POPULATION DIFFERENTIATION IN THE CHESTNUT BLIGHT FUNGUS, CRYPHONECTRIA-PARASITICA, IN EASTERN NORTH-AMERICA

Isolates of Cryphonectria parasitica were sampled from American chestn ut trees in 13 locations in eastern North America to study genetic dif ferentiation and gene flow among subpopulations of this pathogen. Usin g data from six unlinked restriction fragment length polymorphism loci , we found significant differences in allele frequencies at all loci a mong the 13 subpopulations. Thirty-one percent of the total gene diver sity was attributed to differences among subpopulations (G(ST) = 0.31) Genetic differentiation was examined for two subsets of data: one fro m between two subpopulations in Michigan, which were outside the natur al range of the host, and the other from among nine subpopulations fro m ecologically similar sites (disturbed sites resulting from clearcutt ing and burning were eliminated). There were significant differences i n allele frequencies at two loci between the two Michigan subpopulatio ns, even though these sites were only 16 km apart. Differentiation was high between these subpopulations because two loci were fixed or near ly fixed for different alleles (G(ST) = 0.81) There was also different iation among the nine subpopulations from ecologically similar sites ( G(ST) = 0.20); significant differences in allele frequencies were foun d at five of the six loci. We attribute this level of differentiation to restricted gene flow among subpopulations. We did not, however, est imate gene flow quantitatively from these data because populations did not appear to be at equilibrium. There was no correlation between the estimated average number of migrants per generation between each pair of nine subpopulations and geographic distance. This correlation woul d be expected to be negative if populations were at equilibrium and ge ne flow was restricted to short distances (isolation by distance). The lack of correlation was interpreted as nonequilibrium conditions; the refore, gene flow would be overestimated.