Examining the response of different wildfire properties to changes in climate and CO2 levels at the Last Glacial Maximum

Climate change and atmospheric CO₂ levels can influence wildfire properties through separate and potentially contrasting impacts on vegetation and climate. One way to examine the sensitivity of global wildfire properties to changes in climate and CO₂ levels is using an out-of-sample experiment, such as the Last Glacial Maximum (LGM; 21 ka BP). Charcoal records show reduced burning at the LGM, when CO₂ levels were ~ 185 ppm and the climate was cooler and drier. In this analysis, we isolated out the potential effects of LGM CO₂ levels and LGM climate on the spatial patterns of global wildfire properties.

Using three statistical models, we conducted simulations of the spatial distribution of global burnt area, fire size and fire intensity under four scenarios: modern climate/modern CO₂ levels, LGM climate/LGM CO₂ levels, modern climate/LGM CO₂ levels and LGM/ modern CO₂ levels. We used outputs from three coupled ocean–atmosphere models representative of the range of simulated LGM climates. The ecophysiological effect of CO₂ levels was explicitly accounted for through vegetation inputs. Gross primary productivity (GPP) and land cover were derived for the LGM and modern climate keeping either CO₂ levels at 395 ppm (modern), or setting them to 185 ppm, using the P Model, a first-principles model of GPP which allows continuous acclimation of photosynthetic parameters to environmental variations, and the BIOME4 equilibrium global vegetation model.

Our results show a reduction in burnt area under LGM CO₂ levels, both with modern and LGM climate inputs. In the case of the warmest of the LGM climate scenarios, this reduction was of the same magnitude as the combined LGM climate/LGM CO₂ levels scenario. However, the driest and coldest LGM climate scenario produced a reduction in burnt area even with modern CO₂ levels, and the largest reduction in burnt area with LGM CO₂.  The reduction was primarily driven by changes in vapour pressure deficit (VPD). Fire size increased under LGM climates, due to changes in wind and VPD. The lower CO₂ values at the LGM had no impact on fire size. Fire intensity increased under LGM climates and LGM CO₂ levels, with both effects of similar amplitude and changes driven primarily by VPD, GPP and diurnal temperature range. 

We compared our outputs with sedimentary charcoal records from the Reading Palaeofire Database (RPD). Overall, the burnt area LGM CO₂ levels/LGM climate scenario showed the greatest agreement, though depending on how cold and dry the LGM climate was, this agreement was either equal to LGM CO₂ levels or LGM climate alone. These results suggest that whilst there was reduced global burning at the LGM, there may have been larger and more intense fires. They also highlight the importance of the ecophysiological effect of CO₂ levels on fuels, a major control of burnt area and fire intensity regardless of climate. They point to the importance of including this effect in process-based fire models, as well as the importance of accurately estimating the amplitude of projected change for different climate variables in order to increase the reliability of future projections.