We focused on expansion of pollen, charcoal, speleothem isotope and lake water-balance records of past climate changes. We addressed known limitations in pollen-based quantitative climate reconstructions methods, in particular the problem that regression techniques compress reconstructions towards the middle of the sampled climate range by accounting for sampling frequency and upweighting taxa with narrow climate ranges. Reconstructions of moisture-related variables using modern data cannot account of the physiological impacts of changing atmospheric CO2 on photosynthesis and overestimate dryness when CO2 is low. We developed a physiology-based method to correct this bias. We developed a new, quantitative approach to reconstruct past vegetation that takes account of within-biome compositional variability and can be used to make more realistic assessments of modelled vegetation changes. Sedimentary charcoal records are used to provide a semi-quantitative estimate of fire but are difficult to compare with modelled outputs. We developed a calibration of the charcoal record to produce quantitative estimates of burnt area.
Assessments of vegetation and fire models in both modern and past climate states show that they are only moderately successful in reproducing observations. We have analysed the causes of this and used novel analytical techniques to pinpoint areas for improvement. We developed a model to predict fire size, intensity and burnt area using climate, vegetation, topographic and ignition-related variables, and accounts for the influence of human activities on fire under modern conditions. We have also made progress towards a new vegetation model by using eco-evolutionary optimality theories to develop simple treatments of key processes including photosynthesis, respiration, gross primary production, allocation and plant hydraulics.
A major focus of the project was on evaluating the role of changes in atmospheric CO2 on vegetation, fires and other land surface properties. We have shown that lowering CO2 by ca 100 ppm influences vegetation productivity, causing a substantial reduction in forest cover over much of the world and increasing runoff. This lower CO2 is also the driver for reductions in burnt area at the LGM. It also has a significant impact on fire intensity, although atmospheric drying is also important for this.
The magnitude (and even the sign) of land-surface feedbacks to climate is a major source of uncertainty for climate predictions. Modern observations are too short to provide strong constraints on these feedbacks. We used the palaeorecord to provide estimates of the feedbacks from wildfires and biospheric greenhouse gas emissions. The positive carbon cycle feedback from increased fire is a large contribution to the overall climate-carbon cycle feedback on centennial time scales. Our estimates of greenhouse gas feedbacks show these are not well represented in current models, in particular published high- and low-end values for CO2, are unrealistic.
