THE USE OF NEAR INFRA-RED SPECTROSCOPY FOR MONITORING QUALITY COMPONENTS IN PLANT BREEDING PROGRAMS. Font, R.

, Del Ro, M. and De Haro, A.(*), Institute of Sustainable Agriculture (IAS-CSIC), Av. Alameda del Obispo s/n, 14080 Crdoba, Spain. (*) Corresponding author. Email: cs9habaa@uco.es BACKGROUND In the last 30 years, Near Infrared Spectroscopy (NIRS) has been widely used as a fast and accurate method for qualitative and quantitative analysis of biological and non-biological materials in the agriculture, food, textile, petrochemical and pharmaceutical fields. International Standards Committees have formally accepted methods using NIRS for the analysis of many compounds in plants (Batten, 1998). The accuracy of NIR predictions is usually high for many components of agricultural and food interest, sometimes being of the same order than that of the reference method. The Department of Agronomy and Plant Breeding (DAPB) at the Institute of Sustainable Agriculture (IAS, CSIC), has been applying NIRS for many years to the agricultural field, and recently to environmental studies. With respect to the first topic, the research labour of DAPB has been focussed in breeding for multiple seed quality components of different plant species of agronomical interest, as legumes (De Haro et al. 1989) and oilseed Brassica (Font et al. 2003). Wet chemical analyses of seed storage components are expensive and time consuming, need trained staff, and implies the use of hazardous chemicals. NIRS technique has many advantages, as to perform the simultaneous analysis of several quality traits at a low cost, in a short time and without using chemicals in the process, thus NIRS can be considered as an important tool to speed up analyses of breeding materials. MATERIALS AND METHODS Plant Material Plant material was formed by the oilseed species Brassica juncea, B. carinata and B. napus, and the legumes Vicia faba and Cicer arietinum (Fig. 1). These species were grown over different years in Crdoba (Spain), harvesting individual plants and storing their seeds for NIRS analysis. Quality parameters of most interest determined by NIRS in the seed were: fatty acid composition, total oil, crude protein (CP), acid detergent fiber (ADF), neutral detergent fiber (NDF), bulk density (BD), 1000 seed weight, and diverse anti-nutritive factors as sinigrin (SIN), gluconapin (GNA) and tannins. Reference Methods of Analysis Reference values for NIRS calibrations and validations were performed according with official methods of analysis (AOAC 1990; ISO 1992) NIRS analysis The NIRS analysis consisted in the following steps: 1. All the samples belonging to each plant species were scanned in an NIRS spectrometer model 6500 (Foss-NIRSystems, Inc., Silver Spring, MD, USA) in reflectance mode, equipped with a transport module, and their NIR spectra recorder as log (1/Reflectance)in the range from 400 to 2500 nm, as independent files (Fig. 2). Seed samples were placed for the analysis in a small ring cup (35 mm diameter) (Part number IH- 0350, Foss-NIRSystems, USA). (Fig. 3).

2. A group of samples representative of the whole set in each file were selected by their spectral features (Mahalanobis H distance), and then were split in a calibration and validation sets. The selected samples were analysed by the reference methods for the components of interest. 3. Calibration equations for the different components were developed on the calibration sets by using Modified Partial Least Square (MPLS) regression (GLOBAL v. 1.50 program, WINISI II, Infrasoft International, LLC, Port Matilda, PA, USA), with different mathematical pretreatments (derivatives, scatter correction) of the original spectra. The equations obtained in the calibration process were then validated on the validation sets, to test the prediction ability of each one over samples of the same characteristics. Those equations presenting the highest prediction ability, i.e., high coefficient of determination in the external validation (r2) and low standard error of prediction (SEP), were selected and used to predict unknown samples. 4. Plant material with interesting chemical or physical features identified by applying NIR equations, was collected and analysed at successive years, and added to the previous spectral libraries to increase both the existing variability and the robustness of each equation. RESULTS AND DISCUSSION Plant breeding programs usually involve the analysis of seed samples of thousand of individuals to identify the target genotypes. This labour is often expensive and timeconsuming and in addition, specialized personal with the proper knowledge of the standard methods used is required. The NIRS technique is one of the fastest physical methods of analysis, together with Nuclear Magnetic Resonance (NMR). Each sample is scanned in approximately two minutes and simultaneous predictions can be made for many traits. The NIRS technique has a central role in the labour at DAPB for the last fifteen years. Thousand of seed samples are analysed by NIRS every year and multiple quality components predicted, a labour that would not be possible to be carry out by using standard wet methods of analysis. This work has yielded the development of a database of calibration equations for many quality components of the seed of oilseed Brassicas and legumes. Table 1 shows the r2 and SEP of the different NIR equations developed at our Department for monitoring the quality components in the plant species above mentioned. All the NIR mathematical models resulted in SEPs and r 2 that were indicative of excellent (r2>0.90) or good precision equations (0.70<r2<0.89) (Shenk and Westerhaus 1996), with enough accuracy for predicting the seed components for screening purposes (Fig. 4).

Table 1. Validation statistics for seed quality components in Brassica carinata (Bc), B. juncea (Bj), Vicia faba (Vf) and Cicer arietinum (Ca) for the selected NIR equations. Seed quality trait (units) oil (% dw) erucic acid (% oil) oleic acid (% oil) linoleic acid (% oil) linolenic acid (% oil) ADF (% dw) NDF (% dw) SIN (mol g-1 dw) GNA (mol g-1 dw) Tannins (% dw) CP (% dw) BD (kg m-3 ww) 1000 seed weight (g ww) 0.98 0.9 0.7 0.9 0.90 0.9 0.9 0.9 0.68 30.5 5 0.41 0.5 Bc 0.92 0.91 0.80 0.73 0.80 0.92a/0.83* r2 Bj 0.9 0.9 0.9 0.9 0.7 0.9 0.9 7.83 15.2 6 12.7 5 0.50 0.9 0.9 Ca 0.9 Vf Bc 1.49 3.56 4.50 2.48 1.30 0.95a/0.82* SEP Bj 0.91 2.51 2.17 1.90 0.85 0.6 1.2 1.3 1.1 Ca 0.2 Vf

*from multispecies calibration with B. carinata, B. juncea and B. napus. CONCLUSION NIRS is a valuable tool in routine analysis in the different plant breeding programs carried out at our Department. NIRS can be considered as an excellent alternative to standard techniques for quick measurement of seed quality components when a large number of analyses have to be performed, as is needed in germplasm screenings and plant breeding programs. In addition, seed samples can be analysed undestructively, which is of crucial importance in the case of scarce or valuable seeds. FUTURE In recent years DAPB, in close collaboration with other national and international institutions, has been applying NIRS to the environmental field, in studies concerning the suitability of NIRS to the analysis of trace elements (As, Pb, Cu, Zn) in different biological

(plants, seawater and continental water animal species) and non-biological (soil) matrices (Font et al. 2002). Preliminary results point out that NIRS is able to discriminate correctly samples with low, medium and high contents of these trace elements in the matrices studied, which encourages for additional research labour in this field, widening the range of NIRS applications. REFERENCES AOAC. Official Methods of Analysis of the Association of Official Analytical Chemists. Edited by K. Helrich. 15th Edition. Published by AOAC. Arlington Virginia 22201 USA. 1990 Batten, G.D. Plant analysis using near infrared reflectance spectroscopy: the potential and the limitations. Australian Journal of Experimental Agriculture. 1998, 38, 697-706. De Haro, A., Lpez-Medina, J., Cabrera, A., Martn, A. Determination of tannin in the seeds of Vicia faba by NIR. In Recent Advances of research in antinutritional factors in legume seeds. Wageningen, The Netherlands. 1988, 172-175. Font, R., del Ro, M., de Haro, A. Use of near infrared spectroscopy to evaluate heavy metal content in Brassica juncea cultivated on the polluted soils of the Guadiamar river area. Fresenius Environmental Bulletin. 11(10a), 2002, 777-781. Font, R., del Ro, M., Fernndez, J.M., de Haro, A. Acid detergent fiber analysis in oilseed Brassicas by near-infrared spectroscopy. Journal of Agricultural and Food Chemistry, 2003. In Press. ISO norm. Rapeseed Determination of glucosinolate content Part 1: Method using high performance liquid chromatography. ISO 9167-1, 1990, 1-9. Shenk, J.S.;Westerhaus, M.O. Calibration the ISI way. In Near infrared spectroscopy: The future waves. Davies, A.M.C., Williams, P.C., Eds.; NIR publications: Chichester, UK, 1996, pp 198-202.

LEGENDS FOR FIGURES 1 TO 3. Fig. 1. Inflorescence of Brassica juncea Fig. 2. Intact seeds of Brassica species ready to be analysed by NIRS Fig. 3. NIR Spectra of different Brassica seed samples

PROTEIN (NIRS) (% DW)

17 16 15 14 13 12 11 10 9 8 7 6

ADF (NIRS) (% DW)

B. juncea B. carinata B. napus

30 28 26 24 22 20 18 16 16 18 20 22 C. arietinum

9 10 11 12 13 14 15 16 17

ADF (laboratory) (% DW)

PROTEIN (laboratory) (% DW)

24

26

28

30

Fig. 4. Scatter validation plots for ADF in Brassica sp. (A), and for crude protein in C. arietinum (B).