will be ad hoc and provide useful allelic diversity but not necessarily from the ancestral area of domestication.
In other cases, such as that of citrus crops (Citrus spp.), existing classifications are substantially untested in the field. IBPGR has put emphasis on relatives of the lime, lemon, and orange and will be organizing fieldwork to this end—not knowing whether the taxonomy works or not. Obviously, the field strategy may have to be modified to fit new knowledge and adapted as necessary. The collection and conservation work will no doubt greatly add to information on the origins, evolution, and distribution of Citrus gene pools. This has already proved true for beans of the genus Phaseolus (G.Debouck, IBPGRI/Centro Internacional de Agricultura Tropical, Colombia, personal communication, 1986), coles (Brassica) (wild gene pool of the Mediterranean), perennial soya bean (Glycine) (IBPGR collaborative work in Australia; A.H.D.Brown, Commonwealth Scientific and Industrial Research Organization, Australia, personal communication, 1986), African eggplants (Solanum spp.) (Lester et al., 1986), and cucurbits (Cucurbitaceae) (L.Merrick, University of California, Davis, personal communication, 1986). Numerous other crops will be worked on in a similar way.
Harlan and de Wet (1971) defined primary, secondary, and tertiary gene pools on the basis of the degree to which closely and distantly related species can be cross-bred. Hitherto a gene-pool approach has been taken in defining IBPGR work. This has obviously been the best strategy, because wild progenitors and their related species have increasingly been shown to be valuable sources of genes for crop improvement. Others, such as wheat, are used for transferring entire chromosomes or parts of chromosomes. Both somatic hybridization and conventional crossing are likely to be more successful among closely related species, but embryo rescue techniques are already commonly used where the parents are less closely related, e.g., in wide crosses between wheat or barley and related wild grasses.
Many advances will continue to come from breeding closely related plant species, but where breeding is far advanced, even species of the tertiary gene pool are useful. IBPGR has recognized this for two groups: first, wild grasses of the tribe Triticeae, which includes wheat, barley, rye, forages, and dozens of species in more than 20 genera distributed throughout the world, and second, the East Asian-Pacific-Australian species of soya bean of the tertiary gene pool. In other cases, the wider gene pool has been used very little; hence, conservation priorities have been imprecisely defined. Wild plants related to maize in the genus Tripasacum are unlikely to be used in maize breeding before the next century.
Integral to crop genetic conservation, but a step subsequent to collection, is description and evaluation of germplasm. Although this activity lags behind, it is an ongoing, long-term process. In autumn 1985, IBPGR held a small strategy workshop to look at the collections and study how they could be used more effectively. It has initiated, in particular, research and development to see how best to work on subsets of large collections, thereby accelerating work. Physiological or otherwise cryptic variation must be thoroughly assessed during this process.