Resiliency and vulnerability of boreal peatlands to wildland fire: identifying patterns through depth of burn, carbon loss, and environmental controls

Abstract

Boreal peatlands are globally significant carbon (C) sinks that have accumulated C over millennia but are increasingly threatened by climate-driven changes in fire regimes and moisture levels. Soil C losses from peat combustion remain uncertain due to challenges in quantifying below-ground combustion, limiting representation in global C budgets and models. This thesis 1) quantifies spatial variability in C loss during boreal peatland fires, and 2) identifies ecological conditions driving variability and peatland resistance or vulnerability to combustion. Research integrates field data, bi-temporal airborne lidar, and statistical modelling from the 2016 Horse River Wildfire in Alberta’s Boreal Plains. A synthesis of existing knowledge produced conceptual models of pre- and post-fire feedbacks impacting peatland combustion and recovery. These frameworks define two conceptual peatlands: resilient – hydrologically connected with low soil bulk density and moderated water tables, resulting in low severity combustion and moisture-retaining negative feedbacks; and vulnerable – fragmented or drained, with high bulk density, fluctuating water tables, and shrub encroachment, resulting in deep burns and drying positive feedbacks. Lidar ground classification accuracy assessments across unburned and burned, regenerating peatlands revealed negligible mean offsets: 0.00 m in burned to 0.01 m in unburned peatlands, with RMSEs of 0.09 m to 0.19 m, respectively. These findings support the utility of lidar for detecting elevation changes from peat combustion. Using these validated data, depth of burn (DOB) was estimated across peatland types and ecotones, averaging 0.08 ± 0.06 m, with deepest combustion in bog ecotones (0.09 ± 0.07 m). Statistical models revealed that top drivers of DOB variability depended on peatland type. In bogs, DOB was associated with topography and morphology, while in fens and swamps, where groundwater connectivity reduces the influence of topography on water tables, differences were associated with vegetation, hydrology, and disturbance. C losses across peatlands estimated using field-based soil C data and lidar-derived DOB showed that soil C losses (-2.11 ± 5.09 kg m-2) were substantially greater than vegetation losses (-0.38 ± 0.32 kg m-2), with bog ecotones identified as hotspots for C combustion loss (-16.5 kg m-2). Comparison with Landsat differenced Normalized Burn Ratio (dNBR) revealed that while vegetation losses related moderately to dNBR, soil losses did not, demonstrating the limitations of optical indices for below-ground combustion. Comparisons with estimates from the Canadian Model for Peatlands highlighted the need to explicitly include ecotones in C models – particularly under a changing climate.