Further studies on growth and reproduction of many tree species are needed to know the regeneration patterns of tropical forests (Clark & Clark 1987, Heideman 1989). Thus, the need for studies on the c. 3% of Atlantic forest that remains in Brazil is acute, particularly in the rarely studied swamp habitats (Scarano et al. 1997). We studied the canopy tree species Calophyllum brasiliense Camb. (Clusiaceae) in flooded and unflooded habitats of a coastal lowland rain forest, in order to describe demography, sexual expression, phenology and flower and fruit production. We report the differences of such traits among three contiguous habitats found in the lowlands - unflooded forest, freshwater-flooded forest and mangrove.

Hornbills are important dispersers of a wide range of tree species. Many of these species bear fruits with large, lipid-rich seeds that could attract terrestrial rodents. Rodents have multiple effects on seed fates, many of which remain poorly understood in the Palaeotropics. The role of terrestrial rodents was investigated by tracking seed fate of five hornbill-dispersed tree species in a tropical forest in north-east India. Seeds were marked inside and outside of exclosures below 6–12 parent fruiting trees (undispersed seed rain) and six hornbill nest trees (a post-dispersal site). Rodent visitors and seed removal were monitored using camera traps. Our findings suggest that several rodent species, especially two species of porcupine were major on-site seed predators. Scatter-hoarding was rare (1.4%). Seeds at hornbill nest trees had lower survival compared with parent fruiting trees, indicating that clumped dispersal by hornbills may not necessarily improve seed survival. Seed survival in the presence and absence of rodents varied with tree species. Some species (e.g. Polyalthia simiarum) showed no difference, others (e.g. Dysoxylum binectariferum) experienced up to a 64% decrease in survival in the presence of rodents. The differing magnitude of seed predation by rodents can have significant consequences at the seed establishment stage.

Foraging and attendance of birds at army-ant swarm raids were studied in the tropical rain forest of Los Tuxtlas, Mexico. Sixty-eight raiding swarms were intercepted over a four-year period of which 57% were Eciton burchelli and 43% Labidus praedator. A total of 461 birds (37 species/12 families) were recorded at swarms of E. burchelli and 208 birds (34 species/10 families) were recorded at L. praedator swarms. The mean number of bird species detected per swarm was 7.2 at E. burchelli raids and 5.6 at L. praedator raids. Red-throated ant-tanagers (Habia fuscicauda) were most frequently seen at the swarms of both army-ant species. Other important species were the golden-crowned warbler (Basileuterus culicivorus), the white-breasted wood wren (Henicorhina leucosticta) and the Kentucky warbler (Oporornis formosus). Swarms of both army-ant species were active year-round. Birds weighing between 20 and 40 g dominated perches closest to the ground and the central zone of the swarms, richest in animal prey. Birds weighing less than 20 g occupied higher perches and tended to forage in more peripheral zones. At Los Tuxtlas raiding swarms of army ants are important in the foraging ecology of 44 bird species, including 12 North American migrants.

Data on litterfall, litter layers accumulated on top of the mineral soil and layer mineralization collected for three years in a tropical rain forest at Pernambuco, Brazil, were implemented on a simulation model. Litterfall was collected biweekly using 11 collectors 1 × 1 m. Every three months, 20 litter mat samples, 0.5 × 0.5 m, were collected, divided into the L, F and H layers and the CO2 evolution from each litter layer was determined in the laboratory. Litterfall, in the three years, averaged 7.8, 8.3 and 8.2 Mg ha-1 y-1, most of it leaves. Litter mat masses varied widely from place to place (15–90 Mg ha-1) and the overall averages were 5.6, 7.6 and 26.1 Mg ha-1 for the L, F and H layers, with CO2 evolution averages of 2.27, 0.507 and 0.123 mgC g litter C-1 day-1. According to the model, the L layer had a high turnover rate, losing 4.7 Mg ha-1 y-1 through mineralization and 3.4 Mg ha-1 y-1 transferred to the F layer. Values for the F and H layers were 1.4 and 1.2 mineralized and 2.0 and 0.8 Mg ha-1 y-1 transferred. Thus, it would take 2.1, 6.7 and 39 years for newly fallen leaves to be mostly fragments, to be transformed to humus and to be incorporated to the soil organic matter, respectively. Variations of litterfall throughout the three years would have little effect on the system which was not very sensitive to litterfall changes, except for the top layer.

For sustainable logging to be achieved in tropical forests, there must be successful gap-phase regeneration to restock the logged-over area. This study examined three aspects of gap-phase regeneration in selectively logged lowland swamp forest of northeast Costa Rica. First, logging gaps were censused immediately after extraction to determine the density of advanced regeneration. Stem density and basal area of residual trees ≥ 10 cm dbh in logging gaps was >85% lower than undisturbed forest, and all trees in gaps had sustained structural damage. The common canopy species Pentaclethra macroloba (Fabaceae) was the most abundant species in gaps whereas the timber tree Carapa nicaraguensis (Meliaceae) was absent from all censused gaps. This suggests that canopy replacement, particularly by Carapa, will depend on trees <10 cm dbh or by seed input into logging gaps. Second, the diversity of the understorey was compared with 6-y old single-tree and multiple-tree logging gaps. Multiple-tree logging gaps were the most diverse, but dominated by two ruderal species; however many shade-tolerant species were present in those gaps. This indicates that controlled selective logging can result in a localized shift in species composition, but that logging gaps should return to pre-logging composition with time under a carefully implemented, controlled harvesting regime. Finally, this study found a significant effect of a fringing Carapa tree on logging gap seedling density. Thus, seed arrival into gaps is a barrier to logging gap regeneration, particularly for a large-seeded tree species. Gap-phase regeneration by a large-seeded tree species in managed forest would benefit from seed broadcasting into gaps.

Flooded savannas in south-west Venezuela consist of seasonal, marshy and flooded grassland communities, the distribution of which depends on soil level. Seasonal grasslands are dominated by tuft-forming, rhizomatous C4-grasses (exemplified by Paspalum chaffanjonii in this study), while flooded grasslands are dominated by stoloniferous grasses which develop rooted, floating culms during the rainy season (Leersia hexandra and Hymenachne amplexi-caulis). The interface between these two communities is a marshy grassland dominated by stoloniferous grass species which tolerate flooding to a depth of 10 to 25 cm (Panicum laxum and Leersia hexandra). All perennial grass species in flooded grasslands behave as typical C3 plants, while marshy grasslands are dominated by Panicum laxum, a species with reduced photo-respiration, low RuBP/PEP-carboxylase ratios, and low absolute RuBP-carboxylase activity compared to C3 grasses. It also has lower photosynthetic rates than the other grass species. Hymenachne amplexicaulis appears to be the least drought tolerant from the species selected, in accordance with its distribution in the wettest side of the flooding gradient studied. The other species showed marked reduction in relative water content and pronounced increase in leaf proline content in drought experiments lasting 6 to 30 days. Alcohol dehydrogenase increased markedly in response to anaerobiosis in the root environment in the tuft forming grasses, while the stoloniferous species with ascending culms were least affected by this treatment, probably as a result of better aeration inside the culms and also to the production of adventitious roots in upper nodes. Nitrate reductase activity increased as a result of anaerobiosis in the roots but not in the leaves, of all species except Leersia hexandra. No difference among the species was found in malate accumulation, or the activity of malic enzyme.

We investigated the links between slope variation and heterogeneity in tree growth and stand structure. The study was conducted in an undisturbed wet evergreen forest in India. We selected three plots situated on gentle slopes and three plots situated on steep slopes. Tree growth was considered as mean annual diameter increment. Stand structure was considered in relation to several characteristics: trunk diameter distributions, tree heights, and spatial distributions of stems and crowns. Except for the spatial structure, analyses were performed on samples including either all the species or only the dominant species (Vateria indica L.). Diameter growth varied according to slope intensity and to tree size: growth of large trees was clearly higher on steep slopes whereas growth of small trees proved lower or similar. Diameter distributions were different among the six plots, but the difference cannot be related to slope intensity. On the contrary, tree heights differed according to slope intensity: for a given diameter, trees were taller on gentle slopes than on steep slopes. Finally, stem spatial distributions showed strong regularity on gentle slopes whereas they were random on steep slopes. The observed changes between gentle and steep slopes suggest differences at tree level through a modification of its shape, but also differences in the major processes responsible for stand dynamics: competition, establishment and survival.

Six rhizotrons in an Eucalyptus tetrodonta savanna revealed seasonal changes in the abundance of fine roots (≤ 5 mm diameter). Fine roots were almost completely absent from the upper 1 m of soil during the dry season, but proliferated after the onset of wet-season rains. At peak abundance of 3.9 kg m−2 soil surface, fine roots were distributed relatively uniformly throughout 1 m depth, in contrast with many tropical savannas and tropical dry forests in which fine roots are most abundant near the soil surface. After 98% of cumulative annual rainfall had been received, fine roots began to disappear rapidly, such that 76 d later, less than 5.8% of peak abundance remained. The scarcity of fine roots in the upper 1 m of soil early in the dry season suggests that evergreen trees may be able to extract water from below 1 m throughout the dry season. Persistent deep roots together with abundant fine roots in the upper 1 m of soil during the wet season constitute a ‘dual’ root system. Deep roots might buffer atmospheric CO2 against increase by sequestering carbon at depth in the soil.

Mass-flowering plant species are often pollinated by social bees that are able to use the abundant resource by recruiting workers from their colonies. In this study we surveyed pollinators on the mass-flowering perennial crop coffee (Coffea arabica) in its native range in Ethiopia. Previous studies in areas where coffee is introduced often find the social honeybee, Apis mellifera, to be the dominant pollinator. In those areas, the bee-species composition visiting coffee varies with a higher bee diversity closer to forest or in less modified habitats. We surveyed pollinators of coffee under different shade-tree structures, by collecting hoverflies and bees landing on coffee flowers in 19 sites in south-west Ethiopia. We found the native honeybee (A. mellifera) to be the dominant visitor of coffee flowers in all sites. Honeybee abundance was not affected by the local shade-tree structure, but was positively affected by the amount of coffee flower resources. Other pollinators were positively affected by complex shade-tree structures. To conclude, the honeybee is clearly the dominant pollinator of coffee in Ethiopia along the whole shade-tree structure gradient. Its high abundance could be a consequence of the provision of traditional bee hives in the landscape, which are colonized by wild swarming honeybees.

Habitat fragmentation causes a sharp increase in the forested area affected by edge effects (Murcia 1995). Compared with the core of the forest, edges present higher litterfall rates, air and soil humidity reduction, higher temperature and increased wind incidence (Bierregaard et al. 1992, Didham & Lawton 1999, Laurance et al. 1998, Matlack 1993). These features of edges may increase the probability of fire occurrence, especially if the surrounding vegetation is composed of grasses (D'Antonio & Vitousek 1992, Freifelder et al. 1998). After a fire, the damaged edge will grow substantial amounts of herbaceous vegetation, extending the inflammable area into the forest, and thereby creating a positive feedback system of fire susceptibility and intensity (Cochrane & Schulze 1999, Cochrane et al. 1999). Fires may affect the structure and composition of the vegetation (Cochrane & Schulze 1999, Didham & Lawton 1999, Sanaiotti & Magnusson 1995), favouring, in the long-term, the dominance of the community by species characteristic of disturbed habitats and making the habitat unsuitable to primary forest species (Malcolm 1994, Possingham et al. 1994). Unfortunately, there are few studies on the effects of fires on neotropical small-mammal populations (Borchert & Hansen 1983, Ojeda 1989, Vieira & Marinho-Filho 1998). The objective of this study was to analyse the influence of a fire on populations of two rodent species, Akodon cursor (Winge) and Oecomys concolor (Wagner) in two fragments of Atlantic Forest in Brazil.

The diets of larvae and juveniles (cut-off size: 20 mm) of Limnothrissa miodon (Boulenger) were studied at 2-h intervals during four 24-h cycles (two during the rainy and two during the dry season) in the Bukavu Basin, Lake Kivu. Larvae and juveniles are both diurnal feeders, and they both feed essentially on copepods, which are the most abundant prey in littoral areas, but they show a marked preference for larger organisms (cladocerans), when available. This indicates a rather intense resource sharing, and probable competition between life stages, which is partly solved through spatial behaviour and segregation. During the dry season, when food resources are scarcer, larvae and juveniles make slightly offset inshore migrations during the day, so that they feed at different places, on copepods of different sizes. Because diel migrations are size-structured, it is suggested that cannibalistic pressure (by adults over juveniles, and by juveniles over larvae) may also be a driving factor for these migrations. Plankton samples revealed that there were few cladocerans, while copepods were abundant and their size was similar to that prior to the introduction of L. miodon in Lake Kivu, thereby suggesting that some form of dynamic equilibrium has been reached in the food web.

This study compares leaf anatomy, chlorophyll content and photosynthetic induction rates for seedlings of five dipterocarp species growing both by a path and in the understorey of a Bornean heath forest. Hemispherical photographs were used to estimate the light level. Although three of the five species showed significantly higher photosynthetic capacity in high light conditions, there were no significant within-species differences in induction rates. Average induction times to reach 50% (T50%) and 90% (T90%) of maximum photosynthetic rate (Amax) were about 1.5 and 9 min for Shorea pachyphylla. In contrast, these were 18 and 37 min respectively for Dipterocarpus borneensis, and 12 and 25 min for Shorea multiflora. Intermediate values were recorded for Hopea pentanervia and Cotylelobium burckii. There was an overall weak and negative correlation of induction rate with stomatal density. Three species showed more rapid induction loss in their leaves from the path edge vs. the understorey. The results suggest that photosynthetic acclimation can influence some aspects of a leaf's dynamic response to sunflecks, such as Amax and induction loss, while not affecting overall induction rates. This study also shows significant differences among diptero carp species in photosynthetic capacity, induction responses and leaf structure, and in acclimation on these traits.

Fruit and seed consumers can both positively and negatively affect plant recruitment through seed dispersion and seed predation, respectively. In turn, fruits influence the abundance and distribution of consumers sustained by local plant assemblages. These interactions are key processes in plant recruitment and the dynamics of tropical forests, where most plants depend on dispersal by frugivorous animals (Corlett 1998). An understanding of these interactions and the functional role of particular seed-dispersing animals is increasingly important nowadays, given that human impact on tropical forest ecosystems may negatively impact seed dispersal and forest regeneration in both natural and human-altered forests (Wright et al. 2000).

In a 30-ha permanent plot of tropical evergreen forest at Varagalaiar, Indian Western Ghats, all trees ≥ 30 cm girth at breast height (gbh) were examined for the presence of lianas ≥ 1 cm dbh. The plot contained 13 445 trees in 152 species and 11 200 lianas in 75 species. Twenty-eight per cent of trees supported lianas and the mean number of lianas per tree was 0.38 ± 0.72. Association analysis between lianas and trees of 16 tree families and 20 abundant tree species indicated that tree susceptibility to lianas was better pronounced at species rather than at family level. Overall, at Varagalaiar site, the aggregation of lianas followed neither Poisson nor clumped distribution. Among the four dominant tree families Dipterocarpaceae and Clusiaceae fit to the negative binomial model better than Euphorbiaceae and Meliaceae. These four families differed in their susceptibility level with 34.6% , 36.7% , 24.1% and 27.7% of trees ≥ 30 cm gbh respectively supporting lianas. At the species level, the proportion of trees with lianas was positively correlated with the mean branch free bole height of trees ≥ 30-40 and ≥ 40 cm gbh classes, and the mean number of lianas per tree was also positively correlated with the mean branch bole height of trees ≥ 30 cm gbh. Of the 16 abundant families, Euphorbiaceae contributed 31% and dominated the lower canopy, but its susceptibility to lianas was lower when compared to most other families.

The pattern of coarse-root distribution was analysed in the woody plant community along a 200-m edaphic gradient on a Kalahari sand woodland catena in Zimbabwe. The root systems of 45 trees and shrubs were excavated, mapped, and digitized to analyse rooting depth and architecture. Patterns of change in the above-ground community were also identified along this transect. Rooting depth increased, and the centre of mass of the root profile shifted towards the maximum rooting depth as a function of distance up the catena. The data also suggest that interspecific variation in rooting depth may increase up the catena. The below-ground pattern was accompanied by above-ground changes: species richness and basal area also increased up the catena. It is hypothesized that increasing soil depth allows greater coexistence of alternative water-use strategies, resulting in the observed increase in species richness up the catena.

Damage caused by elephants was monitored in two woodland stands of Zakouma National Park (Chad) between February and March 1998. The Acacia seyal savanna was more severely damaged (29.8% of trees damaged of which 13.2% severely) than the Combretaceae savanna (26.5% of trees damaged of which only 4.2% severely). Nearly all severely damaged trees showed resprouts (respectively 86.8% and 88.5% in Combretaceae and A. seyal savannas). Both low damage rate and low mortality rate indicate that no serious ‘elephant problem’ occurred in Zakouma National Park. Elephants selected trees to damage according to species and height. Such selection also occurred for severe damage in A. seyal savanna, but only height affected severity of damage in Combretaceae savanna. In both savannas, the plots close to a water point were the most frequently damaged. In Combretaceae savanna, the distribution of damaged trees was clumped. On the other hand, spatial auto-correlation was not significant for the severity or the frequency of damage in A. seyal savanna. Neither tree density nor diversity of woody species affected the spatial patterns of damage or severe damage.

A pteropodid bat community was surveyed in a West African rain forest using mist-nets set for 3 mo in 2 consecutive years (1996-97). The captures were carried out in the understorey and in the canopy for a total of 5054 mist-net hours and were analysed to answer three main questions: (1) to what extent does the bat assemblage vary along the vertical axis, (2) does the observed vertical stratification depend on wingspan (with largest bats preferring canopy openings to the cluttered understorey) and (3) was the vertical stratification repeatable from one year to the next? Nine bat species were reported, among which two were captured exclusively in the canopy where the total capture rate tended to be higher. The community structure did not differ between the four canopy stations. Four species significantly favoured canopy, two significantly favoured understorey and two were opportunistic regarding vertical stratification. Wing size and canopy-preference index were not significantly correlated and except for Eidolon helvum, index values for the 2 sampling years were correlated. Important inter-annual and seasonal variations of species richness and capture rate were observed and these are discussed in relation to rainfall patterns and migratory behaviour of the species concerned.

The orchids Dendrobium infundibulum and Cymbidium insigne are pollinated by the bumblebee Bombus eximius, which also pollinates the ericaceous shrub Rhododendron lyi, A study of the three plants and the bumblebee in maquis-like mountainous vegetation in northern Thailand suggests that they form a floral mimicry system with two mimics (the orchids) to one model (R. lyi). The flowers of the two orchid species offer no food reward to pollinators. Flowers of the three species show definite morphological and visual resemblance, and the species overlap to a great extent in flowering phenology and habitat Mimic flowers were much less numerous than model flowers, which occurred abundantly in the area. The rate of pollinaria removal was very low in both orchid species, resulting in a relatively small percentage of fruit set, especially in D. infundibulum. Infrequency of pollinator visits was, however, somewhat counterbalanced by long-lasting flowers. The pollinaria of the two orchid species are differently positioned on the bee, hence interference between the different reproductive systems is minimized. This is the first report which indicates floral mimicry from tropical Asia.

Over recent decades, studies on post-dispersal seed predation have revealed some general patterns despite dealing with a highly variable phenomenon (reviews in Crawley 1993, Hulme 1998). Most of these studies, however, were carried out with plant species whose population density range from low to moderately high (Hulme 1998). Only rarely has seed predation been studied in very dense stands of a given plant species (Schupp 1988a, b, 1992). In such a situation, seed predators might respond to the local availability of seeds in a manner that differs from their response under conditions of low seed density. For instance, a high density of seeds beneath a fruiting plant may attract seed predators, causing a positive correlation between predation and seed density (Janzen 1970). Schupp (1992), however, noted that at a population scale, a high density of seeds could satiate predators, resulting in a negative correlation between predation and seed density (see also Burkey 1994).

Although resource specialization occurs along a continuum, species are often defined as either specialists or generalists. In general, specialists are more prone to extinction than generalists and, thus, are often the first species to be lost when habitats are modified. The two-toed sloth (Choloepus hoffmanni) and the three-toed sloth (Bradypus variegatus) are arboreal herbivores distributed across the Neotropics. The two-toed sloth is considered a generalist while the three-toed sloth is more specialized. Both species inhabit shade-grown agro-ecosystems but, at least at one study site, only the two-toed sloth was viable. To quantify specialization in sloth species and explore how it influences population viability, we characterized the resource use for 68 adult and 12 subadult sloths across 3 y. The two-toed sloth used 14 tree species relatively uniformly across habitats, while the three-toed sloth largely depended on only two species of tree regardless of habitat type. Both species selected for patches of intact tropical forest, strongly avoided monocultures regardless of spatial scale and generally used cocoa similarly in proportion to availability. However, the sloth species differed in their use of cattle pastures, with the two-toed sloth selecting for pastures and the three-toed sloth avoiding them. Overall, the two-toed sloth exhibited greater plasticity in tree and habitat use, which is likely contributing to its enhanced resilience within this modified agro-ecosystem.