Fallen tree trunks (‘nurse logs’) are important recruitment sites for trees in temperate forest, however nurse log use is seldom reported in tropical forests. We predicted that logs should be important for the regeneration of small-seeded tropical pioneer species because surface leaf litter and competition with established vegetation reduces the establishment success of these species from soil seed banks. In a survey on Barro Colorado Island, Panama, we found that pioneer seedlings were present on logs in 40 of 95 recent treefall gaps. In gaps where seedlings were present on logs, seedling density was not significantly different from adjacent areas of soil. However, species composition was significantly different; logs were disproportionately colonized by smaller-seeded and wind-dispersed species. In growing-house experiments using 12 species, we found that wood substrate had little effect on seed germination. In contrast, seedling growth was 50% lower on decayed wood than soil. Furthermore, species growth rates on wood were not significantly correlated with growth rates in soil (df = 10, r = 0.48). If establishment on logs eventually leads to recruitment to the forest canopy, then logs may promote the maintenance of diversity by favouring a different group of species from those that recruit in soil.

Despite the importance of treefall gaps in tropical forest dynamics, few studies have followed gap-phase processes for more than 2 y. We monitored, for five years, the growth and survival of many seedlings of Tetragastris panamensis (Engler) O. Kuntze, Protium panamense (Rose) I. M. Johnston, and Desmopsis panamensis (Rob.) Saff. (three common tree species of the Panamanian tropical moist forest) in artificially created treefall gaps and under intact-canopy control plots. On these same plots, we also monitored light levels using hemispherical photographs taken over an 8-y period. Seedling height growth was faster in gaps than under intact canopies during the first 42 mo of regrowth, then declined to rates similar to those under intact canopies. Light levels in gaps similarly rose and fell, returning to pre-gap levels by month 48. Only Tetragastris panamensis showed higher survival in gaps, contrary to the general assumption that seedling survival is enhanced by gap creation. Our results document important transition points that mark the conclusion of the gap phase, suggesting an estimate of 4y for the duration of the gap phase in this tropical forest.