Abstract: Cecropia (Cecropiaceae) is a Neotropical genus of pioneer plants. A review of bat/plant dispersal interactions revealed that 15 species of Cecropia are consumed by 32 species of bats. In French Guiana, bats were captured in primary and secondary forests, yielding 936 fecal samples with diaspores, among which 162 contained fruits of C. obtusa, C. palmata, and C. sciadophylla. A comparative morphological and anatomical study of fruits and seeds taken directly from herbarium specimens, bat feces, and an experimental soil seed bank was made. Contrary to previous reports, the dispersal unit of Cecropia is the fruit not the seed. Bats consume the infructescence, digest pulp derived from the enlarged, fleshy perianth, and defecate the fruits. The mucilaginous pericarp of Cecropia is described. The external mucilage production of Cecropia may facilitate endozoochory. The exocarp and part of the mesocarp may be lost after passage through the digestive tract of bats, but fruits buried for a year in the soil seed bank remain structurally unchanged. Fruit characters were found to be useful for identifying species of bat-dispersed Cecropia. Bat dispersal is not necessary for seed germination but it increases seed survival and subsequent germination. Fruit structure plays a significant role in seed longevity.

Abstract: Premise of the study: Many flowers are pollinated by potentially hungry insects, yet flowers also contain gametes and embryos which must be protected from predation. Microscopic calcium oxalate crystals in plant tissues have been proposed to protect against herbivory. Aroids, which have an unusual diversity of such crystals, also exhibit diverse pollination strategies. Many species have pollinators that do not feed while visiting the flowers, while other species, especially those pollinated by beetles, offer sterile staminodia as food rewards. We examined flowers of 21 aroid species with various pollination strategies to test the hypothesis that crystals protect vital gametes and embryos while allowing consumption of food bribes.Methods: Aroid inflorescences collected from the field or from greenhouse material were sectioned, cleared, and examined by bright field and polarization microscopy.Key results: All species examined, regardless of pollination strategy, arrayed crystals around unshed pollen and ovules. Less vital tissues, such as odoriferous appendages, had few crystals. Staminodia offered as food to beetle pollinators, however, differed greatly between species in their crystal contents. Some had minimal crystals; some had crystals in patterns suggesting they limit beetle feeding; still others had abundant crystals in no obvious pattern.Conclusions: The results are consistent with crystals protecting against insect predation of gametes and embryos. However, the role of crystals in food-bribe staminodia is unclear. They may limit and direct feeding by beetles in some species, while in others they might have no protective role. © 2012 Botanical Society of America.

Abstract: Fungus-growing attine ants use natural-product antibiotics produced by mutualist actinobacteria as 'weedkillers' in their fungal gardens. Here we report for the first time that fungus-growing Allomerus ants, which lie outside the tribe Attini, are associated with antifungal-producing actinobacteria, which offer them protection against non-cultivar fungi isolated from their ant-plants. © 2011 Springer Science+Business Media B.V.

Abstract: Over-exploitation and fragmentation are serious problems for tropical forests. Most sustainable forest management practices avoid clear-cuts and apply selective logging systems focused on a few commercial species. We applied a simulation model to estimate the impact of such selective logging scenarios on the genetic diversity and demography of four tropical tree species from French Guiana. The simulations used data on genetic and demographic composition, growth, phenology and pollen and seed dispersal obtained for Dicorynia guianensis, Sextonia rubra, Symphonia globulifera and Vouacapoua americana at the experimental site in Paracou. Whereas Symphonia globulifera serves as a model for a species with low logging pressure, the other three species represent the most exploited tree species in French Guiana. In simulations with moderate logging, typical for French Guiana, with large cutting diameter (> 60 cm diameter) and long cutting cycles (65 years), the two species V. americana and Sextonia rubra were not able to recover their initial stock at the end of the rotation period, with a large decrease in the number of individuals and in basal area. Under a more intensive logging system (cutting diameter > 45 cm diameter, cutting cycles of 30 years) that is common practice in the Brazilian Amazon, only Symphonia globulifera showed no negative impact. Generally, the differences between the genetic parameters in the control scenarios without logging and the logging scenarios were surprisingly small. The main reasons for this were the overlapping of generations and the effective dispersal ability of gene vectors in all species, which guarantee relative homogeneity of the genetic structure in different age classes. Nevertheless, decreasing the population size by logging reduced the number of genotypes and caused higher genetic distances between the original population and the population at the end of the logging cycles. Sensitivity analysis showed that genetic changes in the logging scenarios were principally determined by the growth, densities and cutting diameter of each species, and only to a very small extent by the reproductive system including factors such as pollen and seed dispersal and flowering phenology. (c) 2006 Elsevier Ltd. All rights reserved.

Abstract: This study was designed to explicitly formulate the effect of census interval and plot size on the variability of descriptors of tropical forest structure (stand density, basal area, aboveground biomass [AGB]) and dynamic (tree growth, mortality and recruitment rates, biomass fluxes). A unique dataset from a broad plot network (37.5 ha) surveyed every 2 yr over a 16-yr period was used to develop and parameterize a new statistical model predicting the coefficients of variation for each forest descriptor. More than 90 percent of the inherent variability of these coefficients was predicted using a simple model including plot size and census interval in a Bayesian modeling framework. All descriptors of forest structure varied by <10 percent for plot sizes 42 ha. Among the descriptors of forest dynamics, AGB loss was the most variable. The number of 6.25 ha plots required to estimate its mean, over a 16-yr period, within a 20 percent error of the mean remains above four. This contrasts with a relative constant flux of biomass entering the plot through tree growth and tree recruitment. Tree growth was remarkably well estimated with <15 percent variability for a 2-yr census in a plot of 2 ha. This study provides an easy method to assess dataset limitations in efforts to estimate descriptors of forest structure and dynamic, which is of primary importance to decipher any clear consequences of global change in tropical forests.