Predicting plant attractiveness to pollinators with passive crowdsourcing.

Area of Science:

• Ecology
• Conservation Biology
• Horticulture

Background:

• Global pollinator decline necessitates enhancing resources in managed landscapes.
• Restoration efforts often focus on planting flowering species to support pollinators.
• Identifying optimal plant species for pollinator attraction is challenging and logistically demanding.

Purpose of the Study:

• To test if the abundance of online floral images with identifiable pollinators correlates with plant attractiveness in field trials.
• To evaluate crowdsourced image data as a screening tool for pollinator habitat restoration.
• To identify candidate plant species for supporting wild bee conservation.

Main Methods:

• Utilized Google Image searches to analyze pollinator visitation in photographs for 43 plant species.
• Recorded observations of Apis (honeybees), non-Apis (wild bees), and syrphid flies from initial search results.
• Employed Generalized Linear Models (GLMs) with image observations and bloom period as predictor variables for field-observed abundances.

Main Results:

• Non-Apis bee field observations were positively associated with their presence in Google Image searches (pseudo-R² = 0.668).
• Syrphid fly field observations showed a weak positive association with their frequency in images.
• Apis bee observations were not associated with image data but showed a slight association with bloom period.

Conclusions:

• Passively crowdsourced image data shows potential as a screening tool for identifying plants attractive to wild bees.
• This approach can streamline efforts in pollinator habitat restoration, particularly for wild bee conservation.
• Expanding the understanding of plant-pollinator interactions through such methods can accelerate research for creating pollinator-supportive landscapes.