Plant litter decomposition is one of the most important processes in terrestrial ecosystems, as it is a key factor in nutrient cycling. Decomposition rates depend on environmental factors, but also plant traits, as these determine the character of detritus. We measured litter decomposition rate for 57 common tree species displaying a variety of functional traits within four sites in primary and four sites in secondary tropical forest in Madang Province, Papua New Guinea. The phylogenetic relationships between these trees were also estimated using molecular data. The leaves collected from different tree species were dried for two days, placed into detritus bags and exposed to ambient conditions for two months. Nitrogen, carbon and ash content were assessed as quantitative traits and used together with a phylogenetic variance– covariance matrix as predictors of decomposition rate. The analysis of the tree species composition from 96 quadrats located along a successional gradient of swidden agriculture enabled us to determine successional preferences for individual species. Nitrogen content was the only functional trait measured to be significantly positively correlated with decomposition rate. Controlling for plant phylogeny did not influence our conclusions, but including phylogeny demonstrated that the mainly early successional family Euphorbiaceae is characterized by a particularly high decomposition rate. The acquisitive traits (high nitrogen content and low wood density) correlated with rapid decomposition were characteristic for early successional species. Decomposition rate thus decreased from early successional to primary forest species. However, the decomposition of leaves from the same species was significantly faster in primary than in secondary forest stands, very probably because the high humidity of primary forest environments keeps the decomposing material wetter.

Networks are a convenient way to represent many interactions among ecological entities. The analysis of ecological networks is challenging for two reasons. First, there is a plethora of measures that can be applied (and some of them measure the same property). Second, the implementation of these measures is sometimes difficult. We present ’EcologicalNetworks.jl’, a package for the ‘Julia’ programming language. Using a layered system of types to represent several types of ecological networks, this packages offers a solid library of basic functions which can be chained together to perform the most common analyses of ecological networks.

There is growing evidence that top–down biotic factors play an important role in plant community dynamics and are able to maintain the high plant diversity of primary tropical forests. However, the top–down impact by herbivores on the accumulation of plant biomass, richness and the community assembly process remains poorly known for the initial stages of secondary ecological succession. Here we test the top–down effects of natural enemies on the biomass, richness, diversity, community composition, trait dynamics and randomness during the assembly process on an early successional community in a lowland tropical forest in Papua New Guinea. We initiated secondary succession on 36 vegetation plots (5 × 5 m) organized in six blocks. After 1 year we evaluated successional community characteristics in each block containing a control plot and five treatment plots: experimentally excluding (a) insects by insecticides, (b) fungal pathogens by fungicides, (c) vertebrate predators and ants with exclosures and traps and experimentally adding (d) low and (e) high levels of a generalist herbivore species. Insects were responsible for increased plant diversity, shifts in plant community composition and a decrease in community weighted mean of leaf dry matter content. The lack of herbivores resulted in a diversifying effect but did not affect plant biomass and stem density. No other treatment had a significant effect on plant diversity during succession. The highest level of artificially increased herbivore density reduced plant biomass, while the fungicide treatment decreased stem density of woody plants. Using a null model approach, we showed that the removal of insects and their vertebrate predators increased randomness during the community assembly processes for woody plants, whilst the same effect was not found for the removal of fungi or elevated herbivore abundances. Synthesis. Early rainforest succession was not entirely plant driven. During the early succession of lowland tropical forest, insect herbivores enhanced plant diversity and caused shifts in community structure by promoting species with acquisitive leaf traits. As expected, the early successional communities were highly variable, but insects and vertebrate predators reduced randomness during the assembly process. Insect and vertebrate communities can thus impact the regeneration dynamics of tropical forests. In contrast, fungal pathogens were less important in maintaining high diversity of successional communities in our system.