UC Berkeley

High-latitude ecosystems play a key role in the global carbon cycle. There are large stocks of carbon stored in permafrost soils, protected by freezing temperatures and saturated soil conditions. However, the arctic is the fastest warming region on the planet, and as permafrost thaws much of the carbon will become vulnerable to microbial decomposition. At the same time, warming air temperatures, changing precipitation patterns, increasing nutrient availability, and rising CO2 concentrations will lead to dramatic changes in vegetation dynamics and ecosystem carbon uptake. Because of strong feedbacks and system complexity, current estimates of the impact of climate change on regional net carbon balance are highly uncertain. Indeed, high-latitude carbon cycling was identified in the IPCC AR6 report as one of the largest sources of uncertainty in the global carbon cycle.

This thesis explores large- and fine-scale controls on high-latitude carbon cycling and examines uncertainties associated with regional estimates of carbon fluxes. This work relies heavily on ecosys, a process-rich mechanistic ecosystem model that has been extensively tested at high-latitudes. After a brief introduction, background on ecosystem ecology and the structure and process representation of ecosys is given. Then, an analysis on the large-scale controls on high-latitude carbon cycling, and how those controls will change with climate change, is presented through the lens of seasonality using ecosys simulations of Alaskan ecosystems. Next, key drivers of fine-scale variability of permafrost distribution, vegetation dynamics, and carbon cycling in a discontinuous permafrost watershed are identified using a sensitivity analysis of ecosys. Finally, machine learning models trained on ecosys outputs are shown to inaccurately predict both current and future high-latitude carbon balances.