Passively crowdsourcing images online for measuring broad-scale fly (Diptera) floral interactions and biodiversity

Abstract

Flies (Diptera) represent one of the largest and most important groups of pollinators on the planet; however, little is known about the interactions between flies and flowers compared to well-known pollinators, such as bees. Understanding pollinator assemblages is key to conserving biodiversity and ecosystem services, but monitoring Diptera is time and cost intensive. Using photographs of blooming flowers taken by photographers worldwide and uploaded on internet repositories, we built a dataset of 1,275 images of fly-flower visitations and extracted fly and flower taxonomic information, flower characteristics (shape and color), and fly activity (pollen carrying and foraging). The resulting dataset shows taxonomic and other biases but can still provide an initial overview of factors that affect pollination by Diptera. We identified 22 families of flies, with blow flies (Family Calliphoridae) most represented (29%) and 63 families of flowers, with Asteraceae (42%) and Apiaceae (21%) as the most common. Using logistic regression, we found that the likelihood of flies carrying pollen was determined by the interaction between flower color and shape: pollen-carrying was more likely when elongate cluster flowers were green-yellow. Fly foraging on flowers was determined by flower color: flies were more likely to feed on green-yellow and white flowers. Overall, Syrphidae flies were less likely to forage for nectar than non-Syrphidae, but were more likely to carry pollen. While biases exist in crowdsourced data, we show that data from photographs collected through citizen science offers potentially valuable information for monitoring pollinator-flower interactions and augment our understanding of pollinator ecology in an era of global insect declines.

Keywords

• data mining
• photos
• fly pollinator
• myophily
• species richness