Food and Environment Safety - Journal of Faculty of Food Engineering, Ştefan cel Mare University – Suceava Year IX, No3 - 2010 105 NEAR INFRARED SPECTROSCOPY – AN ALTERNATIVE TO DETERMINE THE CRUDE FIBER CONTENT OF FORAGES Monica HĂRMĂNESCU1, Alexandru MOISUC1, Iosif GERGEN2 1 Banat’s University of Agricultural Sciences and Veterinary Medicine, Faculty of Agriculture, Timisoara, Calea Aradului nr. 119, RO-300645, Romania, monicaharmanescu@yahoo.com 2 Banat’s University of Agricultural Sciences and Veterinary Medicine, Faculty of Food Technology, Timisoara, Calea Aradului nr. 119, RO-300645, Romania. Abstract: In our days NIR spectroscopy represent a promising alternative to the chemical methods for crude fiber contents of forages. The main objective of this study was to obtain a NIR calibration model for prediction this parameter of forages harvested in June 2009 from hill permanent grassland (Grădinari, Caraş-Severin District). The experimental field was organized in ten experimental trials fertilized organic, mineral, and organo-mineral. The floristic composition of forages from this period was determined gravimetrically. From Poaceae were present Festuca rupicola and Calamagrostis epigejos. Fabaceae family was represented by Trifolium repens and Lathyrus pratensis. From other botanical family: Rosa canina, Filipendula vulgaris, Galium verum and Inula britanica. Like input data for NIR calibration were used the results for this qualitative parameter by chemical method and the reflectance values from 150 NIR spectra for all analysed samples. Partial last square (PLS) regression was used to obtain the “NIR - Total Fiber” model, implemented in Panorama program (version 3, LabCognition, 2009). The statistical parameters (R2=0.80; RMSEC=2.73) and the differences between references and predicted values situated in range 0.03 and 9.24% suggest a medium quality of calibration model, but it is promising to use it to predict the crude fiber contents of forages from grassland in this period of year using higher number of samples for calibration. Key words: forages quality, complex fertilizers, PLS- NIR model, grassland. Introduction Since Romania becomes part of European Union the national authorities must adjust the legislation to those European also regarding the quality and safety of the food. The 150/2004 low from our country transpose partially the CE 178/2002 Rule, which establish the most important recommendations on the qualities of row matter, from vegetal and animal origin, destined to obtain the food [8]. The security and quality of food with animal’s origin must be discussed in direct correlation with the requests for forages quality used like animal’s feed. The forages from permanent grassland represent the cheaper source of feed for ruminants. Fiber content of these forages, alongside of protein, lipids, minerals and vitamins, represents one of the most important parameter which characterize the forages quality [3]. Ruminants have the capacity to digest and use crude fiber like source of nutrients [2]. For example the cellulose content of forages must be in their ratio between 23- 25% [1], representing the key in intestinal transit, in stimulation of satiety sensation [2], like energetic source in animal’s metabolism [6]. The chemically determination of this parameter request a high reagents consumption, qualified human resources to perform the operation, a long time to obtain the results [7]. An important Food and Environment Safety - Journal of Faculty of Food Engineering, Ştefan cel Mare University – Suceava Year IX, No3 - 2010 106 alternative for these chemical models is represented in our days by NIR Spectrometry, a non-destructive method which allows obtaining very fast the final results (appreciatively 3 minutes), without reagents consumption, medium pollution and samples destroying [4, 5]. Experimental The forages were harvested in June 2009 from hill permanent grassland (Grădinari, Caraş-Severin District). The soil of permanent grassland was Calcic Luvisol and the annual average temperature around 10.4oC. The experimental field was organized in ten fertilized trials in randomized plots, in multiple stage blocks with five replications. It was used for mineral fertilization: 15:15:15 NPK complex, ammonium nitrate, potassium salt, superphosphat). Like organic fertilizer was chose fermented sheep manure. The fermented sheep manure was applied at each two years, even the mineral fertilizers yearly. The fertilization process was made during the period 2003-2008. The ten trials were: V1-unfertilized trial, V2-20 t/ha sheep manure, V3-40 t/ha sheep manure, V4-60t/ha sheep manure, V5-20 t/ha sheep manure + 50P2O5(Kg/ha), V6-20 t/ha sheep manure + 50P2O5 (Kg/ha) + 50 K2O (Kg/ha), V7- 20 t/ha sheep manure + 50 P2O5 (Kg/ha) + 50 K2O (Kg/ha) + 50N(Kg/ha), V8-100 N (Kg/ha) + 50 P2O5 (Kg/ha) + 50 K2O (Kg/ha), V9-150 N (Kg/ha) + 50P2O5(Kg/ha) + 50 K2O (Kg/ha), V10 - (100+100)N (Kg/ha) + 50 P2O5 (Kg/ha) + 50K2O(Kg/ha). The floristic composition of forages from the ten trials for this period of year was determined gravimetrically. From Poaceae familly dominant was Festuca rupicola (varied between 16.00 – 52.00%), followed by Calamagrostis epigejos (5.00-13.00%). Fabaceae family was represented mainly by Trifolium repens (dominant) and Lathyrus pratensis. From other botanical family were present Rosa canina (7.00- 18.00%), Filipendula vulgaris (3.00- 9.00%), Galium verum (3.00-7.00%) and Inula britanica (5.00%). NIR calibration model was obtained by PLS (Partial Last Square) regression, implemented in Panorama software (Variant 3, LabCognition, 2009). Like input data were selected the chemical data for crude fiber content, determined by JAOAC 962.09/1990 [7] method (samples are sequentially refluxed in dilute base followed by dilute acid), and the reflectance values from 150 NIR spectra. The V670 Spectrophotometer by Abble- Jasco was the instrument used to scan the spectra in the range 800-2500 nm, and than was selected with Panorama software three spectral ranges favourable to perform the calibration model for crude fiber determination. These spectral ranges were specific for the overtones of fundamental frequencies of OH bound, characteristic for fiber compounds (Table 1). Table 1. Calibration data for the „NIR-CF” model with 3 spectral ranges No. Selected spectral ranges Number of wavelengths 1 [1282.5 .. 1433.5] 303 2 [1542.0 .. 1949.0] 815 3 [2263.0 .. 2356.0] 187 „NIR-CF” - NIR-crude fiber model Food and Environment Safety - Journal of Faculty of Food Engineering, Ştefan cel Mare University – Suceava Year IX, No3 - 2010 107 For all the grounded dried samples the chemical results and NIR spectra were obtained in triplicate. Results and Discussion Statistical parameters for “NIR-CF” model with the three selected spectral ranges are presented in Table 2: Table 2. Statistical parameters for „NIR-CF” model with three selected spectral ranges R2 0.80 RMSEC 2.73 SD 4.75 These parameters suggest a medium quality of “NIR-CF” model, but were better than the case when were used the entire spectral domain (R2 = 0.7355, RMSEC = 3.28, SD = 4.39). The quality of NIR calibration model is underlined also in the graphical presentation of prediction for crude fiber by „NIR-CF” model with three spectral ranges (Figure 1) and also by the differences between the chemical results and those predicted for control samples. The control samples were harvested in the same period of year and grassland and conditioned in the same manner with those used to perform the „NIR-CF” calibration model (Table 3). Table 3. The results of crude fiber (%) prediction for the control samples forages (June2009) by „NIR- CF” calibration model with 3 spectral ranges Crude fiber (%) Control sample’s name Real (chemical method) Predicted (NIR model) Differences between Real - Predicted 101a 32.10 34.11 -2.01 101b 43.99 34.75 9.24 101c 18.85 24.24 -5.39 101d 31.06 26.92 4.14 102a 34.57 34.38 0.19 102b 24.89 27.08 -2.19 102c 25.46 23.27 2.19 102d 28.25 26.31 1.94 103a 36.45 36.35 0.10 103b 32.08 32.28 -0.20 103c 21.07 23.42 -2.35 103d 28.22 28.68 -0.46 104a 38.85 34.97 3.88 104b 32.98 30.26 2.72 104c 20.72 23.36 -2.64 104d 30.82 26.35 4.47 105a 35.17 34.70 0.47 105b 20.73 29.55 -8.82 105c 24.33 25.05 -0.72 105d 32.08 30.91 1.17 106a 33.01 33.91 -0.90 106b 22.60 27.64 -5.04 106c 24.74 23.06 1.68 106d 31.02 26.77 4.25 Food and Environment Safety - Journal of Faculty of Food Engineering, Ştefan cel Mare University – Suceava Year IX, No3 - 2010 108 107a 36.41 36.30 0.11 107b 29.78 31.26 -1.48 107c 20.03 24.18 -4.15 107d 31.62 29.87 1.75 108a 36.44 37.69 -1.25 108c 22.42 24.18 -1.76 108d 29.55 30.21 -0.66 109a 33.75 38.03 -4.28 109b 28.22 30.83 -2.61 109c 24.37 24.34 0.03 109d 36.07 35.46 0.61 Figure 1. Prediction of crude fiber by the „NIR-CF” model with 3 selected spectral ranges The differences between values obtained by chemical method and those predicted by „NIR-CF” model were situated between 0.03 and 9.24%. Almost 31.43% from these values were under 1.00%; 40.00% in range 1.01-3.00%; 17.14% between 3.01- 5.00%; and 11.43% in range 5.01-9.24%. These results indicate a medium quality of performed calibration model, but encourage us to continue these researches using a high number of samples to characterize better the concentrations of this qualitative parameter of forages from the permanent grassland Conclusion The PLS regression model “NIR-CF” with three selected spectral ranges for the determination of crude fiber of forages harvested in June 2009 had a medium quality. But this model promised to be used with success to determine routinely this parameter for the samples harvested in this period of year from the permanent grassland after the enrichment with a higher number of forages samples. Food and Environment Safety - Journal of Faculty of Food Engineering, Ştefan cel Mare University – Suceava Year IX, No3 - 2010 109 Acknowledgement The authors are grateful to CNCSIS - UEFISCSU (Romania) for financial support (PD project / 28 Sept.2010: On the applications of spectroscopic and chromatographic methods to establish the effects pf fertilisation on the quality of forages from grasslands). References 1. DRINCEANU, D., 1994, Alimentatia animalelor, Ed. Eurobit, p: 16-17. 2. GEORGESCU, GH., MĂRGINEAN, GH., PETCU, M., 2007, Cartea producătorului şi procesatorului de lapte, vol. 2, Editura Ceres, Bucureşti, ISBN: 978-973-40-0773-8. 3. LINSKENS H.F, JACKSON J.F., 1989, Plant fibers. Modern Methods of Plant Analysis, vol. 10, Springer-Verlag Berlin Heidelberg, Germania. 4. MCCLURE W.F., 1992, Making Light Work: Advances in Near-Infrared Spectroscopy, eds. Murray I., and Cowe I.A., VCH Publishers, New York, p.4-7. 5. SUGIYAMA J., MCCLURE W.F., HANA M., 1992, in Advances in Near-Infrared Spectroscopy, eds. Murray I., and Cowe I.A., VCH Publishers, New York, p.61-66. 6. VAN SOEST, P. J., 1994, Nutritional Ecology of the Ruminant, 2nd ed. Cornell University Press, Ithaca, NY. 7. *** JAOAC Official Methods of Analysis, 1990, 962.09 – Fiber (Crude) in Animal Feed. Ceramic Fiber Filter method, edited by Herlich Kenneth, 15 Edition, publshed y Association of Official Analytical Chemists, Arlington, Virginia, SUA. 8. www.immromania.ro