Résumé
Conventional wisdom holds that parasites evolve more rapidly than their hosts and are therefore locally adapted, that is, better at exploiting sympatric than allopatric hosts. We studied local adaptation in the insect‐transmitted fungal pathogen Microbotryum violaceum and its host plant Silene latifolia. Infection success was tested in sympatric (local) and allopatric (foreign) combinations of pathogen and host from 14 natural populations from a metapopulation. Seedlings from up to 10 seed families from each population were exposed to sporidial suspensions from each of four fungal strains derived from the same population, from a near‐by population (< 10 km distance), and from two populations at an intermediate (< 30 km) and remote (< 170 km) distance, respectively. We obtained significant pathogen X plant interactions in infection success (proportion of diseased plants) at both fungal population and strain level. There was an overall pattern of local maladaptation of this pathogen: average fungal infection success was significantly lower on sympatric hosts (mean proportion of diseased plants = 0.32 ± 0.03 SE) than on allopatric hosts (0.40 ± 0.02). Five of the 14 fungal populations showed no strong reduction in infection success on sympatric hosts, and three even tended to perform better on sympatric hosts. This pattern is consistent with models of time‐lagged cycles predicting patterns of local adaptation in host‐parasite systems to emerge only on average. Several factors may restrict the evolutionary potential of this pathogen relative to that of its host. First, a predominantly selfing breeding system may limit its ability to generate new virulence types by sexual recombination, whereas the obligately outcrossing host 5. latifolia may profit from rearrangement of resistance alleles by random mating. Second, populations often harbor only a few infected individuals, so virulence variation may be further reduced by drift. Third, migration rates among host plant populations are much higher than among pathogen populations, possibly because pollinators prefer healthy over diseased plants. Migration among partly isolated populations may therefore introduce novel host plant resistance variants more often than novel parasite virulence variants. That migration contributes to the coevolutionary dynamics in this system is supported by the geographic pattern of infectivity. Infection success increased over the first 10–km range of host‐pathogen population distances, which is likely the natural range of gene exchange.