Variation in peatland wildfire severity – implications for ecosystem carbon dynamics

Abstract

Globally peatlands contain ca. 550 GT of ancient carbon and there is the potential for a positive feedback between peatland degradation and global climate change. Peatlands cover a substantial area of the British Uplands and the effects of wildfire and traditional managed burning on their ecological integrity are issues of growing debate. Land-managers and conservationists continue to argue over the effects of managed and wild fire on peatland carbon dynamics. Clear differences do exist in the severity of moorland fires and wildfires, though often intense, are not necessarily severe. The aim of our project was to document variation in the severity and ecological effects of a spate of peatland wildfires that burnt during the springs of 2011 and 2012. We identified a number of fire locations that included coverage of variation in major north-south and east-west bioclimatic conditions and peatland types. Fire severity was described using a modified form of the semi-quantitative Composite Burn Index. Fuel consumption was assessed using destructive harvesting of burnt- and unburnt fuel loads. Carbon dioxide and methane fluxes were assessed using samples taken from gas flux chambers. Fire severity varied substantially both within and between individual fires. Average severity varied up to two-fold between fires but as much as three-fold within some fires. Total carbon loss varied substantially between and within wildfires. Average fire-level consumption was 0.64± 0.12 kg C m-2 but this estimate should be treated with caution. Consumption was best described by a mixed effects model than included random intercepts for different fires and plots within fires and random slopes for different plots. The evidence for differences in consumption between fires was weak due to the small sampling size and the substantial within plot and fire variation in fuel load. There was a linear relationship between pre-fire fuel surface load and surface fuel consumption but no obvious difference between the prescribed and wild fires in how the proportion of fuel consumed changed with increasing fuel load. Soil methane fluxes were consistently lower on burnt sites whilst carbon dioxide fluxes were generally higher. Day-night temperature fluctuations in burnt plots were frequently more than twice that seen in unburnt plots. Our results demonstrate substantial variation in fire severity both within and between individual wildfire events. Assessment of the effects of wildfires requires intensive sampling efforts and drawing robust conclusions about the implications of wildfire on ecosystem C dynamics is thus fraught with difficulties.