Leveraging an open platform for accessible, next generation ecosystem service models

Background/Question/Methods

Designing policies and management to sustain ecosystems requires understanding how human activities and landscape change affect ecosystem services in sometimes locally specific ways, and how these interactions mediate how global phenomenon like climate change and urban expansion. Spatial modelling is a powerful tool that allows us to map these processes across large landscapes but building useful long-term models can be challenging due to several challenges such as (i) existing ecosystem services models are often context-specific and must be entirely rebuilt to for new contexts and (ii) it can be difficult to integrate new data into already existing models. To help resolve these issues we use a model-hosting platform, Artificial Intelligence for Environment and Sustainability (ARIES), which focuses on data and model interoperability. ARIES uses a semantic coding language that allows users to build meta-models that optimally select from general or locally-specific models and datasets to produce landscape-wide outcomes. Developing an extensive database of data and models will allow future large-scale modelling efforts to focus on tailoring these approaches to specific socio-ecological systems.

Results/Conclusions

We developed models to explore three key ecosystem services (carbon storage, pollination, and recreation) across Canada using the ARIES framework. Canada is the 3rd largest country in the world and has dramatic variation in ecological conditions. Relying on the advantages of ARIES, these Canada-wide models use previously developed parts of models tailored to fit the local Canadian ecological context. We first estimate supply and demand for each ecosystem service and identify areas of potential mismatches as a baseline case. We then explore the impacts of 4 scenarios that are derived from two axes of change: a low and high climate change impact (rpc 4.5 and 8.5 scenarios, that specifically affect areas where farmers grow crop within the county) combined with two different urban expansion scenarios (rural growth vs urban expansion). Using these scenarios, we identify changes in supply, demand, and mismatches of each ecosystem service relative to our current (baseline) landscape. Finally, these models can be easily modified to integrate new processes and datasets, making these a powerful tool to explore current and future statuses of ecosystem services.