A total of 537 crops from four coexisting granivorous doves (Columbina minuta, C. passerina, C. talpacoti and Scardafella squammata), collected on seven sampling dates in a Venezuelan savanna, were examined to assess food partitioning. These closely related doves are resident birds, and overlap broadly both in daily activity patterns and habitat use; in circumstances like this, segregation in the use of food resources often occurs. It was found that: (a) differences in body dimensions did not account for variations in seed size taken by doves; they all ate seeds in the same size range and they showed high similarity in seed size preferred; (b) the importance of different seed sizes in the diets changed seasonally with similar tendencies for all four species, (c) there were differences up to 4-fold in mean seed size consumed by individuals of the same species on the same date; (d) seeds of 63 plant species were consumed by the four dove species and constituted from 98 to 99.5% of the diets; (e) high dietary overlap both in prey size and prey type was found throughout the year. The dietary relationships found in this study support the idea that these birds forage opportunistically, and that some differences found among their diets may be the result of random seed sampling from the heterogeneous seed pool available.

A large number of newly published and unpublished hectare plots in Amazonia and the Guiana Shield area allow an analysis of family composition and testing of hypotheses concerning alpha-diversity in the south American rain forest. Using data from 94 plots the family-level floristic patterns in wet tropical South America are described. To test diversity patterns, 268 plots are used in this large area. Contrary to a widely held belief, western Amazonian plots are not necessarily the most diverse. Several central Amazonian plots have equal or even higher tree diversity. Annual rainfall is not a good estimator for tree diversity in the Amazonia area and Guiana shield. Plots in the Guiana Shield area (and eastern Amazonia) usually have lower diversity than those in central or western Amazonia. It is argued that this is not because of low rainfall or low nutrient status of the soil but because of the small area of the relatively isolated rain forest area in eastern Amazonia and the Guiana Shield. The low diversity on nutrient-poor white sand soils in the Amazon basin is not necessarily due to their low nutrient status but is, at least partly, caused by their small extent and fragmented nature.

Although herbivory in ground-rooted flora is well documented, current knowledge of the herbivore pressure on vascular epiphytes remains mostly anecdotal. Here, we present the results of a 3-year study on the herbivory in a population of the epiphytic bromeliad Vriesea sanguinolenta. In different years, 26–61% of all epiphytes showed traces of herbivore attack, while up to 4.4% of the entire leaf area of the epiphyte population was consumed annually. The recorded levels of damage to photosynthetic tissue, mostly caused by the larvae of Napaea eucharilla (Riodinidae, Lepidoptera), indicate that vascular epiphytes may be regularly and sometimes even lethally attacked by insect herbivores. The level of damage is comparable to ground-rooted tropical flora, which certainly does not support the prevalent notion of low and negligible levels of herbivory in vascular epiphytes.

During the flooding of primary lowland rain forest by the filling of a hydroelectric reservoir in French Guiana, live-trapping was conducted on 175 subsequently formed islands. Different factors affecting the mammal trap efficiency were analysed. The main results show the real complexity of the sampling problem and the influence of the habitat disturbance due to the flooding. The location and/or the type of traps influenced captures of Dasypus novemcinctus, Proechimys spp., Metachirus nudicaudatus and Philander opossum in relation to their foraging and/or locomotion behaviour, and to to their size. The conformation of the islands (height and size) had no influence on the number of captures, nor on the species richness. The number of captures increased with the number of trapnights during the first water inflow and the level stretch. The best trapping success was observed during the wet season, when the strong habitat modification and the forest fragmentation became more significant. It resulted from the reduction of available land area and trophic resources. One year after the beginning of the water inflow, the decrease of the number of captures with the increasing number of trapnights reflected a real decrease of the number of mammals on the islands. The species richness increased with the number of trapnights in all periods, and was also the highest during the wet season.

Forest edges bordering on pasture were created by cutting and burning the surrounding Amazonian lowland rain forest in the dry season (June) of 1990. Litterfall was measured for 3.5 y along transects 10, 50, 100 and 250-m into the forest from the forest edge. Litterfall along the 10-m transects increased by up to 2.5 times that on spatial controls (250-m transects) in the dry season in which the edge was created. In the second dry season after edge creation litterfall at 10-m was lower than on controls, after which it returned to control rates in the second wet season, 1.5 y after edge creation. Litterfall 50-m into the forest was less affected; there was a smaller rise in the dry season in which the edge was cut, and no significant effects after that. At 100-m there was no effect of edge creation on litterfall. Phosphorus concentrations in litterfall were elevated along 10-m transects, 10–12 wk after edge creation, possibly because of reduced retranslocation prior to abscission. The changes in litterfall described here are inextricably linked to the biomass collapse, which occurs near forest-fragment edges in the same area.

Species composition and turnover that have occurred in a series of permanent sample plots established during the 1930s and 1940s in Budongo, a semi-deciduous Ugandan forest, are reported. The plots were established as part of a sequence first used to describe forest succession, five of which have been maintained and which were last measured in 1992-1993. One plot (plot 7) provides 53 y of data from old-growth pristine forest. Plot 15 was established in wooded grassland at the forest edge and is now closed high forest. Evaluation of the remaining three plots is complicated by silvicultural interventions carried out in the 1950s. Forty species have been added since the first evaluations and a total of 188 tree species (over 80% of Budongo's forest tree flora, and including two exotics) has now been recorded from within the plots. The pattern of shade-tolerance in the original plot series conforms to patterns expected for succession with an increasing proportion of shade-tolerant species with development, and large stems appearing to ‘lag behind’ smaller stems in this respect. The time series data are less consistent, and while plot 7 increased in the proportion of shade-tolerant stems through time, the proportion of shade-tolerant species actually declines. Stem-turnover (the mean of mortality and recruitment) slowed with implied successional stage. Most species have a higher recruitment than mortality rate and stem numbers have thus increased in all plots. This is most pronounced in the putatively ‘early successional’ plot. Stem size structure has changed within the plots, with an increased proportion of smaller stems. Species show different rates of turnover and these vary from plot to plot and period to period. In plot 7, the overall mortality rate decreased with initial stem size. Estimates imply that some tree species may easily live longer than 500 y after reaching 10 cm DBH, and that 1000 y is possible. The importance of large trees in determining forest dynamics is illustrated by the finding that death of only seven stems in plot 7 contributed over 60% of net basal area losses recorded over the 53-y observation period. Many of the observed patterns were not predicted and could only have been found by long-term studies.

Tropical Cyclone ‘Rona’ crossed the coast of the Daintree lowlands of northeastern Australia in 1999. This study reports on its impact on forest canopy openness at six lowland rain forest sites with contrasting management histories (old-growth, selectively logged and regrowth). Percentage canopy openness was calculated from individual hemispherical photographs taken from marked points below the forest canopy at nine plots per site 3–4 mo before the cyclone, and at the same points a month afterwards. Before the cyclone, when nine sites were visited, canopy openness in old-growth and logged sites was similar, but significantly higher in regrowth forest. After the cyclone, all six revisited sites showed an increase in canopy openness, but the increase was very patchy amongst plots and sites and varied from insignificant to severe. The most severely impacted site was an old-growth one, the least impacted a logged one. Although proneness to impact was apparently related to forest management history (old-growth being the most impacted), underlying local topography may have had an equally strong influence in this case. It was concluded that the likelihood of severe impact may be determined at the landscape-scale by the interaction of anthropogenic with meteorological, physiographic and biotic factors. In the long term, such interactions may caution against pursuing forest management in cyclone-prone areas.

Tropical cyclonic storms with sustained winds above 120 km h−1 are called hurricanes, typhoons or cyclones depending on their geographic location. They can cause considerable damage to forests. This damage may be in the form of pruned and fallen trees from intense winds (Boucher et al. 1990, Walker et al. 1992), defoliation from a combination of winds and torrential rains (Vandermeer et al. 1997), or mortality from marine inundation of low-lying land (Gardner et al. 1991). Occasionally, extensive defoliation of forests can also occur from wind-driven saltwater when winds are onshore and precipitation is insufficient to dilute the seaspray (Chen & Horng 1993). Below I report the dramatic consequences of an unusual seaspray-laden typhoon on the vegetation of the western Micronesian island of Guam.