Evaluating long-term effects of prescribed fire regimes on carbon stocks in a temperate eucalypt forest

• Long-term experiment of repeated prescribed fires in mixed-species eucalypt forest.
• Significantly greater total C stocks in control than prescribed fire treatments.
• Fire-associated decreases in C stocks increased with fire frequency.
• Total C stock decreases were greater in autumn than spring fire treatments.
• Fire-related variables were consistently important to explaining C stock decreases.

Prescribed fires are a common management practice in the temperate forests of Australia, but effects on total forest carbon (C) of long-term prescribed fire regimes, involving multiple repeat fires, remain under-examined. This study quantified C stocks in multiple pools after 27 years of a long-term prescribed fire experiment in a mixed-species eucalypt forest in south-eastern Australia. The experimental design included five replications of each of five treatments – a long-unburnt Control, plus a factorial combination of two fire frequencies (c. 3-yearly ‘High’, c. 10-yearly ‘Low’), and two fire seasons (Spring, Autumn) – encompassing up to 7 low-intensity repeat fires over the 27 years.Overall, total C stocks (the sum of all pools: Above-ground biomass, Dead wood, Litter, and Soil) were significantly greater in Control than Fire treatments. Mean total C stock differences (Control minus Fire) were 36 Mg ha−1, and increased with both fire frequency (46 Mg ha−1, Control versus High frequency treatments) and fires in autumn rather than spring (42 Mg ha−1, Control versus Autumn treatments). Mean differences had wide 95% confidence intervals (e.g. 4–67 Mg C ha−1, Control versus Fire), indicating considerable uncertainty about the magnitude of effects of prescribed fire regimes on total C in these native forests. Weighted averaging of linear multiple regression models was used to identify the most important variables for explaining proportional C stock differences ((Control–Fire)/Control), and involved consideration of 85 explanatory variables including measures of fire intensity, fire severity, fuel, pre- and post-fire rainfall, fire weather, and topography. The best regression model explained 80% of variation in total C stock differences between Control and Fire treatments. Consistently important explanatory variables were those representing or associated with fire intensity (flame angle, wind speed, Forest Fire Danger Index), and number of prescribed fires. Measures of fuel load, pre- and post-fire rainfall, and, in particular, fire severity, were less important explanatory variables. The study provides indications of management options for minimising C stock decreases associated with prescribed fire regimes in temperate eucalypt forests; notably, wide fire intervals (closer to 10 than 3 years), and utilising strategies to minimise mean fire intensity, including burning in moist (spring) rather than dry (autumn) conditions.