Aims Mechanisms contributing to species coexistence have at least one of two modes of action: (i) stabilization of populations through restoring forces and (ii) equalization of fitness across individuals of different species. Recently, ecologists have begun gleaning the relative roles of these by testing the predictions of neutral theory, which predicts the properties of communities under pure fitness equalization. This null hypothesis was rejected for forests of southern Ontario based on large-scale (∼100 km) spatial synchrony evident in the fossil pollen record over the entire Holocene, and the argument that a species' relative abundance would instead vary independently at such distances in the absence of stabilizing mechanisms. This test of neutral theory was criticized based on the idea that the synchrony might be produced by dispersal alone. Here, I revisit this test of neutral theory by explicitly calculating the synchrony expected in these forests using a novel simulation method enabling examination of the distribution of a species over large spatial and temporal scales.Methods A novel neutral simulation algorithm tracking only the focal species was used to calculate the neutral expectation for spatial synchrony properties examined empirically by Clark and MacLachlan [(2003) Stability of forest biodiversity. Nature 423:635–8] using fossil pollen data from eight lake sites. The coefficient of variation (CV) in a species' relative abundance across the eight sites (initiated at about 10% with a small CV) was calculated for 10 runs over a 10 000 year time interval. The CV reflects the level of spatial synchrony in that less synchronous dynamics should lead to more variation across space (a higher equilibrium CV), and in particular, a greater increase in the CV over time from a small initial value. A ‘two dimensional t’ fat-tailed dispersal kernel was assumed with parameters set to the median derived from seed trap data for deciduous wind-dispersed trees. Robustness of results to assumed dispersal distance, density of trees on the landscape, site sizes, age at maturity and starting spatial distribution were checked.Important Findings In contrast to the prediction of Clark and MacLachlan that, in the absence of stabilization, the CV across the sites should increase over time from levels observed at the beginning of the Holocene, under fat-tailed dispersal my neutral model robustly predicted only a brief (50 years) and small increase in the CV. I conclude that purely fitness-equalized species coexistence cannot be rejected based on the observed lack of increase in the CV across the eight sites in southern Ontario over the Holocene. Synchronous variation in environmental factors could alternatively explain the observed synchrony without the need for stabilization. However, neither dispersal nor environmental synchrony seems likely explanations for the quick widespread recovery of Tsuga in the Holocene after its seeming decimation, likely due to a pest outbreak.