Increasing parameter accuracy of an agriculturally focused, spatially-explicit bee abundance model

2015 Annual Report for GNE14-076

Project Type: Graduate Student
Funds awarded in 2014: $14,652.00
Projected End Date: 12/31/2016
Grant Recipient: University of Maine
Region: Northeast
State: Maine
Graduate Student:
Faculty Advisor:
Frank A. Drummond
University of Maine, Dept of Biological Sciences
Faculty Advisor:
Dr. Cynthia Loftin
University of Maine

Increasing parameter accuracy of an agriculturally focused, spatially-explicit bee abundance model


Efforts to promote wild bee habitat around lowbush blueberry fields in Maine may be more effective if wild bee abundance in the landscape surrounding these fields could be predicted accurately. The InVEST Crop Pollination Model is a spatially-explicit model that uses land cover data to predict bee abundance in crop and non-crop cover types in agricultural landscapes. Application of this model requires bee habitat suitability scores that are typically informed by expert knowledge.  Uncertainty that results from expert-derived scores potentially affects model prediction accuracy.  My research uses field surveys to quantify wild bee communities in non-crop land cover types in Maine’s lowbush blueberry growing regions.  I hypothesize that field data-derived parameters will improve accuracy of model-predicted bee abundance compared to accuracy of expert opinion-derived parameters. This research will be the first comparison of expert opinion and field-collected bee community data used to parameterize the InVEST Crop Pollination model. The information gained from this research will also be applied in a tool developed for lowbush blueberry growers to assess wild bee abundance in the landscapes surrounding their crop fields.

Objectives/Performance Targets

Objective 1: Quantify bee community richness and abundance in non-blueberry cover types throughout Maine’s lowbush blueberry growing region.


We conducted an extensive field survey to assess wild bee communities throughout Maine’s lowbush blueberry growing region in summer 2015. Surveys were conducted in 4 blueberry growing regions of the state; one region had two blocks of survey sites for a total of five survey blocks. Each block contained eight sites, one of each land cover type previously identified as important for wild bees and used in existing runs of the InVEST Crop Pollination model. Sites were selected in the largest possible patch size surrounded by similar landscape context and were at least 1 km apart to survey independent bee communities. Surveys were completed three times: in early, mid, and late summer to align with indices used in InVEST Crop Pollination modeling.  Each survey consisted of a 24 hour bowl trapping period and 30 minutes of live netting wild bees along a 100 m transect. We also recorded species and patch area for all blooming flowers during each survey period. We assessed nesting suitability of each survey site at the beginning of the study period, for which we ranked presence of open soil and woody debris that serve as homes for wild bees.


We collected 1,723 bee specimens across 40 survey sites, all of which were identified to the most specific level possible. All identifications made to the genus level were verified to species by a bee taxonomy expert. Statistical analyses of these data will begin in early 2016.


Objective 2: Compare predicted bee abundance from a spatially-explicit bee abundance model informed by expert opinion to abundance informed by bee community field data.


We applied the InVEST Crop Pollination model to additional areas of the Maine lowbush blueberry growing region to establish a region-wide basis of model output comparison. The comparison of expert opinion-informed parameters to field data-informed parameters will begin in early 2016 pending statistical analyses of the extensive data set collected for Objective 1.


Objective 3: Develop a pollination tool for lowbush blueberry growers that presents information on native bee communities in and around crop fields, including an estimate of pollination service on their crop fields


We developed a web-based tool called BeeMapper using current output of the InVEST model based solely on expert opinion. I conducted six testing sessions of the initial BeeMapper prototype with individual growers or small groups of growers in late winter 2015. Growers selected for testing use a diverse range of crop management techniques and manage a wide range of crop fields.


We used feedback from these testing sessions to revise BeeMapper, which we then presented to growers in fall 2015. We continue to develop BeeMapper; we are writing tool documentation and building a web site to host the tool that will be tested by growers in winter 2016.  We will publish the BeeMapper User’s Guide and the methodology used to create the tool in a Technical Bulletin through the Maine Agricultural and Forest Experiment Station. Launch of BeeMapper for public access is dependent on obtaining the most accurate InVEST Crop Pollination Model output as described in Objectives 1 and 2 and is expected to be completed by December 2016.


We have collected a robust data set on wild bees in non-crop land cover types in Maine that we will use to increase the accuracy of parameters used in the InVEST Crop Pollination Model. We have developed a grower-focused web tool displaying wild bee abundance across the landscape that was well-received by growers and is generating interest from other stakeholders in Maine, particularly those involved in beekeeping and pollinator conservation.

Impacts and Contributions/Outcomes

The grower testing sessions for BeeMapper demonstrated that lowbush blueberry growers are interested in promoting wild bee populations around their fields for crop pollination and conservation purposes. We have established a relationship with growers around BeeMapper through a transparent participatory development process, which encourages positive reception of the tool throughout the grower community. Grower feedback led to drastic improvements to the initial prototype of BeeMapper and ultimately made a more grower-friendly tool.  BeeMapper will close an information gap regarding wild bees across Maine’s lowbush blueberry landscape and help growers pursue pollinator conservation practices. Additionally, field-collected data will more accurately inform the predictive accuracy of the InVEST Crop Pollination Model applied to the lowbush blueberry landscape in Maine.


We gave multiple presentations of work for this project in 2015 in both academic and non-academic settings. Collaborating with stakeholders to disseminate scientific output is an exciting process; demonstrating our experiences to other academics has been fruitful and connecting with the public has motivated us to pursue other arenas of outreach using our data. The connection between land cover type and bee abundance is important for managing wild bees around crop fields, but it is also relevant for determining sources of honey bee forage for beekeepers. We look forward to continuing our collaborations with stakeholders and completing our analyses to more accurately quantify this connection and encourage pollinator conservation in agricultural settings.


Academic Presentations:


Du Clos, B., Hanes, S.P., Groff, S.C., Loftin, C.S., and F.A. Drummond. BeeMapper: a tool for grower assessment of wild bee abundance. Wild Blueberry Research and Extension Workers 2015 Annual Conference. 10/2015. Bar Harbor, ME.


Du Clos, B., Hanes, S.P., Groff, S.C., Loftin, C.S., and F.A. Drummond. A tool for grower assessment of wild bee abundance in the Maine wild blueberry landscape. Student Conference on Conservation Science-New York, 10/2015. New York, NY. Poster.


Du Clos, B., Hanes, S.P., Groff, S.C., Loftin, C.S., and F.A. Drummond. Development of a Pollinator Habitat Assessment Tool in Maine’s Wild Blueberry Landscape. Northeast Natural History Conference, 4/2015. Springfield, MA.


Hanes, S.P., and B. Du Clos. How Maine Blueberry Growers Use Wild Bees: Learning from a Sustainability Transition in Progress. Senator George J. Mitchell Center for Sustainability Solutions Seminar, 3/2015. Orono, ME. Invited speakers.


Du Clos, B., Hanes, S.P., Loftin, C.S., and F.A. Drummond. A web-based tool for grower assessment of native bee abundance in the wild blueberry production landscape. Maine State Wild Blueberry Growers Meeting, 3/2015. Bangor, ME. Poster.


Du Clos, B., Hanes, S.P., Loftin, C.S., and F.A. Drummond. A web-based tool for grower assessment of native bee abundance in the wild blueberry production landscape. USGS Maine Cooperative Fish and Wildlife Research Unit Annual Coordinating Committee Meeting, 3/2015. Orono, ME.


Du Clos, B., Hanes, S.P., Loftin, C.S., and F.A. Drummond. A web-based tool for grower assessment of native bee abundance in the wild blueberry production landscape. Maine Sustainability and Water Conference, 3/2015. Augusta, ME. Poster.


Public Outreach Presentations:


Honey bees and the landscape: Choosing an optimal apiary site. Maine State Beekeepers Association, 11/2015. Hampden, ME.


The BeeMapper web tool: assessing bee habitat around wild blueberry fields. University of Maine Cooperative Extension Wild Blueberry Field Day, 7/2015. Jonesboro, ME


Mapping bees in Maine’s landscape. Androscoggin Beekeepers Club, 4/2015. Auburn, ME.


Dr. Cynthia Loftin
Unit Leader and Associate Professor of Wildlife Ecology, USGS Maine Cooperative Fish and Wildlife Research Unit
5755 Nutting Hall, Rm 210
Orono, ME 04469
Office Phone: 2075812843
Dr. Frank Drummond
Professor of Insect Ecology
5751 Murray Hall
Orono, ME 04669
Office Phone: 2075812989