Community structure, macrohabitat selection, and patterns of species co-occurrence were examined during a 14-month study of small mammals in the Cerrado Province of central Brazil. Data were collected from mark-recapture grids in brejo and gallery forest, and from live-trapping and specimen collection in all habitat types within cerrado (campo limpo, campo sujo, cerrado [s.s.], cerradão, brejo, valley-side wet campo, and gallery forest). Gallery forest supported the highest species richness, most complex vertical distribution of species, highest level of trophic diversity, and highest macroniche diversity. Degree of habitat selection varied widely. All habitat types supported both rodents and marsupials, although marsupials tended to be much less common in the grasslands (campos) than in woodlands (cerrado) and forests (cerradão, gallery forest). Some species, such as Didelphis albiventris, occurred in all habitat types, while others were much more restricted. Oryzomys bicolor, for example, ocur-red only in gallery forest. No habitat type had a completely distinct fauna: overlap in species composition always occurred with at least one other habitat type. Because of the great variability of habitats, and the fact that subsets of the mammal fauna were frequently habitat specific, the overall species richness of any portion of mixed cerrado vegetation is remarkably high.

Tropical rain forest is described as a mosaic of ecologically uniform patches. A method is proposed to recognize and inventory the patches and to draw mosaic maps, based on the size and architecture of canopy trees. Mosaic maps from three different forest localities in Indonesia are compared in terms of the relative area occupied by each kind of patch, and the surface range and number of constituent patches. The comparison shows that the mosaic patterns are specific to the sites and can be used for the characterization of patch dynamics in different forest types. Possible implications for forest management are discussed in the light of the mosaic maps.

Phenological observations were made in a Sumatran rain forest during three years (1980–1982). Phenological changes followed a consistent seasonal pattern. The abundance of young leaves and the fall of leaf litter peaked between December and February (first dry season); flowers were most abundant between January and April (first dry and first wet sea son), and ripe fruits in July-August (the second dry season). The fruit of strangling fig trees showed peaks in April and October, both wet season months. Within the study area there was variation in both the phase and the amplitude of the phenological cycles. One year, 1981, displayed mast flowering and fruiting. The observations indicate that the conditions for production were better during the mast year, a finding that facilitates our understanding of the evolution of mast fruiting.

Diurnal and nocturnal censuses were conducted on animal visitors at a single strangler Ficus aff. cotinifolia (H.B.K.) tree during the entire period of ripe fruit availability. Fruit fall to the ground was measured with fruit traps, and chemical analyses were conducted on the figs. Fourteen species of birds, four species of nonvolant mammals and two species of bats consumed the figs. Of the total number of ripe figs accounted for during the sampling period (N= 1.9 X 105), 42% fell to the ground, 45% were removed by mammals and 13% by birds. Frugivores such as the howling monkey Alouatta palliata, the cacomistle Bassariscus sumichrasti, the keel-billed toucan Ramphastos sulphuratus and fruit-eating bats removed the largest amounts of figs from the tree and deposited seeds of 13 plant species other than figs below the tree crown.

Relations between soil nutrient availability, the occurrence of root symbioses and tree species composition in tropical Africa are discussed in the light of new evidence. Transects from humid areas at the equator towards the deserts can be summarized as follows:

High availability of nitrogen, low availability of phosphorus, low pH and aluminium toxicity probably restrict the competitiveness of nitrogen-fixing species in the forests on highly weathered and leached soils in humid west and central Africa. Ectomycorrhizal species attain dominance in species-poor forests on the soils poorest in phosphorus, while species with vesicular-arbuscular (VA) mycorrhiza form diverse forests on more fertile soils. Nitrogen-fixing species, which mostly are VA mycorrhizal, are more common in woodlands, where nitrogen is lost in grass fires, than in forests, which have a more closed nitrogen cycle. Low availability of phosphorus, however, limits the growth of nitrogen-fixing species in moist savannas. Ectomycorrhizal species are dominant, possibly because of a competitive advantage conferred by the large storage capacity of the fungal mantles under conditions of discontinuous nutrient supply. Nitrogen-fixing species become more abundant towards drier areas with an increase in soil phosphorus and a decline in soil nitrogen. These conditions culminate at the desert fringe, where nitrogen-fixing species reach their most conspicuous development.

It is concluded that soil nutrient availability and root symbioses interact and strongly affect the composition of vegetation communities.