Habitat Use and Thermal Biology of the Land Mullet Egernia major, a Large Scincid Lizard from Remnant Rain Forest in Southeastern Australia

Copeia, 2000(4), pp. 931 939 Habitat Use and Thermal Biology of the Land Mullet Egernia major, a Large Scincid Lizard from Remnant Rain Forest in Southeastern Australia A. KLINGENBÖCK, K. OSTERWALDER, AND R. SHINE Land Mullets (Egernia major, Scincidae) are large (60 cm total length), powerful glossy black lizards that are restricted to rain forest and associated habitats in southeastern Australia. We conducted the first ecological study of these spectacular animals to evaluate the degree to which anthropogenic activities might threaten population viability. Twelve adult lizards were captured in the Barrington Tops area of eastern New South Wales and implanted with miniature temperature-sensitive radiotransmitters. The lizards were released at their sites of capture and located daily for the next six weeks. In conjunction with surveys of surrounding areas, our data document strong patterns in habitat selection at various spatial scales. The lizards are most abundant in ecotonal forest rather than in either dense rain forest or more open eucalypt-dominated areas. They are most abundant in areas with high numbers of large fallen logs and spend most of their time in or near these logs. Radio-tracked animals were generally located close to clearings (including roads) but actively avoided crossing roads. Land mullets are diurnal heliotherms, basking for long periods each day to achieve body temperatures around 30 C. Their black color increases heating rate, and their large size retards cooling, thus facilitating retention of high temperatures as they forage in cool forest habitats. The animals ability to take advantage of anthropogenic disturbance (partial clearing of vegetation) suggests that E. major populations are likely to be viable as long as suitable habitat (especially large fallen timber) is available. LIKE many other habitats in Australia, the rain forests of the southeast have experienced massive changes since the arrival of Europeans 200 years ago. Palynological and historical records suggest that rain forest covered extensive areas of coastal and near-coastal habitats at this time (White, 1986). Clearing for timber extraction and agriculture resulted in a rapid decline in both the size and connectedness of rain-forest patches, such that today these dense forests cover only a small fraction of their former area (Recher et al., 1986; White, 1986). Conservation of the southern forests has thus become a major public issue and has attracted enormous publicity (e.g., Lindenmayer et al., 1991). Remarkably, however, the biology of the fauna in these areas, including the rain-forest endemics that may well be threatened by habitat loss, remains poorly known (Gibbons and Lindenmayer, 1997). This is particularly true of rain-forest reptiles, which have attracted less study than have the more charismatic marsupials (e.g., Lindenmayer et al., 1990, 1991) and birds (Kavanagh and Bamkin, 1994; Kavanagh and Turner, 1994). One of the most striking examples of this dearth of scientific information is a giant scincid lizard, the Land Mullet, Egernia major. These large (to 60 cm total length, 670 g), glossy black lizards are among the most spectacular reptiles in Australia and among the largest 5% of lizard species in the world (Pough, 1980). They are widely distributed along coastal eastern Australia and are locally common in remnant rain forest. Nonetheless, there is essentially no published information on the ecology of this species (Shea, 1999). Their apparent restriction to a significantly threatened habitat type (remnant rain forest) suggests that these animals may be potentially vulnerable to anthropogenic disturbance. Hence, information on the basic biology of the species may provide a useful basis for future management decisions. Our study constitutes the first field-based research on land mullets and focuses on two aspects of the species biology that may be particularly significant for conservation planning. First, we describe patterns of habitat use, with particular reference to the lizards reliance on fallen timber as refugia and their avoidance of roads. Second, we describe the species thermal biology. The distribution of heliothermic (sun basking) species such as E. major is likely to depend upon the availability of suitable basking sites, especially in dense rain forest with little penetration of sunlight to the forest floor (Sar- 2000 by the American Society of Ichthyologists and Herpetologists

932 COPEIA, 2000, NO. 4 torius et al., 1999). Thermal constraints may be especially important for large lizards, which are likely to require long periods of basking before they can attain preferred body temperatures (Hillman, 1969; Vitt et al., 1997, 1998). Thermal requirements might, therefore, influence habitat selection by E. major and, hence, be relevant to their management. MATERIALS AND METHODS Study species. Egernia major is the largest skink in Australia, with an average adult snout vent length (SVL) of around 30 cm. The tail is about 115% of SVL (Shea, 1999). Land mullets (so named for their morphological resemblance to this type of fish) are fast moving, muscular, diurnal, heliothermic lizards. The species is restricted to rain forests and wet sclerophyll forests in coastal and near-coastal areas from southeastern Queensland in the north to Gosford (New South Wales) in the south (Wilson and Knowles, 1988; Ehmann, 1992). The dorsal and lateral surfaces of adult land mullets are glossy dark brown to jet black. The venter is pale, with variable dark flecks and an orange tinge on the throat. Juveniles have a few pale spots with dark outlines laterally on the neck and about the axilla. Study area. Our study area of 25 ha was located on the southern border of Barrington Tops National Park, 250 km northeast of Sydney (New South Wales). The climate ranges from cool temperate at higher elevations to warm temperate at lower elevations. Annual mean temperatures are between 24 C (maximum) and 10 C (minimum). Average annual rainfall is 1500 mm. The region surrounding the study area is mostly within State Forest and comprises a diversity of forest types: moist coastal eucalypt forest, dry coastal eucalypt forest, subtropical rain forest, warm temperate rain forest, cool temperate rain forest, and dry rain forest. The study site was covered primarily with subtropical rain forest. Radiotelemetry methods. We caught 12 animals (7 males, 5 females) by hand in October 1998. Based on our observations, these comprised most of the adult animals in our study area. We transported the lizards to the University of Sydney for surgical implantation of temperaturesensitive radiotransmitters (models PD-2T and BD-26T, Holohil Pty. Ltd., battery life 4 to 6 months). Telemeters weighed 1.8 3.4 g, 1% of lizard body mass. Prior to implantation, each transmitter was encapsulated and calibrated against known temperatures. Compared to other lizard species with which we have worked, land mullets required higher doses of anaesthetic (halothane at concentration of 5%) and took longer to achieve surgical levels of anesthesia ( 20 min). The thick scales and strong abdominal musculature of these animals also necessitated minor changes to usual surgical procedures. We sutured each muscle layer individually and used surgical glue to strengthen the external sutures (Klingenböck and Osterwalder, 1999). We commenced radio-tracking the lizards 7 10 days after their release at the initial point of capture. Each animal was located telemetrically between 50 and 62 times over the period 27 October to 15 December. The tracking gear consisted of a three-element Yagi antenna, a Titley Electronics Australis 26K receiver and a device to measure pulse intervals (Titley Electronics Microprocessor Period Meter). Each time we located a lizard, we took the following data: date, time, location, pulse interval, percentage of body outside refuge, percentage of body in sun, behavior, distance to other lizard(s), macrohabitat, and microhabitat. To obtain approximately equal sample sizes at each time of day, we divided the day into hourly intervals from 0800 to 2100 h. We attempted to locate each animal the same number of times in each time interval. We attempted to randomize the sequence in which lizards were located, within constraints set by logistical factors. Analyses of habitat use. We examined habitat use at two spatial scales (Aebischer et al., 1993). First, we compared areas that contained E. major to nearby areas not used by the species, to identify macrohabitat features associated with the presence of lizards. Second, we examined the sites used by lizards compared to other unused sites within the animals home ranges. We carried out two different sets of comparisons at this spatial scale: one concerning vegetational associations, and one involving substrate type. In combination, these data allow us to identify factors that are likely to influence the species distribution and abundance. Macrohabitat comparisons. Three types of forest were distinguishable within the overall study area: dense rain forest, open eucalyptus forest, and an intermediate ecotonal type, hereafter referred to as mixed forest. Lizards were only seen commonly in the mixed forest. This distribution suggested that the mixed forest may differ in consistent ways from the other areas in terms of factors important for the lizards. In an

KLINGENBÖCK ET AL. ECOLOGY OF RAIN-FOREST LIZARDS 933 attempt to identify the nature of those differences and hence potential determinants of lizard distribution on an across-habitat spatial scale, we carried out a quantitative survey of attributes that seemed potentially relevant to the biology of E. major. This survey involved the selection of 10 plots (each measuring 20 20 m) within each of four habitat types. Two of the types were dense rain forest and open eucalypt forest. In both of these areas, the plot locations were selected at random. The other two types were both located within the mixed forest. Ten of these plots (the used plots) were each centered on a location at which a telemetered lizard had been recorded. The other 10 ( unused ) plots were places where lizards were never sighted. Lizards were later seen in four of these unused plots, so we reclassified these as used and surveyed an additional four unused areas. In total, then, we obtained data on 44 plots: 10 in the dense forest, 10 in the open forest, 10 unused plots in the mixed forest, and 14 used plots in the mixed forest. We searched each of these plots to locate potential lizard refugia. We counted all hollow logs large enough to provide shelter for an adult E. major, and we also recorded log length and diameter and whether they were covered in moss. Canopy cover in the middle of each plot was quantified by photographing the canopy with a digital camera (Nikon Coolpix 900). The resulting photograph was transformed from color into black and white using a computer program (Photoshop 4.0) that allowed us to estimate the average canopy cover (histogram function, luminosity). Vegetation associations. At a finer spatial scale, the use of vegetation associations was examined within the mixed forest by comparing locations of radio-tracked lizards with 80 randomly chosen sites. We recognized the following categories. (1) Open forest relatively open areas with grassy understory and considerable light penetration through the canopy. Clearings within this habitat (as in all others) were counted separately. (2) Dense forest denser component of the ecotone, with an understory of shrubs and Lomandra bushes. (3) Shrub any lizard found under a shrub was classified into this habitat type. (4) Clearings an area 10 m 2 cleared of vegetation by natural causes or human influences. Roads, parking lots, and lawns fall into this category. Although the separate treatment of clearings within habitat types might appear arbitrary, most clearings had distinct boundaries. As our results show, these areas may be of particular importance to the lizards. Microhabitat use. The following substrates were distinguished: leaf litter, weeds, mossy logs, bare logs, and bare soil/rocks. In a similar manner to the macrohabitat analyses (above), we compared substrate usage by the lizards (from locations determined during radio-tracking) to characteristics scored at an array of 80 random points. Avoidance of roads. To test whether the lizards avoid crossing roads, we developed a random walk model. Null distributions of random walks were generated for each lizard, using the actual data on daily distances moved by that animal during the telemetry study. These daily distances were combined in a random sequence, using a random direction of travel between successive days. One hundred random walks were calculated for each lizard, starting at the place where the animal was located one week after its release. For each of these random walks, the number of times the lizard s path crossed a road was counted. This procedure generates a frequency distribution of road crossings expected under the null model (no avoidance of roads), which can then be compared with the number of actual road crossings that each lizard exhibited during our study. Thermoregulation. To quantify the range of body temperatures potentially available to the lizards, we recorded ambient temperatures using miniature Hobo-temp data loggers set to record temperatures every 10 min. Three data loggers were used, to obtain representative information on (1) air temperature in the shade, (2) retreat temperature deep inside a hollow log (to measure the lowest available temperature), and (3) operative temperature of a lizard model in a site exposed to full sun throughout the day. For this last purpose, we placed the tip of the thermistor probe inside a hollow copper model. Temperatures inside this model (painted black to mimic reflectance of the lizards) provided an indication of the maximum temperatures available to a lizard that was exposed to that situation for an indefinite period (Peterson et al., 1993; Vitt and Sartorius, 2000). Measurements were taken over the period from 4 November until 16 December 1998. Body temperatures of E. major were determined from pulse intervals of the transmitters, which were recorded each time a radio-tracked animal was located in the field.

934 COPEIA, 2000, NO. 4 TABLE 1. CHARACTERISTICS OF FOUR MAIN HABITAT TYPES (DENSE FOREST, MIXED FOREST EITHER USED OR UNUSED BY TELEMETERED LIZARDS, AND OPEN FOREST) BASED ON SURVEYS OF 10 REPLICATE PLOTS WITHIN EACH TYPE (EXCEPT THAT 14 SITES WERE EVALUATED IN THE USED PLOTS). The table shows mean values for each variable plus standard deviation in parentheses. The final two columns show results from a one-factor ANOVA for ln-transformed data with habitat type as the factor, to test the null hypothesis that the four types of plots differed with respect to the characteristic involved. Characteristic No. of logs Log diameter Log length Canopy cover Dense forest 3.10 (2.2) 0.58m (0.7) 6.40m (3.1) 85% (2.3) Mixed forest Used Unused Open forest F-value P-value 5.80 (2.5) 0.46m (0.1) 9.56m (3.7) 72% (9.4) 3.07 (1.5) 0.41m (0.1) 6.10m (2.8) 74% (9.3) 2.78 (1.1) 0.41m (0.2) 4.20m (2.0) 52% (11.8) F 3,39 3.98 F 3,39 0.27 F 3,39 3.44 F 3,39 20.52 0.014 0.85 0.026 0.0001 RESULTS Macrohabitat comparisons. We used a one-factor ANOVA, with habitat type as the factor, to compare the four areas with respect to canopy cover, number of logs per plot, and the mean length and diameter of logs (Table 1). All dependent variables were ln-transformed prior to analysis to improve normality of the distributions of these variables. Strictly speaking, the results from this analysis should be interpreted as evidence for location effects rather than habitat effects, because the replicate plots within each habitat type were all in the same general area. To this extent, the design of the survey introduces the problem of pseudoreplication (Hurlbert, 1984). Unfortunately, logistical problems made it impossible to carry out the survey on a broader spatial scale with truly independent replicates. Despite this problem, the patterns that were documented may provide a useful baseline for future, more comprehensive studies. The results from our ANOVAs show clearly that the three habitat types differ significantly with respect to several characteristics. (1) Number of logs per area the plots used by lizards within the mixed forest contained more logs than did any of the other habitat types (Table 1). Posthoc tests (Fisher s PLSD) confirmed the significance of these differences, and the lack of any significant differences among mean values for any of the other three areas. Thus, lizards were most abundant in habitats that contained many potential shelter sites. (2) Log size mean diameter of logs did not differ significantly among the four habitat types, but the overall lengths of logs differed (Table 1). Posthoc tests showed that logs were significantly longer on average in the areas used by lizards than in nearby areas that were not used or in open eucalypt forest. However, mean log length did not differ significantly between lizard areas and dense forest. (3) Canopy cover not surprisingly, the dense rain forest had significantly greater mean canopy cover than did the mixed forest, which in turn had greater cover than did the open forest. Within the mixed forest, canopy cover did not differ significantly between areas inhabited by lizards versus nearby areas where lizards were not seen (posthoc tests, all other comparisons significant at P 0.05). Several of these results suggest that the availability of logs may play an important role in the life of E. major. In particular, areas occupied by lizards contained more logs and larger logs than did nearby areas where lizards were scarce or absent. This association between lizards and logs can also be examined in another way. Rather than comparing among plots with and without lizards, the distance to the closest log for all lizard locations can simply be compared with the same parameter for random points. The result is highly significant (ANOVA, F 1,289 147.43, P 0.0001), showing that E. major are found closer to logs than would be expected under the null hypothesis of no preference. Vegetation associations. Comparisons of attributes of sites used versus not used by lizards, within the boundaries of the animals usual home ranges, identified several differences. These divergences were greater than would be expected under the null hypothesis of no difference between the two types of sites (contingency table analysis based on actual counts, although Fig. 1 shows proportions for clarity; combining minor categories to avoid very low sample sizes per cell; 2 23.08, df 5, P 0.003). In particular, the lizards frequently used small clearings within the forest, although such sites are relatively scarce overall (based on the sample of random points). This result suggests that E. major may not only use natural clearings caused by tree fall but also

KLINGENBÖCK ET AL. ECOLOGY OF RAIN-FOREST LIZARDS 935 Fig. 1. Macrohabitat use by radio-tracked land mullets (Egernia major), compared to randomly selected sites within the same study area. The following categories were distinguished: clearing (cl), dense forest (df), clearing within dense forest (df cl), open forest (of), clearing within open forest (of cl), shrubs (sh), clearing among shrubs (sh cl), and other. artificial clearings due to anthropogenic activities. To see whether indeed land mullets selectively use human-made clearings, we calculated the distance from the center of each animal s home range (i.e., center of the 10% harmonic mean area, calculated from the telemetry location data) to the nearest road or cleared area. These distances could then be compared to the same parameter calculated for 50 randomly placed points within the field area. Consistent with the overall use of cleared areas (above), the home range centers were significantly closer to the clearings (forest edges and roads) than were the random points. The mean distance from a lizard s activity center to the nearest road was 17 m (SD 2.2 m), whereas the equivalent distance from a random point averaged 51 m (SD 3.9 m). One-factor AN- OVA confirmed that these two distances differ significantly (F 1,60 8.31, P 0.01). Microhabitat use. Our radio-tracked lizards displayed nonrandom use of different types of substrates. Land mullets were found on bare logs and bare soil more often than would be expected under the hypothesis of random use of microhabitats (Fig. 2; combining minor categories to avoid low sample sizes per cell; 2 23.79, df 4, P 0.0001). Avoidance of roads. Our observations of freeranging lizards indicated that, although they were usually found close to clearings, they rarely ventured out into open areas. Thus, although the lizards spent much of their time close to roads, they seemed reluctant to cross them. To test this hypothesis, we estimated the probability Fig. 2. Microhabitat use by radio-tracked land mullets (Egernia major), compared to randomly selected sites within the same study area. The following categories were distinguished: leaf litter, weeds, mossy logs, bare logs, and soil and rocks. that lizards would cross roads under a random walk model (see Materials and Methods). If the lizards do not avoid crossing roads, we expected that the radio-tracked lizards should have crossed roads about as often in real life as they did in the random walk model. Ten of the 12 lizards crossed roads less frequently than predicted by the random walk model. The probability of this bias arising by chance is 0.02 (Fisher s Exact Test: Seigel, 1956). Thus, we concluded that our telemetered animals actively avoided crossing roads. Thermal characteristics of the habitat. There was considerable day-to-day variation in weather conditions over the course of our study. Minimum daily shade temperatures ranged from 4.3 19.6 C and maximum shade temperatures from 12.4 34.0 C. However, minimum operative temperatures showed little variation either with time of day or among days with different weather conditions. This relative constancy reflects the fact that the probes were deep inside the shelter logs and hence well buffered from outside conditions. In contrast, operative temperatures inside copper models placed in full sunlight fell to low levels overnight but rapidly warmed during the day. On hot clear days, the models exceeded 36.4 C (the highest lizard body temperature recorded during the study) by 0830 h in the morning. Indeed, the models often remained at more than 45 C for much of the day. However, on cool rainy days, the models never attained the temperature range typical for active E. major. Body temperatures of radio-tracked lizards. Lizard temperatures showed strong diurnal cycles (Fig. 3). On warm days (average daily temperature

936 COPEIA, 2000, NO. 4 Fig. 3. Lizard body temperatures and associated operative and ambient temperatures during the radiotelemetry period. Data on body temperatures are combined for all 12 radio-tracked lizards over warm days (mean air temperature 20 C; upper graph) and cool days (mean air temperature 16 C; lower graph). Data are also provided for temperatures from a probe inside a hollow log ( shelter ) and for operative temperatures of a physical model in full sunlight ( model ). 20 C, see upper panel of Fig. 3), the lizards basked in the early to midmorning period (0800 1100 h) and reached temperatures close to 35 C. They then maintained their body temperatures near this level (within the range of 28 35 C), shuttling between sun and shade. The comparison between model temperatures and lizard temperatures suggests that the animals maintained these high thermal levels for as long as possible. Lizard temperatures did not fall below the preferred level until model temperatures declined below this point. At this time of the day, the animals were often found near shelters. Overnight the animals temperatures fell to ambient levels, similar to temperatures within their shelters. The lower panel of Figure 3 shows temperatures recorded during 12 cold days (average daily temperature 16 C) during the field period. Even on these days, the animals often attained body temperatures close to 35 C, but usually only in the afternoon. On cool and rainy days, the animals frequently did not emerge from their shelters. Although the animals were clearly able to maintain high, relatively constant body temperatures in good weather, they were nonetheless constrained to a significant degree by the availability of basking opportunities. Some days were too cool and overcast for the animals to achieve their preferred temperature range. The strongest evidence that weather constrained lizard thermoregulation during our study comes from the lack of lizard activity on cold days and the significant positive correlation between mean body temperature of E. major and average daily temperature (n 640, r 0.21, P 0.001). The lower mean body temperature on colder days reflects the lower proportion of lizards basking on these days. The maximum body temperature achieved on cool days was similar to the maximum temperature on hot days (Fig. 3). If analysis is restricted to data gathered at times of day when a wide range of operative temperatures were available (model temperature 40 C), the preferred thermal range of this species can be defined more precisely. The most striking aspect was the consistency of this preferred range and of the maximum temperatures attained (Fig. 3). All of the radio-tracked lizards displayed very high and very similar body temperatures. Variation around the overall mean temperature of 29.9 C was remarkably low (SE 0.4). Analysis of the data revealed no significant difference between male and female lizards in terms of their mean selected temperatures (ANOVA on ln-transformed temperatures, F 1,205 1.03, P 0.31) or their variance in body temperature (ANOVA on the standard deviation, F 1,10 0.02, P 0.90). The similar standard deviation for both sexes suggests that neither sex tended to thermoregulate more carefully than the other. DISCUSSION Although our data were derived from a relatively small number of lizards in only one area, they provide the first detailed information on habitat use and thermoregulation of these elusive and poorly known rain-forest animals. Most of our results were clear and were evident from different techniques and at different spatial scales. Hence, we believe that the factors identified by our analysis do indeed play a causal role in determining the distribution and abundance of E. major. Our study confirmed that land mullets are diurnal heliotherms, and the telemetry data reveal a surprisingly precise thermal preferendum. The lizards spend much of their time basking and are able to attain preferred thermal

KLINGENBÖCK ET AL. ECOLOGY OF RAIN-FOREST LIZARDS 937 levels on most days. Cool cloudy days constrain the animals in this respect, as does the thick rain-forest canopy and the lizards large body size (and hence relatively slow rate of heating). The same situation has been described in large heliothermic lizards in rain-forest habitats of South America (Hillman, 1969; Vitt et al., 1997; Sartorius et al., 1999). Paradoxically, the large body size of land mullets may be thermally advantageous as well as disadvantageous. Although this large body size retards heating, it also retards cooling, so that the lizard can remain at relatively high temperatures without frequent basking. Physiological abilities to retard cooling are also more effective in large reptiles than in smaller animals (e.g., Grigg et al., 1979). The combination of these two effects may enable adult lizards to remain warm while foraging in cool, dense rain-forest habitats. This size-related effect (as well as vulnerability to predation) may explain why juveniles restrict their activity primarily to the vicinity of basking sites (Hillman, 1969; Klingenböck and Osterwalder, 1999). The dark coloration of this species may also facilitate heat uptake during basking (Bartholomew, 1982). Many cold-climate reptiles are darker colored than their warmer-climate relatives, and studies on reptile skin show that darker colors accelerate heating (Gibson and Falls, 1988). The closest relative to E. major is Egernia frerei, a lighter colored species of open forests and seasonally dry woodland (Cogger, 1992). Clearly, the lizards used available habitat in a highly nonrandom way at various spatial scales. They were largely restricted to sites that provided three major advantages: large hollow logs as refuge sites, relatively thick vegetation (presumably affecting both food supply and the lizards vulnerability to predation), and enough open areas to facilitate behavioral thermoregulation (i.e., basking in direct sunlight). The combination of these three factors successfully predicts the distribution of E. major in our study area at the following spatial scales. (1) At a scale of kilometers (broad habitat types), the lizards prefer mixed forest with intermediate canopy cover to either more open or very dense forest. The mixed forest provides basking opportunities but also considerable shelter at ground level. Logs are abundant here, as they are in most of the other areas that were surveyed. (2) At a scale of tens or hundreds of meters (i.e., within the mixed forest), the lizards select areas with high availability of logs. Canopy cover is not an important determinant of lizard distribution at this scale, because most areas within this habitat type provide abundant opportunities for basking. However, proximity to open areas that provide more hours of direct sunlight per day appears to enhance habitat suitability for the lizards. (3) At a scale of meters (i.e., within mixedforest areas with large logs and good basking sites), the lizards prefer to bask upon bare wood or bare soil rather than on leaf litter or mossy logs. Thermoregulatory needs may drive these preferences, if lizards on drier ground can warm more rapidly than those on damp substrates. In total, these analyses suggest that land mullets will be most abundant in ecotonal areas or patches within the dense forest that have been opened to sunlight by processes such as windthrow or human interference. The relative importance of shelter and thermoregulatory opportunities may well vary across the geographic range of these lizards. Because our study was conducted on a population close to the southern limit of the species range, thermal factors may have played a more significant role than they would under warmer climatic conditions. Patterns similar to those that we have documented in land mullets occur in many other forest ecosystems, with heliothermic animals using both natural and human-made openings in the forest because of the increased availability of basking sites (e.g., Vitt et al., 1997, 1998; Sartorius et al., 1999). The time scale of our study was too brief to investigate longer-term influences on the distribution of this species. For example, the virtual absence of these lizards from the open eucalypt forest might reflect not only their reluctance to spend time in areas where they are relatively exposed to predators but also the greater frequency of wildfires in these habitats. According to local residents, fires regularly move through the open forest but rarely penetrate the mixed or dense forest. In the area we surveyed, fire scars were evident on standing trees and on logs in the open forest and on the edges of the mixed forest but not in any other habitats. The higher frequency of fires in the open habitats might also explain why logs were on average shorter and less common in these areas than in the sites used by lizards (see Table 1). The association of land mullets with very large hollow logs is a central aspect of the ecology of this species. Logs function as protection against predators (such as large varanid lizards) and may also serve as lizard highways, based on our observation of animals moving through the interior of these logs or directly beside them. The logs also serve as a focus of social interaction, because several different individuals (of different ages) may spend much of their time within and beside the same log. In many

938 COPEIA, 2000, NO. 4 cases, the logs that were used may be very old, and some were rotting away. Judging from ax and chainsaw marks, many of these logs were generated during logging activities in the area, which last occurred in the 1970s (Forestry Commission of NSW, 1995). It would be of great interest to know more about the time scale of production and persistence of these logs. Our results offer a strong contrast with those from similar systems in Brazil where anthropogenic tree harvesting involves tree removal and, hence, does not result in the addition of logs as shelter sites (Vitt et al., 1998). Thus, the ecological impact of tree harvesting may depend upon the degree to which anthropogenic activities generate novel shelter items, as well as modify biophysical parameters (Vitt et al., 1997, 1998; Sartorius et al., 1999). Our results have several implications for the conservation of E. major and of other log-associated forest fauna. Importantly, these lizards are clearly able to take advantage of some forms of anthropogenic disturbance. In particular, activities that generate open clearings adjacent to habitats with thick cover may well enhance habitat quality for land mullets, because the resulting mosaic provides opportunities for thermoregulation close to shelter. Such areas are especially favorable if logs produced during tree felling are left in place on the forest floor. In practice, clearing of vegetation may have both positive and negative impacts (Vitt et al., 1998). Extensive cleared areas are unsuitable for the lizards, but roads through dense forest may effectively connect patches of suitable habitat from the lizards perspective. The animals are reluctant to cross such roads but may nonetheless use them as corridors of thermally suitable microhabitat to penetrate further into dense forest than would otherwise be the case. On balance, current levels of anthropogenic disturbance are unlikely to threaten the viability of this species continued existence. As noted above, perhaps the most crucial issue concerns the large logs that play such a central role in the ecology of E. major. Processes that affect the rates at which such logs are generated and broken down, and their effective lifespan as shelters for land mullets and other species, deserve further study. ACKNOWLEDGMENTS We thank B. Lewis, M. Roberts, and the staff at Barrington Guest House for their generous support and hospitality, which made our study possible. We also thank P. Whitaker for performing the surgery, M. Elphick and P. Harlow for logistics, and A. Widmer and G. Shea for useful advice. Valuable comments on the manuscript were provided by L. J. Vitt and an anonymous referee. The work was funded by the Australian Research Council. All procedures were carried out under the University of Sydney Animal Care Committee protocol L04/11 98/3/ 2863. LITERATURE CITED AEBISCHER, N. J., P. A. ROBERTSON, AND R. E. KENWARD. 1993. Compositional analysis of habitat use from animal radio-tracking data. Ecology 74:1313 1325. BARTHOLOMEW, G. A. 1982. Physiological control of body temperature, p. 167 211. In: Biology of the Reptilia. Vol. 12. Physiology C: physiological ecology. C. Gans and F. H. Pough (eds.). Academic Press, London. COGGER, H. G. 1992. Reptiles and amphibians of Australia. 4th ed. Reed Books, Sydney, New South Wales, Australia. EHMANN, H. 1992. Encyclopedia of Australian animals. Reptiles. Angus and Robertson, Sydney, New South Wales, Australia. FORESTRY COMMISSION OF NSW. 1995. Environmental impact statement (EIS), Barrington Tops area. Vol. A. Forestry Commission of NSW, Sydney, New South Wales, Australia. GIBBONS, P., AND D. B. LINDENMAYER. 1997. Conserving hollow-dependent fauna in timber-production forests. Environmental Heritage Monograph Series 3. NSW National Parks and Wildlife Service, Sydney, New South Wales, Australia. GIBSON, A. R., AND J. B. FALLS. 1988. Melanism in the common garter snake: a Lake Erie phenomenon, p. 233 245. In: The biogeography of the island region of western Lake Erie. J. F. Downhower (ed.). Ohio State Univ. Press, Columbus. GRIGG, G. C., C. R. DRANE, AND G. P. COURTICE. 1979. Time constants of heating and cooling in the eastern water dragon, Physignathus lesueurii, and some generalizations about heating and cooling in reptiles. J. Therm. Biol. 4:95 103. HILLMAN, P. E. 1969. Habitat specificity in three sympatric species of Ameiva (Reptilia: Teiidae). Ecology 50:476 481. HURLBERT, S. H. 1984. Pseudoreplication and the design of ecological field experiments. Ecol. Monogr. 54:187 211. KAVANAGH, R. P., AND K. L. BAMKIN. 1994. Distribution of nocturnal forest birds and mammals in relation to the logging mosaic in south-eastern NSW, Australia. Biol. Conserv. 71:41 53., AND R. J. TURNER. 1994. Birds in eucalypt plantations: the likely role of retained habitat trees. Australian Birds 28:32 41. KLINGENBÖCK, A., AND K. OSTERWALDER. 1999. Ecology and conservation biology of a large rainforest skink (Egernia major, Scincidae). Diploma thesis, Geobotanical Institute, Zurich, Switzerland. LINDENMAYER, D. B., R. B. CUNNINGHAM, M. T. TAN- TON, A. P. SMITH, AND H. A. NIX. 1990. The con-

KLINGENBÖCK ET AL. ECOLOGY OF RAIN-FOREST LIZARDS 939 servation of arboreal marsupials in the montane ash forests of the central highlands of Victoria, south-east Australia: III. The habitat requirements of lead-beater s possum Gymnobelideus leadbeateri and models of the diversity and abundance of arboreal marsupials. Biol. Conserv. 56:295 315.,,,H.A.NIX, AND A. P. SMITH. 1991. Habitat requirements of the mountain brushtail possum and the greater glider in the montane ash-type eucalypt forests of the central highlands of Victoria. Aust. Wildl. Res. 17:476 78. PETERSON, C. R., A. R. GIBSON, AND M. E. DORCAS. 1993. Snake thermal ecology: the causes and consequences of body-temperature variation, p. 241 314. In: Snakes: ecology and behavior. R. A. Seigel and J. T. Collins (eds.). McGraw-Hill, Inc., New York. POUGH, F. H. 1980. The advantages of ectothermy for tetrapods. Am. Nat. 115:92 112. RECHER, H. F., D. LUNNEY, AND I. DUNN. 1986. A natural legacy. Ecology in Australia. Pergamon Press, Rushcutters Bay, New South Wales, Australia. SARTORIUS, S. S., L. J. VITT, AND G. R. COLLI. 1999. Use of naturally and anthropogenically disturbed habitats in Amazonian rainforest by the teiid lizard Ameiva ameiva. Biol. Conserv. 90:91 101. SEIGEL, S. 1956. Nonparametric statistics for the behavioural sciences. McGraw-Hill, Tokyo, Japan. SHEA, G. M. 1999. Morphology and natural history of the land mullet (Egernia major). Aust. Zool. 31: 351 364. VITT, L.J.,AND S. S. SARTORIUS. 2000. HOBOs, tidbits and lizard models: the utility of electronic devices in field studies of ectotherm regulation. Funct. Ecol. 13:670 674.,P.A.ZANI, AND A. C. M. LIMA. 1997. Heliotherms in tropical rain forest: the ecology of Kentropyx calcarata (Teiidae) and Mabuya nigropunctata (Scincidae) in the Curu-Una of Brazil. J. Trop. Ecol. 13:199 220.,T.C.S.AVILA-PIRES, J.P.CALDWELL, AND V. R. L. OLIVEIRA. 1998. The impact of individual tree harvesting on thermal environments of lizards in Amazonian rain forest. Conserv. Biol. 12:654 664. WHITE, M. E. 1986. The greening of Gondwana. Reed Books Pty Ltd, Sydney, New South Wales, Australia. WILSON, S. K., AND D. G. KNOWLES. 1988. Australia s reptiles. Collins Books, Sydney, New South Wales, Australia. (AK,KO) GEOBOTANICAL INSTITUTE, FEDERAL IN- STITUTE OF TECHNOLOGY, 8092 ZURICH, SWIT- ZERLAND; AND (RS) BIOLOGICAL SCIENCES A08, UNIVERSITY OF SYDNEY, NEW SOUTH WALES 2006 AUSTRALIA. E-mail: (RS) rics@bio.usyd. edu.au. Send reprint requests to RS. Submitted: 2 Dec. 1999. Accepted: 21 March 2000. Section editor: A. H. Price.