Fresh Green Seed Yield and Seed Nutritional Traits of Vegetable

Published November, 2002

Fresh Green Seed Yield and Seed Nutritional Traits of Vegetable Soybean Genotypes M. S. S. Rao,* A. S. Bhagsari, and A. I. Mohamed ABSTRACT

et al., 1971). Fresh or frozen vegetable soybean can be cooked just like sweet pea (Pisum sativum L.) or lima bean (Phaseolus limensis L.), either stir fried or added to stews and soups. They are nutritious and rich in phytochemicals beneficial to humans. Masuda (1991) compared vegetable soybean quality with that of green peas (P. sativum) and reported nearly 56% more protein content for vegetable soybean than that of green peas. The combination of low oil content and the relatively high protein content of fresh green soybean seeds makes them particularly desirable to the health conscious people seeking low fat, high protein snacks (Brar and Carter, 1993). The USA currently imports more than 10 000 Mg of frozen edamame each year and this is estimated to increase to 25 000 Mg by 2005 (Lin, 2001). Soybean with 78 to 220 g isoflavone g⫺1 dried seed weight, depending upon isoflavone type (Mohamed et al., 2001), is one of the few natural sources of isoflavones. Messina (2001) summarized the results of several clinical studies that showed the association of soyfoods, particularly soy isoflavones with reduction in blood serum cholesterol levels, reduction in the risk of cardiovascular diseases in humans, reduction in mammary and prostrate cancers in women and men, respectively, and increased bone density and reduced osteoporosis among menopausal women. Because of these nutraceutical benefits of soybean and the recent approval of soybean protein extract as dietary supplement by the Food and Drug Administration, the demand for soyfoods may continue to increase over the long term. In the USA, specialty soybean is a niche market commodity that fetch a premium ranging from $18 to $589 per Mg above the market price of commodity soybean (Carter and Wilson, 1998). Limited consumer base and lack of suitable cultivars are some of the factors limiting vegetable soybean production in the USA. There is a need to introduce, evaluate, and characterize vegetable soybean cultivars of domestic and Asian origin for cultivation in the USA. Development of improved soybean cultivars for vegetable purpose offers potential for expanding the domestic and international soybean market. In the USA, information on agronomic and nutritional characteristics of vegetable soybean is limited, although there is a long history of trying to promote soybean as a vegetable crop (Carter and Shanmugasundaram, 1993). A chronology of vegetable soybean including its history in the USA has recently been published (Shurtleff and Lumpkin, 2001). A number of large-seeded Japanese, Korean, and Chinese vegetable soybean cultivars were reportedly renamed and released in 1930s and 1940s for cultivation in the USA. A few vegetable soybean cultivars were developed for adapta-

Edamame (pronounced eh-dah-MAH-meh) are large-seeded soybean [Glycine max (L.) Merr.] harvested as green pods at the R6 stage when the seed are approximately 80% matured. The demand for edamame as fresh or frozen vegetable is increasing worldwide. Currently, lack of adapted edamame cultivars is one of the major factors limiting its commercial production in the southeastern USA. A need exists, therefore, to evaluate and identify Asian vegetable soybean genotypes for potential production and/or as a source of vegetable traits for breeding suitable cultivars. In a 4-yr study, six Japanese edamame cultivars, four large-seeded Japanese plant introductions, two Chinese vegetable soybean cultivars, and two adapted U.S. cultivars were evaluated for fresh green pod and seed yields and seed composition at the R6 stage. The genotypes were planted in a randomized complete block with four replications and were harvested at the R6 stage. The mean fresh pod and seed yields were 18.5 and 9.6 Mg ha⫺1, respectively. The PI 181565, ‘Tambagura’, ‘Shangrao Wan Qingsi’, and PI 200506 with fresh pod and seed yields in excess of 20 and 10 Mg ha⫺1, respectively, offer potential for commercial production in Georgia. The seed oil and protein contents ranged from 130.7 to 155.8 and 333.2 to 386.0 g kg⫺1, respectively. The mean glucose content was 67.1 g kg⫺1, whereas the mean phytate content was 12.6 g kg⫺1. Fresh pod weight was the major yield determinant (R2 ⫽ 0.88). The Japanese edamame genotypes may be utilized as a source of vegetable soybean traits in breeding programs.

S

oybean continues to be the major source of protein and edible oil in the world. It yields more protein per hectare than most other crops and accounts for more than 63% of high protein meal and 28% of the total edible oil supply worldwide (Golbitz, 2000). In the USA, soybean is mainly used as a source of protein in livestock and poultry feed, whereas soybean oil is used for human consumption and industrial purposes. Despite a worldwide increase in demand for soybean meal and oil, the commodity soybean price has been declining mainly because of increasing production in South American countries (Golbitz, 2000). Therefore, the identification of soybean for specialty uses may be pivotal to the future of the soybean industry in the USA. Vegetable soybean, called edamame (pronounced eh-dah-MAH-meh) in Japan or Mao dou in China (Shurtleff and Lumpkin, 2001), are large-seeded soybean with a sweet, nutty flavor that can be eaten as snack either boiled in salt water or roasted similarly to peanut (Arachis hypogaea L.) seed. In Asia, where edamame is an important vegetable, farmers harvest fresh green pods along with the stems when the pods are fully filled and just before turning yellow (Shanmugasundaram et al., 1991). This stage corresponds to the R6 stage of soybean development (Fehr

M.S.S. Rao and A.S. Bhagsari, Agric. Res. Stn., Fort Valley State Univ., Fort Valley, GA 31030; A.I. Mohamed, Agric. Res. Stn., Virginia State Univ., Petersburg, VA 23806. Received 11 Oct. 2001. *Corresponding author ([email protected]).

Abbreviations: DAP, days after planting; LAI, leaf are index; PAR, photosynthetically active radiation; MG, maturity group.

Published in Crop Sci. 42:1950–1958 (2002).

1950

RAO ET AL.: VEGETABLE SOYBEAN YIELD AND NUTRITIONAL TRAITS

tion to the USA by crossing Japanese and Korean vegetable soybean cultivars with high yielding, pest and shatter resistant grain soybean cultivars adapted to the U.S. environments. Carter and Shanmugasundaram (1993) reported limited breeding efforts in the USA. A few large-seeded soybean cultivars particularly suited for vegetable purposes have been released by Iowa State University (Bernard, 2001). Although several cultivars through crossing American cultivars with Japanese foodgrade cultivars have been developed for production in the USA, none of these cultivars have become popular because the American consumer has not adapted to the taste and flavor (Carter and Shanmugasundaram, 1993). In the USA, Japanese cultivars belonging to MG I- through III have been reported to be suitable for production in Washington, Oregon (Miles, 2001), and Colorado (Johnson et al., 1999). No such efforts have been reported for identifying edamame cultivars for production in the southeastern USA. While breeding edamame genotypes adapted to the southeastern USA is a long-term possibility, selection of edamame cultivars developed for production in the Asian countries for potential production in the southeastern USA could be a viable option in the short term. The objective of this research was to evaluate 10 Japanese vegetable soybean genotypes, including four plant introductions with desirable vegetable soybean traits, two Chinese vegetable soybean cultivars, and two adapted U.S. soybean cultivars for LAI, light interception, fresh green seed yield, and green seed compositional traits in the southeastern USA. MATERIALS AND METHODS Twelve vegetable soybean genotypes and two elite U.S. commodity-type cultivars were planted 12 May 1995, 27 May 1996, 5 June 1997, and 9 June 1998 in a randomized complete block with four replications on Norfolk loamy sand (fine loamy, kaolinitic, thermic Typic Kandiudelts) soil at the Agricultural Research Station Farm, Fort Valley State University, GA. Each plot consisted of three 6 m-long rows planted 0.75 m apart. Nitrogen fertilizer in the form of ammonium nitrate at 28 kg ha⫺1 and herbicide, trifluralin [2,6-dinitro-N,N-dipropyl-4-(trifluoromethyl)benzenamine] at 1.8 L ha⫺1 were soilincorporated prior to planting each year. Insect control was achieved through three applications of Ambush [Syngenta Crop Protection, Inc., Greenboro, NC (3-phenoxyphenyl) methyl(⫹/⫺)cis/trans-3-(2,2-dichloroethenyl)-2,2-dimethylcyclopropanecarboxylate], at 585 mL ha⫺1 and a single application of carbaryl [(1-naphthyl N-methylcarbamate)] and acephate (O,S-dimethyl acetylphosphoroamidothioate) each at 1.12 kg ha⫺1. Irrigation was applied by overhead sprinkler system throughout growing season as needed.

Flowering Date of first flower was recorded each year. Thirty days after planting (DAP), the crop was checked at 2-d interval to record flowering.

Interception of Photosynthetically Active Radiation (PAR) The interception of PAR was recorded with a Sunfleck PAR Ceptometer (Decagon Devices, Inc., Pullman, WA). In

1951

each plot, the total PAR incident on crop canopy and that reaching the base of the crop were recorded. The difference between the incident PAR and transmitted PAR was calculated as that intercepted by the crop canopy. Two observations were recorded for each plot and the average of the two expressed as percentage is reported as PAR intercepted by the crop canopy.

Leaf Area Index (LAI) and above Ground Biomass Plants from 0.5-m row were harvested from each plot for estimating LAI and above ground biomass at the R1 stage (Fehr et al., 1971), when the crop was at peak vegetative to early flowering phase. The leaf area was determined with leaf area meter (LI-COR 3100, LI-COR Inc., Lincoln, NE) and LAI was calculated as the ratio of leaf area to land area.

Fresh Green Seed Yield and Yield Components at the R6 Stage Each year, plants from 0.5-m rows were sampled from each plot when the crop was at the R6 stage [when the pods are fully developed but still green and immature with seeds still green and about 80% matured (Fehr et al., 1971)] to determine the green seed yield and yield components. The harvested plants were separated into stems, leaves, and pods. Fresh weight of the pods with seeds were recorded. The pods were then shelled to determine shell and green seed fresh weights. The fresh weight of 100 green seeds was recorded and number of seeds per pod and per per square meter were computed. All seed yield data were expressed on fresh weight basis.

Determination of Oil, Protein, Glucose, and Phytate Fresh green seeds harvested at the R6 stage were packed in plastic bags and shipped overnight in frozen condition to Virginia State University for the determination of green seed compositional traits. Moisture content was determined by drying the seeds at 105 to 110⬚C in an air oven (Fisher Isothermal Oven model 350) to a constant weight and then percent moisture content was calculated. Five grams of fresh green seeds were homogenized in a mixture of hexane and isopropanol (3:2 v/v) to extract oils at room temperature as described previously (Mohamed et al., 1995). After removal of the solvent under nitrogen, the oil was weighed. The remaining defatted meal was placed in the oven at 60⬚C overnight to remove excess solvents, then the dried meal was used to determine total concentrations of protein, glucose, and phytate. A ¨ CHI 430 digestor was used for the digestion process using BU H2O2/H2SO4 (4:1 v/v). Total N2 content was determined by means of the Nessler reagent (AOAC, 1984). The protein content was calculated by multiplying the total nitrogen by a factor of 6.25. Total phytate was extracted according to the procedures reported by Mohamed et al. (1986). The phytate was then purified on an anion exchange resin (Dowex - 1⫻ 8 Cl⫺ ) to remove inorganic phosphorus, di-, tri-, and tetraphosphate inositol. Total soluble sugar from the meal was extracted with deionized H2O. Protein was eliminated from the extract by the addition of acetonitrile to a final concentration of 50%, followed by centrifugation. Then, acetonitrile was eliminated and concentrated samples were diluted and used for sugar analysis by the phenol sulfuric acid assay (Dubois et al., 1956) and absorbance measured at 490 nm. Glucose is the second major carbohydrate (Tsou and Hong, 1991) and is a good indicator of sweetness of vegetable soybean. Therefore, glucose was used as a standard. All seed compositional traits have been expressed on dry weight basis.

1952

CROP SCIENCE, VOL. 42, NOVEMBER–DECEMBER 2002

Table 1. Plant introduction numbers†, country of origin, pedigree, seed dry weight, and maturity group of soybean genotypes used in this study. Cultivar–line PI 181565 Shangrao Wan Qingsi Tambagura§ Akiyoshi PI 200506 Houjaku Guanyun Da Hei Dun Hutcheson¶ PI 416981 Tomahomare§ PI 417427 Tousan - 122 Ware¶ Mian Yan

PI number

PI 561382 PI 187154 PI 561383 PI 594182 PI 561378 PI 518664 PI 561391 PI 561397 PI 548627 PI 561339

Pedigree/country of origin Collected from Japan Collected from Japan Collected from Japan Collected from Japan Collected from Japan Collected from Japan Collected from China USA V68-1034 (York ⫻ PI 71506) ⫻ Essex. Collected from Japan Collected from Japan Collected from Japan Collected from Japan USA PI 80837 ⫻ V63-76 (Hill ⫻ D53-354); D53-354 is form D49-2525 (S-100 ⫻ CNS) ⫻ L46-5679 (Lincoln ⫻ Richland). Collected from China

Seed dry weight‡

Maturity group†

g 100⫺1 22.0 23.5 46.5 26.3 24.0 24.8 26.1 16.4 21.6 26.1 22.0 22.3 19.1

VII V VII V VII VI V V V V VI V IV

21.7

III

† Information source: http://www.ars-grin.gov. ‡ Based on 4-yr data at FVSU (Rao et al., 2002. Unpublished). § Information source: Dr. R. Nelson, USDA/ARS, Urbana, IL. ¶ Commodity grain soybean; Information source: Dr. G.R. Buss, VA Polytech. Inst., Blacksburg, VA.

Statistical Analysis The data were subjected to statistical analysis with SAS software [SAS, 2000]. PROC MIXED analysis was carried out using Year and Year ⫻ Cultivar as random effects. Year ⫻ Cultivar was used to test differences between cultivars. The important yield contributing components were determined by stepwise regression analysis.

RESULTS AND DISCUSSION Genotype Selection All genotypes, except Hutcheson, were selected on the basis of large-seed dry weight and or protein content. All six Japanese cultivars and two Chinese cultivars used in this study were developed for vegetable use in Japan and China, respectively. The plant introductions, all collected from Japan (Table 1), were selected for their seed size larger than that of commodity grain soybean (seed dry weight 苲 150 mg seed⫺1 ). Among control cultivars, Ware is a large-seeded cultivar adapted to southeastern USA and Hutcheson is an elite U.S. cultivar belonging to MG V. It produced consistently high yields across four locations in the southeastern USA in a concurrent study (Rao et al., 2002).

Flowering The onset of flowering among the 14 genotypes ranged from 37 (Mian Yan) to 55 (PI 181565) DAP (Table 2). Mian Yan and Ware flowered within 40 DAP, whereas PI 181565, Shangrao Wan Qingsi, Tambagura, Akiyoshi, PI 200506, and Houjaku flowered between 50 and 55 DAP. Remaining genotypes flowered between 40 and 49 DAP. In the southeastern USA, soybean genotypes belonging to MGs VI, VII, and VIII generally produce higher yields than cultivars of lower maturity groups (Rao and Bhagsari, 1998) and are preferred for planting as summer crop (Boerma and Mian, 1998). Thus, on the basis of maturity group, all geno-

types, including the Japanese cultivars, may be suitable for production in Georgia.

Leaf Area Index The LAI ranged from 4.1 for PI 416981 to 5.5 for Shangrao Wan Qingsi (Table 2). Thus, all genotypes produced LAI greater than 4.1 considered adequate for full interception of incident PAR by the crop canopy and maximum dry matter production (Sakamoto and Shaw, 1967). PI 416981 produced significantly lower LAI than Shangrao Wan Qingsi. The differences between all other genotypes were not significant. The Japanese cultivars produced a mean LAI of nearly 5.0 whereas the American cultivars had a mean LAI of 4.6. The Japanese genotypes, particularly Tambagura, Tomahomare, and Akiyoshi produced larger leaves than Table 2. Days to flowering, leaf area index, intercepted photosynthetically active radiation (PAR), and biomass† at R1 stage of soybean genotypes, 1996–1998. Genotype PI 181565 Shangrao Wan Qingsi Tambagura Akiyoshi PI 200506 Houjaku Guanyun Da Hei Dun Hutcheson PI 416981 Tomahomare PI 417427 Tousan 122 Ware Mian Yan LSD (P ⫽ 0.05) Mean

Flowering

Leaf area index

Intercepted PAR

Biomass

DAP‡ 55 54 54 53 53 50 49 48 47 46 46 43 40 37 8 48

Ratio 4.7 5.5 5.1 4.9 4.3 5.2 5.2 5.1 4.1 4.8 4.6 4.3 4.6 4.4 1.3 4.8

% 83 89 86 85 87 90 83 87 82 86 87 85 83 86 NS 89

g m ⫺2 322 362 370 385 316 347 357 332 287 338 318 305 325 308 NS 337

† Biomass excluding roots. R1 stage: peak vegetative to early flowering (Fehr et al., 1971). ‡ Days after planting. NS ⫽ Not significant.

1953


the U.S. cultivars perhaps because of their relatively longer vegetative phase and their larger seed size.

Interception of Photosynthetically Active Radiation The interception of PAR by the 14 genotypes varied between 82 and 90% of that incident upon the crop at R1 stage (Fehr et al., 1971) (Table 2). All genotypes intercepted similar levels of PAR. Although the genotypes produced an average LAI of 4.8, the interception of PAR was less than 90% of that incident upon the canopy. The Japanese cultivars, were observed to have planophyll leaf architecture, which usually results in poor interception as the larger horizontally disposed leaves restrict light penetration to lower layers of the crop canopy.

Biomass The mean biomass at R1 stage varied between 287 g m⫺2 for PI 416981 and 385 g m⫺2 for Akiyoshi on dry weight basis (Table 2). The PI 416981, which had a lower LAI and percent intercepted PAR also produced smaller amount of biomass than Akiyoshi, Tambagura, and Shangrao Wan Qingsi. The Japanese genotypes, which produced higher LAI, also produced more biomass than the U.S. cultivars. Generally, later flowering cultivars also produce greater biomass than early flowering cultivars, but in this study, such a relationship was not clear.

Fresh Green Seed Yield and Yield Components at R6 Stage

nyun Da Hei Dun, and Hutcheson. The genotypes that achieved the R6 stage within 120 DAP may be categorized as early. PI 417427, Tomahomare, and Houjaku achieved the R6 stage significantly earlier than Tambagura but significantly later than Mian Yan. These genotypes which took about 120 to 121 d to reach R6 stage could be categorized as medium range, whereas remaining genotypes, Akiyoshi, Tambagura, Shangrao Wan Qingsi, PI 181565, and PI 200506 which attained the R6 stage after 124 DAP may be classified as late under Georgia conditions. Genotypes Mian Yan and Ware may be better suited for planting early in the season as a short season cash crop. Planting early and late maturing genotypes in a sequence will enable the farmer to market fresh vegetable soybean over a longer duration. The time of harvesting is a critical factor in determining consumer acceptability and marketability of fresh vegetable soybean (Mbuvi and Litchfield, 1995). The optimum time for harvesting fresh vegetable soybean to combine the best product quality with maximum yield is a function of a dynamic relationship between maturity, yield, and quality parameters. Quality properties such as color, texture, and seed size of vegetable soybeans are a function of development time (Mbuvi and Litchfield, 1995). Since these quality parameters do not peak at the same time, it is necessary to compromise time of harvest of green beans. Shanmugasundaram et al. (1991) reported that the optimum time for harvesting green beans was when the pods are still green, immature, and tight with fully developed immature green seeds. This stage coincides with the R6 stage of soybean development as staged by Fehr et al. (1971). Thus, harvest at R6 stage is very critical for ensuring bean yield and quality.

Number of Fresh Green Pods

Number of Days to R6 Stage There were significant genotypic differences for days to achieve R6 stage, when the green pods could be harvested (Table 3). The average number of days from planting to R6 stage ranged from 99 (Mian Yan) to 134 (Tambagura). The two early flowering cultivars, Mian Yan and Ware, also achieved R6 stage significantly earlier than all other genotypes, except Tousan-122, Gua-

The mean number of green pods across years ranged from 890 for Tambagura to 1552 for PI 181565 (Table 3). PI 181565, Hutcheson, and Shangrao Wan Qingsi produced significantly greater number of pods than all other genotypes. Tambagura, Tousan - 122, and Guanyun Da Hei Dun produced fewer pods than most other genotypes.

Table 3. Fresh green seed yield and yield components of soybean genotypes at the R6† stage, 1995–1998. Genotype PI 181565 Shangrao Wan Qingsi Tambagura Akiyoshi PI 200506 Houjaku Guanyun Da Hei Dun Hutcheson PI 416981 Tomahomare PI 417427 Tousan - 122 Ware Mian Yan LSD (P ⫽ 0.05) Mean

R6 Stage DAP‡ 129 128 134 124 132 121 108 112 117 121 120 114 104 99 12 118

Fresh green pods No. m⫺2 1552 1451 890 1179 1108 1046 983 1432 1178 1366 1049 975 1112 1244 392 1183

Mg ha⫺1 21.7 21.2 22.0 19.7 18.2 17.6 17.3 17.9 16.0 21.5 16.0 16.3 14.6 18.7 5.8 18.5

Fresh green seeds No. Pod⫺1 1.8 1.6 1.3 1.6 1.6 1.6 1.5 2.2 1.8 1.5 1.6 1.8 1.7 1.6 0.3 1.7

† R6 stage: Pods are developed but seeds are immature and about 80% matured (Fehr et al., 1971). ‡ Days after planting.

No. m⫺2 2604 2290 1193 1871 1859 1706 1532 3173 2162 1994 1807 1720 1826 1834 653 1969

Green seed fresh weight

Green seed yield

mg seed⫺1 420 513 948 566 508 50.4 575 315 441 582 494 498 396 471 95 517

Mg ha⫺1 11.1 11.5 11.1 9.9 9.5 8.7 8.9 10.0 8.6 11.6 8.9 8.6 7.3 9.0 3.0 9.6

1954


Fresh Green Pod Yield The fresh green pod yield varied between 14.6 Mg ha⫺1 and 21.7 Mg ha⫺1 and the differences among the genotypes were significant (Table 3). Tambagura had the highest mean pod yield among the genotypes whereas Ware produced the lowest. The average green pod yield across genotypes was 18.5 Mg ha⫺1. The green pod yield of Tambagura, PI 181565, Tomahomare, and Shangrao Wan Qingsi was in excess of 20 Mg ha⫺1 and differed significantly from that of Ware. The fresh green pod yield of the rest of the genotypes did not differ significantly. PI 181565, Shangrao Wan Qingsi, and Tomahomare which produced more pods than most other genotypes also had high pod yields. Tambagura produced significantly higher pod yield than PI 416981, PI 417427, and Ware but similar to all other genotypes. Although Hutcheson also produced a similar number of pods as the higher yielding genotypes, its pod yield was considerably lower perhaps because of its smaller seeds and earlier maturity. On the other hand, Tambagura produced high pod yield, although it had significantly fewer pods than PI 181565, Shangrao Wan Qingsi, Hutcheson, and Tomahomare because of its heavier seeds. The pod yields of genotypes tested in this study were considerably higher than those reported for vegetable soybean breeding lines in Taiwan (Shanmugasundaram et al., 1991). In this study, the mean pod yield ranged from 15 to 22 Mg ha⫺1 compared with 10 to 13, 6 to 9, and 6 to 10 Mg ha⫺1 during spring, summer, and autumn seasons, respectively, in Taiwan. This could be due to the planting of Group V cultivars in the tropics (Taiwan), which normally results in low plant height and low yields. For example, edamame cultivar Blueside grown in Taiwan is of MG V (T.E. Carter, Jr. 2001. Personal communication). The vegetable soybean improvement program at AVRDC has reportedly increased potential pod yields of some Taiwanese edamame varieties to about 24 Mg ha⫺1 (Shanmugasundaram and Yan, 1999). Konovsky et al. (1996) evaluated 36 edamame genotypes composed of 32 Japanese, three U.S., and 1 Taiwanese genotypes for heritability of yield and quality traits in Washington and reported gross yields ranging from 11.2 to 13.6 Mg ha⫺1 and net yields of around 7.2 to 8.4 Mg ha⫺1.

Number of Fresh Green Seeds Tambagura had the lowest number of green seeds per pod, whereas Hutcheson had significantly more seeds per pod than all other genotypes (Table 3). Tambagura retained significantly fewer seeds per pod than did Ware, Tousan 122, PI 181565, PI 416981, and Hutcheson. The number of seeds per pod is one of the important quality characteristics that determine the marketability and profitability of edamame. Pods with more than two seeds are generally preferred and fetch premium prices in the Asian markets (Shanmugasundaram et al., 1991). In this study, only Hutcheson retained more than two seeds per pod. The PI 181565 had significantly greater number of seeds per pod than Tomaho-

mare, Guanyun Da Hei Dun, and Tambagura but was similar to the rest. A similar number of seeds per pod was reported for several vegetable soybean genotypes grown in Virginia (Mebrahtu et al., 1997) and Washington (Konovsky et al., 1996). The number of seeds per pod and seed weight are generally negatively related as they compete for the same resources. A compensatory mechanism between number of seeds per pod and seed weight may have been operative in the present study since Hutcheson with smaller, lighter seeds could retain more seeds per pod than vegetable soybean genotypes which produced heavier seeds. The number of seeds per pod is one of the yield determinant of soybean. Shanmugasundaram et al. (1991) reported a significant linear relationship between seed fresh weight and pod length (r 2 ⫽ 0.674) and pod width (r 2 ⫽ 0.689) suggesting that these two physical characteristics could be useful selection criteria for breeding vegetable soybeans with more seeds per pod. The number of seeds per unit area is a function of number of pods per unit area and is an important yield determinant. Hutcheson produced a significantly greater number of seeds than all other genotypes except PI 181565 (Table 3). Tambagura which produced fewer pods and fewer number of seeds per pod also had fewer number of seeds per square meter than half of the tested genotypes. PI 181565 produced number of seeds per square meter similar to Shangrao Wan Qingsi, PI 416981, and Tomahomare, but significantly more than the rest of the genotypes.

Fresh Green Seed Weight The mean green seed fresh weight ranged from 315 mg seed⫺1 for Hutcheson to 948 mg seed⫺1 for Tambagura (Table 3). Tambagura, Tomahomare, Akiyoshi, Shangrao Wan Qingsi, Guanyun Da Hei Dun, PI 200506, and Houjaku had a mean seed fresh weight above 500 mg seed⫺1. The green seed fresh weight of Tambagura was three times greater than that of Hutcheson and it was twice that of most other genotypes. Although, Tambagura produced fewer pods and fewer seeds per pod, it had higher pod yield because of heavier seeds. On the other hand, Hutcheson had a greater number of pods and seeds per pod, but its pod yield was not as high as many other genotypes because of smaller seeds. Seed fresh weight is an important yield determinant and quality parameter that determines consumer acceptability (Shanmugasundaram et al., 1991; Mbuvi and Litchfield, 1995). Generally, seed quality characteristics achieve their peak levels when the seed size is also at its maximum. The mean fresh seed weight of the genotypes tested in this study with the exception of Hutcheson and Ware, was higher than that reported for vegetable soybean cultivars in Taiwan (Chen et al., 1991). However, Shanmugasundaram et al. (1991) reported higher seed fresh weights for some vegetable soybean breeding lines in Taiwan.

Green Seed Yield All genotypes produced high fresh green seed yield which ranged from 7.3 for Ware to 11.6 Mg ha⫺1 for

1955


Tomahomare (Table 3). Ware produced a significantly lower green seed yield than Tomahomare, Shangrao Wan Qingsi, PI 181565, and Tambagura, but was similar to the rest of the genotypes. Tomahomare, PI 181565, Tambagura, Shangrao Wan Qingsi, and Hutcheson had mean green seed yield of above 10 Mg ha⫺1. The overall mean seed yield across genotypes was 9.6 Mg ha⫺1. Variations in the mean green seed yield between genotypes could be attributed to variation among yield components. Hutcheson produced greater number of fresh green pods and seeds m⫺2 which contributed to its high fresh green seed yield. In Tambagura it was mainly the seed fresh weight that resulted in higher fresh pod weight and green seed yields. In Guanyun Da Hei Dun, Tomahomare, and Shangrao Wan Qingsi it was a combination of number of green pods and seeds per square meter and green seed fresh weight that was responsible for higher green seed yield than many of the genotypes. Ware, Tousan-122, and Mian Yan, which had relatively fewer pods and seeds, and lighter seeds also had lower green seed yields.

Seed Compositional Traits The oil content ranged from 130.7 to 155.8 g kg⫺1 on dry weight basis (Table 4). The PI 416981, Akiyoshi, and Hutcheson had a significantly greater oil content than Tomahomare, Houjaku, Tambagura, and Shangrao Wan Qingsi. The mean oil content was about 29% lower than that (20 g kg⫺1 ) of mature commodity soybean seed (Burton, 1990). The oil contents of the genotypes in this study were lower than those reported for selected Japanese soyfood cultivars (Brar and Carter, 1993). The seed protein content varied between 333.2 and 386.0 g kg⫺1 on dry weight basis (Table 4). The protein content of Akiyoshi was significantly higher than Tousan-122 and Houjaku. Tousan - 122 had a significantly lower protein content than most other genotypes except Hutcheson, PI 200506, and Akiyoshi. The Japanese cultivars appeared to have a slightly higher protein and a lower oil content than some of the PI lines and the two U.S. cultivars. The protein content of matured soybean seed could range from 360 to 500 g kg⫺1 (Clarke and Wiseman, 2000) with a mean of about 415 g kg⫺1 on

dry weight basis (Burton, 1990). In this study, the mean protein content of the fresh immature seeds was 360.4 g kg⫺1, which was more than 86% of that of matured soybean seed. The glucose content ranged from 60.3 to 74.0 g kg⫺1 (Table 4). Tsou and Hong (1991) reported that the sugar content of vegetable soybean in Taiwan ranged from 7.35 to13.71 mg g⫺1. Hutcheson had a significantly higher glucose content than Guanyun Da Hei Dun, PI 181565, Akiyoshi, PI 417427, Tomahomare, and Mian Yan. Whereas, Guanyun Da Hei Dun had a significantly lower glucose content than Tousan - 122, Ware, PI 416981, Hutcheson, Tambagura, and Shangrao Wan Qingsi. Although Hutcheson had a higher seed glucose content than many genotypes, it does not qualify as a vegetable soybean since it has small seeds and does not possess pod and seed physical traits that characterize vegetable soybean (Shanmugasundaram, 1991). The total soluble sugar content of the fresh green beans is an important nutritional trait that directly influences the organoleptic properties of edamame and determines consumer acceptability. The sugar content of soybean is considered to be at its peak level at the R6 stage (Masuda, 1991). Phytate, calcium–magnesium–potassium salt of inositol hexaphosphoric acid, commonly known as phytic acid occurs in certain cereal and legume seeds (Reddy et al., 1982), including soybean (Mebrahtu et al., 1997). Phytate is the main source of phosphorus in soybean seed and is known to form complexes with phosphorus, proteins, and minerals such as Ca, Mg, Zn, and Fe (Reddy et al., 1982). This reduces the bioavailability of these minerals, affect seed germination and seedling growth, and cause deficiencies in nonruminant animals. In this study, the mean phytate content was 12.6 g kg⫺1 and ranged from 10.8 to 13.9 g kg⫺1 (Table 4). Tambagura and Haujaku had relatively lower phytate content than Akiyoshi, PI 200506, Guanyun Da Hei Dun, Tomahomare, and Mian Yan. The rest of the genotypes did not differ significantly. Phyic acid content of certain cereals and legumes has been reported to vary between 1.4 and 20.5 g kg⫺1 (Reddy et al., 1982). The phytate content of the genotypes studied here were considerably lower

Table 4. Fresh green seed† oil, protein, sugar, phytate, and moisture content of soybean genotypes at the R6 stage, 1995–1998. Genotype

Oil

Protein

Glucose

Phytate

Moisture

11.4 12.6 10.8 13.3 13.8 10.8 13.9 12.5 12.3 13.2 12.8 11.9 12.9 13.5 02.4 12.6

543 548 561 560 548 542 539 534 559 556 551 555 556 549 50 550

kg⫺1

PI 181565 Shangrao Wan Qingsi Tambagura Akiyoshi PI 200506 Houjaku Guanyun Da Hei Dun Hutcheson PI 416981 Tomahomare PI 417427 Tousan - 122 Ware Mian Yan LSD (P ⫽ 0.05) Mean

140.6 134.1 132.1 155.3 144.9 131.6 143.4 155.4 141.1 130.7 155.8 147.1 141.3 139.9 17.2 142.4

353.5 363.9 359.3 386.0 371.3 347.8 362.5 370.1 354.0 361.4 354.5 333.2 361.6 366.2 35.4 360.4

g 64.6 70.0 71.8 64.7 65.9 65.4 60.3 74.0 69.5 64.0 64.8 70.0 70.3 64.4 09.1 67.1

† R6 stage: Pods are developed but seeds are immature and about 80% matured (Fehr et al., 1971).

1956


Table 5. Coefficients of correlation between various parameters of soybean genotypes, 1995–1998.

Parameter† At the R1 stage LAI vs. PAR LAI vs. Biomass at flowering At the R6 stage Green seed yield vs. No. of green pods Green seed yield vs. No. of green seeds m⫺2 Green seed yield vs. Green pod weight Green seed weight vs. Days to R6 stage Oil vs. Protein Oil vs. Glucose Oil vs. Phytate Protein vs. Glucose Protein vs. Phytate

Coefficient of correlation (r ) 0.49*** 0.68*** 0.84*** 0.74*** 0.94*** 0.38** ⫺0.40*** ⫺0.62*** ⫺0.37*** ⫺0.48** ⫺0.24***

** Indicates significance at P ⫽ 0.01. *** Indicates significance at P ⫽ 0.001. † R1 stage: peak vegetative to early flowering (Fehr et al., 1971); R6 stage: Pods are developed but seeds are immature and about 80% developed, (Fehr et al., 1971).

than those reported for several vegetable soybean genotypes harvested at R6 stage in Virginia (Mebrahtu et al., 1997). The moisture content of fresh green seeds ranged from 539 to 561 g kg⫺1, but the differences between genotypes were not significant (Table 4). Seed moisture content is another critical factor that affects time of harvest and is an integral part of organoleptic characteristics of vegetable soybean (Mbuvi and Litchfield, 1995).

Relationships among Yield Contributing Parameters and Seed Yield R1 Stage The intercepted PAR was significantly correlated with leaf area index (Table 5). Biomass at the R1 stage was significantly correlated with LAI. Thus, LAI was critical for both interception of PAR and biomass production. Energy absorption into the plant system is in part dependant upon the total LAI and in part on the interception of PAR by the leaf canopy. In soybean, it has been shown that the maximum photosynthetic rate is partly dependant upon the distribution of radiation within the canopy. Thus, canopy structure is critical for both interception of PAR and its absorption by the plant for photosynthesis. Most of the interception of PAR by soybean canopy occurs at the top leaf surface (Sakamoto and Shaw, 1967) because of planophyll orientation of leaf architecture of soybean. Intercepted PAR, at least, during the pre-anthesis phase has been shown to be linearly related to LAI and dry matter accumulation in soybean (Shibles and Weber, 1965). In soybean, maximal interception of PAR has been reported to occur at a LAI of 3.0 and maximal dry matter production as the LAI approaches 4.0. In the present study, most of the genotypes achieved a mean LAI of 4.0 or greater which may have been adequate for maximal interception of PAR and higher levels of photosynthesis. This may have been the reason for high biomass at final harvest.

R6 Stage The green seed yield at R6 stage was significantly correlated with number of green pods and seeds (Table 5). Board et al. (1996) reported that seed yield at maturity of late planted soybean was more strongly correlated with number of pods than with seeds per pod or seed size. The fresh green seed yield showed a greater correlation with pod yield than with number of pods and seeds, perhaps, because pod yield is the product of number of pods and seeds per pod. The fresh green seed weight showed a positive correlation with number of days to R6 stage. The longer duration to attain R6 stage helped seed development resulting in heavier seeds. Thus Japanese cultivars Tambagura, Shangrao Wan Qingsi, Akiyoshi, PI 181565, and PI 200506 which took longer time to attain R6 stage also had heavier seeds. Among seed compositional traits, oil and protein contents were negatively correlated (Table 5). Vegetable soybean seed oil and protein content have been shown to be negatively correlated. Oil and protein were both negatively correlated with glucose and phytate contents. Shanmugasundaram et al. (2001) reported similar negative correlations based on data from 20 trials carried out during different seasons between 1996 and 1999 in Taiwan. At R6 stage the seed sucrose contents are high and oil content low (Masuda and Harada, 2000).

Fresh Green Seed Yield Determining Components The differences in maturity groups may have been an important factor influencing fresh green pod and seed yields. Stepwise regression analysis of the yield components, excluding maturity group, indicated that at the R6 stage, fresh pod weight was the major determinant of yield with an R2 value of 0.88 followed by number of seeds per square meter, 100-seed fresh weight, and seeds per pod in that order of importance (Table 6). Path analysis of data from several experiments comprising a range of agronomic treatments showed seed number as the most important yield determinant of soybean grown in southeastern USA (Board et al., 1997).

CONCLUSIONS In summary, most of the Japanese edamame cultivars and some of the PI lines tested produced high LAI of above 4.0, but intercepted only 89% of the PAR incident upon the crop canopy, perhaps because of poor plant architecture (planophyll type) and possibly wide row spacing. Most of the Japanese cultivars flowered later and produced higher LAI than the two U.S. genotypes used in this study. The U.S. cultivars Ware and Hutcheson were of maturity groups IV and V, respectively, and, therefore, flowered earlier than the Japanese cultivars before achieving full canopy closure. The genotypes tested in this study produced a mean fresh green pod yield of 18.5 Mg ha⫺1 and fresh green seed yield of 9.6 Mg ha⫺1. PI 181565, Tomahomare, Shangrao Wan Qingsi, and Tambagura produced fresh green pod yield


1957

Table 6. Seed yield determining components at the R6† stage based on stepwise regression analysis. Source Intercept Green pod weight Seeds m⫺2 100-Green seed wt. Seed per pod

Partial slope

Standard Error

Cumulative R2

Probability ⬎ F

955.40321 0.27537 0.22514 7.27645 38.11610

6.00585 0.01978 0.01820 0.64812 19.59368

0.8777 0.9014 0.9374 0.9388

⬍0.0001 ⬍0.0001 ⬍0.0001 0.0530

† R6 stage: Pods are developed but seeds are immature and about 80% matured (Fehr et al., 1971).

of more than 20 Mg ha⫺1 and fresh green seed yield greater than 10 Mg ha⫺1. The Japanese cultivars and PIs gave pod and seed yields similar to or better than the adapted cultivar, Hutcheson, indicating their possible adaptation to Georgia and the southeastern USA. Tambagura had the largest seed on fresh weight basis and this component was probably the key to high pod and seed yields of this cultivar. Genotypes that produced more pods and seeds also produced high pod and seed yields. The protein and oil contents of fresh vegetable seed were about 86 and 33% of that of matured soybean seed, respectively. This study has not only helped identify several potential high yielding vegetable soybean cultivars for production to cater to the needs of soyfood industry, it also provides valuable information that could be used for further improvement of soybean for food uses through classical breeding combined with modern molecular biological approaches as suggested for the improvement of food-grade soybean (Boerma and Mian, 1998). Poor germination, uneven plant stands, and a greater susceptibility to stink bugs [Nezara viridula (L) and Euschitus servus (Say)], were some of the problems observed in this study. Poor germination and lack of uniform plant stand could be minimized by planting seeds into moist seed beds as opposed to the application of irrigation after planting. Edamame harvesting is time specific, laborious, and time consuming. Currently, edamame production in the USA is confined to small-hectarages mainly aimed at niche markets because of lack of harvesting machinery for large scale commercial production combined with the need for specific handling and storage requirements to preserve fresh bean quality and flavor. Research focused on making this crop amenable to canning could help increase its production in the USA. Research aimed at identification of off-flavor causing phytochemicals and improving flavor by either eliminating the causative factors through conventional and/or molecular marker assisted breeding or through improved processing and storage procedures is in progress in Asia (Kitamura, 1995). There is a need for similar research in the USA for developing nutritious, high yielding vegetable soybean cultivars with taste and flavor acceptable to the U.S. consumer. ACKNOWLEDGMENTS This work was part of a Regional Soybean Research Project, “Improvement of soybean for food uses” sponsored by the Association of Research Directors of 1890 Institutions with funding from USDA/CSREES. The authors acknowledge all members of the project for any indirect contribution they may have made towards this work. The authors are thankful to Drs. T.E. Carter, Jr., USDA/ARS, Raleigh, NC; G.R Buss, Virginia Tech, Blacksburg, and J.M. Joshi, Univ. of Maryland

Eastern Shore, Princess Anne, MD, for providing the seed of genotypes used in this study. The authors also thank Mr. B. Mullinix for assistance with statistical analysis and Mr. Z. Blake for assistance with field work.

REFERENCES AOAC. 1984. Official methods of analysis, 13th ed. p. 14–15. Association of Official Analytical Chemists, Washington DC. Bernard, L.R. 2001. Breeding vegetable soybeans in the Midwest. p. 21. In T.A. Lumpkin and S. Shanmugasundaram (Compilers) Int. Vegetable Soybean Conference, 2nd, Tacoma, WA. 10–12 Aug. 2001. Washington State Univ., Pullman, WA. Board, J.E., W. Zhang, and B.G. Harville. 1996. Yield rankings for soybean cultivars grown in narrow and wide rows with late planting dates. Agron. J. 88:240–245. Board, J.E., M.S. Kang, and B.G. Harville. 1997. Path analysis identify indirect selection criteria for yield of late-planted soybean. Crop Sci. 37:879–884. Boerma, R., and R.M.A. Mian. 1998. Application of DNA markers for selection of intractable soybean traits. Korea Soybean Digest 15(2):106–121. Brar, G.S., and T.E. Carter, Jr. 1993. Soybean Glycine max (L.) Merril. p. 427–763. In G. Kalloo and B.O. Bergh (ed.) Genetic improvement of vegetable crops. Pergamon Press, New York. Burton, J.W. 1990. Development of high-yielding high protein soybean germplasm. p. 109–117. In R.F. Wilson (ed.) Designing value-added soybeans for markets of the future. American Oil Chemists’ Society, Baltimore, MD. Carter, T.E., Jr., and S. Shanmugasundaram. 1993. Vegetable soybean (Glycine). p. 219–239. In J.T. Williams (ed.) Pulses and vegetables. Chapman and Hall, New York. Carter, T.E., Jr., and R.F. Wilson. 1998. Soybean quality for human consumption. p. 1–16. In A. James (ed.) Soybeans role in Australia. Proc. Australian soybean conference, 10th. 15–17 Sept. 1998. CSIRO Tropical Agriculture, St. Lucia, Australia. Chen, K-F., S.H. Lai, and Chen, S-T. 1991. Vegetable soybean seed production technology in Taiwan. p. 45–52. In S. Shanmugasundaram (ed.) Vegetable soybean: research needs for production and quality improvement. Asian Vegetable Research and Development Center, Taiwan. Clarke, E.J., and J. Wiseman. 2000. Developments in plant breeding for improved nutritional quality of soya beans I. Protein and amino acid content. J. Agric. Sci. (Cambridge) 134:111–124. Dubois, M., K.A. Gilles, J. K. Hamilton, P.A. Roberts, and F. Smith. 1956. Colorimetric method for determination of sugars and related substances. Anal. Chem. 28(3):351–356. Fehr, W.R., C.E. Caviness, D.T. Burmood, and J.S. Pennington. 1971. Stage of development descriptions for soybeans, Glycine max (L.) Merrill. Crop Sci. 11:929–931. Golbitz, P. 2000. Soya bluebook plus. p. 340–341, 365. Soyatech Publications, Bar Harbor, ME. Johnson, D., S. Wang, and A. Suzuki. 1999. Edamame: A vegetable soybean for Colorado. p. 385–387. In J. Jules (ed.) Perspectives on new crops and new uses. ASHS Press, Alexandria, VA. Kitamura, K. 1995. Genetic improvement of nutritional and food processing quality in soybean. JARQ 29:1–8. Konovsky, J., D.W. Evans, and T.A. Lumpkin. 1996. Heritability of yield, plant architecture, and quality traits of edamame: the vegetable soybean. Soybean Genet. Newsl. 23:243–249. Lin, C-C. 2001. Frozen edamame: Global market conditions. p. 93–96. In T.A. Lumpkin and S. Shanmugasundaram (Compilers) Int. Vegetable Soybean Conference, 2nd, Tacoma, WA. 10–12 Aug. 2001. Washington State Univ., Pullman, WA.

1958


Masuda, R. 1991. Quality requirement and improvement of vegetable soybean. p. 92–102. In S. Shanmugasundaram (ed.) Vegetable soybean: Research needs for production and quality improvement. Asian Vegetable Research and Development Center, Taiwan. Masuda, R., and K. Harada. 2000. Carbohydrate accumulation in developing soybean seeds- sucrose and starch levels in 30 cultivars for soyfoods. p. 67–68. In The Third International Soybean Processing and Utilization Conferences. 15–20 Oct. 2000. Japanese Food Science and Technology, Tsukuba, Japan. Mbuvi, S.W., and J.B. Litchfield. 1995. Green soybeans as vegetable: Comparing green soybeans with green peas and lima beans; and maximized harvest time determinations using mathematical modeling. J. Veg. Crop. Prod. 1:99–121. Mebrahtu, T., A. Mohamed, and A. Elmi. 1997. Accumulation of phytate in vegetable-type soybean genotypes harvested at four developmental stages. Plant Foods Human Nutr. 50:179–187. Messina, M. 2001. An overview of the health effects of soyfoods and soybean isoflavones. p. 117–122. In T.A. Lumpkin and S. Shanmugasundaram (Compilers) Int. Vegetable Soybean Conference, 2nd, Tacoma, WA. 10–12 Aug. 2001. Washington State Univ., Pullman, WA. Miles, C. 2001. Edamame variety trials in Werstern Washington. p. 123–126. In T.A. Lumpkin and S. Shanmugasundaram (Compilers) Int. Vegetable Soybean Conference, 2nd, Tacoma, WA. 10–12 Aug. 2001. Washington State Univ., Pullman, WA. Mohamed, A.I., P.A.J. Perera, and Y.S. Hafez. 1986. New Chromophore for phytic acid determination. Cereal Chem. 63:475–476. Mohamed, A.I., H. Bhardwaj, A. Hamma, and C. Webber, III. 1995. Chemical composition of kenaf (Hibiscus cannabinus L.) seed Oil. Ind. Oil Crops Prod. 4:157–165. Mohamed, A., M.S.S. Rao, and T. Mebrahtu. 2001. Nutritional and health benefits of vegetable soybean: Beyond protein and oil. p. 131–134. In T.A. Lumpkin and S. Shanmugasundaram (Compilers) Int. Vegetable Soybean Conference, 2nd, Tacoma, WA. 10–12 Aug. 2001. Washington State Univ., Pullman, WA. Rao, M.S.S., and A.S. Bhagsari. 1998. Variation between and within

maturity groups of soybean genotypes for biomass, seed yield, and harvest index. Soybean Genet. Newsl. 25:103–106. Rao, M.S.S., B. Mullinix, M. Rangappa, E. Cebert, A.S. Bhagsari, V.T. Sapra, J.M. Joshi, and R.B. Dadson. 2002. Genotype-environment interactions and yield stability of food-grade soybean genotypes. Agron. J. 94:74–82. Reddy, N.R., S.K. Sathe, and D.K. Salunkhe. 1982. Phytates in legumes and cereals. Adv. Food Res. 28:1–92. Sakamoto, C.M., and R.H. Shaw. 1967. Light distribution in soybean canopies. Agron. J. 59:7–9. SAS Institute Inc. 2000. SAS/C Onlinedoc, Release 7.00. SAS Institute Inc., Cary, NC. Shanmugasundaram, S., S-T. Cheng, M-T. Huang, and M-R. Yan. 1991. Varietal improvement of vegetable soybean in Taiwan. p. 30–42. In S. Shanmugasundaram (ed.) Vegetable soybean: research needs for production and quality improvement. Asian Vegetable Research and Development Center, Taiwan. Shanmugasundaram, S., and M-R. Yan. 1999. AVRDC vegetable soybeans for nutritional security, income generation, and soil sustainability. p. 450. In H.E. Kauffman (ed.) Proc. World Soybean Research Conference VI, Chicago, IL. 4–7 Aug. 1999. National Soybean Research Lab, University of Illinois, Urbana, IL. Shanmugasundaram, S., M-R. Yan, and R-Y. Yang. 2001. Association between protein, oil, and sugar in vegetable soybean. p. 157–160. In T.A. Lumpkin and S. Shanmugasundaram (Compilers) Int. Vegetable Soybean Conference, 2nd, Tacoma, WA. 10–12 Aug. 2001. Washington State Univ., Pullman, WA. Shibles, R., and C.R. Weber. 1965. Leaf area, solar radiation interception, and dry matter production by soybeans. Crop Sci. 5:575–577. Shurtleff, W., and T.A. Lumpkin. 2001. Chronology of green vegetable soybeans and vegetable-type soybeans. p. 97–103. In T.A. Lumpkin and S. Shanmugasundaram (Compilers) Int. Vegetable Soybean Conference, 2nd, Tacoma, WA. 10–12 Aug. 2001. Washington State Univ., Pullman, WA. Tsou, S.C.S., and T.L. Hong. 1991. Research on vegetable soybean quality in Taiwan. p. 103–107. In S. Shanmugasundaram (ed.) Vegetable soybean: research needs for production and quality improvement. Asian Vegetable Research and Development Center, Taiwan.