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]

Understanding the ecological rules structuring the organization of species interactions is a prerequisite to predicting how ecosystems respond to environmental changes. While the ecological determinants of single networks have been documented, it remains unclear whether network ecological rules are conserved along spatial and environmental gradients. To address this gap, we reconstructed 48 plant–herbivore interaction networks along six elevation gradients in the Central European Alps in Switzerland, using DNA metabarcoding on orthoptera faeces. We developed hypotheses on the ecological mechanisms expected to structure interaction networks, based on plant phylogeny, plant abundance, leaf toughness, leaf nitrogen content and plant metabolomics. We show that plant phylogenetic relationships and species abundance have the greatest explanatory power regarding the structure of the ecological networks. Moreover, we found that leaf nitrogen content is a key determinant of interactions in warmer environments, while phenolic compounds and tannins are more important in colder environments, suggesting that determinants of species interactions can shift along environmental gradients. With this work, we propose an approach to study the mechanisms that structure the way species interact with each other between bioregions and ecosystems. [ABSTRACT FROM AUTHOR]

European grasslands face strong declines in extent and quality. Many grassland types are priority habitats for national and European conservation strategies. Countrywide, high spatial resolution maps of their distribution are often lacking. Here, we modelled the spatial distribution of 20 permanent grassland habitats at the level of phytosociological alliances across Switzerland at 10x10 m resolution. First, we applied ensemble models to provide distribution maps of the individual habitat types, using training data from various sources. Copernicus Sentinel satellite imagery and variables describing climate, soil and topography were used as predictors. The performance of these models was assessed based on the true skill statistics with a split‐sampling of the data. Second, the individual maps were combined into countrywide maps of the most and second most likely habitat type, respectively, using an expert‐based weighting approach. The performance of the combined map for the most likely habitat type was assessed via an independent testing dataset and a comparison of the predicted habitat‐type proportions with extrapolations from field surveys. Most individual maps had useful to excellent predictive performance (TSS ≥ 0.6). For most grid cells in the combined maps, the most and second most likely habitat types were either ecologically closely related or representing two grassland types along a nutrient gradient. The same was true for omission errors. We found good agreement between the predicted and estimated proportions from field surveys. The area of raised bogs appears to be underestimated, while dry grasslands showed highest agreement. This work highlights the potential of earth observation data at fine spatial and temporal resolution to map habitats at broad scales, thereby providing the foundation for diverse conservation applications. A particular challenge remains in capturing the transition from nutrient‐poor to nutrient‐rich grasslands, which is highly important for biodiversity conservation. [ABSTRACT FROM AUTHOR]

Ecologically meaningful predictors are often neglected in plant distribution studies, resulting in incomplete niche quantification and low predictive power of species distribution models (SDMs). Because environmental data are rare and expensive to collect, and because their relationship with local climatic and topographic conditions are complex, mapping them over large geographic extents and at high spatial resolution remains a major challenge. Here, we propose to derive environmental data layers by mapping ecological indicator values in space. We combined ~6 million plant occurrences with expert‐based plant ecological indicator values (EIVs) of 3600 species in Switzerland. EIVs representing local soil properties (pH, moisture, moisture variability, aeration, humus and nutrients) and climatic conditions (continentality, light) were modelled at 93 m spatial resolution with the Random Forest algorithm and 16 predictors representing meso‐climate, land use, topography and geology. Models were evaluated and predictions of EIVs were compared with soil inventory data. We mapped each EIV separately and evaluated EIV importance in explaining the distribution of 500 plant species using SDMs with a set of 30 environmental predictors. Finally, we tested how they improve an ensemble of SDMs compared to a standard set of predictors for ca 60 plant species. All EIV models showed excellent performance (|r| > 0.9) and predictions were correlated reasonably (|r| > 0.4) to soil properties measured in the field. Resulting EIV maps were among the most important predictors in SDMs. Also, in ensemble SDMs overall predictive performance increased, mainly through improved model specificity reducing species range overestimation. Combining large citizen science databases to expert‐based EIVs is a powerful and cost–effective approach for generalizing local edaphic and climatic conditions over large areas. Producing ecologically meaningful predictors is a first step for generating better predictions of species distribution which is of main importance for decision makers in conservation and environmental management projects. [ABSTRACT FROM AUTHOR]