Index | Search | Home | Table of Contents

link to pdf file

Holser, R.A. and G.A. Bost. 2002. Extraction of lipid components from seeds of perennial and woody Hibiscus species by supercritical carbon dioxide. p. 550–555. In: J. Janick and A. Whipkey (eds.), Trends in new crops and new uses. ASHS Press, Alexandria, VA.

Extraction of Lipid Components from Seeds of Perennial and Woody Hibiscus species by Supercritical Carbon Dioxide

R.A. Holser and G.A. Bost*

*We acknowledge the sponsorship of this project by USDA/SBIR grant 2001-33610-10402.


The genus Hibiscus exhibits great diversity in the production of natural materials, with both edible and industrial applications. Components such as flowers and green tissues may be consumed as specialty foods, while the bast fibers (cellulosic), core fibers (lignitic and cellulosic), and seed oils (primarily poly-unsaturates) are a renewable source of industrial materials and edible oils.

In this study, a total of 22 seed genomes of perennial and woody species and hybrids of Hibiscus were selected and characterized for lipid components, using supercritical carbon dioxide (CO2) as the extractant in order to avoid the use of organic solvents. These samples include 6 hybrid genomes (developed from native American Hibiscus species and subspecies) and 13 native American Hibiscus species and subspecies, all of which are perennials with annually produced canes. Three non-native woody Hibiscus species were also characterized: two Asiatic (H. syriacus and H. mutabilis), and one Pan-Pacific (H. hamabo hamabo). The data presented herein are part of an on-going USDA-funded SBIR study which includes chemical characterization of Hibiscus flower pigments (Puckhaber et al. 2002), and seed oils, seed meals, and seed proteins. The perennial and woody Hibiscus used in these studies are also being evaluated for fiber applications (industrial and dietary), nutraceutical products, and bioremediation applications.

The suite of perennial and woody Malvaceae species (native American species in boldface) grown at our production farm (USDA Hardiness Zone 8–9) includes: Abelmoschus manihot, A. moschatus, Abutilon hybridum, Kosteletzkya virginica, Pavonia hastata, P. lasiopetalus, and Sida spinosa, as well as the following Hibiscus species: H. aculeatus, H. coccineus, H. hamabo hamabo, H. hamabo tileaceous, H. laevis (multiple forms), H. moscheutos moscheutos (multiple forms), H. moscheutos grandiflorus, H. mutabilis (multiple forms), H. striatus lambertianus, H. syriacus (multiple forms), Malvaviscus arboreus arboreus, and M. a. drummondii. In addition, we have developed hybrids of North American native species (we currently grow ~125 hybrids).

In the US, most of these Malvaceae genera have tended to be regarded strictly as garden plants. However, their potential for other applications is substantial, as indicated above. The emphasis in this paper is on the characterization of the seed oils for these genomes, and comparison of the seed-oil yields and fatty acid composition results obtained herein with other published conventional oils (canola, cotton, crambe, peanut, soybean, and sunflower) and other “new” oilseeds in the New Crops oilseed database.


Preparation of Seed Samples

Selected samples of whole-seed were separated from Hibiscus seedpods which were harvested by genome in early October, 2001. Subsamples of seed were mechanically ground to a nominal particle diameter of 0.1 mm with a Wiley mill. Ground seed samples were stored in a –20°C freezer until extractions were performed.


Lipid components were extracted from 5 g samples of ground seed using supercritical carbon dioxide and a mixture of carbon dioxide modified with ethanol (Montanari et al. 1999). Extractions were performed with an ISCO model 3560 supercritical fluid extractor. This instrument was programmed to perform the sequential extraction of neutral and polar lipids by the addition of 15% ethanol. Neutral lipids were extracted with carbon dioxide at 80°C and 5370 MPa for 45 min. Polar lipids were subsequently extracted with a mixture of carbon dioxide and 15% ethanol at the same temperature and pressure. Fractions were collected and stored at –20°C.


High performance liquid chromatography (HPLC) was used to analyze extracts for major neutral and polar lipid components. A 5-µm silica column, 4.6 mm × 250 mm (Alltech, Deerfield, Illinois), was used with a solvent gradient of hexane/isopropanol/water with ultraviolet (UV) and evaporative light scattering detectors (ELSD). This system was able to resolve both neutral and polar lipid classes (Moreau et al. 1990). Peak identification was achieved by comparison of retention time to commercial standards (Avanti Polar Lipids, Alabaster, Alabama).

An aliquot of each collected triglyceride fraction was converted to the corresponding fatty acid methyl ester (FAME). This was performed by trans-methylation of the triglyceride with sodium methoxide and recovery of the ester in hexane. Analysis was by gas chromatography, Hewlett-Packard 5890, with a flame ionization detector.


The analytical results indicated that manipulation of both temperature and pressure could extend the solvating power of both solvent systems to achieve recoveries of predominately nonpolar components, e.g., triglycerides, moderately polar, and polar components.

The oil yield was expressed as the weight percent of the total extracted lipids determined gravimetrically from the collected fractions on a dry basis of whole ground seed. These data are presented in Table 1 for 22 species. The values range from a low of 8.5% for a species indigenous to Madagascar (H. calyphyllus) to nearly 20% for the hybrid BOSTx®HHHybrid ‘Georgia Rose’™. Excluding these two, the average oil yield across the remaining entries examined was 11.4% (Table 1).

Table 1. Nonpolar fatty acid methyl exter analyses (FAME) profiles for BOSTx®HHHybrids species and hybrids.

ID Species and hybrids Oil yield as % seed wt. Fatty acids (mole %) Total % recovered Total %
C14z C16 C16:1 C18 C18:1 C18:2 C18:3 C20 C20:1 C22
3 H. laevis (normal pod) 12.67 0.26 17.48 0.34 0.00 31.31 48.61 0.57 0.52 0.00 0.00 99.09 80.83
5 H. laevis (balloon calyx) 10.99 0.28 17.85 0.42 0.00 30.61 32.78 0.69 16.59 0.00 0.31 99.53 64.50
11 BOSTx(r) 'Razberri Rhapsody' 12.34 0.20 5.04 11.20 0.25 24.76 9.74 7.04 39.07 0.76 0.37 98.43 52.74
12 BOSTx(r) 'Razberri Ruffles' 10.43 0.44 21.10 0.36 0.00 29.63 45.45 0.56 0.42 0.00 0.00 97.96 76.00
13 BOSTx(r) 'Mary Todd Lincoln' 11.34 0.00 17.87 32.11 0.00 0.00 47.29 0.63 0.46 0.00 0.00 98.36 80.03
21 H. moscheutos 'Lowrey's Pink' 14.77 0.00 16.89 0.00 0.00 32.57 50.52 0.00 0.00 0.00 0.00 100.00 83.09
27 BOSTx(r) 'Nathan's Star' 11.98 0.00 16.60 0.00 0.00 28.98 54.42 0.00 0.00 0.00 0.00 100.00 83.40
28 BOSTx(r) 'Georgia Rose' 19.95 0.00 19.60 0.00 0.00 25.66 53.80 0.03 0.24 0.00 0.00 99.34 79.49
31 BOSTx(r) 'Grace Coolidge' 9.79 0.00 18.90 0.00 1.64 27.18 49.47 0.00 0.00 0.00 0.00 97.19 76.64
38 H. laevis (std. leaf) 10.29 0.00 18.40 0.00 3.35 23.20 55.10 0.00 0.00 0.00 0.00 100.05 81.65
43 H.mutabilis 'single pk' (det) 8.87 0.00 17.93 0.00 2.88 17.83 61.35 0.00 0.00 0.00 0.00 99.99 79.18
44 H. mutabilis 'Dbl. Pinks' 9.06 0.00 19.00 0.00 0.00 16.30 64.90 0.00 0.00 0.00 0.00 100.20 81.20
46 H. calyphyllus 8.54 0.00 23.50 0.43 0.00 18.40 55.30 0.68 0.00 8.53 0.00 98.31 83.34
53 H. moscheutos-1 'Lowry's Pink' 13.05 0.00 14.85 0.16 1.68 31.85 49.70 0.62 0.50 0.00 0.00 99.34 81.55
54 H. moscheutos-3 'Arkansas' NR 0.00 16.83 0.00 0.00 29.07 53.05 0.00 0.00 0.00 0.00 98.95 82.12
56 H. moscheutos -4 'Hairy Pod'   0.15 16.48 0.17 1.33 25.86 55.14 0.00 0.00 0.39 0.00 99.50 81.54
58 H. moscheutos -5 NR 0.00 18.06 4.06 29.50 48.37 0.00 0.00 0.00 0.00 0.00 99.99 52.43
60 H. striatus lambertianus 16.23 0.46 0.00 23.84 0.00 8.41 15.35 0.54 48.41 0.00 0.35 97.36 48.14
61 H. dasycalyx NR 0.14 17.30 0.16 0.00 30.08 51.80 0.34 0.20 0.00 0.00 100.03 82.38
62 H. dasycalyx NR 0.13 19.09 0.24 1.86 28.70 42.32 0.00 6.62 0.29 0.00 99.27 71.56
64 H. dasycalyx NR 0.34 16.61 0.00 3.42 26.71 52.96 0.00 0.00 0.00 0.00 100.00 79.67
83 H. laevis 16.01 0.00 17.60 0.00 0.00 28.88 51.78 0.69 0.00 0.00 0.00 98.95 81.35

zC14=myristic, C16=palmitic, C16:1=palmitoleic, C18=stearic, C18:1=oleic, C18:2=linoleic, C18:3=linolenic, C20=arachidic, C20:1=arachidoleic, C22=behenic.

The results obtained show a high degree of unsaturation across all entries examined. Oleic, linoleic, and linolenic fatty acids appear as predominate components, with only minor amounts of the saturated fatty acids measured. Trace amounts of C14 and C20 typically appear, although one hybrid exhibits an unusual abundance (39.1%), of C20 (No.11, BOSTx®HHHybrid ‘Razberri Rhapsody’™). The distribution of the major phospholipids [phosphatidyl-ethanolamine (PE); phosphatidic acid (PA); phosphatidylserine (PS); phosphatidyl-choline (PC); and lysophosphatidylcholine, (LPC)] was found to vary significantly among the Hibiscus entries examined. These data, as well as the currently available % oil data from 14 entries, are compiled in Table 2 and are expressed as percent of total phospholipids.

Table 2. Polar phospholipid profiles for BOSTx®HHHybrids species and hybrids.

ID Species and hybrids % of DAG detected % of total phospholipids
27 BOSTx(r) 'Nathan's Star' 0.0 100.0 0.0 0.0 5.6 51.3 0.0
28 BOSTx(r) 'Georgia Rose' 100.0 0.0 0.0 0.0 5.6 44.0 50.5
31 BOSTx(r) 'Grace Coolidge' 12.1 87.9 0.0 0.0 29.1 70.9 0.0
43 H. mutabilis 'single pk' (det) 21.5 78.5 4.4 2.7 9.2 17.6 66.2
44 H. mutabilis 'Dbl. Pinks' 14.5 85.5 4.4 0.0 0.0 56.5 39.0
54 H. moscheutos-3 'Arkansas' 68.5 31.5 13.0 ND 20.5 8.3 58.1
56 H. moscheutos -4 'Hairy Pod' 94.7 5.3 4.6 15.6 23.0 24.6 32.2
58 H. moscheutos -5 56.8 43.2 6.3 4.2 4.6 34.7 50.2
60 H. striatus lambertianus 62.7 37.3 4.6 2.3 5.8 47.4 40.0
61 H. dasycalyx 66.9 33.1 5.1 3.5 2.9 59.4 29.1
62 H. dasycalyx 66.9 33.1 5.1 3.5 2.9 59.4 29.1
64 H. dasycalyx 39.0 61.0 7.6 0.0 7.0 52.3 33.2
83 H. laevis 48.5 52.5 2.1 7.9 51.5 15.4 0.0

DAG= diacylglycerol
MGDG = monogalactosyldiacylglycerol (as % of DAG detected)
DGDG = digalactosylacylglycerol
PE = phosphatidylethanolamine
PA = phosphatidic acid
PS = phosphatidylserine
PC = phosphatidylcholine
LPC = lysophosphatidylcholine (as % of total phospholipids)

The extraction and analysis of Hibiscus seed oils revealed large amounts of unsaturated fatty acids. The presence of the nutritionally essential fatty acids suggests potential applications as an edible oil. The oil yields, however, were relatively low when compared to commodity oilseeds such as soybean, which averages about 18.6% (from NCAUR New Crops Database, circa 1999).

Interestingly, one of the hybrids (BOSTx®HHHybrid ‘Georgia Rose’™) in this data set exhibited 19.95% oil content—almost twice the average oil content of the other cultivars analyzed in Table 1—and appears to offer the possibility of improving oil yield through a program of selective breeding. This hybrid was not specifically crossed to improve oil content, but exhibited an impressive increase over the average of oil content of the other species that were examined for % oil (as percent of total seed weight).

Several other genomes showed respectable oil yield, including H. striatus lambertianus (16.23%) [No. 60], H. laevis (16.01%) [No. 83], BOSTx®HHHybrid ‘Mary Todd Lincoln’™ (14.80%) [No. 13], H. moscheutos ‘Lowry’s Pink’ (14.77%) [No. 21], and (13.05%) [No. 53]. The lowest oil yields in the data set presented in this document were found in H. mutabilis ‘Single Pink’ (8.87%) and H. mutabilis ‘Double Pink’ (9.06%).

Additional information such as average seed yield per plant, number of seed pods per plant, number and size of seed per pod, and planting density is needed to extrapolate these results to oil yield per hectare. These data are available for review at the website. These data (while not complete for all of our cultivars) indicate that seed yields in fields with 2–3 year-old crowns (1–5 canes per plant) should yield about 224–1680 kg/ha of seed; more mature crowns (10+ canes/crown) are expected to yield 1792–6720 kg/ha.

Average seed size across entries discussed herein appears to be an inverse function of pod size; as a general rule, the larger the seed (at least within a given Malvaceae taxon), the fewer number of seeds are produced per pod. This relationship is clearly shown in the species Hibiscus seeds, but the relationship is not as obvious for the hybrid genomes, partly because hybrids tend to have larger seeds than those of their parent species. In essence, this means that seedoil production must be evaluated not just on the % oil in the seed, but also on the total seed yield for a given crop. Without good biometric data, it is difficult to “normalize” seed production across genomes (within Malvaceae and/or in comparison to more conventional oil seeds) without also fully characterizing the relationship between seed and pod characters and oil yield.

As fields of perennial Hibiscus crops are established much more quickly and reliably from rooted stock than from seed, all seed produced would be available as a harvest product. Vegetative propagation is achieved by a combination of root crown divisions during the winter dormancy period and rooted green cuttings during the growing season.

All hybrids and many of the species selections are vegetatively propagated by cuttings rather than seed, because propagules of mature canes reach maturity much faster than seedlings; and clonal propagation assures uniformity of product.

For full scale production, two different field setup strategies are anticipated for production of specific commercial targets. These correspond to: (1) production fields dedicated to fresh produce (edible flowers and pods, for local gourmet restaurants and bulk production of petal extracts for natural food colorants and mucilages); (2) production fields dedicated to seed production for use in manufacturing seed oils, seed meals, and refined protein and oil isolates; and (3) end-of-season cane and wood harvests (for industrial fiber, biomass fuels, and/or bioethanol markets) from both field setup types.

It is also possible to produce green pods (as an okra substitute) and/or mature oilseed from the same plants that are harvested for flower colorants. This can be accomplished by hand-pollinating the flower styles after petals are removed for use in fresh food and food colorant applications.

The ability to harvest 5–6 months of multiple, high-value fresh products for gourmet markets offsets the disadvantage of having to hand pick the fresh products. In addition, large-volume production of dehydrated petals (for use in the natural food colorant market), fresh and/or dried petals (for the mucilage market), and root and green stem mucilages from the discards generated during crown and stem propagation have potential for additional high-volume value-added products.

The analysis of the extracts for polar lipids (Table 1) suggests that the seed would be a significant source of phospholipids that would be easily recovered by a selective extraction. The use of carbon dioxide/ethanol mixtures is appropriate for compounds destined for the edible or nutraceutical markets. Natural products recovered by such environmentally acceptable processes can be obtained in high purity and offer some advantage in market position. These products would be included in the natural or nutraceutical category to be marketed directly as supplements or formulated into functional foods.

The additional end-of-season harvest of dormant cane—which can be machine harvested readily with conventional equipment—and the perennial habit of the genomes (which makes annual plowing and planting unnecessary) also greatly enhance the potential profits from the crop. The annual cane harvest alone (at a planting density of 6173 crowns/ha for the hybrid perennial native American caning types) is projected to be about 15.7 tonnes (t)/ha. For “woody” species (e.g., H. hamabo, H. mutabilis, H. syriacus, H. tileaceous, and Malvaviscus arboreus arboreus), crowns are planted at a density of about 1234–2963/ha, and dormant branches and/or trunks can be harvested on a 3 to 5 year cycles, depending on the genome. The woody species are expected to generate about 22.4–33.6 t/ha per periodic harvest.


Potential Oil Seed Yields in Malvales

A tabulation of available oil seed data from genera and species of Malvaceae in Order Malvales is summarized at for comparison to the data presented in this report. These data include average % oil per genome, as well as percent saturated and unsaturated oil, and the average number of seed per carpel or pod. Malvales species are presented alphabetically by family; genera are listed alphabetically within each family, and species are listed alphabetically within each genus. The table presents data for 27 genera encompassing 100 Malvaceaous species. Percent oil yields from conventional oil seeds, including cotton (26.3%), are summarized at the end of the table, along with average % oil yields for crambe (32%–37%), sunflower (37.6%–53%), canola (33.8%–41.9%), soybean (18.6%) and peanut (43.1%).

The % oil in the available Malvaceae oilseed literature ranges from 5.71% (H. tileaceous) to 52.43% (Pachira aquatica). The % oil data presented in this investigation ranges from a low of 8.87% (for H. mutabilis single pink) to a high of 19.95% (for BOSTx®HHHybrid ‘Georgia Rose’™). However, average % oil for seeds of a given genome is meaningless with regard to production yield unless the relevant biometric data for plant habit (annual, biennial, perennial, or tree form), size of pods, number of seed per pod, size of seeds, planting density, and production life is also available.

Distribution of Suitable Habitats for Production of Malvales Crops

A comprehensive summary of representative genera and species in order Malvales, along with their distribution, habit, and habitats is also available at Many of the species of Malvales listed in this table are perennial or arboreal species that are suitable for sustainable agricultural (using zero runoff protocols) in temperate mild and subtropical zones (USDA zones 8b, 9, and 10) of the United States, much of Mexico, the more northern (or higher altitudes) in the Central Americas (above latitude 15°N), and the Caribbean Islands.

Other suitable climates in the Americas include coastal, riverine, and lacustrine habitats in Southern South America (below latitude 15°S). Interior higher-altitude lakes, swamps, and broad riverine floodplains would also be suitable for production of these oil crops.

Southern Australia, Southern Africa, and Madagascar (also below latitude 15°S), the coastal regions of the Mediterranean, and riverine, lucustrine, and floodplain areas of Europe (e.g., Spain, France, and England) are also excellent climatic regions for many of these species. Suitable habitats and latitudes are also available above 15°N in coastal, riverine, and lacustrine habitats of Arabia, India, Berma, Thailand, China, northern to subtropical Vietnam, northern Phillipines, North and South Korea, and Japan.

The fact that production field(s) can be simultaneously (or sequentially) harvested for edible flowers, natural food colorants, emollients, mucilages, dietary, and/or industrial fibers (for production of fiber boards, absorbants, papers, cordage, biomasss ethanols, and biomass fuels) should make the introduction of many of these new Malvaceae species a viable alternative to timber trees, especially where conventional annual crops are unsuitable, or where deforestation has resulted in severe loss of topsoil and nutrients.

Other genera/species listed on the website are more suited to USDA Zones between 8–3 (e.g., Alcea, Altheae, Callirhoe, Lavatera, Malva, Sidalcea, Sphaeralcea, and Tilia). None of the species and hybrids analyzed in this report are from tropical genomes (between longitudes 15°S and 15°N).

Most Malvacea are not suitable for desertic habitats, but can be successfully introduced in areas of high topographic relief, as long as fields are contoured along topographic lines (using zero-runoff protocols), and where adequate rainfall and/or irrigation water (preferably supplied by flooding or drip irrigation, rather than sprinklers) is available.


Pressurized extraction techniques provide a rapid method to separate a range of both polar and nonpolar natural products from Hibiscus seeds with environmentally benign solvents. The results obtained in this Phase I investigation have demonstrated the technical feasibility for recovery and isolation of Hibiscus (and other Malvaceae species) seed components by the environmentally benign process of supercritical carbon dioxide extraction. The seeds require a minimum of processing prior to extraction (simple mechanical size reduction in a Wiley mill or hammermill food processor), and the extracts obtained are solvent free and suitable for edible products.

The predominant lipid components identified here (e.g., the seed oils and the phospholipids) could be brought to market without difficulty. These natural products would increase the potential economic viability for the introduction of species and hybrids of various Hibiscus species as a cultivated crops. Agronomic issues—such as plant spacing (by genome), water requirements, contouring and row grade protocols for crop production on slopes (i.e., terrace alignments and concomitant row grades), irrigation delivery systems, and fertilizer and pesticide management requirements (preferably with organic protocols for both fertilizers and pesticides)—would require attention to complete a full economic evaluation. Planting and pest management guidelines can and should be developed in a straightforward manner from field trials under both organic and conventional protocols.