Centro de Investigaciones Agrarias de Mabegondo, CIAM, Apartado 10, 15080 A Coruña, Spain
SUMMARY
Near infrared reflectance spectroscopy (NIRS) has been used in CIAM since 1990 to analyse forages including grasses, mixed swards, forage maize, vetch-oats mixtures and silages from maize and mixed swards. This present work summarizes updated NIR calibrations to determine main nutritive parameters in green herbages (dry material) and in herbage silages, both in fresh and dry material. Crude protein was the most precise determination in green forages, with determination coefficients R2= 0.99 for calibration and validation and standard errors of calibration, SEC= 0.42 and validation, SEP= 0.52. Analysis of fresh herbages and silages was less precise than that of dry samples, mainly for validation results.
Keywords: calibration, digestibility, fresh silages, protein.
INTRODUCTION
Near infrared reflectance spectroscopy (NIRS) has shown to be a very fast, precise and non-destructive technique for routine analysis of various agricultural materials including forages (Norris et al., 1976; Sinnaeve et al., 1994). It has been used by CIAM since 1990 to analyse forages both in animal feeding (Castro, 1994; Castro et al., 1997; Castro, 1998) and in plant breeding trials (Castro and Moreno-González, 1993; Castro and Oliveira, 1996). The aim of this present work is to summarize updated calibrations for the main components determined in routine analysis of mixed swards and both dried or fresh herbage silages.
MATERIALS AND METHODS
NIRS calibrations were developed by MPLS regression (Martens and Naes, 1987) of chemical data on spectral data (2nd derivative of SNV and D-trended spectra) in a set of samples (calibration set). Calibration samples of green herbages were selected by adding cocksfoot (Dactylis glomerata L) and mixed swards (Lolium perenne L + Trifolium repens L) samples to perennial ryegrass (Lolium perenne L) calibration samples (Castro and Oliveira, 1996). Selection was performed by using the SELECT algorithm of NewISI software (InfraSoft International, 1995). In 1996, an approach was made to calibrate NIRS for the analysis of fresh herbage silages. Calibration samples came from commercial farms and were evaluated in vivo by CIAM (Castro et al., 1997). Spectra of frozen subsamples was recorded. A subsample was taken to determine pH and volatile components. The remaining material was dried in an Unitherm oven at 80°C, to determine dry matter (DM), and ground to pass 1 mm in a Christy and Norris mill. The spectra of dried material was recorded. All calibrations are periodically updated by adding selected routine samples (SELECT) to the calibration set. Only a few of these samples can be processed in vivo and they are chosen taking into account their in vitro digestibitily. Another set of samples of each type of forage was selected at random from routine samples to validate calibrations (validation set).
Calibration and validation samples were analysed by wet chemistry to determine organic matter (OM), crude protein (CP), acid (ADF) and neutral (NDF) detergent fibre, and in vitro organic matter digestibility (IVOMD). In vivo organic matter digestibility (OMD) values result from measures carried out in the CIAM feed unit on six sheep fed ad libitum for a period of six days. Volatile compounds were also determined in herbage silages and water soluble carbohydrates (WSC) in green forages. All determinations were carried out on dried material by conventional methods and results were reported on a dry matter basis
RESULTS AND DISCUSSION
Calibration equations for green herbages, dried and fresh silages are summarized in Tables 1 and 2. All equations were obtained by MPLS (Modified Partial Least Squares) regression of chemical and in vivo data on the second derivative of reflectance data. Scatter correction SNV (Standard Normal Variate) and De-trending (Barnes et al., 1989) of spectral data was made in all cases. The spectral range was 400 to 2 000 nm for green forages and 1 100 to 2 500 nm for herbage silages, but the water region (1 870-1 950 nm) was eliminated when analysing fresh samples of silages. Two passes of outliers elimination were performed in fresh analysis and down-weighting outliers was the best treatment in dry analysis.
Crude protein was the most precise of green herbage determinations. Precision of fresh herbage analysis was similar to that of dry material for calibration, but it was less precise for validation except for dry matter determination.
Table 1. NIRS equations for green herbage forages analysis.
|
|
Calibration |
Validation |
|||||||||
|
Component |
N |
Mean |
Sd. Dv. |
R2 |
SEC |
SECV |
N |
Mean |
Sd. Dv. |
R2 |
SEP |
|
OM |
169 |
90.99 |
2.42 |
0.94 |
0.61 |
0.91 |
160 |
90.93 |
1.92 |
0.87 |
0.79 |
|
CP |
169 |
10.75 |
4.32 |
0.99 |
0.42 |
0.66 |
160 |
11.50 |
4.40 |
0.99 |
0.52 |
|
ADF |
169 |
30.55 |
4.84 |
0.98 |
0.74 |
1.15 |
160 |
29.17 |
4.05 |
0.93 |
1.12 |
|
NDF |
62 |
54.64 |
6.26 |
0.99 |
0.78 |
2.64 |
44 |
53.84 |
6.74 |
0.90 |
2.17 |
|
WSC |
157 |
18.72 |
8.68 |
0.91 |
2.64 |
2.83 |
143 |
17.39 |
6.07 |
0.91 |
2.07 |
|
IVOMD |
63 |
73.43 |
3.23 |
0.64 |
1.93 |
2.05 |
|
|
|
|
|
N= number of samples.
Sd.Dv.= standard deviation.
SECV= standard error of cross-validation.
R2= determination coefficients.
SEC and SEP= standard errors of calibration and prediction.
Table 2. NIRS equations for herbage silages analysis.
| |
Calibration |
Validation |
|||||||||
|
Component |
N |
Mean |
Sd. Dv. |
R2 |
SEC |
SECV |
N |
Mean |
Sd. Dv. |
R2 |
SEP |
|
Dry Samples |
|||||||||||
|
OM |
369 |
87.08 |
4.74 |
0.91 |
1.46 |
1.65 |
96 |
89.96 |
2.36 |
0.90 |
0.87 |
|
CP |
365 |
12.55 |
2.90 |
0.83 |
1.18 |
1.29 |
91 |
11.52 |
2.22 |
0.89 |
0.75 |
|
ADF |
369 |
38.40 |
5.61 |
0.88 |
1.96 |
2.07 |
94 |
38.33 |
5.22 |
0.94 |
1.40 |
|
NDF |
369 |
54.87 |
7.20 |
0.88 |
2.46 |
2.75 |
92 |
58.59 |
6.56 |
0.92 |
1.86 |
|
IVOMD |
282 |
63.66 |
8.49 |
0.81 |
3.75 |
3.97 |
85 |
63.10 |
7.29 |
0.90 |
2.25 |
|
OMD |
137 |
65.73 |
6.76 |
0.81 |
2.97 |
3.13 |
|
|
|
|
|
|
pH |
112 |
4.34 |
0.41 |
0.77 |
0.20 |
0.23 |
74 |
4.55 |
0.38 |
0.51 |
0.29 |
|
Ammonia-N |
112 |
0.26 |
0.16 |
0.75 |
0.08 |
0.10 |
80 |
0.26 |
0.16 |
0.31 |
0.14 |
|
Lactic acid |
100 |
5.74 |
2.91 |
0.87 |
1.06 |
1.38 |
79 |
1.63 |
0.88 |
0.56 |
5.19 |
|
Acetic acid |
100 |
2.26 |
1.27 |
0.49 |
0.91 |
1.02 |
80 |
0.51 |
0.26 |
0.14 |
1.76 |
|
Butyric acid |
100 |
1.23 |
1.06 |
0.74 |
0.54 |
0.67 |
80 |
0.52 |
0.35 |
0.17 |
1.51 |
|
Fresh Samples |
|||||||||||
|
OM |
294 |
87.69 |
3.87 |
0.75 |
1.93 |
2.15 |
94 |
89.98 |
2.35 |
0.48 |
1.96 |
|
CP |
300 |
12.40 |
2.74 |
0.82 |
1.17 |
1.36 |
96 |
11.45 |
2.26 |
0.58 |
2.04 |
|
ADF |
304 |
37.91 |
5.81 |
0.86 |
2.15 |
2.40 |
94 |
37.87 |
5.23 |
0.80 |
2.36 |
|
NDF |
297 |
54.34 |
6.75 |
0.81 |
2.91 |
3.33 |
92 |
58.84 |
6.57 |
0.79 |
3.55 |
|
DM |
280 |
27.38 |
10.06 |
0.99 |
1.20 |
1.33 |
95 |
31.43 |
11.47 |
0.98 |
1.68 |
|
IVOMD |
266 |
64.41 |
8.53 |
0.84 |
3.42 |
3.94 |
86 |
63.17 |
7.63 |
0.77 |
3.77 |
|
OMD |
83 |
64.88 |
6.31 |
0.80 |
2.83 |
3.85 |
|
|
|
|
|
|
pH |
91 |
4.33 |
0.37 |
0.80 |
0.17 |
0.21 |
74 |
4.55 |
0.38 |
0.51 |
0.30 |
|
Ammonia-N |
93 |
0.24 |
0.13 |
0.53 |
0.09 |
0.11 |
80 |
0.26 |
0.16 |
0.16 |
0.15 |
|
Lactic acid |
86 |
5.77 |
2.87 |
0.78 |
1.29 |
1.76 |
79 |
1.63 |
0.88 |
0.35 |
3.84 |
|
Acetic acid |
79 |
2.25 |
0.92 |
0.33 |
0.75 |
0.78 |
80 |
0.51 |
0.26 |
0.11 |
1.28 |
|
Butyric acid |
83 |
1.23 |
1.02 |
0.92 |
0.30 |
0.47 |
80 |
0.52 |
0.35 |
0.35 |
1.60 |
N= number of samples.
Sd.Dv.= standard deviation.
SECV= standard error of cross-validation.
R2= determination coefficients.
SEC and SEP= standard errors of calibration and prediction.
REFERENCES
Barnes, R.J., Dhanoa, M.S. & Lister, S.J. 1989. Standard normal variate transformation and De-trending of near infrared diffuse reflectance spectra, Applied Spectroscopy, 43: 772-777.
Castro, P. 1994. Espectroscopía de reflectancia en el infrarrojo próximo (NIRS) y evaluación nutritiva de pastos, Tesis Doctoral en microfichas Nº 408, Servicio de Publicaciones, Universidad de Santiago, Spain.
Castro, P. 1998. Análisis de ensilados mediante NIRS: ¿En fresco o en seco?, XXXVIII Reunión Científica SEEP, 1-5 June 1998, Soria, Spain.
Castro, P., Castro, J., González-Arráez, A. & Flores, G. 1997. Analysis of mixed sward silages by near infrared reflectance spectroscopy (NIRS), Proceedings of XVIIIth. International Grassland Congress, 8-19 June 1997, Saskatoon, Saskatchewan, Canada.
Castro, P. & Moreno-González, J. 1993. Análisis de maíz forrajero mediante espectroscopía de reflectancia en el infrarrojo próximo, XXXII Reunión Científica de la SEEP, 1992, Pamplona, Spain.
Castro, P. & Oliveira, J.A. 1996. Breeding for quality: Analysis of perennial ryegrass by NIRS, Proceedings of XVIth EGF Meeting, 1996, Grado, Italy.
InfraSoft International, 1995. NIRS 3, Routine operation, calibration development, and network system management software for near infrared instruments, Version 3.11, Perstorp Analytical inc., Silver Spring, MD, USA.
Martens, H. & Naes, T. 1987. Multivariate calibration by data compression. In: Near-infrared technology in the agricultural and food industries. P. Williams and K. Norris Ed., American Association of Cereal Chemists, Inc., St. Paul, Minnesota, USA.
Norris, K.H., Barnes, R.F., Moore, J.E. & Shenk, J.S. 1976. Predicting forage quality by near infrared reflectance spectroscopy, Journal of Animal Science, 43: 889-897.
Sinnaeve, G., Dardenne, P., Agneesens, R. & Biston, R. 1994. The use of near infrared spectroscopy for the analysis of fresh grass silage, Journal of near infrared spectroscopy, 2: 79-84.
