Forest ecology suffers from a lack of long-term community records. Preserved pollen data are richer in such information than is generally realized. By applying suitable statistical techniques to pollen records, one can learn much about competition, succession, and population dynamics in past tree communities. In this study, preserved pollen records from Everitt Lake, Nova Scotia, are analyzed as time series. Time domain studies reveal the post-fire responses of individual tree taxa. Correlograms yield models of past forest succession patterns. The models explain some effects of changing fire frequency, thus suggesting mechanisms by which fire, competition, and climate combine to produce long-term forest composition changes. Frequency domain studies suggest relationships between disturbance cycles, stand composition, and forest mosaics. Fire frequencies are seen to be highest where fire-dependent species abound and most regular where tree stands have uniform, not mixed, composition.