Results/Conclusions We found that the importance of environmental requirements and dispersal limitation varied by spatial scale.  At the site level (10.3 ha), violet abundance was determined equally by environmental and spatial factors.  Violet distribution was concentrated in partially shaded understory, but a seed addition experiment revealed that many suitable locations were unoccupied.  Furthermore, surveys revealed that natural recruitment mainly occurred within 30 cm of adults.  Site level variability gradients are thus longer than typical dispersal distances for the violet.  At the violet-patch level (fine-scale area of <30 m), only environmental factors explained abundance.  Density, size and fecundity were higher in drier, shadier and nutrient-rich microsites, but these factors did not influence distribution.  Variability in the environment was seemingly random over distances of <1 m.  Nutrient variability (COV) was twice that of other environmental factors.  We conclude that the discordance between distribution vs. abundance and performance is due to this non-linear and spatially unpredictable environmental variability.  Specialization is presumably prevented by gene flow between high and low performance individuals.  This violet appears to possess an intermediate persistence strategy, balancing its lack of fine-scale environmental specialization with the ability to find good microsites by lottery.  Examination of other species at this site by similar methods revealed that each partitions the broad and fine-scale variability differently and no dominant species appears capable of capturing all the fine-scale resource patches.  Competitive exclusion is thus unlikely, making fine-scale specializations by the non-dominant species unnecessary for coexistence.  Niches are not totally absent, but are most detectable at broad-scales where resource availability (e.g. soil depth and canopy cover) is more spatially predictable.