Linkage of in situ ruminal crude protein degradation with ruminal degradation of amino acids and phytate from different soybean meals in dairy cows

The objectives of this study were to determine the range in ruminal degradability of crude protein (CP) and intestinal digestibility of rumen undegradable protein (ID RUP ) in commercial soybean meal (SBM) and to investigate the range in in situ ruminal AA and phytate (InsP 6 ) degradation and their relationship to CP degradation. An in situ study was conducted using 3 lactating Jersey cows with permanent rumen cannulas. Seventeen SBM variants from Europe, Brazil, Argentina, North America, and India were tested for ruminal CP and AA degradation, and in vitro ID RUP . Nine variants were used to investigate the ruminal degradation of InsP 6 . The estimated rapidly degradable fraction ( a ) of CP showed an average value of 4.5% (range: 0.0– 9.0%), the slowly degradable fraction ( b ) averaged 95% (91–100%), and the potential degradation was complete for all 17 SBM variants. The degradation of fraction b started after a mean lag phase of 1.7 h (1.1 h–2.0 h) at an average rate ( c ) of 10% h −1 , but with a high range from 4.5% to 14% h −1 . Differences in the degradation parameters induced a considerable range in CP effective degradation at a rumen passage rate of 6% h −1 (CPED 6 ) from 38% to 67%; hence, the concentration of rumen undegradable protein (RUP) varied widely from 33% to 62%. The range in AA degradation between the SBM variants was high, with Ser showing the widest range, from 28% to 96%, and similar for the other AA. The regression equations showed close relationships between CP and AA degradation after 16 h of in situ incubation. However, the slopes of the linear regressions were significantly different between AA, suggesting that degradation among individual AA differs upon a change in CP degradation. The concentrations of InsP 6 and InsP 5 (myo-inositol pentakisphosphate) in bag residues in the in situ study decreased constantly with longer ruminal incubation times. The ruminal degradation parameters of InsP 6 ranged from 11% to 37% for fraction a , 63% to 89% for fraction b and from 7.7% to 21% h −1 for degradation rate c with average values of 21%, 79%, and 16% h −1 , respectively. The calculated InsP 6 effective degradation at a rumen passage rate of 6% h −1 (InsP 6 ED 6 ) varied from 61% to 84% among the SBM variants. Significant correlations were detected between InsP 6 ED 6 and CPED 6 and between InsP 6 ED 6 and chemical protein fractions A, B1, B2, B3 and C. Linear regression equations were developed to predict ruminal InsP 6 degradation using CPED 6 and chemical protein fractions B3 and C selected by step-wise selection procedure. It was concluded that there was a high range in CP, AA, and InsP 6 degradation among commercial SBM, suggesting that general de-gradability values may not be precise enough for diet formulation for dairy cows. CP degradation in SBM may be used to predict rumen degradation of AA and InsP 6 using linear regression equations. CP and InsP 6 degradation could also be predicted from the chemical protein fractions.


INTRODUCTION
Intensive milk production systems rely on highnutrient-density diets, including a large proportion of protein feed, to meet the requirements of dairy cows.An important goal in diet formulation for lactating dairy cows is to maximize microbial protein synthesis in the rumen and meet the demands of RUP and essential AA (EAA) for optimal milk production without wasting dietary protein, thereby improving N efficiency.For this purpose, the feeding value of these protein feeds and its range must be known and/or predicted with confidence.What we do know is that different feedstuffs vary widely in ruminal protein degradation, intestinal digestibility, and AA composition of RUP (NASEM, 2021).Such differences also exist within one feedstuff, and studies have shown a wide range in the rumen degradation behavior of CP and AA in rapeseed meal (RSM) (Steingass et al., 2013), dried distillers' grains with solubles (DDGS) (Westreicher-Kristen et al., 2013), lupin grains (Titze et al., 2019) and pea grains (Titze et al., 2021).
Although soybean meal (SBM) is one of the most common protein feeds in the diets of lactating dairy cows, studies on the range in RUP in SBM from different processing plants are scarce.Some studies evaluated ruminal CP and AA degradation of SBM and the influence of an additional chemical or thermal treatment (Harstad and Prestløkken 2000;Borucki Castro et al., 2007;Awawdeh et al., 2007) or compared SBM to other protein feedstuffs (Maxin et al., 2013;Mjoun et al., 2010).However, no study has examined the range in the feeding value of standard solvent-extracted SBM of different origins without additional chemical or physical treatment using an extensive sample set.
Moreover, there has been substantial progress in the understanding of AA nutrition and metabolism in ruminants over the past few years (Schwab and Broderick 2017) and considering of the degradation of individual AA to refine the N supply to animals is important.However, it has been suggested that prediction of the AA composition of RUP from the AA profile of the feed protein is not possible (Cozzi et al., 1995) whereas Steingass et al. (2013) and Westreicher-Kristen et al. (2013) showed that ruminal CP and AA degradation are closely related in RSM and DDGS, and that degradation of individual AA can be accurately predicted from CP degradation.Therefore, we hypothesized that the prediction of AA degradation from CP degradation might also be possible for SBM.
A close relationship between CP and phytic acid (myo-inositol 1,2,3,4,5,6 hexakis dihydrogen phosphate, InsP 6 ) disappearance in RSM and SBM has been reported (Haese et al., 2017).Rapeseed and soybeans are relatively high in P concentration and most of the P is present in the salt form of InsP 6 (phytate) and is located in a protein-rich structure.While InsP 6 degradation was found to vary in a pattern similar to that of CP degradation in RSM (Haese et al., 2022), range in ruminal InsP 6 degradation from commercial SBM in relation to CP degradation has not yet been studied.Our second hypothesis was that InsP 6 degradation is related to CP degradation in SBM.
The aims of this study were as follows: (1) to determine the range in the in situ ruminal degradation of CP and the intestinal digestibility of rumen undegradable protein (ID RUP ) for different SBM; (2) to investigate the range in in situ ruminal AA and InsP 6 degradation and their relationship with CP degradation; (3) to predict CP and InsP 6 degradation from chemical protein fractions.

Soybean Variants
Seventeen commercially available solvent-extracted SBM variants from Europe (n = 6), Brazil (n = 6), Argentina (n = 2), North America (n = 2), and India (n = 1) were tested for ruminal CP and AA degradation and in vitro ID RUP .Nine samples (no. 2, 4, 5, 9, 10, 11, 12, 14, 18) were chosen because they covered a wide range of CP degradability and used to investigate ruminal degradation of InsP 6 , including analysis of less phosphorylated inositol phosphates (InsP 3-5 ).The variants were part of a larger set from a companion study that investigated the AA digestibility of 18 SBM in cecectomized laying hens (Siegert et al., 2023).Siegert et al. (2023) contains a detailed description of all chemical analyses of the sample material used in the present study, as well as all analytical values for each variant, including DM, CP, crude fat, crude ash, crude fiber, aNDFom, ADFom, NDF-N, ADF-N, starch, sugar, GE, AA, tannins, protein dispersibility index, KOH solubility, trypsin inhibitor activity, urease activity, in vitro solubility, inositol phosphate isomers, and particle size distribution.The numbering of the variants used in the present study is the same as that used by Siegert et al. (2023).However, for one variant (no.3) not enough material was left from the study by Siegert et al. (2023) which explains why we have the number 18 although only 17 variants were investigated.For the 17 SBM under study, concentrations (g kg -1 DM) of CP and of the first limiting AA for milk production, Met,Lys,and His,ranged between 498 and 564,, respectively (Table 1).The average InsP 6 concentration of the 9 SBM variants was 14.0 g kg -1 DM (12.3-16.8g kg -1 DM) and average InsP 5 concentration was 2.6 g kg -1 DM (1.9-4.0 g kg -1 DM).Traces of InsP 4 was detected (up to 0.6 g kg -1 DM) in the SBM but InsP 3 was not detectable.Protein fractionation was based on the method described by Licitra et al. (1996) and analyzed as described by Titze et al. (2019).Protein fractions were calculated according to the CNCPS (Sniffen et al., 1992) and ranged (% of CP) between 0.0 and 5. 2, 0.3-20.3, 71.1-85.8, 3.0-20.8, and 0.8-1.8

Animals and Experimental Design
The experiment was conducted in accordance with German animal welfare legislation and approved by the Regierungspräsidium Stuttgart, Germany (approval code V352/18 TE).Three lactating Jersey cows with an average body weight of 470 kg (standard deviation [SD]: 70 kg) fitted with rumen cannulas were used for rumen in situ incubation.Cows were milked twice daily at 0500 and 1600 h and had a mean milk yield of 16 kg d -1 (SD: 0.5 kg d -1 ) during the experimental period of 17 d until the incubation of all samples was finished.
The cows were housed in groups in a freestall barn with cubicles covered with rubber mats and chopped straw, and had free access to feed and water.During the experimental period cows consumed on average 16 kg DM per day (SD: 3.4 kg d -1 ) from a total mixed ration that contained 25% corn silage, 25% grass silage, 17% grass hay, 16% mixed concentrate, 7% RSM, 6% toasted soybeans, 2% barley straw, and 1% of a mineral mix on a DM basis.The diet contained (per kg DM) 6.3 MJ net energy for lactation, 133 g CP, and 4.3 g P.

In Situ Procedure and Determination of ID RUP
The in situ procedure was based on the method proposed by Madsen and Hvelplund (1994) with minor modifications (Seifried et al., 2017).Samples (8 g of DM) were weighed into polyester bags (Ankom Co, Macedon, NY, USA; pore size: 50 ± 3 μm; internal dimensions: 10 × 20 cm).After conditioning in warm tap water (~39°C) for 1 min, bags were inserted into the rumen for time spans of 2,4,6,8,12,16,24,48, and 72 h.The incubation scheme was the same for all cows, meaning that bags of all variants for a specific time span were incubated at the same day and time in all 3 cows.However, variant and time span were at least replicated on 2 different days in all cows.After removal from the rumen, bags were rinsed with cold tap water and frozen.After the end of all in situ incubations bags were thawed and washed in a washing machine (type WM14A160; Siemens GmbH, Munich, Germany) for 15 min without centrifugation, dried at 60°C for 24 h and weighed.For the determination of the initial 0 h time point, 3 bags per SBM variant were rinsed as described previously without ruminal pre-incubation.As leftovers in the bags decrease with longer ruminal incubation time and to obtain sufficient residue for each variant at any incubation time for chemical analysis, the number of bags per cow and time point differed; however, a minimum of 9 measurements (3 bags × 3 cows) were performed at each time point.The residues per cow and incubation time were pooled and pulverized using a vibrating cup mill (PULVERISETTE; Fritsch GmbH, Idar-Oberstein, Germany) before further analysis.The water solubility of each SBM was determined over filter paper according to Madsen and Hvelplund (1994) and used to estimate small particle losses by subtraction from the initial 0 h loss.
To estimate ruminal degradation kinetics and ED, near-infrared spectroscopy (SpectraStar 2500X, Software: Unity InfoStar Version 3.11.1;Unity Scientific, Brookfield, CT, USA) was used to determine the N concentration in all SBM samples and residues obtained from the bags.From a total of 425 samples of the current study, a database of 333 samples, including SBM and their in situ bag residues (222 for calibration, 111 for validation) with N concentrations determined by chemical analysis (VDLUFA 2006), was used for new calibration development according to the approach described by Krieg et al. (2018).Reference values ranged from 1.92% to 13.60% N on a DM basis.The best performance was achieved using a wavelength segment of 730-2450 nm and 1st derivative of the spectra.The calibration used had a standard error of calibration of 0.15% DM and a standard error of prediction of 0.21% DM, and the validation step showed an R 2 of 1.00 be-tween the measured and predicted values.Slope and intercept of the validation set were 0.999 and 0.002, respectively and not different from 1 and 0, respectively (P < 0.05).Crude protein was calculated as total N × 6.25.
Amino acids were determined in SBM and bag residues after incubation for 16 h.Determination on an L-8900 amino acid analyzer (VWR, Hitachi Ltd., Tokyo, Japan) followed sample oxidation and acid hydrolysis according to the protocol described by Rodehutscord et al. (2004).Residues of 16 h ruminal incubation were also used for in vitro determination of ID RUP based on the 3-step method of Calsamiglia and Stern (1995) and measured as described by Grubješić et al. (2020).
Nine SBM variants and their corresponding bag residues were analyzed for InsP 3-6 .Samples were extracted using the method described by Zeller et al. (2015), with slight modifications, as described by Sommerfeld et al. (2018).Briefly, 0.05 g of each sample was extracted with a solution of 0.2 M EDTA and 0.1 M sodium fluoride (pH = 8) under agitation for 30 min and was centrifuged at 12,000 × g for 15 min.After centrifugation following 2 extractions, the supernatants were combined and 1 mL therefrom was centrifuged at 14,000 × g for 15 min and filtered.After centrifugation again at 14,000 × g for 30 min, the isomers of InsP 3-6 in the filtrates were measured using high-performance ion chromatography (ICS-3000 system, Dionex, Idstein, Germany).

Calculations and Statistical Analysis
Samples were created by pooling residues of all bags from cow j, SBM sample i at incubation time t.From these samples, degradation of CP, InsP 6 and AA (y) was calculated from the quantity of nutrients in the sample after incubation y bag residue ijt ( ) and the quantity before the incubation started y feed ij The quantities of y bag residue ijt and y feed ij were calculated from the amount of DM in the bags pooled within the sample multiplied by the associated analyzed concentrations of CP, InsP 6 or a specific AA.
For each variant and cow an exponential model including lag time (lag) was used to fit the ruminal degradation of CP and InsP 6 (Ørskov and McDonald 1979;McDonald 1981) as where Deg (t) represents the degradation (%) of CP or InsP 6 at time t (h), a (%) is the rapidly degradable fraction, b (%) is the slowly degradable fraction over time, c (% h −1 ) is the degradation rate of b, and lag (h) is the duration until the beginning of the degradation of b.
The CP effective degradation at a rumen passage rate of 6% h -1 (CPED 6 ) and that of InsP 6 (InsP 6 ED 6 ) were calculated using a ruminal outflow of k = 6% h -1 and the following equation (Wulf and Südekum 2005) and RUP 6 was calculated as For estimation of in vitro ID RUP calculation was as follows where N soluble is the quantity of soluble N determined in vitro (mg) and N incubated is the total N incubated with pepsin and pancreatin (mg).
Model parameters for the in situ data were estimated using an iterative least-squares procedure in GraphPad Prism (version 5.00, GraphPad Software Inc., San Diego, CA, USA).
All other statistical analyses were performed using SAS software (SAS System for Windows, Version 9.4, SAS Institute, Cary, NC, USA).To evaluate if the variability of the degradation parameters and CPED 6 , InsP 6 ED 6 and AA degradation after 16 h was equal to zero a chi-squared test was applied to test the null hypothesis H 0 : σ 2 ≤ 0.01 over the variance of a normally distributed population.The null hypothesis could be rejected for all corresponding measures.Therefore, the degradation parameters and CPED 6 , InsP 6 ED 6 and AA degradation after 16 h between SBM variants were compared using a one-factorial approach with the MIXED procedure using the following model: where Y ij is the observed degradation in the sample of cow j and SBM sample i, μ is the intercept, SBM i is the fixed effect of the SBM sample i (i = 1-17 or 1-9), A j is the random effect of animal (j = 1, 2, 3), and e ij as residual error of Y ij .In the case of a significant global F-test for differences in SBM levels, individual differences between the SBM means were determined using Tukey's HSD test.To determine the correlation, PROC CORR was used, and linear regression equations for InsP 6 ED 6 were tested using stepwise selection in the REG procedure.
For comparison of individual AA degradation and prediction of individual AA degradation from CP degradation, data from the 17 SBM variants and 3 cows were used, too.Two models were fitted.First, model ( 6) was fitted to each AA to estimate the mean degradation.Note that residuals showed heterogeneous variances as expected from for proportions.Therefore, data were logit-transformed before analysis.To compare degradations from 2 AA, a bivariate model was fitted to each pair of AAs.The model can be described as where k is the index of AA.Therefore, the 2 parameters μ k represent the 2 AA degradations from the pair of AAs considered.For the random effects and the error effects, unstructured 2 × 2 variance-covariance matrices were fitted with AA specific variances on the diagonal and a covariance on the off-diagonal.The bivariate analysis accounts for the correlations due to analyzing measured traits from the same pooled bag sample.Again, observations were logit-transformed before analysis.Means from (6) and mean comparisons from (7) were used to perform a Fishers LSD test.Results were presented via letter display (Piepho 2012).Finally, means were back-transformed for presentation purpose only.Standard errors were back-transformed using the delta method.Note that AA degradation was estimated alternatively using the method of (White et al., 2017).They proposed to predict AA degradation by the ratio of undegradated total AA (TAA ij ) and undegraded CP (CP ij ) using the following equation: For the ratio of undegradability of individual AA (uAAi) to total AA (TAA), data were logarithmically transformed.Both fractions were proportions that can be binomial distributed.According to (Katz et al., 1978), the ratio of 2 binomial distributed variables is approximately normal distributed.Therefore, model ( 6) and ( 7) were applied to logarithmically transformed data and no deviations from the assumed normality were found when checking residual plots.As before, means and mean differences were used to create a letter display.Afterward, means and their standard error were back-transformed for presentation purpose only.
To predict individual AA degradation from CP degradation, model ( 6) and ( 7) were extended by fitting average CP across cows as a covariable.The models can be described as follows: where CP mean i is the CP degradations of SBM variant i averaged across cows, β 1 and β 1k are the general or AA specific slope parameters.All other terms were analogously defined as in model ( 6) and ( 7) including the use of index k for the AA used.Note that β 1 was reported through out the paper.Again, 2 × 2 variance-covariance matrices for all random effects and the error were assumed.Unfortunately, estimated variance parameters for animal effects were generally small resulting in convergence problems, if one or both variances were bounded at zero in the final REML iteration.To get convergence in all bivariate models, the covariance between animal effects was dropped from the model.Statistical significance was set at P ≤ 0.05 in all cases.For all analysis aiming in predicting AA degradation a cross validation was added to evaluate how good the model may fit in future.Therefore, data were split in 17 subsamples each from a single SBM variant.Sixteen subsamples were used to estimate the model, and predict data from the remaining subsamples.The cross validation can be considered as leaf one SBM variant out cross validation.Data from cross validation were used to estimate the coefficient of variation.Additionally, the root mean squared error between predicted and observed AA degradation was estimated for both analysis of AA degradation.

RESULTS AND DISCUSSION
This study aimed to characterize ruminal CP, AA, and InsP 6 degradation, as well as ID RUP for different commercial solvent-extracted SBM.Furthermore, the relationship between CP degradation, AA degradation and InsP 6 degradation was determined, and the prediction of CP and InsP 6 degradation from the chemical protein fractions was investigated.

CRUDE PROTEIN AND AMINO ACIDS
Water-soluble CP corrected for small particle losses in the in situ study was, on average, 3.6% and was only slightly lower than the uncorrected rapidly degradable fraction (a) which showed an average value of 4.5% (Table 2).For this reason, and because 6 out of 17 samples (No. 6,11,12,15,16,and 18) showed lower rapidly degradable than water-soluble fraction (Supplemental Table 2, https: / / doi .org/ 10 .5281/zenodo .7804535),no correction for small particles was applied, as only minor effects on kinetic parameters and CPED were observed (values not shown).This is consistent with the results of Benchaar et al. (2021), who found differences of only 1.5%-points (pp) for the CPED of SBM with or without correction for small particles, as fraction a in their study was only reduced from 27.7% to 24.4%.This indicates that although measurement of the water-soluble fraction should always be performed, correction of in situ CP degradability values for small particle losses is not generally advisable for every class of feedstuff.
Common SBM is characterized by a well-balanced EAA pattern like that of ruminal bacteria and rich in Lys with a slight deficiency in Met (Miranda, 2019).However, because of the extensive ruminal degradation generally described in feed libraries (67% for SBM; NASEM, 2021), the utilization of SBM by ruminants as a source of RUP and metabolizable EAA from RUP is limited.In this study, the slowly degradable fraction (b) averaged 95%, and the estimated maximum degradation (a+b) was complete for all samples.This is consistent with previous studies showing that almost the entire CP in SBM is potentially degradable by rumen microorganisms if no further chemical or physical treatment is applied (Benchaar et al., 2021;Borucki Castro et al., 2007;Cozzi et al., 1995).The degradation of fraction b started after a mean lag phase of 1.7 h (range: 1.1-2.0h) at an average rate of 10% h -1 , but with a high range of the c value from 4.5% to 14% h -1 .Differences in degradation parameters induced a considerable range in CPED 6 with values from 38% to 67%; hence, the concentration of RUP 6 varied widely from 33% to 62% for the 17 SBM under study.Although the degradation parameters, CPED, and RUP were determined using a defined approach in the present study, they were within the range of published values for solvent-extracted SBM among different studies, or partly higher (Benchaar et al., 2021;Awawdeh et al., 2007;Cozzi et al., 1995).For instance, Harstad and Prestløkken (2000) and Benchaar et al. (2021) observed degradation rates of 5.9% and 11.5% h -1 and a CPED of 52.2% or 66.2%, respectively.It cannot be ruled out whether discrepancies in published values of CP degra- dation for standard SBM are due to the sample material or if methodological alterations in the study design may have induced such differences among experiments, as small modifications in every step of the in situ procedure can impact the results of rumen degradation measurements (GfE, 2022).However, we incubated all samples using the same standardized procedure, and the influences of the study design on the ranges in ruminal degradation kinetics and ED were therefore unlikely or at least negligible.
The large differences in CP degradation may be a function of the soybean genotypes used and the processing conditions, which include the intensity of moisture and heat treatment.Faldet et al. (1992) found that RUP content in SBM increased at higher temperatures (120-160°C) and heating durations (10-120 min).To the best of our knowledge, the influence of soybean genotype on the ruminal CP degradability of SBM has not yet been studied.However, the composition of genotypes in SBM might differ considerably because the sample material came from different regions with varying vegetation lengths, determining which soybean variety could be grown.Bachtiar et al. (2022) found differences in in vitro DM digestibility among 30 different soybean genotypes from Indonesia, and Ayasan et al. (2019) detected different gas production rates for 5 genotypes from Turkey.Differences in the feeding value may be mainly due to ranges in the chemical composition.In our study CP degradation was significantly correlated with the chemical protein fractions, which will be discussed in more detail in a subsequent section.Nevertheless, the causes of the differences in the amount of RUP, although not fully understood, could also have led to differences in intestinal digestibility among the SBM variants.However, the ID RUP was high for all SBM and ranged between 87% and 96%, with an average value of 93% (Table 2), which is similar to the value of 91% for solvent-extracted SBM presented by NASEM (2021).The overall high values of ID RUP are consistent with previous results.Awawdeh et al. (2007) observed an average ID RUP of 82% using the same in vitro procedure as in our study and Harstad and Prestløkken (2000) found intestinal indigestibility of RUP measured with the mobile nylon bag technique to be 1.4% and 1.6% for differently processed SBM products without significant differences between samples.
With sufficient knowledge, rations can be balanced for individual AA, thereby driving the implementation of low-protein diets for economic and ecological reasons.White et al. (2017) argued that incorporating AA degradability into nutrition supply-requirement models is challenging because even an extensive literature search yields an incomplete database for important feeds.Until now, it has not been possible to consider the degradability values for individual AA in any feed evaluation system, and it was deemed that a broader database would be needed before the differential profile of the RUP fraction of the feedstuff could be predicted with confidence (NASEM, 2021).Liebe et al. (2018) suggested that values on AA degradation should also include ruminal degradation of total AA (TAA) and White et al. (2017) proposed an approach of normalizing undegradability of individual AA (uAAi) as a proportion of TAA to allow better integration of AA degradability values into feed libraries based on RUP.We characterized a wide range of solvent-extracted SBM for AA degradability, and degradability of TAA showed a LS mean of 83% while LS means of individual AA differed between 80% for Ser and 84% for Glx (Table 3).Differences in degradation between in- dividual AA were partly significant, also in case when uAAi was expressed as a proportion of TAA (Table 3).
The average ratio of uAAi to TAA was highest for Ser (1.12) and higher than 1.0 for Ala, Gly, Leu, Ser, Thr, and Tyr for all SBM.For Met (n = 16), Val (n = 15), Cys (n = 13), and Asx (n = 13) the majority of samples also showed higher undegradability of these AA than their corresponding TAA (Table 3 and Supplemental  Table 3, https: / / doi .org/ 10 .5281/zenodo .7804535).On the other hand, Glx showed the lowest undegradability ratio (0.90) and Arg, and Lys also tended to be lower than 1.0 in all 17 SBM, and 15 out of 17 SBM showed lower undegradability for His compared with TAA.The average ratio of undegradability for Ile, Phe and Pro tended to be 1.0.These results were in accordance with the reviewed data set of White et al. (2017) who found that undegradability of His tends to be lower than that of TAA whereas undegradability of Arg and Lys depended on the feed category.Susmel et al. (1989) and Harstad and Prestløkken (2000) found the highest ED for Arg, Glu and Lys in SBM.Differences among individual AA have also been reported by other authors, and Lys is often considered the most degradable AA (Borucki Castro et al., 2007;O'Mara et al., 1997).Glu was reported to be more degradable than the other AA in SBM products by O'Mara et al. (1997) and in the study of Cozzi et al. (1995) Arg, His and Lys were the only AA with consistently lower concentrations after ruminal exposure for 0, 8, 12, 16, and 24 h compared with the original profile.In contrast, reports on the least degradable AA in SBM are more variable.Harstad and Prestløkken (2000) found that Ser, together with other AA, was less degradable than the sum of AA taken together.However, Met and the branched-chain AA were most often investigated to be the least degradable AA in SBM (Harstad and Prestløkken 2000;Borucki Castro et al., 2007;Maxin et al., 2013) and other feed categories (White et al., 2017;Erasmus et al., 1994;Mijoun et al., 2010).Generally, some studies indicated that EAA are degraded more slowly than non-EAA which may be due to the different distribution of these AA in the feed proteins (Cozzi et al., 1995).Other authors suggested that ruminal bacteria utilize peptides containing hydrophilic AA faster than hydrophobic peptides (Grieswold and Mackie 1997).However, results from other studies do not support these assumptions and showed that the degradation of individual AA is feed dependent (Depardon et al., 1995;Susmel et al., 1989).Therefore, more research on this topic is warranted as a targeted supplementation of rumen-protected AA instead of increasing several protein feeds in the diets may prevent animals from diseases and enhance their productivity while lowering environmental pollution (Kim and Lee 2021;Khan et al., 2022) Compared with the differences between the average values of AA degradation within one SBM, the range in AA and CP degradation between the SBM variants was large.Among the meals studied, the degradability of the first limiting AA for milk production, Met, Lys, and His varied between 34 and 96%, 37-97%, and 34-97%, respectively.The other AA, whether essential or not, also showed high range (Supplemental Table 3; https: / / doi .org/ 10 .5281/zenodo .7804535).
The variable degradation of CP and AA observed in our study may indicate that the supplied amount of RUP and, hence, metabolizable EAA from RUP differ considerably among sources.Therefore, accurate information on ruminal degradability, instead of using average values for diet formulation, is of utmost importance to meet the animal's protein requirements and maximize animal performance while minimizing N losses.
Although the magnitude of the difference between CP and individual AA degradation was not large, it was significant in most cases, suggesting that the degradation of AA in SBM differs from that in CP (Table 3).The regression equations and accuracy of the cross validations showed close relationships between CP degradation and the degradation of Met and Lys (R 2 = 0.96; Table 4) and other AA showing R 2 values between 0.94 and 0.97 (Table 4).The slopes differed among the AA, suggesting that the degradation of each AA varied in magnitude with changes in CP degradation.Slopes and intercepts of the linear regression lines were all higher than 1 and 0, respectively.This can maybe be explained by part of CP that is nondegradable." To the best of our knowledge, this is the first study to investigate the relationship between CP and AA degradation in SBM using the applied approach.Borucki Castro et al. (2007) tested 4 differently treated SBM products and found that within each product, the ED of individual AA was variable and not constant across the SBM.However, statistical validation of the first statement was missing in Borucki Castro et al. (2007), and 3 of their SBM products underwent additional treatment to enhance RUP, which may have altered AA degradation compared with standard solvent-extracted SBM.As previously discussed, other authors have reported differences in the degradation of individual AA degradation within the same meal.The application of a mean degradability value for all AA or the degradability of CP may lead to a biased calculation of the supply of EAA to the duodenum of cows.Linear regressions, as used in our study, can estimate the rumen degradation of all AA based on the CP degradation of SBM (Figure 1 and Supplemental Figure 1, https: / / doi .org/ 10 .5281/zenodo .7804535).The diagnostic plots of the linear regression showed that the studentized residuals were not correlated with the prediction values.Slopes of studentized residuals regressed on the predicted degradation values for each AA were not significantly different from zero in any case.This implies that the error of the prediction did not depend on the AA degradation of the sample indicating an unbiased model with whom prediction of AA degradation is possible with similar accuracy over the entire range.Moreover, accuracy of cross validation to validate the model fit for future prediction showed overall high R 2 (0.94-0.97) and low RMSE (2.92-4.42)(Table 4).This approach has previously been used to estimate AA degradation, and hence, the AA composition of the RUP of DDGS (Westreicher-Kristen et al., 2013) and RSM (Steingass et al., 2013).Hence, the linear regression approach for the prediction of AA degradation based on CP degradability appears to be applicable to different feeds.White et al. (2017) also showed a strong relationship between N (or CP) and TAA degradability over a wide range of feed categories and proposed that future feed libraries could include prediction of undegradability of EAA based on RUP content by multiplication with the ratios of EAA normalized to TAA.However, when AA degradability of the 17 SBM from the present study was predicted from the ratio (uAAi/TAA) of each AA (Table 3) and their corresponding RUP after 16 h (values not shown), RMSE of the prediction model was 4.42% and therefore higher compared with 2.40% of the regression approach.Though, cross validation of both approaches to evaluate how good the model may fit in future showed similar RMSE with 17.2% and 17.3% for the TAA and the linear regression approach, respectively.In conclusion, both approaches show that usage of CP degradability or undegradability measured by in situ procedure or maybe, in the future, estimated from laboratory measurements like chemical protein fractions, could be a robust predictor for ruminal degradability of AA driving their implementation in feed evaluation systems.

PHYTATE
Overall, concentrations of InsP 6 and InsP 5 in the bag residues of the 9 SBM decreased with the incubation time from 0 h (InsP 6 : 14.4 g kg -1 DM; InsP 5 : 2.5 g kg -1 DM) to 24 h (InsP 6 : 3.9 g kg -1 DM; InsP 5 : 0.7 g kg -1 DM) (Figure 2).After 48 h of ruminal incubation, the InsP 6 concentration was reduced to less than 1 g kg -1 DM, and InsP 5 was not detectable in most bag residues.InsP 4 was detected at very low concentration (0.1 g kg -1 DM) until 2 h of ruminal incubation but was barely detectable from 4 h of incubation onwards.InsP 3 was not detected in any bag residue.These changes over time are consistent with the results of previous in Estimates with at least one identical letter were not significantly different from each other (P < 0.05).
3 Asp, Asn, and Glu, Gln, respectively, were detected together because the Asn and Gln side groups were lost during acid hydrolysis (Fontaine, 2003).
situ studies, showing that InsP 6 degradation in SBM begins soon after incubation and proceeds almost completely within 48 h without any accumulation of partially dephosphorylated InsP isomers (Haese et al., 2017).However, a wide range was observed in the in situ degradation parameters of InsP 6 .The ruminal degradation parameters ranged from 11% to 37% for fraction a, 63% to 89% for fraction b and 7.7% to 21% h -1 for degradation rate c (Table 5) with average values of 21%, 79%, and 16% h -1 , respectively.The calculated InsP 6 ED 6 varied considerably among the SBM, ranging from 61% to 84%.The solubility and degradation rate of InsP 6 from treated oilseed meals are strongly affected by processing conditions such as toasting/desolventizing time (Haese et al., 2022), heating (Konishi et al., 1999;Wang et al., 2018) and formaldehyde treatment (Park et al., 1999).Such processing effects may at least partly explain the range in InsP 6 ED 6 among the SBM in the present study because different processing conditions in the oil plants can be assumed.The meals were obtained from the market, and no information on their origin, except the country, was available.Thus, intrinsic differences in the processed soybeans and agronomic conditions may also have contributed to the observed range.

CORRELATIONS AND RELATIONSHIP BETWEEN DEGRADATION AND CHEMICAL PROTEIN FRACTIONS
Significant correlations were detected between the estimated parameters and ED of CP degradation with the chemical protein fractions.The lag phase of CP degradation was positively correlated with the chemical protein fraction B3 (r = 0.60).The CP degradation rate correlated with the chemical protein fractions B1 (r = 0.80), B3 (r = −0.88)and C (r = −0.56)and the same 3 fractions showed significant correlations with CPED 6 (B1: r = 0.76; B3: r = −0.90;C: r = −0.66).Differences in chemical protein fractions could be partly due to the range in soybean genotypes as protein composition varies between soybean varieties (Zilić et al., 2011;Fehr et al., 2003;Pesic et al., 2005).However, unpublished data from our department showed that the chemical protein fractions B1, B2, and B3 are especially influenced by thermal and hydrothermal treatments (sample material and processing conditions described in Kaewtapee et al. (2017)), with B1 decreasing from 73% in raw soybean to less than 1% with increasing intensity of processing conditions, whereas B2 and B3 increased from 22 to 73 and 0% to 16%, respectively.Chemical protein fractions A and C were almost unaffected.These results are consistent with those of previous studies on the influence of microwave irradiation, electron beam irra-  Estimates with at least one identical letter were not significantly different from each other (P < 0.05).
1 Asp, Asn, and Glu, Gln, respectively, were detected together because the Asn and Gln side groups were lost during acid hydrolysis (Fontaine, 2003).
diation, and roasting on the chemical protein fractions of whole soybeans (Akbarian et al., 2014;Golshan et al., 2019).The application of dynamic models for precise diet formulation depends, in addition to other factors, largely on the estimates of feedstuffs feeding values.For example, CNCPS models CPED based on chemical protein fractions and defines the degradation rates for A, B1, B2, and B3 to calculate RUP values for a given passage rate (Sniffen et al., 1992).Using individual degradation rates for chemical protein fractions from CNCPS version 6.5 for the 17 SBM samples under study, the calculated CPED 6 values ranged between 62% and 67% which is, although a significant correlation was found (r = 0.73), much less variable than in situ CPED 6 (Table 2).Hence, in some cases, the RUP values for the SBM were considerably underestimated by this prediction model.Therefore, regression analysis was applied to estimate the in situ CPED 6 directly from chemical protein fractions without the use of fractional degradation rates.The following equation was developed based on stepwise selection using only significant variables as CPED 6 (%) = 69.90-1.54 • B3 (B3 in % of CP) (R 2 = 0.82; RMSE = 3.43).
When examining RSM, Haese et al. ( 2022) calculated significant correlations between InsP 6 ED and CPED and between InsP 6 ED and protein fractions B1, B2, and C.These relationships can be confirmed for SBM of the present study where a significant correlation was observed between InsP 6 ED 6 and CPED 6 (r = 0.88) and between InsP 6 ED 6 and chemical protein fractions A, B1, B2, B3, and C (A: r = −0.91;B1: r = 0.69; B2: r = 0.70; B3: r = −0.85;C: r = −0.91).For SBM, InsP 6 was found to interact naturally with the protein in the seed (Hídvégi and Lásztity, 2002;Prattley and Stanley, 1982) which may be strengthened by the further formation of protein-phytate complexes during mechanical processing (Wang et al., 2018), which could explain the observed association between InsP 6 and CP degradation.
In contrast to the high digestibility of RUP from SBM in the intestine, the post-ruminal digestion of InsP 6 is very low (Chi et al., 2022), and degradation of InsP 6 in the large intestine does not vary with the flow of InsP 6 (Ray et al., 2012(Ray et al., , 2013)).Thus, increasing the amount of rumen undegraded InsP 6 does not increase P available for the animals but excretion of InsP 6 in feces, which shows the contradicted interest of high RUP content for supporting high production and milk quality.
Based on the observed range in the degradation of CP and InsP 6 and the significant correlation between them, we suggest equations to estimate ruminal InsP 6 degradation of commercial SBM depending on CP degradation values or CP fractionation when formulating diets, which has also been proposed for RSM (Haese et al., 2022).The linear regression equation for InsP 6 ED (y) depending on CPED 6 (x) was y (%) = 0.76x (%) + 34.7 (R 2 = 0.77; RMSE = 4.34; P < 0.05) (Figure 3).For the linear regression based on chemical protein fractions, variables were selected as described above and resulted in InsP 6 ED 6 (%) = 103 − 0.76 • B3 − 15.4 • C (R 2 = 0.92; RMSE = 2.80; P < 0.05).While both approaches seem to be applicable, equations based on chemical protein fractions showed remarkably higher accuracy (higher R 2 and lower RMSE) than the regression equation based on CPED, which allows a more precise prediction of InsP 6 ED for individual SBM.Furthermore, an approach based on chemical protein fractions is less time-consuming, and the use of animals for in situ experiments is not necessary.To evaluate the applicability of one general equation for different feedstuffs, we also calculated InsP 6 ED for SBM in the present study using the equations proposed for RSM by Haese et al. (2022), which were y = 1.01x -2.38 for InsP 6 ED (y) depending on CPED (x) and InsP 6 ED (%) = −233.6− 0.38 • B1 + 6.45 • B2 + 0.74 • C at a rumen passage rate of 5%/h.Nevertheless, the calculation based on equations developed for RSM resulted in a considerable underestimation and overestimation of InsP 6 ED by CPED and chemical protein fractions, respectively (values not presented here), implying that the equations for one oilseed meal type may not be applicable to another.Oilseeds differ in their InsP 6 to protein ratio (Kies et al., 2006), InsP 6 distribution (Prattley and Stanley, 1982;Gillespie et al., 2005), and the occurrence of inorganic cations (Hídvégi and Lásztity, 2002) within the protein storage vacuole, which can influence the interaction between InsP 6 and CP and the degradation of InsP 6 and CP.In addition, storage proteins in seeds exhibit different chemical or physical features and reactions, such as the aggregation or denaturation of storage proteins during the production of oilseed meals, which can further affect the degradation of associated components (Yiu et al., 1983).Therefore, feedstuff-specific equations may be necessary to estimate the InsP 6 ED.

CONCLUSIONS
Different SBM showed considerable range in ruminal degradation parameters and the ED of CP and InsP 6 .Data on CP, AA, and InsP 6 degradability can be used to extend the databases for SBM as an important protein feed for dairy cows and to support the implementation of AA degradability values in protein evaluation systems.Close relationships between the CP degradation and degradation of individual AA and InsP 6 were detected, and regression equations were proposed and can be used to estimate the degradation values of CPED and InsP 6 ED of SBM based on chemical protein fractions.a-f Estimates with at least one identical letter were not significantly different from each other (P < 0.05). 1 a = rapidly degradable fraction; b = slowly degradable fraction; c = degradation rate; InsP 6 ED 6 = calculated effective degradation of InsP 6 at a rumen passage rate of 6% h -1 .
2 SEM = standard error of the mean.
Titze et al.: Amino acid and phytate degradation of SBM Table 3. Adjusted median and their standard error 1 of in situ degradability of CP, total amino acids (TAA) and individual AA (AAi) after 16 h with corresponding undegradability ratios 2 of soybean meal (n = 17 variants).The values for the individual soybean meal variants are presented in Supplemental Table 3 (https: / / doi .org/ 10 .5281/zenodo .7804535).

Figure 1 .
Figure 1.Diagnostic plots for predictions of ruminal degradation of Met (A) and Lys (B) from the degradation of CP after 16 h ruminal incubation.Filled symbols and solid lines represent observed versus predicted data, and clear symbols with broken lines represent studentized residuals versus predicted data.Intercepts of the regression lines are not different from 0 and slopes of the regression lines are not different from 1 (observed vs. predicted) and 0 (studentized residuals vs. predicted) (P > 0.99) (n = 51).The diagnostic plots for the other AA are presented in Supplemental Figure 1A-O (https: / / doi .org/ 10 .5281/zenodo .7804535).

Figure 2 .
Figure 2. Mean concentrations of inositol phosphates including phytate (InsP 6 ) and the sum of less phosphorylated inositol phosphate isomers (InsP 5 ) in the bag residues of 9 soybean meal variants at different time periods of incubation in the rumen (g kg -1 DM).
Titze et al.: Amino acid and phytate degradation of SBM
Titze et al.: Amino acid and phytate degradation of SBM
1Water solubility determined with filter paper; a = rapidly degradable fraction; b = slowly degradable fraction; c = degradation rate; CPED 6 = calculated effective degradation of CP at a rumen passage rate of 6% h -1 ; RUP 6 = calculated rumen undegradable CP at a rumen passage rate of k = 6% h -1 .2ID RUP = Intestinal digestibility of RUP after 16 h incubation in the rumen.
Titze et al.: Amino acid and phytate degradation of SBM Titze et al.: Amino acid and phytate degradation of SBM

Table 4 .
Parameter estimates of the linear regression parameters and model fit critera of AA degradation (y, %) from CP degradation (x, %) after 16 h of ruminal incubation (n = 17 soybean meal variants)