1Forest Research, Alice Holt Lodge, Farnham, Surrey GU10 4LH, UK; 2Department of Plant Pathology, Cornell University, lthaca, NY 14853, USA

Both Dutch elm disease (DED) and chestnut blight (CB) have been introduced into new, susceptible host populations in Europe and North America, resulting in destructive epidemics. Deleterious dsRNA- based fungal viruses have been active in the pathogen populations, in some cases bringing about a natural remission of the disease. These viruses therefore offer biological control potential in natural or genetically modified form. In both pathogens, vegetative incompatibility (vc) restricts virus spread, and vc diversity is related to the probability of virus spread. Successful control may depend upon intimate understanding of the genetic structure and dynamics of the pathogen and virus populations and an ability to 'bypass' the pathogen's vc system.

The DED pathogen, Ophiostoma novo-ulmi, usually spreads at epidemic fronts as a single vc type clone, replacing another DED pathogen, O. ulmi, in the process. In Europe viruses spread extensively within these frontal clones. However, the fungus diversifies rapidly (ca 5 years) into numerous new vc types, followed by a marked decline in virus frequency. Recently, transient O. novo-ulmi x 0. ulmi hybrids of low fitness have been detected at epidemic fronts. The possibilities (i) that the appearance of novel vc types in O. novo-ulmi involves introgression of vc genes from O. ulmi; and (ii) that the viruses are also acquired from O. ulmi, are under investigation. In North America, dominant O. novo-ulmi vc clones have persisted and overall vc diversity has remained low: yet genetic crosses reveal a minimum seven vc (vic) loci and a potential for much greater vc diversity. Virus frequency has also remained very low, despite extensive contact with heavily virus-infected O. ulmi.

O. novo-ulmi viruses range from mild to severe, hence moderate viruses or virus mixtures could be deployed for biocontrol. The viruses disrupt mitochondrial function, and their gene action is being investigated for genetic manipulation. O. novo-ulmi populations having large stable vc clones (e.g. eastern North America) could be targeted with either natural or genetically modified viruses; the saprotrophic phase of O. novo-ulmi would tend to promote their spread. O. novo-ulmi populations with high vc diversity (e.g. Europe) could be targeted with viruses reconstructed to bypass the pathogen's vc barriers (see CB, below).

In the CB pathogen, Cryphonectria parasitica, there is a correlation between the success of hypovirulence (caused by hypoviruses) and low vc type diversity. Populations in Europe and Michigan have low vc type diversity relative to those in eastern North America and those in eastern Asia (the probable centre of origin of CB). Recent studies on the genetics of vc types have revealed seven vc (vic) loci. Three vic loci inhibit virus transmission, but only unidirectionally; that is, viruses may be transmitted from one isolate to another, but not back again; two vic loci inhibit in both directions; and one has no effect on transmission. On this basis, it is now possible to estimate the expected probability of virus transmission at the population level. Not surprisingly, the expected probability of transmission is highly negatively correlated to vc type diversity. Empirical tests of these predictions are being conducted. Studies are also under way to estimate the gene flow of viruses between vc types in some populations.

Efforts to overcome barriers to virus transmission because of high vc type diversity have been initiated in the USA with genetically engineered isolates of C. parasitica. Full-length CDNA transcripts of Cryphonectria hypovirus 1-EP713 have been transformed into C. parasitica, conferring the hypovirulence phenotype. Transgenes are transmitted to nearly 100% of the conidia and to 50% of ascospore progeny.Transmission into ascospores is a means of introducing viruses into recombinant vc types in the field. Preliminary field trials are under way, but virus persistence so far is low.