Calibration strategies for prediction of amino acid content of poultry feeds
The characterization of poultry feeds requires the knowledge of major chemical compounds (Crude protein, Fat, Starch, etc.), but amino acids (AA) have also to be measured since they are one of the most important quality factors. In particular essential AA such as Lysine (LYS), Methionine (MET), Cysteine (CYS) or Tryptophane (TRY) must reach sufficient levels to allow optimal performance of animals. The chemical determination of AA by classical methods (HPLC) is time consuming and expensive. It was therefore decided to investigate the potential of NIRS for AA determination. A characteristic of AA levels is that they are strongly linked to crude protein level, but that this relationship is sometimes broken due to the use of different raw materials or purified AA in diets. Therefore most attempts to build calibrations for AA failed to provide robust models. In the present experiment, 95 poultry feed (SET1) from East Africa were analysed for LYS, MET, CYS and TRY levels by reference method. An additional set of 65 samples (SET2) was obtained by adding various levels of pure AA the original feed. The 160 samples were ground (0.5mm sieve) and diffuse reflectance spectra were measured on a FOSS NIRSystem 6500 spectrometer in small ring cups. Calibration equations were built after mathematical pre-processing of spectral data (SNV and detrend with 2nd derivative of spectra) without visible wavelengths. Partial Least Squares regression (modified PLS in WINISI software) was found to be the most efficient method for calibrations. Calibration equation on the whole dataset led to R2 values of 0.99; 0.96; 0.96; 0.95 and SECV values of 0.73; 0.62; 0.45; 0.26 g/kg for LYS, MET, CYS, TRY respectively. These performances are satisfactory given the repeatability of reference AA values. Equations calculated on SET1 (non supplemented samples) alone were more precise with SECV values of 0.72; 0.32; 0.25; 0.16 g/kg respectively, but they were unable to predict SET2 (supplemented samples). It is concluded that NIRS can be used for assessment of AA in poultry feed with an accuracy which allows quality control. A very interesting point is that the strategy of including pure amino acids lead to calibration models which are not based on AA-Protein relationship, and which are therefore resistant to qualitative gaps between samples, including pure AA supplementation which is a common practice in poultry feeds.
Main Authors: | , , , , |
---|---|
Format: | conference_item biblioteca |
Language: | eng |
Published: |
IM Publications
|
Subjects: | Q54 - Composition des aliments pour animaux, alimentation des animaux, volaille, acide aminé, http://aims.fao.org/aos/agrovoc/c_429, http://aims.fao.org/aos/agrovoc/c_6145, http://aims.fao.org/aos/agrovoc/c_342, |
Online Access: | http://agritrop.cirad.fr/531001/ http://agritrop.cirad.fr/531001/1/document_531001.pdf |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Summary: | The characterization of poultry feeds requires the knowledge of major chemical compounds (Crude protein, Fat, Starch, etc.), but amino acids (AA) have also to be measured since they are one of the most important quality factors. In particular essential AA such as Lysine (LYS), Methionine (MET), Cysteine (CYS) or Tryptophane (TRY) must reach sufficient levels to allow optimal performance of animals. The chemical determination of AA by classical methods (HPLC) is time consuming and expensive. It was therefore decided to investigate the potential of NIRS for AA determination. A characteristic of AA levels is that they are strongly linked to crude protein level, but that this relationship is sometimes broken due to the use of different raw materials or purified AA in diets. Therefore most attempts to build calibrations for AA failed to provide robust models. In the present experiment, 95 poultry feed (SET1) from East Africa were analysed for LYS, MET, CYS and TRY levels by reference method. An additional set of 65 samples (SET2) was obtained by adding various levels of pure AA the original feed. The 160 samples were ground (0.5mm sieve) and diffuse reflectance spectra were measured on a FOSS NIRSystem 6500 spectrometer in small ring cups. Calibration equations were built after mathematical pre-processing of spectral data (SNV and detrend with 2nd derivative of spectra) without visible wavelengths. Partial Least Squares regression (modified PLS in WINISI software) was found to be the most efficient method for calibrations. Calibration equation on the whole dataset led to R2 values of 0.99; 0.96; 0.96; 0.95 and SECV values of 0.73; 0.62; 0.45; 0.26 g/kg for LYS, MET, CYS, TRY respectively. These performances are satisfactory given the repeatability of reference AA values. Equations calculated on SET1 (non supplemented samples) alone were more precise with SECV values of 0.72; 0.32; 0.25; 0.16 g/kg respectively, but they were unable to predict SET2 (supplemented samples). It is concluded that NIRS can be used for assessment of AA in poultry feed with an accuracy which allows quality control. A very interesting point is that the strategy of including pure amino acids lead to calibration models which are not based on AA-Protein relationship, and which are therefore resistant to qualitative gaps between samples, including pure AA supplementation which is a common practice in poultry feeds. |
---|