The science behind Beescape tools
If you have questions about how to use Beescape tools, please watch our video tutorials or browse our FAQs.
If you would like additional information about tool parameters for research purposes, please check our articles or contact us.
Jump to topic
The Beescape Map Tool enables users to predict the quality of pollinator habitat at US locations. Specifically, this mapping tool estimates spring, summer and fall floral resources in addition to nesting habitat and insecticide load. Sites with high scores for floral and nesting habitat and low scores for insecticide load are predicted to be higher quality habitat for pollinators.
This tool combines remote-sensing data with other data sources to generate habitat predictions.
Cropland Data Layer
The tool uses the USDA-NASS Cropland Data Layer (CDL) for its base map of what crops and natural habitats are in your landscape. These data are created through a combination of survey work of crop fields by the USDA and satellite collection of land cover data for natural areas. This is the best available national map of US cropland, but it has some inaccuracies at the local level. It provides data at a resolution of 30 meters (about 1/5th of an acre), so it misses detail for land cover and land use at finer scales. Also, the use of satellite information to classify land cover is not perfect so sometimes will misclassify land cover. Finally, it takes about a year to classify the land cover from the satellite and provide the data to the public, so that the base map typically represents the previous year’s data. We use the resulting CDL layer as an input into predictive models that translate the land cover into bee forage resources, nesting habitat, and insecticide toxic load.
To see what crops are on the map for your selected area (go to the beescape app), toggle the Crop Layer button on and off at the lower left of the page. You can change the opacity of the Crop Layer with the sliding bar underneath the button.
Forage and Nesting Scores
Bees require two basic types of resources to persist on a landscape: nesting substrates (for wild bees) and floral resources (for honey bees and wild bees) . Bees move between nesting habitats and foraging habitats across seasons and their foraging distances, in combination with arrangement of different habitats, affect their individual fitness, hive health and population persistence. The model we use predicts nesting habitat and the quality of forage in spring, summer and fall and does so for each in two steps.
First, it translates a map of land cover into a floral quality index (0-100) for each floral season (spring, summer, fall) that represents the density and supply of floral resources provided each land cover type at each location on the map. Data used to parameterize the model come from a past study that estimated wild bee habitat quality across the continental United States. In this study, fourteen wild bee experts provided estimates of forage and wild bee nesting quality provided by each land cover type, and we use the average of their estimates to parameterize the translation of land cover into nesting and floral quality values. The model used in this paper has been empirically tested globally and shown to predict wild bee abundance.
Second, the model represents foraging behavior of bees to summarize the quality of a landscape around a selected site (which represents a colony or nest site). The model uses the information from the first step to determine the average floral and nesting quality around the site using a distance decay function. This means that locations closer to a nest site are more likely to be visited than those farther away. Higher scores indicate higher quality forage or nesting sites within a bee’s foraging range (3 to 5km) and are predictors of greater relative bee abundance for wild bees or likely higher quality hives.
Here is an example of how the model translates satellite information into floral resource quality. The satellite imagery shows the western edge of State College, PA and the rural landscape just outside the city. This imagery is then translated into distinct land cover types within the Crop Data Layer. We then translate each land cover into relative floral quality value for spring, summer and fall.
Insecticide Load Scores
The insecticide load score reflects the expected ‘toxic load’ of insecticides applied surrounding a given location (representing a colony or nest site). These scores are generated in a multi-step process. First, we use data on insecticide use from the U.S. Geological Survey and data on crop acreage from the U.S. Department of Agriculture to estimate the average per-hectare use of > 100 insecticide active ingredients on each type of cropland for each state. Insecticide use is then translated into honey bee lethal doses and summed across insecticides to generate a single value expressed in billion lethal doses applied per hectare, which we scaled (x100) to be similar in value to the forage and nesting scores. We use these data in combination with land cover data to generate a map of predicted insecticide toxic load. Similar to the method for forage quality, we then assume that locations closer to a hive or nest site are more likely to be visited than those farther away, to generate a value at each site that represents the insecticide toxic load in the entire foraging range (3 to 5km). Higher scores are predicted to negatively influence bee and/or hive health.
There are several limitations and sources of uncertainty in the insecticide load score. Importantly, the score is based on all insecticides applied in a landscape, but does not account for the reality that bees will encounter only a small proportion of the total insecticides applied. Thus, these scores should provide information on the relative amount of insecticide toxicity in the landscape. Moreover, the score reflects agricultural insecticide use, and so excludes other kinds of insecticide application (e.g. homeowner use, mosquito spraying). The score also scales insecticide use by short-term toxicity to adult bees, and assumes that insecticides have additive effects. Sublethal effects, effects on developing bees, and synergistic effects therefore may not be fully captured by this score. Finally, patterns of insecticide use are predicted based on state averages from surveys conducted in recent years and so do not reflect local variability in farmer decision-making.
Landscape quality ranking
We ranked landscape scores as low, medium, or high based on the distribution of scores across all the states. Landscape scores are considered ‘low’ if they are less than or equal to the 25th percentile, ‘medium’ if they are between the 25th and 75th percentile, and ‘high’ if they are equal to or greater than the 75th percentile. The percentile thresholds were calculated from the landscape scores in each state (PA, IN, IL, and WV) and then averaged. We included West Virginia to have equal representation of Mid-Atlantic (PA and WV) and Midwest (IL and IN) states.
Temperature fluctuations and precipitation affect colony survival. Summer growing conditions are important predictors of winter colony survival because they strongly impact the availability of floral resource across seasons and, therefore, overall colony health at the onset of winter. If a growing season was too warm (or warm for too long) or too cool (or too cool for too long) then colony survival is predicted to be lower.
Beekeepers across the state of Pennsylvania contributed apiary survival data from thousands of managed colonies. The Beescape Development team combined this information with meteorological data to develop this prediction tool for assessing relative PA colony survival rates for a given year (Calovi et al, 2021). Survival rates are presented as a function of Growing Degree Days (GDD; a metric of temperature across seasons) for the current and past several years. If temperatures are ideal, a given year's trend line will overlap the grey GDD range. If temperatures are too high or low, the trend line will rise into red or fall into blue GDD territory, respectively, predicting increased colony stress and lower survival.
Although the BeeWinterWise Colony Survival Tool is currently limited to the state of Pennsylvania, we hope to add additional states as new data become available. This is made possible by beekeepers like you from across the nation. To contribute colony data to this project, create a Beescape account today!