Ecosystem adaptation is the dominant source of uncertainty in projections of rangeland production responses to climate change

Andrew Felton, Department of Wildland Resources and The Ecology Center, Utah State University, Logan, UT, Robert Shriver, University of Nevada-Reno, Reno, NV and Peter Adler, Department of Wildland Resources and the Ecology Center, Utah State University, Logan, UT

Background/Question/Methods
A major goal of contemporary ecology is providing society with information about the potential impacts of climate change on ecosystem structure, function, and services. Models play a central role in addressing this goal: we use them to translate predicted changes in climate into changes in ecological processes. The speculative nature of this endeavor guarantees large uncertainty. Quantifying uncertainty and determining its sources is therefore essential, but studies rarely consider uncertainty resulting from our limited understanding of relevant ecological processes. We partitioned sources of uncertainty for end-of century projections of forage production across six rangeland ecoregions of the western United States. We leveraged remotely sensed time series of herbaceous primary production and combined them with historical weather data to build ecoregion-specific spatiotemporal statistical models relating forage production to weather. Running future climate projections through this model, we projected late-century changes in mean forage production. Consistent with previous work, we considered a high and low emissions scenario (RCP 4.5 and 8.5) and eleven global circulation models (GCM) to grasp uncertainty related to climate forcing and climate models. Extending previous work, we quantified the uncertainty attributed to the degree in which these ecosystem could adapt to climate change, focusing on the two extremes of possibilities: 1) no ecosystem adaptation by end of century, implemented through a purely time-series approach, and 2) complete ecosystem adaptation, implemented through a space-for-time approach.

Results/Conclusions
End-of-century projections in mean forage production varied spatially both within and across ecoregions but were highly sensitive to the degree of adaptation. Specifically, the assumption of no ecosystem adaptation via the time-series approach was pessimistic; reductions in mean forage production were consistently projected across the western US, and especially in the southwest. However, when we assumed complete adaptation in ecosystem properties via the space-for-time approach, the models projected forage increases in two of the six ecoregions, and much smaller decreases elsewhere. We compared the uncertainty attributed to site, parameter, RCP, GCM, and ecosystem adaptation in each ecoregion and found that ecosystem adaptation was consistently and by far the dominant source of uncertainty in projections of changes in forage production. Our results indicate that the rate at which ‘slow’ ecosystem properties will shift and adapt to contemporary climate change­ poses the largest challenge in projecting resulting changes in ecosystem functions, such as primary production. Meeting this challenge will require fresh and collaborative thinking among paleoecologists, experimental ecologists, modelers, and theorists.