Effect of some variables on oil extraction yield from Mexican pumpkin seeds

Abstract

The effect of particle size (0.59 and 1.19 mm), meal to solvent ratio (1:5, 1:10, 1:15, and 1:20 w/v) and contact time (0, 1, 2, and 5 h) on oil extraction yield and efficiency from pumpkin (Cucurbita pepo) seeds using n-hexane were evaluated. Fatty acids profile was determined for the extracted oil. Results were analyzed by response surface methodology. Particle size, meal to solvent ratio and contact time, affected (p < 0.05) extraction yield. The yield (422 g/kg) and efficiency (860 g/kg) were highest at 0 h contact time, 0.59 mm particle size and a 1:20 (w/v) pumpkin seed meal:solvent ratio. Pumpkin seed oil contained (755 g/kg) unsaturated fatty acids and an elevated content of linoleic (431 g/kg) and oleic acids (324 g/kg). It is therefore classified as a high linoleic-oleic acids oil, making it a promising substitute for cotton, corn, sesame, sunflower, or soybean oils in our Mexican diet.

Se evaluó el efecto del tamaño de partícula (0,59 mm y 1,19), relación harina solvente (1:5, 1:10, 1:15 y 1:20 p/v) y tiempo de contacto (0, 1, 2 y 5 h) sobre el rendimiento y la eficiencia de extracción de aceite en semilla de calabaza (Cucurbita pepo) usando n-hexano. El perfil de ácidos grasos se determinó en el aceite extraído. Los resultados se analizaron mediante la metodología de superficie de respuesta. Tamaño de partícula, relación harina solvente y tiempo de contacto, afectaron (p < 0,05) el rendimiento de la extracción. Rendimiento (422 g/kg) y eficacia (860 g/kg) fueron mayores a tiempo 0, tamaño de partícula 0,59 mm y 1:20 (p/v) de harina:solvente. El aceite contiene 755 g/kg de ácidos grasos insaturados y un contenido elevado de ácido linoleico (431 g/kg) y oleico (324 g/kg). Por consiguiente, se clasifica como un aceite con un alto contenido de acidos linoleico-oleico, por lo que es un sustituto prometedor para los aceites de algodón, ajonjolí, girasol o de soya en la dieta mexicana.

Palabras claves::

• extracción de aceite
• Cucurbita pepo
• ácidos grasos
• n-hexano

Keywords:

• oil extraction
• Cucurbita pepo
• fatty acids
• n-hexane

Introduction

Fruit seed oils have fatty acid compositions and contain other components which make them beneficial to health and may aid in disease prevention. Seed oils with biological activity include those from raspberry, citrus, watermelon, apple, grape, pomegranate, and pumpkin, among others (Shahidi, 2006). Pumpkin (Cucurbita pepo) seed has received considerable research attention in recent years due to its elevated lipids (28–50%) and unsaturated fatty acids contents; particularly linoleic (43–56%) and oleic (24–38%) acids. It also contains β- and γ-tocopherols (vitamin E), as well as carotenoids such as luteoline and beta-carotene (Bellma et al., 2006). It has recognized antidiabetic (Quanhong, Ze, & Tongyi, 2003), antifungal (Wang & Ng, 2003), antibacterial and anti-inflammatory (Caili, Huan, & Quanhong, 2006), and antioxidant activities (Nkosi, Opoku, & Terblanche, 2006).

Cucurbita pepo is a low-cost and widely distributed produce (Atuonwu & Akobundu, 2010). The majority (63.8%) of worldwide pumpkin production from 2003 to 2008 occurred in seven countries: China, 5900 ton/year = 28.7%; India, 3500 ton/year = 16.9%; Russia, 1100 ton/year = 5.1%; the United States, 793 ton/year = 3.8%; Egypt, 708 ton/year = 3.4%; Ukraine, 686 ton/year = 3.3%; and Mexico, 522.6 ton/year = 2.5% (Food and Agriculture Organization Corporate Statistical Database [FAOSTAT], 2011).

The extraction method determines the feasibility of seed oil production from a given raw material. Oil yield and quality are vital to determine the viability of commercial production. Solvent extraction is traditionally employed to extract oil from oil seeds, and n-hexane is currently preferred worldwide for its efficacy and availability. Solvent extraction is a simple procedure based on the fact that a solute is distributed in two phases according to the equilibrium ratio, determined by the nature of the component and the two phases (Bockish, 1998). To facilitate oil extraction, seed or grain size is reduced by cracking or rolling (Prámparo, Mattea, & Gregory, 2003). Heat treatment before or during extraction causes cell emulsion rupture and reduces oil viscosity. Both of these properties facilitate oil fluidity and movement, and lower oil surface tension, but can negatively affect its chemical quality and increase the oil’s susceptibility to oxidation. Nonetheless, preheating offers clear advantages over other methods such as pressing and aqueous extraction (Evon, Vandenbossche, Pontalier, & Rigala, 2009), and the use of other solvents such as petroleum ether, and mixtures of chloroform/methanol with hexane, acetone, methanol, ethanol, ethyl acetate, and water, among others.

To the best of our knowledge, no reports have been published on the oil extraction from Cucurbita pepo seed that address particle size, solvent ratio, and contact time. The aim of present study was to evaluate the efficiency of pumpkin seed oil extraction with n-hexane, correlating with the effect of particle size, meal:solvent ratio, and contact time, as well as to determine the fatty acids profile of the extracted oil.

Materials and methods

Materials

Cucurbita pepo seed was purchased at a local market in San Juan Bautista Tuxtepec, Oaxaca state, Mexico. The dehulled seeds were cleaned manually to remove foreign material and then dried for 24 h at 60 ± 2°C in a convection oven (Binder ED 115, Germany). Once dried, the seeds were ground in a blender (Oster 465) until the residue passed through No. 16 (0.59 mm) and 30 (1.19 mm) meshes (USA standard test sieve ASTM E-11 Specification W.S. Tyler, Milwaukee, WI, USA). The resulting meal was stored in sealed polyethylene bags at 4°C until used.

Chemical analysis

Pumpkin seed proximate composition was determined in triplicate before seeds were dried following Association of Official Analytical Chemists (AOAC, 1997) methods: moisture (method 925.10); ash (method 923.03); protein (method 920.87); and fats (method 920.39). Crude fiber content was estimated by acid-alkaline digestion (Tejeda, 1992), and nitrogen-free extract was calculated by difference.

Oil extraction

A 25 g sample of seed meal was added to n-hexane (reagent grade) and heated to 60°C for 10 min. Extraction process variables included meal:solvent ratio (1:5, 1:10, 1:15, and 1:20 w/v); contact time (0, 1, 2, and 5 h); and particle size (0.59 mm and 1.19 mm). Post-extraction contact time (resting) was done at room temperature (25–27°C) and under constant agitation at 80 rpm (Cimarec digital stirring hotplates, Thermo Scientific, Model: SP131325, 2555 Kerper Boulevard, Dubuque, Iowa 52001 USA). Samples were initially vacuum filtered (Whatman Filter Paper No. 40, 150 mm Ø, Whatman International Ltd., Maidstone, England) and then filtered twice more with activated charcoal (2 g) and Celite (1.5 g). Solvent was removed with a rotary-evaporator (Model N-1, Eyela, Tokyo Rikakikal Co., Ltd., Japan). Residual solvent was removed by drying in an oven at 60°C for 1 h, followed by nitrogen flushing. The extracted oil was stored in amber flasks under nitrogen at 4 ± 1°C until analyzed. Extraction yield and efficiency were calculated with the equations:

Fatty acids profile

Fatty acids content was quantified with gas chromatography (GC). Raw pumpkin seed oil methyl esters were prepared following the method of Martínez, Vinay, Brieva, Hill, and Garcia (2005), with 1N HCl-methanol. Fatty acids methyl esters were analyzed with a gas chromatographer (Hewlett–Packard 6890) equipped with a flame ionization detector, and using nitrogen as the carrier gas (4 mL/min). A split/splitless injector (1:20 split) and detector ports (both set at 250°C) were employed. Samples (1 μL) were injected into a fused-silica capillary column (HP-Innowax, 60 m, 0.25 mm) at a programmed temperature of 130–230°C at 20°C/min and held for a 47-minute total run time. Fatty acids were identified based on the retention times of true standards (Lipid Standards Sigma-Aldrich: Fatty Acid Methyl Ester mixtures C14:0 – C:22:0, 18917 Supelco, 3050 Spruce Street, Saint Louis, Missouri 63103 USA) injected under the same conditions.

Experimental design and statistical analysis

A completely random design with a factorial arrangement was used (Table 1), with two particle sizes (0.59 and 1.19 mm); four meal:solvent ratios (1:5, 1:10, 1:15, and 1:20 w/v); and four contact times (0, 1, 2, and 5 h). Data were analyzed by response surface using the Design-Expert 7.0.0 package (Statease Inc., Minneapolis, MN, USA). Results were also analyzed by multiple linear regressions. Statistical significance of the regression terms was calculated by ANOVA and linear correlation analysis, both run with the software Statistica v. 8.0 package (StatSoft, Inc.). All experiments were done in triplicate.

Table 1. Experimental design and results of the oil extraction yield and efficiency.

Tabla 1. Diseño experimental y resultados del rendimiento de extracción de aceite y la eficiencia.

Treatment	Particle size (mm)	Contact time (h)	Ratio (meal:solvent)	Yield (g/kg)	Efficiency (g/kg)
1	0.59	0	1:5	275.0 ± 10.6h	559.6
2			1:10	327.8 ± 0.3i	667.1
3			1:15	366.7 ± 15.6jk	746.3
4			1:20	422.0 ± 1.4m	858.8
5		1	1:5	360.3 ± 9.4j	733.1
6			1:10	360.5 ± 1.51j	733.6
7			1:15	380.8 ± 1.58kl	775.0
8			1:20	424.1 ± 1.41m	863.0
9		2	1:5	258.2 ± 2.9gh	525.5
10			1:10	357.9 ± 6.5j	728.3
11			1:15	358.2 ± 5.2j	728.9
12			1:20	421.0 ± 0.7m	856.7
13		5	1:5	353.8 ± 12.9j	720.0
14			1:10	389.6 ± 14.6l	792.8
15			1:15	382.9 ± 18.0kl	779.2
16			1:20	425.0 ± 2.8m	864.9
17	1.19	0	1:5	229.5 ± 10.8e	467.1
18			1:10	260.0 ± 12.0gh	529.1
19			1:15	322.5 ± 14.8i	656.3
20			1:20	364.0 ± 5.7jk	740.7
21		1	1:5	125.9 ± 5.5b	256.2
22			1:10	126.7 ± 3.5b	257.8
23			1:15	153.6 ± 7.6c	312.5
24			1:20	160.6 ± 13.5c	326.7
25		2	1:5	91.5 ± 6.4a	186.2
26			1:10	189.5 ± 4.9d	385.6
27			1:15	191.0 ± 9.9d	388.7
28			1:20	255.5 ± 4.9gh	519.9
29		5	1:5	191.5 ± 3.6d	389.6
30			1:10	231.5 ± 9.2ef	471.1
31			1:15	250.0 ± 11.3fg	508.8
32			1:20	267.0 ± 7.1gh	543.3

Notes: (2 Particle size) × (4 Contact time levels) × (4 Ratio levels) = 32 total treatment combinations. Values are the mean of three replicates ± standard deviation. Different letter in the yield column indicate significant difference (p < 0.05) between treatments.

Notas: (2 Tamaño de la partícula) × (4 niveles de tiempo de contacto) × (4 niveles de relación) = 32 combinaciones de tratamientos totales. Los valores son la media de tres réplicas ± desviación estándar. Diferentes letras en la columna rendimiento indican diferencia significativa (p < 0,05) entre los tratamientos.

Results and discussion

Seed proximate composition

The dehulled pumpkin seed had a high fat (491.4 g/kg) and protein (354.5 g/kg) contents. These values were higher than those reported for C. pepo (Table 2), except for seeds grown in Egypt. Differences in pumpkin seed proximate composition are the result of seed characteristics, genus, species, harvest conditions, degree of fruit maturation, and cultivation zone characteristics. The C. pepo seeds analyzed here from the Tuxtepec region in Oaxaca state, Mexico, are a good alternative oil and protein source with potential applications in innovative food formulations such as cooking oils, as an ingredient in margarine blends, flours for instant soups, cookies, etc. (Atuonwu & Akobundu, 2010; Zdunczyk, Minakowski, Frejnagel, & Flis, 1999).

Table 2. Pumpkin (Cucurbita pepo) seed proximate composition (g/kg, dry basis) compared to previous reports.

Tabla 2. Composición proximal de la semilla calabaza (Cucurbita pepo) (g/kg, base seca) comparado con reportes anteriores.

Component	C. pepo(Mexico^1)	Literature
		C. pepo(Iran^2)	C. moschata(Egypt^2)	Cucurbita spp.(Japan^3)	C. pepo(Africa^4)	C. pepo(Egypt^5)	C. maxima (Saudi Arabia^6)
Protein (N × 6.25)	354.4 ± 43.0	265	240	285–298	280–320	364.7 ± 4.8	392.2 ± 24.6
Crude fat	491.4 ± 02.6	370	430	473–486	239–350	510.1 ± 3.1	436.9 ± 39.2
Crude fiber	23.0 ± 0.8	30	60	–	–	443 ± 5.9	21.3 ± 5.7
Ash	52.7 ± 0.9	12.7	12.2	–	33	321 ± 1.6	51.4 ± 12.3
N-free extract*	78.5 ± 10.2	288.6	230	–	365–379	488 ± 4.2	98.2 ± 27.0

Note: ¹This study; ²Al-Khalifa (1996); ³Yoshida et al. (2006); ⁴Younis et al. (2000); ⁵El-Adaway & Taha (2001); ⁶Alfawaz (2004).

Means ± standard deviation. *Value obtained by difference.

Nota: ¹Este estudio; ²Al-Khalifa, 1996; ³Yoshida et al. (2006); ⁴Younis et al. (2000); ⁵El-Adaway & Taha (2001); ⁶Alfawaz (2004).

Media ± desviación estándar. *Valor obtenido por diferencia.

Effect of particle size, solvent proportion, and contact time

Regression analysis of seed oil yield produced a significant (p < 0.05) regression model with an R² = 0.89 (Table 3). In linear terms, solvent proportion had a significant (p < 0.05) effect and a positive correlation with yield (R = 0.39, p < 0.05). In quadratic terms, particle size and contact time had a significant (p < 0.05) effect, although none of the interactions were significant (p > 0.05). There was a strong negative correlation between particle size and yield (R = −0.79, p < 0.05).

Table 3. Regression coefficients of the response surface models for oil extraction yield.

Tabla 3. Coeficientes de regresión de los modelos de superficie de respuesta para el rendimiento de extracción de aceite.

Regression coefficients	Oil yield
Intercept	2.1E + 17
Linear
X1	12.857
X2	−12.241
X3	48.066
Quadratic
X1^2	−2.1E + 17
X2^2	78.853
X3^2	−0.674
Interaction
X1X2	−8.618
X1X3	5.346
X2 X3	−11.830
R^2	0.889
P of F	< 0.05

Notes: X₁ = Particle size, X₂ = Contact time, and X₃ = Solvent proportion. Bold numbers indicate significant parameter estimates (p < 0.05).

Notas: X₁ = Tamaño de partícula, X₂ = Tiempo de contacto y X₃ = proporción de solvente. Los números en negrita indican las estimaciones significativas de los parámetros (p <0,05).

Oil extraction yield was higher as particle size decreased and contact time increased (Figure 1). Maximum yield (425 g/kg, with 864.9 g/kg extraction efficiency) was attained with the smallest particle size (0.59 mm) and longest contact time (5 h). There was a linear response in both meal:solvent ratio and particle size (Figure 2): yield increased as particle size decreased and meal:solvent ratio increased. When analyzed together, both meal:solvent ratio and contact time exhibited a saddle-shaped behavior with minimum values occurring near 3 h and increases at both higher and lower contact times (Figure 3 ). An interaction effect was observed since yield increased as solvent proportion increased at short contact times, and when both contact time and meal:solvent ratio were high. However, the ANOVA showed no differences (p > 0.05) between treatments 4, 8, 12, and 16 (Table 1), indicating that contact time had no effect (p > 0.05). The overall maximum yield was 421–425 g/kg with 860 g/kg efficiency.

Figure 1. Experimental (points) and predicted (mesh) oil extraction yield as a function of contact time and particle size.

Figura 1. Experimental (puntos) y predicho (malla) del rendimiento de la extracción de aceite en función del tiempo de contacto y el tamaño de partícula.

Figure 2. Experimental (points) and predicted (mesh) oil extraction yield as a function of particle size and solvent proportion (w/v)

Figura 2. Experimental (puntos) y predicho (malla) del rendimiento de la extracción de aceite en función del tamaño de partícula y la proporción de solvente (p/v)

Figure 3. Experimental (points) and predicted (mesh) oil extraction yield as a function of contact time and solvent proportion (w/v).

Figura 3. Experimental (puntos) y predicho (malla) del rendimiento de la extracción de aceite en en función del tiempo de contacto y la proporción de solvente (p/v).

The effect of particle size is associated to an increase in cellular damage as particle size decreases. This favors removal of the oil on the particle surface and diffusion of n-hexane within the particle (Patricelli, Assogna, Emmi, & Sodini, 1979; Şaşmaz, 1996). Pre-extraction heat treatment causes expansion and rupture of cell structures, which enhances material plasticity and permeability, it also facilitates oil release and thus increase yield (Li, Bellmer, & Brusewitz, 1999). This can be noted in our results, which show that extraction rate slowed down as particle size increased, and improved with smaller particle sizes (Figures 1 and 2). In addition, maximum oil yield (425 g/kg) occurred with the 0.59 mm particle size. This tendency has been reported previously: Matos and Acuña, (2010) found that small (0.5 mm) particles provide high oil yield, although they also cloud the oil, are difficult to separate and often remain as suspended particles, which is an undesirable trait. In another study, the limiting step was proposed to be the diffusive stage, with higher extraction efficiency as particle size diminished and temperature increased (Patricelli et al., 1979).

The meal:solvent ratio marks the difference in oil yield up to a certain extent (Kwiatkowski & Cheryan, 2002). A high initial extraction rate is attributed to rapid solution of the oil on the solid’s surface and a higher conduction mass transference force anticipated by the high solvent concentration. A slower rate can be attributed to a lower motive force resulting from a lower solvent concentration. Extraction with solvents is a mass transfer process in which materials (oils) are moved from one phase to another to separate one or more compounds from a mixture (Giraldo, Velásquez, & Cuartas, 2010). The meal:solvent ratio is one of the most important variables in the extraction process, such that at higher solvent proportions the mass transfer coefficient increases, producing greater oil extraction (Seth, Agrawal, Ghosh, Jayas, & Singh, 2007). Material transfer from a particle by solvent effect is called leaching or percolation. Leaching involves a complex mechanism that implies transfer of a solvent to the surface of solid particles, penetration of the solvent into the solid matrix, incorporation of the solute into the solvent by diffusion, and transfer of the solute into the bulk solvent (Adu-Amankwa, 2006). Diffusion determines the effect of contact time. The speed at which equilibrium is reached and the oil extraction rate are influenced by oil diffusion into the solvent, particle size and internal structure. Most of the oil is extracted within the first 30 min, leaving residual oil (<1%) that requires long extraction times to be removed (Bernardini, 1981).

The highest yield observed in this work (425 g/kg) is comparable to yields reported in the relevant literature. For example, oil yield from C. pepo seeds from Africa using n-hexane was 350 g/kg refluxed for 16 h/60°C in a Soxhlet extractor (Younis, Ghirmayb, & Al-Shihry, 2000). With peach pits, a 570 g/kg yield was attained with a 3 h/60°C hexane extraction and 0.71 mm particle size (Hernández & Mieres, 2005), while 445 g/kg yield was accomplished with a 90 min extraction time, 0.5 mm particle size and 1:4 meal:liquid ratio and 65°C (Matos & Acuña, 2010). A 443 g/kg extraction yield was produced from neem (Azadirachta indica A. Juss) using an n-hexane extraction at 50°C/3 h and a 0.425–0.71 mm particle size, (Liauw et al., 2008). Extremely high (950–990 g/kg) n-hexane extraction yields can be produced by using a Soxhlet device for a 24 h period (Achten et al., 2008). The extraction method used in the present study produced lower yields than the above methods, however, long extraction times and prolonged exposure to heat treatment can degrade the oil. The present method is therefore more appropriate for use with high oil content (50%) materials such as pumpkin seeds. In addition, the short exposure time to the temperature used hereby (60°C/10 min) preserved oil quality while still helping to shorten extraction time; this represents both energy savings and reduced operating costs.

Oil characterization

The main fatty acids in the studied pumpkin seed were linoleic (C18:2) and oleic acids (C18:1). Unsaturated fatty acids accounted for 755 g/kg of the total, while saturated fatty acids represented 245 g/kg (Table 4 ). Overall oil composition (431 g/kg linoleic and 324 g/kg oleic) was similar to data from previous reports: 430 g/kg linoleic and 340 g/kg oleic in C. pepo from Iran (Al-Khalifa, 1996); and 420 g/kg linoleic and 380 g/kg oleic in C. pepo from Greece (Tsaknis, Lalas, & Lazos, 1997). In contrast, other authors have reported linoleic acid levels near or above 500 g/kg in cucurbits: 550 g/kg in C. pepo from Egypt (El-Adaway & Taha, 2001); 503 g/kg in C. pepo from Africa (Younis et al., 2000); 530 g/kg in C. moschata (Al-Khalifa, 1996); 520 g/kg in C. maxima (Alfawaz, 2004); and 465.8 g/kg in C. pepo (Delaš, 2010). In other studies, oleic acid is reported to be the main fatty acid in pumpkin seed oil: 410–460 g/kg oleic and 334–343 g/kg linoleic acids in pumpkin seeds from Italy and Libya (El-Gharbawi, 1978); 466–604 g/kg oleic acid in different C. pepo lines (Idouraine, Kohlhepp, & Weber, 1996); and 504 g/kg oleic acid in pumpkin seed (Zdunczyk et al., 1999). However, this type of fatty acid is not essential in the human diet because the organism can synthesize it. Polyunsaturated n-3 fatty acids are vital to human health and are known to aid in preventing certain diseases, particularly cardiovascular disorders (De Lorgeril et al., 1994; Hartman, 1995; Leaf & Kang, 1998; Yoshida, Tomiyama, Hirakawa, & Mizushina, 2006). Polyunsaturated omega-6 fatty acids are also necessary in the human diet but excessively high intakes can cause heart problems, asthma, some types of cancers, arthritis, and depression. Omega-3 fatty acids reduce inflammation while omega-6 fatty acids cause inflammation, meaning that a chronic imbalance between the two series of fatty acids can constitute a health risk (Calder, 2001). Both types also play a crucial role in brain function and growth in children, emphasizing again that a balance between them is required in a healthy diet.

Table 4. Fatty acids composition of pumpkin seed oil.

Tabla 4. Composición de ácidos grasos del aceite de semilla de calabaza.

Fatty acid (g/kg)	Crude
Palmitic (C16:0)	158 ± 1.1
Stearic (C18:0)	87 ± 2.2
Oleic (C18:1)	324 ± 7.1
Linoleic (C18:2)	431 ± 21.5
Total unsaturated fatty acids	755 ± 28.6
Total saturated fatty acids	245 ± 3.3

Note: Values represent the mean of 3 replicates ± standard deviation.

Nota: Los valores representan la media de 3 repeticiones ± desviación estándar.

The most abundant (158 g/kg) saturated fatty acid in pumpkin seed oil was palmitic (C16:0). This coincides with the (80–150 g/kg) reported by other authors in this raw material (Al-Khalifa, 1996; Delaš, 2010; El-Adaway & Taha, 2001; Idouraine et al., 1996; Tsaknis et al., 1997; Younis et al., 2000; Zdunczyk et al., 1999). The differences between the pumpkin seed oil fatty acids profile observed in the present study and those in previous reports can be attributed to factors such as variety, origin, and drying conditions, among others.

Conclusions

Cucurbita pepo seed contains both high protein content (354.5 g/kg) and crude fat content (up to 491.4 g/kg), suggesting that it can be a valuable source for oil and protein extraction. All three studied variables (solvent proportion, particle size, and contact time) and their interactions affected (p < 0.05) oil yield from n-hexane extraction. Under the studied conditions, the highest overall yield (860 g/kg extraction efficiency) was observed at 0 h of contact time, 0.59 mm particle size, and a 1:20 (w/v) meal:solvent ratio. In addition, the short exposure to the temperature employed (60°C/10 min) preserved oil quality while still helped to shorten extraction time; this represents both energy savings and reduced operating costs. The extracted pumpkin seed oil contained large amounts of poly- and mono-unsaturated fatty acids and low contents of saturated fats, making it a good unsaturated fats source. This oil has potential applications as cooking oil or as an ingredient in margarine blends or highly unsaturated oil substitutes.