indicated that nitrogen-corrected values were also proportional to the AME or TME value in the duck feedstuffs. Therefore, the 4 ME measures (AME, AME n, TME, and TME n ) in the calibration samples had almost the same relationship with their chemical composition. Correlation analyses (Table 4) showed that the AME, AME n, TME, and TME n were highly negatively correlated with CF, ADF, and NDF contents, and moderately correlated with GE content (average of the 4 ME r values were 0.905, 0.915, 0.95, and 0.55, respectively) in the calibration samples. In contrast, no significant correlations were found with CP, ether extract, and ash content. The GE content was also correlated positively with CF, ADF, Table 5. Equations of prediction of the ME (kcal/kg of DM) values of corn according to neutral detergent fiber (NDF) and gross energy (GE) contents (%, DM basis), Experiment 1 ME Equation R 2 RSD, 1 kcal/kg P-value AME 2,299.1 41.6 NDF + 0.394 GE 0.9181 38.6 <0.0001 AME n 2,509.8 40.4 NDF + 0.330 GE 0.9200 37.6 <0.0001 TME 2,606.0 41.4 NDF + 0.384 GE 0.9154 39.2 <0.0001 TME n 2,708.2 40.3 NDF + 0.325 GE 0.9188 37.8 <0.0001 1 Residual standard deviation.
RESEARCH NOTE 1607 Table 6. Comparison of ME contents in corn determined by using the in vivo method and prediction model, 1 Experiment 2 AME, kcal/kg AME n, kcal/kg TME, kcal/kg TME n, kcal/kg Test corn Observed Predicted Difference Observed Predicted Difference Observed Predicted Difference Observed Predicted Difference 1 3,662 3,678 16 3,621 3,612 9 3,922 3,941 19 3,794 3,789 5 2 3,679 3,738 58 3,637 3,667 30 3,939 4,000 61 3,811 3,843 32 3 3,827 3,817 9 3,768 3,746 22 4,086 4,081 5 3,941 3,922 19 4 3,771 3,752 18 3,698 3,684 14 4,030 4,016 14 3,871 3,860 11 5 3,835 3,812 22 3,778 3,743 35 4,096 4,076 20 3,952 3,919 33 6 3,823 3,825 3 3,776 3,755 21 4,081 4,089 8 3,948 3,931 17 Statistics Mean 3,766 3,770 3,713 3,701 4,026 4,034 3,886 3,877 SD 78 58 72 57 77 59 71 56 P-value 0.7509 0.2504 0.5285 0.3724 1 The observed ME values were determined with the in vivo method; the predicted ME values were calculated according to the neutral detergent fiber and gross energy contents of corn. NDF, CP, and ether extract contents (average of r values were 0.71, 0.72, 0.67, 0.62, and 0.69, respectively). However, the ME content was negatively correlated with GE content in our study, which did not agree with the results of Lessire et al. (2003). In those chemical compositions that were significantly correlated with the ME of calibration samples, any 2 of NDF, CF and ADF had highly significant correlations between themselves (r 0.95). Correlation between the ME and NDF contents was greatest among those 3 fiber contents, which indicated that the effect of CF and ADF contents on the ME content of calibration samples could be explained by NDF content. Therefore, the present results indicated that the ME content of corn calibration samples in adult ducks might be largely dependent on NDF and GE content. Similar observations have also been reported on the ME content of corn in cockerels (Lessire et al., 2003). Using correlations and stepwise regression analysis, the equations to predict ME content of corn in ducks were established according to the significant linear relationship between ME, NDF, and GE content (Table 5). The equations based on NDF and GE contents for predicting AME, AME n, TME, and TME n contents had high accuracy with R 2 > 0.90 and RSD <40 kcal/kg, which indicated that only less than 10% of the observed variation in the ME content of corn calibration samples resulted from factors other than NDF and GE content. This suggested that the accuracy of the prediction model for ME was close to that observed in vivo during a classic metabolizable energy experiment, which was in accordance with the previous results of Lessire et al. (2003) and Dale (1994), who used cockerels as the test animal. To test the suitability of these models (Table 5) to predict the ME content of a normal corn sample, the ME content of 6 samples of corn was measured by both the in vivo method and prediction models. Our results showed that the chemical compositions (Table 3) and ME content determined by using the in vivo method of 6 samples were all in the range of the 30 corn calibration samples (Table 2). The maximum absolute difference between ME determined by the in vivo method and the prediction model was 61 kcal/kg (Table 6). The fluctuation of ME measurements among 4 replications with the in vivo method in one sample of corn ranged from 31 to 178 kcal/kg. This suggested that the accuracy of prediction models for ME was close to that obtained in vivo. Therefore, the prediction models established from 30 corn calibration samples as described in this article can be used to predict the ME content of corn for ducks with acceptable accuracy. ACKNOWLEDGMENTS We gratefully acknowledge the financial support of Basic Science Research Program (ywf-td-4), State Science and Technique Support Project (2006BAD12B01 1), and the State Commonwealth Research Project (2005DIB4J033) in China. We also wish to thank W. Huang, L. Zhao, Y. W. Dong, Q. J. Wang, X. H. Jiang, M. Zhao,
1608 ZHAO ET AL. and R. P. Wang (Institute of Animal Sciences, Chinese Academy of Agricultural Sciences, Beijing) for their help in force feeding and excreta samples collection. REFERENCES Adeola, O., D. Ragland, and D. King. 1997. Feeding and excreta collection techniques in metabolizable energy assays for ducks. Poult. Sci. 76:728 732. Association of Official Analytical Chemists (AOAC). 1990. Official Methods of Analysis. Association of Official Analytical Chemists, Washington, DC. Carré, B. 1990. Predicting the dietary energy value of poultry feeds. Pages 283 300 in Feedstuff Evaluation. J. Wiseman, and D. J. A. Cole, ed. Butterworths, London, UK. Chinese Feed Database (CFD). 2005. Tables of Feed Composition and Nutritive Values for Poultry in China. The Center of Chinese Feed Database Information, Beijing. http://www. Chinafeeddata.org.cn Accessed May 2007. (in Chinese) Connor, J. K., A. R. Neill, and K. M. Barram. 1976. The metabolizable energy content for the chicken of maize and sorghum grain hybrids grown at several geographical regions. Aust. J. Exp. Agric. Anim. Husb. 16:699 703. Dale, N. 1994. Relationship between bushel weight, metabolizable energy, and protein content of corn from an adverse growing season. J. Appl. Poult. Res. 3:83 86. Dale, N., and D. Jackson. 1994. True metabolizable energy of corn fractions. J. Appl. Poult. Res. 3:179 183. Fan, H. P. 2003. Comparative study of the digestion of feed nutrients between cockerel and drake. MS thesis. Chinese Academy of Agricultural Sciences, Beijing. Francesch, M., K. Bernard, and J. M. McNab. 2002. Comparison of two direct bioassays using 3-week-old broilers to measure the metabolisable energy of diets containing cereals high in fibre: Differences between true and apparent metabolisable energy values. Br. Poult. Sci. 43:580 587. Kaps, M., and W. Lamberson. 2004. Biostatistics for Animal Science. CABI Publishing, Cambridge, UK. King, D., D. Raglang, and O. Adeola. 1997. Apparent and true metabolizable energy values of feedstuffs for ducks. Poult. Sci. 76:1418 1423. Leeson, S., J. D. Summers, and T. R. Daynard. 1977. The effect of kernel maturity at harvest as measured by moisture content, on the metabolizable energy value of corn. Poult. Sci. 56:154 156. Leeson, S., A. Yersin, and L. Volker. 1993. Nutritive values of 1992 corn crop. J. Appl. Poult. Res. 2:208 213. Lessire, M., J. M. Hallouis, B. Barrier-Guillot, M. Champion, and N. Femenias. 2003. Prediction of the metabolisable energy value of maize in adult cockerel. Br. Poult. Sci. 44:813 814. Mollah, Y., and E. F. Annison. 1981. Prediction of metabolisable energy of wheat, maize and sorghum in poultry diets from chemical compositon. Proc. Nutr. Soc. Aust. 6:137. National Research Council. 1994. Nutrient Requirements of Poultry. 9th rev. ed. Natl. Acad. Press, Washington, DC. SAS Institute. 1990. SAS/STAT User s Guide: Statistics. SAS Institute Inc., Cary, NC. Sibbald, I. R. 1976. A bioassay for true metabolizable energy in feedingstuffs. Poult. Sci. 55:303 308. Sibbald, I. R. 1982. Measurement of bioavailable energy in poultry feeding stuffs: A review. Can. J. Anim. Sci. 62:983 1048. Van Soest, P. J. 1963. Use of detergents in the analysis of fibrous feeds. II. A rapid method for the determination of fiber and lignin. J. Am. Off. Anal. Chem. 46:829 835. Van Soest, P. J., J. B. Robertson, and B. A. Lewis. 1991. Methods for dietary fiber, neutral detergent fiber, and nonstarch polysaccharides in relation to animal nutrition. J. Dairy Sci. 74:3583 3597.