Centre for Plant Biodiversity Research, CSIRO-PI, PO Box 1600, Canberra, ACT 2601, Australia

Somewhat paradoxically, the effective long-term in situ conservation of wild relatives of crops for their potential as sources of disease resistance actually requires effective conservation of the pathogens against which resistance is being sought. Failure to maintain the selective pressure applied by pathogens has long-term consequences as (i) the frequency of existing resistance genes becomes fixed (at 0 or 1); and (ii) novel resistances that arise de novo through mutation or intra-genic recombination during meiosis may never increase to a detectable frequency.

Recent developments in our understanding of co-evolved host-pathogen systems suggest that many of these survive as a consequence of natural fragmentation through time and space. In such a metapopulation scenario, the incidence and severity of disease caused by any given pathogen may vary substantially between populations in space and within populations in time. Careful studies of a number of interactions involving rusts and smut diseases have shown this epidemiological asynchrony to occur over a variety of distances including as small as a few hundred metres. At individual sites both host and pathogen have a finite chance of extinction, so that over time the continued existence of both partners in the interaction is dependent on a dynamic process of extinction, migration and recolonization.

The geographic size of interaction metapopulations is not likely to be constant across all host-pathogen combinations. Rather, it will vary according to a range of life-history attributes of both host and pathogen and their interaction (e.g. relative dispersal distances, survival mechanisms, mating system, host longevity). Differences in these characters may lead to differences in the way resistance genes are distributed. Thus constraints imposed by mating system parameters and perenniality of hosts may lead to the retention of high frequencies of resistance and the presence of many resistance genes within individual populations long after environmental conditions have changed and the pathogen has become locally extinct. In contrast, marked differences between different geographic areas in the phonology of an association may lead to the development of quite distinct metapopulations with very different evolutionary trajectories. Examples of the consequences of the interplay of host and pathogen life histories and the environment will be provided by reference to work on Linum, Glycine, Avena and Gossypium.

Recognition of these dynamics and the geographic scale of interaction metapopulations will have a marked impact on strategies for the in situ conservation of wild crop relatives.