Published in Crop Sci. 44:93-97 (2004).
© 2004 Crop Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
CROP BREEDING, GENETICS & CYTOLOGY
Use of Near-Infrared Reflectance Spectroscopy for Selecting for High Stearic Acid Concentration in Single Husked Achenes of Sunflower
Leonardo Velasco*,
Begoña Pérez-Vich and
José M. Fernández-Martínez
Instituto de Agricultura Sostenible, Apartado 4084, E-14080 Córdoba, Spain
* Corresponding author (ia2veval{at}uco.es).
 |
ABSTRACT
|
|---|
Half-seed analysis by gas-liquid chromatography (GLC) allows a nondestructive evaluation of the fatty acid composition of oilseeds, since in most cases the analyzed portion is representative of the whole seed. However, the sunflower (Helianthus annuus L.) mutant CAS-14 exhibits a high stearic acid concentration nonuniformly expressed along the longitudinal axis of the seed. No analytical technique is available for nondestructive selection at the single seed level in this mutant. The objective of the present research was to study the potential of near-infrared reflectance spectroscopy (NIRS) for analyzing the fatty acid composition of single husked achenes of sunflower and to evaluate its performance in a selection program for high stearic concentration in materials derived from CAS-14. A calibration set containing 2510 single husked achenes from a broad spectrum of breeding materials was developed. Reliable equations were developed for stearic acid, with r2 = 0.80 and ratio of the standard error of cross validation (SECV) to the standard deviation (SD) of 0.45, oleic acid (r2 = 0.89, SECV/SD = 0.34), and linoleic acid (r2 = 0.91, SECV/SD = 0.30). The calibration equations were applied to the analysis of 8109 husked achenes within a selection program for high stearic acid concentration, 503 of them being further analyzed by GLC to monitor the performance of NIRS. The results revealed a close relationship between NIRS and GLC data, with r2 of 0.83 for stearic acid, 0.92 for oleic acid, and 0.93 for linoleic acid concentration. It was concluded that NIRS can be used reliably for nondestructive selection for these major fatty acids at a single-seed level.
Abbreviations: GLC, gas-liquid chromatography • NIRS, near-infrared reflectance spectroscopy • SECV, standard error of cross validation
|
INTRODUCTION
|
|---|
IN GENERAL, the fatty acid composition of seed oils is governed by the genotype of the developing embryo (Downey, 1987). Accordingly, selection for fatty acid composition can be conducted at the single-seed level, provided that single-seed fatty acid composition can be accurately measured in a nondestructive way. Downey and Harvey (1963) developed the half-seed technique for nondestructive analysis of the fatty acid composition of single seeds of rapeseed (Brassica napus L.). The technique consists of the excision and further analysis by gas chromatography of one of the cotyledons, which does not affect the viability of the embryo. Variations of this technique have been developed for most oilseeds, including sunflower (Conte et al., 1989).
In sunflower, half-seed analysis requires the removal and analysis of a small seed portion distal to the embryo. Successful utilization of the half seed technique in this crop necessarily requires that the fatty acid composition of the analyzed seed (husked achene) portion be representative of the fatty acid composition of the whole seed. This is generally true in sunflower germplasm. However, a recently developed high stearic sunflower mutant CAS-14 (Fernández-Moya et al., 2002) exhibits strong variation for fatty acid composition along the longitudinal axis of the seed (Fernández-Moya et al., 2003). Accordingly, the half-seed technique cannot be used in this mutant as an estimation of the fatty acid composition of the whole-seed oil.
To conduct a single-seed selection for high stearic acid content in materials derived from the mutant CAS-14, we have focused on the potential of NIRS for nondestructive analysis of stearic acid content in husked achenes of sunflower. Sato et al. (1995) demonstrated the feasibility of NIRS for measuring the concentration of linoleic acid in the oil of single husked achenes of sunflower. Velasco et al. (1999) reported that NIRS permitted the discrimination of single unhusked achenes of sunflower for oleic and linoleic acid concentration in the seed oil. However, the potential of NIRS for estimating the stearic acid concentration in single unhusked or husked achenes of sunflower has not been evaluated. Pérez-Vich et al. (1998) conducted a comparative study using sunflower oil, meal, husked achenes, and unhusked achenes (bulk samples of about 10–15 achenes) to assess the accuracy of NIRS for analyzing all major fatty acids present in sunflower oil (palmitic, stearic, oleic, linoleic). The study demonstrated that NIRS is efficient in discriminating stearic acid levels in all four types of samples analyzed. The greatest NIRS accuracy for stearic acid estimation was obtained for oil samples, followed by meal, husked achenes, and unhusked achenes. Pérez-Vich et al. (1998) found that NIRS was not reliable enough for analyzing stearic acid concentration in unhusked sunflower achenes, which was attributed to the low resolution of unhusked achene spectra in comparison with husked achene spectra.
The objective of the present research was to study the potential of NIRS for analyzing the fatty acid composition of single husked achenes of sunflower and to evaluate its performance in a selection program for high stearic content in materials derived from CAS-14.
|
MATERIALS AND METHODS
|
|---|
NIRS calibration equations for concentrations of individual fatty acids were developed from a set of 2510 single husked achenes from a wide range of sunflower breeding materials and environments, including standard fatty acid profiles, high oleic acid, and high linoleic acid types, as well as materials derived from high palmitic acid mutants (Osorio et al., 1995; Fernández-Martínez et al., 1997) and from medium and high stearic acid mutants such as CAS-4, CAS-8, CAS-3 (Osorio et al., 1995), and CAS-14 (Fernández-Moya et al., 2002). The mutants CAS-4, CAS-8, and CAS-3 exhibit increased stearic acid concentration in the seed oil uniformly expressed along the seed. However, CAS-14 seeds are not uniform for increased stearic acid levels, showing a decreasing gradient from the distal extreme of the seed to the embryo, which contains the lowest stearic acid levels (Férnandez-Moya et al., 2003). The seeds were chosen from plants grown in a wide range of environments, including several years and different locations.
Single sunflower achenes were manually husked and placed into an NIRS standard ring cup (ref. IH-0307, Infrasoft International, Port Matilda, PA) containing a Teflon adaptor specifically designed for sunflower achenes. Absorbance spectra (log 1/R) from 400 to 2500 nm were recorded at 2-nm intervals on a monochromator NIRSystems model 6500 (NIRSystems, Inc., Silver Spring, MD).
For reference, sunflower husked achenes were placed into 2-mL vials where they were finely crushed with a stainless steel rod. The oil was extracted and methylated following the one-step method described by Garcés and Mancha (1993). Fatty acid methyl esters were analyzed on a Perkin-Elmer Autosystem gas-liquid chromatograph (Perkin-Elmer Corporation, Norwalk, CT) equipped with a 2-m long column packed with 3% SP-2310/2% SP-2300 on Chromosorb WAW (Supelco Inc., Bellefonte, PA). The oven, injector, and flame ionization detector were held at 195, 275, and 250°C, respectively. The carrier gas was nitrogen at a flow of 20 mL min–1. The analysis time was 12 min.
Calibration equations for the individual fatty acids palmitic, stearic, oleic and linoleic acid were developed from a set of 2510 single husked achenes scanned by NIRS and further analyzed by the reference method. Mathematical procedures on the spectral information were performed with WinISI II software (Infrasoft International, LLC., Port Matilda, PA). Original reflectance spectra were corrected before calibration by applying second derivative transformation, standard normal variate transformation, and de-trend scatter correction. The second derivative was calculated from the log (1/R) spectra at gaps of 5 data points (10 nm) and a smoothing over segments of 5 data points (2,5,5,1). This combination was selected after having tested three additional math treatments (1,5,5,1; 1,10,10,1; and 2,10,10,1) with and without spectral corrections.
The NIRS calibration equation for stearic acid was used for selecting for high stearic acid concentration in the whole seed in a total of 8109 single husked achenes. These were F2, F3, F4, and F5 seeds from crosses between the sunflower mutant CAS-14 and several breeding lines, either with standard fatty acid profile or with high oleic acid content. A selection cut-off of 260 g kg–1 stearic acid was arbitrarily defined, so that seeds having NIRS stearic acid above the cut-off were a target for selection, whereas seeds having NIRS stearic acid below the cut-off were discarded. After NIRS analyses, a set of seeds was further analyzed by GLC to monitor the performance of NIRS calibration equations in routine analysis. The set included seeds with NIRS stearic acid concentration below the cut-off of 260 g kg–1 as well as seeds with NIRS stearic acid concentration above the cut-off. In the former case, seeds were randomly picked from the whole population, whereas in the latter only seeds from F4 and F5 populations containing a large proportion of high-stearic acid seeds were included. This strategy was adopted to preserve seed for the breeding and selection program. The low proportion of seeds with high stearic acid levels in the F2 and F3 populations precluded random destruction for GLC analyses. In total, 392 seeds with NIRS stearic acid concentration below 260 g kg–1 and 111 seeds with NIRS stearic acid concentration above 260 g kg–1 were selected to be analyzed further by the reference method.
|
RESULTS AND DISCUSSION
|
|---|
Development of NIRS Calibration Equations
The calibration set contained large variation for the concentration of the major fatty acids palmitic, stearic, oleic, and linoleic acid. Palmitic acid ranged from 19 to 459 g kg–1 (Fig. 1), exhibiting a bimodal distribution in which the high palmitic acid levels (>250 g kg–1) corresponded to materials derived from the high palmitic acid mutants CAS-5 and CAS-12. No seed with palmitic acid concentration from 150 to 250 g kg–1 could be analyzed because the mutants do not segregate within this range (Pérez-Vich et al., 1999). Stearic acid ranged from 8 to 535 g kg–1, with the highest levels corresponding to materials derived from the CAS-14 mutant. Oleic acid in the calibration set ranged from 22 to 916 g kg–1, whereas the range of variation for linoleic acid concentration was between 9 and 788 g kg–1 (Fig. 1).
View larger version (35K):
[in this window]
[in a new window]
|
Fig. 1. Histograms of palmitic, stearic, oleic, and linoleic acid concentration (g kg–1) in a calibration set of 2510 sunflower seeds (husked achenes).
|
|
Cross-validation statistics were calculated for the calibration equations developed for the four major fatty acids (Table 1). The calibration equation for palmitic acid did not show adequate validation statistics, as the coefficient of determination between NIRS and GLC data was low (r2 = 0.52) and the ratio of the standard error of cross validation to the standard deviation of the calibration set (SECV/SD) was high (0.69). The calibration equation for stearic acid was much more accurate, exhibiting an r2 of 0.80 and a SECV/SD ratio of 0.45. Even better validation statistics were obtained for oleic (r2 = 0.89, SECV/SD = 0.34) and linoleic acid concentration (r2 = 0.91, SECV/SD = 0.30).
View this table:
[in this window]
[in a new window]
|
Table 1. Cross validation statistics of NIRS calibration equations for concentration of individual fatty acids in a set of 2510 single seeds (husked achenes) of sunflower.
|
|
No previous results have been reported on developing NIRS calibration equations for estimating the concentration of the saturated stearic and palmitic acid in single husked achenes of sunflower. At a larger-sample scale, Pérez-Vich et al. (1998) developed accurate calibration equations for both fatty acids by analyzing from 10 to 15 husked achenes per sample, obtaining an r2 of 0.92 and a SECV/SD of 0.28 for both fatty acids in cross validation. In that work, the calibration set contained a similar variability for fatty acid composition to that used in the present research, except for stearic acid concentration. The latter ranged from 14 to 303 g kg–1 in the calibration set used by Pérez-Vich et al. (1998), whereas seeds containing between 8 and 535 g kg–1 stearic acid have been used in the present work. Therefore, it can be speculated that the expected loss in spectra resolution associated with the drastic reduction of the sample size from 10 to 15 husked achenes to one single husked achene caused the insufficient response of NIRS to differences in palmitic acid concentration observed in the present work, whereas this effect might have partly been compensated for by the use of a wider range of variation in the case of stearic acid.
The close relationship between NIRS and GLC data for oleic and linoleic acid concentration confirms the previous results obtained by Sato et al. (1995) in the analysis of linoleic acid concentration in single husked achenes and Velasco et al. (1999) in the analysis of oleic and linoleic acid concentration in single unhusked achenes of sunflower. Velasco et al. (1999) reported r2 in cross validation of 0.91 for oleic acid and 0.92 for linoleic acid, and SECV to SD ratios of 0.30 for oleic acid and 0.29 for linoleic acid, which are very similar to the statistics obtained in the present work. Therefore the analysis of husked achenes instead of unhusked achenes did not present clear advantages for these fatty acids.
Total lipid concentration was not determined in the single husked achenes used for calibration. Accordingly, it is not possible to determine whether the total lipid concentration in the seeds might influence NIRS discrimination of different fatty acid profiles. The utilization of a calibration set including a large variability for fatty acid profile, genetic background, and environmental conditions suggests that the prediction of individual fatty acid levels is based on specific absorbance properties of the different fatty acid molecules, rather than on indirect effects of the total lipid concentration or another seed constituent.
Evaluation of the Performance of the Calibration Equations in a Selection Program
NIRS calibration equations for stearic, oleic, and linoleic acid concentration were used for analyzing these fatty acids in single husked achenes within a program focused on the development of sunflower germplasm with very high stearic acid concentration, both in high oleic and high linoleic acid backgrounds. From a total of 8109 husked achenes analyzed by NIRS, a set of 503 of them was picked to be analyzed further by GLC to monitor the performance of NIRS calibration equations in routine analysis. The results revealed a close relationship between NIRS and GLC data for stearic, oleic, and linoleic acid concentration. The coefficient of determination between NIRS estimated and actual GLC values was 0.83 for stearic acid, 0.92 for oleic acid, and 0.94 for linoleic acid, whereas the standard error of performance (SEP) was 42.0 g kg–1 for stearic acid, 83.2 g kg–1 for oleic acid, and 66.2 g kg–1 for linoleic acid (Fig. 2). These results are very close to those obtained in cross validation during the development of the calibration equations, which revealed that the equations were robust enough to be used in routine analysis.
View larger version (23K):
[in this window]
[in a new window]
|
Fig. 2. Scatter plots of NIRS estimation of concentrations of individual fatty acids vs. GLC values in a set of 503 intact sunflower seeds (husked achenes).
|
|
Because the calibration equations to estimate the concentration of individual fatty acids in single husked achenes of sunflower were developed to be used in selection, we compared the observed performance of NIRS prediction with that expected from the validation statistics obtained during the development of the calibration equations. We focused on stearic acid concentration, since the performance of NIRS calibration equations for estimating oleic and linoleic acid concentration in single sunflower achenes has already been studied (Velasco et al., 1999). In our selection program for high stearic acid concentration, a selection cut-off of 260 g kg–1 stearic acid concentration, as determined by NIRS, was arbitrarily established. At a 95% confidence level, an interval for GLC stearic acid concentration from 2 SECV below the selection cut-off to 2 SECV above the selection cut-off was expected. Since the SECV was 46 g kg–1, seeds with up to 352 g kg–1 GLC stearic acid were expected within the nonselected group (NIRS stearic acid <260 g kg–1), whereas seeds with up to 168 g kg–1 GLC stearic acid were expected in the set of selected seeds (NIRS stearic acid 
260 g kg–1).
The set of 503 seeds used for monitoring the performance of NIRS equations included 392 seeds with NIRS stearic acid below the cut-off (nonselected seeds) and 111 seeds with NIRS stearic acid above the cut-off (selected seeds). After GLC analyses, the set of nonselected seeds averaged 154 g kg–1 stearic acid, ranging from 21 to 351 g kg–1, whereas the set of selected seeds averaged 324 g kg–1 stearic acid, exhibiting a range for this fatty acid from 228 to 440 g kg–1. In both cases, the selection errors were found within the expected intervals of ±2 SECV, which demonstrated the good performance of the NIRS calibration equation in routine selection for high stearic acid concentration in single husked achenes of sunflower. The efficiency of NIRS selection for high stearic acid concentration is most clearly observed in histograms of the GLC data for stearic acid concentration in both sets of selected and nonselected seeds (Fig. 3).
View larger version (35K):
[in this window]
[in a new window]
|
Fig. 3. Histograms of GLC values for stearic acid concentration in single sunflower seeds previously analyzed for stearic acid by NIRS in a non-destructive way. GLC analyses were conducted to monitor the performance of selection for high stearic concentration on the basis of NIRS information, with a selection cut-off of 260 g kg–1.
|
|
In summary, the results of the present research demonstrated for the first time the high efficiency of NIRS for selecting for stearic acid concentration in single husked achenes of sunflower and confirmed the good performance of NIRS in measuring oleic acid and linoleic acid concentrations. Since NIRS analyses are faster and more cost-effective than GLC analyses, NIRS can be used alternatively to GLC to speed up selection programs. Furthermore, NIRS offers the possibility of a nondestructive analysis of the fatty acid composition of the whole seed in materials with nonuniform fatty acid composition along the seed, for which GLC cannot be used.
|
ACKNOWLEDGMENTS
|
|---|
The authors thank Eva Valdecantos, Antonia Escobar and Gloria Fernández for excellent technical assistance. The work was funded by Advanta Seeds B.V.
Received for publication December 15, 2002.
|
REFERENCES
|
|---|
- Conte, L.S., O. Leoni, S. Palmieri, P. Capella, and G. Lercker. 1989. Half-seed analysis: Rapid chromatographic determination of the main fatty acids of sunflower seed. Plant Breed. 102:158–165.
- Downey, R.K. 1987. Genetic manipulation of oilseed quality. p. 669–676. In P.K. Stumpf et al. (ed.) The metabolism, structure, and function of plant lipids. Plenum Press, New York.
- Downey, R.K., and B.L. Harvey. 1963. Methods of breeding for oil quality in rape. Can. J. Plant Sci. 43:271–275.
- Fernández-Martínez, J., J. Osorio, M. Mancha, and R. Garcés. 1997. Isolation of high palmitic mutants on high oleic background. Euphytica 97:113–116.
- Fernández-Moya, V., E. Martínez-Force, and R. Garcés. 2002. Temperature effect on a high stearic acid sunflower mutant. Phytochemistry 59:33–38.[Medline]
- Fernández-Moya, V., E. Martínez-Force, and R. Garcés. 2003. Temperature-related non-homogeneous fatty acid desaturation in sunflower (Helianthus annuus L.) seeds. Planta 216:834–840.[ISI][Medline]
- Garcés, R., and M. Mancha. 1993. One-step lipid extraction and fatty acid methyl esters preparation from fresh plant tissues. Anal. Biochem. 211:139–143.[ISI][Medline]
- Osorio, J., J.M. Fernández-Martínez, M. Mancha, and R. Garcés. 1995. Mutant sunflowers with high concentration of saturated fatty acids in the oil. Crop Sci. 35:739–742.[Abstract/Free Full Text]
- Pérez-Vich, B., J. Fernández, R. Garcés, and J.M. Fernández-Martínez. 1999. Inheritance of high palmitic acid content in the seed oil of sunflower mutant CAS-5. Theor. Appl. Genet. 98:496–501.
- Pérez-Vich, B., L. Velasco, and J.M. Fernández-Martínez. 1998. Determination of seed oil content and fatty acid composition in sunflower through the analysis of intact seeds, husked seeds, meal and oil by near-infrared reflectance spectroscopy. J. Am. Oil Chem. Soc. 75:547–555.
- Sato, T., Y. Takahata, T. Noda, T. Yanagisawa, T. Morishita, and S. Sakai. 1995. Nondestructive determination of fatty acid composition of husked sunflower (Helianthus annuus L.) seeds by near-infrared spectroscopy. J. Am. Oil Chem. Soc. 72:1177–1183.
- Velasco, L., B. Pérez-Vich, and J.M. Fernández-Martínez. 1999. Nondestructive screening for oleic and linoleic acid in single sunflower achenes by near-infrared reflectance spectroscopy. Crop Sci. 39:219–222.[Abstract/Free Full Text]
Related articles in Crop Science:
- THIS ISSUE IN CROP SCIENCE
Crop Science 2004 44: 1-4.
[Full Text]
TOP
ABSTRACT
INTRODUCTION
