Building a North American Feed Information System (1995)

The value of any data base depends on the regular release of current, time-sensitive information and attentive maintenance. Data bases must be oriented to the end user and must have active professional oversight for the exchange of reliable information. Accessibility and user-friendly operation are the two most important components that determine the extent to which a data base will be used.

There have been rapid advancements in electronic communications technology in recent years resulting in improved function and performance. Faced with increasing reliance on computers and computer networks exclusively for day-to-day activities, many professionals have achieved a tremendous level of improvement in their understanding of the mechanisms and capabilities of these systems. Advancements in both the technology and users' understanding of data base systems have resulted in the development of many useful networks of information exchange. The ability now exists to maintain enormous collections of material, archived from a myriad of various sources over many years, and to continually update, improve, and disseminate the information.

Recognizing the importance of the nutrient composition of feeds as early as 1809, the first table containing such information was published by Thaer, who compared the nutrient value of feed to that of hay (Oltjen, 1992). The 15th edition of Feeds and Feeding included a compilation of the nutrient composition of feeds in the United States. That book was continually revised until the twenty-

second edition was published in 1956 (Morrison, 1956). In 1941 the chairman of the Committee on Animal Nutrition (CAN) of the National Research Council (NRC) asked B. H. Schneider of the West Virginia Agricultural Experiment Station to prepare a report examining the results of all studies on the digestibilities and composition of feedstuffs that had been conducted to that date. Feeds of the World, Their Digestibility and Composition was published as a result of Schneider's work (Schneider, 1947).

In 1952 NRC's Board on Agriculture took on the challenge of addressing the need for data on the nutrient composition of feed by establishing the Feed Composition Committee. Two publications, Composition of Concentrate By-Products (National Research Council, 1956) and Composition of Cereal Grains and Forages (National Research Council, 1958), resulted from the efforts of that committee. In 1959, United States-Canadian Table of Feed Composition: Nutritional Data for United States and Canadian Feeds was published (National Research Council, 1959). That publication contained chemical and biological data for 1,500 feeds. Scientists at Utah State University began developing a system to describe feeds in 1948, and in 1963 they obtained a grant from the National Institutes of Health to develop an international system for naming feeds and for computerization of feed data. The NRC's Committee on Animal Nutrition made extensive use of the feed data and during the period from 1963 to 1971, the Subcommittee on Feed Composition was charged with converting the data to metric equivalents, in addition to updating and adding new data (Oltjen, 1992). The Atlas of Nutritional Data on United States and Canadian Feeds (National Research Council, 1971) was published as a result of the committee's work.

In 1971 the Food and Agriculture Organization of the United Nations (FAO) conducted a survey of ongoing international activities in the field of feed data collection; the systems used to code, store, and retrieve feed data; and possibilities for collaboration on an international basis. Following the survey, FAO invited scientists working in this area to Rome in 1971 and 1972 to explore the feasibility of achieving international cooperation in describing feedstuffs and collecting feed composition data. These meetings brought about INFIC. The aim of that group is to contribute to more responsible animal production throughout the world by improving access to reliable information on the composition, nutritive value, and practical use of feeds for animals. The activities of INFIC began with promoting the establishment and effective operation of cooperating centers for the collection, processing, and dissemination of information on the chemical composition and nutritive value of feeds while addressing general information on practical ways to feed animals and the efficient use of feeds. INFIC makes possible the exchange and dissemination of information within and between countries in a simple, uniform, and unambiguous manner.

Prompted by the need for a participatory INFIC center within the United States, the International Feedstuffs Institute (IFI) was organized at Utah State University in 1972. The concept of a feed data base functioning as a tool for the

exchange of information was adopted. IFI took on the responsibility for collecting and processing feed composition data in the United States. Scientists from IFI collaborated with scientists at other INFIC centers throughout the world. Data were an invaluable resource used by the CAN Subcommittee on Feed Composition and by other subcommittees in preparing and revising animal nutrient requirements publications. Research scientists, practicing agricultural professionals, producers, and educators relied heavily on the feed composition data base, first developed by CAN subcommittees and augmented by IFI at Utah State University. Inclusion of feed composition data in NRC's nutrient requirement publications was a touchstone for the science of animal feeding. The data were and continue to be in demand, with more than 200,000 copies of CAN nutrient requirement and feed composition publications purchased.

Revision of the United States-Canadian Table of Feed Composition was completed in 1982 with the publication of its 3rd edition (National Research Council, 1982b). USDA appointed a study group that same year to review the operations of IFI, determine aspects critical to researchers, and recommend a strategy for future data base activity. The study group emphasized the critical need for the data base by users in the United States and abroad and recommended upgrading the domestic aspect, obtaining missing data, and improving the accuracies of specific values (R. R. Oltjen, Agricultural Research Service, U.S. Department of Agriculture, unpublished data, 1982).

Responsibility for the data bank was subsequently transferred to the National Agricultural Library in 1985 as the Feed Composition Data Bank (FCDB). The data base was maintained, and new values obtained from the published literature were incorporated for the next several years. In response to budget constraints, the activities of FCDB were suspended in 1990, halting an organized nutritional science movement toward facilitating sustainable, efficient, and environmentally sound agriculture.

The data base most recently referred to as FCDB has a solid foundation of data. A vast wealth of nutrient composition and digestibility information is already in the data base; however, an enormous amount of new information has become available in recent years. A comprehensive analysis of the archived material is required to eliminate obsolete and inaccurate data. Because the infrastructure is in place, and because of the wealth of data already available, an improved and revitalized system can realistically be put to use without delay. Nevertheless, improvements need to be made for the present system to become the progressive model described in this report. Review of the former feed data base reveals voids, inadequacies, and discrepancies in information. Variations in feed and by-product nutrients due to genetics, the environment, and processing systems are not reflected in the data base. The bioavailabilities of feeds are not

addressed in the system, nor is information on the presence of naturally occurring mycotoxins in feeds available. There is also no standard method of analysis of feeds, and in general, most feed nutrient values are simply static national averages that provide minimal useful information for today's market. The format of the previous system has become antiquated.

The emergence of the electronic information highway has sparked the development of many different interactive forms of information exchange such as bulletin boards, gopher sites, list servers, and data bases. Telecommunications capability has become widespread, making access to a network a fairly effortless task. On the basis of the experience gained from the operation of the former feed data base and critical evaluation of other, similar data bases, several design options exist for the new North American feed information system.

Analysis of existing data bases can prove to be beneficial in the development of a new system. Operational functions, budgetary considerations, and staffing are areas that should be closely assessed because they may direct future plans of action. Additionally, an evaluation of the difficulties that other data bases have experienced can provide insight into avoiding similar difficulties.

Many centralized computer data base systems permit user flexibility in the storage and retrieval of information. An example of a such a system is the Germplasm Resources Information Network (GRIN), a cooperative effort by state, federal, and private organizations established to allow researchers access to information on an extensive collection of germplasm samples (Mowder and Stoner, 1989). This network is a means for recording, maintaining, evaluating, and searching for germplasm data through a central site and is now available globally through World Wide Web on the Internet. In addition, links have been established with other data bases, such as the plant genome data base and the maize data base. Similarities exist between GRIN and the North American feed information system regarding data base function and the variety and format of the data that they contain. Information is contributed to GRIN by institutions and organizations throughout the research community (National Seed Storage Laboratory, Regional Plant Introduction Stations, and National Germplasm Repositories) working with germplasm. Data base users include plant breeders and scientists from the public and private sectors. Information from other agricultural germplasm projects is being added to GRIN, including information on livestock, poultry, insects, microbes, and forests, expanding the scope of users. Much like the GRIN data base, a North American feed information system would provide professionals in the public and private sectors access to a large body of informa-

tion on feeds. Maintained within a preexisting laboratory by scientists working with an annual budget of $1.2 million, which covers the cost of a site license, maintenance of equipment, and labor, the GRIN data base has become a successful repository of information (J. D. Mowder, U.S. Department of Agriculture, personal communication, 1994). With a staff of 13 who are largely responsible for programming under a central management authority, GRIN has undergone modifications since its inception. Improved methods of data preparation and evaluation, verification of accessions, and revision of network format are the primary target areas identified by the National Academy of Sciences in a 1991 report drafted by the Committee on Managing Global Genetic Resources, Managing Global Genetic Resources: The U.S. National Plant Germplasm System (National Research Council, 1991). These recommendations have been acted on and have been put into place, resulting in a significant advancement in the usefulness and productivity of GRIN. Just as GRIN evolved into a strong system of accessible, accurate information from its initial form, the North American feed information system must also undergo critical evaluation and change.

The utilities of GRIN and the North American feed information system are parallel; however, because of the differences in the data in the two systems, methods of data acquisition are distinct. GRIN seeks germplasm information in accordance with established priorities and standards on the basis of observed deficiencies and predicted needs. Conversely, a North American feed information system has relied on voluntary data contributions, reports, and values from the published literature for its data. A better method of securing the type of data relevant to a North American feed information system is to commission data from companies, government agencies, and organizations within the industry responsible for developing, manufacturing, producing, and supplying feeds, in addition to using voluntarily contributed, valid data. An analogous approach for data procurement is used by USDA's Nutrient Data Laboratory (NDL).

The National Nutrient Data Bank, maintained by NDL within the Agricultural Research Service, compiles, validates, and disseminates data on the composition of foods. Data are verified by analysis of food samples through contracts with private laboratories and universities. Blind tests among laboratories are performed to verify and validate the values that are generated, maintaining quality control. Approximately 10 percent of the $1.4 million total annual budget is allocated for contract services to generate sound nutrient assessments of food. The majority of base funds for NDL are devoted to maintaining a staff of 23 professionals employed to develop and maintain specific food-related projects and oversee data entry, verification, and validation (W. R. Wolf, Nutrient Data Laboratory, U.S. Department of Agriculture, personal communication, 1994). Much of the information on the nutrient composition of foods is generated at the

request of interested parties. For instance, the National Heart, Lung, and Blood Institute provided funds for information on the fatty acid content of various foods, and the Agricultural Marketing Service funded analyses of commodity foods (W. R. Wolf, Nutrient Data Laboratory, U.S. Department of Agriculture, personal communication, 1994). The methods of data acquisition and verification used by NDL have direct application to a North American feed information system. Feed data should be obtained and validated in a manner that addresses the requests and support provided by the feed industry. NDL has encouraged partnership with the food industry and is contributing to further development of an industry-initiated standard format for the exchange of data on products, realizing that benefits extend to both the scientific and business communities. Similarly, a strong partnership should be cultivated between a North American feed information system and the feed industry. Active participation by the feed industry in providing feed nutrient composition information is crucial for developing an effective system. Today, the human food industry is more willing to contribute to NDL what it previously considered to be proprietary analytical data. This information has become extremely valuable for food product manufacturers in terms of the recently developed nutrition labeling laws. In addition, companies are now able to create internal infrastructures for data maintenance and analytical procedures used to evaluate products, with the economic advantage of having readily accessible current nutrient information for consumers. Feed companies face situations that are almost identical to those faced by the food industry with respect to nutrient specifications and consumer demand for information. Participation in a North American feed information system will help the feed industry realize the same types of marketing and economic benefits currently experienced in the human food sector.

As with GRIN, NDL has undergone many modifications and continues to direct efforts at improving the usefulness and the quality of its data. Originally developed as a data base available on a mainframe computer, NDL information more recently became available on electronic bulletin boards and computer diskettes. NDL, working with industry, is moving toward making food product data accessible through telecommunications networks on the Internet. NDL is in the strategic planning phase of this modification. Options for networking NDL data base or the proposed North American feed information system include the following:

1. a centralized data base system at a central location for data input and maintenance of the system by system staff,

2. a virtual system with the capability that all contributors can input data

from their respective sites, without the need to funnel information through a central location, or

3. a regional system composed of several selected regional centers assigned to contribute and evaluate information.

The utility of a feed data base has been recognized not only in North America but also in Europe. The French Feed Data Base, established in 1989 and managed by the French Association for Animal Production (Association Française de Zootechnie) is supported jointly by the Ministry of Agriculture and dozens of public research and technical institutions, feed manufacturers, and raw materials suppliers. The French Feed Data Base is a microcomputer-based data base that runs on relational data base management software with structured query language applications. Feed manufacturers, realizing their dependence on this information, supply 80 percent of the data to the French system; the remaining 20 percent is obtained from research organizations and the scientific literature. Likewise, the U.S. feed industry has the capacity to supply a large amount of data to a North American feed information system, and they, too, have a continuous need for information on the nutrient composition of feeds.

Buzzard and colleagues (1991) posed six practical questions for consideration when evaluating a nutrient data base. In establishing the criteria for the development of a North American feed information system, the subcommittee applied these questions to the former feed data base and used their answers as a guide for identifying deficiencies, proposing solutions, and developing the recommendations presented in this report.

1. “Does the data base contain all of the foods and nutrients of interest? ” The former data base contained information on the nutrient composition of most common feedstuffs but was incomplete with regard to all feed nutrients. The proposed system would be comprehensive, providing detailed information on all feeds and nutrients for which accurate information is available. Additionally, information on underutilized feeds and the bioavailabilities of nutrients and, ultimately, information on the presence of mycotoxins and contaminants in feeds and a new feed classification system would be included.

2. “Is the data base complete for the nutrients of interest?” With the advent of sophisticated, reliable methods of nutrient analyses, more information will be available. Included in the proposed system will be information on specific feed nutrients such as amino acids, fatty acids, vitamins and minerals, and fiber fractions. Because of the state of technology, data from many of these analyses were unavailable, inaccurate, or absent from the former system.

3. “Do the foods included in the data base provide adequate specificity for the accurate assessment of the nutrients of interest?” Specificity with regard to individual nutrients was not adequate to obtain valid estimates of the nutrients of interest in the former system because the individual methods of analysis were not specified, nor was the number of samples analyzed provided. A new North American feed information system would provide specific information on the method of analysis used for each entry, and standard statistics would be provided to allow for representative assessment of the data.

4. “Is the nutrient data base kept up to date with the changing marketplace and the availability of new nutrient data?” Without a formal structure for data base oversight or staff and working on a restrictive budget with outdated equipment, updating the former feed data base was difficult. With adequate staff and funding, a proposed oversight group should provide a manageable system and should facilitate the routine updating of the information in the data base.

5. “Are manufacturers contacted routinely for new information on reformulations of existing products?” A strong partnership with the feed industry is proposed in the development of a North American feed information system. The human foods industry has recognized the value of analyzing and compiling this information in recent years, and similarly, the feed industry will experience many of the same benefits. Thus, rather than relying solely on the scientific literature for information, the feed industry may be more likely than in past years to contribute information.

6. “What quality control procedures are used to ensure the accuracy of the nutrient data base?” By far, the lack of quality control was the greatest weakness of the former system. Generated by a lack of financial support and relatively no oversight structure, quality control in the previous system deteriorated rapidly. A defined mechanism of quality control, with suggestions for analytical methodology and sources of data acquisition, verification, and validation, is proposed in the formation of a North American feed information system.

Without question, significant breakthroughs in electronic forms of information exchange have greatly enhanced the capabilities of the scientific and working communities. Data bases like GRIN and the National Nutrient Data Bank are prime examples of progress from data printed in tabular form, limited in scope to the number of columns that could fit on a printed page and containing tables categorized by topic with no tabular integration possible, to interactive systems containing current, reliable information. Partnerships between technical information networks and the industries that use them are successful endeavors, as is apparent in the French Feed Data Base system. A North American feed information system must also become a source of current, reliable information contributed to and used by all segments of the agricultural community as well as the rest of society.