1. Introduction
Rapeseed is the second most abundant oilseed produced after soybean in the world, with production of more than 68 million tons in 2019 [
1]. Double-low rapeseed cake (DLRSC), a co-product of expeller-press extruded double-low rapeseed to produce oil, is an attractive, cost-effective feed resource for animals [
2]. Double-low rapeseed cake is produced without solvent extraction, resulting in high residual oil content (10–15%) in DLRSC [
3]. Hence, DLRSC has a high energy value and may be a valuable source of amino acids (AA) in swine diets. Double-low rapeseed cake has been widely used in swine feed formulation to reduce diet cost and has no negative effects on the growth performance of pigs [
4,
5].
Energy value for DLRSC varies greatly due to its diverse chemical composition [
6]. In vivo measurement of energy value is not only time consuming and expensive, but the measured values are only applicable to the specific samples evaluated in the experiment. Prediction equations that estimate the energy value based on chemical composition can be used to rapidly and accurately estimate energy value of feed ingredients [
7,
8]. However, to our knowledge, there is no available information on the prediction of energy content in DLRSC for growing pigs. Furthermore, for these equations to be effective, validation studies using animal tests are needed [
8,
9]. There is, however, no clear agreement on which method to use to verify the accuracy of a prediction equation. Commonly used methods include caloric efficiency [
10], cross-validation [
11], and repeated experimentation [
12,
13] methods. In this experiment, the caloric efficiency method was used to verify the accuracy of the ME prediction equation.
We hypothesized that the energy prediction equations can be established and used to accurately predict the metabolizable energy (ME) of DLRSC. Therefore, the objective of this study was to: (1) develop prediction equations for ME based on the chemical compositions and energy values of 10 different DLRSC samples previously analyzed in our laboratory; and (2) verify the accuracy of the best-fit prediction equation for ME using the caloric efficiency approach.
4. Discussion
Li et al. [
6] determined the chemical compositions and ME content of ten DLRSC samples. Based on those data, ME prediction equations for DLRSC were developed in the present experiment. Previous studies successfully established prediction equations for energy content of flaxseed expellers [
20], sorghum grains [
21], barley [
22], and corn [
23]. The development of such prediction equations contributed to rapidly and accurately determining the energy content of feed ingredients. However, there is little information about prediction equations for DLRSC. To our knowledge, the results of the present study provide the first prediction equations for ME based on the measured chemical composition of DLRSC. More accurate prediction of energy values may improve diet formulation and reduce diet cost. In addition, due to the relatively complicated test process of the net energy evaluation system, the energy evaluation of feed ingredients is still based on digestive energy and ME.
It is not surprising that NDF and CF were the first and second predictors, because many reports have indicated that dietary fiber is a key factor affecting the energy content of a diet [
24]. Higher fiber has been shown to reduce the energy content of DLRSC [
6,
25]. Gross energy was also one of the predictors of the best-fit prediction equation, which may be attributed to the fact that EE is a primary determinant of GE. Crude fat is not only a more digestible component in the intestine, but its presence can also improve the digestibility of other nutrients [
26]. In this experiment, the optimal model to predict ME included three predictors: NDF, CF, and GE. However, a previous report indicated that the optimal model for predicting the ME value of canola meal, 00-rapeseed meal, and 00-rapeseed expellers fed to growing pigs was ME = –630.8 + 14.13 × ash + 5.02 × CF + 3.45 × ADF + 1.03 × digestible energy (R
2 = 0.98) [
27]. This difference may be due to the prediction equation established in the previous study being based on three rapeseed co-products, while the equation from the present study is based only DLRSC. The energy content of different rapeseed co-products varies greatly due to the variations in concentrations of nutrients in the seeds and differences in oil extraction procedures [
27]. Therefore, it may be more accurate to base prediction equations of energy content on analyses of each rapeseed co-product.
Increasing the levels of dietary DLRSC had no effect on the growth performance of growing pigs, which agrees with results from studies with weaned pigs [
4,
5]. However, in contrast, there are reports that increasing dietary expeller-extracted canola meal content linearly decreased ADG and ADFI of growing pigs [
28] and growing–finishing pigs [
29]. These differences may be directly related to the concentrations of total glucosinolates in the diet. The DLRSC in the present Experiment 2 contained 9.38 μmol/g glucosinolates. Therefore, the calculated concentration of glucosinolates in the diet with 21% of DLRSC was 1.97 μmol/g diet, which was below the generally accepted glucosinolates tolerance level (2.0–2.5 μmol/g) for growing pigs [
25,
30]. However, the concentrations of total glucosinolates were 5.22 μmol/g in the diet containing 22.5% expeller-pressed canola meal [
29] and 2.75 μmol/g in the diet containing 30% expeller extracted canola meal [
28], which are greater than the highest glucosinolates tolerance level for growing pigs. In addition, increasing levels of extruded
B. juncea expeller increased growth performance of weaned pigs due to a decrease in dietary net energy value and SID Lys/net energy ratio [
31]. In general, glucosinolates in rapeseed co-products are considered as a limiting factor for their utilization in swine diets. However, when concentration of the dietary glucosinolates are below tolerance level, feeding DLRSC to replace soybean meal did not affect the growth performance of growing pigs fed diets balanced for SID Lys/ME ratio and SID AA systems.
Although the best ME prediction equation successfully fits the DLRSC samples used in the development of the model, there is no guarantee that ME values can be accurately predicted when this equation is applied to chemical composition data from other DLRSC samples. Therefore, validation of this equation is warranted. Some researchers have proposed the concept of caloric efficiency to verify the accuracy of the prediction equation. The assumption of this approach is that if the energy value assigned to a test ingredient is accurate, regardless of the ingredient inclusion level, a similar caloric efficiency will be calculated among diets [
32,
33]. Increasing the inclusion levels of dietary DLRSC in Experiment 2 had no effect on the caloric efficiency of ME, which indicated that the predicted ME value of DLRSC was accurate. In addition, concentrations of total glucosinolates in the diets were lower than the recognized maximum tolerance level, and dietary SID Lys/ME ratio and SID essential amino acids/SID Lys ratio were kept constant to meet requirements. Thus, energy value of diet may be the main factor affecting growth performance of pigs. Further, growth performance of pigs was not affected by increasing the levels of DLRSC, indicating that the actual ME values of four experimental diets were relatively consistent and close to the predicted ME values used in diet formulation. Therefore, the best-fit prediction equation obtained in Experiment 1 may be used to accurately calculate the ME value of DLRSC for growing pigs. In addition, prediction equations of energy content have been verified for dried distillers’ grains with solubles, using the cross-validation method [
11], as well as for lipids and full-fat rice bran, using repeated experimentations [
12,
13]. Furthermore, some studies have determined the relative accuracy of ME prediction equations for de-oiled corn distillers’ dried grains with solubles [
8] and corn [
34] by comparing the effects of methods for calculations for metabolizable energy (prediction equation or nutrient composition table) of diet formulation on growth performance and carcass quality. However, each of these methods have advantages and shortcomings. There is no clear agreement on which method should be used to verify the accuracy of the prediction equation. Continued research is required to develop improved methods to verify the accuracy of the prediction equation.
Increasing dietary DLRSC reduced concentrations of T3 and T4 in serum in the present study, as also reported for pigs fed diets containing canola meal [
35], expeller-extracted canola meal [
28], and double-low rapeseed meal [
36]. The break-down products of glucosinolates, such as oxazolidinethione and isothiocyanate, may impair function of the thyroid gland to decrease secretion of thyroid hormones [
25,
28]. Thyroid hormones are required for the normal growth and development of muscle, and their deficiency may inhibit growth of pigs [
37]. However, in this experiment, the concentration of thyroid hormones in pigs fed DLRSC diets was not decreased sufficiently to reduce growth performance.
The ATTD values of nutrients in the diet decreased as DLRSC levels increased. Similarly, weaned pigs fed up to 200 g/kg expeller-pressed canola meal or canola press-cake or up to 240 g/kg extruded
Brassica juncea expeller had lower nutrient digestibility values [
4,
5,
31]. Furthermore, similar results were reported for pigs fed diets containing solvent-extracted meal [
38]. The reduction in the ATTD of nutrients was likely due to the increase in fiber content as the dietary DLRSC levels increased. Greater dietary fiber intake can increase evacuation rate and decrease the transit time of nutrients in the intestine [
39,
40]. Increasing dietary fiber content may also increase endogenous excretion and decrease ATTD of nutrients [
41]. In addition, decreases in nutrient digestibility may also be related to increases in concentrations of glucosinolates in the diet [
38].