Search

Plant–pollinator interactions are ecologically and economically important, and, as a result, their prediction is a crucial theoretical and applied goal for ecologists. Although various analytical methods are available, we still have a limited ability to predict plant–pollinator interactions. The predictive ability of different plant–pollinator interaction models depends on the specific definitions used to conceptualize and quantify species attributes (e.g., morphological traits), sampling effects (e.g., detection probabilities), and data resolution and availability. Progress in the study of plant–pollinator interactions requires conceptual and methodological advances concerning the mechanisms and species attributes governing interactions as well as improved modeling approaches to predict interactions. Current methods to predict plant–pollinator interactions present ample opportunities for improvement and spark new horizons for basic and applied research.

Deep learning and particularly Convolutional Neural Networks (CNN) in concert with remote sensing are becoming standard analytical tools in the geosciences. A series of studies has presented the seemingly outstanding performance of CNN for predictive modelling. However, the predictive performance of such models is commonly estimated using random cross-validation, which does not account for spatial autocorrelation between training and validation data. Independent of the analytical method, such spatial dependence will inevitably inflate the estimated model performance. This problem is ignored in most CNN-related studies and suggests a flaw in their validation procedure. Here, we demonstrate how neglecting spatial autocorrelation during cross-validation leads to an optimistic model performance assessment, using the example of a tree species segmentation problem in multiple, spatially distributed drone image acquisitions. We evaluated CNN-based predictions with test data sampled from 1) randomly sampled hold-outs and 2) spatially blocked hold-outs. Assuming that a block cross-validation provides a realistic model performance, a validation with randomly sampled holdouts overestimated the model performance by up to 28%. Smaller training sample size increased this optimism. Spatial autocorrelation among observations was significantly higher within than between different remote sensing acquisitions. Thus, model performance should be tested with spatial cross-validation strategies and multiple independent remote sensing acquisitions. Otherwise, the estimated performance of any geospatial deep learning method is likely to be overestimated.

The study of mutualistic interaction networks has led to valuable insights into ecological and evolutionary processes. However, our understanding of network structure may depend upon the temporal scale at which we sample and analyze network data. To date, we lack a comprehensive assessment of the temporal scale‐dependence of network structure across a wide range of temporal scales and geographic locations. If network structure is temporally scale‐dependent, networks constructed over different temporal scales may provide very different perspectives on the structure and composition of species interactions. Furthermore, it remains unclear how various factors – including species richness, species turnover, link rewiring, and sampling effort – act in concert to shape network structure across different temporal scales. To address these issues, we used a large database of temporally‐resolved plant–pollinator networks to investigate how temporal aggregation from the scale of one day to multiple years influences network structure. In addition, we used structural equation modeling to explore the direct and indirect effects of temporal scale, species richness, species turnover, link rewiring, and sampling effort on network structural properties. We find that plant–pollinator network structure is strongly temporally‐scale dependent. This general pattern arises because the temporal scale determines the degree to which temporal dynamics (i.e. phenological turnover of species and links) are included in the network, in addition to how much sampling effort is put into constructing the network. Ultimately, the temporal scale‐dependence of our plant–pollinator networks appears to be mostly driven by species richness, which increases with sampling effort, and species turnover, which increases with temporal extent. In other words, after accounting for variation in species richness, network structure is increasingly shaped by its underlying temporal dynamics. Our results suggest that con

Most studies of plant–animal mutualistic networks have come from a temporally static perspective. This approach has revealed general patterns in network structure, but limits our ability to understand the ecological and evolutionary processes that shape these networks and to predict the consequences of natural and human‐driven disturbance on species interactions. We review the growing literature on temporal dynamics of plant–animal mutualistic networks including pollination, seed dispersal and ant defence mutualisms. We then discuss potential mechanisms underlying such variation in interactions, ranging from behavioural and physiological processes at the finest temporal scales to ecological and evolutionary processes at the broadest. We find that at the finest temporal scales (days, weeks, months) mutualistic interactions are highly dynamic, with considerable variation in network structure. At intermediate scales (years, decades), networks still exhibit high levels of temporal variation, but such variation appears to influence network properties only weakly. At the broadest temporal scales (many decades, centuries and beyond), continued shifts in interactions appear to reshape network structure, leading to dramatic community changes, including loss of species and function. Our review highlights the importance of considering the temporal dimension for understanding the ecology and evolution of complex webs of mutualistic interactions.

Ecosystem service research is high on the policy agenda. Strategies to synthesize individual success stories and derive generalized results to provide guidance for policymakers and stakeholder is central to many science-policy initiatives, such as Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services and The Economics of Ecosystems and Biodiversity. However, generalization requires the documentation of basic information on methods and results of case studies, which might not be present throughout all case studies. We used a quantitative review based on a random sample of studies published in the ISI Web of Knowledge between 1996 and 2016 to identify blind spots in ecosystem service research that might hinder the generalization. We structured our analysis along critical questions about five facets that characterize the holistic ideal of ecosystem services research: (i) social-ecological validity of ecosystem data and models, (ii) consideration of trade-offs between ecosystem services, (iii) recognition of off-site effects, (iv) comprehensive and shrewd involv ement of stakeholders, and (v) relevance and usability of study results for the operationalization of the ecosystem service concept in practice. Results show that these facets were not addressed by the majority of case studies including more recent studies. Clusters of ecosystem services studied together were prone to different blind spots. To effectively operationalize the concept of ecosystem services, the blind spots need to be addressed by upcoming studies. A list of critical questions is provided to raise the awareness of the blind spots both for synthesis of existing knowledge and for future research agendas.

In ecology, the true causal structure for a given problem is often not known, and several plausible models and thus model predictions exist. It has been claimed that using weighted averages of these models can reduce prediction error, as well as better reflect model selection uncertainty. These claims, however, are often demonstrated by isolated examples. Analysts must better understand under which conditions model averaging can improve predictions and their uncertainty estimates. Moreover, a large range of different model averaging methods exists, raising the question of how they differ in their behaviour and performance. Here, we review the mathematical foundations of model averaging along with the diversity of approaches available. We explain that the error in model‐averaged predictions depends on each model's predictive bias and variance, as well as the covariance in predictions between models, and uncertainty about model weights. We show that model averaging is particularly useful if the predictive error of contributing model predictions is dominated by variance, and if the covariance between models is low. For noisy data, which predominate in ecology, these conditions will often be met. Many different methods to derive averaging weights exist, from Bayesian over information‐theoretical to cross‐validation optimized and resampling approaches. A general recommendation is difficult, because the performance of methods is often context dependent. Importantly, estimating weights creates some additional uncertainty. As a result, estimated model weights may not always outperform arbitrary fixed weights, such as equal weights for all models. When averaging a set of models with many inadequate models, however, estimating model weights will typically be superior to equal weights. We also investigate the quality of the confidence intervals calculated for model‐averaged predictions, showing that they differ greatly in behaviour and seldom manage to achieve nominal coverage

1.Forest recovery following management interventions is important to maintain ecosystem functioning and the provision of ecosystem services. It remains, however, largely unclear how aboveground biomass (AGB) recovery of species-rich tropical forests is affected by disturbance intensity and post-disturbance (remaining) tree-community attributes, following logging and thinning interventions. 2.We investigated whether annual AGB increment (∆AGB) decreases with management-related disturbance intensity (disturbance hypothesis), and increases with the diversity (niche-complementarity hypothesis) and the community-weighted mean (CWM) of acquisitive traits of dominant species (biomass-ratio hypothesis) in the remaining tree community. 3.We analysed data from a long-term forest-management experiment in the Brazilian Amazon over two recovery periods: post-logging (1983-1989) and post-thinning (1995-2012). We computed the ∆AGB of surviving trees, recruit trees and of the total tree community. Disturbance intensity was quantified as basal area reduction and basal area remaining. Remaining diversity (taxonomic, functional and structural) and CWM of five functional traits linked to biomass productivity (specific leaf area, leaf nitrogen and phosphorous concentration, leaf toughness and wood density) were calculated for the post-intervention inventories. Predictors were related to response variables using multiple linear regressions and structural equation modelling. 4.We found support for the disturbance hypothesis in both recovery periods. AGB increment of survivors and of the total tree community increased with basal area remaining, indicating the importance of remaining growing stock for biomass recovery. Conversely, AGB increment of recruit trees increased with basal area reduction because changes in forest structure increased resource availability for young trees. We did not find consistent support for the niche-complementarity and biomass-ratio hypotheses, possibly because

BackgroundSpecies richness affects processes and functions in many ecosystems. Since management of temperate grasslands is directly affecting species composition and richness, it can indirectly govern how systems respond to fluctuations in environmental conditions. Our aim in this study was to investigate whether species richness in managed grasslands can buffer the effects of drought and warming manipulations and hence increase the resistance to climate change. We established 45 plots in three regions across Germany, each with three different management regimes (pasture, meadow and mown pasture). We manipulated spring warming using open-top chambers and summer drought using rain-out shelters for 4 weeks.ResultsMeasurements of species richness, above- and below-ground biomass and soil carbon and nitrogen concentrations showed significant but inconsistent differences among regions, managements and manipulations. We detected a three-way interaction between species richness, management and region, indicating that our study design was sensitive enough to detect even intricate effects.ConclusionsWe could not detect a pervasive effect of species richness on biomass differences between treatments and controls, indicating that a combination of spring warming and summer drought effects on grassland systems are not consistently moderated by species richness. We attribute this to the relatively high number of species even at low richness levels, which already provides the complementarity required for positive biodiversity–ecosystem functioning relationships. A review of the literature also indicates that climate manipulations largely fail to show richness-buffering, while natural experiments do, suggesting that such manipulations are milder than reality or incur treatment artefacts.

The conservation of amphibians is of increasing relevance due to their ongoing and rapid decline. Alterations of floodplains are a major contributor to these declines in temperate Europe. In this study, we assessed the factors determining the abundance of four anuran species (Rana arvalis, Pelobates fuscus, Bombina bombina and Hyla arborea) in a dynamic floodplain inCentral Europe during two hydrologically markedly differing years. We identified species responses to habitat characteristics using zero-inflated models. Only pond surface area explained the abundances of all species investigated in both years. Pond hydroperiod was the second most informative variable, determining site selection of all species except R. arvalis in both years independent from inter-annual landscape variability. Temporal variability in habitat use was determined by water chemistry and morphology of ponds in the year with frequent floods, whereas aspects of vegetation were more important in the year with the lower water level. Our results underline the importance of accounting for temporal variability of habitat use as facilitated by habitat heterogeneity in conservation planning.

Increased warming in spring and prolonged summer drought may alter soil microbial denitrification. We measured potential denitrification activity and denitrifier marker gene abundances (nirK, nirS, nosZ) in grasslands soils in three geographic regions characterized by site-specific land-use indices (LUI) after warming in spring, at an intermediate sampling and after summer drought. Potential denitrification was significantly increased by warming, but did not persist over the intermediate sampling. At the intermediate sampling, the relevance of grassland land-use intensity was reflected by increased potential N2O production at sites with higher LUI. Abundances of total bacteria did not respond to experimental warming or drought treatments, displaying resilience to minor and short-term effects of climate change. In contrast, nirS- and nirK-type denitrifiers were more influenced by drought in combination with LUI and pH, while the nosZ abundance responded to the summer drought manipulation. Land-use was a strong driver for potential denitrification as grasslands with higher LUI also had greater potentials for N2O emissions. We conclude that both warming and drought affected the denitrifying communities and the potential denitrification in grassland soils. However, these effects are overruled by regional and site-specific differences in soil chemical and physical properties which are also related to grassland land-use intensity.

Plant diversity is globally threatened by anthropogenic land use including management and modification of the natural environment. At regional and local scales, numerous studies world-wide have examined land use and its effects on plant diversity, but evidence for declining species diversity is mixed. This is because, first, land use comes in many variations, hampering comparisons of studies. Second, land use directly affects the environment, but indirect effects extend beyond the boundaries of the land in use. Third, land-use effects greatly depend on the environmental, historical and socio-economic context.To evaluate the generality and variation of studies’ findings about land-use effects, we undertook a quantitative synthesis using meta-analytic techniques.Using 572 effect sizes from 375 studies distributed globally relating to 11 classes of land use, we found that direct and indirect effects of land use on plant diversity (measured as species richness) are variable and can lead to both local decreases and increases. Further, we found evidence (best AIC model) that land-use-specific covariables mostly determine effect-size variation and that in general land-use effects differ between biomes.Synthesis and applications. This extensive synthesis provides the most comprehensive and quantitative overview to date about the effects of the most widespread and relevant land-use options on plant diversity and their covariables. We found important covariables of specific land-use classes but little evidence that land-use effects can be generally explained by their environmental and socio-economic context. We also found a strong regional bias in the number of studies (i.e. more studies from Europe and North America) and highlight the need for an overarching and consistent land-use classification scheme. Thereby, our study provides a new vantage point for future research directions.

&nbsp

The determination of temporal niche dynamics under field conditions is an important component of a species' ecology. Recent developments in niche mapping, and the possibility to account for spatial autocorrelation in species distributions, hold promise for the statistical approach explored here. Using species counts from a landscape-scale benthic monitoring programme in the western Dutch Wadden Sea during 19972005 in combination with sediment characteristics and tidal height as explanatory variables, we statistically derive realised niches for two bivalves, two crustaceans and three polychaetes, encompassing predators, suspension and bottom feeding functional groups. Unsurprisingly, realized niches varied considerably between species. Intraspecific temporal variation was assessed as overlap between the year-specific niche and the overall mean niche, and this analysis revealed considerable variation between years. The main functional groups represented by these species showed idiosyncratic and wide variability through the study period. There were no strong associations between niche characteristics and mean abundance or body size. Our assessment of intraspecific niche variability has ramifications for species distribution models in general and offers advances from previous methods. 1) By assessing species' realized niches in the multivariate environmental space, analyses are independent from the relative availability of particular environments. Predicted realized niches present differences between years, rather than annual differences in environmental conditions. 2) Using spatially explicit models to predict species habitat preferences provide more precise and unbiased estimates of speciesenvironment relationships. 3) Current niche models assume constant niches, whereas we illustrate how much these can vary over only a few generations. This emphasizes the potentially limited scope of global change studies with forecasts based on single-time species distribution snaps

AimThe species–area relationship (SAR) is a prominent concept for predicting species richness and biodiversity loss. A key step in defining SARs is to accurately estimate the slope of the relationship, but researchers typically apply only one global (canonical) slope. We hypothesized that this approach is overly simplistic and investigated how geographically varying determinants of SARs affect species richness estimates of vascular plants at the global scale. LocationGlobal. MethodsWe used global species richness data for vascular plants from 1032 geographical units varying in size and shape. As possible determinants of geographical variation in SARs we chose floristic kingdoms and biomes as biogeographical provinces, and land cover as a surrogate for habitat diversity. Using simultaneous autoregressive models we fitted SARs to each set of determinants, compared their ability to predict the observed data and large-scale species richness patterns, and determined the extent to which varying SARs differed from the global relationship. ResultsIncorporating variation into SARs improved predictions of global species richness patterns. The best model, which accounts for variation due to biomes, explained 46.1% of the species richness variation. Moreover, fitting SARs to biomes produced better results than fitting them to floristic kingdoms, supporting the hypothesis that energy availability complements evolutionary history in generating species richness patterns. Land cover proved to be less important than biomes, explaining only 36.4% of the variation, possibly owing to the high uncertainty in the data set. The incorporation of second-order interactions of area, land cover and biomes did not improve the predictive ability of the models. Main conclusionsOur study contributes to a deeper understanding of SARs and improves the applicability of SARs through regionalization. Future models should explicitly consider geographically varying determinants of SARs in order to improv

Although temporal heterogeneity is a well-accepted driver of biodiversity, effects of interannual variation in land-use intensity (LUI) have not been addressed yet. Additionally, responses to land use can differ greatly among different organisms; therefore, overall effects of land-use on total local biodiversity are hardly known. To test for effects of LUI (quantified as the combined intensity of fertilization, grazing, and mowing) and interannual variation in LUI (SD in LUI across time), we introduce a unique measure of whole-ecosystem biodiversity, multidiversity. This synthesizes individual diversity measures across up to 49 taxonomic groups of plants, animals, fungi, and bacteria from 150 grasslands. Multidiversity declined with increasing LUI among grasslands, particularly for rarer species and aboveground organisms, whereas common species and belowground groups were less sensitive. However, a high level of interannual variation in LUI increased overall multidiversity at low LUI and was even more beneficial for rarer species because it slowed the rate at which the multidiversity of rare species declined with increasing LUI. In more intensively managed grasslands, the diversity of rarer species was, on average, 18% of the maximum diversity across all grasslands when LUI was static over time but increased to 31% of the maximum when LUI changed maximally over time. In addition to decreasing overall LUI, we suggest varying LUI across years as a complementary strategy to promote biodiversity conservation.

Understanding how landscape characteristics affect biodiversity patterns and ecological processes at local and landscape scales is critical for mitigating effects of global environmental change. In this review, we use knowledge gained from human-modified landscapes to suggest eight hypotheses, which we hope will encourage more systematic research on the role of landscape composition and configuration in determining the structure of ecological communities, ecosystem functioning and services. We organize the eight hypotheses under four overarching themes. Section A: ‘landscape moderation of biodiversity patterns’ includes (1) the landscape species pool hypothesis—the size of the landscape-wide species pool moderates local (alpha) biodiversity, and (2) the dominance of beta diversity hypothesis—landscapemoderated dissimilarity of local communities determines landscape-wide biodiversity and overrides negative local effects of habitat fragmentation on biodiversity. Section B: ‘landscape moderation of population dynamics’ includes (3) the cross-habitat spillover hypothesis—landscape-moderated spillover of energy, resources and organisms across habitats, including between managed and natural ecosystems, influences landscape-wide community structure and associated processes and (4) the landscape-moderated concentration and dilution hypothesis—spatial and temporal changes in landscape composition can cause transient concentration or dilution of populationswith functional consequences. Section C: ‘landscape moderation of functional trait selection’ includes (5) the landscape-moderated functional trait selection hypothesis—landscape moderation of species trait selection shapes the functional role and trajectory of community assembly, and (6) the landscape-moderated insurance hypothesis—landscape complexity provides spatial and temporal insurance, i.e. high resilience and stability of ecological processes in changing environments. Section D: ‘landscape constraints on conservati, Understanding how landscape characteristics affect biodiversity patterns and ecological processes at local and landscape scales is critical for mitigating effects of global environmental change. In this review, we use knowledge gained from human-modified landscapes to suggest eight hypotheses, which we hope will encourage more systematic research on the role of landscape composition and configuration in determining the structure of ecological communities, ecosystem functioning and services. We organize the eight hypotheses under four overarching themes. Section A: ‘landscape moderation of biodiversity patterns’ includes (1) the landscape species pool hypothesis—the size of the landscape-wide species pool moderates local (alpha) biodiversity, and (2) the dominance of beta diversity hypothesis—landscapemoderated dissimilarity of local communities determines landscape-wide biodiversity and overrides negative local effects of habitat fragmentation on biodiversity. Section B: ‘landscape moderation of population dynamics’ includes (3) the cross-habitat spillover hypothesis—landscape-moderated spillover of energy, resources and organisms across habitats, including between managed and natural ecosystems, influences landscape-wide community structure and associated processes and (4) the landscape-moderated concentration and dilution hypothesis—spatial and temporal changes in landscape composition can cause transient concentration or dilution of populationswith functional consequences. Section C: ‘landscape moderation of functional trait selection’ includes (5) the landscape-moderated functional trait selection hypothesis—landscape moderation of species trait selection shapes the functional role and trajectory of community assembly, and (6) the landscape-moderated insurance hypothesis—landscape complexity provides spatial and temporal insurance, i.e. high resilience and stability of ecological processes in changing environments. Section D: ‘landscape constraints on conservati

Land use is increasingly recognized as a major driver of biodiversity and ecosystem functioning in many current research projects. In grasslands, land use is often classified by categorical descriptors such as pastures versus meadows or fertilized versus unfertilized sites. However, to account for the quantitative variation of multiple land-use types in heterogeneous landscapes, aquantitative, continuous index of land-use intensity (LUI) is desirable. Here we define such a compound, additive LUI index for managed grasslands including meadows and pastures. The LUI index summarizes the standardized intensity of three components of land use, namely fertilization, mowing, and livestock grazing at each site. We examined the performance of the LUI index to predict selected response variables on up to 150 grassland sites in the Biodiversity Exploratories in three regions in Germany (Alb, Hainich, Schorfheide). We tested the average Ellenberg nitrogen indicator values of the plant community, nitrogen and phosphorus concentration in the aboveground plant biomass, plant-available phosphorus concentration in the top soil, and soil C/N ratio, and the first principle component of these five response variables. The LUI index significantly predicted the principal component of all five response variables, as well as some of the individual responses. Moreover, vascular plant diversity decreased significantly with LUI in two regions (Alb and Hainich). Inter-annual changes in management practice were pronounced from 2006 to 2008, particularly due to variation in grazingintensity. This rendered the selection of the appropriate reference year(s) an important decision for analyses of land-use effects, whereas details in the standardization of the index were of minor importance. We also tested several alternative calculations of a LUI index, but all are strongly linearly correlated to the proposed index. The proposed LUI index reduces the complexity of agricultural practices to a single d