Pollination supply models from local to global scale

GIMÉNEZ-GARCÍA, ÁNGEL; ALLEN-PERKINS, ALFONSO ; BARTOMEUS, IGNASI; BALBI, STEFANO ; KNAPP, JESSICA; HEVIA, VIOLETA; WOODCOCK, BEN A.; SMAGGHE, GUY; MIÑARRO, MARCOS; EERAERTS, MAXIME; COLVILLE, JONATHAN F.; HIPÓLITO, JULIANA; CAVIGLIASSO, PABLO; NATES-PARRA, GUIOMAR; HERRERA, JOSÉ M.; CUSSER, SARAH; SIMMONS, BENNO I.; WOLTERS, VOLKMAR; JHA, SHALENE; FREITAS, BRENO M.; HORGAN, FINBARR G.; GARIBALDI, LUCAS A.

Web Ecology

Copernicus GmbH

Año: 2023

Ecological intensification has been embraced with great interest by the academic sector but is still rarely taken upby farmers because monitoring the state of different ecological functions is not straightforward. Modelling tools can representa more accessible alternative of measuring ecological functions which could help promote their use amongst farmers and otherdecision-makers. In the case of crop pollination, modelling has traditionally followed either a mechanistic or a data-driven5 approach. Mechanistic models simulate the habitat preferences and foraging behaviour of pollinators, while data-driven modelsassociate georeferenced variables with real observations. Here, we test these two approaches to predict pollination supply, andvalidate these predictions using data from a newly released global dataset on pollinator visitation rates to different crops. Weuse one of the most extensively used models for the mechanistic approach, while for the data-driven approach, we select amonga comprehensive set of state-of-the-art machine-learning models. Moreover, we explore a mixed approach, where data-derived210 inputs, rather than expert assessment, inform the mechanistic model. We find that, at a global scale, machine-learning modelswork best, offering a rank-correlation coefficient between predictions and observations of pollinator visitation rates of 0.56. Inturn, the mechanistic model works moderately well at a global scale for wild bees other than bumblebees. Biomes characterisedby temperate or Mediterranean forests show a better agreement between mechanistic model predictions and observations,probably due to more comprehensive ecological knowledge and therefore better parameterization of input variables for these15 biomes. This study highlights the challenges of transferring input variables across multiple biomes, as expected given thedifferent composition of species in different biomes. Our results provide clear guidance on which pollination supply modelsperform best at different spatial scales – the first step toward bridging the stakeholder-academia gap in modelling ecosystemservice delivery under ecological intensification.