In an effort to understand variations in ecological patterns among lowland tropical rain forests, Alwyn Gentry and colleagues synthesized data sets from four of the premier Neotropical field stations—La Selva (Costa Rica), Barro Colorado Island (Panama), Cocha Cashu (Peru), and the Biological Dynamics of Forest Fragmentation Project (Brazil). To promote the kind of geographically comparative tropical ecology advocated in the 1990 Gentry book, the Organization for Tropical Studies and the Smithsonian Tropical Research Institute organized a course in 2001 that visited each of these field stations. Papers from some of the studies resulting from this course are highlighted in this special section. These studies are notable for the consistent methods applied across forests, and they underscore the acute need and bright future for comparative tropical ecology. Key site characteristics for each of the field stations are summarized here.

We contrasted the structure and biomass of four lowland Neotropical forests (La Selva, Costa Rica; Barro Colorado Island, Panama; Cocha Cashu, Peru; and KM41, Brazil) to determine if commonalities exist within and among forests differing in latitude, rainfall, seasonality, and soil fertility. We examined the effect of soil fertility specifically by measuring the density and basal area of trees, lianas, and palms on two soil types differing in fertility at each site. We used allometric relationships to estimate the contribution of the various life-forms to total aboveground biomass (AGB) and compared two relationships for trees 30 cm diameter or greater. Estimated liana density and AGB were similar among sites, but the density and AGB of trees and palms, estimated using diameter alone, differed significantly. Basal area and AGB of trees 10 cm diameter at breast height (DBH) or greater differed among forests and averaged 30.2 m²/ha and 250 Mg/ha, respectively. Cocha Cashu and KM41 had higher tree basal area and AGB than La Selva or Barro Colorado Island. Across forests, lianas and small trees (1–10 cm DBH) each contributed between 4 and 5 percent of the total AGB and small palms contributed ca 1 percent. Many forest inventories ignore lianas, as well as trees and palms less than 10 cm DBH, and therefore underestimate AGB by ca 10 percent. Soil type had little influence on the forest structure within sites, except at Cocha Cashu where total AGB was much higher and liana density much lower on the more fertile old floodplain Entisols than the terra firme Oxisols. Although total stem density, basal area, and some biomass components differed significantly among forests, they seemed less variable than other quantitative measures (e.g., species richness).

Variation in climate and soils results in inter-site differences in the assemblages of tree life history strategies within a community, which has important implications for ecosystem structure and dynamics. I investigated interspecific and inter-site variation in wood specific gravity—an easily measured indicator of tree life history strategy—in four Neotropical forests and analyzed its correlates. Mean wood specific gravity (oven-dry weight divided by fresh volume, sometimes also referred to as wood density in the literature) differed significantly among sites, varying inversely with soil fertility and independently of rainfall, seasonality, and temperature. Mean wood specific gravity values were much higher at Kilometer 41, Manaus, Brazil, where soils are extremely poor, than at Cocha Cashu, Peru, Barro Colorado Island, Panama, or La Selva, Costa Rica, where soils are better and mortality rates of trees are higher. Within sites, wood specific gravity varied widely among species. On Barro Colorado Island, among-species variation was significantly, albeit weakly, negatively correlated with sapling and tree mortality and relative growth rates. Altogether, the results suggest that the distribution of tree life history strategies in a community varies substantially among sites, with important consequences for community and ecosystem properties such as aboveground carbon stores.

Understory light availability is an important component of the environment for canopy tree seedlings. Understanding what generates variation in understory light availability represents one step in understanding processes regulating seedling banks, the beginning capital of regeneration in tropical forests. Near the forest floor, variation in light availability may be strongly influenced by the structure and composition of the understory. I examined abundance of understory foliage and patterns of light attenuation in five Neotropical forests (La Selva Biological Station, Costa Rica = LS; Barro Colorado Nature Monument, Panama = BCI; Cocha Cashu Biological Station, Manu National Park, Peru = CC; Kilometer 41, Manaus, Brazil = KM41; and Nouragues, French Guiana = NG). I used hemispherical canopy photos to estimate light availability at two heights (0.65 and 1.7 m) above the forest floor and counted foliage between these heights. Light availability differed significantly among sites, and site rankings depended on measurement height. At 0.65 m, NG had significantly lower mean light availability than CC, with KM41, LS, and BCI equivalent and intermediate. At 1.7 m, light availability was still lowest at NG, and highest at CC, but LS had levels similar to CC while KM41 and BCI remained intermediate. Across sites, light attenuation was positively related to foliage density. LS had the most understory foliage and the highest light attenuation between 0.65 and 1.7 m. KM41, NG and CC had intermediate amounts of foliage and intermediate light attenuation. BCI had the lowest density of understory foliage and no difference between light at the two measurement heights. My results show significant effects of understory foliage on understory light and indicate that the selective environment of canopy tree seedlings differs among sites.

Even though many forest plants spend all or a significant portion of their lives in the forest understory, few studies have compared understory composition, structure, and resource availability among forests. We used standardized transect-based methods to compare small sapling densities (10–50 cm tall), understory vegetation cover, canopy openness, and nutrient availability in non-gap portions of four lowland Neotropical forests: La Selva, Costa Rica (LS), Barro Colorado Island, Panama (BCI), Cocha Cashu, Peru (CC), and north of Manaus, Brazil (KM41). Sites differed significantly in all variables except canopy openness. LS had high palm and non-fern herb cover and low density of small saplings (0.7–1.6/m²) compared to other sites. CC had high fern cover, whereas BCI had low cover in all categories of understory vegetation (palms, ferns, and non-fern herbaceous plants). BCI, CC, and KM41 had similar small sapling densities, ranging from 4.8–7.5/m². Within each forest, cation (Ca, Mg, K, and Na) availability was usually higher on more fertile soil orders (Inceptisols, Alfisols, and Entisols) than on more weathered soil types (Ultisols and Oxisols). Extractable P was highest at LS and CC and lowest on BCI (no data for KM41). Spatial autocorrelation was present for some variables in some transects to distances beyond our detection ability (>25 m). Understory palm cover was negatively correlated with small sapling density at fine (1 m² quadrat) and coarse spatial scales (among forests), although across forests the effect of palms was due entirely to the difference between LS and the other three forests. These results provide cross-site support for the hypothesis that understory cover by palms decreases the density of small saplings that comprise the advance regeneration of the forest.

Composition of pioneer tree species was recorded for 23 to 76 treefalls in the forests of La Selva (LS), Barro Colorado Island (BCI), Cocha Cashu Research Station (CC), and KM41 to evaluate the hypothesis that distance to secondary forests influences the composition of pioneer tree species at scales of thousands of hectares and to assess the relative importance of distance to seed sources compared to two microenvironmental factors: soil and gap size. Discriminant function analysis (DFA) was used to assess the relative importance of the landscape and environmental factors in controlling the presence in gaps of Cecropia species that are typical of secondary forests regenerating after large-scale disturbances. Principal component analysis (PCA) was used as an ordination technique based on pioneer species composition of each gap. The PCA ordination axes were regressed against soil type, gap size, forest type, and distance values of each site to explore the relationship between composition and control factors. At the four sites, the variable best correlated to the discriminant function of gaps with and without Cecropia was the distance to secondary forests, although such discriminant function was statistically significant at only one site. In three of the four sites (with the exception of BCI, where there are no typical young secondary forests), distance to secondary forests was correlated with at least one axis of the PCA (P < 0.06). Other significant correlations included soil variables (CC and KM41), gap size (KM41), and forest type (secondary vs. primary; BCI). Overall, the results provide some support for the hypothesis that at the spatial scale of this study, distance to young secondary forests (both natural and anthropogenic) influences treefall gap composition at a level comparable to endogenous and exogenous microenvironmental factors.

Hurricanes are important disturbance events in many forested ecosystems. They can have strong effects on both forest structure and animal populations, and yet few studies have considered the impacts on plant–animal interactions. Reduction of canopy cover by severe winds increases light availability to understory plants, providing an opportunity for increased growth. An increase in light availability should cause an increase in annual production of leaves and a more even production throughout the year (i.e., less seasonality in production). This change will affect the availability of food resources to folivorous insects that feed primarily on young leaves; outbreaks of these insects could nullify the temporary advantage of increased understory light levels. On 21 September 1998, Hurricane Georges struck Puerto Rico, providing an excellent opportunity to determine the effect of the hurricane on leaf production and herbivory in the forest understory by comparing post-hurricane data with data obtained from a previous study conducted at the same site. Eight species were analyzed at El Verde Field Station, a wet forest site in eastern Puerto Rico. For the eight species combined, there was an increase in number of leaves produced after the hurricane and a more even seasonal pattern of leaf production, as predicted. Levels of herbivory were much lower (2.03%) after the hurricane compared with pre hurricane conditions (16.05%), indicating that increased light availability to understory plants was not offset by increased herbivory. Lower levels of herbivory were possibly due to herbivore satiation, changes in leaf chemistry, changes in herbivore populations, changes in herbivore predator populations, or a combination of two or more of these factors.

Although herbivory can affect plant fitness, few studies have investigated the effects of herbivory on the fitness of plants that reproduce vegetatively via fragmentation. Plants that reproduce in this manner are common in tropical wet forest understories, where they are subject to damage from falling branches and other debris. For these species, the ability to reproduce via fragmentation may be affected by the extent of herbivory. In this study, we determined the effects of natural and artificial herbivory on plant fragments. The leaves of four tropical understory plant species were either damaged artificially or chosen based on the extent of existing damage. Effects of real and artificial herbivory ranged from strong to no effects depending on the plant species and biotic conditions. Insect-damaged fragments showed stronger effects than artificially damaged fragments. Herbivory may be an important predictor of mortality in reproductive fragments of understory plants.

This study investigated the seed dispersal of the canopy palm Attalea oleifera in a 3500 ha fragment of the Brazilian Atlantic Forest. We monitored ten fruiting palms during six months, when the following mammals were recorded feeding on Attalea fruits and/or seeds: (1) Brazilian squirrels (Sciurus aestuans, 78 diurnal records); (2) white-eared opossums (Didelphis albiventris, 8 nocturnal records); (3) black-humped agoutis (Dasyprocta prymnolopha, 3 diurnal records); (4) pacas (Agouti paca, 1 nocturnal record); (5) rice rats (Oryzomys capito, 11 nocturnal records); and (6) short-tailed opossums (Monodelphis americana, 4 diurnal records). Fruit removal on the ground reached 71.9 ± 21.9 percent after a three-month period, and most fruits were moved less than 5 m from parental palms. Squirrels generally moved fruits and endocarps 5–10 m from parents, whereas opossums, pacas, and agoutis moved 91.7, 78, and 74.1 percent of fruits less than 5 m, respectively. Oviposition on endocarps by the bruchid Pachymeros cardo ranged between 62.1 and 84.3 percent of endocarps previously manipulated by white-eared opossums, squirrels, and rice rats, whereas 57.9 percent of old endocarps collected around reproductive palms were attacked by bruchids. Moreover, oviposition on endocarps varied from 42.3 percent in the case of single endocarps to 70.9 percent of endocarps in piles. Attalea seeds abundantly germinated beneath parental palms despite the high levels of predation, and distribution of both seedlings and adults suggested that squirrels played an important role in palm recruitment at the study site.

We examined morphological and genetic data for Chirostoma grandocule, the most abundant silverside fish from Lake Pátzcuaro, Michoacán, México, as part of a larger project to clarify the genetic structure of the genus. We analyzed 19 morphometric, 8 meristic, and 11 putative enzyme-coding loci from C. grandocule samples from two different localities within the lake during three different years (Janitzio 1996, 1999, and Ichupio 1997). Because morphometric attributes were all correlated, scores derived from reciprocal averaging analysis were used to describe the variation in fish morphology. Discriminant analysis using 10 morphometric and 8 meristic variables revealed significant differences between samples (Janitzio and Ichupio, χ² = 558.4 and 531.5, P < 0.001, respectively). The utilization of morphometric features for the reclassification of organisms into the site of origin was highly effective (x̄ = 98.2%). Allozyme data resolved seven polymorphic loci (P_0.99) and levels of heterocigosis that ranged from 0.0134 to 0.0319 (x̄ H_e = 0.0234). Weir and Cockerham's F-statistics provided evidence of low but significant heterogeneity between sites and years (jackknifed θ_S = 0.0252, SD = 0.0126); a pairwise comparison of θ_S values revealed that the sample from Ichupio was highly distinct compared to the samples from Janitzio (0.0400 vs. 0.0033, P < 0.05), indicating intra-lacustrine differentiation between the north and south samples. Genetic co-ancestry distances between samples ranged from 0.0154 to 0.0341. The genetic and morphometric variation detected in C. grandocule's samples are discussed in the light of differences in sex ratio and of the contrasted environmental and physio-geographical intra-lacustrine conditions.

In the highly fragmented landscape of Los Tuxtlas, Mexico, we studied seed rain, seed bank, and vegetation established beneath the canopies of five fig trees (Ficus spp.) isolated in pastures. The area beneath the canopy of each fig tree was enclosed with a barbed wire fence to exclude cattle, and weeding with machete and herbicide spraying were stopped. Annual seed rain beneath the canopy of the five trees was 20,936 seeds (1066 seeds/m²) belonging to 149 species. Seed rain was dominated by woody species dispersed by frugivorous birds and bats. Most captured seeds (94%) were small (seed diam < 3 mm). A total of 6290 individuals from 78 species germinated from the seed bank (1258 seeds/m²). Ruderal herbs dominated the seed bank (52% of spp. and 77% of seeds). After three years of exclusion, a dense, 4–5 m tall understory comprised of woody plants developed beneath the canopy of the five fig trees. A total of 621 plants (4.6 plants/m²) from 95 species were recorded along 15 transects placed beneath the canopy of the fenced trees. Primary forest and pioneer tree and shrub species dominated species richness (55%) and abundance (37%) of established vegetation, but secondary and pioneer species had higher overall biomass. Most of the established plants (68%) had relatively large seeds (>3 mm). When cattle was excluded and management practices were stopped, remnant trees in Los Tuxtlas pastures effectively favored the early establishment of late successional species, including large-seeded canopy and subcanopy trees.

Habitat fragmentation can have consequences on plant reproduction through several interactions with insects. We studied the effects of habitat fragmentation in the proportion of aborted, predated and undamaged seeds of Acacia aroma and Cercidium praecox. We performed this study in the Chaco Serrano of Argentina. We found that fragmentation effects differed between species. In small patches, the number of aborted seeds of A. aroma increased and the number of pre-dispersal predated seeds and undamaged seeds of C. praecox decreased and increased, respectively. However, higher numbers of aborted seeds in small patches did not lead to fewer available seeds for dispersion. For these species, seed predation by bruchids was about 35 percent and aborted seeds were around 7 percent. Because of the elevated levels of damage and the high specificity of bruchids, the changes caused by habitat fragmentation may be very important for the populations of these two species. Habitat fragmentation effects on pollination and pre-dispersal seed predation of these legumes may determine higher availability of seeds for dispersion in fragmented habitats.

We investigated the impact of seed predation by large macaws (Ara spp.) on Brazil nut, the seed of Bertholletia excelsa (Lecythidaceae). Counts of macaw-damaged fruit below 50 focal trees in a Peruvian Brazil nut harvesting concession indicated that macaws destroyed about 10 percent of the concession's crop. We compared this impact to other sources of variation in profits from harvesting and suggest methods to compensate harvesters while encouraging them to conserve macaws in their concessions.

Information on an undescribed keroplatid fly (tribe Orfeliini) from Cameroon is presented. Its myrmecophagous larvae live in hollow stems of an understory myrmecophytic tree. Data on the biology of this fly and its interaction with one of the plant's strictly associated ant species Cataulacus mckeyi (Myrmicinae) showed that the number of keroplatid larvae per tree was not limited by occupation competition with the ant. Through predation on workers, keroplatid larvae may have a sizeable impact on colonies of C. mckeyi.