RESUMO
Extreme disturbance events, such as wildfire and drought, have large impacts on carbon storage and sequestration of forests and woodlands globally. Here, we present a modelling approach that assesses the relative impact of disturbances on carbon storage and sequestration, and how this will alter under climate change. Our case study is semi-arid Australia where large areas of land are managed to offset over 122 million tonnes of anthropogenic carbon emissions over a 100-year period. These carbon offsets include mature vegetation that has been protected from clearing and regenerating vegetation on degraded agricultural land. We use a Bayesian Network model to combine multiple probabilistic models of the risk posed by fire, drought, grazing and recruitment failure to carbon dynamics. The model is parameterised from a review of relevant literature and additional quantitative analyses presented here. We found that the risk of vegetation becoming a net source of carbon due to a mortality event, or failing to realise maximum sequestration potential, through recruitment failure in regenerating vegetation, was primarily a function of rainfall in this semi-arid environment. However, the relative size of an emissions event varied across vegetation communities depending on plant attributes, specifically resprouting capacity. Modelled climate change effects were variable, depending on the climate change projection used. Under 'best-case' or 'most-likely' climate scenarios for 2050, similar or increased projections of mean annual precipitation, associated with a build-up of fuel, were expected to drive an increase in fire activity (a 40-160% increase), but a decrease in drought (a 20-35% decrease). Under a 'worst-case' climate scenario, fire activity was expected to decline (a 37% decrease), but drought conditions remain similar (a 5% decrease). These projected changes to the frequency of drought and fire increase the risk that vegetation used for carbon offsetting will fail to provide anticipated amounts of carbon abatement over their lifetime.