Ecological Resilience of Restored Mediterranean-Climate Woodlands to Experimental Fire.

The ability of restored sites to recover from subsequent disturbances is a key component of restoration success. Resilience is achieved when a restored site returns to its pre-disturbance state, rather than shifts to a different one. In restored fire-prone ecosystems, the drivers of post-fire plant responses and resilience of plant assemblages to fire are underexplored. Exploration of these responses is used to predict and measure the resilience of restored ecosystems to disturbance, including whether the disturbance response was desirable or not. We implemented fine-scale experimental fires in a post-mining restoration chronosequence 14-27 years of age in Banksia woodlands, Western Australia. We sought to understand the effects of restoration age, fire impact, and soil conditions on post-fire regeneration and survival of restored Banksia woodland plant assemblages. To assess early-stage resilience to fire, we calculated four descriptors of ecosystem state: plant species density, species diversity, rarefied richness and functional redundancy, and compared how these changed following fire across the restoration ages and in comparison to nearby reference Banksia woodland. Ordinations and indicator species analyses were used to compare restored and reference sites. In restoration sites, restoration age, fire impact and soil conditions had little effect on plant regeneration and survival. Changes in diversity, rarefied richness and functional redundancy pre- to post-fire in restored sites were typically similar to or less than that observed in reference sites. Broadly, our findings demonstrate the incomplete resilience of restored Banksia woodland to fire. Resprouters typically demonstrated poor resilience, through significant decreases in diversity and rarefied richness following fire in restored sites. They were under-represented in restored Banksia woodlands, so further investigations into the establishment of resprouters in restored environments are required. Our findings also highlight the importance of utilising reference data and a broad range of descriptors to fully understand responses of restored plant assemblages to fire.