Abstract:We evaluated the effects of forest fragmentation on herbivory on central Amazonian trees. Levels of herbivory were measured on leaves from a total of 1200 saplings from 337 species. There was a positive and significant effect of forest fragment area on herbivore damage, with plants from continuous forest having twice as much damage as plants in the smallest fragments studied (1 ha). Measurements of herbivory rates on two species, however, indicate that the effect can be species-specific. Forest area had a positive and linear effect on rates of herbivory in Henriettella caudata (Melastomataceae), whereas in Protium hebetatum (Burseraceae), rates of herbivory were greater in the 10-ha fragments than in the 1-ha fragments and in continuous forest. There is no evidence that the nutritional and defensive characteristics of the leaves of the species we studied changed as a result of forest fragmentation, at least not in a manner consistent with the observed herbivory patterns. Herbivore predation levels, measured with artificial caterpillars, also showed no significant relationship with forest area. Therefore, neither of these top-down and bottom-up forces could explain the observed patterns of herbivory. It is suggested, instead, that forest fragmentation may affect the dispersal of insect herbivores, and reduce their abundances on small forest isolates. Altered patterns of herbivory on tree saplings may have important consequences for forest structure and dynamics.

Empirical and theoretical work has suggested that plants can change their compensatory responses to herbivory as they develop. However, such a relationship is likely to be more complex than previously thought since the amount and type of damage a plant receives can also change as the plant develops. Here, we determined the survival, growth, and reproductive output of plants (Actinocephalus polyanthus) from different ontogenetic stages that received variable levels of natural or simulated herbivore damage. Juvenile plants and nonreproductive adults in which leaves were damaged showed full vegetative compensation, whereas pre-reproductive plants were not able to replace the lost leaves. However, these same pre-reproductive plants produced more inflorescences and thus more seeds and seedlings than control plants. In contrast, damage to vegetative and/or reproductive structures during the reproductive phase resulted in a negative effect on seed and seedling production. Herbivory effects on plant survival, growth, and reproduction during the vegetative and pre-reproductive phases were independent of the amount of damage. However, during reproduction, the magnitude of these effects was strongly influenced by the amount of damage and the reproductive stage of the plant at the time of the damage. In short, our results demonstrate that the survival, growth, and reproductive responses to herbivory of A. polyanthus can be dependent on the timing and/or intensity of damage. The reproductive response of A. polyanthus to our simulated herbivory treatments during the pre-reproductive phase represents an example of overcompensation. Furthermore, it indicates that vegetative regrowth is not necessarily a driving factor for tolerance.Abstract in Spanish is available with online material.

River damming has created fragmented landscapes in parts of the Amazon basin. The resulting decrease in forest area could directly affect amphibian species if large areas of habitat are required to guarantee the presence of specific types of breeding sites. Here, we describe the anuran assemblages on islands created by damming of the Tocantins River twenty years ago in the eastern Amazon basin. We surveyed 10 undisturbed islands varying in size from 3 to 2140 ha and located at distances of up to 6.7 km from the margin of the reservoir. We identified 32 frog species, of which 15, 14, and 3 have aquatic, semiaquatic, and terrestrial development, respectively. The number of frog species increased significantly with island area but was not affected by island distance from the margin. Species with aquatic or semiaquatic development tended to be absent from the smaller islands, regardless of the degree of isolation from the mainland. These findings emphasize the need to preserve specific microhabitats on smaller land‐bridge islands to maintain amphibian diversity in reservoir environments.

Construction of hydroelectric dams in tropical regions has been contributing significantly to forest fragmentation. Alterations at edges of forest fragments impact plant communities that suffer increases in tree damage and dead, and decreases in seedling recruitment. This study aimed to test the core-area model in a fragmented landscape caused by construction of a hydroelectric power plant in the Brazilian Amazon. We studied variations in forest structure between the margin and interiors of 17 islands of 8-100 hectares in the Tucuruí dam reservoir, in two plots (30 and >100m from the margin) per island. Mean tree density, basal area, seedling density and forest cover did not significantly differ between marginal and interior island plots. Also, no significant differences were found in liana density, dead tree or damage for margin and interior plots. The peculiar topographic conditions associated with the matrix habitat and shapes of the island seem to extend edge effects to the islands' centers independently of the island size, giving the interior similar physical microclimatic conditions as at the edges. We propose a protocol for assessing the ecological impacts of edge effects in fragments of natural habitat surrounded by induced (artificial) edges. The protocol involves three steps: (1) identification of focal taxa of particular conservation or management interest, (2) measurement of an "edge function" that describes the response of these taxa to induced edges, and (3) use of a "Core-Area Model" to extrapolate edge function parameters to existing or novel situations. Keywords: core-area model, forest edges, fragmentation, hydroelectric dams Estrutura da floresta em ilhas artificiais no reservatório da Usina Hidrelé-trica de Tucuruí, Brasil: Um teste do modelo de área nuclear resumo A construção de usinas hidrelétricas em regiões tropicais tem contribuído significativamente para a fragmentação da floresta. As alterações nas bordas de fragmentos florestais causam profundos impactos na comunidade de plantas, tais como, o aumento em de árvores mortas ou danificadas e a diminuição do recrutamento de plântulas. Este estudo tem como objetivo testar o modelo de área nuclear (core-area model) em uma paisagem de floresta fragmentada resultante da formação do lago da Usina Hidrelétrica de Tucucui na Amazônia. Foram medidas as variações na estrutura em 17 ilhas, com tamanhos variando de 8 a 100 hectares. Em cada ilha foram colocadas duas parcelas, a 1ª parcela a 30 metros da margem e a 2ª parcela a mais de 100 metros da margem da ilha. A densidade e a área basal de árvores, a densidade da regeneração natural e a cobertura do dossel da floresta não foram significativamente diferentes entre as parcelas da borda e do interior das ilhas. Também não foram encontradas diferenças significativas na densidade de lianas, no número de troncos mortos ou árvores danificadas entre as parcelas da borda e do interior das ilhas. A grande declividade do terreno das ilhas, a matriz circundante (água) e a forma irregular da ilhas po...