Are longunburnt eucalypt forest patches ... - Wiley Online Library

Applied Vegetation Science 14 (2011) 172–180

Are long-unburnt eucalypt forest patches important for the conservation of plant species diversity? T.D. Penman, M. Beukers, R.P. Kavanagh & M. Doherty

Keywords Fire management; Plant community dynamics; Prescribed fire; Wildfire Abbreviations AIC = Akaike Information Criterion; ANOSIM = analysis of similarities; BFBPI = bush fire behaviour potential index; NSW = New South Wales; PB = prescribed burn; SE = standard error; SF = sclerophyll forest; SLRV = spatially lagged response variable; WF = wildfire Nomenclature Harden (1993) Received 17 January 2010; Accepted 27 July 2010. Coordinating Editor: Amy Symstad

Penman, T.D. (corresponding author, [email protected]) & Kavanagh, R.P. ([email protected]. gov.au): Forest and Rangeland Ecosystems, Industry and Investment NSW, PO Box 100, Beecroft, NSW 2119, Australia Penman, T.D.: Bushfire Cooperative Research Centre, Level 5, 340 Albert St, East Melbourne, Victoria 3002, Australia Beukers, M. ([email protected]. gov.au): Parks and Wildlife Group, NSW Department of Environment, Climate Change and Water, PO Box 1967 Hurstville, NSW 2220, Australia Doherty, M. ([email protected]): CSIRO Sustainable Ecosystems, GPO Box 284, Canberra, ACT 2601, Australia

Abstract Question: Are long-unburnt patches of eucalypt forest important for maintaining floristic diversity? Location: Eucalyptus forests of southeastern New South Wales, Australia. Methods: Data from 976 sites representing a range of fire history from three major vegetation formations – shrubby dry sclerophyll forest (SF), grassy dry SF and wet SF – were analysed. Generalized linear models were used to examine changes in species richness with increasing time since wildfire and analysis of similarities to examine changes in community composition. Chi-squared tests were conducted to examine the distribution of individual species across four time since fire categories. Results: Plant species relationships to fire varied between the three formations. Shrubby dry SF supported lower plant species richness with increasing time since wildfire and this was associated with shifts in community composition. Grassy dry SF showed significant shifts in community composition and species richness in relation to time, with a peak in plant species richness 20–30 yr post fire (either prescribed fire or wildfire). Wet SF increased in species richness until 10–20 yr post wildfire then displayed a general declining trend. Species richness in each vegetation type was not related to the fire frequencies and fire intervals observed in this study.

Conclusions: Long-unburnt (30–50 yr post wildfire) forests appeared to play a minor role in the maintenance of plant species diversity in dry forest systems, although this was more significant in wet forests. Maintenance of a range of fire ages within each vegetation formation will assist in maintaining floristic diversity within regions.

Introduction Fire regimes (sensu Gill 1975) are a major factor in determining the composition and structure of vegetation communities worldwide (Bond & van Wilgen 1996; Gill 1999; Bond & Keeley 2005). Management of fire within natural areas is a contentious issue (Gill 2001; McCarthy & Lindenmayer 2007), particularly where there is a risk of damage to property and loss of human life (Gill & Williams 1996; Fernandes & Botelho 2003). Conversely, some perceive all fire to be an ecologically destructive

172

force that should be eliminated (Schultz 2008). Conservation managers are faced with the challenge of prescribing and implementing fire regimes that satisfy the economic and social requirements for a region, while achieving a positive outcome for biological diversity (Morrison et al. 1996; Bradstock & Gill 2001; Bradstock et al. 2005). Often fire management strategies are developed in the absence of detailed scientific data for most species and plant communities. While all aspects of the fire regime (sensu Gill 1975) have the potential to affect biological

Applied Vegetation Science Doi: 10.1111/j.1654-109X.2010.01108.x r 2010 International Association for Vegetation Science

Role of long-unburnt eucalypt forest patches

T.D. Penman et al.

diversity, most of the attention to date has been on the impacts of fire frequency (Cary & Morrison 1995; York 2000; Franklin et al. 2001; Gundale et al. 2005). Frequent homogeneous fire, particularly in heath communities, will result in a reduction in woody shrub diversity and an increase in the diversity of herbaceous species (Keith 1996; Woinarski et al. 2004; Andersen et al. 2005; Penman et al. 2008b) resulting in a change to vegetation structure. Changes to vegetation structure can affect habitat quality and populations of native fauna (Monamy & Fox 2000; Catling et al. 2001; Fox et al. 2003). Relatively little attention has been given to the effect of long periods without fire on the composition of vegetation communities. Hypothetical changes in standing plant species richness with increasing time since fire are presented by Gill (1999) based on the initial floristics model of Egler (1954), later termed ‘complete initial floristics model’ (Wilson et al. 1992). These models predict an initial increase in species richness immediately after a fire as species regenerate from seed or lignotubers. Following this period, a plateau occurs and then a subsequent decline in standing species richness. If the long-term declines represent a permanent loss of species from the area, then this represents a loss of biological diversity. Some plant species lost from the standing community may persist as viable seed stored in the soil (Keith 1996). Under such a scenario, long-unburnt patches would contribute little to the maintenance of regional vascular plant diversity even though they may contribute to faunal diversity. Alternatively, if the net loss in total species richness is associated with gains in new species, then long-unburnt patches would have an important role in maintaining vascular plant diversity within a region. Here we aimed to examine whether long-unburnt patches of Eucalyptus forest were important for the conservation of floristic diversity. To achieve this, we modelled the long-term changes that occurred in aboveground vegetation communities with increasing time since wildfire and related these to existing models of plant community dynamics. Three major vegetation formations (Keith 2004) were considered for southeastern Australia: shrubby dry sclerophyll forests (SF), grassy dry SF and wet SF. From this information, we discuss fire management strategies for conserving vascular plant diversity.

west, and by the Pacific Ocean in the east (Fig. 1). A large proportion of the study area is forested, publicly owned land, either State Forest or National Park (Fig. 1), with the remainder being agricultural land or townships. Detailed descriptions of the study region are provided in NSW National Parks and Wildlife Service (1998, 2000) and Keith & Bedward (1999). Vegetation data for a total of 967 sites spread across the study area were used in the analysis (Fig. 1). All data were originally collected for the purposes of native vegetation classification and mapping (Keith & Bedward 1999; Beukers & Miles 2005; Gellie 2005; Tozer et al. 2006). The sites used in this analysis were from three major formations: shrubby dry SF (n = 458), grassy dry SF (n = 130) and wet SF (n = 379), based on the associations described in Beukers & Miles (2005). At all sites, a 400 m2 plot

Materials and Methods The study area encompassed 19 215 km2 of southeastern Australia between longitudes 148.0–150.01 E and latitudes 35.5–37.51 S. The area is bounded by the Victorian border in the south, the mouth of the Clyde River near Batemans Bay in the north, Cooma and Braidwood in the

Fig. 1. Map of the study area and location of the study sites. SF = sclerophyll forest.


173


T.D. Penman et al.

(20 m 20 m) was used to record all vascular plant taxa present (following taxonomy of Harden 1993) using a modified Braun-Blanquet cover abundance score (Keith & Bedward 1999). Standing species richness (hereafter species richness) was calculated as the number of unique species or recognized subspecies occurring on the plot. Sampling was not undertaken during winter as seasonal conditions could influence the number of taxa recorded, particularly for annuals. Fire history data were derived for combined databases from the NSW Department of Environment and Climate Change and Forests NSW. These databases contain records of the occurrence and extent of both prescribed fire (since 1967–1968) and wildfire (since 1946). Sites were only included in this analysis if there were records of wildfires within the last 50 yr (i.e. all sites in this analysis were burnt and the maximum time since wildfire was 50 yr). This time period was chosen because we could not be certain of fire history prior to 1960. We calculated the time since wildfire, prescribed fire or any fire, the number of wildfires, prescribed fires or any fires since 1960 and the minimum interval between successive wildfires, prescribed fire or any fire since 1960. In the case of sites with only one fire, the fire interval was set as the maximum of the time since the last fire or 50 (maximum time since fire) minus the time since the last fire. A summary of these data is presented in Table 1. A limitation of the data is that the boundaries of earlier fires were not accurately recorded and not all prescribed fires undertaken by either NSW National Parks and Wildlife Service or Forests NSW were accurately mapped in these databases. Fires that have been omitted are likely to be of smaller area, lower intensity, and occurred earlier than more recent fires. Two other variables were included in the analysis to account for natural variation in vascular plant species richness in the landscape. First, we calculated a spatially lagged response variable (SLRV) to account for the spatial autocorrelation in the dataset (Haining 2003). This vari-

able was calculated for each site by summing the product of a weight and the response of all sites within the distance over which spatial autocorrelation was observed divided by the sum of the weights for that site. Weights used in this study were the inverse distance between sites. The distance over which spatial autocorrelation occurred was calculated using a semi-variogram (Haining 2003) and determined to be 10 km. The SLRV provides a relative measure of species richness in the neighbouring 10 km, which is independent of vegetation formation and environmental variables. Second, we derived an index of bushfire behaviour potential (BFBPI) which accounts for both slope, aspect and vegetation structure (NSW National Parks and Wildlife Service 2008a, b), all of which are known to influence native plant species richness in the region (Penman et al. 2008b). Species richness models were developed in a generalized linear model framework with a Gaussian distribution. We wished to examine how plant species richness differed between sites with varying fire histories. Eleven models were run for each of the three vegetation types (Appendix S1). The response variable was the square root of plot-based vascular plant richness with the transformation made to achieve normality. Predictor variables tested were the SLRV, BFBPI and the range of fire variables. Four categories of time since fire were used: 0–10 yr, 10–20 yr, 20–30 yr and 30–50 yr. The 30–50 yr time since fire category was necessary as there were insufficient data to support both 30–40 yr and 40–50 yr categories separately. Time since fire was converted to a categorical variable for two reasons. First, categories permit non-linear responses in plant species richness to time since wildfire. Second, data for the occurrence of wildfires were clustered according to larger fire years and small fire years because of the strong influence of weather and drought cycles. This meant there was not an even spread of data for time since fire when taken as a continuous variable. Categories provided a more even spread of data across the varying time since fire categories. In any given model, only a

Table 1. Mean ( standard error) frequency of fires, fire interval and time since fire over the last 50 yr recorded at each floristic sample site in each of the three vegetation formations. WF = wildfire, PF = prescribed fire, AF = any fire, SF = sclerophyll forest. Formation

Fire variable

WF

PF

AF

Shrubby dry SF

Number Interval (yr) Time since fire (yr) Number Interval (yr) Time since fire (yr) Number Interval (yr) Time since fire (yr)

1.37 0.03 24.09 0.56 19.66 0.40 1.12 0.03 27.80 0.90 20.52 0.08 1.26 0.03 24.40 0.55 21.69 0.51

0.90 0.05 37.16 0.09 31.20 0.95 0.64 0.08 40.60 1.50 35.66 1.69 0.65 0.05 40.46 0.90 35.13 1.01

2.26 0.06 16.36 0.66 14.17 0.38 1.75 0.09 18.82 1.21 17.81 0.81 1.91 0.06 17.11 0.66 17.71 0.55

Grassy dry SF

Wet SF

174



T.D. Penman et al.

single fire history variable was included as they were all correlated to some extent and we wished to avoid the effects of multi-collinearity (Chatterjee et al. 2000). The support for models was determined using the Akaike information criterion (AIC) (Akaike 1973), which balances the benefits of models fitted with the costs of model complexity. We considered all models within two AIC points of the best model (i.e. lowest AIC), as these models are considered to have substantial support (Burnham & Anderson 2002). Analyses were conducted in the R statistical package v 2.8.1 (R-Development Core Team, R Foundation for Statistical Computing, Vienna, AT). We used a chi-square test to determine whether the occurrence of individual species changed with increasing time since fire. For each species, a 2 2 contingency table was created using the four time-since-fire categories with counts being the number of sites at which each species was recorded across all plots in each category. Within each formation, responses were assessed for all species occurring on 30 or more plots. We considered responses significant if the test resulted in P o 0.10. Trends in species richness were categorized in relation to the following ‘time since fire’ categories: recent (0–10 yr or 0–10 yr plus 10–20 yr); mid (10–20 yr, 20–30 yr or both 10–20 yr and 20–30 yr); and old (30-50 yr). Analyses were conducted in the R statistical package v 2.8.1 (R-Development Core Team). Comparisons of the community composition at varying times since fire were made for the three vegetation formations using analysis of similarities (ANOSIM) (Clarke 1993). The cover abundance score was used as the species response for each site and the Bray–Curtis similarity measure was applied to these values (Bray & Curtis 1957). Comparisons were made for time since wildfire and time since any fire; time since prescribed fire was not considered because of the lack of response in the univariate analysis (see below). We adopted a more conservative P-value of significance (P = 0.01) because of the large number of post hoc comparisons (6). ANOSIM calculates the magnitude of differences between two samples based on the R-statistic, which ranges from 0 to 1. If the similarities between and within groups are, on average, equal then R = 0, but if all sites within a group are more similar to each other than to sites in other groups then R = 1 (Clarke 1993). We consider both the R statistic and the P-value for comparisons as it is possible for R to be significant ‘yet inconsequentially small’ with large samples (Clarke 1993) All analyses were conducted using the PRIMER package v 6.1.6 (Clarke & Warwick 2001).

Results A total of 1094 species were recorded throughout the study area, with five to 123 species observed per plot and a mean

of 34.2 13.2 (SD) species per plot. Sixty-six of these species were classed as trees, of which the most frequently encountered were Eucalyptus cypellocarpa (360 plots), Eucalyptus sieberi (359 plots) and Eucalyptus muelleriana (294 plots). A total of 377 shrub species were observed of which Leucopogon lanceolatus (472 plots), Persoonia linearis (465 plots) and Platysace lanceolata (366 plots) were the most common. A total of 533 herbaceous species were recorded in the study area, with the most commonly occurring being Lomandra longifolia (571 plots), Poa meionectes (518 plots), Dianella caerulea (451 plots) and Viola hederacea (448 plots). Forty-seven fern species were recorded of which Pteridium esculentum (639 plots) was the most common, with Blechnum catilagineum the next most common, but occurring only on 138 plots. Forty-nine vine species were recorded in the study, the most frequent being Billardiera scandens (378 plots) and Glycine clandestine (352 plots). The remaining species (n = 22) were classified as aquatic species or epiphytes. Fire variables influenced plant species richness in all three vegetation formations considered. The best set of models (i.e. within 2 AIC points of the best model) for shrubby dry SF included either time since the last wildfire or time since any fire (Appendix S1). These models explained 42% and 38% of the variance in the data, respectively. Both models exhibited a significant, almost linear, decline over time (Fig. 2a). For the grassy dry SF, three models were included in the best set of models (Appendix S1). The set included models 1 and 2, which did not include any fire variables (only the SLRV [model 1] or BFBPI [model 2] variables), and model 5, which included SLRV, BFBPI and time since any fire. In model 5, species richness remained stable for 0–20 yr after fire, increased significantly in the 20-30 yr period, and then returned to original values in the period 3050 yr after fire (Fig. 2b). Models explained 18 % (models 1 and 2) or 25% (model 5) of the variance in the dataset. Time since wildfire was not significant at the P = 0.05 level in any model for grassy dry SF. Only one model comprised the best set of variables for the wet SF: this was model 3, which included SLRV, BFBPI and time since wildfire. This model explained 27% of the variance in the dataset. Species richness increased significantly from the 0–10 yr period to the 10–20 yr period, after which, species richness returned to similar values in the 20–30 yr period, then rose again in the 30–50 yr period (Fig. 2c). When included in a model for any formation, time since prescribed fire, interval between fires (WF, PB, Any fires) or the number of fires (WF, PB, Any fires) were not significantly related to vascular plant species richness at the P = 0.05 level. For all formations, there was a positive relationship between species richness and the SLRV. Only


175


T.D. Penman et al.

Fig. 2. Predicted changes in plant species richness with increasing time since fire (a) shrubby dry sclerophyll forest (SF) versus time since wildfire, (b) grassy dry SF versus time since any fire, and (c) wet SF versus time since wildfire. Dashed lines represent 95% confidence interval.

in the shrubby dry SF was there a significant negative relationship with the BFBPI. In other formations, no relationship with this variable was observed, but

176

removal of this variable did not result in a significantly better model (i.e. DAIC o 2). Analyses were repeated for species richness in each of the following categories in each formation: herbs, woody shrubs and trees. Models in the best subset of models (presented above) were run again on the three different strata and predictions were made (Fig. 2). In the shrubby dry SF, both the herbs and the woody shrubs showed declining trends over time. The herbs decreased earlier, with plots 20–30 yr since wildfire and 30–50 yr since wildfire having significantly reduced herb species richness compared with the plots 0–10 yr since wildfire (P o 0.05). In contrast, the shrubs showed an overall decline, but only the plots 30–50 yr since wildfire had significantly lower shrub species richness than the plots 0–10 yr since wildfire (P o 0.01). Tree species richness did not vary with time since fire. Identical patterns were seen for time since any fire. In the grassy dry SF, herb species richness appeared to be responsible for the trends seen in overall plant species richness. In these forests, there was a significant increase in herb species richness in the plots 20–30 yr after wildfire (P o 0.05) compared with the plots 0–10 yr since any fire. Shrub species richness in these forests was significantly greater in the plots 10–20 yr and 20–30 yr since any fire than in the plots 0–10 yr since any fire (P o 0.05). Tree species richness did not vary with time since fire. In the wet SF, shrub species richness was significantly greater in all time-periods compared with the plots 0–10 yr since wildfire, although there were some fluctuations in these values. Herb species richness increased in the plots 10–20 yr since wildfire compared with the plots 0–10 yr since wildfire, although the effect was marginal (P = 0.06). Plots in the 20–30 yr and 30–50 yr since wildfire categories had similar values to the plots 0–10 yr after wildfire. Similarly, tree species richness showed a statistically significant increase in the plots 10–20 yr after wildfire (P o 0.05), although the effect is not considered biologically meaningful (i.e. o 1 species). The average number of exotic species per plot in grassy dry SF varied very little in relation to time since fire, with a minimum of 0.88 0.31(SE) in the 0–10 yr category and a maximum of 1.27 0.32 (SE) in the 20–30 yr category. This represents a mean increase of only 0.5 species per plot, less than 10% of the observed increase. The responses of individual species varied between formations. A total of 125 species were recorded on 30 or more plots in the shrubby dry SF (15 trees, 45 woody shrubs and 65 herbaceous species including ten vines and one fern). All species were recorded more frequently on the recently burnt sites (0–10 yr since wildfire, 122 species; 0–20 yr since wildfire, two species), except Corymbia maculata, which was recorded equally across all four timeperiods. In the grassy dry SF, three trees, four woody



T.D. Penman et al.

shrubs and 30 herbaceous species (including three vines and one fern) occurred on 30 or more plots. The majority of species (34) showed no significant differences with time since any fire or time since wildfire. One herb, Pratia purpurascens, was more common in the 0–10 yr and 10–20 yr since fire categories and both Sigesbeckia orientalis (herb) and Angophora floribunda (tree) were more common in the 30–50 yr since fire category. There was a more varied response for the 101 species recorded at 30 or more plots in the wet SF sites (nine trees, 30 shrubs and 62 herbaceous species including 17 vines and nine ferns). Sixty-two species showed no response to changing time since wildfire. Four herb species (Hypericum gramineum, Geranium potentilloides, Poranthera microphylla and Veronica calycina) were more common in the recently burnt sites, with the vine Clematis glycinoides and the obligate seeding shrub Pittosporum undulatum less common in these sites compared with the other three categories. Twenty-one species were more common in the mid-range sites (10–20 yr and 20–30 yr post wildfire). These comprised three tree species, five resprouting shrubs, three obligateseeding shrubs, five resprouting herbs, two obligate-seeding herbs, two vines and one fern. Nine species were less common in the mid-range sites: two tree species, four resprouting herbs and three ferns. Eight species were more common in the 30–50 yr since wildfire category: four resprouting shrubs (Acacia melanoxylon, Bursaria spinulosa, Lomatia myricoides and Synoum glandulosum), one obligate-seeding shrub (Hakea eriantha) and three resprouting herbs (Galium propinquum, Hydrocotyle laxiflora and Lepidosperma laterale). Shifts in plant community composition were not consistent across all three formations. Increasing time since wildfire resulted in significant shifts in the plant species composition of shrubby dry SF (Table 2), with significant differences also observed in the comparison between the 20–30 yr and the 30–50 yr categories. There was an overall difference between the communities with varying time since any fire, but the post hoc comparisons found no significant differences between any of the categories (Table 3). In the grassy dry SF, there was no overall effect for either the time since wildfire or time since any fire (Tables 2 and 3). Post hoc comparisons found a difference between the categories 20–30 yr and 30–50 yr since wildfire (Table 2). In wet SF, there was an overall effect of increasing time since wildfire and time since any fire. Post hoc comparisons found differences between the 20–30 yr since wildfire category and both the 0–10 yr since wildfire and the 30–50 yr since wildfire categories and between the 10–20 yr and 30–50 yr since wildfire categories (Table 2). Significant differences were also found between the 20–30 yr and 30–50 yr since any fire categories (Table 3).

Table 2. Analysis of similarities (ANOSIM) comparisons of plant communities between categories for time since wildfire. Values presented are the R-statistic in the lower triangle and the P-value in the upper triangle. Shaded values represent significant differences at the P = 0.01 level. R\P

0–10 yr

10–20 yr

20–30 yr

30–50 yr

(a) Shrubby dry sclerophyll forest, overall R = 0.031, P = 0.028 0–10 yr o 0.001 o 0.001 o 0.001 10–20 yr 0.1069 0.798 0.688 20–30 yr 0.1535 0.0155 o 0.001 30–50 yr 0.0884 0.0123 0.0896 (b) Grassy dry sclerophyll forest, overall R = 0.060, P = 0.071 0–10 yr 0.106 0.158 0.229 10–20 yr 0.135 0.255 0.219 20–30 yr 0.092 0.027 0.006 30–50 yr 0.060 0.049 0.141 (c) Wet sclerophyll forest, overall R = 0.044, P o 0.001 0–10 yr 0.468 0.002 0.214 10–20 yr 0.0043 0.022 0.010 20–30 yr 0.09 0.0321 o 0.001 30–50 yr 0.0228 0.0357 0.1144

Table 3. ANOSIM comparisons of plant communities between categories for time since any fire. Values presented are the R-statistic in the lower triangle and the P-value in the upper triangle. Shaded values represent significant differences at the P = 0.01 level. R\P

0–10 yr

10–20 yr

20–30 yr

30–50 yr

(a) Shrubby dry sclerophyll forests, overall R = 0.026, P = 0.031 0–10 yr 0.351 0.262 0.240 10–20 yr 0.004 0.372 0.224 20–30 yr 0.036 0.016 0.195 30–50 yr 0.030 0.032 0.034 (b) Grassy dry sclerophyll forests, overall R = 0.060, P = 0.072 0–10 yr 0.276 0.091 0.627 10–20 yr 0.025 0.022 0.300 20–30 yr 0.079 0.134 0.342 30–50 yr 0.027 0.029 0.009 (c) Wet sclerophyll forests, overall R = 0.045, P o 0.001 0–10 yr 0.088 0.12 0.005 10–20 yr 0.008 0.14 0.030 20–30 yr 0.044 0.042 o 0.001 30–50 yr 0.051 0.046 0.205

Discussion Fire is important factor in Australian ecosystems, but the effects of fire vary between vegetation formations. All three vegetation formations varied with increasing time since fire, although only shrubby dry SF and wet SF appeared to conform to the initial floristic model predictions of Egler (1954) and Gill (1999). Although we observed no significant effect of either fire interval or fire frequency on plant species richness in the three formations considered, this study did not have good power to detect these effects, as the


177


T.D. Penman et al.

fire intervals were uniformly high and the fire frequencies uniformly low in all formations. Changes in species richness for both the shrubby dry SF and wet SF appeared to conform with the trends of the initial floristics model. That is, plant species richness is expected to decline in relation to time since disturbance, in this case fire. Consistent with other studies in the region (Penman et al. 2008b, 2009b) herb species declined more rapidly than shrub species, reflecting the greater longevity of shrubs compared with herbs. The expected peak in species richness was not seen in the data for shrubby dry SF, probably because the sites were grouped into decadal categories in relation to time since fire. Penman et al. (2009a) found that species richness in shrubby dry SF peaks approximately 5 yr after a wildfire and then declines for at least 33 yr (the period over which that study was conducted). In the current study, the peak in species richness would have been incorporated into the first category of 0–10 yr post fire. A different trajectory was seen in the wet SF, with the time taken to plateau for wet SF being greater than the 0-5 yr reported for shrubby dry SF (Fig. 2; Penman et al. 2009a). A peak in wet SF species richness 10–20 yr post wildfire was followed by a fluctuating but decreasing trend. Natural fire intervals in these forests are thought to be significantly longer than in dry sclerophyll forests (Kenny et al. 2004) and therefore it is likely that the changes observed in shrubby dry SF occur more rapidly than in wet SF. Changes in floristic species richness with increasing time since wildfire were relatively small at the plot scale for both shrubby dry SF and wet SF (Fig. 2). Species richness only considers the presence of a species in the plot and does not consider the relative abundance of species. The initial floristics model predicts that all species occur at a site in the years immediately after a fire and through changes in the relative abundances of species (including disappearing from the visible plant community) the community changes over time. Our analysis of the changes in community composition and species analysis supported this, particularly for the shrubby dry SF (Table 2). Results for the grassy dry SF were least related to time since fire (Appendix S1) and did not conform to the initial floristics model. In grassy dry SF, species richness increased in the older time since fire categories (20–30 yr and 30–50 yr) compared with the younger time since fire categories. Differences in the older time since fire categories in grassy dry SF represented invasions of species into these sites, possibly supporting a relay floristics model where successive groups of species invade a community as conditions become suitable (Clements 1916; Connell & Slatyer 1977). Other studies in grass-dominated systems have found that woody shrubs or exotic species invade longunburnt patches (Swaine et al. 1992; Bond & van Wilgen

178

1996; Woinarski et al. 2004), but this was not the case in our study. The increase in species richness in grassy dry SF in the 20–30 yr period was largely driven by an increase in herbaceous species richness (Fig. 2b) although, in the species analysis, only one herb was found to be recorded more frequently in the older time since fire category (S. orientalis). Lunt (1997) reported similar results from grassy dry SF, with significant shifts in community composition with long periods without fire driven by changes in herbaceous species composition. Long-unburnt sites play a relatively minor role in the maintenance of plant species diversity in the dry sclerophyll forests of the region, but the opposite is true for wet SF. Only in the wet SF did we identify a number of species that were more common in the long-unburnt sites. The limitations of the data and the analytical approach taken (i.e. forests burnt more than 50 yr ago could not be incorporated) did not allow for the consideration of rarely occurring species, which warrant further attention. Results of this study indicate that strategies to ensure that patches of long-unburnt wet SF exist within the region will be important for maintaining regional floristic diversity. We acknowledge that the pursuit of high vascular plant biodiversity is not always the preferred objective in sustainable forest management. Long-unburnt patches of forest create habitat structure in a form that is not always available in the recently burnt patches of forest (McCarthy et al. 1999; Catling et al. 2001). The conservation of total biodiversity will therefore require long-unburnt patches of these formations within the landscape. Other reasons for including long-unburnt patches in landscapes may include the conservation of threatened species, refugia or corridors, maintenance of water quality or timber quality. Management strategies that allow for the occurrence of fire are likely to be more important for these systems than the maintenance of long-unburnt areas. However, the data are insufficient to enable comment on the frequency at which fires should occur. All three communities responded to wildfire, with more varied responses to prescribed fire. Species richness in shrubby dry SF was significantly affected by both time since wildfire and time since any fire, but not time since prescribed fire, suggesting that the effect of prescribed fire on vascular plant diversity in shrubby dry SF is relatively small (Penman et al. 2008a, b). Prescribed fire was found to have no effect on floristic diversity in wet SF. However, only 153 of the 379 wet SF sites in the dataset had prescribed fire recorded in the last 50 yr. The two reasons for this are (1) management recognition that wet SF requires a long interval (30–50 yr) between fires to maintain species diversity between fires (Kenny et al. 2004) and (2) their high moisture levels reduce the ability of management to safely implement prescribed fires. In contrast to the other two formations, the type of the fire in



T.D. Penman et al.

grassy dry SF was not important, suggesting that prescribed fire could be used to achieve positive ecological responses in these systems. Further research is needed to determine whether prescribed fires can be safely implemented in shrubby dry SF and wet SF in a manner that will elucidate a wildfire response in vascular plants.

Acknowledgements The study was based on a dataset owned and maintained by NSW Department of Environment, Climate Change and Water. The Bushfire Co-operative Research Centre funded the analysis and publication of these results. Alan York, Richard Thornton, Andy Stirling, Frank Lemckert, Brad Law and two anonymous referees provided valuable comments on an earlier draft of this manuscript.

References Akaike, H. 1973. Information theory as an extension of the maximum likelihood principle. In: Petrov, B.N. & Csaki, F. (eds.) Second international symposium on information theory. pp. 267–281. Akademiai Kiado, Budapest, HU. Andersen, A.N., Cook, G.D., Corbett, L.K., Douglas, M.M., Eager, R.W., Russell-Smith, J., Setterfield, S.A., Williams, R.J. & Woinarski, J.C.Z. 2005. Fire frequency and biodiversity conservation in Australian tropical savannas: implications from the Kapalga fire experiment. Austral Ecology 30: 155–167. Beukers, M. & Miles, J. 2005. A classification of vegetation assemblages found on the far south east of NSW. Internal Report: NSW Department of Environment and Climate Change. Bond, W.J. & Keeley, J.E. 2005. Fire as a global ‘herbivore’: the ecology and evolution of flammable ecosystems. Trends in Ecology & Evolution 20: 387–394. Bond, W.J. & van Wilgen, B.W. 1996. Fire and plants. Chapman & Hall, London, UK. Bradstock, R.A. & Gill, A.M. 2001. Living with fire and biodiversity at the urban edge: in search of a sustainable solution to the human protection problem in southern Australia. Journal of Mediterranean Ecology 2: 179–195. Bradstock, R.A., Bedward, M., Gill, A.M. & Cohn, J.S. 2005. Which mosaic? A landscape ecological approach for evaluating interactions between fire regimes, habitat and animals. Wildlife Research 32: 409–423. Bray, J.R. & Curtis, J.T. 1957. An ordination of the upland forest communities of Southern Wisconsin. Ecological Monographs 27: 325–349. Burnham, K.P. & Anderson, D.R. 2002. Model selection and multimodel inference: a practical information-theoretic approach. Springer, New York, NY, US. Cary, G.J. & Morrison, D.A. 1995. Effects of fire frequency on plant species composition of sandstone communities in the

Sydney region: combinations of inter-fire intervals. Australian Journal of Ecology 20: 418–426. Catling, P.C., Coops, N.C. & Burt, R.J. 2001. The distribution and abundance of ground-dwelling mammals in relation to time since wildfire and vegetation structure in southeastern Australia. Wildlife Research 28: 555–564. Chatterjee, S., Hadi, A.S. & Price, B. 2000. Regression analysis by example. John Wiley and Sons, New York, NY, US. Clarke, K.R. 1993. Non-parametric multivariate analyses of changes in community structure. Australian Journal of Ecology 18: 117–143. Clarke, K.R. & Warwick, R.M. 2001. Change in marine communities: an approach to statistical analysis and interpretation. 2nd ed. PRIMER-E, Plymouth, UK. Clements, F.E. 1916. Plant succession. Carneige Institute, Washington, DC, US. Connell, J.H. & Slatyer, R.O. 1977. Mechanisms of succession in natural communities and their role in community stability and organization. American Naturalist 111: 1119–1144. Egler, F.E. 1954. Vegetation science concepts. I. Initial floristic composition – a factor in old-field vegetation development. Vegetatio 4: 412–417. Fernandes, P.M. & Botelho, H.S. 2003. A review of prescribed burning effectiveness in fire hazard reduction. International Journal of Wildland Fire 12: 117–128. Fox, B.J., Taylor, J.E. & Thompson, P.T. 2003. Experimental manipulation of habitat structure: a retrogression of the small mammal succession. Journal of Animal Ecology 72: 927–940. Franklin, J., Syphard, A.D., Mladenoff, D.J., He, H.S., Simons, D.K., Martin, R.P., Deutschman, D. & O’Leary, J.F. 2001. Simulating the effects of different fire regimes on plant functional groups in Southern California. Ecological Modelling 142: 261–283. Gellie, N.J.H. 2005. Native vegetation of the Southern Forests: South-east Highlands, Australian Alps, South-west Slopes, and SE Corner bioregions. Cunninghamia 9: 219–253. Gill, A.M. 1975. Fire and the Australian flora: a review. Australian Forestry 38: 4–25. Gill, A.M. 1999. Biodiversity and bushfires: an Australia-wide perspective on plant-species changes after a fire event. In: Gill, A.M., Woinarski, J.C.Z. & York, A. (eds.) Australia’s biodiversity – responses to fire: plants, birds and invertebrates. Environment Australia biodiversity technical paper no. 1. pp. 9–53. Department of the Environment and Heritage, Canberra, ACT, AU. Gill, A.M. 2001. Economically destructive fires and biodiversity conservation: an Australian perspective. Conservation Biology 15: 1558–1560. Gill, A.M. & Williams, J.E. 1996. Fire regimes and biodiversity: the effects of fragmentation of southeastern Australian eucalypt forests by urbanisation, agriculture and pine plantations. Forest Ecology and Management 85: 261–278. Gundale, M.J., DeLuca, T.H., Fiedler, C.E., Ramsey, P.W., Harrington, M.G. & Gannon, J.E. 2005. Restoration


179


T.D. Penman et al.

treatments in a Montana ponderosa pine forest: effects on soil physical, chemical and biological properties. Forest Ecology and Management 213: 25–38. Haining, R. 2003. Spatial data analysis: theory and practice. Cambridge University Press, Cambridge, UK. Harden, G.J. 1993. Flora of New South Wales, volumes 1–4. New South Wales University Press, Kensington, AU. Keith, D.A. 1996. Fire-driven extinction of plant populations: a synthesis of theory and review of evidence from Australian vegetation. Proceedings of the Linnean Society of New South Wales 116: 37–78. Keith, D.A. 2004. Ocean shores to desert dunes: the native vegetation of New South Wales and the ACT. Department of the Environment and Conservation, Hurstville, NSW, AU. Keith, D.A. & Bedward, M. 1999. Native vegetation of the South East Forests region, Eden, New South Wales. Cunninghamia 6: 1–218. Kenny, B., Sutherland, E., Tasker, E. & Bradstock, R. 2004. Guidelines for ecologically sustainable fire management. NSW National Parks and Wildlife Service, Hurstville, NSW, AU. Lunt, I.D. 1997. Effects of long-term vegetation management on remnant grassy forests and anthropogenic native grasslands in south-eastern Australia. Biological Conservation 81: 287–297. McCarthy, M.A. & Lindenmayer, D.B. 2007. Info-gap decision theory for assessing the management of catchments for timber production and urban water supply. Environmental Management 39: 553–562. McCarthy, M.A., Gill, A.M. & Lindenmayer, D.B. 1999. Fire regimes in mountain ash forest: evidence from forest age structure, extinction models and wildlife habitat. Forest Ecology & Management 124: 6. Monamy, V. & Fox, B.J. 2000. Small mammal succession is determined by vegetation density rather than time elapsed since disturbance. Austral Ecology 25: 580–587. Morrison, D.A., Buckney, R.T., Bewiock, B.J. & Cary, G.J. 1996. Conservation conflicts over burning bush in south eastern Australia. Biological Conservation 76: 167–175. NSW National Parks and Wildlife Service. 1998. Eden Fauna Modelling. A report undertaken for the CRA/RFA steering committee. NSW National Parks and Wildlife Service, Internal Report. NSW National Parks and Wildlife Service. 2000. Modelling Areas of Habitat Significance for Vertebrate Fauna and Vascular Flora in the Southern CRA Region – A project undertaken as part of the NSW Comprehensive Regional Assessments. NSW National Parks and Wildlife Service, Internal Report. NSW National Parks and Wildlife Service. 2008a. Fire management manual, policy and procedures for fire management. NSW Department of Environment and Climate Change, Hurstville, NSW, AU. NSW National Parks and Wildlife Service. 2008b. Wadbilliga national park, fire management strategy. NSW Department of Environment and Climate Change, Merimbulla, NSW, AU.

180

Penman, T.D., Binns, D.L. & Kavanagh, R.P. 2008a. Quantifying successional changes in response to forest disturbance. Applied Vegetation Science 11: 261–268. Penman, T.D., Binns, D.L., Shiels, R.J., Allen, R.M. & Kavanagh, R.P. 2008b. Changes in understorey plant species richness following logging and prescribed burning in shrubby dry sclerophyll forests of south-eastern Australia. Austral Ecology 33: 197–210. Penman, T.D., Binns, D.L., Brassil, T.E., Shiels, R.J. & Allen, R.M. 2009a. Long-term changes in understorey vegetation in the absence of wildfire in south-east dry sclerophyll forests. Australian Journal of Botany 57: 533–540. Penman, T.D., Binns, D.L. & Kavanagh, R.P. 2009b. Patch occupancy modelling as a method for monitoring changes in forest floristics: a case study in south-eastern Australia. Conservation Biology 23: 740–749. Schultz, B. 2008. Prescribed Burning in Western Australian Forests. Available at: http://www.green.net.au/boycott/archive/ wdchip2.htm. Accessed April 2009. Swaine, M.D., Hawthorne, W.D. & Orgle, T.K. 1992. The effects of fire exclusion on Savanna vegetation at Kpong, Ghana. Biotropica 24: 166–172. Tozer, M.G., Turner, K., Simpson, C., Keith, D.A., Beukers, P., MacKenzie, B., Tindall, D. & Pennay, C. 2006. Native vegetation of southeast NSW: a revised classification and map for the coast and eastern tablelands. Version 1.0. NSW Department of Environment and Conservation and NSW Department of Natural Resources Sydney. Wilson, J.B., Gitay, H., Roxburgh, S.H., King, W.M. & Tangeny, R.S. 1992. Egler’s concept of ‘Initial floristic composition’ in succession – ecologists citing it don’t agree what it means. Oikos 64: 591–593. Woinarski, J.C.Z., Risler, J. & Kean, L. 2004. Response of vegetation and vertebrate fauna to 23 years of fire exclusion in a tropical Eucalyptus open forest, Northern Territory, Australia. Austral Ecology 29: 156–176. York, A. 2000. Long-term effects of frequent low-intensity burning on ant communities in coastal blackbutt forests of southeastern Australia. Austral Ecology 25: 83–98.

Supporting Information Additional supporting information may be found in the online version of this article: Appendix S1. Akaike Information Criterion outputs for models of plant species richness in each vegetation formation. Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.
