Marketing and Delivery of Quality Grains and Bioprocess Coproducts

NON-TECHNICAL SUMMARY: New meaning is being given to quality and quantity of grains by the rapid growth in demand. In the bioprocess area, the major need is to consign products to their optimal use, and to process in such way as to optimize the distribution of key nutrients. For example, corn may be preprocessed before ethanol fermentation to remove certain high quality protein and oil, for diversion to food or feed use while increasing the volumetric capacity of the fermentation. More than ever, industrial supply chain and quality management principles will be applied to the management of now high-value and scarce raw materials. In the food/feed area, the introduction of supply from diverse world locations has created quality, purity and safety concerns. These two are addressable by a combination of instrumentation (analytics) and organized, potentially auditable, supply chain systems. This project will work on both the analytical side of quality/quantity and the management systems side. A validated feedstock measurement system to track biochemical properties relevant to ethanol production into product yield and quality will be developed. A procedure and template for converting alternative or industry-specific quality management system formats to ISO 9000-2000 certifiable formats will be created. A bulk material traceability template based on ISO 22005 standard, will be applied to several grain scenarios. The bio-security website training program will be expanded, and also delivered on site as needed. Grain storage/management related materials from ISU, other Universities, and commercial suppliers are being organized into a multi-level training program. A quality/quantity based assessment of the potential to meet grain supply needs for both food and fuel production will be developed. The impacts of these efforts will be 1) to improve the efficiency of bioprocess industries, critical as raw material prices rise and margins shrink; 2) to capture operating efficiencies throughout the market chain, as they arise from the intensive operations study necessary for quality management/supply management systems; and 3) to develop accepted practices that would allow the rapid release of biotechnology events intended to increase the rate of grain yield increase, while not disrupting trade with customers that choose not to accept biotech products. The greater value of grains has significantly increased the economic value of achieving these goals.

<P>
APPROACH: Iowa State has developed laboratory capability to calibrate a wide range of near infrared instruments and to collect very large spectral databases for these instruments. Calibrations for measuring subunit (amino acid, fatty acid, etc) factors of corn and soybeans will be extended. The chemometrics and information technology necessary to optimize use of very large databases will be applied to several forms of spectroscopy. Image analysis technology will be applied to single seeds, low-concentration food safety/biotechnology factors, and biomass characterization. The chemometrics of transferring spectral data among brands of NIRS, from bulk sample applications to single seed applications, and from single beam spectrometers to individual pixels of image analysis will be developed. A validated feedstock measurement system to track biochemical properties relevant to ethanol production into product yield and quality will be developed. A procedure and template for converting alternative or industry-specific quality management system formats to ISO 9000-2000 certifiable formats will be created. A bulk material traceability template based on ISO 22005 standard, will be applied to several grain scenarios. A universal lot identification and aggregation system will be tested. The system and accompanying cost-benefit analysis procedure will be harmonized with similar work being done in the EU. The process will be applied to the management of less-than-fully approved GM events. Improve value of corn fermentation co-products by removing oil to increase feed quality of DDGS and to generate alternative oil source for biofuels. We are studying how seed preparation and additional treatments may maximize oil distribution into the liquid fraction (stillage), and how the oil in the stillage can be effectively recovered by mechanical means (centrifugation) after chemical or enzymatic treatments. Operate quality analysis testing/instrument calibration services to support research and marketing activities. Achieve ISO 17025 certification with related statistical control of data management. Emphasize high throughput and instrument support services for the public plant breeding community. Provide training in storage and biosecurity issues. The bio-security website training program will be expanded, and also delivered on site as needed. Grain storage/management related materials from ISU, other Universities, and commercial suppliers are being organized into a multi-level training program. This program has been delivered onsite by Extension field specialists, and will become podcasts and web module(s). Decision making spreadsheets will be emphasized. Develop a raw-material quality/quantity based assessment of the potential to meet grain supply needs for both food and fuel production.

<P>

PROGRESS: 2007/01 TO 2007/12 <BR>
Modified fatty acid soybeans were a $100 million product in 2007. Separation from commodity soybeans is necessary to preserve the unique value and functionality of oil modifications, such low linolenic acid, or low total saturated fats. There are at least 40 companies manufacturing or developing near infrared instruments that could be applied to grains. Several provide more spectral information at higher levels of precision than units now used in trade and processing. Harmonization of databases and calibration methodologies could accelerate the acceptance of new technologies with greater capabilities. Standard methods are the basis for legal trade and arbitration. For NIRS, and the properties that it measures, to become accepted in legal metrology, its capability for standardized use must be demonstrated and scientifically accepted. Publication in refereed methods compendia is one means by which scientific acceptance is registered. Accurate traceability of bulk materials will increase compliance with bioterror regulations and improve brand protection. One incident, such as the Diamond Pet Food issue, can easily create tens of million of dollars or more in legal and sales loss costs. The work with the elevator demonstrated that ISO-compliant traceability systems generated net revenue (2-5 cents per bushel) from efficiencies and conformance to specifications, greatly in excess of costs. This means the traceability issue can be a business benefit rather than a cost. In Iowa, we expect that 500-700 million additional bushels will be stored in-state, with less tolerance for deterioration than was previously acceptable in the grain market. An overall reduction of 1% in damaged kernels on the 2.2 billion bushels of corn grown in Iowa represents approximately $60-100 million. The cost of achieving this is balanced by the more than doubling of corn value within the last 18 months, as a result of ethanol production. Producing soy protein fractions rich in either of the two major proteins (glycinin and beta-conglycinin) is key to delivering healthier and better performing soy protein products. We are working with commercial companies to adopt our simple process. GSSP has the potential to be a very effective means of identity preserving soybeans during their processing, Unlike alternative processes, little protein denaturation occurs during oil recovery and the meal can be used to produce organic or certifiable non-GMO soy protein ingredients for which customers in Europe and Asia are willing to pay premium prices. U.S. processors could again become a preferred source of food-grade soybean meal. Our GFP containing corn enables rapid evaluation of new corn fractionation procedures. Some of the new fractionation procedures being developed may be suitable for on-farm use reducing the risk of contaminating commodity corn with unapproved transgenics. The collaborative study on wet-milling identifies laboratories and protocols that are effective in discriminating corn lines with superior wet-milling characteristics. This will enable corn breeders to develop value-enhanced corn for the wet milling industry.
<BR> <BR>
IMPACT: 2007/01 TO 2007/12 <BR>
Whole soybean calibrations for four NIRS brands were developed to measure linolenic acid and total saturated fats. In one case, the standard error of prediction was less than 0.5% pts for each factor, which meant that discrimination to the nearest whole percentage (eg 1% vs 2%) was possible. Rancidity factors were measureable by NIRS using a simple, disposable sample cup placed under the beam of a reflectance NIRS spectrometer. Measures of free fatty acids (FFA), total polar materials (TPM), and conjugated dienoic acids (CDA), typical indices of oil degradation, were analyzed in daily oil aliquots taken from soybean oils with different linolenic acid concentrations used to fry French fries. The coefficients of determination (R2) when compared to the standard methods were 0.973 for FFA, 0.984 for TPM, and 0.902 for CDA. The 10-year soybean protein and oil database of Iowa State University was used to quantify the progressive increase in overall variation among units on individual samples as more brands and configurations of NIRS are introduced to a system. Standard errors increased from 0.30 and 0.25 percentage points (for protein and oil, respectively) when only one transmission NIRS model was used to 0.90 and 0.80 percentage points when three transmission and three reflection instrument models were used. The transfer of prediction models among NIRS brands is not explored. In a network with Foss Infratecs (1229 and 1241) and Dickey-John OmegAnalyzer G's.Partial least squares regression calibration technique was applied to soybean moisture, oil, and protein. Robust techniques using the spectral information from all the network instruments (moisture: r2 = 0.987, SEP = 0.278; protein: r2 = 0.963, SEP = 0.518) performed as well as calibrations developed using traditional optical correction techniques (r2 = 0.988, SEP = 0.279; protein: r2 = 0.958, SEP = 0.561). Based on the development of a triticale protein calibration over four years, several validation scenarios were evaluated (cross validation, a validation set coming from the calibration set with two sizes (10 and 25%) and a validation set containing only next year samples) performed by two regression methods (Partial Least Squares and Least Squares - Support Vector Machine Regression). The next year validation method provided the real accuracy of the calibration; RPD values were half those of same year calibration. There was not a significant difference between 10% and 25% of the calibration set as validation set. Three production cycles were required to obtain a stable calibration that was not likely to need changing in the following cycle. Fractionated soy protein ingredients from solvent-extracted soybean meal and gas-supported screw pressed (GSSP) meal were compared. High-sucrose soybean lines in our fractionation process produced fractions with unique compositions. In a collaborative study of inter-laboratory testing of wet-milling properties of corn, our data compared well with three other laboratories. We also demonstrated the utility and effectiveness of transgenic corn containing green fluorescent protein (GFP) as a tool to test efficiencies of new dry-milling procedures.
<BR> <BR>
PROGRESS: 2006/01/01 TO 2006/12/31 <BR>
NIRS calibration equations for soybean oil fatty acid profile in whole soybeans were developed using partial least squares (PLS), artificial neural networks (ANN), and support vector machines (SVM) methods. Equations for total saturates had the highest predictive ability (r2 = 0.91 - 0.94) Palmitic acid models (r2 = 0.80 - 0.84) were usable for research applications. Equations for stearic (r2 = 0.49 - 0.68), oleic (r2 = 0.76 - 0.81), linoleic (r2 = 0.73 - 0.76), and linolenic (r2 = 0.67 - 0.74) acids could be used for sample screening. SVM models produced more accurate predictions than those developed with PLS regression. ANN calibrations were not significantly different from the other methods. NIRS calibrations were developed for whole soybean amino acid profiles using partial least squares (PLS), artificial neural networks (ANN), and support vector machines (SVM) regression methods and for five models of NIR spectrometers. Validation of models gave r2 from 0.04 (tryptophan) to 0.91 (leucine and lysine). The predictive ability of NIRS models was directly determined the correlation of amino acids to reference protein. No NIRS soybean amino acid calibration outperformed the protein regression. Four NIRS calibration validation scenarios were evaluated with two regression methods (Partial Least Squares and Least Squares Support Vector Machine Regression). Only next year validation sets provided the real accuracy of the calibration; RPD values were half those of same year calibration validation sets. Multiple-brand soybean moisture and protein calibrations were done for two transmission-type instruments. A new Robust Calibration Method gave excellent results both within and among brands. Pooled calibration accuracy was equal to or better than individual calibrations. The best known optical matching procedure for NIRS units (Piecewise Direct Standardization) was able to make instrument brand 1 calibrations transfer to brand 2 without loss in accuracy, but not vice versa. Representatives of twenty Iowa ethanol plants were interviewed concerning corn buying. At this time, twenty-three dry grind plants and four wet mills are expected to produce 1.5 billion gallons of ethanol in Iowa in 2006. Ten new dry grind plants, one new wet mill plant, and six expansions are under construction. The current plants and those under construction will use 1.61 billion bushels of corn annually. Thirty-three new plants or expansions have been announced. If all announced facilities are built and operating at capacity by 2010, fuel ethanol production of 7.5 billion gallons per year in and near Iowa would require 2.68 billion bushels of Iowa corn. Sixty-two percent (range 5% - 100%) of the corn is purchased directly from farmers. Operators expressed a desire to source corn directly from farmers. However, most had absolute limits on quality, unlike the general acceptance policies of traditional elevators. Plants had onsite storage for about 5% (range 2% - 12%) of annual corn use and 2.5% (range 1.6% - 4.7%) of annual DGS production. The need for ongoing training for current and future ethanol plant workers was stressed.
<BR> <BR>
IMPACT: 2006/01/01 TO 2006/12/31 <BR>
The requirement of only one instrument model in a given trading system may be eliminated, which will reduce testing costs, and provide incentives for new technology development. True calibrations (beyond correlations) will have a major impact on genetic screening programs for secondary factors, by identifying unique samples that would have otherwise been missed. Calibration monitoring systems can be simplified, and the cost of calibration maintenance can be reduced by more effective targeting of truly unique or independent samples. The ability to simultaneously align and expand spectra from multiple brands would be a major step toward both harmonization of results and the creation of universal databases that would allow rapid introduction of new technologies/instruments. Based on the results of the ethanol study, new Iowa State Extension efforts in grain storage training, and in supply chain evaluation were started. Programs will be operational in 2007. Plants could increase ethanol output by about 0.1 gal/bu if inputs were more consistent. For a 100 MM gal/yr plant, this would be 4,000,000 more gallons of ethanol or about $8 million per year. DGS quality would be more consistent as well.
<BR> <BR>
PROGRESS: 2005/01/01 TO 2005/12/31 <BR>
The 2005 US soybean crop was lower in protein (34.9%) and significantly higher in oil (19.4%) than in previous years. Processors should be able to meet target meal protein levels and benefit from higher oil yields. However, the lower protein, especially in the northern corn belt, increased the quality differences between US and Brazilian beans in international markets where protein and oil are the key measures of quality. The database from a private variety testing service was expanded to include 2005 crop soybeans. This database contains yield and quality data from approximately 2500 soybean plots annually. There was little correlation between quality (protein plus oil) and yield (r<0.3). Six protein and oil pricing strategies, all in use by one or more users, continued to reward very different combinations of oil and protein. From soybean meal (n=151) and soybean (n=102) samples collected at importing points, U.S. soybean meal was more consistent with higher digestibility, lower fiber, and with better quality protein than SBM of other major export origins (Argentina, Brazil and India) Protein quality carried through from whole soybeans to soybean meal, for a given origin. While the average crude protein contents were higher for the SBM from Brazil, the percentage of total digestible amino acids was highest for the SBM from U.S. and China. The U.S. SBM was higher in the total of the 5 key limiting amino acids for both poultry and swine feed uses, which when coupled with higher digestibility, would give U.S. meal an advantage in rations balanced on amino acids. U.S. soybeans were lower in protein than Brazilian soybeans, but higher than Argentine soybeans. U.S. soybeans were lower in oil content than soybeans from either of the South American origins. The U.S. soybeans likely sourced from the Western United states (exports to Asia, Mexico) were significantly lower than the national averages from the domestic survey, in both protein and oil. These samples were quite similar, however, to the survey samples collected from the four states most likely to originate Asian and Mexican shipments. The crude protein disadvantage of U.S. soybeans was offset by higher concentrations of the essential amino acids in the lower protein beans. The index of 5 key amino acids, as a percent of the total protein was an effective indicator of protein quality. The calibration equations for soybean linolenic acid and total saturates were modified. In on-site testing, the Bruins unit achieved standard errors of prediction for linolenic acid and total saturates of 0.8 and 0.7 percentage points, respectively. The Bruins unit was also effective in predicting soybean moisture, protein, oil and fiber contents. Corn calibration equations for a reflectance-based diode array analyzer, Perten DA7200, were improved. Standard errors of prediction of 0.35 and 0.38 percentage points were achieved for corn protein and oil, respectively.
<BR> <BR>
IMPACT: 2005/01/01 TO 2005/12/31 <BR>
Asian customers are clearly happy to have us do the survey and bring the results to them. Sales have been made because of this customer focus. Processors are gradually instituting component pricing systems. However, the increasing gap between US and South American soybeans has been accentuated by freight patterns to Asia favoring Pacific Northwest shipments. New applications for NIR were demonstrated. The very high-speed diode array technology made large increases in measurement capabilities when put on line at a genetics company. The fatty acid application supported marketing of modified soybeans that produce oil that does not have trans-fat producing hydrogenation.
<BR> <BR>

PROGRESS: 2004/01/01 TO 2004/12/31 <BR>
The 2004 U.S. soybean crop had slightly lower than average protein (35.2%) and slightly higher than average oil (18.7%) contents. The variability in protein and oil content was lower than 2003 and similar to the long-term average variability. Yields and total production were the highest in 20 years. Linolenic acid and total oil content varied across environments, for the same variety. Growing conditions in 2004 increased linolenic acid levels 0.2-0.5 % pts. (of the oil) beyond expectations from low linolenic varieties. A three year database from a private variety testing service contains annual yield and quality data from approximately 2500 soybean plots across the corn belt and east coast growing regions. There was little correlation between quality (protein plus oil) and yield (r<0.3). Six protein and oil pricing strategies, all in market use, rewarded very different combinations of oil and protein, which gives confusing signals to the genetics industry. Calibration models for determination of amino acid concentration in whole soybeans were developed using five NIR spectrometers and three regression methods. No consistent correlation could be established between spectral data and concentrations of amino acids beyond the inherent correlation of amino acids to crude protein. Calibrations for linolenic acid in whole soybeans were developed for the Foss Infratec series of NIR units and are now being updated to use the Neural Network algorithm. Corn and soybean proximate analysis calibrations, developed for Bruins Omega (monochromator) and Perten DA7200 (diode array) units, had accuracy equal to or better than the Official Infratec calibrations on the same samples. Post regression slope and bias (to units other than the calibration masters) did not work for diode array units. The largest producer owned grain handling firm in Iowa, Farmers Cooperative Elevator Company, Farnhamville, Iowa has taken the national lead in application of quality management systems (ISO9000) for agricultural marketing. Its Odebolt elevator and the central grain marketing department became ISO 9000 certified on July 7, 2004, The initial estimate of $2 profit per $1 invested in the QMS has remained, and a major study of performance benchmarks has begun. A spreadsheet-based perpetual inventory system has enabled the company to capture higher value by meeting customer specifications with less material marketed as lower quality. Traceability has emerged as a major concern in all food markets; one of the statistics in the FC system is an index of traceability, to measure how precisely grain from the farm can be tracked to individual food lots sold by second and third stage processors <BR> <BR>
IMPACT: 2004/01/01 TO 2004/12/31 <BR>
Asian customers have appreciated the survey and presentation of results. Sales have been made because of this customer focus. Processors are gradually instituting component pricing systems. We created new applications for NIR, which showed that the very high-speed diode array technology could increase measurement capabilities. The fatty acid application will be necessary to market modified soybeans that eliminate trans-fat producing hydrogenation. The US grain industry is adapting to food safety, biosecurity, consumer right to know and other market pressures with QMS systems that will simultaneously improve efficiency.

<BR> <BR> PROGRESS: 2003/01/01 TO 2003/12/31 <BR>
In 2003 the corn stewardship poster project was expanded to include nine Midwestern states and a national website. Thirty-seven partners from university extension programs, corn producer organizations, grain handler associations, and three seed companies cooperated on the project that promoted the channeling of Market ChoicesSM corn. NIRGrainNet software was tested, using corn and soybean samples on which moisture and protein was being measured with 3 Foss Infratec 1225/1229 analyzers. Spectral data and predictions on three models (PLS, LWR, and ANN) were reported in real time and compared. The nonlinear models were more accurate than the PLS models, but the best accuracy (25% improvement over any model individually) was obtained by either selecting the best model for each sample/constituent situation or by averaging the results of the three models. The neural network model created a 25-30% reduction in standard error of prediction compared to any PLS model, for all corn and soybean factors. The standard deviations across like units were generally reduced by the neural network models. Seven types of spectral noise were evaluated in Partial Least Squares regression, Locally Weighted Regression, and two types of Artificial Neural Network calibrations - random noise, path length, gain, wavelength axis, bandwidth variations, and stray light. Three were significant: baseline shift, wavelength shift, and multiplicative noise. The overall robustness of each model was compared by assessing the dimension of the stability area (constant standard error). Data pre-processing (SNV, genetic algorithms) played a large role in the model robustness. Artificial neural networks were the most robust, whereas the local models were the least robust in most cases. An Iowa grain company and Iowa State University have created a guideline procedure for upgrading entry level quality management systems to ISO 9000-2000. The ISO system expanded the initial system to include management supervision and feedback, in addition to operationally based requirements. Spreadsheet inventory management and traceability tools were developed. U.S. soybean producers, representing 30 soybean production states, provided 1204 samples of 2003 crop soybeans for protein and oil analysis. Average U.S. protein and oil contents for 2003 were 35.65% and 18.66% respectively (on a 13% moisture basis). These are slightly above the long-term U.S. averages of 35.42 % protein and 18.60 % oil. The variability (standard deviation) within states, regions, and the U.S. was substantially higher than in 2002. There has been virtually no change in soybean quality over the 19 years of the survey despite steady yield increases. Oil and protein are inversely correlated, but with varying magnitudes and statistical significance among years and regions. The tradeoff between protein and oil has ranged from 4:1 to 1:1. About 20-25% of samples in any situation will be above average in yield and above average in total components (protein plus oil).
<BR> <BR>
IMPACT: 2003/01/01 TO 2003/12/31 <BR>
This project is providing the measurement, certification and operations management tools necessary to support differentiation of grain markets according to end use. NIRS calibrations and software will enable rapid measurement of more detailed traits to a higher level of accuracy than previously known. Quality management systems will enable the market to reduce operating cost and fill the needs of specialized users at significantly lower cost than expected. The US will be able to meet traceability and food safety requests of Europe and the Pacific Rim at reasonable and competitive cost. US plant breeders and marketers can adopt business strategies that will better target varying customer needs for quality, as opposed to attempts at competing for low cost, generic commodity sales.
<BR> <BR>
PROGRESS: 2002/01/01 TO 2002/12/31 <BR>
In response to a mailed request producers, representing all 29 soybean-producing states, provided samples of 2001 crop soybeans for analysis. Average U.S. protein and oil contents were 35.5% and 19.3%, respectively, which were both above long-term averages of 35% protein and 19% oil. These soybeans will produce and average of 43.2 lbs of 48% protein meal and 11.4 lbs of oil per bushel. The combination of oil and protein, indicative of total value, was the highest in the 17 years of the survey. Soybean fatty-acid calibrations were developed for four brands of NIRS using 1971 samples from the 1993-2001 crop years. The calibrations had Relative Performance Determinants (RPD, standard deviation of the data divided by prediction standard error) in the range of 3-6 for 18:0, 16:0, 18:1, 18:2, and 18:3 fatty acids. A database of amino acid levels has been collected from 605 widely samples of 1993-2001 crop soybeans. Of the essential amino acids, lysine and threonine were highly correlated (r > 0.8) with protein. Cysteine, methionine, and tryptophan levels were not correlated with protein (p=0.05). This means that low protein soybeans such as those from the Western Corn Belt can be effective for poultry nutrition even if the standard 48% crude protein dehulled meal cannot be produced from them. Three Infratec NIRS analyzers were connected to a remote server through Internet based software during the 2001 corn and soybean crop season. PLS, neural network (NN), and locally weighted regression (LWR) calibrations were run simultaneously. Both the NN and the LWR models were more repeatable and accurate relative to chemistry than the instrument standard PLS calibrations. The largest producer-owned grain-handling firm in Iowa, Farmers Cooperative Elevator Company, Farnhamville, instituted a complete quality management system at one of its 35 grain elevators, located in Odebolt, Iowa. Iowa State University (Dr. Hurburgh) was the trainer. The company is now expanding the program to four other elevators, plus its feed business. The format of Quality Systems Evaluation (American Institute of Baking) was used, but the Odebolt system is now being converted to ISO 9000-2000. The company is generating at least $2 in annual profits for every $1 invested in the system. A Pioneer Hi-Bred International, Inc.-Iowa State University (ISU) team produced a 20-minute video, Planter Clean-Out Procedures for Corn and Soybeans (VID 41) and companion publications (Pm 1902-1908), that illustrate the clean-out procedures for seven different planters. An ISU Extension team produced a 20-minute video (EDC 41) Combine Clean-Out Procedures for Identity Preserved Grain and a companion publication (Pm 1923) that illustrate the clean-out procedures for a conventional threshing system combine. In 2002, the corn channeling poster project was expanded from Iowa to include Illinois and Indiana.
<BR> <BR>
IMPACT: 2002/01/01 TO 2002/12/31 <BR>
This project is providing the measurement, certification and operations management tools necessary to support differentiation of grain markets according to end use. NIRS calibrations and software will enable rapid measurement of more detailed traits to a higher level of accuracy than previously known. Quality management systems will enable the market to reduce operating cost and fill the needs of specialized users at significantly lower cost than expected.

<BR> <BR>
PROGRESS: 2001/01/01 TO 2001/12/31 <BR>
U.S. soybeans from the 2001 crop were highly variable in composition, but on average equal to long-term trends. Dry weather reduced yields and increased protein content of corn. New calibration algorithms were developed to optimize performance of near-infrared analyzers when large amounts of calibration data were available. These algorithms were applied to the identification of genetically modified grains. They were also included in copyrighted internet networking software for measurement instruments. A combination HPLC-spectroscopy method was developed for soybean isoflavone measurement. A web-based quality assurance system was developed for a large country elevator. The elevator will be applying for ISO 9000 certification in 2002. Information on impact and response to biotechnology concerns (StarLink, RoundUp Ready) was created and web distributed.
<BR> <BR>
IMPACT: 2001/01/01 TO 2001/12/31 <BR>
Various programs and technologies were created to support identity-preservation and/or traceable product distribution of U.S. grain. These technologies will be essential to resolve consumer concerns and to support marketing of high value grains.

<BR> <BR>

PROGRESS: 2000/01/01 TO 2000/12/31 <BR>
The annual soybean quality survey showed the highest protein and oil values in the 16-year history of the survey. Over 16 years, U.S. soybean yields have increased 0.5 bu/a/yr while the average composition remained steady. Low protein soybeans contain higher levels of the s-containing amino acids, limiting their use to poultry. Near-infrared calibrations were developed to identify GMO from non-GMO soybeans. Accuracy was 93%. An NIR image analyzer was calibrated to measure the level and distribution of moisture and oil in single corn kernels. A producer-owned LLC company was supported in its development of supply chains for specialty grains. A survey showed that 9.5% of Iowa corn acres and 5.6% of Iowa soybean acres were specialty crops in 2000.
<BR> <BR>
IMPACT: 2000/01/01 TO 2000/12/31 <BR>
Development of specialized marketing systems with measurement and certification procedures will be critical to U.S. agriculture as it adapts to changing customer needs. These results support both marketing and genetic development of specialty grains.
<BR> <BR>
PROGRESS: 1999/01/01 TO 1999/12/31 <BR>
The annual soybean quality survey showed below average protein and oil levels - much below average in western growing regions. Meal quality will be difficult to maintain. A major soybean processor began its oil-based pricing schedule, developed from 1998 ISU studies. Approximately 30% of soybeans qualified for the oil-only premium, but inclusion of protein will be necessary to reflect true value. Several organizations, including Iowa State University, have developed NIR calibrations for grains. If different NIR calibration equations (to measure the same product and constituent) are used by traders, error will be increased 20-200% compared to use of the same equations at all trading points. Performance targets for NIR calibrations were developed and recorded as an AACC guidelines method. Locally-weighted regression methods gave more accurate calibrations than PLS or neural networks on large datasets. NIR was used to identify Roundup-Ready (a genetically altered trait) soybeans.
<BR> <BR>
IMPACT: 1999/01/01 TO 1999/12/31 <BR>
The soybean market has begun conversion to component pricing. A national plan for NIR calibration uniformity has been started. It may be possible to screen for genetic modifications at the first point of sale.
<BR> <BR>
PROGRESS: 1998/01/01 TO 1998/12/31 <BR>
The annual soybean quality survey showed much above average protein and oil levels, caused by very favorable growing conditions. Nearly all growing regions should be able to produce 48% protein meal this year. Analysis of Iowa-location strip trial genetic evaluations supported the soybean data and indicated very low protein levels in corn as crops exhausted their nitrogen supplies. An inexpensive near-infrared analyzer proved effective at screening corn and soybeans for basic proximate analysis factors. For any near-infrared model, the use of more than one calibration (from different origin databases) will increase intermarket variability by 25-175%. A study of corn modifications showed that feed-ingredient based genetic enhancements will create short-run values increases but will be offset by losses in soybean markets. Starch-based modifications are industrial and have long-term competitive potential for corn producers. Selection among currently available soybean varieties can benefit processors with incremental gains in protein and oil. However, only about 25% of current germplasm appears to be both high yielding and high composition, which means that selection will increase costs to genetics firms.
<BR> <BR>
PROGRESS: 1997/01/01 TO 1997/12/31 <BR>
The annual soybean quality survey showed reduced protein levels in 1997 soybeans, but higher oil levels. A similar corn survey, new for 1997, produced samples from 18 states. While there were no major regional patterns, the protein range of 5.6% - 10.8% was surprising for normal, non-specialty corn (white, high-oil, waxy, flint, popcorn). A large database of grain quality data was brought online through the Iowa Grain Quality Initiative. New standardization and master selection procedures were developed for near-infrared analyzers. Over 80 Infratec units were standardized for elevator use; variability among units was less than 33% of variability relative to chemical references. A sampling of soybean meal from world origins showed U.S. meal to be superior in fiber, protein digestibility, and overall consistency. A software program was developed to assist producers in evaluating grain contract production opportunities.
<BR> <BR>
PROGRESS: 1996/01 TO 1996/12 <BR>
The second year of quality testing for grower-sponsored corn and soybean strip plot trials verified last years' finding that quality differences could cause economic value differences about 20% as large as could yield differences. The annual U.S. soybean survey showed 1996 protein content to be 0.5 percentage points higher and oil content 0.5 percentage points lower than long-term averages. The near-infrared analyzer network grew to 72 units in eight states. The analyzers had a standard deviation across units of about half the standard error of prediction relative to chemistry reference. New algorithms were developed for temperature compensation of near-infrared spectra and for standardization of optical data across near-infrared units.
<BR> <BR>
PROGRESS: 1995/01 TO 1995/12 <BR>
Data from corn and soybean strip-plot trials in Iowa showed that intrinsic quality variations among varieties could represent about 20-30% as much economic value as yield variations. U.S. soybean protein and oil contents were close to 10-year averages but highly variable because of weather/growing conditions. Late planting and early frost made early harvest estimates of oil yields unreliably high. Updated corn and soybean calibrations for two near-infrared analyzer brands gave excellent accuracy on 1995 samples. A computer-connected network of 20 elevators with NIRS analyzers was started. An industry-university task force on grain quality was created.
<BR> <BR>
PROGRESS: 1994/01 TO 1994/12 <BR>
Average protein and oil content of 1994 U.S. soybeans was at normal levels, withsome central cornbelt locations showing oil contents in excess of 20% (basis 13% moisture). Over nine years of survey data, oil content was consistently more variable than protein content. Mandatory additional cleaning of soybeans would not yield operational benefits in excess of costs. The current U.S. Grade definition of foreign material was effective at differentiating non-grain from grain material. A formal procedure for calibration and standardization of several brands of near-infrared analyzers was developed and field-tested. Analysis of raw spectral data showed that the present procedure of incorporating hot and cold samples in near-infrared calibration sets can never be totally effective for temperature compensation.