account for 90% of the supply of food plants by weight, calories, protein, and fat for most of the world's countries (Prescott-Allen and Prescott-Allen 1990). Just three crops—wheat, rice, and maize—account for roughly 60% of the calories and 56% of the protein consumed directly from plants (Wilkes 1985). Relatively few species that have not already been used as foods are likely to enter our food supply, but many species now consumed only locally are likely to be introduced into larger markets and grown in different regions. For example, the kiwi fruit was introduced into the United States as recently as 1961; within 20 years, US sales had grown to some $22 million per year (Myers 1997).
Although relatively few species are consumed for food, their productivity in both traditional and modern agricultural systems depends on genetic diversity within the species and interactions with other species found in the agroecosystem. Claims that such biodiversity "archives" can serve as substitutes for biodiversity in natural habitats are more fanciful than factual. Genetic diversity provides the raw material for plant breeding, which is responsible for much of the increases in productivity in modern agricultural systems. In the United States from 1930 to 1980, plant breeders' use of genetic diversity accounted for at least the doubling in yields of rice, barley, soybeans, wheat, cotton, and sugarcane; a threefold increase in tomato yields; and a fourfold increase in yields of maize, sorghum, and potato. An estimated $1 billion has been added to the value of US agricultural output each year by this widened genetic base (OTA 1987). Breeders rely on access to a wide range of traditional cultivars and wild relatives of crops as sources of genetic material that is used to enhance productivity or quality. Different landraces can contain genes that confer resistance to specific diseases or pests, make crops more responsive to inputs such as water or fertilizers, or confer hardiness enabling the crop to be grown in more extreme weather or soil conditions.
Much of the genetic diversity available for crop breeding is now stored in a network of national and international genebanks administered by the UN Food and Agriculture Organization, the Consultative Group on International Agricultural Research, and various national agricultural research programs, such as the US Department of Agriculture's National Seed Storage Laboratory in Fort Collins, Colorado. The value of these genebanks for agricultural improvement is substantial. For example, in a presentation to this committee,2 Evenson and Gollin estimated the present net value of adding 1,000 cataloged accessions of rice landraces to the International Rice Research Institute's genebank at $325 million (on the basis of empirical estimates that these accessions would generate 5.8 additional new varieties, which would generate an annual $145 million income stream with a delay of 10 years). As important as they are in agriculture,