In all forests there is a cycle initiated by disturbance. We may arbitrarily recognize gap, building, and mature phases (Watt 1947, Cousens 1974, Whitmore 1975, 1978, 1982), thereby designating forests as spatial mosaics of structural phases which change over time as a result of dynamic processes. Gaps, openings in the forest canopy, drive the forest cycle. Very tiny gaps may be filled by lateral ingrowth of surrounding trees. Usually, though, trees grow up from seedlings to form an immature forest of saplings and poles which grow on and develop into mature trees. The mature phase may enter a fourth degenerate phase as trees become senile, but often is destroyed at a stroke by some external factor. Differences in sizes of gaps result in differences in species composition of the next cycle. Present evidence suggests that, in all forests, tree species fall into one or other of two groups. In small gaps seedlings that became established in the shade of the closed forest are released (i.e., commence height growth). By contrast, a quite different group of species colonizes large gaps. Seeds of these species germinate only in the open, so seedlings occur only after formation of gaps. Swaine and Whitmore (1988), while recognizing that there are no universally accepted terms, propose that these groups be called climax (non-pioneer) and pioneer, respectively. Pioneer species only regenerate in large gaps. When a mature canopy of pioneers enters the degenerate phase, small gaps develop, and these are closed by growth of climax species that became established under them. The next cycle is composed of climax species that, unlike the pioneers, can regenerate in situ. The gap phase is thus the most important part of the growth cycle for the determination of floristic composition. Competition among tree species and their different light requirements at the building and mature phases play a lesser role. Intensive study of forest dynamics since the early 1970s suggests this paradigm is applicable to many forests at all latitudes (Whitmore 1988). There are, however, numerous embellishments, as well as many alternatives, and these lead to the great diversity seen even within a single forest biome. A few examples include: clonal spread occurs in some temperate forests; in some tropical rain forests big gaps may be invaded by woody climbers that arrest the next growth cycle; both pioneer and climax species differ in longevity (hence size) and in autecology (e.g., microsite preferences for seedling establishment).