No abstract

A lack of standard and internationally agreed procedures for higher‐tier risk assessment of plant protection products for bees makes coherent availability of data, their interpretation, and their use for risk assessment challenging. Focus has been given to the development of modeling approaches, which in the future could fill this gap. The BEEHAVE model, and its submodels, is the first model framework attempting to link 2 processes vital for the assessment of bee colonies: the within‐hive dynamics for honey bee colonies and bee foraging in heterogeneous and dynamic landscapes. We use empirical data from a honey bee field study to conduct a model evaluation using the control data set. Simultaneously, we are testing several model setups for the interlinkage between the within‐hive dynamics and the landscape foraging module. Overall, predictions of beehive dynamics fit observations made in the field. This result underpins the European Food Safety Authority's evaluation of the BEEHAVE model that the most important in‐hive dynamics are represented and correctly implemented. We show that starting conditions of a colony drive the simulated colony dynamics almost entirely within the first few weeks, whereas the impact is increasingly substituted by the impact of foraging activity. Common among field studies is that data availability for hive observations and landscape characterizations is focused on the proportionally short exposure phase (i.e., the phase where colony starting conditions drive the colony dynamics) in comparison to the postexposure phase that lasts several months. It is vital to redistribute experimental efforts toward more equal data aquisition throughout the experiment to assess the suitability of using BEEHAVE for the prediction of bee colony overwintering survival. Environ Toxicol Chem 2019;38:2535–2545. © 2019 The Authors. Environmental Toxicology and Chemistry published by Wiley Periodicals, Inc. on behalf of SETAC

No abstract

The quantification of risk (the likelihood and extent of adverse effects) is a prerequisite in regulatory decision making for plant protection products and is the goal of the Xplicit project. In its present development stage, realism is increased in the exposure assessment (EA), first by using real-world data on, e.g., landscape factors affecting exposure, and second, by taking the variability of key factors into account. Spatial and temporal variability is explicitly addressed. Scale dependencies are taken into account, which allows for risk quantification at different scales, for example, at landscape scale, an overall picture of the potential exposure of nontarget organisms can be derived (e.g., for all off-crop habitats in a given landscape); at local scale, exposure might be relevant to assess recovery and recolonization potential; intermediate scales might best refer to population level and hence might be relevant for risk management decisions (e.g., individual off-crop habitats). The Xplicit approach is designed to comply with a central paradigm of probabilistic approaches, namely, that each individual case that is derived from the variability functions employed should represent a potential real-world case. This is mainly achieved by operating in a spatiotemporally explicit fashion. Landscape factors affecting the local exposure of habitats of nontarget species (i.e., receptors) are derived from geodatabases. Variability in time is resolved by operating at discrete time steps, with the probability of events (e.g., application) or conditions (e.g., wind conditions) defined in probability density functions (PDFs). The propagation of variability of parameters into variability of exposure and risk is done using a Monte Carlo approach. Among the outcomes are expectancy values on the realistic worst-case exposure (predicted environmental concentration [PEC]), the probability p that the PEC exceeds the ecologically acceptable concentration (EAC) for a given fraction of habitats, and risk curves. The outcome can be calculated at any ecologically meaningful organization level of receptors. An example application of Xplicit is shown for a hypothetical risk assessment for nontarget arthropods (NTAs), demonstrating how the risk quantification can be improved compared with the standard deterministic approach.

Persistent global urbanization has a direct relationship to measurable artificial light at night (ALAN), and the Defense Meteorological Satellite Program has served an important role in monitoring this relationship over time. Recent studies have observed significant declines in insect abundance and populations, and ALAN has been recognized as a contributing factor. We investigated changes in nightlight intensity at various spatial scales surrounding insect traps located in Orbroicher Bruch Nature Reserve, Germany. Using a time series of global nighttime light imagery (1992–2010), we evaluated pixel-level trends through linear regressions and the Mann–Kendall test. Paired with urban land cover delineation, we compared nightlight trends across rural and urban areas. We utilized high-resolution satellite imagery to identify landscape features potentially related to pixel-level trends within areas containing notable change. Approximately 96% of the pixel-level trends had a positive slope, and 22% of pixels experienced statistically significant increases in nightlight intensity. We observed that 80% of the region experienced nightlight intensity increases >1%, concurrent with the observed decline in insect biomass. While it is unclear if these trends extend to other geographic regions, our results highlight the need for future studies to concurrently investigate long-term trends in ALAN and insect population decline across multiple scales, and consider the spatial and temporal overlaps between these patterns.