Predicting metabolizable energy of normal corn from its chemical composition in adult Pekin ducks.

Abstract

Two experiments were conducted to establish 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 crude fiber (r = -0.905), ADF (r = -0.915), and NDF (r = -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.

Full text

Page 1

Research Note
Predicting Metabolizable Energy of Normal Corn from its Chemical
Composition in Adult Pekin Ducks
F. Zhao,1H. 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
lish 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 re-
gions of China were used for testing the accuracy of ME
prediction models. The results indicated that the AME,
AMEn, TME, and TMEnwere negatively correlated with
Twoexperimentswereconductedtoestab-
Key words: duck, corn, metabolizable energy, prediction model
2008 Poultry Science 87:1603–1608
doi:10.3382/ps.2007-00494
INTRODUCTION
In2005,thetotalyieldofChinesecornwas>130,000,000
t. Corn is the principal energy source for ducks, compris-
ing >40% by weight of the duck diet in China. To produce
an accurate evaluation of the ME content of corn for poul-
try diet formulation, a considerable number of studies
have been conducted to predict the ME content of corn
based on its physical characteristics or chemical composi-
tion. Using the adult rooster as a test animal, many re-
searchers 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 Anni-
son, 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 composi-
tion and used to predict ingredient ME value are more
accurate than models based upon physical characteristics
ofthetestingredient.However,therehasbeennouniform
model predicting the ME content of corn for birds based
©2008 Poultry Science Association Inc.
Received December 4, 2007.
Accepted April 9, 2008.
1Corresponding author: zsummit@163.com
1603
crude fiber (r = −0.905), ADF (r = −0.915), and NDF (r =
−0.95) contents, and moderately correlated with gross en-
ergy (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 maxi-
mumabsolutedifferencebetweentheinvivoMEdetermi-
nations 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.
onthechemicalcomposition.Severalfactorscanaffectthe
accuracy of ME prediction models, which subsequently
influences their successful use. One such factor is the
sample size for regression analysis; another is the repre-
sentativeness 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 mod-
els from smaller sample sizes may have greater R2and
less residual standard deviation (RSD), but may not be
as accurate as other models developed with a greater
number of samples (Carre ´, 1990). On the other hand,
the range of ME and chemical composition contents of
samples obviously affect the accuracy of the prediction
model.Forexample,lowvariationintheMEandchemical
composition contents of calibration sample set might pro-
vide an incorrect prediction model (Carre ´, 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
AMEandTMEcontentsofcornforducksshouldimprove
the accuracy of calculating ME in diet formulations for
ducks.Therefore, thisstudyutilized aseries ofcalibration
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.

Page 2

ZHAO ET AL.
1604
Table 1. Composition and nutrient content of the corn-soybean meal-
based diet fed during the wash-out period between ME determinations
Item%
Ingredient
Corn
Soybean meal
Soybean oil
Sodium chloride
Limestone
Calcium phosphate
DL-Methionine
Lysine?HCl
Vitamin-mineral premix1
Analyzed nutrient content
DM
CP
Crude fiber
Ether extract
Ash
Calculated nutrient content
ME,2kcal/kg
Lys
Met
Calcium
Total phosphorus
70.87
23.37
1.42
0.30
1.23
1.70
0.07
0.04
1.00
89.68
16.79
3.07
4.12
5.66
2,950
0.82
0.32
0.90
0.60
1Supplied per kilogram of diet: vitamin A (retinyl acetate), 2,500 IU;
vitamin D3, 400 IU; vitamin E (DL-α-tocopheryl acetate), 10 IU; vitamin
K3, 0.50 mg; thiamin, 1.80 mg; riboflavin, 4 mg; pyridoxine?HCl, 3 mg;
vitamin B12(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 (CuSO4?5H2O), 8 mg; iron (FeSO4?7H2O), 80 mg; zinc (ZnSO4),
40 mg; manganese (MnSO4?H2O), 60 mg; selenium (Na2SeO3), 0.15 mg;
iodine (KI), 0.35 mg.
2Calculated value according to the AME of roosters.
MATERIALS AND METHODS
Duck ME Assay
All procedures were approved by the animal care and
welfarecommittee ofInstituteofAnimal 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 ac-
count 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 pe-
riod between ME trials, water and a corn-soybean meal-
based diet (Table 1) were available for ad libitum con-
sumption. 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
regulatingthe ratioofwater tofeedstuffwith alaboratory
nonsteam press pellet mill, and were then air-dried until
the water content was <14% before force feeding. A stain-
less 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)inatemperature-controlledroom(25°C)andprovided
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,andcornstarch(Table2)toprovideawidedistribu-
tion of proximate nutrient compositions that spanned the
range of values previously observed for 427 samples of
Chinese normal corn, excluding high-oil corn.
SeparateMEtrialswereconductedundersimilarcondi-
tions from October 2005 to January 2006 to measure the
MEcontentsoftheeachofthe30corncalibrationsamples.
One hundred and twenty 18-wk-old White Pekin drakes
ofsimilarweight(3.8to4.0kg)providedbytheWaterfowl
Research Center of Chinese Academy of Agricultural Sci-
ences (Beijing) were selected and randomly divided into
10 groups of 12 birds each. Each group contained 4 repli-
cates of 3 ducks per replicate. One of the 10 groups was
usedforthedeterminationofendogenouslossesandeach
of the 9 remaining groups was used to determine the ME
content of 1 calibration sample. After the ME determina-
tions 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
10groupsof12birds(4replicatesof3ducks)todetermine
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
10groupsof12birds(4replicatesof3ducks)todetermine
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 composi-
tion. Six corn cultivars were randomly selected from the
maingrowing areasof China(Table3) totest theaccuracy
of ME prediction model established in Experiment 1. The
ME contents of each of the 6 corn cultivars were deter-
mined 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-

Page 3

RESEARCH NOTE
1605
Table 2. Composition of calibration samples, Experiment 1
Ingredients, % Analyzed nutrients on DM basis1
Calibration
sample
CornCorn
hull
Corn
germ
Corn
starch
Ether
extractCorn gluten mealCF CP AshADFNDFGEAME AMEn
TME TMEn
%kcal/kg
3,535
3,342
3,241
3,365
3,779
3,492
3,162
3,447
3,431
3,193
3,395
3,222
3,327
3,395
3,386
3,405
3,508
3,346
3,424
3,368
3,384
3,503
3,405
3,460
3,365
3,455
3,376
3,507
3,598
3,639
3,415
128
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Mean
SD
42.24
45.08
54.08
66.19
68.35
71.08
74.42
74.61
75.65
76.72
76.91
77.94
78.21
79.24
81.00
82.25
83.02
83.03
83.51
83.52
84.90
85.15
85.15
85.36
85.53
86.63
86.74
88.50
91.62
100.00
8.01
6.86
6.00
0.26
3.33
3.52
0.28
3.15
—
0.09
2.20
—
4.92
2.02
0.87
4.38
—
0.92
3.28
1.27
0.18
—
1.11
2.01
0.50
—
1.58
—
2.09
—
9.78
20.15
20.80
15.44
—
8.15
24.41
11.87
10.87
22.67
15.13
20.77
16.17
12.82
17.76
11.40
9.30
15.17
12.54
13.68
14.86
6.68
9.00
9.58
12.77
9.20
11.60
7.41
4.19
—
—
—
—
—
0.27
—
0.89
—
1.16
—
—
0.21
0.70
—
0.37
—
2.64
0.88
0.67
1.53
0.06
1.49
—
3.05
1.20
0.35
0.08
2.85
2.10
—
39.97
27.91
19.12
18.11
28.05
17.25
—
10.37
12.32
0.52
5.76
1.08
—
5.92
—
1.97
5.04
—
—
—
—
6.68
4.74
—
—
3.82
—
1.24
—
—
1.9
3.1
3.5
2.9
1.1
2.2
3.5
2.3
2.4
3.4
2.8
3.7
3.5
2.8
3.5
2.5
2.5
3.4
2.9
2.9
3.3
2.2
2.4
2.8
2.7
2.4
2.6
2.3
2.0
1.6
2.7
0.6
10.8
11.3
11.5
8.5
8.9
10
9.9
10.6
8.7
9.8
10.7
9.7
13
10.5
10.2
12.6
9.2
10.4
11.9
10.8
10
9
9.5
10.8
10.2
9.2
10.9
9.1
10.5
9.3
10.3
1.1
2.3
3.0
3.4
3.1
2.4
3.0
4.5
3.2
3.7
4.2
3.6
4.3
4.6
3.5
4.3
3.8
4.8
4.2
3.9
4.4
3.8
4.3
3.2
5.3
4.3
3.8
3.3
5.0
4.8
3.6
3.9
0.7
0.8
0.9
1.0
1.1
1.1
1.1
1.1
1.1
1.1
1.2
1.1
1.2
1.2
1.2
1.4
1.2
1.5
1.1
1.3
1.1
1.1
1.3
1.3
1.2
1.3
1.4
1.2
1.4
1.3
1.3
1.2
0.2
2.9
5.1
5.4
4.1
1.8
3.5
6.8
4.0
3.8
6.4
4.5
6.2
5.6
4.3
5.7
4.2
4.2
5.5
4.4
4.9
5.2
3.3
3.9
4.6
4.3
4.1
4.2
3.8
3.4
2.5
4.4
1.1
11.1
16.9
18.4
15.4
6.0
11.8
21.8
13.9
13.6
21.1
15.8
20.4
16.6
15.3
18.3
14.5
14.7
17.5
15.1
16.6
16.5
12.4
13.2
14.5
15.0
14.0
14.7
12.6
11.9
9.4
15.0
3.3
4,493
4,571
4,621
4,488
4,458
4,518
4,642
4,558
4,517
4,590
4,582
4,599
4,664
4,584
4,623
4,598
4,608
4,603
4,643
4,610
4,587
4,541
4,489
4,654
4,582
4,506
4,595
4,579
4,592
4,489
4,573
3,610
3,419
3,316
3,417
3,851
3,560
3,214
3,509
3,482
3,253
3,473
3,274
3,415
3,468
3,448
3,480
3,570
3,412
3,500
3,436
3,451
3,556
3,462
3,506
3,429
3,514
3,425
3,561
3,669
3,712
3,480
130
3,873
3,679
3,575
3,683
4,111
3,820
3,477
3,773
3,743
3,515
3,739
3,534
3,679
3,729
3,708
3,739
3,828
3,681
3,761
3,702
3,718
3,818
3,723
3,767
3,689
3,776
3,686
3,820
3,932
3,977
3,742
130
3,711
3,515
3,414
3,543
3,952
3,665
3,338
3,623
3,605
3,368
3,572
3,396
3,504
3,569
3,559
3,578
3,680
3,525
3,598
3,545
3,562
3,678
3,579
3,635
3,538
3,630
3,550
3,680
3,774
3,816
3,590
12855
1Mean 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-
mmscreenbeforeanalysis.Thefeedstuffandexcretawere
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
Table 3. Growth location, variety, and nutrients of test corn samples, Experiment 2
Test corn samples
Item123456
Growth location
Variety
Color
Nutrient,1DM basis
GE, kcal/kg
CP, g/kg
Ether extract, g/kg
CF, g/kg
Ash, g/kg
NDF, g/kg
ADF, g/kg
Heilongjiang
Baidan 9
White
Shanxi
Nongda 108
Yellow
Hebei
Haideng 3
Yellow
Shandong
Yedan 981
White
Shaanxi
Zhengdan 958
Yellow
Xingjiang
Dika 656
Yellow
4,4914,5464,5414,524
102
4,5084,520
85
39
16
22
94
33
85
45
15
12
85
27
94
41
13
14
65
25
86
39
13
13
63
23
81
36
11
12
61
22
37
15
16
79
25
1Mean of 3 determinations per sample. CF = crude fiber; ADF = acid detergent fiber; NDF = neutral detergent
fiber; GE = gross energy.
of feedstuffs were determined according to the procedure
described by Van Soest (1963) and Van Soest et al. (1991),
respectively. The AME, AMEn, TME, and TMEnof the
samples were calculated according to the procedure de-
scribed by Adeola et al. (1997).
Statistical Analysis
Possible relationships between chemical composition
and ME content were analyzed with correlation and step-

Page 4

ZHAO ET AL.
1606
Table 4. Correlation coefficients1between chemical composition2and ME of calibration samples, Experiment 1
Ether
extractCF CP AshADFNDFGE AMEAMEn
TME
CP
Ether extract
Ash
ADF
NDF
GE
AME
AMEn
TME
TMEn
0.35
0.38
−0.08
0.96
0.95
0.71
−0.90
−0.91
−0.90
−0.91
0.05
−0.26
0.31
0.26
0.62
−0.23
−0.29
−0.23
−0.29
0.60
0.43
0.38
0.69
−0.26
−0.24
−0.26
−0.24
−0.05
−0.07
0.16
0.15
0.18
0.15
0.17
0.98
0.72
−0.91
−0.92
−0.91
−0.92
0.67
−0.95
−0.95
−0.95
−0.95
−0.54
−0.56
−0.54
−0.55
0.99
0.99
0.99
0.99
0.990.99
1Correlation whose absolute value is more than 0.38 is significantly different from zero at P < 0.05.
2CF = crude fiber; ADF = acid detergent fiber; NDF = neutral detergent fiber; GE = gross energy.
wise regression analysis using the CORR and REG proce-
dures 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 Lamb-
erson (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
AMEnwas 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)
AMEnvalues 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 AMEnof 30 sam-
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 EquationR2
RSD,1kcal/kg
P-value
AME
AMEn
TME
TMEn
2,299.1 – 41.6 × NDF + 0.394 × GE
2,509.8 – 40.4 × NDF + 0.330 × GE
2,606.0 – 41.4 × NDF + 0.384 × GE
2,708.2 – 40.3 × NDF + 0.325 × GE
0.9181
0.9200
0.9154
0.9188
38.6
37.6
39.2
37.8
<0.0001
<0.0001
<0.0001
<0.0001
1Residual standard deviation.
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 Carre ´
(1990).
The correlation coefficients between AME and AMEn,
TME, and TMEnof calibration samples were significantly
high (0.99; P < 0.05; Table 4). These results were in accor-
dance with the findings of Sibbald (1982), Lessire et al.
(2003), and Francesch et al. (2002). Our study also indi-
cated that nitrogen-corrected values were also propor-
tional to the AME or TME value in the duck feedstuffs.
Therefore, the 4 ME measures (AME, AMEn, TME, and
TMEn) in the calibration samples had almost the same
relationship with their chemical composition. Correlation
analyses (Table 4) showed that the AME, AMEn, TME,
and TMEnwere 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,

Page 5

RESEARCH NOTE
1607
Table 6. Comparison of ME contents in corn determined by using the in vivo method and prediction model,1Experiment 2
AME, kcal/kg
AMEn, kcal/kg
TME, kcal/kg
TMEn, 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
1The 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). How-
ever, the ME content was negatively correlated with GE
content in our study, which did not agree with the results
ofLessireetal.(2003).Inthosechemicalcompositionsthat
were significantly correlated with the ME of calibration
samples,any2ofNDF,CFandADFhadhighlysignificant
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 de-
pendent 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
establishedaccordingtothesignificantlinearrelationship
between ME, NDF, and GE content (Table 5). The equa-
tions based on NDF and GE contents for predicting AME,
AMEn, TME, and TMEncontents had high accuracy with
R2> 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 accu-
racy 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 pre-
dict 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 be-
tween ME determined by the in vivo method and the
prediction model was 61 kcal/kg (Table 6). The fluctua-
tion 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 predic-
tion models for ME was close to that obtained in vivo.
Therefore,thepredictionmodelsestablishedfrom30corn
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 StateCommonwealth
(2005DIB4J033) in China. We also wish to thank W. Hu-
ang,L.Zhao,Y.W.Dong,Q.J.Wang,X.H.Jiang,M.Zhao,
ResearchProject

Page 6

ZHAO ET AL.
1608
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. Offi-
cial Methods of Analysis. Association of Official Analytical
Chemists, Washington, DC.
Carre ´, 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.
ChineseFeedDatabase(CFD).2005.TablesofFeedComposition
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,andK.M.Barram.1976.Themetaboliz-
able 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, metaboliz-
able 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 con-
tent, 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 poul-
try 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 poly-
saccharides in relation to animal nutrition. J. Dairy Sci.
74:3583–3597.