Geographical Variation in Fertility, Phenology, and Composition of the Understory of Neotropical Forests

We compare levels of flowering and fruiting in 55 samples of Neotropical forest understory from 13 sites in 6 countries. Each sample consists of a census of fertile understory plants along a transect. Changes in species richness and density of fertile understory plants are correlated with rainfall and soil fertility. Areas with weak (or no) dry seasons and intermediate to rich soils average 64 fertile plant species and 174 individuals per sample, whereas areas with poor soil and a strong dry season average only 5 fertile species and 8 fertile individuals. Areas with either strong dry seasons and good soils or weak dry seasons and very poor soils have intermediate values. Taxonomic composition of the understory also changes predictably with rainfall and soil fertility. In increasingly stressed forests changes are found in understory structure, with sequential loss of terrestrial herbs, epiphytes, understory shrubs, and lianas. The understory of the poorest soil site consists almost entirely of young trees. The effects of seasonal differences at a given site are small compared with between-site differences. We suggest that the level of understory fertility may provide a simple indicator of overall ecosystem productivity. THE UNDERSTORY OF A TROPICAL RAIN FOREST iS composed of a different set of species than is the canopy. The understory, though often neglected, is an integral and important part of the plant community. For example, understory species constitute 25 percent of the species and 24 percent of the individuals sampled in 1000 m2 at a wet-forest site in western Ecuador (Gentry and Dodson 1987). In a coastal Ecuador moist forest, 2 1 percent of the species and 44 percent of the individuals in a similar sample are restricted to the understory. The importance of understory plants is evident from local species lists: 46 percent of the Rio Palenque species (Dodson and Gentry 1978), 43 percent of the Barro Colorado Island, Panama species (Croat 1978), and 47 percent of the Jauneche (Dodson et al. 1985) species are herbs, shrubs, and small trees. These figures are overestimates, since a number of the species included in such lists are weedy and found rarely, if at all, inside the closed forest, but the overall importance of the understory to rainforest plant diversity is clear. The understory supports a different fauna than does the canopy. Many species of insects, birds, and mammals are restricted to the understory, and understory plants are major food sources for this biota. For example, hermit hummingbirds, major pollinators for many plant taxa, are exclusively understory birds. Such frugivores as birds of the family Pipridae and bats of the genus Carollia are largely dependent on understory fruits (e.g., Snow 1965, Stiles 1981, Fleming 1985). Temperate zone studies have shown that response of the forest understory to environmental gradients can differ from that of the canopy (summary in Whittaker 1977). For the tropics, we now have data that show predictable patterns of change in plant community composition across such environmental gradients as annual precipitation (Gentry 1982a, 1987, in press). Surprisingly little attention has been focused on the easily accessible understory, and few data are available for comparing the structure or floristic composition of the rainforest understory in differ-