USING NEAR INFRARED SPECTROSCOPY FOR DETERMINING PROTEIN CONTENT IN ETHIOPIAN MUSTARD (Brassica carinata A. Braun)

Font1, R., Del Río1, M., Sillero2,A., Arthur3, E., Bancroft3, I., Chinoy3, C., Morgan3, C., De Haro1, A.1Instituto de Agricultura Sostenible (CSIC). Alameda del Obispo s/n. 14080 Córdoba. Spain.2Laboratorio Agroalimentario. J.A. Alameda del Obispo s/n. 14080 Córdoba. Spain.3John Innes Centre. Norwich Research Park, Colney, Norwich, NR4 7UH, UK. INTRODUCTIONEthiopian mustard is an oilseed species with high potential as a crop for Mediterranean (semi-arid) conditions and as a genetic source for characters of agronomic importance. In the last 30 years, Near Infra-Red Spectroscopy (NIRS) has been widely used as a rapid and accurate method for qualitative and quantitative analysis in many fields (Williams et al., 1987).The Department of Agronomy and Plant Breeding of the Institute for Sustainable Agriculture (IAS, CSIC), has been using NIRS for the last fifteen years, to determine of seed quality components in different plant species (De Haro et al., 1989.;Velasco et al., 1992; Font et al., 1998, 2000). The most attractive features of analysis using NIRS are its speed, minimal sample preparation and its non-destructive nature thus making it possible to analyse large number of samples in a short time. The objectives of this work are to test the potential of NIRS to determine the protein content of intact seeds of Brassica carinata, and to apply this technique to evaluate the protein content of a germplasm collection of this species grown in two different environments, Norwich (UK) and Córdoba (Spain).

MATERIAL AND METHODTen plants from 100 accessions of B. carinata, chosen at random from the collections at IAS (Córdoba, Spain) and Centre for Genetic Resources (Wageningen, The Netherlands) were transplanted in March 2000 to field plots at Norwich, UK, and harvested in September 2000. The same 100 accessions were grown in field plots at Córdoba, Spain, from November 2000 to June 2001. To perform NIRS calibration for protein content, 3g samples of intact seed from 2000 plants (100 accessions x 10 plants/accession x 2 localities), were scanned in a NIR spectrophotometer (NIRSystems model 6500, Foss-NIRSystems, Inc., Silver Spring, MD, USA) in the reflectance mode, acquiring their spectra at 2 nm. intervals over a wavelength range from 400 to 2500 nm (VIS + NIR regions). On the basis of their spectral features, a sub-set of 100 samples representative of the whole spectral variability contained in the entire set, were selected for performing NIR calibrations. Reference analytical values for the selected samples were obtained by Kjeldahl (AOAC method 920.87, 1990). Using the program GLOBAL v. 1.50 (WINISI II, Infrasoft International, LLC, Port Matilda, PA, USA), different mathematical treatments (0,0,1,1 (derivative, gap, first smooth, second smooth); 1,4,4,1; 2,5,5,2) were used to correlate spectral and chemical data. Cross-validation was performed on the calibration set to test the ability of the equations obtained to predict the protein content. The equation with the higher ratio of standard deviation (SD) to standard error of cross-validation (SECV) and the higher coefficient of determination (1-VR) was selected as the best equation (Williams et al., 1993). This equation was then applied to the previously acquired spectra to predict the protein content of the different accessions of B. carinata from both localities.

RESULTS AND DISCUSSIONThe selected calibration equation resulted in a standard error of calibration (SEC) of 0.50 % dw, and a coefficient of determination (R2) of 0.99, meaning that the 99% of the variability contained in the data was explained by the calibration model (Table 1). On the basis of the ratio SD/SECV, the second derivative of the raw optical data gave the equation with the higher prediction ability and a very good fit of the data by the model. On the other hand, the calibration equation explained the 98 % of the data variability when it was cross-validated on the calibration set (Fig. 1).

Table 1. Calibration and cross-validation statistics for protein content of B. carinata samples

Calibration Cross-validationn range mean SD SEC R2 SD/SECV 1-VR

100 16.80-37.80 24.23 4.66 0.50 0.99 6.85 0.98

The wavelengths highly participating in the development of the protein equation were chemically assigned to stretching of S-H groups (1740 nm), protein (2052 nm) and C-H stretching and bending of the CH2 groups of oil (1724, 1764, 2308 and 2348 nm).These results show that it is possible to use NIRS to accurately analyse the protein content on intact seed samples of Ethiopian mustard. The use of this non destructive technique represents an important reduction of the analysis time at a low cost and without using hazardous chemicals.Mean seed protein content (% DW) of the B. carinata accessions was significantly greater (P<0,001) at Norwich (25,9%) than at Córdoba (22,7%). However, the range of protein content was greater at Córdoba (18,2-29,9%) than at Norwich (22,4-31,7%) (Fig. 2). These differences are likely to be explained by the different environments, where Córdoba were sunnier, warmer and drier that Norwich. The data obtained in this work will enable us to evaluate the stability of this important seed quality component in different environments (G x E effects).

Fig. 1. Cross-validation scatter plot for protein18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

0

10

20

30

mean= 22.7 %SD= 2.05 %

nº o

f acc

essi

ons

protein (% DW)

Córdoba

18 19 20 21 22 23 24 25 26 27 28 29 30 31 320

10

20

30

mean= 25.9 %SD= 1.70 %

Norwich

Fig. 2. Frequency distribution of protein

REFERENCESASSOCIATION OF OFFICIAL ANALYTICAL CHEMISTS. (1990). In: Official Methods of Analysis. Ed. by Kenneth Helrich.. Fifteenth edition. Arlington, Virginia, USA. Vol. 2. 781-782.DE HARO, A., LÓPEZ-MEDINA, J., CABRERA, A. & MARTÍN, A. (1989). In: Recent Advances of Research in Antinutritional Factors in Legume Seeds. Ed. by J. Huisman, T.F.B. van der Poel and I.E. Liener. Pudoc. Wageningen. 297-300.FONT, R., DEL RIO, M., FERNÁNDEZ-MARTÍNEZ, J.M. & DE HARO, A. (1998). Cruciferae Newsletter 20, 67-68.FONT, R., GUTA, B., DEL RIO, M., DOMÍNGUEZ, J., FERNÁNDEZ-MARTÍNEZ, J.M. & DE HARO, A. (2000). Cruciferae Newsletter 22, 7-8.VELASCO, L., MARTÍN, L.M. & DE HARO, A. (1992). In: Near Infrared Spectroscopy. Ed. by K.I. Hidrum, T. Isaksson, T. Naes and A. Tandberg. Ellis Horwood, England. 287-292.WILLIAMS, P.C. & NORRIS, K.H. (1987). American Association of Cereal Chemists Inc.: St. Paul, MN. 1987. 330 pp.WILLIAMS, P.C. & SOBERING, D.C. (1993). J. NIRS. 1, 25-32.

AcknowledgementsAuthors thank Dr. José Manuel Gálvez from Laboratorio Agroalimentario, Junta de Andalucía, for performing the chemical analyses of the reference samples. This work has been supported by the European Union (Contract no: RESGEN CT99 109-112.