Species Turnover
Models of turnover in species composition of tree/shrub, forb and graminoid communities between burnt and unburnt plots consistently had good explanatory power, which (except for forbs) was higher than equivalent models of relative species richness (trees/shrubs: 32.99 %, forbs: 27.60%, and graminoids: 64.95%). Dissimilarity in species composition of tree/shrub and forb communities was initially marked (Table 3; Fig3a, 3b). Composition of forb communities had returned to that of pre-fire communities approximately ten years after a fire (Fig. 3b), the composition of tree and shrub communities remained dissimilar to that of pre-fire communities ten years after the fire event (Fig. 3a).
Within protected areas graminoid communities exhibited marked turnover immediately following fire events, but these communities resembled pre-fire communities ten tears after the fire event (Table 3; Fig. 3c). Turnover in graminoid species composition in non-protected areas was much lower immediately after a fire than in non-protected areas, but this dissimilarity increased slightly over the ten years following a fire (Fig. 3c). Species turnover in the composition of climber communities did not appear to be associated with any of our predictor variables as the best model had a higher AICc (79.27) than the null model that lacked predictors (AICc 78.14).
Table 3:  Results from model averaging across multiple regression models of species turnover for trees/shrubs, forbs, and graminoids. Results for climbers are not included as no model had a lower AICc than the null model (i.e., one that lacked predictors). Parameter estimates are provided with 95% confidence intervals in brackets.