Ecological forecasts for marine resource management during climate extremes

Climate change is significantly altering marine ecosystems, creating uncertainty for both ecological and human communities. Proactive management strategies require forward-looking information on ecosystem conditions. While sub-seasonal to seasonal forecasting of atmospheric and ocean physics and ecology has advanced significantly, the uptake of these forecasts by end-users remains limited due to factors such as technical debt, uncertainty communication, and forecast compatibility with end-user needs. A common barrier is the assumption that high-resolution regional forecasts are necessary, while global forecasts offer considerable utility and availability. This research addresses this gap by exploring the application of readily available global forecasts in marine ecological forecasting.

The authors reviewed existing literature on climate change impacts on marine ecosystems, the advancements in ocean forecasting, and the challenges in translating forecast products into effective management tools. They highlighted the importance of seasonal to decadal forecasting for marine resource management and the need for flexible management frameworks that can adapt to climate-driven change. The literature review also underscored the limitations of high-resolution regional forecasts in terms of availability and the potential benefits of using global forecasts for regional applications.

The study employed two case studies within the California Current Ecosystem (CCE): 1) the Habitat Compression Index (HCI), which assesses whale entanglement risk by measuring the compression of cool thermal habitat nearshore; and 2) the Temperature Observations to Avoid Loggerheads tool (TOTAL), which guides the timing of drift gillnet fishery closures to minimize loggerhead sea turtle bycatch. Both tools utilize sea surface temperatures (SST). The researchers used both downscaled (<10 km resolution) and global (<100 km resolution) model forecasts with lead times of 0.5–11.5 months. They compared the performance of downscaled versus global forecasts using three skill assessment metrics: correlation coefficient, forecast accuracy, and Symmetric Extreme Dependence Index (SEDI). The NMME provided global SST forecasts, while downscaled forecasts were generated using ROMS, dynamically downscaling the CanCM4 model. The significance of correlation coefficients was assessed using the effective degrees of freedom to account for autocorrelation. Forecast accuracy and SEDI were assessed using bootstrapping.

The study found that global forecasts of SST could be skillfully applied to both management tools, providing accurate predictions up to 11.5 months in advance. For the HCI, forecasts of high compression events (associated with high whale entanglement risk) were skillful from 0.5 to 11.5 months lead time. For the TOTAL tool, closures were skillfully forecast at 6.5 months lead time, with significant skill extending to 11.5 months. Importantly, the global forecast ensemble was generally more skillful than the downscaled ensemble due to a larger number of ensemble members (73 vs. 3). While downscaling improved the skill of individual ensemble members, the larger number of ensemble members in global forecasts better characterized environmental variability. The study also provided a decision-making framework to help practitioners determine whether to invest in regional downscaling for specific applications, considering factors like the required biogeochemical variables, spatiotemporal scale, and accessibility of knowledge and computational resources.

The results demonstrate the value of operational forecast systems in supporting decision-making under uncertain future conditions. The ability to accurately predict ocean conditions using readily available global forecasts, even with relatively coarse resolution, is a significant finding, lowering barriers to implementing ecological forecasting tools globally. The study highlights the potential for proactive management strategies, such as advanced warning of high-risk events, leading to more effective mitigation actions compared to reactive responses. The findings are relevant to various ecosystems and support the development of climate-ready ecosystem-based management approaches.

This research successfully demonstrates the capacity to transition existing marine resource management tools into forecasting systems capable of providing skillful predictions up to 11.5 months in advance. The use of readily available global forecasts offers a cost-effective and logistically feasible approach, particularly beneficial for regions lacking the infrastructure for regional downscaling. Future research should focus on co-developing operational forecast systems with end-users, expanding applications to other ecosystems and species, and further exploring the optimal balance between global and regionally downscaled forecasts.

The study's case studies were limited to the California Current System (CCS), which is characterized by a relatively narrow shelf and exposed to basin-scale dynamics, potentially limiting the generalizability of the findings to other regions. The analysis focused primarily on SST, and incorporating additional biogeochemical fields might enhance forecast skill for certain applications. The relatively small number of ensemble members used for downscaling could also affect the comparison between global and downscaled forecasts.