Iterative model predictions for wildlife populations impacted by rapid climate change.

To improve understanding and management of the consequences of current rapid environmental change, ecologists advocate using long-term monitoring data series to generate iterative near-term predictions of ecosystem responses. This approach allows scientific evidence to increase rapidly and management strategies to be tailored simultaneously. Iterative near-term forecasting may therefore be particularly useful for adaptive monitoring of ecosystems subjected to rapid climate change. Here, we show how to implement near-term forecasting in the case of a harvested population of rock ptarmigan in high-arctic Svalbard, a region subjected to the largest and most rapid climate change on Earth. We fitted state-space models to ptarmigan counts from point transect distance sampling during 2005-2019 and developed two types of predictions: (1) explanatory predictions to quantify the effect of potential drivers of ptarmigan population dynamics, and (2) anticipatory predictions to assess the ability of candidate models of increasing complexity to forecast next-year population density. Based on the explanatory predictions, we found that a recent increasing trend in the Svalbard rock ptarmigan population can be attributed to major changes in winter climate. Currently, a strong positive effect of increasing average winter temperature on ptarmigan population growth outweighs the negative impacts of other manifestations of climate change such as rain-on-snow events. Moreover, the ptarmigan population may compensate for current harvest levels. Based on the anticipatory predictions, the near-term forecasting ability of the models improved nonlinearly with the length of the time series, but yielded good forecasts even based on a short time series. The inclusion of ecological predictors improved forecasts of sharp changes in next-year population density, demonstrating the value of ecosystem-based monitoring. Overall, our study illustrates the power of integrating near-term forecasting in monitoring systems to aid understanding and management of wildlife populations exposed to rapid climate change. We provide recommendations for how to improve this approach.

End-user involvement to improve predictions and management of populations with complex dynamics and multiple drivers.

Sustainable management of wildlife populations can be aided by building models that both identify current drivers of natural dynamics and provide near-term predictions of future states. We employed a Strategic Foresight Protocol (SFP) involving stakeholders to decide the purpose and structure of a dynamic state-space model for the population dynamics of the Willow Ptarmigan, a popular game species in Norway. Based on local knowledge of stakeholders, it was decided that the model should include food web interactions and climatic drivers to provide explanatory predictions. Modeling confirmed observations from stakeholders that climate change impacts Ptarmigan populations negatively through intensified outbreaks of insect defoliators and later onset of winter. Stakeholders also decided that the model should provide anticipatory predictions. The ability to forecast population density ahead of the harvest season was valued by the stakeholders as it provides the management extra time to consider appropriate harvest regulations and communicate with hunters prior to the hunting season. Overall, exploring potential drivers and predicting short-term future states, facilitate collaborative learning and refined data collection, monitoring designs, and management priorities. Our experience from adapting a SFP to a management target with inherently complex dynamics and drivers of environmental change, is that an open, flexible, and iterative process, rather than a rigid step-wise protocol, facilitates rapid learning, trust, and legitimacy.