Attiwill et al. (2014) discussed effects of timber harvesting on the patterns of severity of recent, major fires in south eastern Australian. They concluded that the probability of crown fire partially increased with fuel age in Ash forests, supposedly contrary to results of our recent analyses (Price & Bradstock 2012). In doing so, Attiwill et al. (2013) have erroneously reported our results, which demonstrated that time since logging was inversely related to the probability of high severity crown fires. Attiwill et al. (2014) claimed that our result was only evident for dry forests, not the other types that we investigated (i.e., damp and Ash forests). In doing so they erroneously conflated our statement about weak effects of time since logging, evident in the raw data, with conclusions derived from the selected model that emerged from formal analysis.
Price & Bradstock (2012) investigated the effects of a range of environmental factors (e.g., weather, terrain, forest type) and fuel age (i.e., time since last fire, time since last logging) on the probability on fire severity within four areas burnt in February 2009 in Victoria. Our analyses identified significant effects of weather, forest type (Ash, damp, and dry), time since fire and time since logging, along with a range of interactions between weather, forest type, and terrain (Price & Bradstock 2012, Supporting Information). The effects of time since logging effects model on the probability of crown fire in the selected model were highly significant and negative (i.e., highest probability of crown fire in recently logged forests, Supporting Information). Importantly, no significant interaction between time since logging and forest type was included. Thus, the negative effect of time since logging on crown fire probability occurred in all forest types and was consistent across a range of weather conditions (Supporting Information). We originally illustrated this result for only one forest type (dry forest, Price & Bradstock 2012), but at no time stated or implied that this effect was exclusive to dry forests.
In summary, our analysis shows that recent logging resulted in higher probability of crown fire in a range of forest types, including Ash forest, during the 2009 Victorian fires. The disparity between our results and the putative conclusions of Attiwill et al. (2014) reflects their lack formal analyses. Such analyses are required to discriminate obvious environmental effects (e.g., weather, terrain) on fire severity, from those of land management and time since disturbance. Spatial autocorrelation in fire severity patterns must also be accounted in order to derive valid conclusions (e.g., Murphy & Russell-Smith 2010).
These important omissions render the conclusions of Attiwill et al. (2014) tenuous, given the important influence of weather on fire severity, as demonstrated in recent studies of fire severity in Australian eucalypt forests (e.g., Bradstock et al. 2010; Price & Bradstock 2012; Collins et al. 2014) and other forests elsewhere (e.g., Thompson & Spies 2009). The development of robust conclusions based on their data must therefore await completion of appropriate analyses to account for important, multiple influences on fire severity.