Twelve crossbred barrows (Duroc×Landrace ×Large white) with an initial body weight (BW) of 35.0±2.1 kg were used. All pigs were allocated to two 6×4 Youden square design with 4 periods and 6 diets, and 2 replicated pigs per diet in each period. Each period contained 7 days of diet adaptation and followed by 5 days of feces collection. The treatment diets (Table 1) included a high protein basal (HPB) diet, a low protein basal (LPB) diet, and four experimental diets formulated by substituting 40.0% corn or 25.0% WB at the expense of HPB diet, or 30.0% SBM or RSM at the expense of LPB diet, respectively.

The experiment was conducted in the Swine and Poultry Nutrition Research Center of the National Feed Engineering Technology Research Center (Chengde, China). Pigs were placed in individual metabolic crates (1.4×0.7×0.6 m) that were equipped with a self-feeder, a nipple waterer, and slatted floors to allow for the total, but separate, collection of feces and urine. For the F1 formulation, feed intake of pigs was restricted to 2.2 times of the energy requirement for body maintenance: Daily feed allowance (kg/d) = (419×0.7×3.2×BW^0.75, KJ/d)/NE_feed (KJ/kg); For the F2 formulation, daily feed was offered to pigs at 3% of each pig’s BW. All feeds for both formulations were provided to pigs each day in 2 equal meals at 0800 and 1600. All pigs had free access to water throughout the experiment.

Feed consumption was recorded daily and treatment diets were fed for 12 days. The initial 7 days were considered as the adaptation period to the diet, and urine and feces collection were completed during the following 5 days. A time-based collection method was used [8]. Each feces and urine collection day was 24 h of full collection with no marker used to signify beginning or end of collection. Stainless steel collection trays and urine collection buckets containing 50 mL of 6 N HCl were put under the metabolism cages at 1600 on day 8 and removed at 1600 on day 13. Feces samples and 20% of the collected urine were stored at −20°C immediately after collections. At the end of the experiment, feces and urine samples were thawed and mixed separately within pig and diet, and subsampled for analysis.

Two drying techniques included FD and OD for feces were conducted at the same time. Half of the total feces collected in 5 days were dried in an oven for 72 h at 65°C, and remained in the oven for one more day at 25°C. The remaining half of the fecal samples were transferred into vacuum tubes and frozen in a freeze dryer, where the samples were cooled to −110°C under a vacuum pressure of ≤100 μm for 48 h. Freeze-dried feces also remained in the freeze dryer for one more day at 25°C (total of 72 h of freeze drying). Both oven-dried and freeze-dried feces were ground through a 1-mm screen for chemical analysis.

Feed and feces samples were analyzed for gross energy (GE) via an adiabatic oxygen bomb calorimeter (Parr Instruments, Moline, IL, USA), dry matter (DM) (method 930.15; AOAC 2006) [19], CP (method 984.13; AOAC 2006) [19], ether extract (EE) [20], and ash (method 942.05; AOAC 2006) [19]. Crude fibre (CF), neutral detergent fibre (NDF), and acid detergent fibre (ADF) were determined using fiber bags and fiber analyzer equipment (Fibre Analyzer, Ankom Technology, Macedon, NY, USA) following the procedure described by Van Soest et al [21]. Starch concentration in feed and GE in urine were analyzed using methods described by Zhang et al [22]. Determination of Cr₂O₃ in feed and feces were completed as described by Chen et al [23]. The AIA concentration of feeds and feces were determined using the method of Atkinson et al [24].

For the TC method, digestibility and metabolizability of feed components were calculated according to the following equations [5]:

Where C_input, C_output, and C_urine are the amount of component ingested, and the amount of component voided via the feces and via the urine, respectively.

For the IM methods, digestibility of feed components were calculated using the following equation [5]:

Where CI_input and CI_output are the concentration of index compound in feed and feces, respectively; CC_input and CC_output are the concentration of component in feed and feces, respectively.

The digestibility of components in the test ingredients was determined using the difference method and is calculated according to the following formula described by Kong and Adeola [5]. We assumed that there was no interaction between the digestibility values of components in the test ingredient and those in the basal diet. The calculation was as follows:

In which D_bd, D_td, and D_ti are the digestibility (%) of the component in the basal diet, test diets, and test ingredient, respectively, and P_bd and P_ti are the proportional contribution of the component by the basal diet and test ingredient to the test diet, respectively.

The recovery rate of marker was calculated as described by Jagger et al [25].

Data were checked for normality and outliers were detected and removed using the UNIVARIATE procedure of SAS (SAS Inst. Inc., Cary, NC, USA). To test the effects of different methods within each diet, data were analyzed by one-way analysis of variance (ANOVA) using the general linear model (GLM) procedure of SAS, with pig as the experimental unit and the experimental methods of TC vs AIA vs Cr₂O₃ or OD vs FD was the only main effect included in the model. To test the effects of different methods within each ingredient, data were analyzed by two-way ANOVA using the GLM procedure of SAS, and the statistical model included the main effect of method of F1 vs F2, the main effect of method of TC vs AIA vs Cr₂O₃, and their interaction effect. Since all interaction effects were not significant, only the main effects were shown in the results. Treatment means were calculated using the LSMEANS statement, and were separated using the Tukey-Kramer test. Significant differences were declared at p<0.05.

The recovery rate of AIA ranged from 0.87 to 1.05 among the ten experimental diets. A wide range of 0.78 to 1.17 for the recovery rate of Cr₂O₃ was observed in the ten experimental diets (Table 4). There were no significant differences between the recovery rate of Cr₂O₃ and AIA in the experimental diets except that a greater recovery rate of Cr₂O₃ was observed in LPB diet or SBM diet formulated using F1 formulation (p< 0.05). The big variation in the recovery rate of Cr₂O₃ was in agreement with the previously reports [10,26]. The good recovery rate of AIA indicated that AIA can be used as a good marker for calculation of nutrient digestibility.

Concentration of digestible (DE) and metabolizable energy (ME), and the apparent total tract digestibility (ATTD) of GE, ash, NDF, and ADF in HPB, LPB, and RSM diet in F1 formulation and corn diet in F2 formulation determined by the AIA marker method were lower than those determined by the Cr₂O₃ marker method (p<0.05), and no differences were observed between the results gained by the AIA marker method and those by the TC method (Table 5, 6). The same pattern was shown for the ATTD of CP in the LPB diet. The lower DE and ME values or ATTD of nutrients in feeds evaluated by the AIA marker method compared with the Cr₂O₃ marker method can be explained by the lower recovery rate of AIA marker. No significant differences were observed between the DE and ME values and ATTD of GE evaluated by the TC method and those by the Cr₂O₃ marker method in the HPB diet, but in LPB diet, the DE, ME, and the ATTD of GE determined by the Cr₂O₃ marker method were greater than those calculated by the TC method (p<0.05). This may be mainly due to the high CP level in HPB diet. The dietary protein level was found to be negatively correlated with the ME/DE ratio (r = −0.956), indicating that protein level in diets had a profound effect on the ME obtainable from DE, and thus the conventional methods (e.g. the TC method) lead to underestimated energy content of high protein feeds [6]. In addition, the slightly greater (1.4%) NDF level in the LPB diet compared with the HPB diet may influence the determined results, because fibre is the main factor that could affect the energy values of feed [27]. As for similar previous studies on methodology in evaluating swine diets, various results were reported. The aforementioned effects of calculation methods (TC vs IM methods) on dietary component digestibility were inconsistent with some previous results [9,11,13], but in agreement with others which estimated DE and digestible nutrients using the TC and AIA marker methods [8,11,28]. A range of 97.5% to 98.8% for the ME/DE ratio was observed in the ten experimental diets in the current trial, which was close to the ratio of 96.0% in most complete feeds reported by Noblet [27]. The ATTD of GE, CP, ash, CF, EE, NDF, and ADF in ten experimental diets varied from 81.7% to 93.0%, 81.7% to 92.2%, 32.3% to 66.3%, 36.4% to 66.5%, 51.5% to 80.2%, 42.3% to 71.7%, and 19.2% to 66.1%, respectively (Tables 5, 6). The ATTD of GE falls into the normal range of 70% to 90% which was reported by Noblet [27].

A greater ATTD of CF calculated by the AIA marker method was detected compared with that determined by the TC method in SBM diet formulated by F1 (p<0.05), but the ATTD of CF determined by the TC method was lower than that evaluated by the Cr₂O₃ marker method in LPB diet and corn diet formulated by F2 (p<0.05). Otherwise, no significant differences were observed on estimated energy contents and nutrient digestibility in each treatment diet between different determining methods.

Overall, considering the good recovery rate of AIA marker and the equal performance in estimating nutrient digestibility in diets between TC method and AIA marker method, the AIA is a reliable marker for measuring digestibility of components in swine feed.

The DE concentration in corn diet formulated by F1 was greater when determined using freeze-dried feces compared with that determined using oven-dried feces (p<0.05, Table 7). A greater ATTD of GE (p<0.05) evaluated using freeze-dried feces was observed in SBM diet and RSM diet formulated by F2 compared with that evaluated using oven-dried feces. These findings contradict with those of Jacobs et al [16], who reported no differences among drying techniques on estimating DM, GE, N, C, or S concentrations in pig feces. Some study on poultry also reported no significant differences between the FD and OD techniques on chicken excreta samples in determining the true ME value of poultry diet [17]. However, Wallis and Balnave [29] reported greater energy and N losses when excreta were freeze-dried rather than being oven-dried at 60°C or 80°C. Those results were somehow in accordance with ours, since we had numerical greater DE and ME estimations in almost all the test diets when the FD technique was used, although not significant difference, indicating less fecal energy loss using the OD technique. The current results of no significant effects of drying method on the ATTD of CP in diets was in agreement with those of Jorgensen et al [30], who reported that the two drying techniques (FD vs OD at 70°C) of feces did not affect measurement of protein digestibility. Therefore, the greater DE concentration and ATTD of GE in feeds calculated with FD technique may be not caused by greater nitrogen loss. Based on the above analysis, the OD technique is more recommended considering its lower fecal energy loss and lower cost compared with the FD technique when drying fecal samples.

Greater DE, ME, ME/DE ratio, and ATTD of GE in corn were observed when evaluated based on F2 formulation compared with the F1 formulation (p<0.05, Table 8). When determining the energy concentration and nutrient digestibility of corn, method based on F2 formulation actually belongs to the direct method, while method based on F1 formulation belongs to indirect method. The influence of interaction between the basal diet and test ingredient was smaller for the direct method compared with the indirect method, resulting in greater energy values of corn evaluated based on F2. The DE values of WB, SBM or RSM, and the ME values of SBM or RSM evaluated based on F1 were greater than those determined based on F2 (p<0.05). These findings were contradict with previous studies which compared the effects of different basal diets on determined digestibility coefficients of ingredients and found no differences [6–8]. However, the current findings confirmed that constituent of basal diets could affect the estimated DE and ME values of test ingredients. When determining the energy concentration of ingredients rich in fibre or protein, such as WB, SBM, and RSM, the corn basal diet may have too simple composition, just as that included in F2. The LPB diet in the F1 formulation was more complicated and already contained a portion of the tested ingredients, e.g. SBM and RSM, leading to the real substitution rate of these two feedstuffs to be greater than 30%. This may partial explain the greater DE or ME values of SBM and RSM evaluated using F1 formulation, since a higher inclusion level of test ingredient will lead to more accuracy and greater estimation of energy concentration in the difference method [5]. Furthermore, the lower DE or ME values of WB or RSM evaluated based on F2 may be due to the decreased energy contents of the test diets in the present of fibrous ingredients. It was shown that increased dietary fibre level would decrease the ATTD of DM and GE in diets [31,32], leading to less discrepancy of energy contents between test diet and basal diet in the difference method [5]. The greater DE or ME values of SBM determined based on F1 can be explained by the greater DE or ME values of the SBM test diet, which were caused by 4.0% more CP in SBM test diet formulated by F1. The concentration of DE and ATTD of GE in SBM determined using the AIA marker method were greater than those determined by the Cr₂O₃ marker method (p<0.05), regardless of the formulations (F1 or F2). Overall, the interaction between the energy concentration and digestibility values of components in the test ingredient and the basal diet exists. The formulation procedure can affect the evaluation results of the energy content and nutrient digestibility of feedstuffs, but it depends on the characteristics of the test ingredients.