Predicting metabolizable energy of soybean meal and rapeseed meal from chemical composition in broilers of different ages

Abstract

This study determined metabolizable energy (ME) and developed ME prediction equations for broilers based on chemical composition of soybean meal (SBM) and rapeseed meal (RSM) using a 2 × 10 factorial arrangement of age (11 to 14 or 25 to 28 d of age) and 10 sources of each ingredient. Each treatment contained 6 replicates of 8 broilers. The ME values were determined by total collection of feces and urine. Principal components analysis (PCA) of the chemical composition clearly revealed distinct differences in SBM and RSM based on a principal components (PC) score plot. The nitrogen-corrected apparent metabolizable energy (AMEn) of SBM was higher in broilers from 25 to 28 than 11 to 14 d of age (P = 0.013). Interactions between broiler age and ingredient source affected apparent metabolizable energy (AME) of SBM and ME of RSM (P < 0.05). The ME of SBM in 11 to 14 and 25 to 28-day-old broilers were estimated by crude protein (CP) content (R²≥ 0.782; SEP ≤ 83 kcal/kg DM; P < 0.001). The AME and AMEn of RSM in 11 to 14-day-old broilers were estimated by ether extract (EE), ash and acid detergent fiber (ADF) (R² = 0.897, SEP = 106 kcal/kg DM; P = 0.002), and by EE and ash (R² = 0.885, SEP = 98 kcal/kg DM; P = 0.001), respectively. The AME and AMEn of RSM in 25 to 28-day-old broilers were estimated by ash and ADF (R² = 0.925, SEP = 104 kcal/kg DM; P < 0.001) and by ash and neutral detergent fiber (NDF) (R² = 0.921, SEP = 91 kcal/kg DM; P < 0.001), respectively. These results indicate that ME of these 2 plant protein ingredients are affected interactively by chemical composition and age of broilers. This study developed robust, age-specific prediction equations of ME for broilers based on chemical composition for SBM and RSM. Overall, ME values can be predicted from CP content for SBM, or EE, ash, ADF, and NDF for RSM.

INTRODUCTION

Accurate metabolizable energy (ME) values of feed ingredients are crucial to diet formulation as energy accounts for approximately 75% the cost of broiler diets (Musigwa et al., 2021). Soybean meal (SBM) and rapeseed meal (RSM) are the byproducts of oil extraction and primary protein ingredients for broilers. The ME values vary depending on chemical composition of each source which relates to the varieties, growing environment, agronomy of oilseeds and oil extraction process (Qin et al., 1998; Baker et al., 2011; Banaszkiewicz, 2011; Khajali and Slominski, 2012; Zhou et al., 2021). Thus, many nutritionists have aimed to establish ME prediction models from chemical composition of SBM and RSM for poultry. Janssen (1989) developed an apparent metabolizable energy (AME) prediction model from crude protein (CP), ether extract (EE), and nitrogen free extract for SBM in adult cockerels. The CVB (2018) recommended ash, CP, EE, and crude fiber (CF) to predict the nitrogen-corrected apparent metabolizable energy (AMEn) of SBM for roosters. Furthermore, Alvarenga et al. (2013) validated that the AMEn of SBM calculated by prediction models resulted in better performance compared to the use of tabulated AME value for broilers. These findings suggest prediction models improved the accuracy of AMEn of SBM for broilers. In the ME prediction models for RSM, Toghyani et al. (2014) observed that the AMEn of RSM for broilers was estimated by acid detergent fiber (ADF), neutral detergent fiber (NDF), and ash, and Ding et al. (2019) reported NDF was the best factor to predict the ME of RSM for broilers. These studies commonly indicate that the fiber can predict the ME for SBM and RSM, but the effect of EE, ash, or CP on ME varied across the 2 protein ingredients. However, various ME prediction models for SBM or RSM were developed using different sources, and the selection and contribution of predictors to the ME can depend on the composition of source. In recent years, genetics of oilseed crops and the process of oil extraction has developed rapidly (Hu et al., 2017; So and Duncan, 2021; Rahman et al., 2023), which can alter the chemical composition of SBM and RSM. Additionally, many studies have shown the AME of SBM and RSM depended on age for quick-growing broilers (Shires et al., 1980; Brumano et al., 2006; Bertechini et al., 2019), indicating unique ME prediction models may be warranted for broilers in starter and growing phases. Therefore, the objective of the current study was to investigate the effects of age and source on the ME of SBM and RSM, and to establish AME and AMEn prediction models based on chemical compositions using 10 sources for each ingredient of SBM and RSM sample in China for broilers at 11 to 14 and 25 to 28 d of age.

MATERIALS AND METHODS

All experimental procedures related the use of live broilers were approved by the animal care and welfare committee of the Institute of Animal Sciences, Chinese Academy of Agricultural Sciences (Beijing, China). The code of ethical inspection was IAS 2022-143.

Feed Ingredients and Experimental Diets

A total of 10 SBM and 10 RSM were sampled across China. The chemical compositions were shown in Table 1. A corn-soybean meal basal diet was formulated with a CP of 17.5% to guarantee the CP of experimental diet was less than 25% according to the technical protocol for the determination of ME in feed for Arbor Acre (AA) broilers recommended by Ministry of Agriculture and Rural Affairs of P. R. China (2020). Twenty experimental diets were formulated by mixing 20% of each test ingredient sample with 80% of basal diet (Table 2). Corn, SBM, and test ingredients were ground through a 1.5-mm screen, then pelleted with a non-steam granulator (Type: KL200, Laizhou Chengda Machinery Co., LTD.).

Table 1.

Chemical composition of soybean meals and rapeseed meals (dry matter basis).

Item	GE, Kcal/kg DM	CP, %	EE, %	Ash, %	CF, %	ADF, %	NDF, %
SBM 1	4636	49.31	1.43	6.72	10.04	11.10	19.14
SBM 2	4648	49.98	1.20	6.72	9.55	10.17	18.34
SBM 3	4738	52.54	2.00	6.77	5.39	4.81	15.29
SBM 4	4653	50.15	1.00	6.61	7.58	8.03	15.36
SBM 5	4710	50.53	2.38	6.58	6.94	6.46	11.73
SBM 6	4676	53.31	0.88	6.85	6.23	5.92	7.99
SBM 7	4701	54.04	1.36	6.95	5.65	6.76	11.28
SBM 8	4727	54.89	1.48	6.68	4.34	4.34	7.89
SBM 9	4704	55.14	1.26	6.89	4.04	2.40	7.98
SBM 10	4682	53.47	0.96	6.66	5.53	5.99	9.51
MeanSBM 1-10	4688	52.34	1.39	6.74	6.53	6.60	12.45
SDSBM 1-10	35	2.17	0.47	0.12	2.02	2.62	4.31
MinimumSBM 1-10	4636	49.31	0.88	6.58	4.04	2.40	7.89
MaximumSBM 1-10	4738	55.14	2.38	6.95	10.04	11.10	19.14
RSM 1	5007	37.07	8.70	6.82	18.82	27.77	32.26
RSM 2	4710	45.56	1.58	7.19	15.49	18.45	27.49
RSM 3	4714	44.63	1.50	6.93	14.83	20.32	31.29
RSM 4	4704	44.94	2.00	7.20	14.20	19.17	27.12
RSM 5	4773	42.77	3.43	7.35	15.14	19.00	28.71
RSM 6	4746	44.35	1.14	8.46	24.77	43.65	61.70
RSM 7	4730	44.44	2.65	8.71	16.45	24.83	38.49
RSM 8	4608	42.73	1.11	8.63	18.50	27.20	40.20
RSM 9	4728	44.01	1.50	7.37	15.93	19.76	33.49
RSM 10	4718	43.99	1.17	7.22	15.41	19.79	32.04
MeanRSM 1-10	4744	43.45	2.48	7.59	16.95	23.99	35.28
SDRSM 1-10	102	2.41	2.31	0.72	3.13	7.74	10.23
MinimumRSM 1-10	4608	37.07	1.11	6.82	14.20	18.45	27.12
MaximumRSM 1-10	5007	45.56	8.70	8.71	24.77	43.65	61.70

Abbreviation: SBM, soybean meal; RSM, rapeseed meal; GE, gross energy; CP, crude protein; EE, ether extract; CF, crude fiber; ADF, acid detergent fiber; NDF, neutral detergent fiber; SD, standard deviation.

Table 2.

The composition of the basal and test diets (As-received basis, %).

Items	Basal diet1	Test diets
		Diets 1 to 10	Diets 11 to 20
Ingredients
Corn	67.17
SBM (non-experimental)	26.05
Dicalcium phosphate	1.92
Soybean oil	2.56
Limestone	1.14
Premix1	0.50
Sodium chloride	0.30
L-lysine	0.15
DL-methionine	0.16
L-threonine	0.05
Basal diet		80.00	80.00
Test ingredient
SBM, 1 to 10		20.00
RSM, 1 to 10			20.00
Total	100.00	100.00	100.00
Nutrient content, %2
DM	90.87	90.50∼91.31	90.41∼91.53
GE, Kcal/kg	4092	4111∼4148	4113∼4173
CP	17.71	22.80∼25.06	21.06∼23.02

Abbreviation: SBM, soybean meal; RSM, rapeseed meal; DM, dry matter; GE, gross energy; CP, crude protein.

¹Supplied per kilogram of diet: vitamin A, 8,000 IU, vitamin D₃ 1 000 IU, vitamin E 20 IU, vitamin K₃ 0.8 mg, vitamin B₁ 3 mg, vitamin B₂ 8 mg, vitamin B₆ 5 mg, vitamin B₁₂ 0.02 mg, biotin 0.2 mg, folic acid 0.6 mg, pantothenic acid 10 mg, nicotinic acid 40 mg, Cu (as copper sulfate) 8 mg, Fe (as ferrous sulfate) 100 mg, Mn (as manganese sulfate) 120 mg, Zn (as zinc sulfate) 100 mg, I (as potassium iodide) 0.70 mg, Se (as sodium selenite) 0.30 mg, Choline chloride 1500 mg.

²Values were determined values (Air-dry basis).

Experimental Design

The ME of SBM and RSM were determined using 2 batches of broilers. A total of 528 eight-day-old AA male broilers (BW = 208 ± 1 g) were selected from the first batch and randomly divided into 11 groups with 6 replicates of 8 broilers in each. One group was used to determine the ME of basal diet. The remaining 10 groups were used to determine the ME of experimental diets of 10 SBM. Another 528 twenty-two-day-old AA male broilers (BW = 961 ± 15 g) were selected from the first batch and randomly divided into 11 groups. Each group contained 6 replicates of 8 broilers. One group was used to determine the ME of basal diet. The remaining 10 groups were used to determine the ME of experimental diets of 10 SBM. Similar experiment was designed to determine the ME of basal diet and experimental diets of RSM for broilers selected on 8 (BW = 221±1 g, n = 528) and 22 (BW = 1,123 ± 14 g, n = 528) d of age from the second batch. This resulted a 2 × 10 factorial arrangement of age (11 to 14 or 25 to 28 d of age) and 10 sources within each ingredient (SBM and RSM).

AME Bioassay

The AME bioassays were conducted at the experimental base of the State Key Laboratory of Animal Nutrition and Feeding, the Institute of Animal Science of Chinese Academy of Agricultural Sciences (Beijing, China) in accordance with the technical specifications for white-feather broilers recommended by Ministry of Agriculture and Rural Affairs of P. R. China (2020). Two batches of 1,360 one-day-old AA male broilers were obtained from a local hatchery and provided free access to a commercial starter-grower broiler diet (Charoen Pokphand Feed) and water in 3-tier cages when not participating in an energy balance experiment. The energy balance experiments were conducted over 6 d (8–14 and 22–28 d of age). In each period, 528 AA male broilers with similar body weight were transferred to metabolic cages and randomly divided into 11 groups. They were acclimated with experimental diets for 55 h, followed by withdrawal for 17 h, free access to experimental diets for 55 h, and followed by 17 h fasting. Meanwhile excreta were collected for 72 h by the total collection method. The AME was determined by the difference between gross energy (GE) intake and GE output from feces and urine. The AMEn was calculated by AME corrected to zero nitrogen balance according to Hill et al. (1960). During the fasting period, chickens could freely access glucose saline solution (5% of glucose and 0.9% of sodium chloride). Excreta were dried using forced air in a drying oven at 65 ℃.

Chemical Analysis

Dry matter (DM) was determined using AOAC (2005) methods 930.15 and 925.10. The GE was measured according to ISO 9831:1998 using a Parr 6400 automatic adiabatic calorimeter (Parr instrument company, Moline, IL). The nitrogen was determined by AOAC (2005) method 984.13 using a Kjeldahl nitrogen analyzer (model KDY-9820, Shandong Haineng Scientific Instruments Co., Ltd., Dezhou, China), and the CP content was calculated as nitrogen content × 6.25. The EE content of ingredients was determined according to AOAC (2005) method 954.02 using an automatic fat analyzer (model SOX-406, Shandong Haineng Scientific Instruments Co., Ltd., Dezhou, Shandong). The ash was determined by AOAC (2005) method 942.05 using a muffle furnace at 550 ℃ for 4 h. The CF and ADF were determined by AOAC (2005) methods 978.10 and 973.18, respectively. The NDF was determined by the methods outlined by Van Soest et al. (1991).

Calculation and Statistical Analysis

All data were expressed on DM basis. The AME and AMEn of the experimental diets were calculated by the following formulas (Wang et al., 2022):

$$\text{AME}\left(\text{kcal}/\text{kg}\text{DM}\right)=\left(\text{FI}\times \mathrm{G}{\mathrm{E}}_{\mathrm{d}}-\text{EO}\times \mathrm{G}{\mathrm{E}}_{\mathrm{e}}\right)/\text{FI}$$

$$\text{AMEn}\left(\text{kcal}/\text{kg}\text{DM}\right)=\text{AME}-\left(\text{RN}\times 8.22\right)/\text{FI}$$

in which FI is the feed intake (kg DM); GE_d is the GE content of diet (kcal/kg DM); EO is the output of excreta (kg DM); GE_e is the GE content of excreta (kcal/kg DM); RN (retained nitrogen, g) = diet intake (g) × nitrogen content of diet – excreta output (g) × nitrogen content of excreta; and 8.22 kcal/g is the GE of uric acid per gram of nitrogen (Hill and Anderson, 1958).

The AMEn of feed ingredients were calculated according to the formula of Ye et al. (2022):

$${\text{AME}}_{\text{ti}}\left(\text{kcal}/\text{kg}\text{of}\text{DM}\right)=\left[\text{AM}{\mathrm{E}}_{\text{td}}-\left(1-{\mathrm{P}}_{\text{ti}}\right)\times \text{AM}{\mathrm{E}}_{\text{bd}}\right]/{\mathrm{P}}_{\text{ti}}$$

$${\text{AMEn}}_{\text{ti}}\left(\text{kcal}/\text{kg}\text{of}\text{DM}\right)=\left[\text{AME}{\mathrm{n}}_{\text{td}}-\left(1-{\mathrm{P}}_{\text{ti}}\right)\times \text{AME}{\mathrm{n}}_{\text{bd}}\right]/{\mathrm{P}}_{\text{ti}}$$

in which AME_ti, AME_td or AME_bd is the AME of test ingredient, test diet or basal diet; AMEn_ti, AMEn_td or AMEn_bd is the AMEn of test ingredient, test diet or basal diet; and P_ti is the proportion of DM from the test ingredient in the test diet.

The chemical composition and ME of 20 feed ingredients were summarized using the MEANS procedure of SAS 9.4 (SAS Inst. Inc., Cary, NC). Principal components analysis (PCA) of chemical characteristics and GE of SBM and RSM samples was performed using JMP Pro 16 (SAS Inst. Inc.). The eigenvalues and eigenvectors of the variance matrix and the cumulative variance contribution rate of principal components (PC) were calculated to obtain the load matrix, and the PC values were charted in scatterplots. The GLM procedure (SAS 9.4) was used to analyze the effect of age (d 11–14; d 25–28), ingredient source (10 samples), and interactions on the AME and AMEn of each ingredient. Means were separated by the Tukey honest significant difference. The CORR procedure (SAS 9.4) was performed to calculate correlation coefficients between chemical composition and AME or AMEn of SBM or RSM. The REG procedures (SAS 9.4) were employed to establish prediction equations for AME and AMEn specific for each age and ingredient. Significance was set at P < 0.05. Standard error of prediction (SEP) indicated the quality of prediction equations, and was calculated according to the following equation (Yegani et al., 2013):

$$\text{SEP}=\sqrt{\frac{\sum {\left(\mathrm{Y}-{\mathrm{Y}}^{\prime }\right)}^2}{\mathrm{N}}}$$

where Y and Y’ were the determined and calculated values of AME or AMEn. N is the total numbers of samples.

RESULTS

Principal Component Analysis of Chemical Compositions for Soybean Meal and Rapeseed Meal Samples

The PCA of the chemical composition of the 2 groups of SBM and RSM sample showed that the first 2 PC together explained 92.55% of the total variation (Table 3). The PC1 was mainly associated with CF, ADF, NDF, CP, and ash, and the PC2 was associated with EE and GE (Figure 1). The PC score plot characterized each sample for their position on PC1 and PC2 and clearly identified the 2 main populations: SBM and RSM. The distribution across sources was apparently more consistent for SBM relative to RSM.

Table 3.

Loading vectors of original variables on principal component as estimated by principal component analysis.

Original variables	PC1	PC2
GE, Kcal/kg	0.474	0.8371
CP, %	−0.9211	−0.171
EE, %	0.469	0.8541
Ash, %	0.7251	−0.5321
CF, %	0.9771	−0.141
ADF, %	0.9711	−0.168
NDF, %	0.9441	−0.284
Eigenvalue	4.607	1.871
Proportion	65.819	26.734
Cumulative	65.819	92.552

Abbreviation: GE, gross energy; CP, crude protein; EE, ether extract; CF, crude fiber; ADF, acid detergent fiber; NDF, neutral detergent fiber; PC, principal component.

¹Variables loaded on extracted components (i.e. loading vectors higher than 0.50).

Figure 1.

Principal component analysis of SBM and RSM: scores and loading plots of PCA model constructed from chemical composition as model variables (N = 20). Abbreviation: SBM, soybean meal; RSM, rapeseed meal; GE, gross energy; CP, crude protein; EE, ether extract; CF, crude fiber; ADF, acid detergent fiber; NDF, neutral detergent fiber; PC, principal component.

Metabolizable Energy of Soybean Meal and Rapeseed Meal Diets for Broilers From 11 to 14 and 25 to 28 d of age

The effects of broiler age and ingredient source on the AME and AMEn of the experimental diets were shown in Table 4. In SBM diets, there was an interaction between age and source of SBM on AME (P = 0.019) in which there was no effect of source on AME of SBM diet for broilers from 11 to 14 d of age, but the AME differed across sources of SBM diet for broilers aged from 25 to 28 d. There was no similar interaction affecting the AMEn of SBM diet. The age did not affect AMEn of SBM diets (P = 0.224). No statistical differences were observed in AMEn across SBM diets 6 to 10 or SBM diets 1 to 5, but the AMEn of SBM diet 8 was greater than that of SBM diets 1 to 5 (P < 0.05). There were interactions between age and ingredient source affecting the AME and AMEn of RSM diets (P < 0.05). Greater AME was observed in RSM diet 1 compared to RSM diets 3, 4, and 9 in 11 to 14-day-old broilers (P < 0.05), but no statistical differences of AME were observed across these 4 RSM diets in 25 to 28-day-old broilers. There were no statistical differences in AMEn of RSM diets 3, 4, and 8 in 11 to 14-day old broilers, but the AMEn of RSM diet 8 was lower than these of RSM diets 3 and 4 (P < 0.05) in 25 to 28-day old broilers. In addition, greater AMEn was observed in RSM diet 1 compared to RSM diets 3, 4, 9, and 10 in broilers from 11 to 14 d of age (P < 0.05), but there were no statistical differences in AMEn of these 5 RSM diets at 25 to 28 d of age.

Table 4.

The metabolizable energy (kcal/kg DM) of experimental diets (including the basal diet¹) for broilers from 11 to 14 and 25 to 28 d of age.

Item		SBM diet		RSM diet
Age	Sources of sample	AME	AMEn	AME	AMEn
D 11 to 14	1	3244cdef	3054	3288a	3116a
	2	3242def	3062	3243abc	3047abc
	3	3259bcdef	3087	3188bcd	3004cde
	4	3261bcdef	3077	3175cde	3010cde
	5	3267bcdef	3088	3234abc	3057abc
	6	3277bcdef	3089	3060fg	2923f
	7	3304abcde	3130	3106ef	2931f
	8	3308abcd	3122	3108ef	2941ef
	9	3285abcde	3107	3208bc	3021bcd
	10	3285abcde	3095	3220abc	3041bc
D 25 to 28	1	3229ef	3058	3252ab	3084ab
	2	3245cdef	3075	3263ab	3067abc
	3	3279bcdef	3099	3218abc	3031bc
	4	3205f	3036	3216abc	3048abc
	5	3260bcdef	3079	3223abc	3048abc
	6	3297abcde	3111	3023g	2887f
	7	3300abcde	3110	3089fg	2917f
	8	3361a	3161	3122def	2954def
	9	3332ab	3138	3247abc	3057abc
	10	3321abc	3117	3252ab	3070abc
SEM		15	13	15	14
Age	D 11 to 14	3273	3091	3183	3009
	D 25 to 28	3283	3098	3190	3016
Sources of sample	1	3237	3056d	3270	3100
	2	3244	3068cd	3253	3057
	3	3269	3093bcd	3203	3017
	4	3233	3056d	3196	3029
	5	3264	3083bcd	3228	3053
	6	3287	3100abc	3042	2905
	7	3302	3120ab	3097	2924
	8	3334	3142a	3115	2947
	9	3308	3122ab	3227	3039
	10	3303	3106abc	3236	3056
Variance of sources, P-value
Age		0.155	0.224	0.258	0.244
Sources of sample		<0.001	<0.001	<0.001	<0.001
Age × Sources of sample		0.019	0.100	0.040	0.047

Abbreviation: SBM, soybean meal; RSM, rapeseed meal; AME, apparent metabolizable energy; AMEn, N-corrected apparent metabolizable energy.

¹In the ME determination of SBM, the values of basal diet were 3,465 kcal/kg DM for AME and 3,315 kcal/kg DM for AMEn in 11 to 14-day-old broilers, or 3,458 kcal/kg DM for AME and 3,306 kcal/kg DM for AMEn in 25 to 28-day-old broilers. In the ME determination of RSM, the values of basal diet were 3,543 kcal/kg DM for AME and 3,382 kcal/kg DM for AMEn in 11 to 14-day-old broilers, or 3,525 kcal/kg DM for AME and 3,364 kcal/kg DM for AMEn in 25 to 28-day-old broilers.

Metabolizable Energy of Soybean Meal and Rapeseed Meal for Broilers From 11 to 14 and 25 to 28 d of age

The effects of age and ingredient source on the AME and AMEn of SBM or RSM were shown in Table 5. In SBM samples, there was an interaction between age and source of SBM on AME (P = 0.021) in which there was no effect of source on AME of SBM for broilers from 11 to 14 d of age, but the AME differed across sources of SBM for broilers aged from 25 to 28 d. For example, the AME of SBM 8 exceeded that of other SBM samples except SBM 6, 7, 9, and 10 (P < 0.05), and the AME of SBM 9 to 10 were greater than those of SBM 1 and 4 (P < 0.05) for broilers from 25 to 28 d of age. There was no similar interaction affecting the AMEn of SBM. The AMEn of SBM was lower for broilers from 11 to 14 than 25 to 28 d of age (P = 0.013). No statistical differences were observed in AMEn across SBM 6 to 10 or SBM 1 to 5, but the AMEn of SBM 8 exceeded that of SBM 1 to 5 (P < 0.05). There were interactions between age and source affecting the AME and AMEn of RSM (P<0.05). The AME in 11 to 14-day-old broilers was greater for RSM 1 compared to RSM 3, 4, and 9 (P < 0.05), but no statistical differences of AME were observed across these 4 RSM samples in 25 to 28-day-old broilers. There were no differences in AMEn of RSM 3, 4, and 8 in 11 to 14-day old broilers, but the AMEn of RSM 8 was lower than that of RSM 3 and 4 (P < 0.05) in 25 to 28-day old broilers. In addition, the AMEn of RSM 1 exceeded that of RSM 3, 4, 9, and 10 in broilers from 11 to 14 d of age (P < 0.05), but there were no differences in AMEn of these 5 RSM samples at 25 to 28 d of age.

Table 5.

The metabolizable energy (kcal/kg DM) of soybean meals and rapeseed meals for broilers from 11 to 14 and 25 to 28 d of age.

Item		SBM		RSM
Age	Sources of sample	AME	AMEn	AME	AMEn
D 11 to 14	1	2374de	2030	2287a	2072a
	2	2365de	2071	2061abcd	1725abcd
	3	2451cde	2199	1784cde	1505def
	4	2455cde	2139	1717def	1536def
	5	2495bcde	2200	2034abcd	1797abcd
	6	2548bcde	2216	1192gh	1147g
	7	2662abcd	2395	1375fgh	1149g
	8	2682abcd	2352	1396fgh	1207fg
	9	2581bcde	2291	1869bcde	1584cde
	10	2583bcde	2238	1935abcd	1685bcde
D 25 to 28	1	2335de	2087	2181ab	1982ab
	2	2415cde	2169	2228ab	1891abc
	3	2581bcde	2290	1999abcd	1710bcd
	4	2218e	1981	1994abcd	1795abcd
	5	2489bcde	2195	2052abcd	1822abcd
	6	2670abcd	2353	1081h	1037g
	7	2673abcd	2336	1363fgh	1149g
	8	2977a	2588	1536efg	1341efg
	9	2840ab	2483	2139abc	1830abcd
	10	2789abc	2385	2163ab	1900abc
SEM		74	65	74	69
Age	D 11 to 14	2520	2213	1765	1541
	D 25 to 28	2599	2287	1869	1643
Sources of sample	1	2354	2058d	2234	2027
	2	2390	2120cd	2144	1808
	3	2516	2244bcd	1891	1608
	4	2337	2060d	1855	1665
	5	2492	2197bcd	2043	1810
	6	2609	2285abc	1137	1092
	7	2668	2365ab	1369	1149
	8	2829	2470a	1466	1274
	9	2710	2387ab	2004	1707
	10	2686	2311abc	2049	1793
Variance of sources, P-value
Age		0.019	0.013	0.002	0.001
Sources of sample		<0.001	<0.001	<0.001	<0.001
Age×Sources of sample		0.021	0.103	0.040	0.047

Abbreviation: SBM, soybean meal; RSM, rapeseed meal; AME, apparent metabolizable energy; AMEn, N-corrected apparent metabolizable energy.

Correlation Between Chemical Compositions and Metabolizable Energy in Soybean Meal and Rapeseed Meal

There were significant correlations amongst chemical compositions and ME of SBM and RSM (Table 6). The GE and CP were positively correlated (r = 0.666, P < 0.05), but there were negative correlations between GE and CF or ADF and between CP and fiber components (CF, ADF, or NDF) (|r|≥ 0.815, P < 0.01) for SBM. The fiber components (CF, ADF, or NDF) of SBM were positively related to each other (r≥0.826, P<0.01). The AME and AMEn of SBM positively correlated with the CP (r ≥ 0.884, P < 0.01) but negatively correlated with the fiber components (CF, ADF, or NDF) (|r|≥0.703, P < 0.05) for 11 to 14-day-old broilers. The AMEn of SBM positively correlated with the GE (r = 0.722, P < 0.05) for 11 to 14-day-old broilers. Similarly, the AME and AMEn of SBM positively correlated with the GE and CP (r ≥ 0.662, P < 0.05) but negatively correlated with the fiber components (CF, ADF, or NDF) (|r|≥0.776, P < 0.01) for 25 to 28-day-old broilers. In the RSM samples, positive correlation was observed between GE and EE (r = 0.933, P < 0.01), ash and NDF (r = 0.695, P < 0.05), and between fiber components (CF, ADF, and NDF; r ≥ 0.930, P < 0.01). However, the CP negatively correlated with the GE and EE (|r|≥ 0.817, P < 0.01) of RSM. Both AME and AMEn of RSM negatively correlated with ash and NDF (|r|≥ 0.659, P < 0.05) and AMEn postively correlated with GE and EE (r ≥ 0.652, P < 0.05) for 11 to 14-day-old broilers. The AME and AMEn of RSM negatively correlated with ash, CF ADF and NDF (|r|≥0.644, P < 0.05) for 25 to 28-day-old broilers.

Table 6.

Pearson correlation coefficients between chemical composition and metabolizable energy in soybean meals and rapeseed meals.

Item	GE	CP	EE	Ash	CF	ADF	NDF
SBM
CP	0.666*
EE	0.555	−0.179
Ash	0.169	0.560	−0.270
CF	−0.823⁎⁎	−0.924⁎⁎	−0.046	−0.337
ADF	−0.815⁎⁎	−0.858⁎⁎	−0.120	−0.302	0.966⁎⁎
NDF	−0.564	−0.861⁎⁎	0.151	−0.282	0.856⁎⁎	0.826⁎⁎
AME d11-14	0.599	0.884⁎⁎	−0.122	0.363	−0.836⁎⁎	−0.703*	−0.875⁎⁎
AMEn d11-14	0.722*	0.886⁎⁎	0.024	0.460	−0.872⁎⁎	−0.753*	−0.806⁎⁎
AME d25-28	0.662*	0.930⁎⁎	−0.063	0.374	−0.834⁎⁎	−0.778⁎⁎	−0.834⁎⁎
AMEn d25-28	0.673*	0.922⁎⁎	−0.040	0.408	−0.818⁎⁎	−0.776⁎⁎	−0.808⁎⁎
RSM
CP	−0.817⁎⁎
EE	0.933⁎⁎	−0.908⁎⁎
Ash	−0.458	0.236	−0.369
CF	0.185	−0.241	0.076	0.527
ADF	0.156	−0.193	0.054	0.571	0.982⁎⁎
NDF	−0.088	0.050	−0.215	0.695*	0.930⁎⁎	0.951⁎⁎
AME d11-14	0.579	−0.454	0.570	−0.878⁎⁎	−0.517	−0.609	−0.757*
AMEn d11-14	0.677*	−0.552	0.652*	−0.870⁎⁎	−0.389	−0.477	−0.659*
AME d25-28	0.287	−0.194	0.286	−0.898⁎⁎	−0.714*	−0.795⁎⁎	−0.869⁎⁎
AMEn d25-28	0.369	−0.275	0.359	−0.924⁎⁎	−0.644*	−0.724*	−0.828⁎⁎

Abbreviation: SBM, soybean meal; RSM, rapeseed meal; GE, gross energy; CP, crude protein; EE, ether extract; CF, crude fiber; ADF, acid detergent fiber; NDF, neutral detergent fiber; AME, apparent metabolizable energy; AMEn, N-corrected apparent metabolizable energy.

^⁎⁎Means P<0.01

^⁎Means 0.01≤P<0.05

Prediction Equations of Metabolizable Energy on Chemical Compositions of Soybean Meal and Rapeseed Meal

Equations to predict the AME and AMEn from chemical composition of SBM and RSM were developed by stepwise regression analysis (Table 7). The AME and AMEn of SBM were estimated by CP content (R² ≥ 0.782, SEP ≤ 83 kcal/kg DM; P < 0.001) for 11 to 14 and 25 to 28-day-old broilers. The AME of RSM was estimated by EE, ash, and ADF(R² = 0.897, SEP = 106 kcal/kg DM; P = 0.002), but AMEn was estimated by EE and ash (R² = 0.885, SEP = 98 kcal/kg DM; P = 0.001) for 11 to 14-day-old broilers. The AME of RSM was estimated by ash and ADF (R²=0.925, SEP=104 kcal/kg DM; P < 0.001), and AMEn were estimated by ash and NDF (R² = 0.921, SEP = 91 kcal/kg DM; P < 0.001) for 25 to 28-day-old broilers.

Table 7.

Prediction equations of metabolizable energy (kcal/kg DM) on chemical composition (%, DM basis) in soybean meals and rapeseed meals for broilers from 11 to 14 and 25 to 28 d of age.

Item	Age	Prediction Model	R2	SEP, kcal/kg DM	P-value
SBM	D 11 to 14	AME = 161 + 45.06×CP	0.782	49	<0.001
		AMEn = -243 + 46.93×CP	0.785	51	<0.001
	D 25 to 28	AME = -2729 + 101.80×CP	0.864	83	<0.001
		AMEn = -1785 + 77.79×CP	0.851	67	<0.001
RSM	D 11 to 14	AME = 4016+57.26 × EE-271.49 × Ash-13.89 × ADF	0.897	106	0.002
		AMEn = 3740 + 50.34×EE-306.24 × Ash	0.885	98	0.001
	D 25 to 28	AME = 5164-365.15 × Ash-21.67 × ADF	0.925	104	<0.001
		AMEn = 4481-318.38 × Ash-11.90 × NDF	0.921	91	<0.001

Abbreviation: SBM, soybean meal; RSM, rapeseed meal; CP, crude protein; EE, ether extract; ADF, acid detergent fiber; NDF, neutral detergent fiber; AME, apparent metabolizable energy; AMEn, N-corrected apparent metabolizable energy.

DISCUSSION

The CP of the SBM used in the current study was higher than RSM, and the SBM had less CF than RSM, but there was considerable variation across sources (CV of CP > 4.1% and CF > 18.4% across sources for each ingredient). In oilseeds, most of the protein is stored in the embryo while fiber components are mostly from seed hulls. Soybean and rapeseed contained 10% and 15% of hull, 40% and 21% of protein, 21% and 40% of oil, respectively (Li et al., 2012; Medic et al., 2014; Kaczmarek et al., 2016). After oil extraction, the proportions of hulls in SBM and RSM concentrate to 11% and 30%, respectively (Kaczmarek et al., 2016). As a result, fiber contents of soybean and rapeseed meals are higher than that of the respective oilseed. The CP of SBM was higher than that of RSM which is in accordance with other researches (Florou-Paneri et al., 2014). To show the overall variation of the chemical composition of SBM and RSM more directly, PCA was used to reduce the dimensionality of data sets, improve interpretation, and preserve variability of the data (Jolliffe and Cadima, 2016). Others have applied PCA to the classify feed ingredients, define forage populations, screen pelleting properties of wheat, or evaluate the availability of agro-industrial by-products (Gallo et al., 2013; Kong et al., 2015; García-Rodríguez et al., 2019). In the current study, PCA explored the variable chemical components of 10 sources each of SBM and RSM. The 7 chemical components were integrated into 2 PC, which reflected 92.6% of the original variable information. The 2-dimensional scatter plot of the 2 PC revealed distinct visual segregation by ingredient: SBM with generally high CP content but low fiber concentration, and RSM with generally high fiber and EE content. This indicates that PCA can effectively distinguish different chemical compositions of SBM and RSM.

Nutritionists generally agree that the ME of feed for broilers is affected by the development of digestive tract (Smulikowska, 1998), because digestive tract weight, intestinal villus surface area, and microbial fermentation of cecal digesta all increase as broilers age (Rehman et al., 2007; Li et al., 2020). Furthermore, trypsin and chymotrypsin activities in broilers at 28 d of age were 49% and 89% higher respectively than those at 14 d of age (Lin et al., 2017). The current study revealed that the AMEn of SBM increased with broilers age, which was in accordance with the reports by Shires et al. (1980) and Bertechini et al., 2019 indicating an age-dependent effect on the ME of SBM for broilers. However, the interactive effects between sample source and age were observed on AME of SBM and AME or AMEn of RSM, indicating inconsistent differences in ME for broilers at 2 growth phases. The AME of SBM 6 to 10 were higher for broilers at 25 to 28 d of age compared to SBM 1 to 5 than 11 to 14 d of age. It is noteworthy that SBM 6 to10 contained higher CP and less fiber content than those of SBM 1 to 5 which likely contributed to the age dependent variation as the older broilers could more effectively digest CP. Pope et al. (2023) summarized that CP concentrations across 5 sources of SBM ranged from 44 to 48% and positively correlated with ME. Previous studies report the CP digestibility of SBM increases with age of broilers (An and Kong, 2022). The inconsistent difference in the AME or AMEn of several RSM at the 2 age phases could relate to high fiber content. Fiber has the ability to escape digestion and absorption in the small intestine and contributes to variable digestibility in broilers (Tejeda and Kim, 2021). Others have shown that nutrient utilization was related to the interaction between fiber components and the age of broilers (Jiménez-Moreno et al., 2009).

Generally, the fiber components (CF, ADF, NDF) and ash negatively correlate with ME (Downey and Bell, 1990; Khajali and Slominski, 2012), but GE and EE positively correlate with ME (Livesey, 1995). Stepwise regression eliminates the multicollinearity between various chemical components and ME (Garg and Tai, 2012). In SBM, CP negatively correlated with fiber components (|r|≥0.858), but ME positively correlated with CP (r ≥ 0.884) and negatively correlated with fiber components (|r|≥0.703). These findings indicate multicollinearity between CP and fiber on ME. Consequently, only CP was determined as a predictor in the ME prediction equation of SBM for broilers regardless of age, in accordance with reports by Jiang et al. (2021). For RSM sources, EE, ash, and fiber components were primarily predictors to estimate ME in the current study. Similar results were observed by Toghyani et al. (2014) who reported the ME could be predicted from NDF, ash, and ADF for RSM with 8.3% fat content (R² = 0.91), and Chibowska et al. (2000) who reported ME was negatively correlated with CF whereas positively correlated with EE (AMEn = 10.78 - 0.30 (% CF) + 0.20 (% EE)). This indicates although the nutrient profile of RSM is greatly influenced by the variety, seed quality and processing technology, fiber and EE can effectively estimate its ME value.

CONCLUSIONS

Our study demonstrated clear difference in the ME value of different sources of samples in each of SBM or RSM that were linked to chemical composition and age indicating their interactive effects on the ME. This study developed robust, age specific prediction equations of ME for broilers based on chemical composition for SBM as well as RSM. Overall, ME values can be predicted from CP content for SBM, or EE, ash, ADF, and NDF content for RSM.

DISCLOSURES

The authors declare no conflicts of interest.