Page Banner

United States Department of Agriculture

Agricultural Research Service

Related Topics

Research Project: IMPROVE THE DETECTION OF QUALITY ATTRIBUTES AND CHEMICAL AGENTS IN AGRICULTURAL COMMODITIES

Location: Market Quality and Handling Research

2006 Annual Report


1.What major problem or issue is being resolved and how are you resolving it (summarize project aims and objectives)? How serious is the problem? Why does it matter?
Manufacturers of food products and their suppliers of raw products are under increasing pressure from regulatory agencies and consumers to improve the quality and safety of their products. Estimating a characteristic of a bulk lot is accomplished by measuring the characteristic in a sample taken from the lot. There is always some number of false positives (sellers' risk where good lots are rejected) and false negatives (buyers' risk where bad lots are accepted) associated with using samples to estimate the true characteristic of a lot. Both buyers and sellers want to reduce these risks to the lowest possible level. Methods need to be developed that allow commodity industries and regulatory agencies to evaluate the performance (false positives, false negatives, lots removed, quantity of a characteristic reaching the public) of a sampling plan so that efficient plans can be designed that meet industry or regulatory objectives. Customers have specifically requested assistance to: (1) develop methods to evaluate and design an aflatoxin control program for almonds, pistachios, hazelnuts, Brazil nuts, and farmers' stock peanuts, (2) develop methods to evaluate the performance of sampling plans that measure genetically modified (GM) seed in bulk grain lots, and (3) develop methods to evaluate the performance of sampling plans that measure foreign material in raw shelled peanut lots. The evaluation methods are built upon statistical models that can accurately predict the variability and distribution among replicated sample values taken from a bulk lot. The variability and distribution among sample values for the attributes discussed above will be measured. This information will be used to develop unique computer models to evaluate the performance of sample designs for these industries. These models will be used in technology transfer activities to design sampling plans to more effectively detect these attributes in various commodities.

Manufacturers of food products are exerting greater pressures on suppliers of raw commodities to provide product with lower levels of mycotoxins, GM proteins, and foreign material. Government regulatory agencies are also continuing to reduce mycotoxin legal limits on imported products. It is estimated that 25% of all crops are contaminated with mycotoxins resulting in large economic losses to producers, processors, manufacturers, exporters, and importers. As a result of animal studies, aflatoxin and other mycotoxins have been shown to be toxic and carcinogenic. About 100 some countries have established legal limits for aflatoxin and other mycotoxins to control the levels of mycotoxins in food and feed products. Europe and Asia are setting limits on the percent GM seed found in imported lots. As a result individual commodity industries invest substantial resources to test raw product so that consumer-ready products meet FDA legal limits or manufacturer specifications. StarLink Logistic Inc. have spent an estimate 3 million dollars over the last three years assisting the grain industry to keep StarLink corn out of the food market by testing corn for GM protein. With such large investments of resources and concerns for public and animal health, commodity industries and regulatory agencies want a method to evaluate the performance of sampling plans so that efficient sampling plans can be designed to minimize misclassification of lots. Methods to predict the effects of sample designs on the costs and risks to processors, manufacturers, consumers, exporters and importers are needed to improve product quality and improve consumer safety.


2.List by year the currently approved milestones (indicators of research progress)
Year 1 (FY 2005)

1-- Complete sample collection and send samples to aflatoxin lab for analysis (objective 1)

2-- Complete sample collection and send samples to Federal State grading offices for FM count (objective 2)

3-- Complete collection of samples (objective 4)

Year 2 (FY 2006)

1 -- Complete analytical testing and send test results to MQHRU to begin statistical analysis (objective 1)

2-- Complete FM measurements and send test results to MQHRU to begin statistical analysis (objective 2)

3-- Complete sample collection and send samples to FDA analytical lab for analysis (objective 3)

4-- Complete shelling of all samples and divide the samples into components based on size and damage (objective 4)

Year 3 (FY 2007)

1-- Complete statistical analysis and begin development of a computer model to predict performance of aflatoxin sampling plans for almonds (objecitve 1)

2-- Complete statistical analysis and develop computer model to predict performance of FM sampling plans (objective 2)

3-- Complete analytical testing and send test results to MQHRU to begin statistical analysis (objective 3)

4-- Complete aflatoxin analysis of all components samples based upon size and damage (objective 4)

Year 4 (FY 2008)

1-- Complete computer model and work with industry and regulatory agencies to design sampling plans to meet customer requirements of costs and risks (objective 1)

2-- Work with industry and regulatory agencies to design sampling plans to meet customer requirements of costs and risks (objective 2)

3-- Complete statistical analysis and develop computer model to predict performance of sampling plans (objective 3)

4-- Complete statistical analysis to determine aflatoxin reduction due to sorting by size and damage (objective 4)

Year 5 (FY 2009)

1-- Complete documentation of study (objective 1)

2-- Complete documentation of study (objective 2)

3-- Work with industry and regulatory agencies to design GM sampling plans to meet customer requirements of costs and risks (objective 3)

4-- Work with producers and shellers to establish aflatoxin sampling plans and price schedules based upon aflatoxin levels in farmers’ stock peanuts (objective 4)


4a.List the single most significant research accomplishment during FY 2006.
Single most significant accomplishment during FY 2006 year: The effect of sorting by size and color on partitioning aflatoxin in farmers' stock peanuts into various acceptable and unacceptable shelled peanut grade sizes was determined. Equations were developed that can predict the level of aflatoxin in each of six grade sizes that have been sorted into acceptable and unacceptable categories by electronic color sorters. Working with Marshal Lamb and Joe Dorner, USDA/ARS Dawson Georgia, aflatoxin in 11 grade and color sort categories were determined and regression equations developed to correlate aflatoxin before sorting to aflatoxin in each of the 11 categories. The prediction equation will assist handlers and USDA aflatoxin limits for aflatoxin sampling plans for farmers' stock peanuts.


4b.List other significant research accomplishment(s), if any.
The variability and distribution among 10 sample test results was determined for each of 33 lots of Jamaican ackee fruit contaminated with the natural toxin hypoglycin A (HG-A). The variability associated with the HG-A test procedure was partitioned into sampling and analytical variance components. The variability and distribution information was used to develop a model to predict the performance of sampling plan designs to detect hypoglycin A (HG-A) in Jamaican ackee fruit. Because HG-A is a natural toxin that causes illness in consumers that eat the fruit, FDA is in the process of developing a monitoring program to limit fruit into the U.S. that exceed the FDA guideline. Working with Joyce Saltsman and George Ware of the FDA, the variability and distribution among sample test results was determined and a model was developed. The model will be used by FDA to design HG-A sampling plans that will be used in a monitoring program to evaluate levels of HG-A in ackee fruit imported into the US.

The variability among the 80 sample estimates of foreign material taken from each of 12 lots was compared to estimates predicted by the binomial distribution. The comparisons indicated that the binomial distribution can be used predict the distribution among sample test results for a given percent foreign material in the lot and a given sample size. Working with Francis Giesbrecht (Department of Statistics, NCSU) and Gregg Grimsley (Birdsong Peanuts), a computer model based on the binomial distribution was developed to predict the performance of sampling plan designs to estimate foreign material in peanut lots. The model has been used to assist handlers and food manufactures design sampling plans to detect FM in peanuts and treenuts.

The variability and distribution among 20 sample test results was determined for each of 20 lots of medium grade runner peanuts that contained peanuts with fruity fermented (FF) off-flavor. The variability associated with the FF off-flavor test procedure was partitioned into sampling and measurement variance components. The variability and distribution information was used to develop a model to predict the performance of sampling plan designs to detect FF off-flavor in bulk peanut shipments. The sampling plan will be used by handlers and food manufacturers to identify lots with FF off-flavor intensities that exceed customer specifications. Because FF off-flavor is considered an objectionable flavor by consumers, food manufactures have developed a monitoring program to detect and remove peanut lots from the edible trade that exceed FF off-flavor limits. Working with Timothy Sanders, USDA/ARS and Francis Giesbrecht, Professor of Statistics (retired), NC State University, evaluation models were developed and the variability and distribution among sample test results was determined. The model will be used by the peanut industry to design sampling plans to detect bulk shipments of peanuts with excessive levels of FF of-flavor.

The variability and distribution among 32 sample test results was determined for each of 4 lots with varying percentages of spotted peanut kernels. The variability and distribution information was used to develop a model to predict the performance of sampling plan designs to determine the percentage of spotted peanut kernels in bulk peanut shipments. The sampling plan will be used by handlers and food manufacturers to identify lots with percentages of spotted peanuts that exceed customer specifications. Because spotted peanuts are considered to be objectionable by consumers, food manufactures have developed a monitoring program to measure the percentage of spotted peanuts in a lot and remove peanut lots from the edible trade that have percentages that exceed limits. Working with Timothy Sanders, USDA/ARS and Francis Giesbrecht, Professor of Statistics (retired), NC State University, an evaluation model was developed and the variability and distribution among sample test results was determined. The model will be used by the peanut industry to design sampling plans to detect bulk shipments of peanuts with percentages of spotted peanuts that exceed food manufacturer limits.

The variability and distribution among 20 sample test results was determined for each of 87 lots of shelled corn marketed in 5 locations in Nigeria that were contaminated with fumonisin. The variability and distribution information was used to develop a model to predict the performance of sampling plan designs to detect shelled corn lots contaminated with fumonisin. Working with Bruno Doko and David Byron of the International Atomic Energy Agency (IAEA) in Vienna, Austria, an evaluation model was developed and the variability and distribution among sample test results was determined. Sampling plan will be designed to meet specifications developed by food safety officials in Nigeria and other developing countries to detect contaminated lots in the market place.

A computer model was developed to investigate the effect of various sampling plan designs used by the U.S. and Japan to test shelled corn for aflatoxin in the export market that would reduce lots rejected in Japan. Sample size and accept/reject limits for both plans were varied to demonstrate how these two parameters can be varied to reduce lots rejected in Japan. Japan indicated a concern to the U.S. government and to the grain industry that too many U.S. corn lots were being rejected at import as having aflatoxin levels greater than their national limit of 10 parts per billion B1. USDA/GIPSA and the U.S. grain industry requested that sampling plans be designed for the U.S. and Japan to reduce rejects in Japan. Working with John Pitchford (USDA/GIPSA), Gary Martin (NAEGA), and Arvid Hawk (Cargill), a model was developed and used to demonstrate that harmonization of the U.S. and Japanese aflatoxin sampling plans would drastically reduce lots rejected in Japan. Results of the study are currently under review by the Japanese government.


4c.List significant activities that support special target populations.
None


5.Describe the major accomplishments to date and their predicted or actual impact.
This project (6645-44000-008-00D) is a second year project following a normal progression of research from 6645-44000-007-00D. Major accomplishments from the terminated project included: (1) measuring the variability and distributional characteristics associated with measuring aflatoxin and fumonisin in shelled corn and vomatoxin in wheat that led to the development of a method to evaluate the performance of sampling plans for shelled corn and wheat; (2) development of a method to predict aflatoxin in farmers' stock peanut lots by measuring aflatoxin in the three high risk grade components; (3) determination of the efficiency of the blanching process to reduce aflatoxin in contaminated raw shelled peanut lots; (4) development of a peanut market system model to show the effects of industry processing methods and aflatoxin regulations on the reduction of aflatoxin in peanuts from the farm to the manufacturer.

Computer models were developed to evaluate the performance of aflatoxin sampling plans for almonds and hazelnuts that are currently being used by the commodity organizations and Codex to design aflatoxin sampling plans for tree nuts. Because regulatory agencies such as the FDA and European Commission have identified aflatoxin contamination in several treenuts as a public health concern, the Almond Board of California (ABC), International Nut Council (INC), Brazil Ministry of Agriculture (BMA), and CODEX have indicated a need to design aflatoxin sampling plans for almonds, hazelnuts, pistachios, and Brazil nuts that minimize the misclassification of lots in both the export and domestic markets. Studies were designed in cooperation with Francis Giesbrecht (Department of Statistics, NCSU), Merle Jacobs (ABC), and Guner Ozay (INC) to determine the variability and distributional characteristics among replicated sample test results taken from contaminated almond and hazelnut lots and develop a method to evaluate the performance of aflatoxin sampling plans for tree nuts. As part of the CODEX harmonization process, the Brazil Ministry of Agriculture, INC, and the California Pistachio Board have also reviewed the almond experimental design and have initiated plans to measure variability and distributional characteristics to evaluate the performance of sampling plans for brazil nuts and pistachios.

Determined how much aflatoxin in farmers' stock peanuts is partitioned by shelling plant processes into various shelled peanut grades so that a more effective aflatoxin control program can be developed for the peanut industry. The U.S. peanut industry continues to discuss the establishment of an aflatoxin-sampling plan for farmers' stock peanuts, which requires decisions about the accept/reject limits associated with such a sampling plan. It is generally agreed that the aflatoxin limits for farmers' stock peanuts should relate to the efficiency of shelling plant processes at removing aflatoxin-contaminated peanuts from farmers' stock peanuts in the formation of shelled peanuts. Working with Marshal Lamb and Joe Dorner (USDA/ARS, National Peanut Research Lab), studies were designed to measure how much aflatoxin is removed from farmers' stock peanuts by shelling plant processes in the production of shelled peanuts. Forty runner-type farmers' stock peanut lots, contaminated with aflatoxin, were each shelled, graded into five commercial kernel sizes, and each size category separated into accept/reject components using electronic color sorters. The aflatoxin in each shelled kernel categories was measured and the percentage of total aflatoxin in the farmers' stock peanuts that was partitioned into the 20-shelled kernel categories was determined. Correlations between aflatoxin in each category and aflatoxin in the lot before processing are being determined.

Measured the Cry9C protein distribution among 800 individual StarLink corn kernels that will be used to develop design StarLink sampling plans for FDA, EPA, USDA, and the grain industry. Because StarLink (Cry9C protein) is a genetically modified corn that produces an insecticidal protein, the U.S. Food and Drug Administration (FDA), USDA, and EPA limit the use of StarLink corn to feed-use only and established a zero tolerance for StarLink corn in the human food supply. Since some StarLink corn has been found in the food supply, FDA, EPA, USDA, and corn millers now inspect over one million shelled and milled corn lots each year, destined for human consumption, for the presence of StarLink. It is difficult to determine accurately the levels of StarLink corn in large shipments because of the uncertainty associated with the sampling and analytical methods used in the test procedure, which results in some lots being misclassified. Working with Francis Giesbrecht (Department of Statistics, NCSU) and Mary Trucksess (FDA), the Cry9C distribution among 800 individual kernels has been measured. From the Cr9C distribution, the effect of sample size on the sampling variance can be determined. Knowing the variability and distributional characteristics, StarLink sampling plans can be designed to reduce the number of lots misclassified, which will reduce both health risks to the consumer and economic loss to the processor.

The sampling and analytical errors associated with measuring StarLink in corm meal and flour were determined so that effective StarLink sampling plans can be developed for the grain industry and regulatory agencies. Because StarLink is a genetically modified corn that produces an insecticidal protein, the U.S. Food and Drug Administration (FDA) limits the use of StarLink corn to feed-use only and has established a zero tolerance for StarLink corn in the human food supply. Since some StarLink corn has been found in the food supply, FDA, USDA, and corn millers now inspect shelled and milled corn, destined for human consumption, for the presence of StarLink. It is difficult to determine accurately the levels of StarLink corn in large shipments because of the errors associated with the sampling and analytical methods used in the test procedure, which results in some lots being misclassified. Working with Francis Giesbrecht (Department of Statistics, NCSU) and Mary Trucksess (FDA), the sampling and analytical errors associated with measuring StarLink in corm meal and flour were determined. Once the magnitude of the testing errors were know, the effect of sample size and number of analyses on reducing testing errors and the number of lots misclassified was demonstrated. Knowledge of the measurement errors will reduce both health risks to the consumer and economic loss to the processor.

The percentage of foreign material in 80 samples taken from each of 12 peanuts lots (960 grade samples) was measured and the variability and distributional characteristics among sample test results was determined to provide data to design effective sampling plans to predict the level of foreign material in shelled peanuts. Food manufacturers of peanut products are requiring shellers supply raw peanuts with the lowest possible levels of foreign material. Even though shellers use the latest sorting technology to remove foreign material, it is next to impossible to completely remove all foreign material. Shellers and food manufacturers sample peanut lots to estimate the quantity of foreign material in a lot. Because of the errors in sampling, some lots will be misclassified according to the level of foreign material in the lot. Working with Francis Giesbrecht (Department of Statistics, NCSU) and Gregg Grimsley (Birdsong Peanuts), 80 grade samples were taken from each of 12 peanut lots containing foreign material. The percentage of foreign material in each of the 960 grade samples (12x80) was determined by grading inspectors. The variability among the 80 sample estimates of foreign material per lot will be statistically analyzed so the rate of misclassifications can be predicted as a function of sample size.

Developed theoretical models to predict the effect of sample size on reducing the chances of not detecting lots intentionally contaminated with chemical agents. Because of increased threats of terrorism, the FDA and USDA are concerned with food security or the intentional adulteration of foods with chemical and/or biological agents. The FDA and USDA are developing plans for larger scale inspections or sampling of foods to detect intentional adulteration. Because of errors associated with sampling, it is difficult to accurate estimate the true level of an unwanted agent in foods and as a result some lots will be misclassified. The model was developed in cooperation with Douglas Park (FDA). While these theoretical models are based on previous studies to predict the performance of sampling plans to detect mycotoxins, an unintentional contaminate produced by fungi, they should provide regulatory agencies with guidelines on how to design sampling plans to minimize consumers' risk associated with not detecting contaminated lots.

Determined the percent reduction that various sorting methods have on reducing aflatoxin in processed almonds to demonstrate to the EU that current aflatoxin limits for almonds are too low for unprocessed almonds. The EU indicated they would consider increasing the maximum aflatoxin limit for imported almonds from 5 B1/10 total ppb to 8 B1/15 total ppb (current peanut limits) if documented evidence could be produced showing that processing methods reduce aflatoxin in almonds. At the request of the Almond Board of California, experiments were designed, data were analyzed, and results were documented in the Market Quality and Handling Research Unit Laboratory at N.C. State University in cooperation with Julie Adams and Merle Jacobs (ABC), and Frans Verstraete (European Commission) that described the percent reduction that various sorting methods have on reducing aflatoxin in processed almonds. Aflatoxin reductions in processed almonds was consistently in excess of 90% when using the basic industry sorting methods such as electronic color sorting, hand sorting, gravity table sorting, and blanching to remove aflatoxin-contaminated almonds during processing. Results have been incorporated into a document developed by the Almond Board of California and sent to the European Commission for review.

Evaluated the performance (accuracy and precision) associated with four commercially available test kits at detecting peanut proteins in four different foods to assist FDA develop an inspection program to detect peanut protein in foods. Unintentional contamination of foods with peanut proteins can cause severe allergenic reactions that can result in shock and death. The determination of peanut proteins in foods at various stages of the manufacturing process by analytical methods can reduce the risk of serious reactions in highly sensitized individual by removing contaminated foods from the market system. The U.S. Food and Drug Administration (FDA) is developing an inspection program to randomly test foods for unintentional peanut contamination using commercially available analytical test kits. The efficiency of detecting and removing contaminated foods depends on the performance of the test kits used to quantify peanut proteins in foods. The study was conducted in cooperation with Francis Giesbrecht (Department of Statistics, NCSU), Mary Trucksess and Kristine Williams (FDA). The performance of the test kits will be used to help field inspectors minimize the errors associated with not detecting contaminated lots. Minimization of false negatives will make the food supply safer for individuals allergic to peanut proteins.

Determined the uncertainty associated with sampling plans used to detect ochratoxin A (OTA) in green coffee so that effective sampling plans can be developed for the export market. Since OTA is a toxic and carcinogenic compound that naturally occurs in several foods such as grains, coffee, and grapes, organizations such as FDA, European Union, and CODEX are considering the establishment of maximum limits and sampling plans to detect OTA in several foods, particularly in coffee. Because of errors associated with sampling plans used to detect OTA in coffee, some lots are misclassified causing an unnecessary expense to the processor and a health risk to the consumer. The study was conducted in cooperation with Francis Giesbrecht (Department of Statistics, NCSU), Eugenia Vargas (Brazil Ministry of Agriculture), Garnett Wood (FDA), and Frans Verstraete (European Commission). Using the uncertainty estimates, a computer model was developed to predict the performance of OTA sampling plans for green coffee. The computer model is being used to demonstrate to exporting and importing countries how to design OTA sampling plans to minimize misclassification of coffee lots.

Determined the moisture distribution among individual peanut kernels for 50 groups of 100 peanut kernels to predict the percentage of high moisture peanuts in bulk lots after the curing process based upon the measurement of moisture in a limited number of kernels. Because a few high moisture peanut kernels in bulk lots going into storage can cause quality deterioration and aflatoxin formation, there is a need to know the distribution of moisture content among individual peanut kernels after the curing process. Working with Christopher Butts, statistically analyzed single seed moisture content data for 50 groups of 100 peanut kernels/group in the Market Quality and Handling Research Unit Laboratory at N.C. State University in cooperation with Christopher Butts (National Peanut Research Lab) and Francis Giesbrecht (Department of Statistics, NCSU). Compared three theoretical distributions to the each of the 50 observed distributions and determined the most suitable theoretical distribution to characterize the actual moisture content distribution among peanut kernels. The theoretical distribution will be used to predict the percentage of high moisture peanuts in bulk lots after the curing process to improve quality.

Compared peanut kernel size and aflatoxin contamination among peanuts in the current US # 1 grade and the new proposed US #1 grade to assist USDA/AMS decide on implementing a new definition of a n US #1 grade peanut. Because peanut shellers requested that the USDA,AMS and the Peanut Standards Board (PSB) change the definition of a US #1 runner peanut from peanuts that rides a 16/64 slotted (16S) screen to one that rides a 17/64 round (17R) screen, the kernel size distribution and aflatoxin levels among peanut kernels that would ride a 17R screen were compared to kernels that ride a 16S screen. The study was conducted by the Market Quality and Handling Research Unit Laboratory at N.C. State University working in cooperation with Tom Tichenor, Frank Bodiford, Bobby Joyner, Jim Wendland (USDA,AMS), and Francis Giesbrecht (Department of Statistics, NCSU). Results indicated that peanut kernels that ride a 17R screen contains 21.9% more aflatoxin and 21.4% more smaller kernels than peanuts that ride a 16S screen. Studies were presented to the American Shellers Association and Peanut Standards Board meeting. Peanut Standards Board recommended to USDA-AMS that the 17R screen define a US #1 runner.

Compared the percent damage, percent foreign material, and aflatoxin levels in imported peanut to domestic peanuts to help make U.S. peanut industry more competitive in the world market with foreign origin peanuts. Using grading records provided by the USDA-AMS and Georgia Federal State Inspection service, the percent damage and foreign material were compared in both domestic and imported peanuts were sorted by grade and type peanut in the Market Quality and Handling Research Unit Laboratory at N.C. State University in cooperation with Frank Bodiford, Nate Tichnor, Bobby Joyner, and Howard Valentine. Results showed that percent damage and foreign material were in much smaller amount in domestic peanuts than in foreign origins. Results were presented to a USDA-FAS and American Peanut Council trade workshop with European buyers.


6.What science and/or technologies have been transferred and to whom? When is the science and/or technology likely to become available to the end-user (industry, farmer, other scientists)? What are the constraints, if known, to the adoption and durability of the technology products?
Collected and stored 40,000 records (one record per shelled peanut lot marketed) of aflatoxin and grade data for the 2003 crop year for the Peanut Standards Board (PSB) formally the Peanut Administrative Committee. Provided data analysis of 40,000 lot records showing PSB the extent of the aflatoxin contamination in that crop year. The data based, started in 1975, and contains about 1,200,000 total records of aflatoxin and grade data. PSB and USDA/AMS frequently request analysis of the database to help guide future decisions about the USDA aflatoxin control program for peanuts.

Assisted both the almond industries develop an industry wide aflatoxin-sampling program for both the domestic and export markets. False negatives are a special concern for product in the export market to the EU since their aflatoxin limits are much lower than the U.S. FDA.

Provided sampling plan designs to the Codex Committee on Food Additives and Contaminates (CCFAC), which are under review as possible harmonized aflatoxin sampling plans for treenuts. The harmonization of sampling plans will improve world trade and provide better consumer protection.

Assisted a California pistachio handler develop sampling plans to detect foreign material in shelled pistachios. The effect of several sampling plan designs that varied sample size and accept/reject limits on the risk of accepting lots with foreign material above the tolerance were developed for the handler.


7.List your most important publications in the popular press and presentations to organizations and articles written about your work. (NOTE: List your peer reviewed publications below).
Lamb M.C., Blankenship, P.D., Whitaker, T.B., Butts, C.L. A Note on the Accuracy and Variability of Grading and Marketing High Moisture Farmer Stock Peanuts. Peanut Science, 30:94-99. 2003.

Whitaker, T.B., 2005, Sampling Feeds for Mycotoxin Analysis in The Mycotoxin Blue Book, ed. Durate, Diaz, Nottingham University Press, Bath, England, pp.1-23. 2005.

Vargas, E.A., Whitaker, T.B., Santos, E.A., Slate, A.B., Lima, F.B., and Franca, C.A. Design of sampling plans to detect ochratoxin A in green coffee, J. Food Additives and Contaminates, Vol. 23:62-72. 2005.

Johansson, A.J., Whitaker, T.B., Hagler, W.M., Bowman, D.T., Slate, A.B., and Payne, G. Predicting aflatoxin and fumonisin in shelled corn lots using poor quality grade components, J. Assoc. Official Analytical Chem., Int., Journal of AOAC International, Vol. 89, No. 2, pp. 433-440. 2005.


Last Modified: 9/23/2014
Footer Content Back to Top of Page