Development
Here, we describe the development of the two maps presented in BeeMapper: a land cover map and a predicted wild bee abundance map.
Land cover map
The land cover map is based on the 2004 Maine Landcover Dataset, which is freely available from the Maine Office of GIS (Landcover – MELCD 2004). This dataset has 5m spatial resolution, which captures landscape variation that is important to wild bees. We classified the original map into 8 land cover classes that each provide a unique set of resources for wild bees. We further modified this map to include roads and railroads, enhance wetland diversity, and provide the most extensive wild blueberry coverage. Additional data sources used for the land cover map are:
- Roads: MEDOTPUBRDS (from MEGIS)
- Railroads: RAILROUTESYS (from MEGIS)
- Wetlands: National Wetlands Inventory
- Additional wild blueberry fields: USDA Croplands Dataset
Predicted wild bee abundance map
The abundance map is output from the InVEST Crop Pollination Model (Lonsdorf et al. 2009) applied to Maine’s wild blueberry production landscape. The InVEST Crop Pollination Model requires three components to predict bee abundance across a landscape:
- A land cover map. The land cover map presented in BeeMapper and described above is the same land cover map we used to predict wild bee abundance with the InVEST Crop Pollination Model. We resampled this map to 10m spatial resolution for computational efficiency, with no measurable loss in model performance.
- Habitat suitability values for wild bees within each land cover type within the land cover map. These values were provided by expert opinion. We asked 12 experts to rank eight land cover types for their ability to provide:
- Ground nesting resources (open soil, gravel)
- Cavity nesting resources (rotting logs, dead shrubs with hollow branches)
- Forage resources (pollen and nectar from blooming flowers) in the early, middle, and late summer (3 separate scores). Assessing forage resources throughout the growing season accounts for seasonal variability in the abundance and diversity of blooming flowers and leads to more accurate model predictions.
Expert opinion-derived habitat suitability parameters are listed in Table 1. Each land cover type is ranked on a scale of 0 to 1, with 1 being most suitable habitat for wild bees.
Table 1. Expert-opinion derived parameters for the InVEST Crop Pollination Model applied to Maine wild blueberry
| Land cover type | Ground nesting | Cavity nesting | Late spring forage | Early summer forage | Late summer forage |
|---|---|---|---|---|---|
| Agriculture/grassland | 0.7 | 0.2 | 0.9 | 0.7 | 0.9 |
| Blueberries | 1.0 | 0.4 | 0.4 | 1.0 | 0.5 |
| Coniferous forest | 0.5 | 0.6 | 0.1 | 0.1 | 0.1 |
| Deciduous/mixed forest | 0.6 | 0.9 | 0.7 | 0.5 | 0.4 |
| Deciduous/mixed forest edge | 0.9 | 1.0 | 0.9 | 0.9 | 1.0 |
| Emergent wetland | 0.2 | 0.4 | 0.7 | 0.6 | 0.6 |
| Urban/developed | 0.9 | 0.6 | 1.0 | 0.9 | 1.0 |
| Wetlands/water | 0.1 | 0.1 | 1.3 | 0.2 | .5 |
- Bee life history traits for wild bees associated with the focal crop. We collected information for nesting preference, foraging distance, and flight season of 14 wild bee species found in blooming wild blueberry fields (Table 2). Due to model constraints, we were unable to incorporate bumblebees (Bombus species), an important blueberry pollinator, into the model.
Table 2. Bee life history traits for 14 common wild bee species in Maine wild blueberry
| Species | Family | Nest substrate | Typical foraging distance(yd) | Flight season |
|---|---|---|---|---|
| Andrena carlini | Andrenidae | ground | 653 | Mar-Aug |
| Andrena carolina | Andrenidae | ground | 270 | Apr-Jul |
| Andrena vicina | Andrenidae | ground | 622 | Mar-Aug |
| Augochlorella aurata | Halictidae | ground | 65 | Apr-Oct |
| Colletes inaequalis | Colletidae | ground | 1193 | Mar-Sept |
| Halictus ligatus | Halictidae | ground | 161 | Mar-Nov |
| Lasioglossum acuminatum | Halictidae | ground | 203 | Apr-Oct |
| Lasioglossum cressonii | Halictidae | ground | 68 | Mar-Oct |
| Lasioglossum heterognathum | Halictidae | ground | 17 | Apr-Sept |
| Lasioglossum leucocomum | Halictidae | ground | 34 | Mar-Oct |
| Lasioglossum pectorale | Halictidae | ground | 88 | Mar-Nov |
| Lasioglossum versatum | Halictidae | ground | 86 | Mar-Oct |
| Osmia atriventris | Megachilidae | cavity | 203 | Apr-Jul |
| Osmia inspergens | Megachilidae | cavity | 541 | May-June |
The InVEST Crop Pollination Model predicts bee abundance at each point across the input land cover map for each bee species using bee foraging distance and flight season in tandem with the resources offered by each land cover type throughout the growing season. These maps are then summed to create the model output, an overall predicted wild bee abundance map. Each point in this map is given an index value between 0 and 1, with 1 representing the highest wild bee abundance. We classified this index into five abundance classes for clearer interpretation.
Full details of the development of the land cover map and application of the InVEST Crop Pollination Model to the Downeast wild blueberry growing region can be found in Groff et al. (2016) and Chapin (2014).