Vegetation state change and consequent carbon dynamics in savanna woodlands of Australia in response to grazing, drought and fire: a scenario approach using 113 years of synthetic annual fire and grassland growth

Abstract

A spatially explicit state and transition model for assessing the interactive effects of grazing, fire and climate on carbon dynamics in Australian savannas is described. The model runs on a yearly time step. It is based on a sequential treatment of events within each year, involving, in order, growth of biomass, consumption of biomass by livestock and burning of the remaining fuel (growth minus consumption). The major drivers are 113 years of annual rainfall data, annual modelled rangeland growth, synthetic fire incidence and timing data, and data describing stocking rates in dry sheep equivalents. Baseline carbon stocks are derived from pre-settlement estimates from the VAST steady-state carbon model. State and transition models for nine vegetation zones define vegetation condition and consequent carbon stock. Change is mediated by key indices of driver variables such as grazing, growth and fire, a series of parameters and default thresholds, and a set of rules. The model is run for three 113-year spin-up cycles to establish a set of initial condition grids that represent a plausible synthetic current state. Sensitivity analyses on selected parameters and index thresholds showed that fire and growth thresholds were most important for woodland zones, and utilisation rates and degradation and recovery periods were most important for grassland zones. The model was used to examine a series of scenarios involving changes to grazing pressure and suppression of fire for different climate sequences. Changes to a few parameters enabled simulation of low and high levels of encroachment of woody weeds, and hence carbon accumulation, in the Astrebla (Mitchell) grasslands; and effective capture of the climate-induced variation in grassland condition, and hence potential for soil carbon loss, within Heteropogon spp.-dominated grasslands in northern Queensland. The scenario results suggest that this simple state and transition model with an annual time step provides a potentially useful and flexible scenario model framework for exploration of vegetation and consequent carbon dynamics of the Australian tropical savanna region.