SEED OIL CONTENT ANALYSIS OF ETHIOPIAN MUSTARD (Brassica carinata A. Braun) BY NEAR INFRARED SPECTROSCOPY

Font1, R., Del Río1, M., Fernández1,J., Arthur2, E., Bancroft2,I., Chinoy2,C., Morgan2,C., De Haro1, A.1Instituto de Agricultura Sostenible (CSIC). Alameda del Obispo s/n. 14080 Córdoba. Spain.2John 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 oil 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 of the oil content for the selected samples were obtained by Nuclear Magnetic Resonance (NMR). 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 1.36 % dw, and a coefficient of determination (R2) of 0.93. This indicates that the 93% of the variability contained in the reference chemical data was explained by the calibration model (Table 1). According with the ratio SD/SECV, the second derivative of the raw optical data gave the equation with the highest prediction ability, showing a good fit of the data by the model. The 1-VR coefficient was high, meaning that the calibration equation explained the 92 % of the variability contained in the data, when it was cross-validated on the calibration set (Fig. 1).

Table 1. Calibration and cross-validation statistics for oil contentof B. carinata samples

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

100 25.60-54.60 43.12 5.38 1.36 0.93 3.61 0.92

The main wavelengths used for performing the factors of the oil equation were chemically assigned to 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 determine the oil content on intact seed samples of Ethiopian mustard with enough accuracy for screening and plant breeding purposes. The use of this non destructive technique represents an important reduction of the analysis time at a low cost and without using hazardous chemicals.Inversely to the protein content, mean seed oil content (% DW) of the B. carinata accessions was significantly greater (P<0.001) at Córdoba (44.7%) than at Norwich (33.2%), with similar range in absolute value (Córdoba: 36.60-50.90 %; Norwich: 27.00-40.30 %) (Fig. 2). Similar inverse relationship between protein and oil contents have been previously found in other oilseed crops (Robbelen, 1991). The data obtained in this work, together with those of protein, fibre, fatty acids and glucosinolate content (unpublished results) will enable us to evaluate the stability of the most important seed storage components of B. carinata in different environments (G x E effects).

Fig. 1. Cross-validation scatter plot for oil30 35 40 45 50

0

5

10

15

20

25

mean= 44.7 %SD= 2.48 %

Córdoba30 35 40 45 50

0

5

10

15

20

25

mean= 33.2 %SD= 2.30 %

Norwich

Fig. 2. Frequency distribution of oil

REFERENCESDE 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.RÖBBELEN, G. (1991). In Production and Utilization of Protein in Oilseed Crops. Ed. by E.S. Bunting. Martinus Nijhoff, The Hague, The Netherlands. 3-11.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.

AcknowledgementsThis work has been supported by the European project: “Brassica collections for Broadening Agricultural Use”, including “Characterising and utilising genetic variation in Brassica carinata for its exploitation as an oilseed crop” (Contract no: RESGEN CT99 109-112).