The application of semantic modelling to map pollination service provisioning at large landscape scales

Abstract

Mapping ecosystem services (ES), including crop pollination by wild insect pollinators, is challenging due to the number of variables involved and the spatial-temporal dimensions of their interactions. To enhance the synergistic relationship between pollination service and crop yield in agricultural landscapes, a better appreciation of the spatial dynamics of pollination service provisioning is needed. Spatially explicit modelling approaches have been used to investigate how different land cover types influence the distribution and abundance of wild bee pollinators in agricultural landscapes. However, an integrated dynamic and spatial modelling framework is needed to address the complexities of pollination supply mapping at the landscape scale. The Artificial Intelligence for Environment and Sustainability (ARIES) framework is a collaborative, spatially explicit and integrated tool for ES assessment. We applied a set of high-resolution process-based pollination models within ARIES to represent landscape capacity to supply pollination by wild bees at the local scale in the Canadian prairies. We also developed a systematic approach to perform a global sensitivity analysis by using a surrogate model (Gaussian Process Regression) and variance-based sensitivity analysis for the selected uncertain key parameters of the model. We modelled pollination dynamics through the mechanistic behavior of native bee guilds, including foraging distance, nesting ability, flight activity, the relative importance of bee guilds, and seasonal variation of floral resources. We focused on three guilds, bumblebees, sweat bees and mining bees, which differed by their nesting habits, floral preferences, and flight distances. We found that over 45% of pollination-dependent croplands in our study area lack wild pollination. The global sensitivity analysis revealed the significance of all key parameters, with seasonal activity across guilds identified as the key driving factors. Our results highlight the significance of the ecological role of wild bees in agricultural landscapes and the sensitivity analysis underscores the importance of temporal dynamics in ecological modeling and pollination.