Iheringia Thermoregulation in the Andean lizard Anolis heterodermus ( Squamata : Dactyloidae ) at high elevation in the Eastern Cordillera of Colombia

Low thermal quality environments, such extreme latitudes or high elevation regions, are highly expensive for reptiles in terms of thermoregulation. Thus, physiological adaptations or behavioral adjustments to live in these habitats have evolved in some species. Anolis heterodermus (Duméril, 1851) is an anole lizard that lives at high elevations in the Andes region. In this paper, we attempted to elucidate the thermoregulation strategy of a population of this species from the eastern cordillera of Colombia during wet and dry seasons. We measured body temperatures (Tb), operative temperatures (Te) and preferred temperatures (Tpref). Based on these data, we obtained accuracy (d̄b), environmental thermal quality (d̄e) and effi ciency of thermoregulation (E) indexes. There were no signifi cant diff erences of Tb or d̄b between seasons, sexes, ages, and for Tpref between sexes or ages, but we found diff erences in Te and d̄e between seasons. The indexes suggested high thermoregulatory accuracy, low thermal environment quality and indicated that A. heterodermus was an active thermoregulator in both seasons. Broad ranges of Tb and the species association with microhabitats with high solar radiation suggest eurythermy and heliothermy. Anolis heterodermus lives in a low thermal quality habitat, using exposed perches, which seems the most effi cient thermal microhabitats. We concluded that A. heterodermus performed behavioral adjustment for compensating seasonal variation in the environmental thermal costs.

PALABRAS-CLAVE. Calidad térmica, exactitud termorreguladora, variación estacional, heliotérmia. Thermoregulation in lizards and other reptiles comprises a set of physiological and behavioral phenomena closely related to the environment and to any several aspects of the biology of these ectotherms (Huey & Stevenson, 1979;Huey, 1982). Active thermoregulation imply the selection of microhabitats that provide optimal or nearly optimal temperature required by individuals for performing their biological activities in an eff ective way (Pianka, 1986;Pianka & Vitt, 2003). This strategy involves a series of physiological (reproductive state, body size, sex, and in general their physical condition) and ecological (predation or competence by thermal resources) constraints (Adolph, 1990;Smith & Ballinger, 2001). Thus, when costs associated to active thermoregulation are too high, organisms adopt a thermoconformer strategy, which consists in following environmental temperature, which could be far from the optimal temperature required for their optimal performance (Huey, 1974(Huey, , 1982Huey & Slatkin, 1976).
High altitude and extreme latitude environments impose important challenges for thermoregulatory activities, due to the extreme low temperature and/or high thermal variance that is typical in these environments (Huey & Webster, 1976;Hertz & Huey, 1981;Hertz, 1981;Ibargüengoytía et al., 2010). Very low environmental temperature, relative to the preferred and/or critical temperature of individuals would decrease the available time needed for organisms to effectively basking and thermoregulate. Thus, lizards from highland environments could face the effects of low temperatures using two strategies as extremes of a continuum: by behavioral adjustments, such as searching more exposed and sunny microsites and reducing activity time (Hertz & Huey, 1981;Hertz, 1981;Gvoždík & Castilla, 2001;Gvoždík, 2002) or by physiological adaptation, such as decreased in the critical temperature minimum values to hold out cold environments (Muñoz et al., 2014). Besides, based on data about thermal biology, Sinervo et al. (2010) hypothesized that high elevation reptile species are particularly threatened by climate change, due to their restricted distribution, which would decrease progressively as temperature increase, and as these environments are colonized by competitors or predators from lowlands, pushing up highland species toward an "endless road", at higher elevations.
Both thermoregulatory strategies have been reported in Anolis lizards, which is a Neotropical and highly diversified genus that has successfully colonized many different environments (Losos, 2009). Species of this genus has been used as model organisms to design a methodology to characterize thermoregulatory strategy in reptiles (Hertz et al., 1993); a protocol widely used since then (Díaz & Cabezas-Díaz, 2004;Blouin-Demers & Nadeau, 2005;Hitchcock & McBrayer, 2006;Row & Blouin-Demers, 2006;Herczeg et al., 2008). Studies on Anolis have showed that thermoregulatory strategy varies among species, elevations and seasons, revealing high behavioral and physiological plasticity for dealing with the challenges imposed by the environment (Huey & Webster, 1975;Hertz & Huey, 1981;Hertz, 1981;Hertz et al., 1993;Muñoz et al., 2014).
We carried out this study from April to March 2014, from September to November 2014, and in February 2015. Body temperature data were collected from 9:00 -16:00 h. Body temperature (T b ) was recorded 30 s after each individual was captured by inserting a K thermocouple connected to a digital thermometer (SE ± 0.6 °C) in the lizard's cloaca. In addition, we recorded their snout-vent length (SVL), and the sex of adults, considering as adults individuals with SVL > 55 mm (Miyata, 1983).
In order to obtain operative temperatures (T e ), we used six empty artificial models (green polyvinylchloride [PVC] pipes, 100 mm length, 127 mm diameter) connected to an external temperature data-loggers (Adolph, 1990;Sinervo et al., 2010). Data-loggers recorded T e each minute for one week during wet season in 2014, one week during dry season in 2014, and four weeks during the dry season in 2015; only data from 9:00 to 17:00 h were used for the analyses, considering the period of activity previously recorded and literature (Moreno-Arias & Urbina-Cardona, 2013). Models were distributed in potential lizard microhabitats (branches randomly selected between 0 and 2 m), three under direct sun exposition and three under shaded conditions. Models were previously validated with an adult male lizard (SVL=68.4 mm, 6.5 g). In order to achieve the calibration, lizard, PVC pipe, copper pipe, syringe with water and empty syringe models were kept in a glass terrarium (200 x 400 x 400 mm). Then, we changed the temperature in the terrarium using one 200 W bulb and cooling packs, which were alternated each 10 min during temperature recording. Lizard and models temperatures were recorded for 2.5 h using an USB data-logger, with ultrafine thermocouples adapted to the cloaca of the lizard and inside the models with a micropore tape. PVC model was selected based on the best correlation with lizard body temperature (R 2 = 0.903, n = 300, p < 0.05).
For laboratory phase, we used a subsample of 33 individuals (10 adult males, 11 adult females and 12 juveniles) from the field capture. Lizards were housed in separated terraria (450 x 450 x 450 mm) with natural light conditions. All lizards we provided water and two adult flour beetles (Tribolium castaneum) ad libitum. In order to estimate the preferred temperature (T pref ) interval, we exposed lizards to a vertical thermal gradient (1000 x 500 x 500 mm) divided in four tracks (one track per individual) of 1000 x 250 x 250 mm ( Fig. 1). We used a vertical thermal gradient because our study model is a tree lizard that commonly use vertical branches as perches ( was created with 200 W bulbs in the upper extreme and cooling gels at the base. We provided tree upright branches of 1 m height and perch diameter, according to species perch use (16 + 2.8 mm; 13.4-20 mm diameter) inside each vertical sections of the gradient. This perch device allowed lizards to move up and down (Fig. 1).
Additionally, since tree branches provided to record T pref in these populations might include perches with different diameters, we posteriorly evaluated possible influence of perch characteristics (diameter, slope and surface texture) on lizard T pref . For this we recorded data of adult individuals from two localities: Chicaque Natural Park, municipality of San Antonio del Tequendama (4°36'55.18"N, 74°18'44.14"W, 2,200 AMSL, n = 17) and Ecological Park Matarredonda, municipality of Choachi (4°33'36.07"N, 73°59'58.92"W, 3,350 AMSL, n = 16), in Cundinamarca, Colombia, in April and May 2016. We compared T pref of lizards exposed to thermal gradient and using different perch diameters (10 mm, 20 mm diameter and tree branch or heterogeneus diameter), perch surfaces (tree twig branches vs. balsa wood branches covered with angeo surface), and perch slope (vertical vs. horizontal gradient). These comparisons were performed among perches within same locality, and since we did not observe any differences of T pref recorded under these different perch characteristics (see results), we pooled data per locality and compared T pref among Chicaque and Matarredonda localities and the three localities: Chicaque, Tabio and Matarredonda. Kruskal-Wallis and t tests were used in these comparisons.
Ultrafine thermocouples, connected to an USB datalogger, were adapted to the cloaca of each lizard with the help of micropore tape. Preferred temperature was recorded each minute during two hours. Each lizard remained in the gradient for 30 minutes before to temperature recording for acclimation to experimental conditions. Lizards were ventrally marked with a temporal number using a permanent marker pen and released in the field after preferred body temperature estimation. Preferred temperature range was estimated for each individual including 25-75% interquartile data (Hertz et al., 1993).
We estimated three biophysical indexes based on T b , T e and T pref according to Hertz et al. (1993): index of thermoregulation accuracy (d̄b), environmental thermal quality (d̄e), and thermoregulatory efficiency (E). Differences between T b or T e and T pref (d b and d e , respectively) were estimated with the 25-75% quartiles of all T pref data as follows: if T b and T e are lower than T pref , d b and d e are the deviations between T b or T e and 25% T pref quartile value; if T b and T e are higher than T pref , d b or d e are the deviations between T b or T e and 75% T pref quartile value; finally, if T b or T e are inside 25-75% quartile T pref values, d b or d e are equal to zero. Average values of individual´s d b and model's d e were considered as indexes for the population; values of d̄b and d̄e closer to zero would correspond to a high accuracy of thermoregulation and environment thermal quality, respectively. Thermoregulatory efficiency index E was estimated using the formula (E = 1 -[d̄b/d̄e]); values closer to 1 would correspond to active thermoregulation, those closer to zero would correspond to thermoconformism, while negative values would be interpreted as individuals avoiding good quality thermic microsites due to high costs imposed by other ecological pressures (Hertz et al., 1993;Blouin-Demers & Nadeau, 2005).
Variation of body temperature (T b ) and accuracy of thermoregulation (d̄b) was evaluated between sexes in adults, age stage (adults vs. juveniles), and seasons (wet vs. dry). Additionally, we compared T pref between sexes in adults and seasons. We previously analyzed if T b and T pref were related to body size (SVL), through linear regression. If significantly regressed, residuals of the regression were used to compare these variables between adults and juveniles. Data from all individuals were included in indexes estimation, since values for juveniles and adults were not statistically different (see results). A single averaged value of T pref per individual was used in all tests. Operative temperatures (T e ) and environmental thermal quality (d̄e) were compared between seasons. In addition, operative temperatures were compared between microhabitat types (sunny vs. shaded microhabitats). Mann-Whitney and Student's t test were used in these comparisons. Some data that could not meet parametric assumptions were normalized using square root transformation allowing the use of parametric tests. When the transformation was not sufficient to normalize data non-parametric tests were performed. Statistical analyses were performed using StatSoft, Inc. (2007) STATISTICA, version 8.0 and graphics were made using SigmaPlot (Systat Software, San Jose, CA).

DISCUSSION
Anolis heterodermus exhibited an average T b lower than other anole species distributed under 1,130 m AMSL (Sinervo et al., 2010, supplementary material), and very similar to the T b recorded for species distributed at higher elevations (1,130-2,200 m) in Puerto Rico and La Hispaniola, such as A. cristatellus, A. gundlachi, A. roquet and A. cybotes group (Hertz & Huey, 1981;Hertz, 1981;Hertz et al., Fig. 2. Frequency of body (T b ) and operative (T e ) temperatures during wet and dry season in Anolis heterodermus (Duméril, 1851). Average values are shown by black arrows. Striped area corresponds to averaged preferred temperature (T pref ) interval for both seasons.
Tab. II. Preferred temperature (T pref ) range, index of thermoregulation accuracy (d̄b), environmental thermal quality (d̄e), and thermoregulatory effi ciency (E) at each season and pooled data of both seasons.  Hertz et al., 1993;Lara-Resendiz et al., 2013b;Woolrich-Piña et al., 2015), but it is very similar to T pref recorded in A. gundlachi (24.3-26.1°C) at 1,130 m AMSL in Puerto Rico (Hertz et al., 1993). This can be explained since A. gundlachi is a thermoconformer species that has low operative temperature and is distributed in a high quality habitat. It suggested that it has physiological adaptations to high altitude (Hertz, 1981;Hertz et al., 1993). Apparently, highland Anolis species has low T b 's contrast to lowland species (Sinervo et al., 2010, supplementary material;Muñoz et al., 2014). Similarly, Sceloporus graciosus, at 2,580 m AMSL has a lower T b than its conspecifi c S. occidentalis, at 1,250 ASML. and both species at 2,230 m ASML have similar T b (Adolph, 1990). In contrast, Marquet et al. (1989), observed similar T b among four species of Liolaemus lizards distributed through an altitudinal gradient. However, at intraspecifi c level, T b is similar or slightly low despite altitudinal changes in several Anolis lizards studies (Hertz & Huey, 1981;Hertz, 1981Hertz, , 1992Hertz et al., 1993;Muñoz et al., 2014). Other lizards like Podarcis tiliguerta, Psammodromus algirus, Sceloporus jarrovi and Zootoca vivipara also exhibit the former trend (Van Damme et al., 1989;Van Damme et al., 1990;Díaz, 1997;Gvoždík, 2002). Similarly, at high latitudes Phrynosoma douglassi populations have low T b with respect to other populations (Christian, 1998).
In the other hand, inter-and intraspecifi c T pref in lizards is highly conservative (Huey & Bennet, 1987;Van Damme et al., 1990;Hertz et al., 1993;Díaz, 1997;Labra, 1998;Gvoždík & Castilla, 2001;Gvoždík, 2002;Medina et al., 2009). However, interspecifi c variation in T pref has been observed in Australian geckos and some Chilean Liolaemus lizards (Angilletta & Werner, 1998;Labra, 1998), as well as variation in Takydromus septentrionalis over a latitudinal gradient (Du, 2006). In A. heterodermus T pref seems highly conserved between localities, sexes, ages and seasons; however, we found that individuals from Tabio exhibited wide range of T pref (19.1-30.2°C, = 24.6 + 3.2°C). High individual variation in T pref is not related to heterogeneity in perch characteristics (diameter, slope or surface) in the thermal gradient. It has been observed that variation in perch selection occurs in Anolis species, and that use of perch with diff erent characteristics aff ects individual performance in escape behavior (Scott et al., 1976;Losos & Irschick, 1996); however, it seems that individuals of A. heterodermus in laboratory conditions choose temperature rather than perch characteristics. However, we do not know if perch selection occurs in the field, and if this selection is driven by temperature or other perch characteristics. Then, we cannot explain what determines this wide plasticity in T pref between individuals of this population.
Thus, lizard thermoregulation at high altitudes could be explained by two non-mutually exclusive hypotheses: (1) by behavioral responses to the variation in environmental temperature to compensate the effects of elevation on the thermal environment (Hertz & Huey, 1981;Hertz, 1981Hertz, , 1992Hertz et al., 1993;Smith & Ballinger, 1994;Gvoždík, 2002) and (2) by physiological adaptation to low temperatures at high elevations (Hertz, 1981;Vidal et al., 2008;Ibargüengoytía et al., 2010;Muñoz et al., 2014). Both hypotheses are supported by intraspecific evaluation of T b and T pref in populations at different altitudinal distribution in Anolis (Hertz & Huey, 1981;Hertz, 1981Hertz, ,1992Hertz et al., 1993;Muñoz et al., 2014). However, we did not evaluate if Anolis heterodermus thermal strategy varies through its elevational range, but given that this species occupies localities above 1,600 m ASML, it would be interesting to examine the hypothesis of locally adaptive thermal strategies (behavioral or physiological) over an altitudinal gradient.
In contrast to other lizard species (Hertz et al., 1993;Christian & Bedford, 1995Díaz & Cabezas-Díaz, 2004), T b in Anolis heterodermus did not vary between seasons, probably due to the high within variation observed among individuals, and consequent wide T b range (16.6-31.9°C). This observation suggests that A. heterodermus behaves like a eurythermic organism, with a broad range of optimal temperature. However, this hypothesis requires further research through evaluation of performance and temperature relationship (Van Berkum, 1986;Huey & Bennet, 1987). Alternatively, the explanation might be related to behavioral adjustments in wet season to compensate ecological costs for low quality thermal environment (Christian & Bedford, 1995Díaz & Cabezas-Díaz, 2004), which is consistent with our findings.
Seasonal variation of operative temperature of Anolis heterodermus is similar to the results reported in previous other studies, where lowest T e were observed during the winter periods or wet seasons (Hertz et al., 1993;Christian & Bedford, 1995Díaz & Cabezas-Díaz, 2004). Operative temperatures were usually below T b , revealing the poor thermal quality of the environment (d̄e) for this species, particularly during the wet season. However, the variable T b observed within each season was always inside the T pref range, even during the wet season. Thus, accurate thermoregulation (d̄b) seems to have occurred in this species in spite of the low thermal quality of the environment. This suggests that anole lizards from this population exhibit behavioral adjustment to maintain T b close to its T pref range, particularly during the wet season. A similar strategy was described in Podarcis melisellensis, P. murallis and Phyllodactylus bordai (Grbac & Bauwens, 2001;Lara-Resendiz et al., 2013a). These results reveal that in contrast with the cost-benefit model for thermoregulation proposed by Huey & Slatkin (1976), A. heterodermus exhibit active thermoregulatory behavior even when the cost are higher (low quality environment). These results agree with those of Blouin-Demers & Nadeau (2005), which included several squamates; these authors suggest that if lizards do not behaviorally adjust in these high cost environments; their survival probabilities would be reduced.
Despite this low thermal quality, values of T e closer to T pref were observed in models exposed to sunlight, especially in midday hours, in agreement with the thermoregulatory strategy and activity pattern found in this species, and showing a heliothermic behavior for it. This also suggests that appropriate thermal microhabitats for A. heterodermus, consist of exposed perches at bordering vegetation and high stratum in the inner of each patch (1-2 m) (Moreno-Arias et al., 2010;Moreno-Arias & Urbina-Cardona, 2013).
High elevation species exhibit higher extinction risk under climate change scenarios, and such vulnerability is usually assesed when observing T e values over T pref during most of the day (Sinervo et al., 2010). However, in this studied population of A. heterodermus, operative temperatures (T e ) were mostly below T pref through the day, suggesting that at least in this population, an increase in air temperature (which would affect T e temperatures), would not exceed their thermal physiological threshold. Thus, it would not be immediately affected by climate change. However, this result must be taken with caution. This species should be evaluated using ecophysiological models of extinction risk, as proposed by Sinervo et al. (2010) for a more reliable conclusion about threatening.
In conclusion, Anolis heterodermus can be considered as a heliotermic, potentially eurythermic lizard. It inhabits a low thermal quality environment and exhibits active thermoregulation. Thus, it had to adjust behaviorally to compensate seasonally variation in environmental thermal costs. Evaluation of thermal ecology of this species through an altitudinal gradient and the description of its performance vs. temperature relationship would provide a clearer panorama to estimate the potential impact of environmental temperature increasing on Anolis heterodermus populations, as expected in the coming years due to global warming.