Introduction Rapid Tests for Edible Soybean Quality J.A. Andrews, G Batten and L.G. Gaynor, NSW Agriculture, Yanco Industry specifications for edible soybeans have been based on seed size, condition of the seed coat, total oil and protein, however these alone are not an adequate indicator of quality for all end-users. Different total protein testing procedures have been used by various processors, which have resulted in different protein levels for the same sample. This has resulted in the acceptance and rejection of beans on the basis of unreliable testing methods. If quality soybeans that meet market specification cannot be reliably supplied, producers risk losing both local and export markets. This will reduce profitability, producers will leave the industry, soybean imports will increase and Australia s share in this expanding food market will be lost. Consequently there is a need to develop a standardised specification and breed varieties suited to this market. Method Grain samples used were from lower protein seed produced by non-inoculation of seed at sowing, commercial crops in southern, central and coastal New South Wales, southern Queensland and Victoria, and breeding lines from southern New South Wales. Samples of "good and "inferior" beans were supplied by processors and manufacturers. Total protein and Oil testing: A Near Infrared Spectrophotomer in both transmission (NIT) and reflectance (NIR) mode was used to measure total oil and protein. The Leco method was used to verify the protein values to calibrate the NIR and the Soxlet Oil Extraction method was used to verify the oil calibration on the NIR. Tofu making ability: The RVA (Rapid Visco Analyser ) was used to ascertain tofu making ability. An automatic domestic tofu maker was used as an initial measure of tofu yield. Two traders were used in the final year to verify the recommended testing procedures. Results and Discussion: Close collaboration was maintained with selected representatives from all segments of the industry throughout the project. Specifications common to most end-users were a light hila bean with minimum total protein of 38% @ 13% moisture, minimum seed size of 18 gms per 100 seeds @ DMW, absence of split and wrinkled seed coats and germination >90%. Some tofu markers were using the BRI Australia (Bread Research Institute) RVA test where >400 RVA units were considered to be necessary for good tofu making ability. Generally tofu makers preferred high protein- low oil beans, the minimum acceptable protein:oil ratio being 2:1.
Total protein and oil testing: 2,500 samples were collected and scanned. From these scans, calibration and verification sets were selected using NIR software. NIR protein and oil calibrations were developed and set up on the scanning NIR instrument at Yanco Agricultural Institute. They provide inexpensive, accurate, simultaneous measurement of total protein and oil. NIR instruments are generally available throughout the wheat belt of Australia. They are an affordable asset that will enable traders and marketers to reduce costs by doing their own tests. Growers and traders will now be able measure acceptability of individual loads, accurately target end-users and avoid associated costs of rejected loads at each stage in the chain NIT scans found individual bean samples had very different absorption patterns. Figs. 1 and 2 show the log1/r and 2 nd derivative spectra for good beans while figs. 3 and 4 show the same spectra for bad beans. The lack of uniformity of spectra within the groups of good and bad can be related to visual observations, as beans varied greatly in factors such as size, shape, colour and seed coat lustre. A preliminary calibration was developed which ranked samples as good and bad. This calibration had a R2 of about 0.7 in which view of the variations in figs.1-4 is not considered to be applicable for commercial purposes without further development. It was felt that the influence of physical characters on the ranking reported by the NIT dominated the internal quality of the beans and would be subject to incorrectly accepting/rejecting batches for manufacture. Figure 1: Good beans NIT (log 1 /R) Derivative). Figure 3: Bad beans NIT (log 1 /R) Derivative). Figure 2: Good beans NIT (2 nd Figure 4: Bad beans NIT (2 nd
Considering the cost of a NIT instrument, and expected season-season variations in physical characters (which would necessitate frequent verification of the calibrations) it is felt that NIT technology is not likely to be acceptable technology for classifying beans at receival depots without better calibrations. To utilize NIT technology a receival depot would need to segregate a large number of good and bad (eg 100) samples of beans to develop more reliable calibrations. Such samples were not available during the current study. Tofu making ability: The BRI (Bread Research Institute) RVA tests have become accepted as the industry standard. The reason for the variability of results between laboratories has been ascertained. It was found that to make good tofu, beans must have a minimum 400 RVA units. Figure 5: Examples of BRI - RVA profiles of soybeans suitable and unsuitable for processing. The curve must rise within 5-7minutes and be maintained at 400 RVA units for the duration of the test (Figure 5). Preliminary testing of tofu yield can be successfully carried out in the automatic tofu maker, however other characteristics viz colour, texture, smell, taste, flavour, etc., need to be determined by an experienced tofu manufacturer. Figure 6: Soybean variety viscosity RVU variations.
7S and 11S globulin are the two major proteins involved in making tofu. A method for the crude extraction of these proteins was developed from previous work described in the literature. A number of samples (covering all current varieties considered suitable for tofu production) of variable tofu-making ability were tested using the RVA, resulting in a range of viscosity traces (Fig. 6). As an extension, 7S and 11S globulins were added to raw soybean meal for analysis of pasting ability using RVA. Initial results suggest that 11S increase the height of the first peak of the RVA curve and 7S increases that of the second (Figs. 7-8). Uniform standards for Edible Soybean: These basic guidelines have been developed in conjunction with processors, manufacturers, commercial laboratories and by tests conducted during the project as the basis for quality assessment. Uniform Standard for Edible Soybean Total protein Minimum 38% at 13% moisture Protein:Oil Minimum 2:1 Seed Size Minimum 18gms per 100 seeds, at DMW for soymilk, tofu, tempeh, and miso Minimum 12-14gms per 100 seeds, at DMW for spouting Minimum 8-10gms per 100 seeds, at DMW for natto Hilum colour Light Seed coat Intact, ie no splits or wrinkles Germination >90%
RVA number DMW 400 units BRI (Bread Research Institute) test dry matter weight Figure 7: Effect of 11S protein on Viscosity RVU Viscosity RVU Figure 8: Effect of 7S protein on Two marketeers, who supply large numbers of processors in New South Wales and Victoria, assisted by validating the effectiveness of the tests in 1999. One, who used all recommended tests, had nil rejection rate of beans compared to 60% rejection prior to this project. Another using all tests except the RVA had only 2 loads rejected by endusers. Marketeers and exporters, usually being in close contact with growers, can now provide feed back thus enabling informed management decisions to be made. This will lead to an increase in the quality of edible beans overall. Growers will be able to use these tests to satisfy themselves that any rejected beans have been rejected for quality reasons only. Acknowledgments: The authors sincerely thank GRDC for funding, Dempsters Seed Cleaners, Tofu Shop, Australia Natural Food, Agrilink International, Soyproducts, Tixana, Soyco, Nectar Soy Products, Allgold Foods, Blue Lotus Foods, Mr Ostiojic, Oriental Merchants P/L, Bread Research Institute Australia, Newport Scientific, Phillip Brodie Grains, Bartters Enterprises, Cargill, Annand Robinson, Narrandera Seed Cleaners, Riverlands P/L and Baron Processors for their assistance. Uniform Standard for Edible Soybean Total protein Minimum 38% at 13% moisture Protein:Oil Minimum 2:1 Seed Size Minimum 18gms per 100 seeds, at DMW for soymilk, tofu, tempeh, and miso Minimum 12-14gms per 100 seeds, at DMW for spouting Minimum 8-10gms per 100 seeds, at DMW for natto Hilum colour Light
Seed coat Intact, ie no splits or wrinkles Germination >90% RVA number 400 units BRI (Bread Research Institute) test DMW dry matter weight Figure 7: Effect of 11S protein on Viscosity RVU Viscosity RVU Figure 8: Effect of 7S protein on Two marketeers, who supply large numbers of processors in New South Wales and Victoria, assisted by validating the effectiveness of the tests in 1999. One, who used all recommended tests, had nil rejection rate of beans compared to 60% rejection prior to this project. Another using all tests except the RVA had only 2 loads rejected by endusers. Marketeers and exporters, usually being in close contact with growers, can now provide feed back thus enabling informed management decisions to be made. This will lead to an increase in the quality of edible beans overall. Growers will be able to use these tests to satisfy themselves that any rejected beans have been rejected for quality reasons only. Acknowledgments: The authors sincerely thank GRDC for funding, Dempsters Seed Cleaners, Tofu Shop, Australia Natural Food, Agrilink International, Soyproducts, Tixana, Soyco, Nectar Soy Products, Allgold Foods, Blue Lotus Foods, Mr Ostiojic, Oriental Merchants P/L, Bread Research Institute Australia, Newport Scientific, Phillip Brodie Grains, Bartters Enterprises, Cargill, Annand Robinson, Narrandera Seed Cleaners, Riverlands P/L and Baron Processors for their assistance.