Living in a caatinga-rocky field transitional habitat : ecological aspects of the whiptail lizard Cnemidophorus ocellifer ( Teiidae ) in northeastern Brazil

The ecology of the active forager lizard Cnemidophorus ocellifer (Spix, 1825) was studied to analyze food habits, thermal ecology and habitat use, in the Morro do Chapéu municipality (11o29’S, 41o07’W), state of Bahia, Brazil. Lizards (N = 34) were collected with rubber bands or with an air rifle and, for each individual, we recorded cloacal temperature (Tc), air temperature (Ta) (1 cm above the substrate) and substrate temperature (Ts) (to nearest 0.2°C). We registered the microhabitat used by each animal at the moment of first sight and measured its morphological variables (nearest 0.1 mm). In the laboratory, we registered the number of items of each prey category to the taxonomic level of Order, its dimensions and frequencies. Data showed that, numerically, the category most consumed was Isoptera (84.4%). Volumetrically, the diet was composed predominantly by Orthoptera (27.5%) and Isoptera (21.5%). Prey items that occur aggregated in the environment (termites) were important in the diet of C. ocellifer, a characteristic of active foragers. Males and females did not differ in the types of prey consumed. Cnemidophorus ocellifer had a mean Tc in activity of 37.6 + 1.6°C and the relationship between Tc and ambient temperatures (Ts and Ta) was positive and significant (F2,28 = 4.814; R 2 = 0.256; p < 0.05). Most lizards were first sighted on leaf litter inside shrubs (45.5%) and on leaf litter at shrub edge (42.4%). Cnemidophorus ocellifer had a relatively high mean Tc during activity, with Ts explaining most of the variation in lizard Tc.

Cnemidophorus ocellifer (Spix, 1825) is a bisexual species and is the most widely distributed species in the ocellifer group, occurring in the Cerrado of central Brazil, in the Caatinga of northeastern Brazil, and in restingas along the northeastern Brazilian coast, from Salvador (Bahia) northwards (VANZOLINI et al. 1980, DIAS & ROCHA 2007).Some aspects of the ecology of C. ocellifer have been described for populations from the Caatinga (VITT 1983a(VITT , 1995)), Cerrado (ANDERSON & VITT 1990, VITT 1991, VITT & CARVALHO 1995, MESQUITA & COLLI 2003a,b) and Atlantic Rainforest (coastal Restinga habitats) (DIAS & ROCHA 2004, 2007) biomes.However, no information regarding populations living in transitional habitats has yet been published.In this study we investigate the diet, thermal ecology and habitat use of a population of C. ocellifer at a transitional area between caatinga and "campos rupestres" (rocky fields) habitats in northeastern Brazil.

MATERIAL AND METHODS
The study was carried out in an area at about 1000 m elevation in the Morro do Chapéu municipality (11º29'S, 41º07'W), state of Bahia, northeastern Brazil.The area is at the transition between Caatinga and "campos rupestres" (rocky fields -see EITEN 1992) habitats.The vegetation is predominantly herbaceous and shrubby on rocky and sandy substrates.The climate of the region is warm and dry.During the study period (November-December 2000), mean air temperature was 29.3°C and relative air humidity was 48.9% (values obtained with a thermohygrometer, measured every hour between 7:00 and 17:00 h, under vegetation in the shade, at a height of 1 m ABSTRACT.The ecology of the active forager lizard Cnemidophorus ocellifer (Spix, 1825) was studied to analyze food habits, thermal ecology and habitat use, in the Morro do Chapéu municipality (11º29'S, 41º07'W), state of Bahia, Brazil.Lizards (N = 34) were collected with rubber bands or with an air rifle and, for each individual, we recorded cloacal temperature (T c ), air temperature (T a ) (1 cm above the substrate) and substrate temperature (T s ) (to nearest 0.2°C).We registered the microhabitat used by each animal at the moment of first sight and measured its morphological variables (nearest 0.1 mm).In the laboratory, we registered the number of items of each prey category to the taxonomic level of Order, its dimensions and frequencies.Data showed that, numerically, the category most consumed was Isoptera (84.4%).
Volumetrically, the diet was composed predominantly by Orthoptera (27.5%) and Isoptera (21.5%).Prey items that occur aggregated in the environment (termites) were important in the diet of C. ocellifer, a characteristic of active foragers.Males and females did not differ in the types of prey consumed.Cnemidophorus ocellifer had a mean T c in activity of 37.6 + 1.6°C and the relationship between T c and ambient temperatures (T s and T a ) was positive and significant (F 2,28 = 4.814; R 2 = 0.256; p < 0.05).Most lizards were first sighted on leaf litter inside shrubs (45.5%) and on leaf litter at shrub edge (42.4%).Cnemidophorus ocellifer had a relatively high mean T c during activity, with T s explaining most of the variation in lizard T c .KEY WORDS.Bahia; Cnemidophorus; diet; microhabitat; thermal ecology.above ground).In this area, five other lizard species occur in sympatry with C. ocellifer, the tropidurids Tropidurus hispidus (Spix, 1825), T. erythrocephalus Rodrigues, 1987, T. semitaeniatus (Spix, 1825) and T. cocorobensis Rodrigues, 1987, and the teiid Ameiva ameiva (Linnaeus, 1758).
Lizards were collected from 27 November to 3 December, 2000, with rubber bands and an air rifle, during their period of activity.The collected individuals were euthanized with ether, fixed in formalin (dilution of 1:9) for 48 hours and stored in 70% ethanol.The microhabitat where each lizard was first seen was recorded according to the following categories: a) leaf litter inside shrubs; b) leaf litter at shrub edges; c) open sand; d) uncovered rocks; e) rocks covered by shrubs; and f) leaf litter outside shrubs.The difference between males and females in the frequency of microhabitats used was tested using a chi-square test (ZAR 1999).For each lizard, we recorded its body temperature (T c ), together with air (T a ) (1 cm above ground) and substrate temperatures (T s ), with a quick-reading Schultheis (near 0.2°C) cloacal thermometer.For each individual (prior to fixation), we measured snout-vent length (SVL), head length (HL, measured from posterior margin of tympanum to snout) and head width (HW, taken at the angle of the jaw) using a Vernier caliper (to the nearest 0.1 mm), and body mass (to the nearest 0.001g) with an electronic balance.Morphological differences in SVL between males and females were tested using analyses of variance for one factor (ANOVA).The differences in HL and HW between sexes were tested by analysis of covariance (ANCOVA), using SVL as covariate (ZAR 1999).To estimate the activity of C. ocellifer at the Morro do Chapéu, we performed hourly transects from 07:00 to 18:00 h during one day, looking for lizards.
In the laboratory, stomach contents were identified and arthropods found were categorized to the taxonomic level of Order.Unidentified arthropod remains were grouped in a separate category "unidentified parts of arthropods" and used only in volumetric analyses.Diet composition was estimated based on relative number, volume, and frequency of each prey type in stomachs.We measured the length and width of each prey with Vernier calipers (to the nearest 0.1 mm) and estimated its volume (mm 3 ) using the elipsoid formula: 4/3(prey length/2)(prey width/2) 2 .The number of items was counted and the mean length (mm) and mean volume (mm 3 ) of the five largest items were estimated for each lizard and related to lizard morphology (head width and head length) by simple regression analyses (ZAR 1999).Due to the wide variation, all these values were log-transformed prior to analysis.A relative importance index (I x ) for each prey category was estimated by the sum of the proportional values of volume, number and frequency of occurrence of prey in the diet divided by three (sensu HOWARD et al. 1999).
Differences between males and females in prey consumption, based on mean length, mean volume and number of prey consumed (log-transformed) were tested by analyses of variance (ANOVA) and by analyses of covariance (ANCOVA) with SVL as covariate (ZAR 1999).The diference between numeric proportions of prey categories consumed by males and females was tested with a Paired t-Test (ZAR 1999).
For females we recorded the number of vitellogenic follicles in each ovary, the size and color of the largest follicle (yellow follicles were considered vitellogenic), the presence and size of corpora lutea, and the number and size of oviductal eggs.We measured the length and width of each egg and estimated its volume using the formula for the elipsoid (DUNHAM 1983).Females were considered reproductive if they had vitellogenic ovarian follicles or oviductal eggs.Mean clutch size was estimated by averaging the number of oviductal eggs and vitellogenic follicles (when eggs were absent).We considered the simultaneous occurrence of vitellogenic follicles, oviductal eggs or corpora lutea as an evidence of multiple clutches per year in the studied population.
Variations throughout the day in mean body temperatures of lizards in activity and in mean substrate and air temperatures were tested by analysis of variance (ANOVA) (including only samples with N у 2) (ZAR 1999).We analyzed the effect of environmental temperatures (T a e T s ) on lizard body temperature using a multiple regression analysis (ZAR 1999).The effect of lizard SVL and mass on body temperature was also tested by multiple regression analysis (ZAR 1999) and differences in mean body temperature between males and females were tested by analysis of variance (ZAR 1999).Additionally, the tails of all individuals were examined for evidence of previous autotomy and differences between sexes in the frequency of regenerated tails were tested using the Z-test for proportions (lizards with broken tails or that lost their tails when captured or handled were ignored).
Results of descriptive statistics are presented throughout the text as mean ±1 standard deviation, except for volume and length mean values of the five largest preys, which are presented as mean ±1 standard error of mean.
Lizards used for this study were deposited in the herpetological collection of the Museu Nacional, Rio de Janeiro, Brazil (MNRJ 13816-13849).
All data were tested for homocedasticity of variances and for normality of distributions before performing statistical analyses.
Mean volume and mean length of the five largest items and the number of preys consumed were not correlated with body measurements, except for head length which was positively correlated with prey volume (Tab.II).The number of preys consumed per lizard varied from two to 90 (x= 23.2 ± 23.1, N = 29) and males consumed significantly more prey than females (males: 31.6 ± 27.3, N = 16; females: 12.5 ± 10.6, N = 12; ANOVA: F 1,26 = 4.469, p = 0.004), but after removing the effect of body size there was no sex difference in the mean number of ingested prey (ANCOVA: F 1,1,25 = 3.732, p = 0.065).The numeric proportion of each prey category did not differ between   Mean SVL of reproductive females was 63.4 ± 4.4 mm (range 52.5-69.9mm, N = 13).Clutch size averaged 2.7 + 1.6 (range = 1-6, N = 11, vitellogenic follicles).For females bearing oviductal eggs, mean egg volume was 494.1 ± 40.4 mm 3 (range = 458.5-538.1,N = 3) for the eggs in the right oviduct and 421.4 ± 37.8 mm 3 (range = 381.2-456.3)for the eggs in the left oviduct.Only one female simultaneously bore oviductal eggs and vitellogenic follicles.
The first active C. ocellifer was observed at 7:00 h and, from then on, the number of active lizards increased until a maximum between 10:00 h and 14:00 h, decreasing afterwards The main microhabitats used by C. ocellifer at Morro do Chapéu were leaf litter inside shrubs (45.5%) and leaf litter at shrub edge (42.4%) (Fig. 6).We saw no C. ocellifer on rocky substrates.Males and females did not differ in the proportion of microhabitats utilized (chi-square = 0.111, p = 0.739).
Plants are an important source of water to omnivorous and herbivorous lizards in environments with little availability of free water (e.g.SCHALL & RESSEL 1991, ROCHA 1996a, 2000).The consumption of plant material has been reported for different species of active foraging lizards, including Cnemidophorus species (e.g.SCHALL & RESSEL 1991, MAGNUSSON & SILVA 1993, PAULISSEN & WALKER 1994, SCHALL & DEARING 1994, VITT et al. 1997, SCHALL 2000, DIAS & ROCHA 2007).At the Morro do Chapéu, flowers were the fifth most frequent item (21.9%) in the diet of C. ocellifer.This fact, associated with the presence of intact flowers in stomachs, does not support the hypothesis of accidental ingestion during capture of insects.All flowers found in the stomachs were identical (Cassia sp.).
Despite the apparent differences among sexes in the number of prey consumed, the statistics showed that when the effect of body size was removed from the analysis, no differences were detected.So, for this C. ocellifer population we can infer  that there are no differences among sexes in mean number of prey consumed.Similarly, the frequency of occurrence of prey categories did not differ among sexes, indicating that the diet of males and females are also similar qualitatively, in spite of the larger body size of males compared to females.
The mean number and mean length of prey of C. ocellifer in Morro do Chapéu were similar to those reported for the unisexual C. nativo in Guaratiba, Bahia, x number of preys = 23.0 ± 2.8, x prey length = 8.2 ± 0.4 (MENEZES et al. 2008), and for the bisexual C. deppii Wiegmann, 1834 in Nicaragua, x number of preys = 16.7 ± 5.1; x prey length = 4.4 ± 0.2 (VITT et al. 1993).However, mean prey volume for C. ocellifer in the Morro do Chapéu (Bahia) was higher than other cogeneric species, e.g. C. nativo: x = 20.4 ± 2.2 MENEZES et al. (2008) and C. deppii: x = 15.0 ± 2.6 (VITT et al. 1993).This difference can be associated to the greater consumption of orthopterans (large and voluminous prey) in the Morro do Chapéu when compared to the other cited species (VITT et al. 1993, MENEZES et al. 2008).In general, prey size varies among lizard species and body size of the lizards explains some of this variation (MAGNUSSON et al. 1985, VITT & ZANI 1998).However, in terms of number of prey, larger lizard species do not necessarily eat more prey items (VITT & ZANI 1998).In Cnemidophorus spp., on the contrary, this is not always the case.As Cnemidophorus spp.generally feed on large quantities of Isoptera, the proportion of termites influence the mean prey size per lizard species/population.In general, there is no relation between number, length and mean volume of the prey and the morphology in many species of Cnemidophorus (e.g.VITT et al. 1997, MESQUITA & COLLI 2003a,b, TEIXEIRA-FILHO et al. 2003, DIAS & ROCHA 2007, MENEZES et al. 2008), a consequence of the consumption of large numbers of isopterans, small prey with little variation in body size (MESQUITA & COLLI 2003a, TEIXEIRA-FILHO et al. 2003, DIAS & ROCHA 2007, MENEZES et al. 2008).
Mean clutch size of C. ocellifer at the Morro do Chapéu was similar to that of other conspecific populations and congeneric species in South America, suggesting that clutch size may be phylogenetically restricted for most species [e.et al. 2003 andC. jalapensis -COLLI et al. 2009).
There is sexual dimorfism in body size and head width (males had greater values than females in both cases), similar to other species/populations of Cnemidophorus in Brazil (VITT 1983a, VITT et al. 1997, ROCHA et al. 2000, MESQUITA & COLLI 2003b, TEIXEIRA-FILHO et al. 2003, DIAS & ROCHA 2007).This can be related to the fact that larger bodies and jaws may confer an advantage during intraspecific male-male agonistic interactions (e.g.VITT 1983a, ANDERSON & VITT 1990, ROCHA 1996b This similarity in tail break frequencies among cnemidophorines may be due to phylogenetic similarities in ease of tail shedding.Tail loss in lizards is a highly complex process and affects several physiological activities (DANIELS 1983).As a consequence, loss of tail (whole or part of it) can be energetically expensive, particularly so during the breeding season, because of their potential role as energy storage for reproduction (VITT & COOPER 1986).Similarly, tail loss also impairs locomotion and reduces escape speed in some species (BALLINGER et al. 1979, BROWN et al. 1995).
Body temperatures of active C. ocellifer in the Morro do Chapéu were similar to those recorded for other populations (DIAS & ROCHA 2004, MESQUITA & COLLI 2003a, VITT 1995) and also for other species of Cnemidophorus (VITT & CARVALHO 1995, TEIXEIRA-FILHO et al. 1995, MENEZES et al. 2000, DIAS & ROCHA 2004).Our data is in agreement with the idea of BOGERT (1949) and SCHALL (1977) that closely related lizards tend to have similar body temperatures even living in different habitats or geographical areas.In the Morro do Chapéu, at least the two measured environmental sources of heat (substrate and air) were associated with body temperature of C. ocellifer.Substrate and air temperatures jointly explained about 26% of the variation in the lizard body temperature, but only substrate temperature explained an additional portion of lizard body temperature.This is suggestive that direct insulation (which was not measured here) may also play an important role together with air and substrate temperature.For this population, mean activity temperature was not influenced by lizard size or body mass, similar to other species of lizards (e.g.SMITH & BALLINGER 1994, VRCIBRADIC & ROCHA 1998, KIEFER et al. 2005, IBARGÜENGOYTÍA 2005).
Lack of sexual dimorphism in body temperature of C. ocellifer can be related, among other factors, to the similarity in use of different microhabitats in the study area.Cnemidophorus lizards normally occur in environments with sandy soils and where humidity is relatively low.In these environments, they generally forage in open places such as shrub edges (TEIXEIRA-FILHO et al. 1995, MENEZES et al. 2000, MESQUITA & COLLI 2003a, DIAS & ROCHA 2007).The high frequency with which C. ocellifer in the Morro do Chapéu used the interior of shrubs suggests that this microhabitat may provide an appropriate thermal and foraging environment as well as shelter from potential predators.
We conclude that C. ocellifer at the Morro do Chapéu is an omnivorous lizard that preferentially consumes termites and ZOOLOGIA 28 (1): 8-16, February, 2011 orthopterans.Sexes do not differ in the types of consumed prey.Mean body temperature while active is relatively high, as expected for an active foraging lizard.Ecology of C. ocellifer at the Morro do Chapéu, despite representing a population living in a transitional habitat, did not differ much from previous reports for other conspecific populations.

Figure 1 .
Figure 1.Snout-vent length (mm) of adult males and females of C. ocellifer from the Morro do Chapéu, Bahia.FemaleMale

Figure 2 .
Figure 2. Activity period of C. ocellifer at the Morro do Chapéu, Bahia, during a single sunny day, with mean air temperature of 29.2 ± 4.2°C and mean relative air humidity of 50.0 ±13.9 mm.
Figures 3-5.Body temperatures recorded during activity of C. ocellifer (3) and air (4) and substrate (5) temperatures (°C) throughout the day in the Morro do Chapéu, Bahia.Dots represent observations in different days.

Figure 6 .
Figure 6.Frequency of the different types of microhabitats used by individuals of C. ocellifer at the Morro do Chapéu, Bahia, Brazil.(IS) Leaf litter inside shrubs, (SE) leaf litter at shrub edge, (SA) open sand, (RS) rocks covered by shrubs, (OS) leaf litter outside shrubs, (UR) uncovered rocks, N = 33.

Table I .
Number, volume, and frequency of prey categories in the diet of C. ocellifer (N = 32), at the Morro do Chapéu, Bahia.(I x ) relative importance index.
2 = 0.256; p = 0.016), but only substrate temperature (p = 0.035) explained an additional portion of the variation in body temperature.Body temperature was not significantly (p > 0.05) related to lizard SVL and body mass.

Table II .
Regression statistics relating lizard morphology (log-transformed) to number, mean length (mm) and mean volume of the five largest prey items of C. ocellifer at the Morro do Chapéu.