Long-term post-fire succession of reptiles in an urban remnant in south-western Australia

Introduction

Many reptile species have limited dispersal capabilities (Munguia-Vega et al. 2013; François et al. 2021; Mulhall et al. 2022), which when combined with their often small home ranges and sensitivity to changes in microhabitats and vegetation structure (Craig et al. 2014; Bradley et al. 2022), can render them highly sensitive to disturbance events such as fire (Nimmo et al. 2013; Smith et al. 2013). As a result, fire has been shown to be a significant factor influencing the composition of reptile communities (Clarke et al. 2021; Partridge et al. 2023). Several species have been shown to be most abundant in long unburnt vegetation (Nimmo et al. 2012; Doherty et al. 2015) and recolonisation of burnt areas can be a relatively slow process leading to a source-sink population dynamic between habitat patches of differing post-fire ages (Driscoll 2004; Driscoll et al. 2012; Simms et al. 2019). Hence, understanding long-term successional responses of reptiles to fire is critical for their effective conservation and management (Driscoll et al. 2010b; Clarke et al. 2021).

The conservation and management of reptiles in urban remnants is challenging with reptile communities in even large remnants being relatively depauperate compared to contiguous habitats (How and Dell 2000; Keinath et al. 2017). Adding to the challenge is the fact that reptiles in urban remnants may display different post-fire responses to populations in contiguous habitat, due to the disruption of source-sink dynamics (Driscoll et al. 2010a, 2012). In addition, remnants are often subject to altered fire regimes due to increased fire frequency due to arson (Stenhouse 2001; Hero et al. 2013) and increased fire frequency and severity due to invasion by fire-promoting weeds (Crosti et al. 2007; Fisher et al. 2009; Ramalho et al. 2014). These altered fire regimes combine with other disturbances, such as edge effects (Urbina-Cardona et al. 2006; Garden et al. 2007; McAlpine et al. 2015), to further alter vegetation structure (Ibanez et al. 2017; Gomes et al. 2022). In combination, these pressures, along with a changing climate (How et al. 2022), have the potential to strongly influence post-fire responses of reptiles in urban remnants. However, whether post-fire responses differ between remnants and contiguous vegetation remains poorly understood and the ecological mechanisms behind any potential differences, even less well understood.

Banksia woodlands are a threatened ecological community confined to sandy soils on the Swan Coastal Plain of south-western Australia where the canopy is dominated by species of Banksia (Ritchie et al. 2021). Banksia woodlands support a very rich herpetofauna of over 60 species and woodlands in and close to the major city of Perth have been extensively cleared and fragmented. Furthermore, fire is a natural disturbance in Banksia woodlands although the frequency of fire has increased over recent decades, particularly in urban remnants, so understanding post-fire responses in remnants is critical for the conservation of these reptile communities in remnants. While post-fire responses of reptiles in contiguous Banksia woodlands have been previously studied (Bamford 1995; Valentine et al. 2012), responses in remnants remain less well understood (see Davis and Doherty 2015). This study intended to address this knowledge gap by continuing the study of Davis and Doherty (2015) to explore longer-term (≤15 years) post-fire responses of reptiles in Kings Park, a large Banksia woodland remnant located within the Perth Metropolitan Area and compare the results with studies from contiguous Banksia woodlands. The questions asked were: (1) does the reptile community change over 15 years post-fire; (2) does the abundance of individual species change over 15 years post-fire; and (3) how do any patterns compare to post-fire responses in contiguous Banksia woodlands.

Materials and methods

Results

Effects of fire

Community composition did not differ between unburnt and burnt sites (Pseudo-F_1,96 = 1.58, P = 0.223) but there was a significant treatment by year interaction (Pseudo-F_12,96 = 2.08, P = 0.001) with composition in the first 2 years post-fire being significantly different from sites with the longest time since fire and unburnt sites (Fig. 1). Overall reptile relative abundance showed no difference between unburnt and burnt sites (t₈ = 1.78, P = 0.113) and there was no significant treatment by time interaction (t₁₁₈ = −1.78, P = 0.078), although overall reptile relative abundance was lower in burnt sites in the first 2 years post-fire. Reptile species richness did show a difference between unburnt and burnt sites (t₈ = 3.15, P = 0.014) but this was driven by a significant treatment by time interaction (t₁₁₈ = −3.14, P = 0.002) with species richness lower in the first 3 years post-fire but the difference disappearing after that time (Fig. 2).

Fig. 1.

Results of the Principle Coordinates Analysis of reptile community composition in the first 15 years post-fire showing both individual sites (top) and the mean (±s.e.) of unburnt and burnt sites for each year (bottom). Unburnt sites are shown as circles and burnt sites as triangles with the colour of the shapes becoming progressively darker with increasing post-fire age. Note that community composition differs greatly between unburnt and burnt sites in the first 2 years post-fire but the difference disappears after this time.

Fig. 2.

Overall reptile abundance and species richness for unburnt (solid lines) and burnt (broken lines) sites for the first 15 years post-fire. All values are shown as mean ± s.e.

Among individual species, H. quadrilineata relative abundance showed a significant difference between unburnt and burnt sites (t₈ = 2.66, P = 0.029) but this was due to a significant treatment by time interaction (t₁₁₈ = −2.65, P = 0.009) with relative abundance being generally lower in the first 5 years post-fire, except for the first year (Fig. 3). L. praepedita also differed in relative abundance between unburnt and burnt sites (t₈ = 2.62, P = 0.031) and this was driven by a significant treatment by time interaction (t₁₁₈ = −2.61, P = 0.010) with relative abundance lower in the first 3 years post-fire but the difference disappearing after that time (Fig. 3). L. lineopunctulata relative abundance did not differ significantly between unburnt and burnt sites (t₈ = 2.04, P = 0.075) but there was a significant treatment by time interaction (t₁₈ = −2.05, P = 0.043) with relative abundance being higher in burnt sites more than 8 years post-fire. C. buchananii relative abundance also did not differ significantly between unburnt and burnt sites (t₈ = −2.23, P = 0.057) but there was a significant treatment by time interaction (t₁₁₈ = 2.22, P = 0.028) with relative abundance being higher in burnt sites in the first 12 months post-fire (Fig. 3). No other species showed any difference in relative abundance between unburnt and burnt sites nor a significant treatment by time interaction (Fig. 4, Table S2).

Fig. 3.

Mean relative abundance (±s.e.) in unburnt (solid lines) and burnt (broken lines) sites for the first 15 years post-fire for the four species that showed a significant response to fire.

Fig. 4.

Mean relative abundance (±s.e.) in unburnt (solid lines) and burnt (broken lines) sites for the first 15 years post-fire for the 11 species that showed no significant response to fire.

Discussion

Over the first 15-years post-fire, we found that most post-fire responses of reptiles were relatively short-lived, with no noticeable negative effects of fire on the reptile community for more than 5 years post-fire. Community metrics showed no noticeable response to fire after more than 3 years post-fire and only four species showed significant responses to fire. L. praepedita showed a reduced abundance for 3 years post-fire while H. quadrilineata was the species that showed the longest negative response to fire, with relative abundance reaching pre-fire levels 6 years post-fire. The positive responses to fire were that C. buchananii was more abundant in burnt areas in the first year post-fire and the relative abundance of L. lineopunctulata increased above pre-fire levels after 8 years post-fire. There was also no evidence of negative fire effects on rare species (≤6 individuals captured) that we did not analyse statistically. Furthermore, three of those rare species (Varanus tristis, C. australis, Psuedonaja affinis) have been recorded in a wide range of fire ages in other studies (Bamford 1995; Valentine et al. 2009) and persist in numerous urban remnants (How and Dell 1994, 2000) so they do not appear to be negatively affected by fire. Furthermore, P. affinis is often more abundant in disturbed habitats and feeds extensively on Mus musculus (Maryan and Bush 1996), which often responds positively to fire (e.g. Recher et al. 2009; Kelly et al. 2011). Therefore, the patterns we observed in these rare species are almost certainly due to stochasticity rather than reflecting an actual fire response. As expected, our results are very similar to the earlier study (Davis and Doherty 2015). The only additional information was that our longer-term study identified the post-fire age when H. quadrilineata recovered its post-fire abundance, that Lerista lineopunctulata abundance increased in burnt area in the longer term, and increased our confidence that rare species are not negatively affected by fire.

In general, our study mirrored responses from two other studies conducted in contiguous Banksia woodlands (Bamford 1995; Valentine et al. 2012). Neither of those studies found that overall reptile abundance or species richness showed a significant response to fire; however, Valentine et al. (2012) only examined sites 4 years post-fire or older so their difference from our study was likely due to their different study design. The fact that we recorded initial decreases in overall abundance and species richness could indicate that several, less common species that are resilient to fire are absent from our remnant or that there are fewer source populations to recolonise after fire. Individual species’ responses were also generally similar. P. minor, C. buchananii, Ctenotus fallens and Lerista elegans all showed no response to fire in either study (Bamford 1995; Valentine et al. 2012). One of these species, C. fallens, potentially disperses widely between bushland remnants as evidenced by a low F_st in a previous population genetics study that included Kings Park and this may enable it to recolonise after fire (Virens et al. 2015). In the case of C. buchananii, the lack of observed response was probably because its abundance was mainly related to canopy cover, which regrew rapidly post-fire (Valentine et al. 2012). Our study only found a response to fire by C. buchananii in the first year post-fire and the other studies only looked at sites 2 years or more post-fire so, again, this difference was due to study design. Hemiergis initialis abundance was lower post-fire in contiguous woodlands (Valentine et al. 2012), similar to this study, which was likely due to reduced litter cover in the first few years post-fire (Valentine et al. 2012). This species was not captured by Bamford (1995). Three species’ responses showed small variations between studies. We observed the abundance of L. praepedita was lower in the first 3 years post-fire whereas both studies in contiguous woodlands found no relationship with time since fire (Bamford 1995; Valentine et al. 2012). However, Valentine et al. (2012) only examined sites more than 4 years post-fire, so any difference was most likely due to the different study design, and the species was rarely captured by Bamford (1995). M. greyii and M. obscura were both found to increase in abundance with increasing post-fire age by Valentine et al. (2012) but both Bamford (1995) and this study found no relationship with fire. The relationship for M. obscura in Valentine et al. (2012), however, was driven by a high abundance of this species in sites more than 20 years post-fire, which is older than any burnt sites in this study. It is unclear whether that was also the case for M. greyii but Valentine et al. (2012) commented that this species’ relationship with post-fire age was weak.

Overall, all three studies exploring reptile responses to fire in Banksia woodlands found similar patterns. All differences between species were minor and inconsistent between remnant and contiguous woodlands with many differences appearing to result from different study designs. All three studies found that reptile recovery post-fire was relatively rapid and, although this recovery extended over decades for a few species, these species recovered rapidly and did not appear to be highly vulnerable to short inter-fire intervals. We also found no evidence that individual species’ responses differed consistently between remnants and contiguous woodlands, at least over the post-fire period explored in this study (i.e. <15 years). This suggests that differences in habitat structure (e.g. increased weed cover) between remnants and contiguous habitat (Ramalho et al. 2014) did not result in significantly different post-fire responses. There was limited evidence that the reptile community in remnants displayed short-term (≤3 years) negative responses that were not evident in contiguous woodlands but again, differences in study design render this conclusion speculative. Given that remnants display lower species richness, losing primarily rare and specialised species, it is possible that reptile communities in remnants are less resilient to disturbances like fire, than those in contiguous habitats, but this requires further research. The impacts of climate change on reptiles and interactions with fire, are poorly understood and require future study. How et al. (2022) analysed a 35-year dataset of reptiles in Banksia woodlands near to our study site and found that abundance was significantly correlated with preceding rainfall. They found only limited community compositional change due to annual rainfall variation.

In conclusion, our study found that post-fire responses in a Banksia woodland remnant were generally fairly minor with the few negative responses lasting no more than 5 years and a short-term (1 year) and longer-term (>8 years) positive effect. This suggests the reptile community should be fairly resilient to a wide range of fire regimes including those with short inter-fire intervals down to 3 years. The fact that reptiles occurred, in reduced numbers for two species, in areas immediately post-fire indicated that population connectivity will be maintained even under short-fire intervals. We found no consistent differences in post-fire responses between remnants and contiguous woodlands, although our results suggested the overall reptile community may show a greater negative response to fire in the first 3 years post-fire. Given fragmentation of many habitats is increasing, we encourage future fire research to explored whether differences in post-fire responses exist between contiguous and fragmented habitats. Understanding whether differences exist is likely to be important for the effective conservation of reptile populations in remnants, particularly as climate change is likely to change fire regimes to more frequent, intense fires in many ecosystems. Our study also further emphasises the importance of long-term monitoring and time series data such as this study (Dornelas et al. 2018) in addressing ecological and management questions.

Supplementary material

Supplementary material is available online.