Introduction
An essential first step in understanding the level of threat to a species is to determine its response to disturbance (Shine et al., Reference Shine, Webb, Fitzgerald and Sumner1998; Pike et al., Reference Pike, Croak, Webb and Shine2010; Böhm et al., Reference Böhm, Collen, Baillie, Bowles, Chanson and Cox2013). Habitat loss and degradation are key causes of extinction of threatened wildlife worldwide (Gibbons et al., Reference Gibbons, Scott, Ryan, Buhlmann, Tuberville and Mets2000; Hibbitts et al., Reference Hibbitts, Ryberg, Adams, Fields, Lay and Young2013). Research into the impacts of disturbance on herpetofauna has focused on the global amphibian decline (e.g. Collins & Storfer, Reference Collins and Storfer2003). A similar global decline may also be underway in reptiles but has attracted less research (Gibbons et al., Reference Gibbons, Scott, Ryan, Buhlmann, Tuberville and Mets2000; Böhm et al., Reference Böhm, Collen, Baillie, Bowles, Chanson and Cox2013), although localized declines and extinctions have been well documented (Cogger et al., Reference Cogger, Cameron, Sadlier and Eggler1993; Hecnar & M'Closkey, Reference Hecnar and M'Closkey1998; Shine et al., Reference Shine, Webb, Fitzgerald and Sumner1998; Reading et al., Reference Reading, Luiselli, Akani, Bonnet, Amori and Ballouard2010; Sinervo et al., Reference Sinervo, Méndez-de-la-Cruz, Miles, Heulin, Bastiaans and Villagrán Santa Cruz2010). Specialist, fragile or low-mobility species may be at greatest risk when environments change (Steffen et al., Reference Steffen, Burbidge, Hughes, Kitching, Lindenmayer and Musgrave2009). One such imperilled species is the Endangered Blue Mountains water skink Eulamprus leuraensis (ARASG, 1996).
This iconic reptile is known only from < 60 isolated sites within the montane regions of south-eastern Australia (Blue Mountains and Newnes Plateau; Gorissen, Reference Gorissen2016). It is endemic to a unique peat-swamp habitat, Temperate Highland Peat Swamps on Sandstone, which is also rare (c. 4,000 ha in extent; Hensen & Mahony, Reference Hensen and Mahony2010) and federally listed as Endangered (TSSC, 2005). Genetic studies show low rates of lizard dispersal, and thus gene flow, between these swamps (Dubey & Shine, Reference Dubey and Shine2010), and even less genetic connectivity between the two main regions where such swamps occur (i.e. Blue Mountains water skinks on the Newnes Plateau are genetically distinct from those in the Blue Mountains; Dubey & Shine, Reference Dubey and Shine2010). Climate change is expected to cause hotter and drier weather conditions in the area, reducing water availability (CSIRO & BOM, 2007; IPCC, Reference Stocker, Qin, Plattner, Tignor, Allen and Boschung2013). Longwall mining is an immediate and landscape-scale threat, given its detrimental and often severe impacts on swamps in this region (Aurecon, 2009; Goldney et al., Reference Goldney, Mactaggart and Merrick2010; Enforceable Undertaking, 2011) via groundwater loss through subsidence (and chemical pollution from mine-water discharge), and is listed as a Key Threatening Process in protective legislation (NSW Scientific Committee, 2005). Development (industrial, urban) and damage by vehicles are other threatening processes affecting swamps of the region (Hensen & Mahony, Reference Hensen and Mahony2010; Fryirs et al., Reference Fryirs, Freidman and Kohlhagen2012; Belmer et al., Reference Belmer, Wright and Tippler2015) and may potentially degrade habitat quality for this Endangered reptile. We surveyed swamps exhibiting habitat degradation (and thus altered hydrological regimes), to clarify the impact of hydrological disturbance on E. leuraensis and its swamp habitat.
Study area
The Newnes Plateau and Blue Mountains areas of south-eastern Australia, c. 100 km north-west of Sydney (Fig. 1), have a temperate climate, with mean monthly temperature of 6.3–23.1°C (BOM, 2015), mean annual rainfall of 464–1,450 mm (Keith & Benson, Reference Keith and Benson1988; Whinam & Chilcott, Reference Whinam and Chilcott2002; DEC, 2006), and an underlying sandstone geology (Keith & Benson, Reference Keith and Benson1988).
Locations of the swamp sites surveyed in the Blue Mountains and Newnes Plateau areas of south-eastern Australia (NPWS, 2001; Table 1). BS, MS and XFC1 have both undisturbed and disturbed areas. (Adapted from Gorissen et al., Reference Gorissen, Mallinson, Greenlees and Shine2015)
Methods
Study sites
The swamps sampled included Blue Mountains sedge swamps and Newnes Plateau shrub swamps (Keith & Benson, Reference Keith and Benson1988; Benson & Baird, Reference Benson and Baird2012). Islands within a matrix of sclerophyll woodland and open forest, these swamps are dominated by sedge, shrub and grass vegetation growing on peaty soils (Keith & Benson, Reference Keith and Benson1988; TSSC, 2005; Benson & Baird, Reference Benson and Baird2012). The swamps contain one or more drainage lines, and many are elongate in shape (Benson & Baird, Reference Benson and Baird2012). Nine such swamps were selected for surveys (Fig. 1, Table 1), c. 3,500–120,000 m2 in extent, with mean elevations of 680–1,170 m. Three swamps were pristine at one end but disturbed at the other (BS, MS, XFC1; Plate 1a), providing both control and treatment sites. We therefore sampled 12 sites in total (selected based on their level of disturbance), comprising six pairs of undisturbed and disturbed sites, paired by proximity and availability. All disturbed sites exhibited hydrological degradation, in terms of ground- and/or surface-water quantity (loss through subsidence from longwall mining practices at EWS, (Plate 1b) and JS (Plate 1c,d)), damage by recreational vehicles (XFC1 (Disturbed); Hensen, Reference Hensen2010) or development of infrastructure (MS (Disturbed); Fryirs et al., Reference Fryirs, Freidman and Kohlhagen2012). Surface-water quality was also affected (physical and/or chemical properties) by pollution and/or sedimentation as a result of industrial (BS (Disturbed); N. Belmer et al., unpubl. data) or urban development (BRS; Belmer et al., Reference Belmer, Wright and Tippler2015). Eight of the study sites were on the Newnes Plateau, mainly within State Forests, at c. 1,100 m elevation, and four were in the Blue Mountains, on land managed by the Blue Mountains City Council, at c. 850 m elevation.
Swamp sites on the Newnes Plateau. (a) Pristine, undisturbed bush swamp (XFC1) with a typical dense and live vegetation structure; (b) a disturbed swamp (EWS) with severe groundwater disturbance as a result of subsidence (and chemical pollution) from longwall mining practices; (c & d) the site JS before (2001) and after (2014) the swamp was undermined (2003–2004).
Features of the six paired swamp sites in the Blue Mountains and Newnes Plateau areas of south-eastern Australia (Fig. 1), with swamp type, disturbance, hydrological impact, year of disturbance, elevation and area (Aurecon, 2009; Goldney et al., Reference Goldney, Mactaggart and Merrick2010; Belmer et al., Reference Belmer, Wright and Tippler2015; M. Hensen, pers. comm., 2015).
These 12 swamp sites span the entire known distributional range of the Blue Mountains water skink, which is the sole endemic vertebrate of the region (Fig. 1). Prior to surveying, pilot trapping established presence of the lizard in all undisturbed swamps. Of the three disturbed sites, the species was recorded prior to disturbance at JS and BRS (Fig. 1; NSW OEH, 2015) but not at EWS (Fig. 1). Although we lack a definite record of the species at the latter site, the area has remnant vegetation and some residual peat, typical of natural swamps, and shares common tributaries with neighbouring swamps that contain the species (SS, Fig. 1) or that have records of it (JS, Fig. 1). We inferred the historical presence of E. leuraensis in the disturbed sections of the remaining three swamps (BS, MS, XFC1), each of which contain both a pristine and a disturbed area, based on presence of the lizards in the pristine area of each swamp.
Focal species
Known from < 60 populations, E. leuraensis is restricted in distribution (Gorissen, Reference Gorissen2016) and listed as Endangered under both state and federal legislation (NPWS, 2001). Surveys suggest that the species is a swamp specialist (> 95% of occurrence records from swamps, and none from an equivalent trapping effort in surrounding woodland < 20 m from the swamp boundary; Gorissen et al., Reference Gorissen, Mallinson, Greenlees and Shine2015). These scincid lizards are medium-sized (total length to 21.6 cm, c. 14.8 g), viviparous, and active on warm sunny days and during the hotter months (September to April/May). To escape predation E. leuraensis takes shelter in dense sedgeland tussocks or in holes in the peat substrate (Shea & Peterson, Reference Shea and Peterson1985).
Quantifying faunal abundance
We quantified herpetofaunal and invertebrate abundances at paired swamp sites (disturbed and undisturbed) by mark–recapture over 3-day sessions during November 2014–March 2015. We divided the swamp ecosystem into three survey zones: swamp, transition (swamp margin) and woodland (Gorissen et al., Reference Gorissen, Mallinson, Greenlees and Shine2015; Gorissen, Reference Gorissen2016). Trapping was conducted only on days with a maximum temperature of 20–35°C and no rainfall (BOM, 2015). At each zone in each swamp we set 10 traps c. 10 m apart. Pitfall traps (10 L, 27 × 28 cm; without drift fences; 1 per zone) and unbaited funnel traps (18 × 18 × 75 cm; 9 per zone) were used. Traps were checked in the late afternoon. Herpetofauna were identified to species, and invertebrates to order. Lizards were marked individually for later identification. Our index for population size was the number of individuals captured, excluding same-survey recaptures of individuals. Only live invertebrates were included, and we scored presence rather than abundance (n = 1) if the trap also contained material (usually scats) that served as an attractant to insects. In addition to records from ground traps we made some opportunistic sightings.
Quantifying habitat characteristics
SG recorded a suite of habitat characteristics around each trap for all three zones and all 12 swamp sites. Each quadrat was a circular area of 1 m radius, centred on the trap. Volumetric soil moisture content was calculated as the mean of three spatially randomized measurements from an MP406 Soil Moisture Instant Reading Kit (ICT International, Armidale, Australia). We also recorded the distance from surface water (drainage line or permanent pool ≥ 0.5 m diameter); the proportion of substrate covered by live vegetation, dead vegetation, log, surface water, rock, and bare ground (dirt or mud); the proportion of cover/sunlight penetration at the canopy (> 5 m high) and understorey (0.5–5 m high) levels; and the proportion of substrate exposed to direct sunlight at the sun's zenith. Distance measurements were made using a global positioning system, and canopy cover was approximated using a canopy cover estimation chart (Hnatiuk et al., Reference Hnatiuk, Thackway and Walker2009).
Statistical analysis
In all tests we compared disturbed vs undisturbed sites (i.e. a paired design) to control for locally variable factors. We evaluated normality and used paired statistical tests accordingly (a two-tailed t-test or Wilcoxon paired-sample test) to investigate the effects of swamp type (undisturbed vs disturbed; independent variable) on faunal abundances and habitat characteristics (dependent variables; see Table 2 for statistical tests). When differences in distributions were non-normal a standard log10(x + 1) transformation was applied (Zar, Reference Zar1999). We used total abundance per swamp site as the unit for analyses (Table 2). Non-swamp lizards are taxa that (based on extensive surveys; Gorissen, Reference Gorissen2016) usually inhabit the drier surrounding habitats (woodland and transitional zones) rather than the swamp itself (Gorissen, Reference Gorissen2016). Our analyses of invertebrates included only species that are likely to be the prey of E. leuraensis (Veron, Reference Veron1969; Brown, Reference Brown1991; LeBreton, Reference LeBreton1992, Reference LeBreton1996).
Descriptive statistics for paired test results (two-tailed t-test; Wilcoxon paired-sample test) comparing undisturbed (U) vs disturbed (D) swamp types (for swamp habitat only, unless otherwise stated) in terms of faunal and habitat characteristics.
*Statistically significant at P < 0.05
Results
Impacts on threatened and other fauna
Eulamprus leuraensis occurred in all six undisturbed swamps but we did not record any in the six disturbed sites during our surveys (Fig. 2a, Table 2). Consistently, E. leuraensis was found in swamp habitat (> 95% of records) and was the most abundant herpetofaunal species in this habitat within undisturbed swamps (E. leuraensis, n = 41; other herpetofauna, n = 4). We trapped more non-swamp lizards in the swamp habitat in disturbed sites than in undisturbed sites (Table 2), notably the skinks Acritoscincus platynota (n = 4), Eulamprus heatwolei (n = 3) and Lampropholis guichenoti (n = 4). In undisturbed swamps, in contrast, E. heatwolei was trapped only in the swamp margins. The numbers of original vs regrown tails were similar in lizards from both swamp types (χ2 1 = 0.34, P = 0.56).
Attributes of undisturbed vs disturbed swamp sites surveyed in the Blue Mountains and Newnes Plateau areas of south-eastern Australia (Fig. 1). (a) Mean abundance of the Blue Mountains water skink Eulamprus leuraensis; (b) substrate composition (% of ground covered by live vegetation, dead vegetation, log, water, rock and bare ground) of swamp habitat; and (c) vegetation cover (canopy and understorey) and sunlight penetration to ground level of swamp habitat.
Overall, 12 species of herpetofauna were captured in undisturbed swamps, and nine in disturbed swamps, with an equal species richness in swamp habitat in both disturbed and undisturbed sites (five species). The total abundance of herpetofauna was similar for both swamp types (Table 2).
Within the swamp habitat, invertebrates were equally abundant in disturbed and undisturbed sites (Table 2). Arthropods were the most common invertebrates. Notable species in the swamp included the burrow-forming Sydney crayfish Euastacus australasiensis and the burrow-forming and Endangered giant dragonfly Petalura gigantea, both of which were found in undisturbed but not disturbed swamp habitat.
Habitat degradation
Predictably, soil moisture declined with distance from water (n = 358, r 2 = 0.16, P < 0.0001). Disturbed swamps were further from water and had drier soil, with volumetric soil moisture content approximately half that of undisturbed swamps (Table 2). Substrate composition also changed with degradation, from a swamp habitat with abundant live vegetation and surface water (with almost no rocks, logs or bare ground) to one with more bare ground, rocks and dead vegetation (including logs; Fig. 2b, Table 2). Disturbance was also associated with a reduced understorey density and increased sunlight penetration to ground level within the swamp habitat (Fig. 2c, Table 2). Transition and woodland habitat characteristics were relatively unaffected by disturbance to the swamp (S. Gorissen et al., unpubl. data). Although the direction of these trends is predictable, the shift in habitat characteristics (especially water availability) following disturbance is striking.
Discussion
Our surveys reveal a stark impact of anthropogenic disturbance on the distribution and abundance of the Endangered Blue Mountains water skink. Although found in all of the undisturbed swamps that we surveyed, these lizards were not caught in any of the disturbed swamps. Earlier surveys and records suggest that E. leuraensis occurred previously in at least five of the six disturbed swamps, and probably in all six. The loss of this iconic taxon is consistent with the species' restriction to densely vegetated and wet swamp habitat (Gorissen et al., Reference Gorissen, Mallinson, Greenlees and Shine2015; this study), its lower abundance in urban than in bush swamps (Gorissen et al., Reference Gorissen, Mallinson, Greenlees and Shine2015), and the changes to its habitat that follow anthropogenic disturbance (this study). Most obviously, the disturbed swamps were drier than pristine swamps, resembling habitats that previously surrounded the swamp itself (Gorissen, Reference Gorissen2016). In keeping with that change, disturbed swamps contained lizard species typical of these transitional and woodland habitats, rather than the swamp-specialist E. leuraensis (Gorissen, Reference Gorissen2016).
Our sample size was limited by the small geographical range of the focal species, the limited availability of disturbed swamp habitat, and a paucity of pre-disturbance survey data for most swamps (precluding a BACI (Before–After Control-Impact) experimental design; Quinn & Keough, Reference Quinn and Keough2002). Consequently, three of our sampling sites included both undisturbed and disturbed sub-sites, which potentially were not truly independent. However, these sub-sites were distinct both spatially and in terms of habitat characteristics. If we had failed to find differences between undisturbed and disturbed areas so close together one could argue that migration of lizards, for example, had obscured a pattern that would otherwise have been evident. However, the fact that we detected clear differences in lizard abundance despite the close proximity of the undisturbed and disturbed swamp areas reinforces our conclusions. Four of the six disturbed swamps had lost much of their water, and therefore the simultaneous loss of a water skink is unsurprising. Nonetheless, the congeneric water skink E. heatwolei, whose distribution is less closely tied to waterlogged habitats (Gorissen, Reference Gorissen2016), not only persisted around these swamps but moved into the swamps vacated by E. leuraensis. Manipulative experiments would be needed to explore whether that habitat expansion was facilitated simply by the habitat shift, or by release from interspecific agonistic interactions (Done & Heatwole, Reference Done and Heatwole1977).
The absence of E. leuraensis from disturbed swamps may reflect a physiological dependence on moist conditions (e.g. because of high rates of evaporative water loss; Heatwole & Veron, Reference Heatwole and Veron1977; Greer, Reference Greer1989; Neilson, Reference Neilson2002) or an ecological dependence on specific habitat attributes that persist only where soil moisture levels are high. Again, manipulations could decipher these possibilities. The loss of E. leuraensis from two swamps where water quantity was unaffected but water quality was compromised suggests that habitat features other than simply moisture levels may be important for this Endangered species. In keeping with this interpretation, many aspects of vegetation cover shifted significantly in our disturbed sites (especially when water was removed from the system). However, patches of relatively dense cover remained in some disturbed sites, and the frequency of caudal autotomy—often used as an index of predation risk (Smith & Ballinger, Reference Smith and Ballinger2001) and/or of the frequency of failed predation attempts (Schwarzkopf & Shine, Reference Schwarzkopf and Shine1992)—did not differ between lizards in disturbed vs undisturbed swamps. Similarly, the abundance of invertebrate prey was not modified substantially by habitat disturbance. The vulnerability of E. leuraensis populations in the disturbed swamps may have been exacerbated by the relatively small spatial extent of most of these sites (Table 1).
The relative similarity in total herpetofaunal abundance between disturbed and undisturbed sites suggests that some species (such as E. leuraensis) are negatively affected by anthropogenic disturbance, whereas others (such as E. heatwolei) benefit. The taxa that thrive tend to be generalist woodland species, none of which are categorized as threatened (possibly reflecting the broad availability of this habitat type). The victims of anthropogenic degradation of swamps are the habitat specialists, at least one of which (E. leuraensis) is of major conservation concern (NPWS, 2001). In essence, swamp degradation reduces the availability of a scarce habitat type, replacing it with an already widespread habitat type. The consequences for faunal conservation are unlikely to be positive.
Translocation is a potential but unexplored management response to disturbance, and may assist in the conservation of E. leuraensis. Some of the swamps in this region are at risk of hydrological disturbance as a result of longwall mining on the Newnes Plateau, a stronghold for this Endangered reptile. We could capture lizards from imminently threatened sites and translocate them to novel (currently lizard-free) swamps to maintain genetic diversity within the species (Dubey & Shine, Reference Dubey and Shine2010). Biological aspects of the species that could be capitalized upon for this process include its low dispersal rate, high habitat specificity and high reproductive rate (compared to congeneric skinks in this montane environment). Analogously, Templeton et al. (Reference Templeton, Brazeal and Neuwald2011) translocated populations of the eastern collared lizard Crotaphytus collaris collaris to combat local extinction without recolonization within a patchy habitat.
In summary, the Blue Mountains water skink is a habitat specialist that is restricted to < 60 groundwater-dependent swamps within an area subject to major anthropogenic pressure (e.g. Wright, Reference Wright2011; Benson & Baird, Reference Benson and Baird2012; Fryirs et al., Reference Fryirs, Freidman and Kohlhagen2012; Belmer et al., Reference Belmer, Tippler, Davies and Wright2014, Reference Belmer, Wright and Tippler2015). Habitat loss or degradation of any of these sites would further threaten this already Endangered species. Especially given its low vagility, E. leuraensis may be vulnerable to habitat change wrought by climate change, urban pollution and longwall mining activity (Dubey & Shine, Reference Dubey and Shine2010). To conserve E. leuraensis in the wild, its rare, fragile and distinctive swamp habitat must be shielded from hydrological disturbance such as groundwater loss and surface water contamination. To conserve the genetic diversity of the species and hence maximize its resilience to any localized adverse impacts we must conserve populations from across its current range (and ideally from as many isolated swamps as is practical). To protect this species we must protect its habitat.
Acknowledgements
We thank our fieldworkers, in particular Stephanie and Peter Bamford, Greg Clarke, Willem Gorissen and Heinz Neubauer. We thank Deborah Ashworth, Ian Baird, Michael Hensen, Chris and Julie Jonkers, Martin Krogh and Keith Muir for discussions. Nakia Belmer provided water chemistry data, Ray Mjadwesch provided photographs, and Melanie Elphick assisted with editing. We also thank the Fenner School of Environment and Society (Australian National University), Blue Mountains City Council, Humane Society International, Ecological Society of Australia, Australian Wildlife Society, University of Sydney and the Australian Research Council for funding. All research was conducted under the University of Sydney Animal Care and Ethics Committee authorization #L04/12-2012/3/5890; National Parks and Wildlife Service Scientific Licence #SL101112; and Forestry Corporation Permits #PB52916 and #PB54029.
Author contributions
SG conceived the study and conducted the research. SG, MG and RS designed the study. SG and RS analysed the data and wrote the article.
Biographical sketches
Sarsha Gorissen primarily researches the conservation of wildlife, in particular, the conservation biology and ecology of threatened species to develop better management guidelines. Matthew Greenlees researches the ecology and evolution of Australian herpetofauna and has a broad interest in natural history, with a particular focus on the impacts of anthropogenic activities and invasive species. Richard Shine's research concerns the interface between evolution and ecology, particularly in reptiles, with recent work focusing on major issues in conservation, especially the biology, impact and control of invasive species such as the cane toad.
