Impact of temperature on performance in two species of South African dwarf chameleon, Bradypodion pumilum and B. occidentale

Anthropogenically induced climate change is a primary concern for the continued well-being of our planet. The Intergovernmental Panel on Climate Change (IPCC) predicts that global temperatures will increase by 1–3°C over this century (IPCC, 2007). Moreover, an increase in the frequency of heat waves, extreme heat and droughts is predicted. If organisms are to persist in the face of such rapid climate change they will need to either shift their distribution to areas with appropriate climatic conditions and habitat (Parmesan, 2006; Thomas et al., 2004; Wilson et al., 2005), or adapt to changing local conditions (Hughes, 2000; Parmesan et al., 2000). Reptiles and amphibians, being ectotherms, are considered especially vulnerable to climate change as their physiology and function are dependent on variation in environmental temperature. Consequently, climate change could have a substantial impact on the distribution and long-term persistence of reptiles and amphibians, some species of which are already in decline (Araújo et al., 2006; Gibbon et al., 2000).

South Africa has the third richest lizard biodiversity globally (Myers et al., 2000) and some climate scenarios predict that this region will be markedly impacted by climate change (Beaumont et al., 2011), with an increase of 3–7°C in temperature and a 20% increase or decrease in precipitation by 2100 (Boko et al., 2007). Previous studies (Houniet et al., 2009; Tolley et al., 2009) have predicted shifts in the available niches of some South African reptiles based on different climate change scenarios. For example, a loss of suitable habitat for Bradypodion pumilum by 2080 was predicted whereas the potential suitable habitat for another chameleon, Bradypodion occidentale, may increase by 2080. However, these inferences were based only on environmental variables and species presence/absence records. Yet, the potential for species to respond to shifts in climate (e.g. temperature) is unclear and therefore it is crucial to refine predictions about the adaptive potential of species by investigating the thermal dependence of ecologically relevant traits, information that is rarely incorporated into species distribution models.

Chameleons, like other reptiles, are poikilotherms and consequently their body temperature is dependent on environmental temperature within the limits of behavioral thermoregulation. As muscle performance is dependent on temperature (Bennett, 1985), performance and behavior are also dependent on environmental temperature. To understand how temperature affects ecologically relevant behaviors, we measured a number of ecologically relevant performance traits under different temperature regimes. Performance is defined here as the ability of an animal to execute an ecologically relevant task, involving the physiological and morphological limits of an individual (Lande and Arnold, 1983). Performance traits are considered as ecologically relevant if they are involved in vital activities like foraging behavior, intraspecific interactions and defense against predators (Herrel et al., 2011; Herrel et al., 2013; Measey et al., 2009; Measey et al., 2011). The determination of the effect of temperature on performance traits allows inferences on the physiological flexibility of animals in response to potential climatic shifts. Relationships between the thermal dependence of isolated muscle and that of performance are thought to be linear (Bennett, 1985), suggesting that the underlying physiology may be driving whole-organism responses to variation in temperature.

Our study focused on two species of South African dwarf chameleons, Bradypodion occidentale and B. pumilum. Bradypodion pumilum (Gmelin 1789) is strictly arboreal and two morphs are recognized: one inhabiting woodland and the other inhabiting heathland (Tilbury et al., 2006; Tolley and Burger, 2007). Here, we used the mesic area-adapted forest morph of this species. Bradypodion occidentale (Hewitt 1935) is a semi-terrestrial species that lives in arid regions along the west coast of South Africa (Tilbury et al., 2006; Tolley and Burger, 2007) and shows adaptation to xeric environments. The specific aim of this study was to determine the temperature dependence of selected performance traits (bite force, hand and tail force, and sprint speed) for each species. We predicted that speed-related performance traits should be highly temperature dependent (Bennett, 1980), yet force-related performance should be less temperature dependent (Bergh and Ekblom, 1979; Binkhorst et al., 1977; Herrel et al., 2007; Petrofsky et al., 1981), thus differentially affecting behavior. These predictions are based on previous results (e.g. Bennett, 1985) which show that tetanic and twitch tension are largely temperature independent, whereas rate-dependent processes show a strong thermal dependence. The thermal dependence of rate-dependent processes is explained by the temperature dependence of the enzymatic reactions allowing contraction and relaxation of muscle, whereas contractile proteins and the number of cross-bridges per sarcomere that can be established are not temperature dependent (Bennett, 1985).

In addition, we predicted that these two species would respond differently to different temperatures. To examine this prediction, we evaluated preferred temperatures, temperature optima and temperature performance breadths at 80% and 90% for each performance trait in each species. Temperature performance breadths correspond to ranges of temperatures over which animals are able to achieve at least 80% (T_pb80) or 90% (T_pb90) of their maximal performance. T_pb80 is considered as the range of temperatures beyond which an animal cannot fulfill vital activities (foraging, escaping predators, etc.), and T_pb90 is considered as the range of ideal temperatures for vital activities. Given the different habitats in which the two species live, we predicted that the xeric environment-adapted B. occidentale would be able to perform better at higher temperatures than the mesic area-adapted B. pumilum. Additionally, we compared temperature breadths of each performance trait with climatic predictions to explore the impact of predicted climate change scenarios. By providing data on optimal temperatures and physiological flexibility in addition to behavioral thermoregulation, we aimed to understand how these species could potentially cope with future climate change. As the migration potential of chameleons is considered to be low, physiological flexibility in response to temperature variation is crucial if these animals are to respond to rapid changes in temperature and persist in the face of climate change.

Bradypodion pumilum and B. occidentale live in areas near Cape Town that differ significantly in their climatic and habitat characteristics, as mentioned above (Tolley et al., 2004). Ten Cape dwarf chameleons (B. pumilum) were caught in Stellenbosch near the Eerste River, and 10 Namaqua dwarf chameleons (B. occidentale) were caught at the Tygerberg Nature Reserve, Cape Town, in January. Collecting permits were provided by Cape Nature (permit no. AAA008-00009-0056). One male and nine females were captured for each species. Mean (±s.e.m.) mass and snout–vent length were, respectively, 9.77±1.02 g and 66.53±1.87 mm for B. pumilum and 15.96±1.44 g and 78.87±1.98 mm for B. occidentale. Seven B. occidentale females and one of the B. pumilum females were gravid. The gravid B. pumilum gave birth during the second week of the trials. For each animal, GPS coordinates were recorded and used to release the animals at the exact place of capture at the end of the experiments. Animals were brought back to the Kirstenbosch Research Centre, in Cape Town, and kept in Exo Terra Explorarium cages (Hagen, Montréal, QC, Canada), furnished with branches, and housed in a climate-controlled chamber set at 25°C. Cages were sprayed profusely with water once daily before animals were fed. Chameleons were fed two crickets enriched with vitamins each day following the daily experiments. On rest days, the cages were placed outdoors around 16:00 h, for 1 or 2 h, to provide the animals with natural sunlight. All the experiments were approved by the SANBI Ethics committee (Clearance Certificate no. 003/2011).

Performance was tested at five different temperatures: 15, 20, 25, 30 and 35°C, for bite force and gripping forces, and one extra temperature (40°C) for measurements of sprint speed. Before each trial, a climate-controlled room was set at the desired temperature; chameleons were placed into the room 1 h before experiments to equilibrate. After 1 h, the room temperature was adjusted if the body (cloacal) temperature of the chameleon was still different from the desired test temperature (±1°C). The cloacal temperature of each chameleon was measured before each session using a K-type thermocouple (Digital Thermometer Nicety DT804A; Shenzen AOEOM Technology Co., Shenzen, Guangdong, China). The order of the test temperatures was randomized using Research Randomizer (http://www.randomizer.org) for every performance metric. Maximal performance at each temperature was recorded and used in statistical analysis.

Running speed was tested on a 1 m long padded surface that prevented chameleons from slipping. Animals were stimulated to run maximally by tapping on the padded surface right behind them or by clapping. Lines were drawn every 25 cm and the time needed to travel each 25 cm interval was recorded using a stopwatch (see Herrel et al., 2013). Speed was calculated as the fastest speed over 25 cm (speed₂₅) and 1 m (speed₁₀₀). A run of 1 m was considered to be a session. Three measurement sessions took place at each temperature. After each session, chameleons were allowed to rest for 1 h to avoid fatigue. Performance was recorded between 09:00 h and 17:00 h.

An isometric Kistler force transducer (type 9203, Kistler, Winterthur, Switzerland) mounted on a purpose-built holder and connected to a Kistler charge amplifier (type 5058 A) was used to measure bite force (in N) (Herrel et al., 1999; Herrel et al., 2001a; Herrel et al., 2001b; Measey et al., 2011). Chameleons were manually stimulated to open their mouth by touching the side of the jaw. When five good bites per chameleon were recorded, the session was ended. Animals were given 1 h between sessions and three sessions took place at each temperature.

Tail and hand gripping forces were recorded using a piezo-electric platform (Kistler Squirrel force plate, 0.1 N). A narrow dowel (5 mm diameter) was mounted on the force platform to allow the chameleons to grip (Herrel et al., 2012; Herrel et al., 2013). The force platform was connected to a charge amplifier (Kistler Charge Amplifier type 9865) and forces were recorded at 500 Hz, transferred to the computer, and recorded using Bioware software (Kistler). Recording sessions lasted for 30 s for hand forces and 45 s for tail forces (Herrel et al., 2012). For hand force trials, animals were held horizontally above the set-up. They voluntarily gripped the dowel with their hands and were pulled in the horizontal direction until they released the dowel. For tail force trials, chameleons were held vertically above the set-up to promote voluntary tail gripping on the dowel. Next, they were pulled vertically, until they released the dowel. A low-pass filter (10 Hz) was applied to the force traces to remove high frequency noise and facilitate extraction of peak forces. The largest Z-peaks (for tail forces) and Y-peaks (for hand forces) were measured on the graph and used in the analyses.

The thermal preference set-up was composed of a wooden box (1.26×0.73×0.38 m) with six lanes. Five lights (Eurolux G230 M-infrared 275 W) were used to create a temperature gradient in which chameleons could select their preferred temperature. Room temperature was set at 12°C and the height of the lights was adjusted to obtain a gradient of 56 to 17°C. Wooden sticks were mounted near the bottom of the corridors to allow chameleons to hold on and to move back and forth. The temperature gradient within the lanes was recorded every 10 min using six iButtons (Maxim Integrated, San Jose, CA, USA), placed in the first lane, 20 cm apart. Sessions started at 09:00 h and lasted until 17:00 h. For each trial, five chameleons were put in separate lanes at 08:00 h to habituate to the set-up. Every hour, the body temperature of the chameleons was taken. Chameleons were starved for 24 h before thermal preference trails because feeding state affects temperature preference in lizards (Autumn and De Nardo, 1995; Li et al., 2010).

To establish critical temperatures, only males and non-gravid females were used given the potential effects of extreme temperature on embryonic development. As the room temperature was relatively unstable, two iButtons were placed in the room, 20 cm from the test area to record room temperature every minute during the trials.

Critical maximal temperature (CT_max) was tested in the climate-controlled rooms. First, the room temperature was set at 42°C, based on CT_max data (43°C) previously obtained (Burrage, 1973) for Bradypodion sp., and a vertical stick was used as a support for chameleons to cling to. Individuals were tested individually until they panted or until they presented signs of hyperactivity and stress (Langlois, 1902); these were considered as signs of discomfort and at this point the trials were ended. After 10 min, if none of these signs appeared, chameleons were removed from the room and tested again in a subsequent session in which the same set-up was used, but the room temperature was set at 43°C. The time was recorded when chameleons showed signs of discomfort and the corresponding temperature of the iButton was recorded and considered as the animal's CT_max. After trials, animals were returned to their cages at 25°C and allowed to rest for 1 day.

Critical minimum temperatures (CT_min) were tested in three cold rooms set at different temperatures: 10, 7 and 5°C (±2°C). Four chameleons of the same species were tested together. They were placed in the 10°C room for 30 min. Every 10 min, they were put on their back to test their righting response. When animals were no longer able to right themselves, the time was recorded and chameleons were returned to their cages. If animals were still able to right themselves after 30 min at 10°C, they were transferred to the 8°C room and, finally, to the 5°C room.

Critical temperatures were obtained using the iButtons. The mean temperature of the two iButtons at the cessation times of the experiment were calculated and used in the analysis as critical temperatures. CT_min was 7.1±2.08°C for B. pumilum and 7.7±2.74°C for B. occidentale; CT_max was, respectively, 41.5±0.13 and 42.3±0.22°C. These critical temperatures were then used to establish the temperature–performance curves.

For each individual, its maximal performance across all temperatures was considered as 100% and was used to calculate the percentage performance of the maxima at the other temperatures. Performances curves were plotted using the species means (%) of the relative performance at each temperature and the minimum convex polygon method was used (Van Berkum, 1986) as it avoids discontinuities in the temperature–performance relationships. CT_min and CT_max means, for each species, were used as 0 performance points. From these curves, temperature optima (T_opt), T_pb80 and T_pb90 were obtained (Fig. 1).

Statistical analyses were performed using R (2.15.0 version); sample size was N=10 for each species. The normality of the data was tested using Shapiro–Wilk's tests; if data were not normally distributed, transformations were applied (log₁₀ or power transformations). Effects of temperature on performance were tested for each species separately using repeated measures ANOVA. Differences in temperature performance breadths, between species and between performance traits, were tested using multivariate ANOVA for repeated measures with performance type and species as factors. The interaction between the two factors was also included in the model. Interactions were tested using one-way ANOVA. Post hoc comparisons between performance traits and temperatures were performed using pairwise t-tests and P-values were adjusted with a sequential Bonferonni correction for normally distributed data.

As the distribution of T_opt and T_pb80,min was not normal even after transformation, Friedman tests were used to compare differences between performance traits for these temperatures, and Wilcoxon signed rank tests were used as post hoc comparison tests. A sequential Bonferonni correction was used. Differences between species were tested using Mann–Whitney U-tests. Friedman tests were used to test for differences in preferred body temperature between hours, for each species, and Mann–Whitney U-tests were used to compare mean preferred body temperature between species and preferred body temperature between species for each hour. The significance level of the tests was set at 5%.

Speed was affected by temperature (ANOVA; B. pumilum: speed₂₅: F_5,45=33.92, P<0.001; speed₁₀₀: F_5,45=33.64, P<0.001; B. occidentale: speed₂₅: F_5,45=24.85, P<0.001; speed₁₀₀: F_5,45=21.29, P<0.001), with a trend for increasing speed as temperature increased. Post hoc comparisons showed differences between most temperatures, for both species (Tables 1, 2). Speed performance curves reached a plateau between 30 and 40°C. For speed₂₅, T_opt was 35.95±1.74°C for B. pumilum and 34.00±1.45°C for B. occidentale (Table 3, Fig. 2). For speed₁₀₀, T_opt was 37.05±1.11°C for B. pumilum and 33.50±1.30°C for B. occidentale (Table 3, Fig. 2).

The grip force data show differences between temperatures for both species (ANOVA; B. pumilum: hand force: F_4,36=4.02, P=0.008; tail force: F_4,36=7.09, P<0.001; B. occidentale: hand force: F_4,36=3.98, P=0.008; tail force: F_4,36=12.62, P<0.001). However, post hoc comparisons showed no difference between temperatures for hand force, in contrast to tail force, which was temperature dependent (Table 2).

Temperature dependence of bite force performance was species dependent. Whereas there were no differences between temperatures for B. occidentale (ANOVA; F_4,36=2.15, P=0.094), in B. pumilum bite force was affected by temperature (ANOVA; F_4,36=25.69, P<0.001) and showed an optimum at 25°C (Fig. 2). This peak at 25°C was present for B. occidentale as well but bite force at this temperature was not significantly different from performance levels at other temperatures.

Interactions between performance type and species were significant for T_pb80,max (MANOVA; F_4,85=3.94, P=0.005), T_pb90,min (MANOVA; F_4,85=2.58, P=0.04) and T_pb90,max (MANOVA; F_4,85=3.45, P=0.01) and thus performance effects were tested within species and species effects were tested within each type of performance.

T_opt was not different between species for each performance trait (Tables 3, 4). Performance breadths for speed₂₅, and hand and tail forces were not different between species (Tables 3, 4). There were, however, differences between species for speed₁₀₀ with B. pumilum having a higher T_pb80,min (t-test; P=0.009), T_pb90,min (ANOVA; F_1,18=7.16, P=0.01) and T_pb90,max (ANOVA; F_1,18=4.59, P=0.04) than B. occidentale. In contrast, for bite force, T_pb80,max (ANOVA; F_1,18=19.2, P<0.001) and T_pb90,max (ANOVA; F_1,18=18.3, P<0.001) were higher for B. occidentale than for B. pumilum (Tables 3, 4).

Friedman tests performed for T_pb80,min and T_opt for B. pumilum and B. occidentale showed differences between performance traits (d.f.=4, P<0.001 for all tests), and ANOVA showed differences between performance traits for T_pb80,max (F_4.40=3.6, P=0.01) and T_pb90,min (F_4,40=3.2, P=0.02) but not for T_pb90,max (F_4,40=0.9, P=0.44) for B. pumilum. There were no differences between T_pb80,max (F_4.40=0.4, P=0.8), T_pb90,min (F_4,40=1.4, P=0.26) and T_pb90,max (F_4,40=0.8, P=0.52) between performance traits for B. occidentale. Speed-related traits showed higher T_opt and T_pb minima and maxima than force-related traits (Tables 3, 5), except for T_pb90,max, which was not different between performance traits for both species. Yet, T_pb was narrower for speed than for force (Fig. 3). There were no differences among different measures of speed, over 25 cm and 1 m (P>0.05 for all comparisons), or among measures of force (P>0.05 for most of the comparisons). There were some exceptions, however, with tail forces presenting higher T_pb80,min for both species and higher T_pb90,min for B. occidentale than hand forces (Tables 3, 5). T_pb80,max of B. occidentale was higher for bite force than for tail force (Tables 3, 5). These statistical results show two main types of temperature–performance profiles: one for speed- and another for force-related performance traits.

Preferred body temperature did not vary during the day for either species (Friedman tests; B. pumilum: d.f.=8, P=0.29; B. occidentale: d.f.=8, P=0.47) and was not different between species at each hour (Mann–Whitney U-tests; P>0.1 for all comparisons). Yet, mean preferred body temperature of B. occidentale (30.59±1.39°C) was higher than that of B. pumilum (29.30±1.30°C) (Mann–Whitney U-test; W=4866.5, P=0.02).

T_opt and T_pb at 80% and 90% divide performance traits in two groups: speed-related traits and force-related traits. Moreover, temperature dependence of performance accords well to our a priori predictions; speed-related traits are highly temperature dependent whereas force-related traits are less dependent on temperature. These different performance trait profiles are consistent with studies on the influence of temperature on muscle function (Bennett, 1980; Bergh and Ekblom, 1979; Binkhorst et al., 1977; Petrofsky et al., 1981; Anderson and Deban, 2012) and whole-organism performance (Herrel et al., 2007; Anderson and Deban, 2010). Furthermore, most of the differences found in the comparison of force-related traits involved temperatures under 25°C, whereas temperature independence of force generation by muscle is known to range from 25 to 40°C (Bergh and Ekblom, 1979; Binkhorst et al., 1977; Petrofsky et al., 1981).

The T_opt of speed-related performance traits was largely above the mean temperature encountered by these species in the Western Cape, with a mean temperature of 22°C during the hottest month (February) and 11.8°C during the coldest month (August) (Schulze, 1997). This suboptimal strategy is common for species with low field active temperatures (Bennett, 1980), yet remains poorly understood. Natural selection may not favor individuals with an optimal running performance, given that chameleons rely on cryptic behavior more than fast escape by running when confronted with a predator (Stuart-Fox et al., 2008) (Stuart-Fox, in press). In contrast, T_opt of force performance traits was very close to the mean of currently encountered temperatures in the Western Cape. This result is interesting and could explain the non-optimization of speed performance traits. A previous study showed there is no trade-off between speed and gripping force in Bradypodion (Herrel et al., 2013). Based on these results, we can hypothesize that natural selection favors individuals who present optimal force performance traits, low thermal dependence of these traits, and optimum performance in the range of encountered temperatures. This observation underscores the relative importance of speed and force performance traits in a chameleon's ecology. Running is an anti-predator strategy and allows escape in most animals. Chameleons, however, have developed other strategies such as physical and behavioral camouflage. Consequently, running may be less relevant to the ecology of a chameleon. In support of this hypothesis, our results show that the mean temperature encountered by Bradypodion is not included within T_pb80 and T_pb90 intervals for both speed performance traits whereas it is included in those of force-related traits. Huey and Bennett (Huey and Bennett, 1987) also found that, in nature, reptiles are not always able to run at their maximal speed, which could explain the development of defensive behavior involving biting, which is less temperature dependent (Herrel et al., 2007). Interestingly, mean preferred body temperature for both species is included within T_pb80 intervals for each performance trait and, consequently, chameleons can perform running, biting and gripping at 80% of their maximal capacity at these preferred temperatures. Chameleons thus prefer a ‘trade-off temperature’ at which they can perform correctly over a wide range of performance traits instead of maximizing a single type of performance. Yet, when we consider the smallest temperature interval that contained at least 50% of the preferred temperatures during the preference trials (T_pref50; 30–36°C for B. pumilum and 32–38°C for B. occidentale) it becomes clear that this interval is right-shifted for both species. T_pb90 of speed-related traits is included in the T_pref50 for both species, whereas just T_pb80 of force-related traits is included in the T_pref50 for B. pumilum only. For B. occidentale, tail force temperature breadths are not included within the T_pref50, but T_pb80 of hand force and T_pb90 of bite force are. This suggests that the animals spend quite some time at temperatures away from the optima for force generation. Although counter-intuitive at first, data on the thermal dependence of feeding behavior (Van Damme et al., 1991) show that traits such as gut-passage time, energy intake, fecal output and body mass change are included in the 30–35°C interval, at least for the lizard Lacerta vivipara. This interval is near to the T_pref50 of Bradypodion and may indicate an optimization of digestive physiological processes. Although our study focused on five different performance metrics, the physiology of an animal is clearly complex and data on the temperature dependence of, for example, digestive physiology would be extremely insightful to better understand temperatures selected by animals.

The climatic change scenario A2 for 2071–2100 of Hudson and Jones (Hudson and Jones, 2002) for South Africa predicts an increase in temperature of 3.9°C in summer and 4.0°C in winter. Winter temperatures are predicted to fall within the range of temperatures currently encountered by both Bradypodion species. Predicted summer mean temperatures (26.5°C) would also fall within the T_pb80 and T_pb90 for each force performance trait of both species. However, potential differences in the adaptive potential of the two species appear when we consider predictions for the maximal summer temperature. The maximal summer temperature predicted (32.5°C) will fall outside the temperature performance breadths of tail force for both species, as well as bite force and hand force for B. pumilum. In contrast, it is included in hand force T_pb80 and bite force temperature breadths for B. occidentale. If temperatures do indeed reach these levels, chameleons might not be able to achieve adequate gripping performance. As gripping is likely more pertinent for arboreal species like B. pumilum than for terrestrial species like B. occidentale, B. pumilum will likely be more strongly affected by predicted temperature changes. Bite force is involved in predator defense and, first and foremost, in predation. As such, if temperatures reach their maxima as predicted, B. pumilum might not be able to perform vital activities; B. occidentale, in contrast, could maintain its activity at normal levels.

Interestingly, at the maximum predicted temperature in climate change scenarios, both species would perform at 90% of their maximal running performance, suggesting potential beneficial effects of climate change on running performance. Thus, Bradypodion could potentially compensate for the negative effect on bite force by running to escape predators, for example. Other strategies such as range shifts are likely important in the case of predicted changes. However, previous studies based on the same climatic scenario showed that the suitable habitat of the most probable threatened species, B. pumilum, will be reduced as well (Houniet et al., 2009), whereas the suitable habitat for B. occidentale will increase. Moreover, the migration potential of Bradypodion seems to be low (K.A.T., unpublished). The results from these previous studies combined with our results and climate change predictions highlight a potential threat on the continued persistence of B. pumilum by 2100. However, mechanisms like developmental or reversible thermal acclimation could result in a shift in thermal optima of performance traits in this species. Additional studies on the effects of temperatures encountered by females during pregnancy on the performance of their offspring are needed. Indeed, developmental thermal acclimation and reversible thermal acclimation could be potential adaptive strategies in the face of climate change, but given the results of the few studies on thermal acclimation in lizards, their potential seems to be low. Indeed, thermal acclimation is often limited or lacking entirely (Kaufmann and Bennett, 1989). However, generalists like the chameleons studied here tend to be favored in environments with predictable temperature fluctuations (Gilchrist, 1995), suggesting that these animals may show some potential for acclimation. Clearly, more data are needed to infer the true adaptive potential of Bradypodion chameleons in the face of potential climate change.

Based on our observations, the predicted increase in temperature should not affect B. occidentale performance; this prediction is consistent with the idea that chameleons that radiated in more open habitats developed physiological, morphological and behavioral adaptations to face a higher rate of solar radiation and decrease in water availability (Measey et al., in press). Our results show that B. occidentale had a higher preferred body temperature, and a shift toward higher temperatures of their temperature performance breadths compared with B. pumilum. The CT_max of B. occidentale was also higher, but given the small sample size (N=3) we were unable to test for statistical differences between species. Although B. occidentale and B. pumilum face the same daily maximal temperature, 29.4°C during the hottest day of February in the Western Cape region (Schulze, 1997), the habitat of B. occidentale is more open than the habitat of B. pumilum, providing more solar radiation and less shadow. Consequently, B. occidentale is faced with more extreme conditions compared with B. pumilum, which may explain the observed shift towards higher temperatures. It should be noted, however, that most of the B. occidentale chameleons tested in our study were gravid females and it is known that the physiological state of these females may affect their performance (Bauwens and Thoen, 1981; Cooper et al., 1990; Garland, 1985; Qualls and Shine, 1997; Schwarzkopf and Shine, 1992; Shine, 1980), behavior (Garland and Losos, 1994; Schwarzkopf and Shine, 1992) and thermal preference (Braña, 1993; Daut and Andrews, 1993; Le Galliard et al., 2003; Mathies and Andrews, 1997), which can introduce a bias in our results for this species. Furthermore, B. occidentale tested in our study were collected in the extreme southern part of their distribution area. As such, individuals inhabiting more northerly regions may experience different climatic conditions and thus could present differences in their thermal optima, thermal preferences and adaptive potential.

In conclusion, our results show that chameleons are adapted to their current habitat with shifts in the preferred and critical thermal maximum temperatures. Moreover, thermal performance curves tend to be right-shifted in the xeric environment-adapted species. These results may have important implications for the future persistence of these species under predicted climate change scenarios and may provide input data for refined species distribution models under different climate change scenarios.

We would like to thank the Tygerberg Nature Reserve for permission to capture chameleons and for their help; the South African National Biodiversity Institute and the City of Cape Town for logistical support; all the people at the Leslie Hill Molecular Systematics Laboratory, at Kirstenbosch, Cape Town, for their help and support during the trials; and students of the Funevol team for their constructive criticism during the writing of the manuscript.

 
• CT_max

  critical thermal maximum

 
• CT_min

  critical thermal minimum

 
• T_opt

  temperature optimum

 
• T_pb80

  temperature performance breadth for 80%

 
• T_pb90

  temperature performance breadth for 90%

 
• T_pref50

  interval of preferred temperature