Abstract: High-throughput sequencing of amplicons from environmental DNA samples permits rapid, standardized and comprehensive biodiversity assessments. However, retrieving and interpreting the structure of such data sets requires efficient methods for dimensionality reduction. Latent Dirichlet Allocation (LDA) can be used to decompose environmental DNA samples into overlapping assemblages of co-occurring taxa. It is a flexible model-based method adapted to uneven sample sizes and to large and sparse data sets. Here, we compare LDA performance on abundance and occurrence data, and we quantify the robustness of the LDA decomposition by measuring its stability with respect to the algorithm's initialization. We then apply LDA to a survey of 1,131 soil DNA samples that were collected in a 12-ha plot of primary tropical forest and amplified using standard primers for bacteria, protists, fungi and metazoans. The analysis reveals that bacteria, protists and fungi exhibit a strong spatial structure, which matches the topographical features of the plot, while metazoans do not, confirming that microbial diversity is primarily controlled by environmental variation at the studied scale. We conclude that LDA is a sensitive, robust and computationally efficient method to detect and interpret the structure of large DNA-based biodiversity data sets. We finally discuss the possible future applications of this approach for the study of biodiversity. © 2019 John Wiley & Sons Ltd

Abstract: Functional traits determine an organism's performance in a given environment and as such determine which organisms will be found where. Species respond to local conditions, but also to larger scale gradients, such as climate. Trait ecology links these responses of species to community composition and species distributions. Yet, we often do not know which environmental gradients are most important in determining community trait composition at either local or biogeographical scales, or their interaction. Here we quantify the relative contribution of local and climatic conditions to the structure and composition of functional traits found within bromeliad invertebrate communities. We conclude that climate explains more variation in invertebrate trait composition within bromeliads than does local conditions. Importantly, climate mediated the response of traits to local conditions; for example, invertebrates with benthic life-history traits increased with bromeliad water volume only under certain precipitation regimes. Our ability to detect this and other patterns hinged on the compilation of multiple fine-grained datasets, allowing us to contrast the effect of climate versus local conditions. We suggest that, in addition to sampling communities at local scales, we need to aggregate studies that span large ranges in climate variation in order to fully understand trait filtering at local, regional and global scales.

Abstract: Environmental DNA (eDNA) metabarcoding has emerged as one of the most efficient methods to assess aquatic species presence. While the method can in theory be used to investigate nonaquatic fauna, its development for inventorying semi-aquatic and terrestrial fauna is still at an early stage. Here we investigated the potential of aquatic eDNA metabarcoding for inventorying mammals in Neotropical environments, be they aquatic, semi-aquatic or terrestrial. We collected aquatic eDNA in 96 sites distributed along three Guianese watersheds and compared our inventories to expected species distributions and field observations derived from line transects located throughout French Guiana. Species occurrences and emblematic mammalian fauna richness patterns were consistent with the expected distribution of fauna and our results revealed that aquatic eDNA metabarcoding brings additional data to line transect samples for diurnal nonaquatic (terrestrial and arboreal) species. Aquatic eDNA also provided data on species not detectable in line transect surveys such as semi-aquatic, aquatic and nocturnal terrestrial and arboreal species. Although the application of eDNA to inventory mammals still needs some developments to optimize sampling efficiency, it can now be used as a complement to traditional surveys.

Abstract: Soil in the tropics is high in diversity, and despite the diversity of Borneo’s forest–soil associations, there is a paucity of data on its soil properties. We investigated the differences between three soil types in the Kabili–Sepilok Forest Reserve, Sabah, Malaysia, encompassing the contrasting alluvial, sandstone and heath forest typologies. We examined the distribution of nutrients between soil types and through soil depths, and assessed the extent of spatial autocorrelation in the three soil types. We confirmed the fertility gradient from alluvial to heath forest soil found by others. Soil elemental concentrations declined in deeper horizons with the exception of exchangeable sodium and aluminium that remained constant through alluvial and sandstone soil profiles. Spatial autocorrelation was present in all three soil types and strongest in the sandstone soil. Overall, we show how bedrock, erosion, leaching and topography influence soil properties across this mosaic of soil types and note their importance in influencing tree communities and their ecological functioning.

Abstract: Selective logging in tropical forests is often perceived as a source of forest degradation and carbon emissions. Improved practices, such as reduced-impact logging (RIL), and alternative timber production strategies (e.g. plantations) can drastically change the overall carbon impact of the wood production sector. Assessing the carbon balance of timber production is crucial but highly dependent on methodological approaches, especially regarding system boundaries and temporality. We developed a temporally-explicit and territory scale model of carbon balance calibrated with long-term local data using Bayesian inference. The model accounts for carbon fluxes from selective logging in natural forest, timber plantation, first transformation and avoided emissions through energy substitution. We used it to compare prospective scenarios of development for the wood sector in French Guiana. Results show that intensification of practices, through increased logging intensity conducted with RIL and establishment of timber plantations, are promising development strategies to reduce the carbon emissions of the French-Guianese wood sector, as well as the area needed for wood production and hence the pressure on natural forests. By reducing logging damage by nearly 50%, RIL allows increasing logging intensity in natural forest from 20 m3 ha−1 to 30 m3 ha−1 without affecting the carbon balance. The use of logging byproducts as fuelwood also improved the carbon balance of selective logging, when substituted to fossil fuel. Allocating less than 30 000 ha to plantation would allow producing 200 000 m3 of timber annually, while the same production in natural forest would imply logging more than 400 000 ha over 60 years. Timber plantation should be preferentially established on non-forested lands, as converting natural forests to plantation leads to high carbon emission peak over the first three decades. We recommend a mixed-strategy combining selective logging in natural forests and plantations as a way to improve long-term carbon balance while reducing short-term emissions. This strategy can reduce the pressure on natural forests while mitigating the risks of changing practices and allowing a diversified source of timber for a diversity of uses. It requires adaptation of the wood sector and development of technical guidelines. Research and monitoring efforts are also needed to assess the impacts of changing practices on other ecosystem services, especially biodiversity conservation.