Two stands of mature evergreen wet forest in Mauritius (Mascarene Islands) were sampled for floristic composition to assess the extent of invasion by weedy exotic phanerogams (Brise Fer, 550 m alt.; Bon Courage, 200–260 m alt.). All individuals ≥ 2.5 cm dbh were recor ded in 50×2 m transects totalling 0.1 ha at each site, and 2×2 m seedling plots totalling 40 m were also sampled at each site. Both forests showed a high degree of invasion. Although exotics constituted only 5% of woody species ≥ 2.5 cm dbh at Brise Fer and 14.5% at Bon Courage, they comprised 34.8% and 20.8% of the individuals, respectively. Seedling plots at both sites were dominated by exotics, which comprised 20.6% of the species and 97.4% of the individuals at Brise Fer, and 22.2% of the species and 73.9% of the individuals at Bon Courage. Comparisons are made with Macabé forest, sampled nearly 50 years ago. These data suggest that unless steps are taken to check the spread of exotics, floristic composition at these sites will shift towards total invasion and degradation as has occurred elsewhere on Mauritius.

Patterns of species composition in a neotropical termite assemblage were studied in relation to early effects of forest fragmentation as well as other sources of environmental and spatial heterogeneity. Termite diversity surveys were carried out at three mainland forest sites, and 10 islands of varying size, in an area of lowland tropical forest that had been flooded 4 y earlier, during the creation of the Petit Saut hydroelectric project in French Guiana. The ‘ghost forest’ of dead emergent trees in the flooded zone was also studied for its influence on island termite assemblages. Results suggested that the effects of forest fragmentation upon the total assemblage, and upon soil-feeders in particular, were subordinate to the influence of forest understorey palm density, and the closely associated gradients of soil humus depth and soil pH. Moreover, gradients for these three variables were uncorrelated with forest fragmentation and probably reflected spatial environmental heterogeneity pre-dating inundation events. Nevertheless, factors associated with forest fragmentation appeared to have had a significant effect on changes in termite species composition across the study site, primarily in structuring the wood- and leaf-litter-feeder assemblage. Effects upon the latter were not apparently a result of influx of species from the ghost forest. Purely spatial variation also influenced β-diversity changes in species composition across the site. In conclusion, the effects of forest fragmentation upon termites appear to have been relatively mild compared with other faunal groups, 4 y after flooding. Nevertheless, we predict that the effects of fragmentation on termite assemblages will ultimately be negative. This study also points to the importance of Amazonian understorey palms in structuring a tropical forest termite assemblage.

Vertebrate attack accounted for 98.4% of all seed loss during the two months following dispersal. This accounted for, across all habitat types, 94% of Bursera, 76% of Spondias, 37% of Swietenia and 25% of Erythrina seeds artificially dispersed. Bursera, Spondias and Swietenia seeds in early successional habitat suffered significantly greater predation than in either older abandoned (30 y) or mature forest plots. Erythrina showed generally low loss to predation in all plots. Seeds attacked were more often removed from, rather than buried or eaten at, the dispersal site. In young secondary habitats, however, seeds of Bursera and Spondias were more frequently eaten at the site. Seeds dispersed individually rather than in larger aggregations (5, 10) were more likely to survive in mature forest and late secondary habitat. This advantage was lost in Bursera and Swietenia when they were dispersed to younger successional habitats.

At the early seedling stage, recruitment of Bursera and Swietenia was highest in the older secondary habitats. Seedlings of Erythrina showed the lowest overall losses to any of the mortality factors identified during the first two months of establishment. Seeds of forest tree species arriving in secondary habitat were more vulnerable to attack by non-flying vertebrates than in mature forest. Survival of seedlings of these species was most closely related to the moisture-conserving status of the habitat.

The dispersal of seeds by howler monkeys (Alouatta seniculus) and the role of dung beetles in determining the fate of the dispersed seeds were studied at Nouragues Station, French Guiana, during three 2-mo periods (April–May, 1995–97). Howler monkeys were observed to disperse seeds of 47 plant species. Monkey dung attracted 57 species of dung beetles (Scarabaeidae), including tunnellers (maximum size 37 mm) and ball rollers (12.6 mm). The rate of seed burial was negatively correlated with seed size (10 plant species) and positively correlated with dung beetle size (six species of tunnellers). More than 80% of small seeds (< 5 mm) and 30% of large seeds (22 mm), were buried by the largest tunnellers. Dung beetles buried 13 to 23% of seeds ranging 8–13 mm in width. The maximum average depth of burial was 28 cm and varied according to the species of dung beetle. The rate of disappearance of large dung clumps varied during the day and was completed by nocturnal dung beetles. Seeds of Chrysophyllum lucentifolium (Sapotaceae) buried at 5 and 10 cm depths did not suffer predation, but showed low germination success compared to seeds left at the soil surface. Dung beetles affected the survival and distribution of a portion of the seeds dispersed by monkeys, and their relative importance in shaping seed fate depended on seed and beetle size.

A study was conducted on the effect of extended land-use on secondary forest biomass accumulation in the Amazon. Structural measurements were made in a series of secondary forest stands, from 4–30 y old, in Brazil and Bolivia. Half of the stands were forest regrowth following clearance and only 1 y of cultivation; the other half were regrowth following 4 y or more of continuous pasture in Brazil and three or more rotations of medium-fallow agriculture in Bolivia. Above-ground live biomass was estimated using published allometric equations. Total biomass ranged from 17 to 207 Mg ha−1. Biomass of pioneer trees was poorly related to stand age, while that of later-successional trees increased linearly with age. Total biomass accumulation in Bolivia averaged 5.4 Mg ha−1 y−1 over the entire age sequence. Biomass accumulation for regrowth following short-term use was not greater than that for regrowth following medium-fallow agriculture. In Brazil, biomass accumulation averaged 9.1 Mg ha−1 y−1 over the first 12 y of regrowth and 5.9 Mg ha−1 y−1 over the entire age sequence. Biomass accumulation was significantly slower, around 5.0 Mg ha−1 y−1, for regrowth following continuous pasture than for regrowth following 1 y of cultivation.

Throughfall volume and interception of bulk precipitation events were measured during individual rain events of differing magnitudes in a primary wet tropical forest at La Selva Biological Station, Costa Rica. The relationship between canopy structure and throughfall were examined to identify key sources of spatial variation. Geostatistical analyses were also used to examine the spatial variation in throughfall, spatial autocorrelation and to determine minimum distances for independence of collectors. Throughfall volume was collected from 56 ground-based (funnel-style) collectors. Throughfall was collected for 26 separate precipitation events during July and August 1998. Per cent cover, distance to nearest tree, distance to nearest leaf were also estimated for each collection point. A weak relationship was found with per cent cover (r2 = 0.11). No relationship was found between throughfall and distance to the nearest leaf above the collector. Estimated interception was 1.88 mm (r2 = 0.94) with increased variance as bulk precipitation increased. A range distance of 45 m was estimated from variograms, strongly suggesting that large tree canopies and gaps are the source of much of the spatial variance in throughfall volume. Interception was reduced by 19% if only spatially independent collectors were used.

Secondary forests account for 40% of all tropical forests yet little is known regarding their suitability as habitat for diurnal large mammals and game birds. This is especially so for second-growth that develops on large areas of degraded land. We address this by investigating assemblages of large-bodied birds and mammals in extensive patches of secondary forest in the Jarí region of the north-eastern Brazilian Amazon, comparing species richness and abundance against that of adjacent undisturbed primary forests. We conducted 184 km of line-transect censuses over a period of 3 mo, and found that although primary and secondary forests held a similar abundance of large vertebrates, the species composition was very different. Secondary forests supported a high abundance of ungulate browsers (0.85 vs 0.44 indiv. per 10 km) and smaller-bodied primates (15.6 vs 4.6 indiv. per 10 km) compared with primary forests. However, large prehensile-tailed primates were absent (black spider monkey Ateles paniscus) or at very low abundance (Guyanan red howler monkey Alouatta macconelli) in secondary forest. The abundance of large frugivorous/granivorous birds was also low in secondary forests compared with primary forests (22.6 vs 37.1 individuals per 10 km, respectively). Faunal assemblages appear to reflect food resource availability. Concurrent vegetation surveys indicated that secondary forests had high levels of terrestrial and understorey browse. Fruit production was largely restricted to pioneer trees such as Bellucia and Inga spp. Although these regenerating forests were an important habitat for large mammals and birds, they were limited in terms of faunal richness, particularly dispersers of large-seeded plants.

The structure, composition and reproductive phenology of a bat assemblage were investigated in the tropical region of Los Tuxtlas, Mexico. A 90-km2 landscape, originally occupied by rain forest, consisted of forest fragments and natural corridors of vegetation surrounded by pasturelands. Artificial strips of vegetation also were present as live fences. Bats were sampled with mist nets set at ground level in 52 forest fragments, three corridors, nine shaded plantations, six unshaded plantations and three live fences. We sampled each site twice, each time for a 5-h period for two consecutive nights. Bats were captured and released and the species identity and reproductive condition of adult females were recorded. Annual fruiting phenology and fruit production was measured with fruit traps in a forest site in three 500 × 20-m transects. We captured 4799 bats of 39 species. At forest sites we recorded 97% of the species and captured 44% of the bats. In corridors we captured 76% of the species and 16% of the total individuals. In the plantations we captured 71% of the species and 32% of the individuals. In the live fences we captured 32% of the species and 8% of the individuals. Three species, Pteronotus parnelli, Carollia brevicauda and Glossophaga soricina, accounted for 52% of all bat captures. Fruit-eating species and fruit-eating insectivores accounted for 65% of captures. Fleshy fruits were available year-round, but a seasonal pattern was evident. Presence of lactating females seemed to be associated, in general, with rainfall and fruiting seasonality, but individual species displayed different patterns ranging from aseasonal polyoestry to seasonal bimodal polyoestry, seasonal polyoestry and seasonal monoestry. Despite extensive fragmentation of the original rain forest, a species-rich bat assemblage still exists in the landscape studied. Short distances among remaining forest fragments and man-made habitats seem to facilitate persistence and reproductive connectivity among members of this bat assemblage.

The diversity, abundance and frequency of vascular epiphytes on the lower trunk were compared between two host groups of a Mexican cloud forest: angiosperm trees (n = 72) and tree ferns (n = 28). The bark of the five most frequent host trees and the root mantle of the two tree ferns were analysed for their thickness, water content, water retention capacity and pH. A total of 55 epiphyte species and 910 individuals were found on the 27 host species. On hosts with a dbh range of 5–10 cm, epiphytes were significantly more diverse (4.3±0.9 species per host) and more abundant (12.5±2.2 individuals per host) on tree ferns than on angiosperm trees (1.9±0.2 species per host and 3.9±0.6 individuals per host). However, these differences were not significant for the dbh class of 10–20 cm, because epiphyte numbers increased on angiosperm trees with larger host size, but not in tree ferns. Most epiphyte species had no preference for any host group, but four species were significantly more frequent on tree ferns and two species on angiosperm trees. The higher epiphyte diversity and abundance on tree fern trunks of the smallest dbh class is attributed to their presumably greater age and to two stem characteristics, which differed significantly between host groups, the thicker root mantle and higher water retention capacity of tree ferns. These bark characteristics may favour germination and establishment of epiphytes.

This study describes forest succession following slash-and-burn agriculture in central Amazonia, based on four chronosequences (22 sites) of 2- to 25-y-old secondary regrowth and mature forest. Biomass accumulated in the form of a saturation curve: 25 y of regrowth restored half of the mature-forest biomass, 75% would be restored after an estimated 175 y. Biomass accumulation was accompanied by a rapid decrease in woody plant density and an increase of the positive skew of plant size distribution. Liana biomass share declined from 5.0% in young to 1.9% in old regrowth, and 1.3% in mature forest, whereas the share of palms was low (0.5%) in young regrowth and high in old regrowth (1.8%) and mature forest (2.2%). 25-y-old regrowth was similar with mature forest in respect to woody plant density and growth-form composition, but both the plant size distribution and the number of stems per plant differed strongly from mature forest. A moderate increase in land use had only minor effects on biomass accumulation, but profoundly changed structural characteristics of regrowth, pointing to a high vulnerability to degradation. Thus, the sustainability of slash-and-burn in central Amazonia may be lower than the rapid initial biomass accumulation would make it seem.

Little is known about how climatic variability affects fragmented forests and their abrupt edges. We contrasted effects of the 1997 El Niño drought between fragmented and continuous forests in central Amazonia, using long-term data on tree mortality. For 23 permanent 1-ha plots, annualized mortality rates of trees ≥ 10 cm diameter at breast height (dbh) were compared among a ‘baseline’ interval of 5-17 y before the drought, a 12-16-month interval during the drought, and a 12-13-month interval after the drought, using repeated-measures ANOVA. We also examined the size distributions of dead trees for each interval. During the drought, average annual tree mortality rose significantly in both forest edges (from 2.44% to 2.93%) and interiors (from 1.13% to 1.91%), and the magnitude of this increase did not differ significantly between edges and interiors. After the drought, tree mortality declined in all plots, but most dramatically on edges. Mortality rates were more variable over time on edges than interiors, and there was no evidence of time lags in mortality. In forest interiors, the size distributions of trees that died did not differ significantly among the three intervals. On edges, however, relatively fewer small (10-15 cm dbh) and more medium-sized (20-30 cm dbh) trees died in the post-drought interval, compared to other intervals. Moreover, forest edges lost a significantly higher proportion of large (≥ 60 cm dbh) trees than did forest interiors. These results suggest that droughts have relatively complex effects on fragmented Amazonian forests. Drought effects in our forest fragments probably were reduced by prior floristic and structural changes near edges and by adjoining regrowth forest that partially buffered edge vegetation from desiccating conditions.

Numbers of lianas ≥2 cm gbh were recorded on trees ≥30 cm gbh in two 4 ha plots of dipterocarp forest. Associations between lianas and trees of the 12 commonest families and 16 commonest species were analysed. Liana density was 882 ha-1. Fifty-seven per cent of trees supported lianas and the mean number lianas per tree was 2.1. Lianas were highly aggregated on individual trees in most families; numbers of lianas per tree followed the negative binomial distribution for the Dipterocarpaceae but not Euphorbiaceae. Dipterocarpaceae and Euphorbiaceae differed in susceptibility, with 41 and 64% of their trees respectively supporting lianas. The proportions of trees with lianas (i.e. % trees laden) were correlated between plots at the species, but not the family, level. They were not related to tree girth; differences in proportions between families and species were maintained for trees in the 30–40 cm gbh class. At the species level, proportions of trees with lianas and the mean number of lianas per tree were significantly inversely correlated with mean branch-free bole height of trees in the 30–40 cm gbh class. Increased branch-free bole height and associated branch shedding may therefore decrease the susceptibility of trees to lianas and explain the lower susceptibility in the Dipterocarpaceae than the Euphorbiaceae.

The relationship between elephant damage and leaf nutrient concentration across tree species was investigated in a semi-arid savanna in western Zimbabwe with the purpose of investigating possible nutritional factors influencing elephant feeding preferences in Kalahari sand woodlands. The presence of elephant damage was recorded in all trees above 1 m in height, and leaf samples were collected from all tree species encountered in 12 vegetation plots during the late dry season. Elephant damage was positively correlated with leaf calcium, magnesium, potassium and protein concentration, but not with sodium, phosphorus or fibre. Tree species associated with sandy soils appeared to be less preferred by elephants and to have lower nutrient concentrations than species occurring on more fertile soils, such as species associated with termite mounds. Elephant damage may suppress recruitment in preferred tree species in these woodlands. This notion was supported by (1) the fact that 44% of all trees surveyed had had their main stems broken by elephants, a situation that leads to a multi-stemmed growth form with limited vertical growth; and (2) the negative correlation between the height of damaged trees and elephant damage across species.

Seedlings from 43 native woody species belonging to different successional groups from the Tibagi River Basin, Paraná State, South Brazil were studied to obtain information on the importance of colonization by native arbuscular mycorrhizal (AM) fungi. The experiment was carried out in a greenhouse for 15 to 45 wk, with soil-mix treatments and four successional groups. The mycorrhizal dependency was 90, 48, 12 and 14% of the pioneer, early secondary, late secondary and climax species, respectively. The content of P, Ca and K was 20, 17 and 23 times greater, respectively, in the leaves of the pioneer species than in the other successional groups. The colonization by AM fungi in field was studied in seedlings of 36 native woody species collected in the interior of the forest of the Mata dos Godoy State Park, and in open area at the beginning of arboreal succession. The mycorrhizal colonization in the field was 55.5, 26.9, 6.1 and 2.2% for the pioneer, early secondary, late secondary and climax species, respectively. To assess the mycorrhizal inoculum potential, rhizosphere soil was collected in the interior of the forest and a gap in the same forest and in a cleared area abandoned for natural regeneration. The inoculum potentials and the spore number in the area at the beginning of succession were 5.6 and 53.4 times greater than in the interior of the forest. The results show that the initial growth of the woody species which take part in the initial phases of succession may be more dependent on the AM fungi, in soils poor in minerals, while those that make up the final succession phases may be less dependent. The potential of the AM fungi inoculum decreases throughout the successional process and there is a relation between the inoculum potential found in the field and the occurrence for the different habitats of the species of adult plants belonging to different successional groups.

In 1998 a survey was made of the termites of the Leuser Ecosystem (in Sumatra, Indonesia), which includes a substantial area (c. 1 million ha) of unbroken primary rain forest. Nine sites and an altitudinal gradient were sampled. Altitude had a significant effect on species richness, which declines with even a 100-m increase in altitude. Species composition too was significantly affected by altitude. Functionally, only termites in one feeding group (II, which forage outside of their nests) were significantly affected by altitude. A mid-altitude termite species assemblage was evident. Longitude also significantly affected species composition; Macrotermitinae were numerous on the western side of the Ecosystem and Termitinae and Nasutitermitinae more numerous to the east of it. This was probably due to differences in rainfall patterns across the Ecosystem.

Fleshy diaspores (fruits, seeds) comprise a large portion of the litter on the floor of tropical forests, and interactions involving litter-foraging ants and diaspores are common in these areas. In this study, the interactions between ants and non-myrmecochorous diaspores (i.e. not adapted to dispersal by ants) were surveyed along a 1.4-km transect in a restinga forest (sandy soil) on the coast of south-eastern Brazil. During 2 y of monthly samplings, 562 interactions involving 48 ant species and 44 species of diaspore (0.02–11.10 g) were recorded. Ant–diaspore associations involved a considerable part of the ground-dwelling ant community. Large ponerine ants individually removed the diaspores up to 13 m, whereas small ants (myrmicines) normally recruited workers and consumed the diaspore on the spot. Ant-derived benefits to diaspores of non-myrmecochorous plants included secondary dispersal (small to medium-sized diaspores), and increased germination success after seed cleaning by ants. Large ponerine ants such as Odontomachus chelifer and Pachycondyla striata were the main seed vectors. Seedlings of three species were associated with nests of O. chelifer. The results indicate that ants play an important role in fruit/seed biology in the restinga forest.

The floral biology, breeding system and pollination of Caryocar brasiliense were studied in the cerrado vegetation of Central Brazil. The large, yellowish-cream, brush-like flowers are pollinated mainly by glossophagine bats (Glossophaga soricina and Anoura geoffroyi). Three non-glossophagine bats (Phyllostomus discolor, Vampyrops lineatus and Carollia perspicillata) and two short probosisced hawk moths (Erinyis ello and Pseudosphinx tetrio) may also act as occasional pollinators. Caryocar brasiliense is self-compatible although it sets significantly more fruits when crossed than when selfed. The natural fruit set (fruit/flower ratio) and seed set (seed/ovule ratio) are 3.1% and 1.0% respectively. Most of the fruits and seeds are formed through the action of the flower visitors, despite the fact that about 20% of the non-visited flowers receive self pollen on at least one stigma.

Community patterns along a successional sequence of abandoned fields within a tropical rain forest environment are described in terms of floristic composition and structure. Temporal patterns were summarized and main axes of variation extracted by multivariate methods.

External factors – length of cropping period, regrowth age, and proportion of the perimeter occupied by forest communities or remnants (here called forested perimeter) – were related to floristic and structural data on 40 sites. Multivariate direct gradient analysis showed overall floristic development to be positively related to regrowth age and forested perimeter, and negatively related to length of cropping period.

Structural development was shown by multiple regression to be positively related to regrowth age and negatively to length of cropping period; but forested perimeter had no significant relation to site structure. The type of crop grown showed no significant relation with floristic or structural development of the regrowth.

These results indicate that the conditions of the vegetation around a site have a great effect on community recovery, particularly in its species composition, whilst structure seems to be affected more directly by parameters which imply time.

Hong Kong is on the northern margin of the Asian tropics. The original forest cover was cleared centuries ago but secondary forest has developed since 1945 at many sites protected from fire and cutting. There are also older forest patches maintained behind villages for reasons of ‘feng shui’, the Chinese system of geomancy. All plants >2 cm dbh were identified and measured in forty-four 400-m2 plots. Detrended correspondence analysis showed a floristic continuum, with the montane sites (>500 m) most distinct and some overlap between lowland post-1945 secondary forest and the feng shui woods. The 30–40 year-old secondary forest is dominated by Persea spp. Montane forest is similar but lacks several common lowland taxa of tropical genera and includes more subtropical taxa. The feng shui woods have the most complex structure and contain some tree species not found in other forest types. Their origin and history is obscure but we suggest that both planting and selective harvesting have had a role in their current species composition

The study presented here investigated the importance of environmental heterogeneity and clumping for the mesoscale distribution patterns of five palm species in old-growth Andean forest (Maquipucuna, Ecuador). Seedlings, juveniles and adults were recorded in 167 plots located throughout the 3-km × 3-km study area. Mountain (Cerro Sosa or Cerro Sta. Lucia), altitude, aspect, topographic-edaphic conditions and forest structure were also recorded. The following questions were investigated: (1) Are the distributions related to environmental heterogeneity? (2) Are the distribution patterns consistent with severe recruitment limitation? The distribution patterns of four species were related to environmental heterogeneity, especially to that created by the mountainous landscape itself, and at least four of the five species also exhibited strong clumping that was clearly linked to spatially restricted recruitment and could not be explained by the environmental parameters. Thus, both niche specialization and recruitment limitation probably play important roles in Andean palm community ecology. Consistent differences among the two mountains in forest structure and the frequency of certain palm stages most likely reflected extractivism.