A New SDM-Based Approach for Assessing Climate Change Effects on Plant-Pollinator Networks

Current methods for studying the effects of climate change on plants and pollinators can be grouped into two main categories. The first category involves using species distribution models (SDMs) to generate habitat suitability maps, followed by applying climate change scenarios to predict the future distribution of plants and pollinators separately. The second category involves constructing interaction matrices between plants and pollinators and then either randomly removing species or selectively removing generalist or specialist species, as a way to estimate how climate change might affect the plant-pollinator network. The primary limitation of the first approach is that it examines plant and pollinator distributions separately, without considering their interactions within the context of a pollination network. The main weakness of the second approach is that it does not accurately predict climate change impacts, as it arbitrarily selects species to remove without knowing which species will truly shift, decline, or increase in distribution due to climate change. Therefore, a new approach is needed to bridge the gap between these two methods while avoiding their specific limitations. In this context, we introduced an innovative approach that first requires the creation of binary climate suitability maps for plants and pollinators, based on SDMs, for both the current and future periods. This step aligns with the first category of methods mentioned earlier. To assess the effects of climate change within a network framework, we consider species co-overlapping in a geographic matrix. For this purpose, we developed a Python program that overlays the binary distribution maps of plants and pollinators, generating interaction matrices. These matrices represent potential plant-pollinator interactions, with a '0' indicating no overlap and a '1' where both species coincide in the same cell. As a result, for each cell within the study area, we can construct interaction matrices for both the present and future periods. This means that for each cell, we can analyze at least two pollination networks based on species co-overlap. By comparing the topology of these matrices over time, we can infer how climate change might affect plant-pollinator interactions at a fine spatial scale. We applied our methodology to Chile as a case study, generating climate suitability maps for 187 plant species and 171 pollinator species, resulting in 2906 pollination networks. We then evaluated how climate change could affect the network topology across Chile on a cell-by-cell basis. Our findings indicated that the primary effect of climate change on pollination networks is likely to manifest more significantly through network extinctions, rather than major changes in network topology.