nations of protective genes. Increasingly, new genes are being identified and transferred from wild relatives of wheat (for example, Cox et al. 1993 and 1994; Sharma and Gill 1983). Wild relatives of cultivated plants have coevolved with the crop pathogens and so are often extremely useful sources of protective genes (Leppik 1970; Wahl et al. 1984). Although many of the wild relatives of wheat are Triticum spp., many are more distant (McIntosh et al. 1995). For example, the protective genes Lr24 and Sr24 came from tall wheat grass, Thinopyrum ponticum (Podp.) (Barkw. & Dewey); and Lr26, Sr31, and Yr9 came from rye, Secale cereale L. Little is known about the biochemistry of genetically based rust-protection, so most breeding programs use phenotypic selection (the presence or absence of the disease) and some use molecular markers to track protective genes.

The main phases of any wheat-breeding program are introduction of genetic variation, inbreeding and selection of useful variants, and extensive field testing of selected variants to determine their agronomic or commercial worth (Baenziger and Peterson 1992). All the standard plant breeding methods are well documented (for example, Fehr 1987; Stoskopf et al. 1993), as are the methods specifically applied to breeding for rust-protection (Knott 1989; MacIntosh and Brown 1997). The most common breeding method for moving one or a few genes into an elite line or cultivar, especially when the genes are being transferred from a wild relative or an unadapted line, is backcrossing. It has been widely used to introduce protective genes into cultivated wheat whether those genes are derived from Triticum spp. or from more distant but sexually compatible relatives.

As mentioned previously, the effect of rusts can be devastating when susceptible wheat cultivars are grown. However, estimating the value of crop resistance to rust accurately is difficult because the widespread growth of resistant cultivars affect the yield-loss estimates. A well-documented estimate (based on potential yield losses due to the disease-infecting susceptible cultivars) of the annual value of having stem rust resistance in wheat grown in western Canada was Can$217,000,000 (Green and Campbell 1979). The annual yield losses due to stem rust have averaged between 15% in Saskatchewan to 25% in Manitoba. in the United States, epidemics are localized, but the yield losses due to plant susceptibility to stem rust were as high as 56.5% in North Dakota and 51.6% in Minnesota in 1935 (Roelfs 1979). For leaf rust, the yield losses due to susceptibility were estimated at 50% in Georgia in 1972. For stripe rust, yield losses due to susceptibility were estimated at 30% in 1961 in Washington. The losses