How can we incorporate species interactions into SDM-based climate change modelling?

Existing approaches for assessing the impacts of climate change on plant–pollinator systems generally fall into two categories. The first relies on species distribution models (SDMs) to generate habitat suitability maps for individual species, which are subsequently projected under future climate scenarios. While useful, this approach evaluates plants and pollinators independently and does not explicitly account for their ecological interactions. The second approach focuses on constructing plant–pollinator interaction networks and simulating the consequences of species loss, either randomly or through the targeted removal of generalist or specialist species, to evaluate network vulnerability. However, this method lacks ecological realism because it does not incorporate actual changes in species distributions driven by climate change. To bridge these approaches, we present an integrative framework that combines species distribution modeling with network analysis. Specifically, we first generated continuous climate suitability maps for plants and pollinators under current and future (2070) climate conditions using SDMs. We then developed a Python-based tool to identify potential spatial co-occurrence between interacting species by overlaying their continuous suitability surfaces and constructing geographically explicit interaction matrices. Applied to a Chilean dataset comprising 187 plant species and 171 pollinator species, our framework indicated that approximately 75% of species are projected to experience reductions in climatically suitable habitat under the SSP585 scenario, with mean declines of about 33% for plants and 25% for pollinators. At the network level, the analysis further suggests that 204 grid cells representing potential interaction networks may completely lose all mutualistic interactions under future climate conditions.