Open Access
How to translate text using browser tools
17 August 2022 Evaluation of prediction of indigestible fiber fraction (iNDF) of whole-crop barley silage by using non-destructive spectroscopic techniques as a fast-screening method: comparison between FTIR vs. NIR
Basim Refat, Peiqiang Yu
Author Affiliations +
Abstract

The objective of this study was to reveal the potential of using Fourier transform mid-infrared (FTIR) and near infrared (NIR) spectroscopy as tools for the determination of indigestible neutral (NDF) fraction (iNDF) of whole-crop barley silage. A total of 48 whole-crop barley silage samples collected from 48 different farms in Western Canada were analyzed for iNDF. Reference values were matched with NIR and FTIR spectra. Spectral data processing (pretreatments) included first derivative, standard normal variate, multiplicative scattering correction, second derivative, and orthogonal signal correction. Prediction equations were obtained from each model using an external validation set. The coefficient of determination for the external validation of iNDF was 0.62 for FTIR and 0.41 for NIR, while the corresponding ratio performance deviation was 1.69 and 1.38 for FTIR and NIR, respectively. Results from this research showed the high potential of applying infrared molecular spectroscopy for the examination of forage plant fiber digestibility. More studies are needed to improve the accuracy and performance of FTIR and NIR spectroscopies in predicting the iNDF of whole-crop barley silage samples.

1. Introduction

Forage fiber digestibility can directly affect animal production (e.,g. milk, meat), rumen function, and animal health. Fiber digestibility of forage is often less than 60%, therefore enhancing the digestibility of forages would result in improving animal performance and sustainability of the production system (Van Soest 1994). Recently, numerous nutritional models have been developed to more precisely predict the nutritive value of forage. Cornell Net Carbohydrate and Protein System (CNCPS) is a mechanistic nutrition system which was generated to predict feed utilization, dairy, and beef cattle nutrition requirements and their performance using accumulated information regarding feed value, digestibility, and metabolism in providing nutrients to meet their requirements (Fox et al. 2004). The main requirement for CNCPS to predict the potentially digestible neutral detergent fiber (NDF) is to estimate the indigestible NDF (iNDF). The CNCPS 6.1 model used acid detergent lignin (ADL) content in calculating iNDF (Tylutki et al. 2008). However, it has been found that iNDF calculated from ADL content could under-estimate the metabolizable energy and over-predict the potential milk production in ruminants (Harper and McNeill 2015). Previous studies showed that the iNDF content is a function of the cross-linkage between hemicellulose and lignin rather than total lignin content, in particular, C4 forages (Van Soest 1994). The recent feeding studies have found that iNDF after long incubation time (240 h using in vitro method or 288 h using in situ method) was highly correlated with dry matter intake of ruminants (van Amburgh et al. 2015). Currently, the updated CNCPS6.55 model recommends using iNDF derived from a longer incubation time of forage to accurately predict the available metabolizable energy content of forages (Higgs et al. 2015). However, in vitro and in situ procedures to estimate the iNDF in forage or fecal samples are time consuming, costly, and limited by intrinsic factors. Thus, other methods need to be evaluated for routine analysis.

A vibrational molecular spectroscopic method has been developed in recent years as a rapid, direct, non-destructive, and non-invasive bioanalytical technique (Yu 2004). In recent years, several studies have reported that the chemometric methods based on reflectance near-infrared (NIR) spectra can accurately predict ruminal degradability of feed nutrients such as fiber NDF and crude protein in plant forage samples (Andrés et al. 2005; Ohlsson et al. 2007; Shi et al. 2019). Nevertheless, there are few reports in which mid-infrared vibrational (MIR) spectrometry is employed to predict the feeding value of forage samples. The MIR spectroscopic methods are able to provide more benefit than NIR analysis because the light absorptivity is greater in the MIR than in the NIR range. Moreover, the MIR range (4000–400 cm−1) contains important vibration information of the fundamental molecules, which are stronger than the overtone absorptions observed in NIR (12500–4000 cm−1) (Manley 2014). As a result, MIR spectroscopy generally has both greater precision and greater sensitivity than NIR spectroscopy and therefore could give better insight into the chemical composition and nutrient profiles of feed samples. In recent years, the chemometric analysis, i.e., principal component analysis and partial least square regression of Fourier transform mid-infrared (FTIR) or NIR data, has been reported to accurately predict several feed nutrients such as lignin, ash, and crude protein content in several forage grasses (Nordheim et al. 2007; Belanche et al. 2014). Nevertheless, vibrational FTIR spectral technology has not yet been used to predict the rumen degradation kinetic parameters of many forages such as barley silage. Furthermore, there are no reports comparing NIR to MIR as analytical tools in predicting the rumen degradation parameters of common forages in ruminants. The objectives of this study were to (i) develop a methodology that employs non-destructive molecular spectroscopy techniques with chemometrics (partial least square [PLS]) as fast analytical tools to reveal indigestible fiber fraction (iNDF) for ruminant systems and (ii) compare two molecular spectroscopic techniques (grading NIR vs. attenuated total reflectance (ATR) - Fourier transform infrared (FTIR) spectroscopy (ATR-FTIR)) in association with chemometrics for the examination and prediction of whole-crop barley silage iNDF for ruminant systems.

2. Materials and methods

2.1. Samples and chemical analysis

Forty-eight whole-crop barley silage samples were used for the study. The whole-crop barley silage samples were collected within 2 years from dairy and beef operators in the western Canada (Saskatchewan and Alberta). All whole-crop barley silage samples represent seven varieties of barley which are commonly used in western Canada by the dairy and beef producers. Selected barley forage varieties included Conlon, which is a smooth awned two row feed and malting type barley; CDC, Copeland, and AC Metcalfe, which are two row malting barley varieties with rough awns; CDC Cowboy and Xena, both are two row feed barley varieties with rough awns; and Falcon and Legacy, which are six row varieties with smooth awns. Barley forages were harvested at the mid-dough stage of maturity. The detailed ensiling procedure has been previously reported (Refat 2018). The detailed sampling procedure of whole-crop barley silage is described by Nair et al. (2016). All whole-crop barley silage samples were oven-dried in a forced-air oven at 55 °C for 2 days. The dried samples were then ground through a 1-mm screen (Christy & Norris 20 cm arm Lab mill, Christy Turner Ltd. Chelmsford, UK). The iNDF content of samples was analyzed as described by Ahvenjärvi et al. (2000). Briefly, the iNDF content of each feed sample was determined following in situ incubations for 288 h in the rumen using eight ruminally cannulated beef heifers (452 ± 10 kg; mean ± standard deviation). The detailed animal and diet information were previously reported (Nair et al. 2016). The incubation of feed samples was repeated twice (two consecutive runs as replicate). Around 3 g of ground sample (1 mm) was weighed in triplicate into 5 × 10 cm size custom-made in situ nylon bags (6 µm pore size, part No. 07 -6/5, Sefar America Inc., Depew, NY, USA) and were randomly assigned to the cows. After removal from the rumen, the bags were rinsed with cold water and oven-dried at 55 °C for 48 h. The feed residues were analyzed for NDF content.

2.2. Spectra acquisition

Vibrational Fourier transform infrared molecular spectroscopy was used for collecting the molecular vibrational spectra. The spectra were generated from the MIR region (4000–800 cm−1) using JASCO Spectramanager II software (JASCO Corporation, Tokyo, Japan) with a spectral resolution of 4 cm−1 and 128 co-added spectral scans. All silage samples were scanned three times (Shi et al. 2019). The detailed procedure and method of the FTIR-ATR including background correction were reported in our previous publications (Shi et al. 2019).

The Unity SpectraStar 2500XL-R spectrometer (Unity Scientific, Brookfield, CT, USA) was used to collect the NIR spectra (680–2500 nm). The spectrometer is operated in reflectance mode which uses a tungsten halogen lamp as a light source. All samples were transferred into a rotating cup with a multi-cup adapter. The Unity InfoStar V3.11.3 software was used to record the NIR spectra.

2.3. Spectral chemometric method and analysis

All the spectra and the reference values for whole-crop barley silage samples were checked, and they were normally distributed. These spectra and the reference values for whole-crop barley silage samples were transferred to the Unscrambler® X version 10.3 software (CAMO Software, Oslo, Norway) (Shi et al. 2019). The raw spectra were transformed by different preprocessing methods, which include orthogonal signal correction (OSC), baseline offset correction, detrending, first derivative (FD), second derivative (SD), multiplicative scattering correction (MSC), and standard normal variate (SNV). For FD and SD, the spectra were derivatized to the first or second Savitsky–Golay derivative (15 units of filter width and 2 polynomial order). The principal component analysis (PCA) was used to remove the outliers and to check spectra patterns. The potential outliers were determined based on Hotelling's T2 or the F-residuals and values in sample sets after the first PCA analysis. Samples that have higher F-residual and Hotelling's T2 values would be deleted from the whole-crop barley silage samples (Lohr et al. 2016). Then, all the remaining barley silage samples were rearranged in an ascending order according to their iNDF values and divided into a calibration set and a prediction set (2:1 ratio) such that each set covers a similar range of iNDF values.

The regression models considering the entire spectral data were constructed using PLS regression. The entire wavelength as well as specific regions were employed for the calibration. The prediction of specific regions such as the fingerprinting region (1000–1500 cm−1) was tested; but they resulted in poor models, and therefore these regions were not applied in the final procedures. During the calibration stage, the model was trained on a random set of samples (containing between 81% and 86% of the samples) using a leave-one-out cross-validation procedure. Furthermore, the best PLS models were further assessed by an external prediction set of samples. Ultimately, the established models were evaluated for their accuracy in terms of the following parameters: root mean square error (MSE) of calibration (RMSEC), cross-validation (RMSECV), and prediction (RMSEP); determination coefficient for calibration (R2C), cross-validation (R2CV), and prediction (R2P); standard error (SE) of calibration (SEC), cross-validation (SECV), and prediction (SEP). The root mean square error is usually used to assess the robustness of the model. Furthermore, the range error ratio (RER) as well as the ratio relative predictive determinant (RPD) were computed to assess the acceptability of the models. The standard deviation of iNDF values in the prediction group was divided by the RMSEP to calculate RPD. The range of iNDF values of the prediction set was divided by the RMSEP to compute the RER (Williams 2003; Shi et al. 2019).

3. Results and discussion

3.1. Sample statistics and spectral characteristics of forage whole-crop barley silage

The iNDF content of whole-crop barley silage samples ranged from 12.1% to 30.0% with an average of 22.1% (Table 1). The large range of reference values in the current study (about 18 units difference) would be very helpful to obtain a reliable calibration model and to better evaluate the performance of the PLS models that will be constructed based on NIR and FTIR techniques (Shi et al. 2019).

Table 1.

Whole-crop barley silage varieties and the number of samples used for chemical analysis of indigestible NDF.

cjps-2022-0054_tab1.gif

In the current study, the remaining samples after removing the outliers were divided into two sets: the calibration set (28 samples) and the prediction set (13 samples). The detailed information on whole-crop barley silage samples statistics for the calibration and prediction sets are shown in Table 2, which includes whole-crop barley silage sample number, max, min, average, range, and standard deviation. The range of iNDF values in the calibration set covers that of the prediction set such that the distribution of samples would be very suitable for the calibration and the external validation procedures (Shi et al. 2019).

Table 2.

Statistical summary of reference values of the indigestible NDF (iNDF) in the calibration and prediction sets of whole-crop barley silage samples.

cjps-2022-0054_tab2.gif

Figure 1 shows the original MIR (Fig. 1a) and NIR (Fig. 1b) spectral patterns of whole-crop barley silage samples. The characteristics of the MIR and NIR spectra for the whole-crop barley silage samples represent different chemical functional groups (e.g., amide I, amide II, C=O, symmetric and asymmetric CH2 and CH3), which correlated with feed nutrients such as moisture, carbohydrates, lipids, and proteins. Our previous study showed that ligneous compounds are associated with the spectra region ca. 1525–1487 cm−1, and within this region one major peak is centered at 1512 cm−1 (Refat et al. 2017). Another region, such as ca. 1487–1189 cm−1, is mainly associated with hemi- and cellulosic compounds and within this region there are three major peaks (1416, 1372, and 1320 cm−1). Furthermore, it was reported that structural carbohydrates related to a region ranged from 1292 to 1189 cm−1 (Li et al. 2016; Refat et al. 2017).

Fig. 1.

MIR (a) and NIR (b) spectra, without pretreatment for whole-crop barley silage samples.

cjps-2022-0054_f1.jpg

A fundamental vibrational transition occurs between the ground (n = 0) and first-excited (n = 1) vibrational states and involves a single vibrational quantum in the MIR region (4000–400 cm−1). Overtone bands involve two quanta of the same vibrational mode, and combination bands involve vibrational quanta of (at least) two different vibrational modes. Overtone and combination bands of X–H (X = C, N, O) stretching modes occur in the NIR region. A previous study by Givens et al. (1992) showed that the region between 1672 and 2254 nm has a positive correlation with in vitro digestion time, such that the spectral intensity of this region increased positively with extending the incubation time of feed in the rumen. Other studies also reported that the region between 1414 and 2400 nm has a positive correlation with rumen degradation fractions in grass silage (Nordheim et al. 2007). To our knowledge, there were no other reports on studying the importance of wavelength and spectral areas in predicting the iNDF of whole-crop barley silage. The current study showed a number of iNDF-correlating spectral regions exist across the whole spectrum, such as 1582, 1914, 2103, and 2251 nm (Fig. 2b). However, the identification of wavelength within NIR or MIR regions should be further studied before any conclusions are drawn.

Fig. 2.

Regression coefficient represents the most important spectra responsible for PLS scores related to indigestible fiber in FTIR (a) and NIR (b) spectroscopies.

cjps-2022-0054_f2.jpg

3.2. Development of NIR and MIR prediction models and evaluating the accuracy and performance of prediction models

MIR calibration and cross-validation for iNDF of whole-crop barley silage are given in Table 3. Calibration performed well for iNDF by FTIR, when the spectra preprocesses by the OSC + SNV method (R2 = 0.82, RMSEC = 1.88, and SEC = 1.92%), OSC method (R2 = 0.83, RMSEC = 1.74, and SEC = 1.72%), and without preprocessing (R2 = 0.89, RMSEC = 1.49, and SEC = 1.51%). Mean results for the prediction for iNDF by FTIR also showed a higher capability of this technique to predict the iNDF in whole-crop barley silage, where the ratio of prediction to deviation was highest in the spectra that were not preprocessed (1.69), intermediate in the spectra that were preprocessed by OSC (1.50), and lowest in the spectra that were processed by OSC + SNV or second derivative methods. The SNV is used to reduce the light scattering and correct baseline offset width, while OSC is able to remove parts that are linearly unrelated (orthogonal) to the response matrix, Y. The results of the current study showed that the preprocessed samples have a lower R2P than samples that were not preprocessed. In the present study, the spectra of whole-crop barley silage were collected with 128 co-added spectral scans to significantly reduce the noise levels, which may improve the quality of spectra. There is no report on the accuracy of FTIR in predicting the iNDF of the whole crop for small grain forages such as wheat, oat, triticale, and whole-crop barley silages. Only one study by Mathison et al. (1999) found that NIR could not predict the lignin content of barley straw, while a reasonable prediction of the degradable fraction of dry matter was observed (R2P = 0.69). A study by Belanche et al. 2014 explored the potential of PLS and FTIR to predict the in situ iNDF of grass-clover and legume forages (n = 110), demonstrating that they found the FTIR could be used to predict iNDF with R2CV = 0.73. They also concluded from their large-scale study that FTIR is able to quantify the concentration of NDF and the effective degradability of NDF, as well as to screen forage samples according to their iNDF content. However, this study only investigated a small group of forages that are commonly grown in Europe and the stage of maturity or varieties were not specified. Therefore, further studies are needed for the evaluation of optimal strategies for FTIR equations development regarding the botanical composition of the plant, stage of maturity at harvest, and conservation methods (silage vs. hay vs. green chopping). The R2 value is commonly used in the multivariate regression, as a powerful tool to evaluate and examine the quality of the fit. In general, an excellent prediction model should have a higher R2 value higher than 0.91 and a good prediction model should have an R2 value between 0.82 and 0.90; while R2 values from 0.66 to 0.91 would indicate an approximate quantitative prediction. However, models with R2 values from 0.50 to 0.65 can perform discrimination analysis (Williams 2003). Furthermore, RPD and RER were used in the current study to assess the prediction set. The RPD and RER were used to standardize RMSEP against standard deviation and range of iNDF values in the prediction subsets, respectively. Models with the RPD value of less than 1.50 indicate that the model is insufficient for practical application (Williams 2003). For RER, a model with values above 10 is excellent, while the model with values between 3 and 10 is considered limited to good, but the poor model has RER values less than 3. According to Table 4, the calibration, cross-validation, and prediction errors were higher for NIR compared with FTIR. Furthermore, the R2P of NIR was lower than that of FTIR. There is no report comparing the capability of NIR vs. FTIR to predict the in situ NDF degradation parameters of forage in the rumen. Recently, Brogna et al. (2018) reported that the NIR technique is able to predict the in vitro iNDF of fecal samples (n = 276; mean = 43.9% dry matter (DM); standard deviation = 6.3% DM), but the performance of the model (R2CV = 085; RER = 14.0) was not as strong as other properties such as NDF (R2CV = 087; RER = 20.2) or acid detergent fiber (ADF) (R2CV = 088; RER = 27.4). Righi et al. (2017) also reported low accuracy of NIR to predict the iNDF (R2P = 0.59; RPD = 1.52) in fecal samples (n = 126; mean = 40.4% DM; standard deviation = 5.6% DM). The low RER or RPD values in the current study and the previous could be due to the fact that the fiber is not a homogeneous fraction, where it contains different chemical elements (digestible (slow + fast digestible fractions) + indigestible fraction) and there are many regions in the spectra that could alter the relationship between the spectra and the reference method. The lower accuracy of the prediction model for both techniques could be also related to the lower range of variability for the reference value, which had a negative effect on the model performance. It is also difficult to compare our results with any other studies, due to different feed samples (fecal samples, species, and conservation method of forage). Furthermore, there is a large variation between the laboratories in terms of the determination of iNDF (i.e., in situ vs. in vitro evaluations), which could result in a large analytical error compared with the NDF and ADF analyses. In addition, the accuracy of reference values for iNDF might be another challenge during the calibration, since many factors affect digestibility (Gilani et al. 2012). In the current study, some models for iNDF obtained excellent R2C (above 0.90), while their cross-validation or external prediction performance was much lower. The result suggested that good calibration performance does not automatically result in a good external prediction. A good fit to a calibration model does not mean that the model itself has to be a good predictor of a given variable (Hell et al. 2016; Shi et al. 2019). Using MIR (FTIR) in the current study led to better results than those obtained with NIR because the MIR provided more specific information due to the sharp bands in the MIR spectral range when compared with broad bands in the NIR spectral region.

Table 3.

Statistical summary of reference indigestible NDF (iNDF) content in the calibration and prediction sets of barley silage samples by FTIR.

cjps-2022-0054_tab3.gif

Table 4.

Statistical summary of reference indigestible NDF (iNDF) content in the calibration and prediction sets of whole-crop barley silage samples by NIR.

cjps-2022-0054_tab4.gif

4. Conclusions

The potential for using MIR (FTIR) and NIR techniques in combination with the PLS method to determine iNDF in whole-crop barley silage was investigated. The regression models for determining iNDF were only able to make an approximate quantitative prediction of iNDF for screening the whole-crop barley silage for their digestible fiber content. Results from this study further indicate that MIR spectroscopy has more potential to predict the digestible fiber content of whole-crop barley silage than NIR spectroscopy. However, more studies are required to improve the performance of spectroscopic models for the accurate prediction of iNDF in whole-crop barley silage.

Acknowledgements

The SRP Chair (PY) research is supported by the Ministry of Agriculture Strategic Research Chair (PY) Program Fund, Natural Sciences and Engineering Research Council of Canada (NSERC-Individual Discovery Grant and NSERC-CRD Grant), Agricultural Development Fund (ADF), Prairie Oat Growers Association (POGA), SaskPulse Growers, SaskMilk, Saskatchewan Forage Network (SNK), Western Grain Research Foundation (WGRF), etc.

Submission declaration

The work described here has not been published previously and is not under consideration for publication elsewhere.

Data availability statement

The data that support the findings of this study are available from the authors upon reasonable request.

References

1.

Ahvenjärvi, S., Vanhatalo, A., Huhtanen, P., and Varvikko, T. 2000. Determination of reticulo-rumen and whole-stomach digestion in lactating cows by omasal canal or duodenal sampling. Br. J. Nutr. 83: 67–77. https://doi.org/10.1017/s0007114500000106.pmid:10703466. Google Scholar

2.

Andrés, S., Calleja, A., Peláez, R., Mantecón, A.R., and Giráldez, F.J. 2005. How can NIRS method be used to predict in situ crude protein and neutral detergent fibre degradation in herbage? J. Anim. Feed Sci. 14: 727–736. https://doi.org/10.22358/jafs/67164/2005Google Scholar

3.

Belanche, A., Weisbjerg, M.R., Allison, G.G., Newbold, C.J., and Moorby, J.M. 2014. Measurement of rumen dry matter and neutral detergent fiber degradability of feeds by Fourier-transform infrared spectroscopy. J. Dairy Sci. 97: 2361–2375. https://doi.org/10.3168/jds.2013-7491Google Scholar

4.

Brogna, N., Palmonari, A., Canestrari, G., Mammi, L., Dal Prà, A., and Formigoni, A. 2018. Technical note: near infrared reflectance spectroscopy to predict fecal indigestible neutral detergent fiber for dairy cows. J. Dairy Sci. 101: 1234–1239. https://doi.org/10.3168/jds.2017-13319.pmid:29248234. Google Scholar

5.

Fox, D.G., Tedeschi, L.O., Tylutki, T.P., Russell, J.B., Van Amburgh, M.E. Chase, L.E., et al. 2004. The Cornell net carbohydrate and protein system model for evaluating herd nutrition and nutrient excretion. Anim. Feed Sci. Technol. 112: 29–78. https://doi.org/10.1016/j.anifeedsci.2003.10.006Google Scholar

6.

Gilani, G.S., Xiao, C.W., Cockell, K.A., Gilani, G.S., Xiao, C.W., and Cockell, K.A. 2012. Impact of antinutritional factors in food proteins on the digestibility of protein and the bioavailability of amino acids and on protein quality impact of antinutritional factors in food proteins on the digestibility of protein and the bioavailability of. Br. J. Nutr. 108: S315–S332. https://doi.org/10.1017/s0007114512002371.pmid:23107545. Google Scholar

7.

Givens, D.I., Baker, C.W., and Zamime, B. 1992. Regions of normalised near infrared reflectance difference spectra related to the rumen digestion of straws. Anim. Feed Sci. Technol. 36: 1–12. https://doi.org/10.1016/0377-8401(92)90081-gGoogle Scholar

8.

Harper, K., and McNeill, D. 2015. The role iNDF in the regulation of feed intake and the importance of its assessment in subtropical ruminant systems (the role of iNDF in the regulation of forage intake). Agriculture, 5: 778–790. https://doi.org/10.3390/agriculture5030778Google Scholar

9.

Hell, J., Prückler, M., Danner, L., Henniges, U., Apprich, S. Rosenau, T., et al. 2016. A comparison between near-infrared (NIR) and mid-infrared (ATR-FTIR) spectroscopy for the multivariate determination of compositional properties in wheat bran samples. Food Control, 60: 365–369. https://doi.org/10.1016/j.foodcont.2015.08.003Google Scholar

10.

Higgs, R.J., Chase, L.E., Ross, D.A., and Van Amburgh, M.E. 2015. Updating the Cornell Net Carbohydrate and Protein System feed library and analyzing model sensitivity to feed inputs. J. Dairy Sci. 98(9): 6340–6360. Google Scholar

11.

Li, X., Hannoufa, A., Zhang, Y., and Yu, P. 2016. Gene-silencing-induced changes in carbohydrate conformation in relation to bioenergy value and carbohydrate subfractions in modeled plant (Medicago sativa) with down-regulation of HB12 and TT8 transcription factors. Int. J. Mol. Sci. 17: 1–14. https://doi.org/10.3390/ijms17050720Google Scholar

12.

Lohr, D., Tillmann, P., Zerche, S., Druege, U., Rath, T., and Meinken, E. 2016. Non-destructive measurement of nitrogen status of leafy ornamental cuttings by near infrared reflectance spectroscopy (NIRS) for assessment of rooting capacity. Biosyst. Eng. 148: 157–167. https://doi.org/10. 1016/j.biosystemseng.2016.06.003Google Scholar

13.

Manley, M. 2014. Near-infrared spectroscopy and hyperspectral imaging: Non-destructive analysis of biological materials. Chem. Soc. Rev. 43: 8200–8214. https://doi.org/10.1039/c4cs00062e.pmid:25156745. Google Scholar

14.

Mathison, G.W., Hsu, H., SoofiSiawash, R., RecinosDiaz, G., Okine, E.K., Helm, J., and Juskiw, P. 1999. Prediction of composition and ruminal degradability characteristics of barley straw by near infrared reflectance spectroscopy. Can. J. Anim. Sci. 77: 519–523. https://doi.org/10.4141/a99-011Google Scholar

15.

Nair, J., Christensen, D., Yu, P., Beattie, A.D., McAllister, T. Damiran, D., et al. 2016. A nutritional evaluation of common barley varieties grown for silage by beef and dairy producers in western Canada. Can. J. Anim. Sci. 96: 598–608. https://doi.org/10.1139/cjas-2016-0032Google Scholar

16.

Nordheim, H., Volden, H., Fystro, G., and Lunnan, T. 2007. Prediction of in situ degradation characteristics of neutral detergent fibre (aNDF) in temperate grasses and red clover using near-infrared reflectance spectroscopy (NIRS). Anim. Feed Sci. Technol. 139: 92–108. https://doi.org/10.1016/j.anifeedsci.2006.11.024Google Scholar

17.

Ohlsson, C., Houmøller, L.P., Weisbjerg, M.R., Lund, P., and Hvelplund, T. 2007. Effective rumen degradation of dry matter, crude protein and neutral detergent fibre in forage determined by near infrared reflectance spectroscopy. J. Anim. Physiol. Anim. Nutr. (Berl). 91: 498–507. https://doi.org/10.1111/j.1439-0396.2007.00683.x.pmid:17988354. Google Scholar

18.

Refat, B. 2018. Molecular structure features and nutrient availability and utilization of barley silage varieties with varying digestible structural carbohydrate in comparison with a new short-season corn silage in high-producing dairy cattle. PhD Thesis, University of Saskatchewan, Canada. Google Scholar

19.

Refat, B., Prates, L.L., Khan, N.A., Lei, Y., Christensen, D.A. McKinnon, J.J., et al. 2017. Physiochemical characteristics and molecular structures for digestible carbohydrates of silages. J. Agric. Food Chem. 65(41): 8979–8991. https://doi.org/10.1021/acs.jafc.7b01032.pmid:28914059. Google Scholar

20.

Righi, F., Simoni, M., Visentin, G., Manuelian, C.L., Currò, S., Quarantelli, A., and De Marchi, M. 2017. The use of near infrared spectroscopy to predict faecal indigestible and digestible fibre fractions in lactating dairy cattle. Livest. Sci. 207: 105–108. https://doi.org/10.1016/j.livsci.2017.10.006Google Scholar

21.

Shi, H., Lei, Y., Prates, L.L., and Yu, P. 2019. Evaluation of near-infrared (NIR) and Fourier transform mid-infrared (ATR-FT/MIR) spectroscopy techniques combined with chemometrics for the determination of crude protein and intestinal protein digestibility of wheat. Food Chem. 272: 507–513. https://doi.org/10.1016/j.foodchem.2018.08.075.pmid:30309575. Google Scholar

22.

Tylutki, T.P., Fox, D.G., Durbal, V.M., Tedeschi, L.O., Russell, J.B. Van Amburgh, M.E., et al. 2008. Cornell net carbohydrate and protein system: a model for precision feeding of dairy cattle. Anim. Feed Sci. Technol. 143: 174–202. https://doi.org/10.1016/j.anifeedsci.2007.05.010Google Scholar

23.

van Amburgh, M.E., Collao-Saenz, E.A., Higgs, R.J., Ross, D.A., Recktenwald, E.B. Raffrenato, E., et al. 2015. The Cornell net carbohydrate and protein system: updates to the model and evaluation of version 6.5. J. Dairy Sci. 98(9): 6361–6380. https://doi.org/10.3168/jds.2015-9378.pmid:26142847. Google Scholar

24.

Van Soest, P.J. 1994. Nutritional ecology of the ruminant. 2nd Edition, Cornell University Press: Ithaca, USA. Google Scholar

25.

Williams, P. 2003. Near-infrared technology——Getting the best out of light. PDK Grain, Nanaimo, Canada. Google Scholar

26.

Yu, P. 2004. Application of advanced synchrotron radiation-based Fourier transform infrared (SR-FTIR) microspectroscopy to animal nutrition and feed science: a novel approach. Br. J. Nutr. 92: 869–885. https://doi.org/10. 1079/bjn20041298.pmid:15613249. Google Scholar
© 2022 The Author(s).
Basim Refat and Peiqiang Yu "Evaluation of prediction of indigestible fiber fraction (iNDF) of whole-crop barley silage by using non-destructive spectroscopic techniques as a fast-screening method: comparison between FTIR vs. NIR," Canadian Journal of Plant Science 102(6), 1130-1138, (17 August 2022). https://doi.org/10.1139/cjps-2022-0054
Received: 7 March 2022; Accepted: 17 July 2022; Published: 17 August 2022
KEYWORDS
choix d’une longueur d’onde
digestibilité des fibres
ensilage d’orge complète
fiber digestibility
mid-infrared and near spectroscopy
spectroscopie dans le moyen et dans le proche infrarouge
wavelength selection
Back to Top