When you imagine a bear in the wild, what do you see? What is the bear’s habitat? Perhaps you imagined a lush meadow in the mountains, a salmon-filled coastal stream, or, an old growth forest with space between the trees for a large mammal to navigate.
Today, wild bears inhabit landscapes dominated by human influence. Under every bear’s footprint is a growing human footprint, whether forest disturbance, climate change, nitrogen deposition, C02 fertilization, or land-use conversion. Removed from the vast majority of their historical range, space for bears continues to shrink, pushing remaining populations toward the edge. Despite a growing understanding of global change, the critical question remains: how can we know when and where our activities are likely to cross a threshold, triggering a state change? As scientists, we are left to determine where these ecological thresholds exist, interpreted in conservation law as fixed in time and space. Yet, it is clear that both ecosystems and their thresholds are dynamic.
Empirical evidence suggests that the combined effects of unprecedented human activity are producing directional change in many forests globally. Not only does humanity require a safe operating space in a warmer and more crowded world, so does all life on earth. Science requires better tools for representing ecosystem dynamics at scales useful to management. In doing so, we may foster a safe operating space for biodiversity by informing a new era in conservation law. These tools may be based on engineering tolerances that incorporate spatiotemporal variability.
While modern ecology has focused on restoring pre-modern landscapes, we are left with a post-modern reality: there is no going back. Our greatest hope lies ahead. We are changing the earth so profoundly that the recent past is no longer a reliable indicator of the future. How then can we forecast forest ecosystems? While supercomputers enable mechanistic modeling, their computation and parameterization costs remain prohibitive. Meanwhile, hybrid models exhibit equilibrium between model specificity and generalizability.
Dynamic forest landscape models – a class of hybrid model – integrate multiple ecosystem process models in a computationally efficient manner, reducing dimensionality by focusing on the most salient dynamics. Using these models, we can simulate forest landscapes on common desktop computers, or even mobile devices. Through the interaction of individual forest stands as cells in a cellular automaton, dynamic forest landscape models can reproduce self-organization and emergence – hallmarks of complex adaptive systems. Upscaling computationally efficient local dynamics produces non-linear dynamics at the landscape scale.
For my research, I am utilizing the LANDIS-II dynamic forest landscape model fused with the TACA-GEM forest regeneration model and LiDAR remote sensing to forecast grizzly bear habitat in Alberta, Canada. We have shown that historical changes to climate and disturbance regimes are already having a pronounced effect on these forests. We are currently extending our work to inform grizzly bear conservation efforts at the landscape scale. In doing so, we hope to improve the resilience of dynamic grizzly bear habitat in an era of increasing human influence.
Blogpost by Adam Erickson, PhD Candidate, University of British Columbia – adam.erickson(at)ubc.ca
Picture courtesy – Adam Erickson