* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Deakin Research Online - DRO
Survey
Document related concepts
Unified neutral theory of biodiversity wikipedia , lookup
Biological Dynamics of Forest Fragments Project wikipedia , lookup
Introduced species wikipedia , lookup
Ecological fitting wikipedia , lookup
Occupancy–abundance relationship wikipedia , lookup
Latitudinal gradients in species diversity wikipedia , lookup
Island restoration wikipedia , lookup
Theoretical ecology wikipedia , lookup
Biodiversity action plan wikipedia , lookup
Transcript
Deakin Research Online This is the published version: Watson, Simon J., Taylor, Rick S., Nimmo, Dale G., Kelly, Luke T., Haslem, Angie, Clarke, Michael F. and Bennett, Andrew F. 2012, Effects of time since fire on birds : how informative are generalized fire response curves for conservation management?, Ecological applications, vol. 22, no. 2, pp. 685-696. Available from Deakin Research Online: http://hdl.handle.net/10536/DRO/DU:30046985 Reproduced with the kind permission of the copyright owner. Copyright : 2012, Ecological Society of America Ecological Applications, 22(2), 2012, pp. 685–696 Ó 2012 by the Ecological Society of America Effects of time since fire on birds: How informative are generalized fire response curves for conservation management? SIMON J. WATSON,1,2,5 RICK S. TAYLOR,3 DALE G. NIMMO,1 LUKE T. KELLY,1,4 ANGIE HASLEM,3 MICHAEL F. CLARKE,3 1 AND ANDREW F. BENNETT 1 Landscape Ecology Research Group, School of Life and Environmental Sciences, Deakin University, 221 Burwood Highway, Burwood, Victoria 3125 Australia 2 Institute for Land, Water and Society, School of Environmental Science, Charles Sturt University, Thurgoona, New South Wales 2640 Australia 3 Department of Zoology, La Trobe University, Melbourne, Victoria 3086 Australia 4 School of Botany, University of Melbourne, Parkville, Victoria 3010 Australia Abstract. Fire is both a widespread natural disturbance that affects the distribution of species and a tool that can be used to manage habitats for species. Knowledge of temporal changes in the occurrence of species after fire is essential for conservation management in fireprone environments. Two key issues are: whether postfire responses of species are idiosyncratic or if multiple species show a limited number of similar responses; and whether such responses to time since fire can predict the occurrence of species across broad spatial scales. We examined the response of bird species to time since fire in semiarid shrubland in southeastern Australia using data from surveys at 499 sites representing a 100-year chronosequence. We used nonlinear regression to model the probability of occurrence of 30 species with time since fire in two vegetation types, and compared species’ responses with generalized response shapes from the literature. The occurrence of 16 species was significantly influenced by time since fire: they displayed six main responses consistent with generalized response shapes. Of these 16 species, 15 occurred more frequently in mid- or later-successional vegetation (.20 years since fire), and only one species occurred more often in early succession (,5 years since fire). The models had reasonable predictive ability for eight species, some predictive ability for seven species, and were little better than random for one species. Bird species displayed a limited range of responses to time since fire; thus a small set of fire ages should allow the provision of habitat for most species. Postfire successional changes extend for decades and management of the age class distribution of vegetation will need to reflect this timescale. Response curves revealed important seral stages for species and highlighted the importance of mid- to late-successional vegetation (.20 years). Although time since fire clearly influences the distribution of numerous bird species, predictive models of the spatial distribution of species in fire-prone landscapes need to incorporate other factors in addition to time since fire. Key words: birds; chronosequence; conservation; disturbance; fire; mallee; semiarid shrubland; southeastern Australia; succession; time since fire. INTRODUCTION Fire is a natural disturbance that influences the structure of ecological communities throughout the world (Bowman et al. 2009). Fire is also used for land management and conservation; for example, to reduce fuel loads to prevent unplanned fires (Gill and Allan 2008), or to provide particular postfire age classes of vegetation to benefit species (Parr and Andersen 2006). The response of many species to fire is poorly understood, and inappropriate fire regimes are a Manuscript received 10 May 2011; revised 28 September 2011; accepted 28 September 2011. Corresponding Editor: R. L. Knight. 5 Present address: Institute for Land, Water and Society, Charles Sturt University, P.O. Box 789, Albury, New South Wales 2640 Australia. E-mail: [email protected] threatening process for species and communities worldwide (e.g., Woinarski 1999, Covert-Bratland et al. 2006, Fuhlendorf et al. 2006, Sara et al. 2006, Slik and Balen 2006). In Australia, inappropriate fire regimes are associated with five extinct species or subspecies of birds, (e.g., Paradise Parrot Psephotus pulcherrimus, Chisholm 1922, Woinarski 1999), and are recognized as a threatening process for more than 50 species (Woinarski 1999). A key challenge for fire management is to develop an understanding of the responses of species to fire so that fire management can be tailored toward species conservation. Knowledge of the pattern of occurrence or abundance of species with time since fire (‘‘fire response curves’’) is essential for fire management (Driscoll et al. 2010). Fire response curves indicate the extent to which a species depends on particular postfire ages and may also 685 686 Ecological Applications Vol. 22, No. 2 SIMON J. WATSON ET AL. FIG. 1. Generalized postfire response curves of bird species, adapted from the literature (e.g., Whelan et al. 2001, Kavanagh et al. 2004). These curves show eight possible relationships between the frequency of occurrence of bird species and time since fire: incline (lowest occurrence in young vegetation, increases monotonically to be highest in old vegetation); decline (highest occurrence in young vegetation, decreases monotonically to be lowest in old vegetation); bell (defined peak in occurrence in mid-aged vegetation); plateau (lowest occurrence in young vegetation, peaks mid-age vegetation, maintained in older vegetation); irruptive response 1 (occurrence peaks in young vegetation, rapidly declines, lowest in mid-aged and old vegetation); irruptive response 2 (occurrence increases and peaks in young vegetation, rapidly declines, lowest in mid-aged and old vegetation); delayed incline (low occurrence until mid- or old-aged vegetation before increasing in occurrence); null (occurrence not influenced by postfire age). identify time-since-fire thresholds necessary to ensure required habitat resources (Keith et al. 2001, Driscoll et al. 2010). ‘‘Generalized’’ fire response curves, whereby a limited number of patterns represent the responses of many species (see Fig. 1), are an attractive prospect for conservation managers, as they suggest that a limited number of postfire ages will provide habitat for a range of species. However, knowledge of temporal responses to fire is scarce for many species, even in fire-prone environments. Furthermore, there have been few empirical investigations to test the extent to which knowing a species’ response can be used to predict its occurrence. A diverse range of faunal responses to time since fire has been documented (Whelan et al. 2001). Some species favor recently burned vegetation (e.g., Hutto 2008), whereas others prefer long-unburned vegetation (e.g., Clarke et al. 2005). Comparison of responses among multiple species may provide insights into the importance of fire in structuring faunal assemblages and the generality of responses among species. Studies of species’ responses to fire over long time frames commensurate with the duration of successional processes remain rare. Additionally, studies that do investigate long time frames seldom do so across broad geographic scales, which may result in localized effects obscuring successional patterns (Johnson and Miyanishi 2008). Investigations of long-term trends at spatial and temporal scales as large as those at which ecosystem processes such as fire can operate are an important complement to detailed short-term and manipulative studies. The former may reveal patterns not apparent at short timescales, allowing for more informed management decisions (Clarke et al. 2010). In this study, we examine the influence of fire on the occurrence of bird species in semiarid Eucalyptus shrublands of southeastern Australia. We use data from systematic surveys at 499 sites representing a 100-year chronosequence in time since fire, to examine the shape of fire response curves over an extended time frame. We investigate whether fire response curves display generality among species in an assemblage and whether they conform to a priori patterns identified in the literature (e.g.Whelan et al. 2001, Kavanagh et al. 2004) (Fig. 1); and examine the role of different successional stages in supporting bird species. Finally, we test the capacity for postfire response patterns to predict species’ distributions across an extensive fire-prone region. METHODS Study area The study area encompassed ;104 000 km2 in the Murray Mallee region of Australia, centered on the intersection of the states of Victoria, New South Wales and South Australia. The region has a semiarid climate March 2012 AVIAN FIRE RESPONSE CURVES with hot, dry summers and mild winters. Climatic variation across the region forms a south to north gradient of increasing aridity (Pausas and Bradstock 2007), with mean annual rainfall declining from ;350 to 200 mm across this gradient. Mean daily maximum and minimum temperatures in the warmest month range from 328 to 338C and 148 to188C, respectively; and in the coolest from 158 to 168C and 48 to 68C, respectively (raw data obtained from the Australian Bureau of Meteorology). The study was conducted within the widespread ‘‘tree mallee’’ vegetation of the region, dominated by a low canopy (,10 m tall) of multistemmed Eucalyptus species, and an understory of shrubs and perennial and ephemeral grasses (Bradstock and Cohn 2002). Here, we recognize three broad vegetation types (on the basis of floristic surveys) for which spatially explicit maps were generated (Haslem et al. 2010): Triodia Mallee, Chenopod Mallee (also incorporating Shrubby Mallee), and Heathy Mallee. Triodia Mallee vegetation occurs on sandy soils: the overstory is dominated by Eucalyptus dumosa and E. socialis, and the understory typically includes the hummock grass Triodia scariosa, and commonly also contains shrubs such as Acacia rigens, A. wilhelmiana, and Beyeria opaca. Chenopod Mallee occurs on soils with higher clay content, typically in swales between dunes. It has a sparse understory of shrubs (e.g., Olearia spp., Zygophyllum spp.) and chenopod species (e.g., Maireana pentatropis, Enchylaena tomentosa var. tomentosa, and M. pyramidata) and an overstory of E. oleosa subsp. oleosa and E. gracilis (Haslem et al. 2010). Sites in Heathy Mallee were excluded from this analysis (see study design). Mallee vegetation is highly flammable (Bradstock and Cohn 2002). Large fires (.10 000 ha) occur regularly in the study region (16 since 1972), with very large fires (.100 000 ha) occurring once every 10–20 years. Smaller fires (,10 000 ha and mostly less than 100 ha) occur every year. The majority of fires are ignited by lightning (S. C. Avitabile, K. E. Callister, L. T. Kelly, L. Fraser, A. Haslem, D. G. Nimmo, S. J. Watson, S. A. Kenny, R. S. Taylor, L. M. SpenceBailey, A. F. Bennett, and M. F. Clarke, unpublished manuscript). Fires in tree mallee vegetation typically remove both understory and canopy vegetation (Noble and Vines 1993). Mallee eucalypts resprout from underground lignotubers following fire, whereas understory plants regenerate via resprouting from surviving vegetative matter (i.e., stems, roots, and lignotubers) and from seed (Parsons 1968, Cheal et al. 1979). Postfire regeneration produces a relatively dense vegetation layer below 2 m until 15 years postfire, after which vegetation cover reduces in the mid stratum (0.5–2 m) as shrubs become sparser and mallee eucalypts develop a canopy above this height. Canopy height reaches a mean of 5–8 m at ;60 years postfire (Haslem et al. 2011). 687 Study design and data collection We surveyed birds at sites representing a chronosequence from ,1 year to 164 years postfire. Time since fire was measured from the year 2007 (i.e., sites burned in 2006 ¼ 1 year since fire). The postfire age of vegetation at each site was ascertained by using two methods. For sites burned after 1972, the year of burning was determined from maps of fire history based on 15 individual years of satellite imagery, combined with local knowledge of exact fire dates (S. C. Avitabile, K. E. Callister, L. T. Kelly, L. Fraser, A. Haslem, D. G. Nimmo, S. J. Watson, S. A. Kenny, R. S. Taylor, L. M. Spence-Bailey, A. F. Bennett, and M. F. Clarke, unpublished manuscript). For sites burned prior to 1972, fire age was estimated using regression models of the strong positive relationship between the diameter of eucalypt stems and years since fire, which were validated with independent data (Clarke et al. 2010). This investigation was part of a study examining the influence of the properties of fire mosaics on flora and fauna; consequently, bird survey sites were arranged in 28 clusters (landscape units), each encompassing 20 sites (mean distance between landscapes ¼ 130 km, range ¼ 6– 260 km; see Haslem et al. [2010] for a map of study landscapes). Within each landscape, sites generally were separated by 500 m to maximize independence (mean distance between sites within landscapes ¼ 1.5 km, range ¼ 0.34–3.61 km). We excluded from analysis 17 sites predicted to be .100 years since fire (i.e., 101–164 years) as there were too few sites to adequately represent this time period. We also excluded all sites in Heathy Mallee vegetation (n ¼ 44) due to poor representation of some age classes: only one site was ,19 years since fire. The resulting data set consisted of 499 sites representing postfire ages between 1 and 100 years: 176 sites in Chenopod Mallee and 323 in Triodia Mallee (see Appendix A for details of the distribution of survey sites in relation to time since fire). Surveys were conducted on four survey rounds, once each in the Austral spring and autumn of 2006/2007 and 2007/2008. All sites were surveyed four times, twice each by two observers (S. Watson and R. Taylor). At each site, all individuals were counted and recorded within a 60 m radius during a 5-min period. Surveys commenced within 15 min of sunrise, during the time of greatest vocal activity for birds. We recorded the distance from the center of the point count to the location of the first detection of individual birds to permit analysis of detectability. For visual detections, distance was measured using an OPTi-LOGIC 800LH laser range finder (Opti-Logic, Tullahoma, Tennessee, USA). For aural detections, distance was estimated by the observer. Observers trained together to ensure comparability of procedures. STATISTICAL ANALYSIS Variation in detectability of different species, or of the same species in different habitats, is a potential source of 688 Ecological Applications Vol. 22, No. 2 SIMON J. WATSON ET AL. variation in ecological studies (Buckland et al. 2001). To ascertain whether detectability was likely to be a problem in interpreting our data, we undertook multiple covariate distance sampling, MCDS (Buckland et al. 2004, Marques et al. 2007), using the program Distance 5.0 release 2 (Thomas et al. 2006). We modeled variation in detectability of individual species with increasing distance from the observer, also incorporating vegetation density as a covariate in the model. Density of vegetation at each site was obtained by recording the number of contacts of vegetation with a 2-m structure pole at each of four height intervals (0 m, 0.5–1 m, 1–2 m, .2 m), at 50 3 1 m intervals along a transect. Species recorded too infrequently to model individually were grouped with more common species that displayed similar detection characteristics, following Alldredge et al. (2007). All species were successfully detected at the farthest extremities of the point count. The lowest probability of detection at a point was for the combined taxa, Variegated Fairy-wren (Malurus lamberti ) and Splendid Fairy-wren (Malurus splendens) (probability of detection ¼ 0.44, 95% CI 0.290.66, N ¼ 72). Vegetation density did not significantly reduce the detectability of any species. Considering these results, and that our primary aim was to assess site occupancy, modeling presence–absence is an appropriate approach that reduces potential influence of differential detectability of species in different postfire ages. We used generalized additive mixed models, GAMMs (Wood 2006), to model the change in species occurrence with time since fire. Generalized additive models (GAMs) are a nonparametric form of regression modeling that use smoothing functions to model nonlinear relationships (Wood 2006, 2008). Because many of the response shapes that we selected a priori are nonlinear (Fig. 1), GAMs presented an appropriate method (Zuur et al. 2009). Models were fitted using presence–absence data to model the probability of occurrence of species, where presence represents the detected occurrence of a species at a site in any of the four survey rounds. We used presence–absence rather than abundance or density because it is less prone to error from localized data anomalies (Buckland et al. 2001) and because the natural rarity of some species results in many zeros across the data set, rendering Poisson and Gaussian models inappropriate. We developed GAMs in a mixed-modeling framework (GAMMs) to account for the spatial clustering of study sites in landscapes. Mixed models reduce the potentially confounding problems of spatial autocorrelation and are recommended where systematic structuring is present in the data (Zuur et al. 2009). Thus, landscape unit was included as a random factor in each species’ model. To ascertain whether species’ response shapes varied between vegetation types (Triodia Mallee vs. Chenopod Mallee), we fitted models allowing a separate smoothed term for time since fire to be fitted for each level of this categorical variable (Wood 2006). We also included vegetation type as a separate factor variable; an additional variable, ‘‘northing,’’ was included to allow for a site’s location along the gradient of aridity in the region. The amount of smoothing used to model the response to time since fire was automatically selected in the model-fitting process (Wood 2008). Wood (2006) warns that P values for smoothed terms are approximated; consequently, we approach P values near 0.05 with caution, following Zuur et al. (2009). The modeled response of the probability of occurrence of each species to time since fire was plotted, and species were then allocated to a response shape based on its similarity to a priori response shapes adapted from Whelan et al. (2001) (see Fig. 1). We evaluated the GAMMs using deviance explained (D2) as a measure of model fit, and seven-fold crossvalidation to test predictive accuracy (Pearce and Ferrier 2000). This procedure involved splitting the data into seven groups (‘‘folds’’), fitting the GAMMs to data from six folds and predicting the occurrence of species to sites in the seventh fold, which is independent of the data used to generate the model. These predictions for sites in the seventh fold were compared with observed responses to test the predictive accuracy of the model. This process was continued iteratively until predictions had been made for all sites. Evaluation of the predictive performance of models was based on the mean discrimination (and standard error) across all folds. Model discrimination was ascertained using the relative operating characteristic (ROC) curve. The area under the curve (AUC) gives a measure of the usefulness of the model in predicting species occurrence. Scores from 0.5 to 0.7 are considered to have some discrimination, 0.7 to 0.9 indicate reasonable discrimination, and 0.9 to 1.0 show very good discrimination (Pearce and Ferrier 2000). All GAMMs were built in the R statistical environment (R Development Core Team 2009), using the ‘‘mgcv’’ package (Wood 2004, Wood 2006). Crossvalidation methods and source scripts were adapted from Elith et al. (2008). RESULTS We detected 70 species of birds during point counts. We modeled species that occurred at .20 sites and that occurred across the entire region (north and south of the Murray River, a biogeographic barrier), resulting in models for 30 of the 70 species. One species (Greyfronted Honeyeater, Lichenostomus plumulus) was recorded only north of the Murray River, and two species (Mallee Emu-wren and Purple-gaped Honeyeater, Lichenostomus cratitius) only south of the river; these species were not included in the analysis. Response of species to time since fire Of the 30 species modeled, 16 showed significant variation in probability of occurrence with time since fire (Table 1). All 16 responded to time since fire in Triodia March 2012 AVIAN FIRE RESPONSE CURVES Mallee vegetation, but only four species displayed a significant response in Chenopod Mallee (Table 1). We assigned each significant time-since-fire response to one of the a priori response shapes (Tables 1 and 2, Fig. 2). There was uncertainty for some species in distinguishing between an inability to detect a significant response and a null response to time since fire. This was particularly evident in the less fire-prone Chenopod Mallee, for which there was a lower sample size, particularly in younger vegetation. Accordingly, we assigned a category of ‘‘nondetection’’ to species that displayed nonsignificant trends, for which a null response may not be appropriate (e.g., where P values were close to alpha, 0.05 , P , 0.1). In Triodia Mallee, 16 species displayed significant responses to time since fire. Of these, eight displayed an incline, five a bell shape, two a plateau, and one an irruptive response shape (Table 2). In Chenopod Mallee, four species displayed significant time-since-fire responses: of these, two species displayed an incline, one species a decline, and one species a plateau (Table 2). Overall, the null response was the most frequently encountered in both vegetation types (Table 2). There was little concordance between responses of species in the two vegetation types (Table 1). Of the species that displayed a significant response to time since fire, only two (Yellow-plumed Honeyeater, Spinycheeked Honeyeater) had the same response shape in both vegetation types. Two species (Chestnut-rumped Thornbill, White-browed Babbler) displayed different response shapes in each vegetation type, and eight species showed a significant response in one vegetation type and a null response in the other (Fig. 3). A further four species responded in one vegetation type and displayed a nonsignificant trend (nondetection) in the other. Evaluation of model fit and predictive discrimination The deviance explained by models ranged from 0.04 to 0.34 (Table 3), suggesting a high level of variability in their explanatory power. Note that the probability of occurrence of several species was also significantly related to the non-fire variables in the model, northing and vegetation type (Table 3; details in Appendix B). The models for the 16 species that showed a significant response to time since fire had AUC values that ranged between 0.54 and 0.86 (Table 3). There was reasonable predictive discrimination for eight species (AUC . 0.7), some predictive discrimination for seven species (0.6 , AUC , 0.7), and predictive discrimination little better than chance for one species AUC ¼ 0.54 (Table 3). DISCUSSION Influence of time since fire on the occurrence of bird species Our findings are consistent with the view that fire is an important disturbance process that influences avifaunal 689 communities (Brawn et al. 2001). The frequency of occurrence of species can continue to change up to 100 years since fire in mallee ecosystems, as has also been found in contemporaneous work on small mammals and reptiles in this system (Kelly et al. 2011, Nimmo et al., in press). This finding emphasizes the importance of interpreting time-since-fire response curves on a temporal scale that corresponds with that at which the disturbance process can operate. In mallee vegetation, the fire return interval is generally .35 years (S. C. Avitabile, K. E. Callister, L. T. Kelly, L. Fraser, A. Haslem, D. G. Nimmo, S. J. Watson, S. A. Kenny, R. S. Taylor, L. M. Spence-Bailey, A. F. Bennett, and M. F. Clarke, unpublished manuscript), and some sites are likely to be .150 years since fire (Clarke et al. 2010). In this study, many sites of similar time since fire were products of different fire events spread across a vast region. Thus, the significant responses show that species respond in a similar way to vegetation of similar postfire age generated by different fire events. This result is notable, given the effects that the complexity of individual fire events (Whelan et al. 2001, Bain et al. 2008, Lindenmayer et al. 2009) and other factors, such as rainfall (Monamy and Fox 2000, Letnic 2003), can have on species succession. However, moderate model fit and predictive ability for several species indicate that factors additional to time since fire also are influential. In addition to general responses of species to fire, represented by the overall frequency of occurrence (or abundance) of a species in postfire age classes, species may respond unexpectedly to particular fire events. For example, the Eastern Bristlebird (Dasyornis brachypterus), considered to prefer older vegetation in temperate heathland of eastern Australia and to take many years for population recovery after fire (Baker 2000), was found to quickly colonize recently burned sites after a fire that burned in a ‘‘patchy’’ manner (Bain et al. 2008, Lindenmayer et al. 2009). Fire response curves derived from large sample sizes and over broad spatial scales can inform management about general responses of species to fire and their preferred seral stages, but may not necessarily be able to predict responses to particular fire events. Shapes of species’ responses to time since fire Modeling results for 30 species of birds revealed six generalized response shapes that resembled those identified a priori in the literature. Similar responses have also been described for birds in mountain rangeland systems in Europe (Pons and Clavero 2010), suggesting that these patterns, and potentially similar processes, may occur in diverse ecosystems. Different postfire responses can result from interactions of species traits with aspects of the environment influenced by fire (Whelan et al. 2001, Moretti and Legg 2009). The shape of a species’ response to fire is influenced by multiple factors. These include the influence of fires on mortality of the species (Whelan et al. 2001); the 690 Ecological Applications Vol. 22, No. 2 SIMON J. WATSON ET AL. TABLE 1. Relationship between species occurrence and time since fire for 30 bird species in the Murray Mallee region of southeastern Australia, derived from generalized additive mixed models. Triodia Mallee Chenopod Mallee Years since file Species Australian Ringneck, Barnardius zonarius Mulga Parrot, Psephotus varius Variegated Fairywren, Malurus lamberti Striated Grasswren, Amytornis striatus Shy Heathwren, Calamanthus cautus Weebill, Smicrornis brevirostris Chestnut-rumped Thornbill, Acanthiza uropygialis Inland Thornbill, Acanthiza apicalis Spotted Pardalote, Pardalotus punctatus Striated Pardalote, Pardalotus striatus White-eared Honeyeater, Lichenostomus leucotis Yellow-plumed Honeyeater, Lichenostomus ornatus White-fronted Honeyeater, Purnella albifrons Spiny-cheeked Honeyeater, Acanthagenys rufogularis Red Wattlebird, Anthochaera carunculata Brown-headed Honeyeater, Melithreptus brevirostris Striped Honeyeater, Plectorhyncha lanceolata White-browed Babbler, Pomatostomus superciliosus Chestnut Quailthrush, Cinclosoma castanotus Gilbert’s Whistler, Pachycephala inornata N Shape F P Maximum occurence Minimum occurrence Years since fire Shape F P 54 null 0.11 0.745 nd 2.96 0.086 36 null 1.25 0.264 null 1.95 0.163 23 null 1.13 0.312 null 1.60 0.207 37 bell 6.84 ,0.001 0–10, 70–100 null 2.09 0.149 69 null 0.16 null 0.10 0.757 null 0.02 0.9 decline 6.47 0.011 null 0.39 0.535 1.71 0.183 20–50 0.692 324 plateau 5.52 ,0.001 114 irruptive 1 4.20 0.017 61 null 2.13 0.145 225 bell 4.19 0.013 153 incline 9.62 ,0.001 90–100 0–10 nd 2.42 0.12 219 bell 3.9 30–50 90–100 null 0.17 0.679 50–100 0–10 286 plateau 21 0.018 ,0.001 113 null 0.9 0.343 147 incline 5.42 0.021 28 null 1.03 46 nd 20–100 0–10 0–10 20–100 30–50 0–10, 90–100 nd plateau 11.59 ,0.001 null 0.18 0.674 incline 5.23 0.022 0.352 null 1.66 0.199 2.88 0.056 null 0 0.971 20 incline 8.04 0.005 90–100 nd 2.67 0.065 39 bell 3.3 0.018 60–80 5.90 0.016 148 null 0.42 0.598 nd 3.03 0.083 null 0.16 0.694 23 incline 11.7 ,0.001 90–100 90–100 0–10 0–10 0–40, 90–100 incline 0–10 Maximum Minimum occurence occurence 0–10 90–100 50–100 0–10 90–100 0–10 90–100 0–10 March 2012 AVIAN FIRE RESPONSE CURVES 691 TABLE 1. Continued. Triodia Mallee Chenopod Mallee Years since file Years since fire F P Maximum occurence Minimum occurrence Shape F P 4.88 ,0.001 20–40 0–10, 70–100 null 0.01 0.924 ,0.001 90–100 0–10 null 0.05 0.819 6.21 0.013 90–100 0–10 null 1.79 0.17 null 1.5 0.226 null 0.18 0.674 132 null 1.11 0.292 null 1.07 0.301 21 null 1.41 0.236 null 2.44 0.119 45 incline 6.48 0.011 null 0.19 0.667 117 null 0.28 0.599 null 1.06 0.304 21 null 0.07 0.796 null 1.87 0.159 53 incline nd 2.34 0.072 Species N Shape Golden Whistler, Pachycephala pectoralis Rufous Whistler, Pachycephala rufiventris Grey Shrike-thrush, Colluricincla harmonica Crested Bellbird, Oreoica gutturalis Grey Butcherbird, Cracticus torquatus Grey Currawong, Strepera versicolor Willie Wagtail, Rhipidura leucophrys Jacky Winter, Microeca fascinans Red-capped Robin, Petroica goodenovii Southern Scrubrobin, Drymodes brunneopygia 21 bell 38 incline 165 incline 101 14.4 15.6 ,0.001 90–100 90–100 0–10 0–10 Maximum occurence Minimum occurence Notes: The number of sites recorded (N ), the assigned response shape with time since fire in Triodia Mallee and Chenopod Mallee, and the statistical significance of the relationship (F, P) in each vegetation type are presented for each species. Significant responses (P , 0.05) are shown in boldface type. For significant responses, the periods of time since fire at which the species display their highest and lowest occurrence are shown. The abbreviation ‘‘nd’’ means that a nonsignificant trend was present, such that a null response was not appropriate. Two ranges of values are displayed for a species’ minimum occurrence if that species displays a bell-shaped response and its minimum occurrence is equally low at two points along the time-since-fire gradient. dependence of the species on resources or vegetation characteristics that are themselves affected by time since fire, such as food (Murphy and Lehnhausen 1998, Barlow and Peres 2004) and the structural complexity of vegetation (Skowno and Bond 2003); and also the ecological traits of the species, such as its fecundity (Friend 1993), dispersal ability (Brotons et al. 2005), and competitive ability (Fox 1982). In the case of generalized response curves, the incline, plateau, and bell responses indicate that fire events initially have a negative effect on the occurrence of a species, but subsequently the species recolonizes over time. In contrast, irruptive and decline responses indicate that fire events initially facilitate the occurrence of a species. Null responses suggest that species are little affected, or are not affected in a consistent way, by most fire events. In the mallee ecosystem, the shapes of several species’ responses to fire tend to correspond with postfire changes in structural attributes of the vegetation on which they rely. Indeed, it is essential to recognize that many bird species do not respond to fire per se, but to the changes in their habitat associated with fire. For instance, the Striated Pardalote, which uses tree hollows for nesting (Higgins and Peter 2002), displays an incline response. The proportion of hollow-bearing stems at sites also increases with time since fire (Haslem et al. 2011). The White-browed Babbler forages under bark on the lower branches of trees and among litter and debris on the ground, and also displays an incline response. Correspondingly, the amount of decorticating bark and litter also is greater in older vegetation (Haslem et al. 2011). The Striated Grasswren, which is closely associated with Triodia hummock grasses that it uses for refuge and nesting (Higgins et al. 2001), displays a bell-shaped response. Hummock grass peaks in cover at 20–40 years postfire (Haslem et al. 2011), and the peak occurrence of the Striated Grasswren corresponds with this pattern. Interspecific competition may also influence postfire response shapes, in particular contributing to declines in species’ occurrence (Fox 1982). For example, of the five species that displayed a bell-shaped response, three (Spotted Pardalote, Golden Whistler, White-eared Honeyeater) have closely related species (Striated 692 SIMON J. WATSON ET AL. TABLE 2. Summary of species response types, given as the number of bird species that displayed each response in Chenopod Mallee and Triodia Mallee. Shape Triodia Mallee Chenopod Mallee Incline Decline Bell Plateau Irruptive 1 Irruptive 2 Delayed Null Not detected 8 0 5 2 1 0 0 13 1 2 1 0 1 0 0 0 20 6 Ecological Applications Vol. 22, No. 2 Pardalote, Gilbert’s Whistler, Rufous Whistler, Yellowplumed Honeyeater) of similar body size that display incline or plateau responses and are most prevalent in older vegetation. These species may be competitors that drive declines in occurrence (i.e., the downward phase of the bell shape) in older age classes. It is important that the shapes of species responses are interpreted within the bounds of the temporal scale of the study. Incline responses, for instance, must eventually resemble plateau or bell-shaped responses over longer time frames: that is, the frequency of occurrence of a species (e.g., Grey Shrike-thrush; Fig. 2a) must eventually stabilize or decline at sites .100 years since fire. Because sites .100 years since fire are rare, it may FIG. 2. Examples of modeled postfire response curves for bird species (mean 6 SE), demonstrating the similarity of observed and hypothetical response shapes: (a) incline response of Grey Shrike-thrush in Triodia Mallee; (b) decline response of Chestnutrumped Thornbill in Chenopod Mallee; (c) bell response of White-eared Honeyeater in Triodia Mallee; (d) plateau response of Yellow-plumed Honeyeater in Triodia Mallee; (e) irruptive 1 response of Chestnut-rumped Thornbill in Triodia Mallee; (f ) null response of Weebill in Chenopod Mallee. March 2012 AVIAN FIRE RESPONSE CURVES 693 their highest frequency of occurrence in mid- or latersuccessional vegetation (.20 years since fire). Some threatened species of mallee birds that were too rare to model in this investigation have also been found to have a preference for vegetation .20 years since fire (e.g., Malleefowl, Leipoa ocellata (Benshemesh 1990), Mallee Emu-wren, Stipiturus mallee (Brown et al. 2009), and Black-eared Miner, Manorina melanotis (Clarke et al. 2005)). This suggests that for many species in this ecosystem, fire displaces populations from sites and they subsequently recolonize over time. In the event of large wildfires, these species will require refuges or source populations nearby where they are able to survive until they are able to recolonize the burned area when it becomes suitable. Although the influence of any given wildfire event on mortality or emigration from sites is not clear, large wildfires in mallee ecosystems dramatically alter vegetation structure and reduce the frequency of occurrence of a substantial proportion of the avian assemblage for decades. Vegetation of .20 years since fire supported a higher occupancy of 14 of the 16 species modeled, suggesting that mid- and later-seral stages are disproportionately important for many species. However, the frequency of occurrence of six of these species declines between 50 and 100 years since fire, and it is likely that there are further reductions at .100 years since fire, highlighting the dynamic nature of the habitats generated by fire in this ecosystem. The rarity of species that prefer early-successional habitats contrasts with some other systems in which TABLE 3. Measures of model performance for models of the relationship between frequency of occurrence of species and time since fire. FIG. 3. Response shapes (mean 6 SE) for species that had different responses to time since fire in different vegetation types (Triodia Mallee, black line; Chenopod Mallee, gray line): (a) Chestnut-rumped Thornbill (irruptive and decline), (b) White-browed Babbler (bell and incline), (c) White-eared Honeyeater (bell and null). be that stabilization or decline will seldom occur before another fire event. Contrastingly, when fire response curves are measured over a long timescale, as is the case here, very short-term postfire population changes (i.e., ,1 year postfire) may not be detectable. For instance, a species that displays an incline response may be present in early-succession stages, but may rapidly decline in the months following fire. Importance of particular seral stages In different ecosystems, vegetation of different postfire age may prove to be of differing importance for maintenance of biodiversity. Here, of the 16 species that displayed a significant response to time since fire, only one (Chestnut-rumped Thornbill) was more likely to occur in early postfire ages. All other species displayed AUC 2 Species D Mean SE Striated Grasswren ,à Weebill Chestnut-rumped Thornbill Spotted Pardaloteà Striated Pardalote ,à White-eared Honeyeater Yellow-plumed Honeyeater Spiny-cheeked Honeyeater Striped Honeyeaterà White-browed Babbler Gilbert’s Whistler ,à Golden Whistlerà Rufous Whistler Grey Shrike-thrush Willie Wagtail Southern Scrub-robinà 0.22 0.05 0.06 0.19 0.14 0.07 0.24 0.05 0.17 0.09 0.13 0.34 0.04 0.06 0.07 0.16 0.81 0.54 0.64 0.72 0.73 0.67 0.76 0.61 0.77 0.66 0.72 0.86 0.60 0.65 0.69 0.75 0.046 0.051 0.060 0.044 0.026 0.045 0.043 0.043 0.033 0.049 0.079 0.060 0.080 0.037 0.050 0.052 Notes: Values presented are the deviance explained (D2) and mean area under the curve (AUC) from seven-fold crossvalidation (SE in parentheses), for species that displayed a significant response to time since fire. Models also account for variation related to geographic location and vegetation type. See Appendix B for statistical details. Significant response to vegetation type at P , 0.05. à Significant response to geographic position (northing) at P , 0.05. 694 Ecological Applications Vol. 22, No. 2 SIMON J. WATSON ET AL. species capitalize on abundant resources that occur postfire (e.g., Murphy and Lehnhausen 1998, Woinarski 1999), or where open-country species (e.g., farmland birds) colonize early-successional habitats (Jacquet and Prodon 2009). In this study, recently burned vegetation (i.e., ,5 years since fire) was generated from fires that occurred during a period of below-average rainfall (raw data obtained from the Australian Bureau of Meteorology); consequently, early-successional habitats may not have experienced postfire resource flushes that can occur in years of higher rainfall. Further, this study was conducted in large reserves of continuous ‘‘natural’’ vegetation where bird species associated with open country (i.e., farmland birds) generally were not abundant in areas close to recently burned vegetation. Early-successional habitats that are generated adjacent to farmland, or in years following good rain, may display increases of bird species in early-successional environments that were not detected in this investigation. Influence of vegetation type on postfire response patterns Little is known about the interaction of fire with other ecological processes (Driscoll et al. 2010). In this study, response patterns of some species differed between vegetation types, and four times as many species showed a significant response in Triodia Mallee than in Chenopod Mallee. Fire behavior varies between vegetation types in this region. Mallee vegetation with Triodia (i.e., Triodia Mallee) carries fire better than other vegetation under typical climatic conditions, due to the flammable nature of Triodia hummocks (Noble and Vines 1993). Postfire temporal changes in understory and midstory vegetation differ between Triodia Mallee and Chenopod Mallee (Haslem et al. 2011). Triodia Mallee tends to have more distinct changes in vegetation structure with time since fire, evidenced by a greater number of significant postfire responses of structural attributes (Haslem et al. 2011). This greater consistency in vegetation structural succession in Triodia Mallee is likely to be a factor that contributes to the higher number of significant responses of species to time since fire in this vegetation type. Particular postfire seral stages often are identified as important for the persistence of particular species (Hutto 1995, Clarke 2005). This investigation reveals that seral-stage preferences of a species may differ between vegetation types. Thus, conservation strategies that aim to benefit species through the provision of particular seral stages will be most effective where they incorporate postfire successional differences among vegetation types. CONCLUSION The ability of time-since-fire models to predict species’ responses across landscapes and regions is a key issue in determining the usefulness of postfire response curves for management purposes. Here, although approximate- ly half of the 30 species modeled showed significant responses to time since fire, model fit varied greatly and was relatively modest. The predictive capacity of these significant relationships was considered reasonable for about half of the species. Thus, caution is required in using time-since-fire models as a primary method for understanding the geographic distributions of species in fire-prone landscapes. Further, in this study only 30 of 70 species detected could be successfully modeled; additional response types may be represented among the other species where they represent aspects of two or more response curves. Generalized fire response curves represent a significant advance in understanding how species respond to fire and in identifying the seral stages in which multiple species are more common. A strong empirical basis for such response curves is essential when time since fire is used as a metric in conservation management, such as to determine whether vegetation should be burned or protected to increase species diversity (Richards et al. 1999) or to create mosaics of age classes (Parr and Andersen 2006). However, the next challenge is to go further and better understand additional factors that interact with time since fire in determining species occurrence, such as postfire rainfall patterns, effects of herbivore grazing on vegetation, and interactions with other species (predators, competitors). ACKNOWLEDGMENTS This research was conducted as part of the Mallee Fire and Biodiversity project. Funding and support for this project was provided by Land and Water Australia, Department for Environment and Heritage (South Australia), Parks Victoria, Department of Sustainability and Environment (Victoria), Mallee Catchment Management Authority (Victoria), New South Wales National Parks and Wildlife Service, Department of Environment and Climate Change (NSW), Lower Murray– Darling Catchment Management Authority (NSW), Natural Heritage Trust, Birds Australia, Australian Wildlife Conservancy, and the Murray Mallee Partnership. We are also greatly appreciative to the Doyle and Barnes families for allowing us access to Petro and Lethero Stations, respectively, and to Jane Elith, who provided advice on statistical analyses. LITERATURE CITED Alldredge, M. W., K. H. Pollock, T. R. Simons, and S. A. Shriner. 2007. Multiple-species analysis of point count data: a more parsimonious modelling framework. Journal of Applied Ecology 44:281–290. Bain, D. W., J. R. Baker, K. O. French, and R. J. Whelan. 2008. Post-fire recovery of eastern bristlebirds (Dasyornis brachypterus) is context-dependent. Wildlife Research 35:44– 49. Baker, J. 2000. The Eastern Bristlebird: cover-dependent and fire-sensitive. Emu 100:286–298. Barlow, J., and C. A. Peres. 2004. Ecological responses to El Niño-induced surface fires in central Brazilian Amazonia: management implications for flammable tropical forests. Philosophical Transactions of the Royal Society of London B 359:367–380. Benshemesh, J. 1990. Management of Malleefowl with regard to fire. Pages 206–211 in J. C. Noble, P. J. Joss, and G. K. Jones, editors. The mallee lands: a conservation perspective. CSIRO, Melbourne, Australia. March 2012 AVIAN FIRE RESPONSE CURVES Bowman, D. M. J. S., et al. 2009. Fire in the Earth system. Science 324:481–484. Bradstock, R., and J. S. Cohn. 2002. Fire regimes and biodiversity in semi-arid mallee ecosystems. Pages 238–258 in R. Bradstock, J. E. Williams, and A. M. Gill, editors. Flammable Australia: the fire regimes and biodiversity of a continent. Cambridge University Press, Cambridge, UK. Brawn, J. D., S. K. Robinson, and F. R. Thompson. 2001. The role of disturbance in the ecology and conservation of birds. Annual Review of Ecology and Systematics 32:251–276. Brotons, L., P. Pons, and S. Herrando. 2005. Colonization of dynamic Mediterranean landscapes: where do birds come from after fire? Journal of Biogeography 32:789–798. Brown, S., M. Clarke, and R. Clarke. 2009. Fire is a key element in the landscape-scale habitat requirements and global population status of a threatened bird: the Mallee Emu-wren (Stipiturus mallee). Biological Conservation 142:432–445. Buckland, S. T., D. R. Anderson, K. P. Burnham, J. L. Laake, D. L. Borchers, and L. Thomas. 2001. Introduction to distance sampling: estimating abundance of biological populations. Oxford University Press, New York, New York, USA. Buckland, S. T., D. R. Anderson, K. P. Burnham, J. L. Laake, D. L. Borchers, and L. Thomas, editors. 2004. Advanced distance sampling: estimating abundance of biological populations. Oxford University Press, New York, New York, USA. Cheal, P. D., J. C. Day, and C. W. Meredith. 1979. Fire in the national parks of north-west Victoria. National Parks Service, Melbourne, Australia. Chisholm, A. H. 1922. The ‘‘lost’’ Paradise Parrot. Emu 22:4– 17. Clarke, M. F., et al. 2010. Ageing mallee eucalypt vegetation after fire: insights for successional trajectories in semi-arid mallee ecosystems. Australian Journal of Botany 58:363–372. Clarke, R. H. 2005. Ecological requirements of birds specialising in mallee habitats. Ornithology and Avian Conservation Laboratory, La Trobe University, Bundoora, Melbourne, Australia. Clarke, R. H., R. L. Boulton, and M. F. Clarke. 2005. Estimating population size of the Black-eared Miner, with an assessment of landscape-scale habitat requirements. Pacific Conservation Biology 11:174–188. Covert-Bratland, K. A., W. M. Block, and T. C. Theimer. 2006. Hairy woodpecker winter ecology in ponderosa pine forests representing different ages since wildfire. Journal of Wildlife Management 70:1379–1392. Driscoll, D. A., et al. 2010. Fire management for biodiversity conservation: key research questions and our capacity to answer them. Biological Conservation 143:1928–1939. Elith, J., J. R. Leathwick, and T. Hastie. 2008. A working guide to boosted regression trees. Journal of Animal Ecology 77:802–813. Fox, B. J. 1982. Fire and mammalian secondary succession in an Australian coastal heath. Ecology 63:1332–1341. Friend, G. R. 1993. Impact of fire on small vertebrates in mallee woodlands and heathlands of temperate Australia: a review. Biological Conservation 65:99–114. Fuhlendorf, S. D., W. C. Harrell, D. M. Engle, R. G. Hamilton, C. A. Davis, and D. M. Leslie. 2006. Should heterogeneity be the basis for conservation? Grassland bird response to fire and grazing. Ecological Applications 16:1706–1716. Gill, A. M., and G. Allan. 2008. Large fires, fire effects and the fire-regime concept. International Journal of Wildland Fire 17:688–695. Haslem, A., K. E. Callister, S. C. Avitabile, P. A. Griffioen, L. T. Kelly, D. G. Nimmo, L. M. Spence-Bailey, R. S. Taylor, S. J. Watson, L. Brown, A. F. Bennett, and M. F. Clarke. 2010. A framework for mapping vegetation over broad spatial extents: a technique to aid land management 695 across jurisdictional boundaries. Landscape and Urban Planning 97:296–305. Haslem, A., L. T. Kelly, D. G. Nimmo, S. J. Watson, S. A. Kenny, R. S. Taylor, S. C. Avitabile, K. E. Callister, L. M. Spence-Bailey, M. F. Clarke, and A. F. Bennett. 2011. Habitat or fuel? Implications of long-term, post-fire dynamics for the development of key resources for fauna and fire. Journal of Applied Ecology 48:247–256. Higgins, P. J., and J. M. Peter, editors. 2002. Handbook of Australian, New Zealand and Antarctic Birds. Volume 6: Pardalotes to Shrike-thrushes. Oxford University Press, Melbourne, Victoria, Australia. Higgins, P. J., J. M. Peter, and W. K. Steele, editors. 2001. Handbook of Australian, New Zealand and Antarctic Birds. Volume 5: Tyrant-flycatchers to Chats. Oxford University Press, Melbourne, Victoria, Australia. Hutto, R. L. 1995. Composition of bird communities following stand-replacement fires in northern Rocky Mountain (U.S.A.) conifer forests. Conservation Biology 9:1041–1058. Hutto, R. L. 2008. The ecological importance of severe wildfires: some like it hot. Ecological Applications 18:1827– 1834. Jacquet, K., and R. Prodon. 2009. Measuring the postfire resilience of a bird–vegetation system: a 28-year study in a Mediterranean oak woodland. Oecologia 161:801–811. Johnson, E. A., and K. Miyanishi. 2008. Testing the assumptions of chronosequences in succession. Ecology Letters 11:419–431. Kavanagh, R. P., R. H. Loyn, G. C. Smith, R. J. Taylor, and P. C. Catling. 2004. Which species should be monitored to indicate ecological sustainability in Australian forest management. Pages 959–987 in D. Lunney, editor. Conservation of Australia’s forest fauna. Second edition. Royal Zoological Society of New South Wales, Mosman, New South Wales, Australia. Keith, D. A., J. E. Williams, and J. C. Z. Woinarski. 2001. Fire management and biodiversity conservation: key approaches and principles. Pages 401–425 in R. A. Bradstock, J. E. Williams, and A. M. Gill, editors. Flammable Australia: the fire regimes and biodiversity of a continent. Cambridge University Press, Cambridge, UK. Kelly, L. T., D. G. Nimmo, L. M. Spence-Bailey, A. Haslem, S. J. Watson, M. F. Clarke, and A. F. Bennett. 2011. Influence of fire history on small mammal distributions: insights from a 100-year post-fire chronosequence. Diversity and Distributions 17:462–473. Letnic, M. 2003. The effects of experimental patch burning and rainfall on small mammals in the Simpson Desert, Queensland. Wildlife Research 30:547–563. Lindenmayer, D. B., C. MacGregor, J. T. Wood, R. B. Cunningham, M. Crane, D. Michael, R. Montague-Drake, D. Brown, M. Fortescue, N. Dexter, M. Hudson, and A. M. Gill. 2009. What factors influence rapid post-fire site reoccupancy? A case study of the endangered Eastern Bristlebird in eastern Australia. International Journal of Wildland Fire 18:84–95. Marques, T. A., L. Thomas, S. G. Fancy, and S. T. Buckland. 2007. Improving estimates of bird density using multiplecovariate distance sampling. Auk 124:1229–1243. Monamy, V., and B. J. Fox. 2000. Small mammal succession is determined by vegetation density rather than time elapsed since disturbance. Austral Ecology 25:580–587. Moretti, M., and C. Legg. 2009. Combining plant and animal traits to assess community functional responses to disturbance. Ecography 32:299–309. Murphy, E. C., and W. A. Lehnhausen. 1998. Density and foraging ecology of woodpeckers following a stand-replacement fire. Journal of Wildlife Management 62:1359–1372. Nimmo, D. G., L. T. Kelly, L. M. Spence-Bailey, S. J. Watson, A. Haslem, J. White, M. F. Clarke, and A. F. Bennett. In 696 SIMON J. WATSON ET AL. press. Predicting the century-long, post-fire responses of reptiles. Global Ecology and Biogeography. Noble, J. C., and R. G. Vines. 1993. Fire studies in mallee (Eucalyptus spp.) communities of western New South Wales: grass fuel dynamics and associated weather patterns. Rangeland Journal 15:270–297. Parr, C. L., and A. N. Andersen. 2006. Patch mosaic burning for biodiversity conservation: a critique of the pyrodiversity paradigm. Conservation Biology 20:1610–1619. Parsons, R. F. 1968. An introduction to the regeneration of mallee eucalypts. Proceedings of the Royal Society of Victoria 81:59–68. Pausas, J. G., and R. A. Bradstock. 2007. Fire persistence traits of plants along a productivity and disturbance gradient in Mediterranean shrublands of south-east Australia. Global Ecology and Biogeography 16:330–340. Pearce, J., and S. Ferrier. 2000. Evaluating the predictive performance of habitat models developed using logistic regression. Ecological Modelling 133:225–245. Pons, P., and M. Clavero. 2010. Bird responses to fire severity and time since fire in managed mountain rangelands. Animal Conservation 13:294–305. R Development Core Team. 2009. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. http://www.r-project.org Richards, S. A., H. P. Possingham, and J. Tizard. 1999. Optimal fire management for maintaining community diversity. Ecological Applications 9:880–892. Sara, M., E. Bellia, and A. Milazzo. 2006. Fire disturbance disrupts co-occurrence patterns of terrestrial vertebrates in Mediterranean woodlands. Journal of Biogeography 33:843– 852. Skowno, A. L., and W. J. Bond. 2003. Bird community composition in an actively managed savanna reserve: Ecological Applications Vol. 22, No. 2 importance of vegetation structure and vegetation composition. Biodiversity and Conservation 12:2279–2294. Slik, J. W. F., and S. V. Balen. 2006. Bird community changes in response to single and repeated fires in a lowland tropical rainforest of eastern Borneo. Biodiversity and Conservation 15:4425–4451. Thomas, L., J. L. Laake, S. Strindberg, F. F. C. Marques, S. T. Buckland, D. L. Borchers, D. R. Anderson, K. P. Burnham, S. L. Hedley, J. H. Pollard, J. R. B. Bishop, and T. A. Marques. 2006. Distance 5.0 release 2. Research unit for wildlife population assessment, University of St. Andrews, UK. http://www.ruwpa.st-and.ac.uk/distance/ Whelan, R. J., L. Rodgerson, C. R. Dickman, and E. F. Sutherland. 2001. Critical life cycles of plants and animals: developing a process-based understanding of population changes in fire prone landscapes. Pages 94 –124 in R. Bradstock, J. E. Williams, and A. M. Gill, editors. Flammable Australia: the fire regimes and biodiversity of a continent. Cambridge University Press, Cambridge, UK. Woinarski, J. C. 1999. Fire and Australian birds: a review. Pages 56–111 in A. M. Gill, J. C. Z. Woinarski, and A. York, editors. Australia’s biodiversity: responses to fire: plants, birds and invertebrates. Biodiversity Technical Paper. No 1. Environment Australia, Canberra, Australia. Wood, S. N. 2004. Stable and efficient multiple smoothing parameter estimation for generalized additive models. Journal of the American Statistical Association 99:673–686. Wood, S. N. 2006. Generalized additive models. An introduction with R. Chapman and Hall, Boca Raton, Florida, USA. Wood, S. N. 2008. Fast stable direct fitting and smoothness selection for generalized additive models. Journal of the Royal Statistical Society Series B 70:495–518. Zuur, A. F., E. N. Ieno, N. J. Walker, A. A. Saveliev, and G. M. Smith. 2009. Mixed effects models and extensions in ecology with R. Springer, New York, New York, USA. SUPPLEMENTAL MATERIAL Appendix A The distribution of survey sites in relation to time since fire in Triodia Mallee and Chenopod Mallee vegetation (Ecological Archives A022-038-A1). Appendix B Relationships of the frequency of occurrence of species with the parametric variables vegetation type and northing, and tests of significance for these variables obtained from the full species models (GAMMs) (Ecological Archives A022-038-A2).