We investigated the variation of stable (δ¹³C) soil carbon isotopes in relation to depth in seven of the most important savanna areas to adjacent contiguous forests in the Amazon region. The δ¹³C of bulk organic matter in all profiles from forested sites increased with soil depth. In forest profiles from Amapá, Alter do Chão, and Roraima, the enrichment was less than 3.5‰ between deeper soil and surface layers, suggesting that C₃ plants have remained the dominant vegetation cover. On the other hand, in forest soil profiles from Humaitá and Carolina sites, the δ¹³C enrichment was greater than 3.5‰, indicating the influence of past C₄ vegetation or a mixture of C₃/C₄ vegetation (woody savanna). The surface δ¹³C values in the savanna profiles were 5–13‰ greater than the comparable forest profiles, indicating the influence of C₄ vegetation. Two kinds of isotopic distribution were observed in deeper layers. The savanna profiles at Alter do Chão, Chapada dos Parecis, and Redenção had relatively constant δ¹³C values throughout the profile, suggesting minor past changes in the vegetation composition. In profiles at Amapá, Roraima, Humaitá, and Carolina, δ¹³C values decreased with depth from the surface and converged with comparable forest values, suggesting more woody savanna in the past than exists currently.

Mineral soils from a chronosequence of landslide scars ranging in age from 1 to more than 55 years in a subtropical montane rain forest of eastern Puerto Rico were used to determine the rate at which labile P capital recovers during primary succession. Nine organic and inorganic soil P fractions were measured using the Hedley sequential extraction procedure. Deep soil cores (9 m) from a nearby site were also analyzed to determine the distribution of P fractions below the solum. Litterfall P was measured for two years in the landslide scars to estimate allochthonous litter P inputs, and published precipitation data were used to estimate annual atmospheric inputs of P to the recovering forests. In the upper solum (0–10 cm), organic matter increased with landslide age, as did resin-Pi, labile P (defined here as resin-Pi HCO₃–Pi HCO₃–Po) and total organic P. Occluded P decreased with increasing landslide age. No significant changes in P concentrations or pools were observed in 10 to 35 or in 35 to 60 cm depth intervals across the chronosequence. Labile soil P increased to approximately two-thirds of the pre-disturbance levels in the oldest landslide scar (>55 yr). Thus, plants, their associated microflora/fauna, and P inputs from off-site substantially altered the distribution of soil P fractions during forest recovery. Across the chronosequence, the increase in labile P accumulated in soil and biomass appeared to be greater than the estimated allochthonous inputs from litter and precipitation, indicating that as the forest developed, some occluded P may have been released for use by soil biota. Resin-Pi and labile P were correlated with soil organic matter content, suggesting, as in other highly weathered soils, organic matter accumulation and turnover are important in maintaining labile P pools. Primary mineral P (apatite) was scarce, even in deep soil cores.

The control of vegetative phenology in tropical trees is not well understood. In dry forest trees, leaf abscission may be enhanced by advanced leaf age, increasing water stress, or declining photoperiod. Normally, it is impossible to dissect the effects of each of these variables because most leaves are shed during the early dry season when day length is near its minimum and leaves are relatively old. The 1997 El-Niño Southern Oscillation caused a ten-week long, severe abnormal drought from June to August in the semi-deciduous forests of Guanacaste, Costa Rica. We monitored the effect of this drought on phenology and water status of trees with young leaves and compared modifications of phenology in trees of different functional types with the pattern observed during the regular dry season. Although deciduous trees at dry sites were severely water stressed (Ψ_STEM < −7MPa) and their mesic leaves remained wilted for more than two months, these and all other trees retained all leaves during the abnormal drought. Many trees exchanged leaves three to four months earlier than normal during the wet period after the abnormal drought and shed leaves again during the regular dry season. Irrigation and an exceptional 70 mm rainfall during the mid-dry season 1998/1999 caused bud break and flushing in all leafless trees except dormant stem succulents. The complex interactions between leaf age and water stress, the principal determinants of leaf abscission, were found to vary widely among trees of different functional types.

Phenological studies were carried out in gallery and montane forests along the Lençóis River in the Chapada Diamantina mountains, state of Bahia, Brazil, between January 1993 and March 1998. Leaf flush, leaf fall, flowering, and fruiting were followed in 54 tree species (N = 243 trees tagged along trails following the river margin and the valley sides). Both gallery and montane forests exhibited a moderate peak in leaf fall during the dry season (August–October), as well as increased leaf flushing in the rainy season (December–April). Three basic modes of leafing activity were noted, although most trees in the gallery forest and in the montane forest were evergreen (82.3 and 57.8% relative frequency, respectively). In both forest types, flowering and fruiting modes were annual, with peaks of activity during the rainy season. The predominantly evergreen mode of leaf fall and leaf flush found in this study was most similar to that seen in tropical wet forests, while the flowering and fruiting modes were more similar to that of dry forests.

The composition, structure, and ecology of epiphyte communities have been intensely studied in tropical rain forests. Studies involving epiphytes, however, are scarce in other vegetational physiognomies. This study examined the composition of an epiphyte community on Vellozia piresiana (Velloziaceae) in a Brazilian rupestrian field, and investigated the factors influencing their abundance and distribution. In July 1998, 98 individuals of V. piresiana were sampled inside three transects in rocky outcrops of Serra do Cipó, Minas Gerais, and their height were measured. All epiphytes growing on these phorophytes were counted and identified. Height and branch diameter at which each epiphyte individual occurred were recorded. Luminosity intensity was measured in the upper canopy, middle canopy, and 10 cm above ground 100 phorophytes in the same population; 712 epiphytes were found growing on V. piresiana, belonging to six species: Vrisea oligantha (Bromeliaceae; 344 individuals), Epidendrum saxatile (251), Constancia cipoencis (50), Prosthechea vespa (31) Sophronites brevipedunculata (30; all Orchidaceae), and Tillandsia usneoides (1; Bromeliaceae). A higher abundance of epiphytes occurred at intermediate heights on the phorophytes, probably because of an intermediate light incidence at these positions. We found epiphytic species across different distributions, both in terms of height and branch diameter of V. piresiana, which may have reflected distinct degrees of tolerance to drought and excessive luminosity. These results indicated that despite the low richness observed in this study, the epiphyte community on V. piresiana presented ecological features similar to that observed for tropical rain forests (e.g., phorophyte partitioning and differential vertical distribution).

The reproductive biology of Pterocarpus macrocarpus Kurz (pradu) was studied in 37-year-old plantation trees in Thailand to determine the causes of seed and fruit loss. Trees flowered at the end of March or early in April at the end of the hot dry season and start of the rainy season. Flowering occurred over about a one-month period. Fruits developed over the next six months during the rainy season and matured at the start of the cool dry season in October and November. Phenology was similar in the four trees that were studied in detail. Racemes averaged 30 flowers each and each raceme was receptive for several days, although each flower was only receptive for one day. After pollination, floral parts were shed over several days and fruits began to develop. Pradu is entomopholous but its insect pollinators were not identified. The stigma is covered by hairs and a secretion is produced. A high proportion of flowers were pollinated. Then, there was a rapid loss of flowers and young fruits. These observations and earlier genetic studies indicate the probability of a high level of self-incompatibility in this predominantly outbreeding species. Pradu may have a very late-acting self-incompatibility mechanism found in many other hardwoods. The zygote remains quiescent for six weeks as the endosperm develops. During this time most of the ovules and fruits abort, suggesting resource allocation preferentially to cross-pollinated ovules. Pradu has a high reproductive potential but a low preemergence reproductive success (0.8), which is common for many hardwood species. The major cause of the low reproductive success was fruit loss during early development. Fruit production may be enhanced by increased cross-pollination among unrelated parent trees. This may be accomplished in seed orchards and seed production areas by the introduction of additional insect pollinators that travel greater distances between trees and by the relatively close spacing of unrelated parent trees.

Observations of hummingbirds feeding at flowers longer or shorter than their bills seem to contradict the view that bill lengths of hummingbirds evolved in concert with the lengths of their flowers. Recent experiments, however, indicate that a hummingbird's ability to feed at artificial flowers of different lengths depends on the widths of the flowers. We examined if the broad range of flower lengths visited by many hummingbird species can be explained by the widths of the flowers. We predicted that both short- and long-billed hummingbirds would include long, wide flower species in their diets, but that short-billed hummingbirds would not include long, narrow flower species because nectar in these species might be beyond the reach of their bills. If so, the slope of the regression for flower width versus flower length should be smaller for flower species visited by longer-billed hummingbirds relative to those visited by shorter-billed hummingbirds. Analyses of data sets for some North American and Monteverde hummingbirds and their food plants were consistent with this prediction, and bill lengths were significantly correlated with the slopes of the regressions of flower width versus length for seven hummingbird species. Comparisons of observed flower use by some Monteverde hummingbird species to flower assemblages generated at random suggest that these significant regressions were not simply a result of allometric relationships between flower lengths and widths, but in some cases reflected active choice by the birds. The two hummingbird–flower data sets also differed significantly in the scaling of corolla width relative to corolla length. In particular, the Monteverde data set contained a large number of long, narrow flower species, which we suggest is a consequence of a different floral evolutionary history and association with long-billed hummingbird species. The evolutionary effects of hummingbirds and their flowers upon one another are more complex than has generally been realized, and a consideration of corolla length jointly with other floral characters may improve our understanding of hummingbird–flower relationships.

Since the discovery of Drosophila pseudoobscura in the tropical highlands of the Colombian Andes during the 1960s, this population has been studied by many evolutionary biologists because of its geographical isolation from the main North American range of this species. We used five highly variable microsatellite loci (DPSX001, DPS2001, DPS3001, DPS3002, and DPS4001) to analyze the genetic structure of three Colombian populations and the genetic relationships with four North American populations. We found that the average heterozygosity was consistent among the three tropical Colombian populations (H = 0.665–0.675), but they had less variability than their North American counterparts. Nonetheless, the genic diversity found in the Colombian populations was higher than that found previously using other genetic markers. The average genic heterogeneity estimate among the Colombian populations (R_ST = 0.042), although statistically significant, was substantially lower than that found among the North American populations (R_ST = 0.088). We identified alleles in the Colombian populations not reported in North American populations, suggesting further divergence between the populations. We estimated that the populations on the two continents diverged ca 80,000 years ago, consistent with independent sequence analyses of these populations but contrary to some suggestions in the literature. Finally, we estimated an average effective population size of the Colombian populations to be on the order of 100,000.

In some leaf-cutting ant species, minim workers ride on the fragments of leaves as they are carried back to the nest from the cutting site. There is convincing evidence that these “hitchhikers” can protect the leaf carriers from attack by phorid (Diptera: Phoridae) parasitoids, but we consider the possibility of other functions for the hitchhiking behavior. It has been hypothesized that the hitchhikers (1) feed on leaf sap from the edges of the cut leaves; (2) ride back to the nest to save energy; (3) get caught on the fragments as they are cut, and hitchhike because they cannot (or will not) get off; and (4) begin the process of preparing the leaf to enter the fungal gardens in the nest, perhaps by removing microbial contaminants. We observed hitchhikers of Atta cephalotes in 14 nests at the La Selva Biological Station in Costa Rica. There was no difference in the proportion of leaf carriers with hitchhikers between day and night. Because the nests we observed were largely nocturnal, more than 90 percent of the hitchhiking occurred at night. The phorid parasitoids are usually considered to be diurnal, so the preponderance of nocturnal hitchhiking suggests other functions in addition to parasitoid defense. Hitchhikers spent more time in the defensive head-up posture during the day, but spent more time in the head-down posture at night. The head-down posture may indicate cleaning or other leaf preparation. The hitchhikers were never observed feeding on sap. Hitchhikers frequently got onto and off of the fragments, and so they were not “marooned.” Few hitchhikers rode all the way back to the nest and were often moving on the leaf fragment; these observations make the energy conservation hypothesis less likely, although we cannot reject it. We conclude that parasitoid defense is an important function of hitchhiking but also that there are probably other functions when parasitoids are absent. Based on available data, the most likely possibility is preparation of the leaf fragment before it enters the nest.

Neotropical herpetofaunas have been studied at selected Central and South American sites, but intra-regional analyses in the Neotropics have not been possible due to lack of long-term data. A particular problem in interpreting herpetofaunal communities is the large geographic distances between the known localities. Such sampling does not take into account the patchiness of the habitat and this prevents analysis of regional diversity. In this study, five Neotropical herpetofaunal communities within a small geographic area (1600 km²) were examined to determine if species composition varies microgeographically. We sampled sites within the Tambopata Province, southeastern Peru. The five sites were surveyed intensively during a two-year period. The herpetofauna of the Tambopata region includes 210 species and, because of extensive sampling, has the highest overall herpetofaunal richness of any known region. Sites were compared using the coefficient of biogeographic resemblance (CBR). The CBRs were similar among all sites but the differences were determined primarily by their positions with respect to the two main rivers of the region and not directly correlated to distance among sites. Sites were also examined by partitioning their species into six microhabitat classes. Sites were found to differ in their proportions of reptiles and amphibians in each microhabitat class. The results of our study were compared to former investigations of various sites within the Tambopata region and elsewhere. This study identified more species than past investigations in all cases. Species composition was shown to vary at a microgeographic level. Based on our findings, we recommend that future studies examine multiple sites within each region to fully comprehend the herpetofaunal communities that exist in complex rain forest habitats.

Distress calls are loud, harsh calls given by some species of birds when they are captured by a predator or handled by humans. We recorded the frequency of distress calls and struggling behavior in 40 species of birds captured in mist nets during the dry season in a Costa Rica cloud forest. We tested the following hypotheses proposed to explain the function of distress calls: (1) calling for help from kin or reciprocal altruists; (2) warning kin; (3) eliciting mobbing behavior; (4) startling the predator; and (5) distracting the predator through attraction of additional predators. Our results did not support the calling-for-help, warning kin, or mobbing hypotheses. Indeed, genera that regularly occurred with kin or in flocks were not more likely to call than non-flocking genera. There was no relationship between calling frequency and struggling behavior as predicted by the predator startle hypothesis. Genera of larger birds tended to call more than smaller birds, providing some support for both the predator distraction hypothesis and predator startle hypotheses. Calls of higher amplitude may be more effective in startling the predator. Distress calls of larger birds may also travel greater distances than those of smaller birds, supporting the predator manipulation hypothesis, but this requires further testing.

Lesser bare-backed bats (Dobsonia minor [Pteropodidae]) are solitary and roost in foliage of understory and subcanopy trees in lowland rain forest. These 70–90 gram frugivorous bats forage in primary and secondary forest and in abandoned gardens. At the Kau Wildlife Area in Papua New Guinea, movements (N = 1041) of four males and four females fitted with radio transmitters were monitored for 1 to 18 months. Mean home range within 30-day sampling periods was 5.1 ha (N = 12). There were no significant differences in home ranges by sex or by dry–wet season. Females, however, had significantly larger mean core-use areas than males (1.43 ± 0.61 and 0.65 ± 0.16 ha, respectively). There was moderate overlap in home range and core-use areas among some simultaneously tracked animals. The long axes of home ranges varied from 150 to 1150 m and the mean was significantly larger in females. Individuals commuted from day roosts to multiple feeding areas, sometimes resulting in disjunct core-use areas and home ranges. Fruits of native Ficus species and the exotic shrub Piper aduncum were staple food items. Piper aduncum grew as dense clusters within early successional habitats, and individual plants ripened 5–20 fruits per night throughout the year. Ficus spp. grew in primary and secondary forest and fruited asynchronously, but individual trees produced tens to thousands of ripe fruits over 7 to 10 days. Three adult female D. minor were tracked over multiple periods spanning 2.5–18 months. Although each female continued to visit a core-use area containing P. aduncum throughout the study, turnover of other core-use areas reflected the ephemeral locations of fruiting fig trees.

The diurnal time budget of four herds of white-lipped peccaries (Tayassu pecari) was studied using direct observation. The study took place from July 1996 to April 1997 in Corcovado National Park, southwestern Costa Rica. We predicted that time budget would be diagnostic of peccaries' welfare regarding food abundance. We considered that a decrease in time spent resting and in social interactions would be indicative of nutritional stress. Peccaries spent 34 percent (±1.7) of their daytime eating, 33 percent (±1.8) moving, 28 percent (±2.3) resting, 3 percent (±0.25) in social interactions, and 2.6 percent (±1.4) in other activities. The time spent resting decreased as the time spent moving and eating increased during the months of fruit scarcity. There were no differences among habitats in the proportion of time used for each activity. In the wet season, peccaries spent more time eating than in the dry season, probably because of the considerable time allocated to rooting. The monthly variation in time spent on social interactions and the frequency of agonistic interactions seemed to be related to breeding rather than fruit availability. While in the study area, peccaries traveled the longest distance in October, which coincided with the lowest fruit availability. In mid-November, the radio-marked herds left the study area and returned in January. Analyses of the peccaries' time budget suggest that fruit scarcity at the end of the wet season affected the peccaries' behavior and probably induced them to travel long distances in search of food.

This paper examines dung beetle communities in remnant patches of tropical deciduous forest at Veracruz, Mexico, as a case study of the effects of tropical deforestation on biodiversity. The two study areas have a common biogeographic history and similar macroclimatic conditions and have been modified by human activities to different extents. The main difference between them is that cattle, which provide the dung beetle's food supply, are present in only one of the areas. Comparison of the dung beetle faunas sheds light on the relative importance of forest cover versus food supply as the principal factor determining the structure and diversity of the fauna. This comparison, which we broaden through an examination of data from other localities in central Veracruz, permits us to speculate about what happens to biodiversity when a tropical deciduous forest undergoes modification of varying type and intensity. Where tree cover has been most modified, native forest species undergo local extinction and are replaced by open area species. On the whole, there has not been a net reduction in species richness (gamma diversity) in the fragmented landscape of central Veracruz, although local species richness (alpha diversity) has diminished.

We assessed species richness and composition of ant assemblages in adjacent montane forest and secondary (anthropogenic) grassland habitats in the central plateau of Madagascar. We used five quantitative methods (leaf litter sifting, two types of pitfall traps, beating low vegetation, and soil digging) and compared methods within and across habitats. Sample-based and occurrence-based accumulation curves demonstrated that the efficiency of ant inventory methods is habitat specific. Litter sifting, however, was the single most efficient method in both habitats. Overall, our analyses of the relative efficiency of methods recommend the use of sifting and beating in the montane forest site, and sifting alone in the grassland site. In four of five methods, more species were collected in the grassland site (31 spp.) than in the forest site (26 spp.). Occurrence-based accumulation curves based on all methods demonstrated that species richness was similar in the two habitats, reaching a maximum difference of approximately one species. Only five species were shared between the grassland and forest sites. The presence of a high number of ant species restricted to the grassland site (18 spp.) is the first record of high endemism in this habitat in Madagascar and may have strong implications for the reconstruction of the natural vegetation types at the time humans arrived. Their presence suggests that a comparable open habitat, such as montane woodland, shrubland, or thicket, was present on Madagascar long before humans developed the secondary grasslands less than 2000 years ago. These results are contrary to the “classical hypothesis” that the central plateau was a continuous region of closed forest. These results support the hypothesis that the montane regions, including the central plateau, once contained areas of habitat with an open structure and that the endemic ants now found in the secondary grasslands were originally native to such a habitat.