The objective of this study was to examine the effects of preemption on rates of competitive displacement and on competitive outcomes for two Typha species (T. latifolia, TL, and T. domingensis, TD) growing along gradients of water depth. Evidence from several experiments indicated that the relative competitive abilities of the two species were size dependent. When TL and TD competed as small seedlings, TD was generally favored while TL was the better competitor when the plants were larger before beginning to compete. Because of this effect, initial density was determined to influence strongly the early outcome of competition. When seeds were sown at high densities, TD gained an early advantage; at lower densities TL was the better competitor. At very low densities, such as when the species competed as clones, TL was very much the better competitor. In the absence of preemption, the results showed that differences in germination, seedling growth, and competitive ability act to create an initial segregation of species at the waterline where establishment occurs. TL is upslope and TD downslope initially, but as the species spread downslope, TL progresses faster because of its greater capacity for vegetative spread, and the result is a widespread overlap along the gradient. Studies of the rate of competitive displacement between established plants yielded an estimate of slightly <6 yr required for complete segregation of two species starting from maximum overlap. Preemption had a strong effect on early growth and distribution of the species. After the initial effects of temporal disadvantage, delayed plants were able to recover under certain circumstances. Both species recovered better in shallow water than in deep water, and, in general, TL recovered better than did TD. Experiments on spatial preemption found that the species moved towards the expected segregation of TL upslope and TD downslope regardless of the initial conditions. This process of recovery was somewhat slower for established stands of adults than for seedlings. Comparison of aerial photographs of natural populations of TL and TD in PAR Pond (South Carolina) in 1980 with distributions in 1984 also suggests that recovery from even large—scale spatial preemption can occur over only a few years time. These results suggest that extreme temporal advantage by either species can lead to extinction of the other species and that the magnitude of the effect depends upon the initial density of seedlings. If both species survive the initial phase of competition and the habitat remains undisturbed, subsequent vegetation spread and competitive displacement can over—come the initial effects of spatial and temporal preemption, resulting in a predictable zonation.