An experiment to test key hypotheses of the drivers of reptile distribution in subalpine ski resorts

Authors


Summary

  1. Alpine and subalpine ecosystems support many endemic species. These ecosystems are increasingly under threat from human-induced disturbances such as habitat loss and fragmentation as a consequence of ski resort development and expansion. However, limited peer-reviewed research has investigated the impacts of ski-related disturbances on wildlife, particularly on reptiles.
  2. To address this knowledge gap, we conducted reptile surveys to determine the patterns of reptile distribution and abundance in Australian ski resorts. Then, using a factorial experimental design, we investigated 1) the influence of temperature and predation in driving observed distributions and 2) how a common ski resort management practice – mowing of modified ski slopes – affected thermal regimes and rates of predation of reptiles on ski runs.
  3. We found that the removal of vegetation structural complexity through mowing resulted in significantly higher rates of predation on plasticine models, as well as significantly altered thermal regimes.
  4. Crucially, mown ski runs had higher maximum ground temperatures that frequently exceeded the recorded critical maximum body temperatures of the target species of lizards. Thus, mowing has the potential to render these areas unsuitable for thermoregulatory purposes for a large proportion of the potential activity period of reptiles.
  5. Together, modifications of the thermal environment and elevated rates of predation appear to explain the avoidance of ski runs by reptiles. To facilitate the persistence of reptiles in disturbed subalpine environments, management plans must focus on implementing strategies that reduce the impact of human activities that alter temperature regimes and predation rates on lizards.
  6. Synthesis and Applications. We suggest that the retention of structural complexity on ski runs (e.g. through the cessation of mowing during peak reptile activity periods) and/or revegetation with native plant communities will concurrently provide refuge from predators and buffer against extreme temperatures, making ski runs more hospitable to reptiles. Based on our findings, we emphasize that effective management strategies targeting subalpine biodiversity conservation require an understanding of the drivers that determine species distributions in these landscapes.

Introduction

Alpine and subalpine ecosystems are significant for biodiversity and support high levels of endemic taxa due to the unique climatic characteristics and geographical isolation (Körner 2004; Nagy & Grabherr 2009). However, the biodiversity found in alpine and subalpine ecosystems is affected globally by major ski developments. There are around 6000 ski resorts and ski areas across the globe, with extensive developments in Europe, Canada, USA and Japan (Vanat 2012). Cumulatively, these ski developments can significantly degrade and fragment alpine and subalpine habitats, threatening the persistence of wildlife in affected environments [WWF (World Wide Fund for Nature) 2005].

The development of ski resorts has a major impact on high mountain environments (WWF 2005), and arguably, the greatest impact is from the construction of groomed ski slopes and ski runs (Wipf et al. 2005; Negro et al. 2010; Caprio et al. 2011). For example, in many North American and Australasian ski areas, native vegetation, rocks, logs and woody debris are removed and soils are extensively compacted during initial construction of groomed ski runs (Ries 1996). The runs may then be seeded with exotic grass species to aid slope stabilization (Tsuyuzaki 1994). During winter, slope grooming is conducted, and mowing is carried out during spring and summer maintenance (one to four times per year; Strong, Dickert & Bell 2002; Kubota & Shimano 2010) to inhibit the establishment of woody vegetation, facilitate early snow coverage and reduce the demand for snowmaking during the winter ski season [PBPL (Perisher Blue Pty Ltd) 2002; Strong, Dickert & Bell 2002]. These activities greatly reduce the structural complexity of the vegetation and ground cover of ski runs and, in turn, affect the distributions and survival of wildlife inhabiting these areas (Laiolo & Rolando 2005; Negro et al. 2009).

Indeed, in a previous study, we found that lizard abundance was greatly reduced on ski runs and this was due – in part – to the removal of rocks, woody debris and tussock grasses (i.e. structural complexity; Chloe F. Sato, unpublished data). However, studies investigating the impacts of ski runs on reptiles are limited (Sato, Wood & Lindenmayer 2013). Thus, the effects of ski run construction and maintenance on underlying ecological drivers important to reptile survival remain unknown.

In other ecosystems, disturbances resulting in the simplification of habitat structural complexity modify underlying biotic processes (e.g. rates of predation and availability of prey; Babbitt & Tanner 1998) as well as abiotic factors (e.g. microclimates; Webb, Shine & Pringle 2005) that directly influence reptile survival (Pianka & Pianka 1970; Huey & Slatkin 1976). However, for subalpine reptiles, the direct and indirect effects of altering habitat complexity may be more pronounced than for faunal communities occurring at lower altitudes. This is because, at higher elevations, reptiles must balance the risk of predation with the need to thermoregulate (usually through direct basking) in extreme and highly variable environmental conditions (Sandercock, Martin & Hannon 2005; Huang & Tu 2008). If reptiles in these ecosystems do not take the opportunities to bask when thermal conditions are optimal, they will not maximize energy assimilation and will fail to maximize growth, time spent foraging or time spent seeking mates (Martin & Salvador 1993; Martin & Lopez 1999; Webb & Whiting 2005). However, basking poses a significant risk to immediate survival as, for many subalpine reptiles, basking opportunities coincide with peak activity periods of predators such as corvids and raptors (Baker-Gabb 1984; O'Brien et al. 2010). Consequently, structural complexity may be crucial to the long-term persistence of reptiles by moderating rates of predation (Stamps 1983; Rubbo et al. 2001; Amo, Lopez & Martin 2007).

Despite the importance of habitat structural complexity in moderating abiotic and biotic processes influencing subalpine reptiles, experimental studies investigating the inter-relationships between these factors are strikingly limited. In this paper, we address this knowledge gap by quantifying the mechanisms underpinning ski run avoidance by lizards. We suggest that two possible drivers related to structural complexity – predation and thermal environments – may strongly influence observed patterns of lizard distribution (see Fig. 1).

Figure 1.

Conceptual summary of the hypothesized relationships between lizard abundance, structural complexity, predation and temperature in subalpine habitats with differing levels of disturbance.

Hypothesis 1 – Predation

We postulated that the reduction in habitat structural complexity on ski runs could potentially increase rates of predation on reptiles because individuals basking on, or dispersing across, these areas are more visible to predators. Several studies suggested that rates of predation increase for a suite of different taxa as habitat structure is removed (see Irlandi 1994; Babbitt & Tanner 1998; Arthur, Pech & Dickman 2005) and animal behaviour is altered to compensate for this increased predation risk (or perception thereof; Cuadrado, Martin & Lopez 2001; Lopez & Martin 2013). Hence, structural complexity plays an important role in providing protection from predators and reducing perceived predation risk, and this could be driving the distributions of reptiles in subalpine ski resorts and adjacent areas.

Hypothesis 2 – Thermal Environments

We postulated that the thermal regime of ski runs dominated by exotic grasses would be altered as structural complexity was reduced. In some instances, the alteration of thermal environments can benefit reptiles as thermoregulatory opportunities may be increased (Huey & Slatkin 1976; Langkilde, O'Connor & Shine 2003). However, we argue that the extreme simplification of habitat structure may fundamentally degrade the thermal quality of grassland habitats for lizards, as intensively disturbed ski runs lose their ability to buffer against extreme temperatures. As a consequence, substrate temperatures may rise to levels much higher than the critical maximum body temperatures of lizards inhabiting subalpine environments (see Spellerberg 1972). This would limit lizard activity periods to earlier or later in the day, potentially preventing sufficient periods for energy assimilation (Huey & Slatkin 1976; Huey, Losos & Moritz 2010) and exposing animals to greater risk of predation (Fox 1978). Due to thermal constraints on intensively modified ski runs, lizards may avoid these habitats in preference for other, more thermally suitable environments.

By testing these two important hypotheses, this paper contributes to our understanding of the key constraints that drive reptile distributions in disturbed subalpine ecosystems. Our findings reveal that both temperature regimes and rates of predation can be significantly affected by human activities and that together these drivers provide a compelling explanation for the avoidance of modified ski runs by reptiles. Based on these important new insights, we propose measures to mitigate potential negative effects of habitat modification on reptile assemblages occurring in and around ski resorts.

Materials and methods

Study Species

Our intervention experiment investigated the effect of habitat structural complexity on two key ecological drivers, thermal environments and predation, which may regulate the occurrence of reptiles in disturbed grasslands on ski runs and in undisturbed grasslands. Accordingly, we selected two grassland-associated reptile species for our experiment. Both the alpine she-oak skink Cyclodomorphus praealtus and the grassland tussock skink Pseudemoia pagenstecheri are diurnally active heliotherms (Spellerberg 1972) that have strong associations with grassland or grassy-heath matrix environments (Green & Osborne 2012), and have been detected previously in skiing areas (Chloe F. Sato, unpublished data). Both species occur in the Australian subalpine region. However, C. praealtus is found only at elevations above 1500 m in the Australian Alps (Green & Osborne 2012), while P. pagenstecheri occurs more widely across south-eastern Australia (Wilson & Swan 2008). The thermal physiology of these lizards has not been specifically investigated, but one closely related sympatric species (woodland tussock skink Pseudemoia entrecasteauxii) can tolerate a wide range of body temperatures (Tmin = 2·2–2·8 °C, Tmax = 41·9–42·5 °C; Spellerberg 1972), as can a larger-bodied sympatric species – the alpine water skink Eulamprus kosciuskoi (Tmin = 2·0–3·2 °C, Tmax = 39·8–40·8 °C; Spellerberg 1972). Given the range of temperatures that P. entrecasteauxii and E. kosciuskoi can endure, it is likely that the critical body temperatures of C. praealtus and P. pagenstecheri fall somewhere within this range.

The diet of C. praealtus and P. pagenstecheri is assumed to consist predominantly of invertebrates (Green & Osborne 2012) but vegetation also may be consumed opportunistically (Brown 1991). Invertebrates are abundant in alpine and subalpine areas, including in disturbed grasslands that have been developed for skiing (Hammelbacher & Mühlenberg 1986; Negro et al. 2009; Rolando et al. 2013), and therefore are not likely to be limiting to reptiles living at high altitudes.

Study Area

We conducted our study in and around the largest alpine–subalpine resort complex in Australia, Perisher Ski Resort (36°24′S 148°24′E; PBPL 2002). The resort is located in Kosciuszko National Park, south-eastern Australia (Fig. 2a). Our grassland survey areas in the national park were characterized by native grasses [predominantly Poa costiniana (Vickery) and Rytidosperma nudiflorum (P. Morris)] and herbs [such as Empodisma minus (Hook.f.); Costin et al. 2000]. Within Perisher Ski Resort, our survey areas were located on subalpine ski slopes dominated by exotic grasses [predominantly Agrostis capillaris (Boiss. & Reuter) and Festuca rubra (L.)]. The ski slopes were almost entirely covered by exotic grasses with occasional patches of bare ground. Mean mid-summer shade temperatures are about 10 °C (but can reach maxima around 30 °C; BOM 2013), while daily mean winter temperatures are around −5 °C (Costin et al. 2000; Green & Osborne 2012). Annual precipitation is > 2000 mm per year with summers usually drier than winters (Green & Osborne 2012) and snow covering the area from mid-June to October.

Figure 2.

(a) Map of the study area in Kosciuszko National Park, south-eastern Australia. (b) Location of the replicate ‘blocks’ within Kosciuszko National Park. Each marker indicates an individual site; numbers above markers indicate a ‘block’ of sites (e.g. 1 = Perisher 1; 2 = Perisher 2; 3 = Perisher 3; 4 = Smiggin 1; 5 = Smiggin 2); letters above markers indicate grassland types within blocks (e.g. a = unmown; b = mown; c = undisturbed). Grey lines designate roads.

Experimental Design

We established a blocked experiment comprised of subalpine grassland habitats only, in two disturbance categories (disturbed/undisturbed). We located the disturbed grassland sites in the centre of ski runs within the lease areas of Perisher Ski Resort. These sites had been cleared of vegetation, graded by heavy machinery and then rehabilitated with a mix of exotic grasses (predominantly A. capillaris and F. rubra). We located undisturbed examples of grasslands outside Perisher Ski Resort. These sites exhibited minimal or no signs of disturbance from ski resort development.

We selected sites to enable a full experimental factorial design, yielding three ‘treatments’: 1) undisturbed grasslands, 2) ‘complex ski runs’ where some structural complexity was retained by leaving long exotic grasses (~28 cm) intact on the ski run and 3) ‘simple ski runs’ that displayed minimal structural complexity after grass was slashed using a ‘whipper snipper’ to ~4 cm above ground level. We replicated the three treatments five times in complete replicate blocks, giving a total of 15 sites (Fig. 2b). We ensured that all sites were 5 × 40 m rectangular plots located on south-facing slopes between 1700 and 1830 m a.s.l.

Lizard Models

Plasticine models have been used previously in many different ecosystems to estimate rates of predation (see Niskanen & Mappes 2005; Vervust, Grbac & Van Damme 2007; Daly, Dickman & Crowther 2008). We constructed models of adult C. praealtus and P. pagenstecheri using non-toxic sculpting clay (Monster Clay; The Monster Makers, Ohio, USA) using mean morphological measurements from specimens caught in the field (C. praealtus:= 30, P. pagenstecheri:= 43; Chloe F. Sato, unpublished data). After construction, we painted the models using non-toxic paint to approximate the true coloration of each species (Fig. 3). In total, we created 100 replica models of each species for use in our study.

Figure 3.

Lizard models used in predation experiment (to scale). The larger-bodied lizard represents C. praealtus and smaller-bodied lizard represents P. pagenstecheri.

Estimating Rates of Predation on Lizards

To test Hypothesis 1 (predation), we established baseline predation rates in each treatment (i.e. undisturbed grassland; complex ski run; simple ski run) before mowing, between 10 January and 14 January 2013. To do this, we placed five plasticine models of each species (C. praealtus and P. pagenstecheri) along a 40-m transect at each site (i.e. 10 models in total at a site, 150 models in total across all sites). We positioned models among (but not completely obscured by) grasses at 4-m intervals along each transect, alternating examples of C. praealtus and P. pagenstecheri. We placed a single camera trap (Scout Guard KG680V; Faunatech Pty Ltd, VIC, Australia; height, 140 mm; width, 102 mm; depth, 74 mm) on each transect to identify the types of predators that were present near the models over approximately 4 days (99 hours), before scoring predation attempts. We considered a predation attempt to include the displacement of the model from the transect, complete removal of the model from the site or visible signs of attack (e.g. bite, scratch or claw marks) on the model. For each model, we recorded 1) whether the model had been attacked, 2) the evidence for predation (e.g. displacement, removal or visible signs of predation), 3) where on the model scratch, bite or claw marks were located and 4) the type of predator that had attacked the model.

After scoring all the predation attempts along a transect, we reset camera traps and replaced lizard models at undisturbed grassland and complex ski run sites. At simple ski run sites, we removed all models from the transect then mowed the grass to ground level (i.e. < 5 cm height) 2·5 m either side of the transect. After mowing, we reset camera traps and replaced the models along the transect, as before. We left all models and camera traps for a further 4 days (99 hours) between 14 January and 18 January 2013, before scoring the new models for predation attempts.

Thermal Environments in Grasslands

To test Hypothesis 2 (thermal environments), we used temperature loggers (Thermochron i-Buttons, Thermodata Pty Ltd, QLD, Australia; diameter, 17·4 mm; height, 5·9 mm) to record the range of ground surface temperatures available to lizards in grassland sites while running our plasticine model experiment. We deployed three temperature loggers at each site where we set plasticine models. We placed one temperature logger at either end of the 40-m transect and one at 20 m, ensuring that the loggers were in contact with the ground but partially covered by grasses to prevent exposure to lengthy periods of direct sunlight. We offset all loggers two metres from the transect to avoid disturbance by animals preying on plasticine models. We programmed loggers to record temperature every 30 minutes and left the loggers in situ for the duration of the experiment (i.e. January 10th to January 18th). After 8 days, we collected all loggers for analysis.

Statistical Analysis

To explore the effect of structural complexity on predation rates, we fitted hierarchical generalized linear models (HGLM; Lee, Nelder & Pawitan 2006) to the numbers of models attacked for each ‘species’ separately and for all models regardless of species. We assumed a quasi-binomial distribution with a logit link function for the response and a beta-distribution with a logit link function for the random component. We included ‘mowing intervention’ (i.e. before mowing or after mowing) and ‘site’ (i.e. undisturbed grassland, complex ski run or simple ski run) as fixed effects and ‘treatment’ nested within ‘block’ as random effects to account for the spatial structure in the data.

To analyse temperature data, we included temperature readings taken between 05:00 and 18:30 (Australian Eastern Standard Time, ‘AEST’) when lizards were likely to be active. For these readings, we calculated the mean, absolute maximum, absolute minimum and range of ground surface temperatures recorded at each data logger, in each grassland type, over the four-day periods before and after mowing. We then used an analysis of variance (anova) to determine whether there were any significant changes in the thermal environments of different grassland types before and after our experimental intervention (mowing).

Finally, we determined the total number of hours that ground surface temperatures exceeded the absolute critical maximum and minimum body temperatures of sympatric lizards (an absolute critical minimum of 2 °C for E. kosciuskoi and an absolute critical maximum of 42·5 °C for P. entrecasteauxii; Spellerberg 1972), as a proportion of the total number of hours available for lizard activity (14 hours), before and after our mowing intervention. We then fitted a quasi-binomial HGLM to investigate whether the calculated proportion of hours differed between treatments, before and after mowing. We included site and mowing intervention as fixed effects, with the proportion of hours as the response variable. To account for the spatial structure, we included day as well as logger position and treatment nested within block as random effects.

We used Genstat 15 (VSN International Ltd) for all statistical computation.

Results

Rates of Predation

During the plasticine predation experiment, the camera traps recorded red deer Cervus elaphus, European hares Lepus europaeus, Australasian pipits Anthus novaeseelandiae, masked lapwings Vanellus miles, little ravens Corvus mellori, Australian ravens Corvus coronoides and European red foxes Vulpes vulpes. Three of these species are known predators of subalpine lizards in Kosciuszko National Park: C. mellori, C. coronoides and V. vulpes (Green & Osborne 1981; Green 2003). However, based on camera trap footage and markings on plasticine models, we conclude that corvids (C. mellori and C. coronoides) were the only predators of our lizard models. Markings consisted of beak and claw marks, predominantly located on the head (= 36), trunk (= 32) and tail (= 24) of models (see Fig. 4 for examples). We did not observe mammalian or reptilian predatory attempts.

Figure 4.

Examples of predation on lizard models. (a) Camera trap image showing C. coronoides preying on a ‘lizard’ in the field. (b) Examples of the types of markings observed on models from corvid predation attempts. Upper lizard depicts the model highlighted in (a).

Before mowing, rates of predation on the models were not significantly different between grassland treatments for either species, or when species were pooled (Fig. 5). After mowing, the rates of predation were altered in each treatment, with the lowest levels of predation occurring in undisturbed grasslands and the highest levels occurring on structurally simple (mown) ski runs (Fig. 5). For individual species, the interaction between mowing intervention and treatment was not significant (C. praealtus: math formula = 3·935, = 0·14; P. pagenstecheri: math formula  = 3·505, = 0·173). However, when predation data were pooled for both species, the interaction was significant (math formula = 7·989, = 0·018; a complete table of estimates from the HGLMs is provided in Table S1 in Supporting Information). This indicates that sites which were mown had significantly higher overall rates of predation compared with baseline (pre-mowing) levels (Fig. 5c). Where structural complexity was not reduced [i.e. undisturbed grassland and unmown (complex) ski runs], overall rates of predation did not differ significantly with intervention (Fig. 5c). Interestingly, predation rates in the undisturbed grasslands were lower after the mowing intervention (Fig. 4), suggesting that corvids learnt that models were inedible in these locations or that the birds may have been attracted to calls from conspecifics at nearby resorts (Bugnyar, Kijne & Kotrschal 2001) where prey items were more obvious due to mowing.

Figure 5.

Estimated mean proportions (±SE) of predation attempts on plasticine models in undisturbed grasslands and on ski runs (‘SR’), for all models and by individual species. (a) Predation of C. praealtus models, (b) predation of P. pagenstecheri models and (c) all predation attempts pooled.

Thermal Environments

Before the mowing intervention, both simple and complex ski runs had higher mean, maximum and minimum ground temperatures than undisturbed grasslands (Table 1). In addition, ski runs had a greater range of ground temperatures compared to undisturbed grasslands (Table 1). We observed these same ground temperature patterns after mowing (Table 1). However, in areas where extreme simplification of structural complexity was undertaken (i.e. simple ski runs), mean and maximum ground temperatures were significantly higher, and the range of ground temperatures was significantly greater than in undisturbed grasslands (Table 1). Mean and maximum ground temperatures on simple ski runs also were higher than on complex ski runs (Table 1), but this difference was not significant.

Table 1. Mean, maximum, minimum and range of temperatures recorded (with least significant differences; ‘l.s.d’) in undisturbed grasslands (C), complex ski runs (CSR) and simple ski runs (SSR), before and after mowing had occurred
Temperature Measure (°C)Before MowingAfter Mowingl.s.dF
CCSRSSRCCSRSSRSame level of treatmentAll other comparisons
  1. *< 0·01; **< 0·001.

Mean 16·620·020·019·122·426·21·132·94F2,42 = 14·64**
Maximum35·438·839·935·038·647·13·086·01F2,42 = 8·08*
Minimum −2·70·90·8−1·11·6−0·21·102·88F2,42 = 5·91*
Range 38·137·939·136·137·047·33·637·03F2,42 = 9·72**

We also found that mowing altered the length of time that ground temperatures exceeded the critical maximum body temperatures of lizards. Before mowing, there were no significant differences between grassland types in the length of time exceeding the critical maximum body temperatures of lizards (control: 0·29 ± 0·21 h day−1; complex: 0·44 ± 0·17 h day−1; simple: 0·45 ± 0·23 h day−1). However, after mowing, sites with grass removed had significantly longer periods where ground temperatures exceeded the critical maximum body temperatures of lizards (undisturbed grassland: 0·12 ± 0·12 h day−1; complex ski run: 0·49 ± 0·25 h day−1; simple ski run: 2·20 ± 0·45 h day−1; math formula  = 75·45, < 0·001; Fig. 6). The length of time that ground temperatures exceeded the critical minimum body temperatures of lizards was not significantly different between grassland types, before or after mowing (math formula  = 3·701, = 0·157; a complete table of estimates from the HGLMs is provided in Table S2).

Figure 6.

(a) Predicted mean proportion of time (±SE) that surface ground temperatures exceeded the critical maximum body temperature (Tb) of lizards in three grassland types, before and after mowing. (b) Mean absolute maximum ground surface temperatures (°C) recorded in three grassland types during lizard activity periods, before and after mowing. The ‘time of day’ is reported as Australian Eastern Standard Time (AEST). The dashed horizontal line represents the absolute critical maximum body temperature of a sympatric lizard (P. entrecasteauxii) recorded by Spellerberg (1972).

Discussion

In this paper, we sought to determine whether two key ecological drivers – thermal environments and predation – are affected by the removal of structural complexity (through mowing) and consequently contribute to the distributions of reptiles in subalpine grasslands observed in a previous study (Chloe F. Sato, unpublished data). We found that both thermal regimes and rates of predation were significantly affected by the removal of structural complexity. Overall rates of predation were significantly higher on ski runs where structural complexity was removed (Fig. 5). In addition, the absolute maximum and mean ground temperatures were higher on simple ski runs, and exceeded the critical maximum body temperatures of lizards for longer periods, than either complex ski runs or undisturbed grasslands (Table 1 and Fig. 6). As both thermal regimes and predation are key drivers that can affect the survival of reptiles in the short term and long term (Martin & Lopez 1999; Webb & Whiting 2005), we argue that these drivers contribute to the avoidance of mown ski runs by lizards.

The thermal environment is an important determinant of reptile occurrence (Webb & Whiting 2005; Daly, Dickman & Crowther 2008). Our results indicate that the removal of grassland structural complexity had a dramatic effect on ground temperatures. While reptiles are well adapted behaviourally and physiologically to cope with thermal variability (Huey, Losos & Moritz 2010), the extreme maximum ground temperatures that occur on mown (simple) ski runs preclude the use of these habitats by lizards for extended periods of the day. As vegetation is reduced to a short herbaceous layer by mowing (i.e. low structural complexity), lizards cannot modify their behaviour to avoid these extreme temperatures by positioning themselves higher in the vegetation or by seeking refuge under cover objects or in areas of shade (Huey 1974). Consequently, simplified ski runs heighten the risk of overheating while foraging, thermoregulating or finding mates, especially during hotter periods of the day. Even though the unmown (complex) ski run still comprises a single herbaceous layer, the increased height and density of the grass resulted in considerable thermal buffering (Fig. 6).

In addition to suboptimal thermal environments, the unsuitability of mown (structurally simple) ski runs is further increased by elevated risks of predation. The extreme simplification of these areas provides no refuge from predatory attempts. Thus, reptiles are likely to perceive the ski runs as high-risk, low-quality habitat. Indeed, the increased perception of predation risk by lizards on ski runs is supported by Amo, Lopez & Martin (2007) who found that lizards moved further and at greater speeds on ski runs in Europe, in response to the low availability of refuges. Additionally, this risk of predation (and the perception of this risk; Lima 1993) is likely to increase as the width of the ski runs increases, because lizards will have to travel further to reach refuges. For large-bodied reptiles (and gravid females), this is particularly important because 1) they may be more susceptible to predation by avian predators that select for larger-bodied individuals during bird breeding seasons (Padilla, Nogales & Marrero 2007) and 2) ski resorts encourage larger concentrations of scavenging birds (Storch & Leidenberger 2003; Jokimaki et al. 2007) that also may prey on lizards in the area. Thus, to facilitate the movements of reptiles throughout disturbed ski resorts, it is essential that additional cover is available on ski runs. Our results indicate that even the absence of mowing provided enough cover to significantly reduce rates of predation. Hence, measures to maximize complexity – such as the cessation of mowing and/or the rehabilitation of native forbs and grasses on ski slopes – will enhance disturbed areas for reptiles.

From our field observations (Chloe F. Sato, unpublished data), we suggest that lizards in ski resorts are favouring alternative habitats to mown ski runs (such as rock outcrops, woodlands, heathlands and undisturbed natural grasslands) where structural complexity is higher and opportunities exist for shuttling between areas of shade and sun. These habitats provide additional cover for avoiding predators and a greater opportunity to avoid extended exposure to extreme temperatures. The selection of habitats other than structurally simple (mown) ski runs by lizards does not pose a threat to their persistence in these disturbed subalpine landscapes. However, if ski resorts continue to expand and new ski runs are constructed using current slope-grooming practices, habitat will increasingly be fragmented and the remaining patches utilized by lizards will progressively become more isolated from one another. As lizards generally have low dispersal abilities (e.g. between 20 m and 57 m depending on the species; see Clobert et al. 1994; Olsson & Shine 2003), mown ski runs with widths > 20 m have the potential to seriously inhibit lizard movement and dispersal, particularly during the summer and autumn months when reproduction and dispersal of young are occurring (Green & Osborne 2012). However, there have been very few studies of lizard movement patterns in alpine or subalpine environments (but see Amo, Lopez & Martin 2007), so it is possible that 20 m is an under- or overestimate of what constitutes a barrier to dispersal, especially considering the effect that factors such as body size, age and sex have on dispersal ability (Olsson & Shine 2003; Warner & Shine 2008).

Clearly, it is important that we begin to consider how the management of ski runs affects underlying ecological processes and constraints as, at present, it is likely that modified ski runs are having significant negative effects on reptiles in affected areas. These impacts may increase with the alterations in thermal regimes associated with climate change. Higher ambient temperatures will lead to higher ground surfaces and a reduced time for reptiles to forage and thermoregulate within their operative temperatures (Huey, Losos & Moritz 2010). This will result in reptiles assimilating less energy for the purposes of foraging, reproducing and avoiding predation (Huey & Slatkin 1976). Higher ambient temperatures also could result in increased predator densities at higher altitudes (Green 2006). As the duration of snow cover decreases, territorial birds that prey on reptiles but are currently restricted to lower elevations (e.g. laughing kookaburra Dacelo novaeguineae) may begin to establish territories at higher elevations (Green 2006). Hence, it is imperative that we further map reptile distribution patterns in subalpine and alpine environments and better elucidate the underlying processes giving rise to these patterns, so we are equipped to effectively manage reptile diversity in these areas, now and in future.

Management Implications

Temperature regimes and rates of predation together provide a compelling explanation for the avoidance of modified ski runs by reptiles. While our results are only directly applicable to Perisher Ski Resort, they are also of importance to ski resorts with similar management practices and similar assemblages of predators. As such, we recommend that to facilitate the persistence of reptiles in disturbed subalpine environments, management plans must focus on implementing strategies that reduce the impact of these key environmental drivers. Our results show that rates of predation were lowest and the thermal environment was most conducive to lizard activity in undisturbed grasslands. It is also in these environments that lizard abundances were highest (Chloe F. Sato, unpublished data). Based on this information, we suggest that undisturbed grasslands located in ski resort lease areas should be preserved wherever possible to facilitate the persistence of reptiles, particularly grassland specialists. There also may be a need to revegetate ski slopes with native plant species to restore linkages between otherwise presently fragmented habitat. This may be imperative for threatened species that have limited distributions such as C. praealtus (TSSC 2009).

Where undisturbed grasslands cannot be retained or are already highly modified, we suggest that the retention of structural complexity on ski runs (e.g. through the cessation of mowing at the very least during peak reptile activity periods) will concurrently provide refuge from predators and buffer against extreme temperatures, making ski runs more hospitable to reptiles. We advise against the intensive management of ski runs (e.g. mowing, slope grooming and vegetation removal), particularly when reptiles are reproductively active and when young are likely to be dispersing during the warmer spring and summer months. We also advise that the development of future ski runs includes the retention of low-growing native plant species and the avoidance of exotic grasses in revegetation.

Conclusions

The effective management of reptile biodiversity in disturbed subalpine landscapes requires an understanding of the ecological constraints driving observed patterns of distribution and abundance. Our study has demonstrated that two key ecological drivers – predation and thermal environments – can be significantly affected by human intervention. In turn, this degrades the quality of grassland habitats on ski runs for reptiles. Based on our results, we suggest that the retention of structural complexity on ski runs will concurrently provide refuge from predators and buffer against extreme temperatures, making ski runs more hospitable to reptiles.

Acknowledgements

This research was supported by the Glenn Sanecki Alpine Ecology Scholarship. Ethical approval was provided by the Australian National University, Animal Experimentation Ethics Committee (Protocol No. S.RE.11.10), and approval to work within Kosciuszko National Park was provided by the Department of Environment and Conservation (Scientific Investigation Licence No. S13155). We would like to thank the national parks and wildlife service staff for assisting with slashing ski runs and Perisher Ski Resort for access to ski resort areas. We would also like to thank L. Rayner for assisting in model construction and reviewing an earlier version of the manuscript.

Ancillary