The lack of recent critiques about terminology has led to the frequent misuse or confusingly varied use of the words that are more or less specific to the field of terrestrial canopy biology. I provide definitions for ca 170 terms and subterms, with translations into four languages. Rather than limit coverage to tree crowns, I define canopy biology as the study of life within any aboveground parts of all plant communities, temperate and tropical. This broadened perspective enables ecologists to consider the entire range of challenges faced by organisms living in aboveground plant life, from just above the rhizosphere to the outer limits of plant growth into the atmosphere. Further, this redefinition may reduce the potential for anthropocentric biases in interpreting life on trees or other plants; encourage the use of alternative ecosystems for hypotheses that may be difficult to address in treetops; and promote more general conceptual thinking about life on vegetation, most notably the importance of scaling in ecology. Among the salient points in terminology: the concept of “stratification” has been criticized in part because strata have been defined many ways, but a flexible application of the word is central to its utility; the source of nutrients is pivotal in distinguishing epiphytes from parasites, rather than the more general issue of an organism's effects on its host; “hemiepiphyte,” as currently used, confounds two radically different life cycle strategies, suggesting a new term, “nomadic vine,” to describe the strategy typical of many aroids; there is a confusion in the literature caused by varied applications of the word “climb;” locomotor terms may have to be modified as more becomes known about forces underlying limb kinematics; and studies of leaping and falling organisms tend to overemphasize arbitrary distinctions between gliding and parachuting to the detriment of the more critical issue of capacity for “controlled descent.”

Despite having one of the world's highest deforestation rates, the tropical forest of Rondônia state in the southwest Amazon is virtually unknown in terms of nutrient dynamics. To fill this knowledge gap, the distribution of nutrients in a humid tropical forest located in the Ecological Reserve of Samuel, Rondônia state, was addressed. A total of 474 trees with diameter at breast height (DBH) >10 cm were mapped at the Ecological Reserve of Samuel, a 20,000-ha area bordering the Samuel Hydroelectric Reservoir, which is located 50 km south of Porto Velho, Rondônia state, Amazon region, Brazil. These individuals belonged to ca 220 species distributed in 41 families. From this total, leaves of 354 trees were collected for chemical compositional analysis; boles of 96 trees and fine branches of 75 trees also were collected. Soil samples were gathered for fertility analysis at two soil depth intervals: 0–50 and 50–100 cm. In general, soils were acidic and very poor in terms of elemental contents. For plants, the highest concentrations of P, K, and Mg were found in leaves and fine branches, and the highest concentration of Ca was in the litter. The lowest nutrient concentration was observed in boles. The nutrient concentrations of leaves collected at Samuel were similar to that found in other terra firme forests of the Amazon region and other tropical forests growing on poor soils. Comparisons of nutrient concentrations among families, genera, and species revealed that only N concentrations were distinct among botanical taxa; the variability found in concentrations of other nutrients were larger and did not reveal any significant differences. Like aboveground biomass, the total nutrient stocks of trees was concentrated in a few large individuals. This fact reveals important implications about the impacts caused by selective logging. Although few trees are cut per hectare in this activity, the effects on the ecosystem structure, due to the loss of nutrients, may be underestimated.

We used a humid tropical elevation gradient to examine the relationships among climate, edaphic conditions, belowground carbon storage, and soil respiration rates. We also compared open and closed canopy sites to increase the range of microclimate conditions sampled along the gradient, and determine the effects of canopy openings on C and P storage, and C dynamics. Total soil C, the light C fraction, and all of the component fractions of the P pool were significantly related to soil moisture, and all but total soil C were also significantly related to temperature. Both labile and recalcitrant soil P fractions were negatively correlated with the light C fraction, while the dilute HCl-extractable P pool, generally thought of as intermediate in availability, was positively correlated with light C, suggesting that P may play an important role in C cycling within these systems. Total fine root biomass was greatest at 1000 m elevation and lowest at 150 m, and was strongly and positively correlated with soil moisture content. Soil respiration rates were significantly and negatively correlated with fine root biomass and the light C fraction. In forested sites, soil respiration rates were strongly and negatively correlated with total belowground C pools (soils roots forest floor). Belowground C pools did not follow the expected increasing trend with decreases in temperature along the gradient. Our results indicated that in humid tropical forests, the relationships among soil C and nutrient pools, soil respiration rates, and climate are complex. We suggest that frequent and prolonged anaerobic events could be important features of these environments that may explain the observed trends.

We used aerial photographs to measure long-term loss of forest cover in dry to mesic forests at Pu'u Wa'aWa'a on the island of Hawai'i. The aerial extent of dense forest decreased 62 percent between 1954 and 1994, and the area covered by grassland increased by 237 percent. We then used field data and modeling to evaluate the consequences of removing cattle to the composition, biomass, and flammability of the grass community. Field studies of exclosures demonstrated that grazing reduced biomass of kikuyu grass (Pennisetum clandestinum) from 770 to 229 g/m², and canopy height from 26.8 to 6.1 cm. Grazed kikuyu was unable to carry fast-spreading or intense fires (maximum flame length 70 cm, covering 1.4 ha after one hour), while ungrazed kikuyu could carry catastrophic fires (flame lengths 3.0 m, covering >75 ha after one hour). In addition, removing cattle from drier areas allowed the spread of fountain grass (Pennisetum setaceum), which created a higher risk of fire due to its architecture and low fuel moistures.

Many species of woody plants in Hawaiian montane forests germinate on nurse logs or epiphytically, often developing aerial roots. This study investigated the role of fallen logs, wood fragments, and other forms of coarse woody debris (CWD; 35 cm diam.) in providing habitat for woody species in a forest with waterlogged soils. Oxidation-reduction potentials (Eh) of root mats and nurse logs (−70.1 to 278.6 mV) were higher than the underlying soil horizons (−203.6 to −128.1 mV). CWD volume varied between 135.9 and 427.9 m³/ha. Live basal area varied between 18.4 and 29.7 m²/ha and increased with total CWD volume. Seedling and sapling abundances on nurse logs were correlated with moss coverage and decomposition class. Moss coverage was the only significant predictor of seedling density on nurse logs, whereas moss coverage and log volume were important for predicting sapling density. The proportion of woody plants established on logs was higher than in a younger Hawaiian montane forest site with well-drained soils.

Gap-phase regeneration of tropical mangroves was studied to test if there were significant environmental, structural, and floristic differences between canopy gaps and adjacent forests with intact canopies. Twenty-five canopy gaps in the low to mid intertidal zone were sampled at two study areas in northern Australia. Canopy gaps at the more equatorial study area were larger and had a higher surrounding canopy. Soil analyses showed that sediment texture exerts strong control on conductivity, pH, organic matter, total P, total N and total S. Some soil factors differed between gaps and intact forest, and PAR was significantly higher in gaps. Despite this, propagule and young seedling abundances did not differ between gaps and adjacent forests due to the dominance of viviparous species. Enhanced recruitment of seedlings to the sapling stage was detected in many canopy gaps, although saplings were also present under mangrove forest canopies. Nevertheless, the overall floristic composition, species richness, and species rank abundance were similar between canopy gaps and forests with intact canopies. There appears to be little evidence for specialized gap colonists in tropical mangrove forest as distinct from other tropical forests. Instead, gap-phase regeneration of mangroves appears to follow a simple “direct replacement” model in which a species population is replaced by members of the same species in forest gaps. The scope for gap-dependent specialization in mangroves via seed dormancy and fast growth rates, common in rain forest pioneers, may be constrained by the intertidal environment in which mangroves occur.

The relationship between canopy gaps and topography was examined in 12 ha of wet rain forest in lowland western Ecuador. Canopy gap and topographical data were recorded on a 10- × 5-m grid system. Each grid point was categorized as belonging to a low canopy gap (≤5 m height), a high canopy gap (5–10 m height), or the non-gap forest (>10 m height). Slope position, altitude, slope, and surface rock all influenced proportional gap area. The gap relationship with slope position and slope angle varied with altitude. Proportional gap area was not related to drainage. Irrespective of altitude, the plateaus had the greatest proportion of low canopy gap followed by the valleys and upper slopes. The mid-slopes and ridges had the lowest proportion of low canopy gap. The proportions of high canopy gap had an almost converse relationship. As altitude decreased, the proportion of low canopy gap increased in the valleys and on the plateaus, but decreased on the ridges and upper slopes. The relationship of slope to the two gap types also varied with altitude, although no clear trend was evident. The processes controlling the observed topographical relationship of gaps are complex, and among other factors, may be linked to the distribution of live and dead treefalls. Topographic differences in the incidence of canopy gaps can potentially influence local species' distributions.

Physical damage caused by litterfall is a source of seedling mortality that can be influenced by environmental disturbances, such as the isolation of forest fragments due to habitat destruction. The impact of physical damage in seedling mortality was evaluated in forest fragments and continuous forests in central Amazonia using artificial seedlings. A total of 11 reserves distributed among three sites were studied: three 1-ha, three 10-ha, two 100-ha, and three continuous forest reserves. In each reserve, ten transects each comprised of ten artificial seedlings were monitored for one year at four-week intervals. Survival curves were compared using failure time analysis, and the effect of reserve size on artificial seedling mortality after one year was analyzed with a categorical model. In the analysis of survival curves, artificial seedling survival was significantly greater in the continuous forest than in the 10-ha reserve at one site only, a result that may have occurred by chance. After one year, artificial seedlings suffered similar mortality due to litterfall among the reserves sampled. Physical damage by litterfall varied from 28 to 35 percent annually in the 1- and 10-ha forest fragments compared to 20 to 24 percent in the continuous forest. Mortality of artificial seedlings in the continuous forest was similar to other tropical sites in which the same method has been used, and may indicate that litterfall plays an important role in seedling mortality in the terra firme tropical forests of central Amazonia.

Isoprene (2-methyl-1,3-butadiene), a volatile organic compound produced by many plants, is the principal source of photochemically active reduced compounds in the troposphere. Emission from tropical forest trees accounts for > 70 percent of the annual global flux of isoprene, and under certain environmental conditions, trees may lose a large fraction of their fixed carbon to isoprene production. It is not known, however, if the production and emission of isoprene serve an adaptive role, or are factors that control isoprene emission from tropical trees well understood. We present results from a study investigating patterns of variability in isoprene emission and photosynthesis in two tropical wet forest tree species, Brosimum utile (H. B. K.) Pitt (Moraceae) and Dussia munda (Leguminosae/Fabaceae). We used leaf-level measurements to investigate the within- and among-individual variability in isoprene emission and photosynthetic rates, and looked at how these rates changed with light intensity. Leaves of both species showed similar responses to increases in light intensity, with photosynthesis appearing closer to saturation than isoprene emission at high light intensities. There was a large difference in both the photosynthetic and isoprene emission patterns of canopy and subcanopy leaves, with canopy leaves consistently showing much greater isoprene flux and photosynthetic rates than subcanopy leaves. This difference was smaller, although still discernible, when differences in specific leaf weight (SLW) were considered. These results suggest that tropical trees exhibit biochemical as well as structural responses to canopy position, with patterns of vertical variability that resemble those seen in temperate trees. These data have important implications for understanding the possible functional significance of isoprene emission and for creating accurate canopy-level isoprene flux models.

Propagules of the red mangrove (Rhizophora mangle L.) were subjected to a variety of growing conditions in which light level, anatomical exposure to light, application of the photosynthetic inhibitor DCMU, and genotype were used to infer possible interactions between autonomous photosynthesis and the utilization of internal reserves for the development of roots and shoots. The outgrowth of these organs from hypocotyls was significantly enhanced under conditions favorable for photosynthesis, and chlorophyll-deficient propagules exhibited reduced development compared to sibling wild types. The results suggest that photosynthesis augments processing of carbohydrates derived from maternal sources during early development and differentiation of the hypocotyl, and in combination represent an adaptive trait.

We studied the reproductive phenology of isolated populations of Spondias mombin to determine the degree of flowering and fruiting synchrony among populations and to examine spatial and temporal variability in fruit production. The study was conducted on six small islands (1.8–3.5 ha) in the Panama Canal. All individuals (10 cm in diameter at breast height (DBH) were marked in 1991 and censused each year through 1994 to record recruits and deaths. All marked trees were censused monthly for flowering and fruiting activity from April 1992 through December 1995. Spondias mombin was seasonal and highly synchronous in flower and fruit production among islands and across years. Proportions of individuals fruiting varied among islands, years, and tree size classes. There was a positive relationship between probability of fruit production and tree size. Synchronous reproductive activity in S. mombin probably was due to responses to proximate environmental cues such as fluctuations in irradiance, but other factors must have been responsible for temporal and spatial variation in reproductive performance. We suggest that this variation may have been due partly to temporal and spatial differences in pollinator abundance.

The mating system in a natural population of Shorea leprosula from Malaysian lowland dipterocarp forest was quantified by allozyme analysis using the multi-locus mixed-mating model. The population was found to be predominantly outcrossed (t_m = 0.837 ± 0.066). Variation in individual multi-locus outcrossing rates (range = 0.55–1.00) may have reflected variation in individual self-compatibility, heterogeneity in the pollen pool, differences in the mating neighborhood of individuals, or population substructure. Departure from the mixed-mating model was evident from the differences in pollen and ovule allele frequencies. Pollen pool heterogeneity and significant levels of biparental mating (t_m − t_s = 0.127, P < 0.01) indicated that the population was most probably genetically substructured. The inbreeding coefficient based on Wright's fixation index for maternal trees (F_a = −0.078) was lower than that obtained for progenies (F_e = 0.089), suggesting considerable selection against selfed progenies during growth of seedlings to reproductive size. Incidence of multiple seedlings was observed in a small proportion of seeds, but genetic analysis and genotype comparisons showed that this was not due to apomixis and may have been caused by multiple fertilization.

Insect activity often tracks moisture gradients. We studied ant activity, size, and diversity along three moisture gradients in a Panamanian rain forest. Ant activity at baits increased by 25 percent from the dry to the wet season, and > 200 percent on a topographical gradient from a ravine to an exposed plateau. Activity varied little from day to night. Three microhabitats—tree trunk, shrub, and litter—showed different responses to these three gradients. The size distribution of the species pool (N = 63) was right skewed, but the average size of ants at a bait was strongly bimodal. Ants active in moister times and places were not significantly smaller. We suggest that gradients of desiccation risk and food availability were the two most likely causes of these patterns. Two temporal niche axes (daily and seasonal) showed little species specialization, but half of the common species could be categorized as litter or plant microhabitat specialists.

Entomophagy in the riodinid butterfly tribe Eurybiini is demonstrated for the first time. Alesa amesis caterpillars and adults possess behavioral and morphological traits for feeding on Homoptera nymphs, and oviposition by A. amesis females is mediated by the combined presence of Camponotus femoratus ants and homopteran nymphs. Caterpillars are entirely entomophagous, and do not eat plant tissues. Alesa amesis caterpillars have distinct behaviors for feeding on their prey, and for soliciting and drinking honeydew secretions from homopteran nymphs. The leg lengths of entomophagous Alesa caterpillars are shown to be longer than phytophagous relatives. The legs of Alesa are used for prey handling and soliciting honeydew secretions. We suggest that elongation of the thoracic legs has been a general consequence of entomophagy in butterfly caterpillars. This study clarifies our understanding of A. amesis and its interactions with multiple species, and points to behavioral and morphological traits important to interpreting the evolution of entomophagy among caterpillars. Our observations establish the likelihood that other members of Eurybiini may be entomophagous, and suggest that entomophagy may have evolved independently among the Nymphidiini and Eurybiini.

Evaporative water loss in Anolis cristatellus from the British Virgin Islands was negatively correlated with the aridity of their habitats. Phenotypic plasticity and/or differential mortality of less well-adapted individuals allowed rapid changes within populations. Here we report the results of two studies intended to differentiate between the two processes. We took advantage of the end of a major drought to study the rapidity with which populations recovered from unusually dry conditions. Population values changed rapidly from those measured in the past. There was no correlation between long-term habitat aridity and measured water-loss rate, but the multiyear correlation between aridity and skin resistance to water loss persisted. We also conducted a common garden experiment in which animals from the wettest and driest habitats were housed under intermediate conditions for one month. We detected no change in skin resistance to water loss in any of the three populations we studied; however, the mass of lizards from Sage Mountain increased significantly during the experiment, and their water loss per unit mass decreased significantly. These results support the existence of both phenotypic plasticity and genetic differences resulting from rapid selection within populations. The magnitude and relative importance of the two, however, remain to be determined.

Between 9 and 22 January 1999, radio-tracking revealed that nine Sturnira lilium (seven females, one lactating, and two males) used hollow trees (N = 5), vine tangles (N = 2), or the bases of palm fronds (N = 1) as day roosts near Lamanai in Belize over 43 roost days. The bats roosted in hollows of four tree species, and the roost entrances ranged from 2.0 to 7.9 m above the ground. Radio-tagged individuals returned to the same roosts day after day, with the exception of a subadult female that used at least three day roosts over the course of the study. In their day roosts, S. lilium were inconspicuous, difficult to flush, and easily overlooked. Radio-tagged bats usually roosted alone and emerged significantly later than bats without radio tags.

As Neotropical forests are increasingly converted to agriculture (especially pastures), little is known about the impacts on microbial biodiversity. To assess such impacts, I compared spore abundance and diversity of arbuscular mycorrhizal fungi (AMF) in soils from lowland evergreen forests and pastures in Nicaragua and Costa Rica. Species composition, dominance–diversity curves, and Simpson's diversity indices were similar for both forest and pasture soils. Of 28 distinct fungus morphospecies, 11 produced more spores in pasture, while only 1 produced more spores in forest. According to species-accumulation curves, local AMF species richness did not significantly decline following conversion of forest to pasture. Because pastures contained a surprising abundance and diversity of AMF spores compared to native forest, a lack of mycorrhizal fungi is unlikely to limit plant succession, restoration, or reforestation in the pastures studied. At the regional scale, however, species-accumulation curves showed significantly greater gamma (G) diversity of spores in forest. In addition to these trends in diversity, species that sporulated more in pasture tended to have small spores, while the one species that sporulated more in forest had the largest spores. Similarly, only large-spored fungi (>300 µm) showed any seasonal variation in spore abundance, being more common in the wet season.

The year 2000 marks 500 years of massive destruction for the Brazilian Atlantic Forest, as a consequence of the European colonization of Brazil. Today, the Atlantic Forest is restricted to ca 98,800 km² of remnants, or 7.6 percent of its original extension. The Atlantic Forest continues to suffer under severe anthropogenic pressure, risking imminent extinction of the remaining species. Our current knowledge indicates that this complex biome contains a species diversity higher than most of the Amazon forests, and also has high levels of endemism. The 13 selected articles in this special issue present data on the natural history, ecology, sustainable management, and conservation of the Atlantic Forest. These articles represent a sample of the research conducted to date in the region and suggest avenues of future research, particularly with regard to conservation alternatives for the remaining portions of the Atlantic Forest. This special issue represents one of the first general references pertaining to the Brazilian Atlantic Forest.

The tree flora of southeastern Brazilian Atlantic forests was investigated according to two main aspects: (a) the variations in floristic composition of both rain and semi-deciduous forests were analyzed in terms of geographic and climatic variables by performing multivariate analyses on 125 existing floristic checklists; and (b) the links of both rain and semi-deciduous forests to Amazonian forests and Cerrados (woody savanna) were assessed. All analyses were performed at the species, genus, and family levels. The information obtained for the 125 forest areas was organized into an environmental database containing geographic and climatic records, and a floristic database containing binary presence records for 2532 species, 520 genera, and 106 families. Canonical correspondence analyses (CCA) were utilized to assess the relationship between geographic and climatic variables, and tree flora composition. Venn diagrams and cluster analyses were used to assess the floristic links to Amazonian forests and Cerrados. The following patterns were detected at all three taxonomic levels: (a) the differentiation between rain and semi-deciduous forests is floristically consistent and strongly correlated with rainfall regime, although transitions may be abrupt to gradual; (b) a north–south differentiation exists for both rain and semi-deciduous forests, probably caused by variations in both temperature and rainfall regime; (c) The flora of semi-deciduous forests also changes with increasing distance from the ocean and the associated increasing rainfall seasonality; and (d) elevation and associated temperatures are strongly correlated with the internal differentiation of both rain and semi-deciduous forests. To a considerable extent, the tree flora of semi-deciduous forests is a subset of the rain forest flora, probably extracting species that are able to cope with a longer dry season. There is greater floristic similarity at the species level between Atlantic rain and semi-deciduous forests than between any of these and either Amazonian rain forests or Cerrados. Nevertheless, semi-deciduous forests and Cerrados show stronger links, particularly at the generic and familial levels. Therefore, there is little floristic ground for viewing Atlantic rain forests as being closer to their Amazonian counterparts than to the adjacent semi-deciduous forests. The most appropriate view of rain and semi-deciduous forests in southeastern Brazil is that of a continuum in tree species distribution. We suggest that the definition of Atlantic forests should be as comprehensive as that of Amazonian forests.

This paper describes the phenology of leaf, flower, and fruit phenology in the Atlantic rain forests of southeastern Brazil. For 17 months, we observed the phenological patterns of trees from two Atlantic forest types at four sites: pre-montane forest (Sites I and IV; the “typical” Atlantic rain forest) and coastal plain forest (Sites II and III). All sites experience a nonseasonal, tropical wet climate, characterized by an annual rainfall usually > 2000 mm and lacking a dry season. We tested for the occurrence (or absence) of seasonal phenological patterns within each site and compared the patterns detected among the four different forest sites using circular statistics. The expected weakly seasonal phenological patterns were not observed for these forests. Flowering and leaf flush patterns of Atlantic rain forest trees were significantly seasonal, concentrated at the beginning of the wettest season, and were significantly correlated with day length and temperature. These results stress the influence that seasonal variation in day length has on ever-wet forest tree phenology. Fruiting phenologies were aseasonal in all four forests. Flowering patterns did not differ significantly among three of the four forest sites analyzed, suggesting the occurrence of a general flowering pattern for Atlantic rain forest trees.

We studied 86 species of hummingbird-pollinated flowers and their pollinators at a coastal lowland site and two highland rain forest sites in southeastern Brazil. The Jaccard index for hummingbird-pollinated floras showed greater floristic similarity between the two highland communities than between floras of high altitudes and that of near sea level. The lowland site had slightly greater richness than either of the two highland sites. Bromeliaceae was the most important family, accounting for ca 36 percent of the bird-pollinated species and comprising 33 percent of the richest flowers. The plant populations bloomed annually for the most part. Both the wet lowland and the dry highland floras showed flowering seasonality, whereas the cloud highland flora had aseasonal flowering. A hermit hummingbird was the major pollinator in the lowland community. A hermit and a non-hermit hummingbird shared most of the floral resources in the two highland communities. Hermit hummingbirds are the major pollinators of hummingbird-pollinated floras in the Atlantic rain forests of southeastern Brazil. These areas may be as rich as Central and other South American areas in hummingbird-pollinated flowers, and altogether present an older development history when compared to the North American flora.

Toucans are prominent components of the tropical American avifauna. Although these birds are very conspicuous, there are few ecological studies focusing on them. In this study, the diets of four sympatric toucans (Ramphastos vitellinus, R. dicolorus, Selenidera maculirostris, and Baillonius bailloni) were assessed by recording feeding bouts at two altitudes in the Atlantic Forest of southeast Brazil. Our results show that toucans are predominantly frugivorous birds (96.5% of the 289 feeding bouts were on fruits). In the lowlands (70 m elev.), only fruits (48 species, 27 families) were recorded, while in the highlands (700 m elev.), toucans were observed feeding on fruits (25 species, 22 families), flowers, leaves, and insects. Non-fruit items were recorded only in the highlands, most of them eaten by B. bailloni. Cecropia glaziovii and Euterpe edulis, two abundant plants in the highland and lowland sites, respectively, and Virola oleifera, a plant that produces lipid-rich arillate fruits, were eaten heavily by the toucans. The number of feeding bouts recorded for R. vitellinus in the lowlands was positively correlated with lipid content of the fruits eaten. The diameters of fruits eaten by toucans varied greatly (range = 0.4–25.0 mm). While the large Ramphastos species not only ate tiny fruits (e.g., Hyeronima alchorneoides) but also large ones (e.g., Virola gardneri), the toucanets ate piecemeal the large fruits that exceeded their gape width, suggesting that gape size did not limit the use of any fruit by the toucans at our study sites.

Given the abundance of litter-foraging ants and fallen fleshy diaspores on the floor of tropical forests, interactions involving them should be common and may render important consequences for the biology of the diaspores. In this study, we surveyed the interactions between ants and non-myrmecochorous diaspores along a 5-km transect in a lowland Atlantic rain forest of southeast Brazil. A diaspore is defined as any seed, fruit, or infructescence that constitutes the unit of dispersal of the plant. During two years of monthly samplings of naturally fallen diaspores, 886 ant–diaspore interactions involving 36 ant species and 56 different species of diaspores (range = 0.05–29.5 g) were recorded. The number of interactions was significantly and positively correlated with rainfall but not with mean temperature. The number of ant species recorded in the interactions was positively associated with both rainfall and temperature. Lipid-rich diaspores attracted a larger ant assemblage than those with lower contents of lipids. The seasonal pattern for ant–diaspore interactions in the Atlantic rain forest is predicted by well known seasonal patterns in ant activity and diaspore production. Other factors that also may affect the observed pattern are the massive and episodic fruiting of some plant species in which diaspores are especially attractive to ants, and a preference for lipid-rich arillate seeds. Interactions between ants and fallen non-myrmecochorous diaspores may be especially common in lowland rain forests in which the abundance of ants is coupled with the year-round availability of fleshy diaspores.

This paper reports a new reproductive mode in anurans, observed for the green treefrog, Hyla leucopygia, a species from the Atlantic Forest of Brazil and a member of the H. albosignata group. The reproductive mode of H. leucopygia is defined as follows: aquatic eggs and embryos in subterranean constructed nests; subsequent to flooding, feeding tadpoles in ponds or streams. We suggest that the new reproductive mode is widespread among the species in the H. albosignata group. The specialized courtship behaviors of H. leucopygia, characterized by vocalizations and a stereotyped sequence of mutual touches between male and female, are described. A review of the reproductive modes of hylid frogs is provided, with 11 modes recognized for the Atlantic Forest hylids and 12 modes for hylids around the world.

We used parsimony analysis of endemicity (PAE) to investigate the distribution patterns of South American lowland forest mammals, focusing on centers of endemism and diversity. We utilized distribution data available for species of marsupials, rodents, and primates that occur in South America. Two grids were superimposed on the distribution maps: a larger one encompassing tropical and subtropical South America, and a smaller and more precise grid covering the Atlantic Forest only. Also, the number of species per quadrat was estimated. The results support the idea that the distribution of this fauna is structured in the form of centers of endemism. In addition, there is good congruence between the areas identified here and those proposed by other authors, reinforcing the value of PAE in this kind of analysis. The hierarchical relationships among areas, however, are not clear and sometimes discordant with other biogeographic analyses.

I present a comparative analysis of the cave fauna composition for two karst areas in the Atlantic tropical rain forest: the extensively studied Ribeira Valley, southeastern Brazil, and the small and poorly known Rio Pardo region, southern Bahia state, northeastern Brazil. These regions also are compared to other Brazilian karst areas. The cave fauna from Rio Pardo appears to be more similar to the fauna of other tropical caves in the Caatinga and Cerrado domains than to the cave fauna of the subtropical Ribeira Valley. Ecological data, including probable trophic relationships, are presented, and special conservation problems related to karst areas, with examples from the Atlantic rain forest cave taxa, are discussed.

Euterpe edulis (Martius) is a single-stemmed, abundant Neotropical palm of the Brazilian Atlantic Forest (Mata Atlântica). The species produces the heart of palm, locally called palmito, which is an important source of income for forest owners. Because of its high market value, its short-term demands of the forest environment, and its important interactions with animal species, this palm is suitable for sustainable management and conservation purposes. In this paper, we present the results from a 15-year investigation of this species, synthesizing results on its demography, seed dispersion, reproductive biology, genetics, and economics. We have linked these results in a proposal for the species' sustainable management and conservation.

Muriquis (Brachyteles arachnoides) are endemic to the Atlantic Forest of southeastern Brazil, where an estimated 1200 individuals are distributed unevenly among only 19 populations. Despite the fact that over two-thirds of all muriquis live at low densities in large and protected, undisturbed forests, virtually nothing is known about the demography of these populations. Consideration of the mechanisms underlying the low extinction probabilities simulated for muriquis inhabiting the small, disturbed forest at the Estação Biologica de Caratinga (EBC), Minas Gerais, where long-term demographic and behavioral studies have been conducted, provides disturbing predictions about the viability of other populations living at lower densities. Local resource competition may be responsible for the female-biased infant sex ratio documented at the EBC under the female-biased dispersal regime exhibited by muriquis. At lower population densities, however, release from competition may lead to male-biased infant sex ratios, and correspondingly slower population growth rates. High densities at the EBC also result in frequent intergroup encounters, permitting muriquis there to avoid close inbreeding through extra-group copulations; but at low densities, opportunities to avoid inbreeding may be more limited. Finally, habitat heterogeneity at the EBC, particularly the greater dietary diversity supported by secondary and regenerating vegetation in disturbed forests, not only may support higher muriqui densities but also provide greater dietary buffers against catastrophic fruit scarcities compared to undisturbed forests. Comparative demographic data on muriqui populations inhabiting undisturbed forests are critically needed to evaluate these predictions and to develop informed management plans for this endangered primate.

We analyzed the avifauna from four areas of the Rio Doce basin (the municipal districts of Nova Era [NE], Antônio Dias [AD], Caratinga [BSC], and Marliéria [RDSP]) located in eastern Minas Gerais state. Based on captures with mist nets, 75 bird species belonging to 15 families were inventoried; however, the species composition varied greatly among the study areas. Fifty species and 231 of the 466 individuals recorded during this study were captured at NE. In the AD area, 33 species and 132 individuals were captured. The other two areas, BSC and RDSP, contributed only 103 captures and 22 species. By analyzing species composition as a function of altitude of the study areas, it was verified that a variation of 500 m altitude was sufficient in determining the difference in existing bird communities. Formicariidae species, such as Drymophila squamata, Thamnophilus punctatus, and Conopophaga melanops, were captured exclusively in the BSC and RDSP areas, located below the 500 m altitude limit while their congeners (D. ochropyga, T. caerulescens, and C. lineata) were found exclusively in the highest areas (NE and AD). Significant seasonal variations in the species composition were observed (Green's test: Q = 12.79 in NE, 12.84 in AD, 12.4 in BSC, and 20.68 in RDSP; P < 0.05), and we believe that such variations could be associated with seasonal movements between the highest and lowest areas of the Rio Doce basin. As such, the impacts of fragmentation to the Atlantic Forest along this stretch of the Rio Doce basin may have affected the natural dynamics of avian movements. The destruction of > 90 percent of the original Atlantic Forest vegetation cover already may have seriously compromised the population viabilities of birds endemic to this area.

Forest structure and tree species composition were studied in four Atlantic seasonal forest fragments ranging from 0.7 to 220 ha, isolated for more than a century in southeastern Brazil. Four physiognomic units were recognized within these fragments: (1) low forest, (2) bamboo forest, (3) high forest, and (4) mature forest. Tree density, basal area, and species diversity (Shannon-Wiener index) increased from low forest to bamboo forest to high forest to mature forest. Liana cover was highest in low forest. Gap-opportunistic species were the dominant ecological group in low forest, bamboo forest, and high forest, while shade tolerant canopy species formed the dominant group in mature forest. These structural and compositional features formed a gradient of forest development from the least developed low and bamboo forests to the most developed mature forest. Low forest is hypothesized to be in a stage of arrested succession and high forest a stage of degraded forest. A conceptual model is suggested to explain how these physiognomic units may be related to one another. Although these fragments were covered mainly by high and low forests and thus had a degraded structure, species diversity remained high in all of them. Fragments thus may undergo degradation long after isolation, even when they have been protected from human disturbance for a considerable period. Nonetheless, they also preserve important samples of local and regional floristic diversity for a long time after isolation.

The Atlantic Forest region (wide sense) includes very complex tropical environments, increasingly threatened by extensive anthropogenic conversion (>90%). Ecologically specialized, short-generation insects (butterflies) are evaluated here as indicators for monitoring community richness, landscape integrity, and sustainable resource use in the region. The > 2100 butterfly species in the Atlantic Forest region have been censused in many sites over 35 years, giving comparable daily, weekly, monthly, and long-term site lists. The 21 most thoroughly studied sites include 218–914 species, of which half can be censused in a week or less. The butterfly communities are divided into six relatively distinct faunal regions, centered in the northeast, the central coastal tablelands, the southeast coastal plain, the mountains plus interior of the southeastern states, the central plateau, and the southern states. Species richness shows the highest values in coastal mountains from 15 to 23°S. Local butterfly communities show a high turnover, with 20 to 40 percent of the species, especially small Lycaenidae and Hesperiidae, recorded only as unstable populations or “tourists.” Easily sampled species in the family Nymphalidae, and especially its bait-attracted subfamilies, are best correlated with the entire butterfly fauna and can be used as surrogates for species diversity. In most butterfly groups, species richness is well predicted by landscape connectivity alone, or by composite indices of environmental heterogeneity, natural disturbance, and (negatively) anthropogenic disturbance. Principal components and redundancy analyses showed that the richness and proportions of different butterfly groups in the local fauna are variably explained by disturbance, seasonality, temperature, vegetation, soils, and landscape connectivity. Various groups thus can be used as rapid indicators of different types of change in the community, its environment, and the landscape. Threatened and rare species also can be used as indicators of the most unique Atlantic Forest communities (paleoenvironments), which need special attention.