Two sites in California, on serpentine soil bearing annual grassland, were studied. Spatial (horizontal and vertical) variation in inoculum potential of vesicular-arbuscular mycorrhizal fungi in soils was measured by a bioassay. Unvegetated gopher mounds had significantly lower inoculum potentials than did undisturbed soils. The decreased inoculum potential in gopher mounds is attributed to the reduced amount of inoculum in subsurface soil brought to the surface during formation of mounds. A difference in inoculum potential between two distinct types of vegetation (grass-or forb-dominated) was also found.