1. Introduction
Soybean meal (SBM), a by-product of soybean oil extraction, is the most commonly used protein feed ingredient and is characterized by high protein content, high nutritional value, good palatability, and high amino acid digestibility in broilers. With the rapid development of intensive broiler production, the demand for SBM has been increasing rapidly. The rising price of SBM has markedly increased the cost of broiler production. The nutritional value of SBM mainly derives from its crude protein (CP) and indispensable amino acid (AA) contents. These differences are largely driven by soybean cultivar/genetic makeup and environmental conditions in the production areas, such as temperature, precipitation, and soil properties, which directly affect the nutritional value of soybeans [
1,
2,
3,
4]. In addition, processing technology is another key factor. Among the processing methods, heat treatment is most widely applied, including extrusion and steam treatment. Romarheim et al. [
5] demonstrated that extrusion can effectively inactivate trypsin inhibitor (TI) in SBM, thereby significantly improving protein digestibility. Moreover, both the temperature and duration of steam treatment are important parameters that determine SBM quality, as they not only affect protein quality but also directly determine the residual level of TI, the main anti-nutritional factor [
6]. Espinosa et al. [
7] reported that autoclaving SBM for 3 to 25 min improved broiler growth performance by effectively degrading TIs. Therefore, to accurately evaluate the nutritional utilization of different SBMs in broilers, it is essential to determine their nutritional values and establish reliable prediction models.
The AAs are essential nutrients for broilers, playing vital roles in regulating diverse physiological functions, including growth, feed conversion efficiency, and health [
8,
9,
10]. However, the intensive development of animal husbandry has contributed to increased nitrogen emissions and environmental pollution. Kriseldi et al. [
11] showed that reducing dietary CP levels through AA supplementation significantly reduced nitrogen emissions. Other studies have shown that insufficient or excessive dietary CP and AA levels can reduce production performance and further increase nitrogen emissions [
12,
13]. Therefore, accurate evaluation of AA availabilities in feed ingredients is critical for precise AA supplementation. Apparent ileal amino acid digestibility (AIAAD) is widely used to determine the concentration of digestible AAs in poultry diets [
14], but it does not account for basal endogenous losses (BELs) of AAs from the gastrointestinal tract [
15]. In contrast, standardized ileal amino acid digestibility (SIAAD) accounts for these inevitable losses. Studies have shown that SIAAD values are more additive than AIAAD values, and thus the use of SIAAD values in feed formulation can greatly improve formulation accuracy and practicality [
16,
17,
18]. Consequently, SIAAD is considered more suitable for assessing AA utilization in poultry [
19]. Recently, Sheikhhasan et al. [
20] evaluated SIAADs in nine different SBMs for fast-growing white-feathered broilers and successfully established prediction models based on chemical composition. However, these models cannot be directly applied to medium-growing yellow-feathered chickens majorly due to significant differences in SIAAD values of SBMs between broiler breeds [
20,
21]. The SIAAD values in SBMs were reported to be consistently and significantly higher for fast-growing white-feathered broilers [
20] than for yellow-feathered chickens [
21] [90.3 vs. 79.9% for valine (Val), 90.2 vs. 83.3% for methionine (Met), 91.7 vs. 79.7% for isoleucine (Ile), 91.7 vs. 80.2% for leucine (Leu), 90.2 vs. 78.6% for threonine (Thr), 92.2 vs. 80.3% for phenylalanine (Phe), 91.8 vs. 76.0% for lysine (Lys), 91.8 vs. 82.6% for histidine (His), 94.1 vs. 84.9% for arginine (Arg), 89.2 vs. 77.6% for aspartic acid (Asp), 90.0 vs. 80.4% for serine (Ser), 93.2 vs. 80.8% for glutamic acid (Glu), 86.5 vs. 77.3% for glycine (Gly), 90.2 vs. 80.7% for alanine (Ala), 80.7 vs. 76.6% for cysteine (Cys), and 89.5 vs. 78.1% for proline (Pro)]. This means that using the SIAADs-predicting models of SBMs for fast-growing white-feathered broilers would over-predict the SIAAD values of SBMs for yellow-feathered broilers. Therefore, it is necessary to establish SIAAD prediction models of SBMs for yellow-feathered chickens. However, to date, no study on this aspect has been reported in yellow-feathered chickens.
Based on this gap, we hypothesized that prediction models for SIAADs in SBMs for medium-growing yellow-feathered chickens could be established using their chemical compositions and AA profiles. To test this hypothesis, we firstly determined the contents of conventional nutrients and anti-nutritional factors in 10 different SBMs, then evaluated their SIAADs in medium-growing yellow-feathered chickens, established prediction models based on their chemical compositions and AA profiles, and finally verified the accuracy of these models.
4. Discussion
In the present study, the chemical compositions and the SIDs of Val, Met, Ile, Leu, Thr, Phe, Lys, His, Arg, Trp, Asp, Ser, Glu, Gly, Ala, Cys, Tyr, and Pro were determined in 10 SBM samples for medium-growing yellow-feathered chickens. Corresponding SIAAD prediction models were then established and validated. The results demonstrate that chemical compositions (CP, EE, CA, ADF, NFE, NDF, PA, GE, and TI) and AA profiles (Cys, Ala, Ser, Trp, Arg, Met, Phe, and Lys) in SBMs could be effectively used to develop prediction models for most AAs in medium-growing yellow-feathered chickens except for His and Pro, because the predicted SID values for His and Pro fell outside the determined values ± SDs, and the predicting models for the SIDs of Val, Thr, and Trp could not be effectively established. To date, although the construction of SIAAD prediction models of SBMs based on their chemical compositions has been reported in fast-growing white-feathered broilers and pullets [
20,
28], no study on this aspect is available in yellow-feathered chickens. Therefore, the present study filled this critical gap. These findings support our initial hypothesis and provide a scientific basis for rapidly predicting SIAADs in SBMs and for formulating precise AA supplementation strategies in medium-growing yellow-feathered chickens.
The chemical compositions of SBMs vary due to factors such as soybean cultivar, growing environment, and processing conditions [
29,
30,
31]. Yang et al. [
32] analyzed 10 SBM samples from nine provinces in China and reported slightly higher GE contents than those observed in our study, possibly due to different sources of SBMs. Lagos and Stein [
33] analyzed 24 SBM samples from China, Argentina, Brazil, USA, and India. Their reported CP, ash, EE, ADF, Val, Ile, Leu, Thr, Phe, Lys, His, Arg, Asp, Ser, Glu, Gly, Tyr, Ala, and Pro contents were generally similar to our results, although their PA, Met, Trp, and Cys levels were somewhat higher and their NDF levels slightly lower. The reason for the above discrepancies may be that the SBM sample size in the current study was relatively small. The present study revealed lower variations (CVs < 10%) in the GE, CP, CA, NFE, Val, Met, Ile, Leu, Thr, Phe, Lys, His, Arg, Trp, Asp, Ser, Glu, Gly, Ala, Cys, Tyr, and Pro contents of the 10 SBM samples. In contrast, elevated levels of variability (CVs > 10%) were exhibited by moisture (13.03%), EE (20.89%), CF (32.32%), NDF (25.70%), ADF (30.46%), PA (20.05%), and TI (17.90%). A comparison with existing research data on the variations in SBM compositions across different regions and varieties showed that the CVs for most chemical compositions in the 10 SBMs analyzed in the current study were similar to those reported in previous studies [
19,
28,
34]. The significant variations in chemical composition across regions and varieties confirmed their diversity, making the data suitable for building precise predictive models.
The AIAAD values do not account for BELs and then lead to inaccurate estimates [
35,
36]. The present study revealed a relatively wide variation in the AIAADs of SBMs, with ranges of 80.0–86.6% for the AID of Val, 74.5–80.6% for the AID of Thr, 76.3–85.4% for the AID of Trp, 80.0–88.2% for the AID of Asp, and so on. The above observed variations can be attributed to the fact that AID values are susceptible to the influence of BELs. In contrast, SIAAD values provide a more accurate estimation of the true availabilities of AAs in feedstuffs, as the SIAADs eliminate the interferences from BELs. De Coca-Sinova et al. [
37] reported that AIAAD values for SBMs were similar to our results for Val and Ala but lower for Arg, His, Ile, Leu, Lys, Met, Thr, Asp, Cys, Glu, Gly, and Ser, while slightly higher for Phe and Ala. These differences may reflect variations in chicken breeds, rearing conditions, and feed batches. Because SIAADs correct for BELs, they more accurately reflect AA utilization and are thus critical for precise dietary AA formulation to minimize feed costs and nitrogen emissions [
38]. Our present results show that different sources of SBMs significantly affected the SID values of Val, Met, Ile, Leu Thr, Phe, Lys, His, Arg, Trp, Asp, Ser, Glu, Gly, Ala, Cys, Tyr, and Pro for medium-growing yellow-feathered chickens. Liu et al. [
21] evaluated SIAADs in one SBM for 105 d-old yellow-feathered chickens, reporting lower SID values for Val, Met, Ile, Leu, Thr, Phe, Lys, His, Arg, Asp, Ser, Glu, Gly, Cys, Ala, and Pro compared with our results. Trp and Tyr were not analyzed by Liu et al. [
21], preventing direct comparison. Sheikhhasan et al. [
20] also evaluated SIAADs in nine SBMs for white-feathered broilers, reporting higher SID values for Thr, Lys, His, Gly, Ala, Tyr, and Pro compared with our results. The SID values for Val, Met, Ile, Leu, Phe, Arg, Asp, Ser, Glu, and Cys were similar between the two studies. Trp was not analyzed by Sheikhhasan et al. [
20], preventing direct comparison. These discrepancies likely resulted from differences in chicken breeds and age, SBM origins and sample number, and physiological status of broilers during different trials.
Establishing SIAAD prediction models for medium-growing yellow-feathered chickens represents an important step toward rapid assessment of SBM quality and precision AA supplementation. In our study, SIAADs were significantly correlated with the chemical compositions and AA profiles of SBMs. For instance, the SIDs of Arg and Gly were positively correlated with GE, suggesting that higher energy availability may promote AA utilization. The SID of Ile was positively correlated with CP but negatively correlated with CF and ADF, while the SIDs of Phe, Lys, His, and Cys were negatively correlated with NDF. These findings were consistent with the evidence that dietary fiber stimulated mucus secretion and increased endogenous AA losses [
39,
40,
41]. Similarly, PA negatively correlated with the SIDs of Leu, Glu, and Ala, in line with previous findings that PA irritated the gastrointestinal mucosa and increased mucin secretion, thereby exacerbating endogenous AA losses [
42,
43]. The SID of Glu was negatively correlated with both NFE and PA, further indicating the inhibitory effects of antinutritional factors on AA digestibility. It is noteworthy that the relationship between chemical composition and digestibility may be influenced by some other factors, such as the different origins and processing conditions of SBM samples. Oviedo-Rondón et al. [
44] demonstrated that SBM nutrient composition, AA digestibility, and energy content were influenced by the country of origin and year of harvest. It is reported that thermal processing conditions affected AA digestibility in SBMs [
45,
46], likely because of formation of indigestible complexes of AA with fiber [
47]. Our SBM samples originated from different countries of origin and were processed under different conditions, leading to greater variations in their compositions and SIAADs. Additionally, the lack of effects of SBM source on SIDs of couple AAs (Ala, Cys and Tyr) was also observed in the current study, which is similar to the report of Sheikhhasan et al. [
20] in fast-growing white-feathered broilers. However, specific reasons for the above lack of effects are unclear, and need a further study.
Furthermore, in the present study, 15 prediction models were established based on their chemical compositions (CP, EE, CA, ADF, NFE, NDF, PA, GE, and TI) and AA profiles (Cys, Ala, Ser, Trp, Arg, Met, Phe, and Lys), with
R2 values ranging from 0.567 to 0.993. Models for the SIDs of Met, Ile, Leu, Phe, His, Asp, Ser, and Cys all had
R2 > 0.90, indicating high predictive reliability. Previous research on prediction models has focused mainly on pigs [
48] and fast-growing white-feathered broilers [
20]. Recently, Cao et al. [
28] also developed prediction models for hens based on SBM compositions. Their identification of GE as a positive predictor, CF and NDF as negative predictors corroborates our findings. Taken together, these results highlight that while antinutritional factors such as PA and dietary fiber inhibit AA utilization, higher GE and CP contents enhance AA digestibility.
To assess the accuracy of the developed SIAAD prediction models, one SBM sample (A4) was randomly selected for validation. With the exceptions of the SIDs of His and Pro, the predicted values for the SIDs of Met, Ile, Leu, Phe, Lys, Arg, Asp, Ser, Glu, Gly, Ala, Cys, and Tyr all closely matched the measured values. These agreements confirm the robustness of most models. To facilitate the application of these prediction models in medium-growing yellow-feathered chicken production, future studies should incorporate a broader range of SBM samples for validation, as the single sample used here is insufficient.
5. Conclusions
The prediction models for the SIDs of Met, Ile, Leu, Phe, Lys, Arg, Asp, Ser, Glu, Gly, Ala, Cys, and Tyr in SBMs for medium-growing yellow-feathered chickens were successfully established in the present study. Except for the models for the SIDs of His and Pro, which showed relatively poor predictive performances, all other models demonstrated high accuracies. Therefore, these accurate models provide a valuable reference for quickly predicting the SIAADs in SBMs for medium-growing yellow-feathered chickens based on their chemical compositions (CP, EE, CA, ADF, NFE, NDF, PA, GE, and TI) and AA profiles (Cys, Ala, Ser, Trp, Arg, Met, Phe, and Lys). Notably, the prediction models for the SIDs of Met, Ile, Leu, Phe, His, Asp, Ser, and Cys achieved R2 values above 0.90, indicating strong predictive performances.