Nicholson_etal_2024_LandscapesOfRisk

These data include observed pesticide residues in pollens gathered by honey bees (Apis mellifera), bumble bees (Bombus terrestris), and mason bees (Osmia bicornis) across 41 sites in southern Sweden from 2019 through 2021 (site network centroid: 55°43'09.2"N, 13°47'12.2"E). From these observed compound concentrations we calculated toxicity-weighted exposure (TWE) to estimate the direct pesticide risk to bees (Knapp et al. 2023), where the TWE for each compound (TWEi) is the ratio of its exposure value (ci) and its respective acute toxicity endpoint (LD50i). We summed TWEis to calculate risk (see Materials and Methods).We analyzed the landscape surrounding our sites at three spatial scales according to the average foraging range for our three genera (Osmia: 500; Bombus: 1,500; and Apis: 2,000 m). Landscape metrics were: proportion agricultural land, the proportion of pesticide-treated area, calculated landscape crop diversity, the interspersion and juxtaposition index, edge density, mean patch area (see Materials and Methods).Using a national data set of pesticide use and agricultural parcel data we calculated landscape-level metrics of pesticide use: total landscape pesticide load, total landscape toxic load (see Materials and Methods).We also used the spatialized pesticide use data as input a previously developed and validated mechanistic model (Lonsdorf et al., 2024) to predict bees' pesticide exposure and subsequent risk (i.e., summed TWEis, as above).These data were used to compare the ability of the three classes of landscape-scale variables to predict this observed pesticide risk: 1) landscape composition and configuration metrics, 2) landscape load based on national pesticide use data, and 3) predictions from the bee pesticide exposure model. For the second and third class we distinguish between potential and realized load and risk (see Materials and Methods).The associated paper has been been published as an article in Journal of Applied Ecology.The columns included in this data are:SiteID - Field Identifier for the sites used in this analysis.BeeSpecies - the bee species from which pollen samples were collected. BB: Bombus terrestris; HB: Apis melifera; SB: Osmia bicornis.Year - When the study took place.Month - Month number of when pollen samples were collected.ObservedPesticideRisk - pesticide risk (summed toxicity weighted concentrations) observed in pollen samples.TotalLandscapeLoad_Realized - The amount of different compounds applied in the landscape within the foraging range of the bee species from which the pollen sample was collected. This variable is calculated for all compounds that were detected in the sample (i.e., realized load). A log transformed and z-standardized version of this variable is also included (this is what was used in statistical models).TotalLandscapeLoad_Potential - The amount of different compounds applied in the landscape within the foraging range of the bee species from which the pollen sample was collected. This variable is calculated for all compounds that were screened for in the sample (i.e., potential load). A log transformed and z-standardized version of this variable is also included (this is what was used in statistical models).TotalLandscapeToxicLoad_Realized - The amount of different compounds applied in the landscape weighting by their toxicity within the foraging range of the bee species from which the pollen sample was collected. This variable is calculated for all compounds that were detected in the sample (i.e., realized load). A log transformed and z-standardized version of this variable is also included (this is what was used in statistical models).TotalLandscapeToxicLoad_Potential - The amount of different compounds applied in the landscape weighting by their toxicity within the foraging range of the bee species from which the pollen sample was collected. This variable is calculated for all compounds that were screened for in the sample (i.e., potential load). A log transformed and z-standardized version of this variable is also included (this is what was used in statistical models).TotalLandscapeRisk_Realized - The level of different compounds that bees are predicted to encounter based on a spatially explicit model of landscape exposure (SEMLE) for bees (Lonsdorf et al. 2024). These exposure predictions are weighted by their toxicity and summed (as above). This variable is calculated for all compounds that were detected in the sample (i.e., realized load). A log transformed and z-standardized version of this variable is also included (this is what was used in statistical models).TotalLandscapeRisk_Potential - The level of different compounds that bees are predicted to encounter based on a spatially explicit model of landscape exposure (SEMLE) for bees (Lonsdorf et al. 2024). These exposure predictions are weighted by their toxicity and summed (as above). This variable is calculated for all compounds that were screened for in the sample (i.e., realized load). A log transformed and z-standardized version of this variable is also included (this is what was used in statistical models).propAg - The proportion of agricultural land within the foraging range of the bee species from which the pollen sample was collected. A z-standardized version of this variable is also included (this is what was used in statistical models).propTreatedAg - The proportion of pesticide-treated agricultural land within the foraging range of the bee species from which the pollen sample was collected. A z-standardized version of this variable is also included (this is what was used in statistical models).CropDiversity - Shannon index of crop types (richness) and their coverage (evenness) within the foraging range of the bee species from which the pollen sample was collected. A z-standardized version of this variable is also included (this is what was used in statistical models).iji - interspersion and juxtaposition index (McGarigal & Marks, 1995). A z-standardized version of this variable is also included (this is what was used in statistical models).References:Knapp, J. L., Nicholson, C. C., Jonsson, O., de Miranda, J. R., & Rundlöf, M. (2023). Ecological traits interact with landscape context to determine bees’ pesticide risk. Nature Ecology & Evolution, 7(4), 547-556.Lonsdorf, E. V., Rundlöf, M., Nicholson, C. C., & Williams, N. M. (2024). A spatially explicit model of landscape pesticide exposure to bees: Development, exploration, and evaluation. Science of the Total Environment, 908, 168146.McGarigal, K., & Marks, B. J. (1995). Spatial pattern analysis program for quantifying landscape structure. Gen. Tech. Rep. PNW-GTR-351. US Department of Agriculture, Forest Service, Pacific Northwest Research Station, 1-122.Nicholson, C. C., Lonsdorf, E. V., Andersson, G. K. S., Knapp, J., Svensson, G. P., Gönczi, M., Jonsson, O., de Miranda, J. R., Williams, N. M., & Rundlöf, M. (2024). Landscapes of risk: a comparative analysis of landscape metrics for the ecotoxicological assessment of pesticide risk to bees. Journal of Applied Ecology. doi: TBD