Search

In the age of big data, scientific progress is fundamentally limited by our capacity to extract critical information. Here, we map fine-grained spatiotemporal distributions for thousands of species, using deep neural networks (DNNs) and ubiquitous citizen science data. Based on 6.7 M observations, we jointly model the distributions of 2477 plant species and species aggregates across Switzerland with an ensemble of DNNs built with different cost functions. We find that, compared to commonly-used approaches, multispecies DNNs predict species distributions and especially community composition more accurately. Moreover, their design allows investigation of understudied aspects of ecology. Including seasonal variations of observation probability explicitly allows approximating flowering phenology; reweighting predictions to mirror cover-abundance allows mapping potentially canopy-dominant tree species nationwide; and projecting DNNs into the future allows assessing how distributions, phenology, and dominance may change. Given their skill and their versatility, multispecies DNNs can refine our understanding of the distribution of plants and well-sampled taxa in general. By modelling the distribution of the entire Swiss flora using deep learning and citizen science data, this study demonstrates a method that predicts flowering phenology and potentially dominant tree species more accurately than commonly used approaches. This approach could enable investigation of understudied aspects of ecology and refine our understanding of plant distributions. [ABSTRACT FROM AUTHOR]

Above- and below-ground herbivory are key ecosystem processes that can be substantially altered by environmental changes. However, direct comparisons of the coupled variations of above- and below-ground herbivore communities along elevation gradients remain sparse. Here, we studied the variation in assemblages of two dominant groups of herbivores, namely, aboveground orthoptera and belowground nematodes, in grasslands along six elevation gradients in the Swiss Alps. By examining variations of community properties of herbivores and their food plants along montane clines, we sought to determine whether the structure and functional properties of these taxonomic groups change with elevation. We found that orthoptera decreased in both species richness and abundance with elevation. In contrast with aboveground herbivores, the taxonomic richness and the total abundance of nematode did not covary with elevation. We further found a stronger shift in above- than below-ground functional properties along elevation, where the mandibular strength of orthoptera matched a shift in leaf toughness. Nematodes showed a weaker pattern of declined sedentary behavior and increased mobility with elevation. In contrast to the direct exposal of aboveground organisms to the surface climate, conditions may be buffered belowground, which together with the influence of edaphic factors on the biodiversity of soil biota, may explain the differences between elevational patterns of above- and below-ground communities. Our study emphasizes the necessity to consider both the above- and below-ground compartments to understand the impact of current and future climatic variation on ecosystems, from a functional perspective of species interactions. [ABSTRACT FROM AUTHOR]

Predicting variation in plant functional traits related to anti-herbivore defences remains a major challenge in ecological research, considering that multiple traits have evolved in response to both abiotic and biotic conditions. Therefore, understanding variation in plant anti-herbivore defence traits requires studying their expression along steep environmental gradients, such as along elevation, where multiple biotic and abiotic factors co-vary. We expand on plant defence theory and propose a novel conceptual framework to address the sources of variations of plant resistance traits at the community level. We analysed elevation patterns of within-community trait dissimilarity using the RaoQ index, and the community-weighted-mean (CWM) index, on several plant functional traits: plant height, specific leaf area (SLA), leaf-dry-matter-content (LDMC), silicium content, presence of trichomes, carbon-to-nitrogen ratio (CN) and total secondary metabolite richness. We found that at high elevation, where harsh environmental conditions persist, community functional convergence is dictated by traits relating to plant growth (plant height and SLA), while divergence arises for traits relating resource-use (LDMC). At low elevation, where greater biotic pressure occurs, we found a combination of random (plant height), convergence (metabolite richness) and divergence patterns (silicium content). This framework thus combines community assembly rules of ecological filtering and niche partition with plant defence hypotheses to unravel the relationship between environmental variations, biotic pressure and the average phenotype of plants within a community. [ABSTRACT FROM AUTHOR]

Plant taxonomic and phylogenetic composition of assemblages are known to shift along environmental gradients, but whether the rate of species turnover is regular or not (e.g., accelerations in particular sections of the gradient) remains poorly documented. Understanding how rates of assemblage turnover vary along gradients is crucial to forecast where climate change could promote the fastest changes within extant communities. Here we analysed turnover rates of plant assemblages along a 2500 m elevation gradient in the Swiss Western Alps. We found a peak of turnover rate between 1800 and 2200 m indicating an acceleration of grassland compositional changes at the transition between subalpine and alpine belts. In parallel, we found a peak in phylogenetic turnover rate in Poales between 1700 m and 1900 and Super-Rosids between 1900 and 2300 m. Our results suggest that changes in abiotic or biotic conditions near the human-modified treeline constitute a strong barrier for many grassland plant species, which share analogous elevation range limits. We propose that this vegetation zone of high ecological transitions over short geographical distances should show the fastest community responses to climate change from the breakdown of barrier across ecotones. [ABSTRACT FROM AUTHOR]

Monitoring and species distribution models (SDMs) are increasingly used to support conservation planning but are rarely projected at a very high resolution for conservation management. In this study, we compared the population distribution and size of five invasive plant species along an 18 km alluvial system in Switzerland, over a period of 11 years. Exhaustive inventories of past (2001) to current (2012) populations showed a massive increase in invaded areas over the eleven years. Impatiens glandulifera and Reynoutria japonica were the species with the largest increases in population number and size. The ecological preferences of each species were then modelled at 1 m resolution, using environmental variables expressing topography, disturbances, dispersal, soil texture and light availability. SDMs successfully depicted the niches at very high resolution. Some of the important predictors (e.g., canopy density, distance to river) would have been unhelpful at a coarser resolution. From these very-high-resolution models, we predicted the potential distribution and abundance of species and derived two indices indicating the amount of habitat still available for future species colonisation, crucial information for management. Large, empty areas were predicted to be suitable for each species, suggesting that the observed increase in population size may continue in the future. The two proposed range-filling indices and abundance models may be used efficiently in future studies at very fine resolution to prioritise eradication efforts in previously invaded areas and controls in areas at high risk of invasion. To our knowledge, this is the first study investigating the efficiency of SDMs to predict invasions at such a fine resolution. [ABSTRACT FROM AUTHOR]