@article{saffer_jones_horner_laginhas_polo_seliger_sanchez_worm_meentemeyer_2025, title={Quantifying uncertainty in forecasts of when and where invasions happen}, volume={27}, url={https://doi.org/10.1007/s10530-025-03573-w}, DOI={10.1007/s10530-025-03573-w}, abstractNote={Abstract To be useful for real-world applications, the results of ecological forecasts must be accompanied by estimates of uncertainty, i.e., measures of how reliable predictions are likely to be. In spread forecasting for invasive species, pervasive unknowns in invasion systems present many potential sources of uncertainty, from initial conditions to the structure of process-based models. Each relevant source must be integrated to accurately account for uncertainty in predictions that inform decision-making. However, the extent to which uncertainty from multiple sources is represented in spread predictions has not been documented. We conducted a literature review and used the standards outlined by the Ecological Forecasting Initiative to assess and summarize current practices in describing, modeling, propagating (i.e., feeding forward), and partitioning (i.e., quantifying the contribution of each source to overall variance) multiple sources of uncertainty in dynamic, spatially interactive forecasts of invasions. We found that 29% of these predictions report uncertainty, and many discuss sources of uncertainty that are not propagated to predictions and therefore underestimate total uncertainty. Many predictions presented uncertainty using “scenarios”, rather than communicating a full range of plausible values. Further, uncertainty partitioning in invasion forecasts is still very limited, possibly due to computational and methodological challenges associated with partitioning uncertainty for dynamic, geospatial predictions. We highlight challenges central to quantifying uncertainty in invasion systems, including poorly measured initial conditions, information gaps from the transnational nature of emerging invasions, mismatched spatial and temporal scales of drivers and modeled processes, and the computational complexity of geospatial data, and the tradeoffs relevant to invasion modelers seeking to quantify uncertainty in their forecasts.}, number={4}, journal={Biological Invasions}, author={Saffer, Ariel and Jones, Chris and Horner, Eli and Laginhas, Brittany and Polo, John and Seliger, Benjamin and Sanchez, Felipe and Worm, Thom and Meentemeyer, Ross}, year={2025}, month={Apr} }