SINTOPY OF TWO Tropidurus LIZARD SPECIES ( SQUAMATA : TROPIDURIDAE ) IN A ROCKY CERRADO HABITAT IN CENTRAL BRAZIL

We studied the ecology of Tropidurus itambere and T. oreadicus that occur syntopically in rocky habitats of Cerrado vegetation in central Brazil during the dry season (April to September 2000). The two species are ecologically similar, but somewhat differentiated in vertical microhabitat use. The two species preferred rocky surface microhabitat. Both species demonstrated a unimodal activity pattern, with a peak between 10 and 15 h. Their diets were similar in composition and prey size. The most frequent item used by both species was ants, whereas the most important preys volumetrically were termites and ants. Small morphological differences observed between the two Tropidurus species could explain minor microhabitat divergence: T. itambere is slightly smaller, heavier, and more robust, and uses lower perches. T. oreadicus is larger, lankier, with longer extremities (tail, foreand hindlegs), and uses a larger vertical microhabitat range. These ecological differences are slight, when compared with those observed between sympatric species of Tropidurus in spatially more heterogeneous landscapes. Considering the slight ecomorphological divergence between the two Tropidurus species and their high abundance in outcrops, we suggest that interspecific territoriality is the mechanism of coexistence.

The Cerrado is the largest savanna and the second largest biome of the Neotropical region, with approximately two million km 2 .The very heterogeneous landscape ranges from grassland to forest habitats, with open vegetation predominating (Eiten, 1978).Fire and edaphic factors strongly influence cover vegetation variety.In the central Brazilian Cerrado, several species of Tropidurus co-occur in assemblages of two to four species (Rodrigues, 1987;Vitt, 1991;Colli et al., 1992;Colli et al., 2002).The large spatial heterogeneity of the Cerrado, with open and closed (forested) habitats side by side, allows sympatric colonization by arboreal and saxicolous Tropidurus species.However, even in homogeneous open habitats, like the rocky cerrado of mountain outcrops in central Brazil, syntopy of pairs of species is a common phenomenon (Rodrigues, 1987).
Considering the premise that morphological adaptations of species reflect their ecological relationships (Ricklefs et al., 1981), we compared morphology, niche dimensions, and niche relationships of two closely-related, syntopic species of Tropidurus (T.itambere and T. oreadicus), correlating morphological differences with the amount of niche partitioning observed in space, food, and time dimensions.

MATERIAL AND METHODS
Tropidurus itambere and T. oreadicus were studied on a mountain top called "Serra dos Pirineus" (Pirenópolis, Goiás; 15º51'S, 48º57'W).The mountain itself is isolated from the others in the area by forested valleys and plantations.Its landscape is a mosaic of scattered quartzite outcrops, surrounded by white sand eroded from the mountains, the elevations of which varies from 900 m to 1200 m.These outcrops are covered by patchy and xeric vegetation, from grass fields to dense shrubs containing a few dispersed trees (rocky cerrado habitat).Both species are very abundant in cerrado rocky outcrops (Van Sluys, 1993;1995;Vitt, 1993;Avila-Pires, 1995), but are extremely rare in the savannalike vegetation covered with white sands between the outcrops (sandy cerrado habitat).
We accumulated field data in 46 working days during the dry season (April to September 2000).Lizards were collected in the outcrops with an air rifle, elastic bands, or by hand, and we tagged individuals for identification.Time of collection, microhabitat use at first sighting and following the collector's approach (safe site), weather conditions (sunny, cloudy, or rainy), kind of activity (at first sighting and subsequently at the safe site), light exposure (direct sun, filtered sun, or shade) and perch height were recorded for each lizard collected.Within 30 seconds of capture, cloacal temperature was recorded by a Minipa  digital thermometer with precision of 0.1ºC.Temperatures of substrate, air (both at 1 cm above the substrate and 1.5 m aboveground) were also recorded using the same thermometer.
In the laboratory, lizards were weighted with Pesola  dinamometers (scales ranging from 0.1 g to 0.5 g of precision).Ten morphological measurements were taken: snout-vent length; tail length; length of regenerated tail if any (all to 1.0 mm) with a linear ruler; body width and height from the middle of the chest; head width at the widest point of the skull; head length from the anterior edge of the tympanum to snout tip; head height at widest point of the skull and hindleg, and foreleg lengths to 0.01 mm using Mitutoyo  digital calipers.Lizards were fixed in 10% formalin.Later, they were dissected.Sex was confirmed based on internal anatomy, stomachs were removed and placed in 70% alcohol for subsequent examination, and reproductive organs were examined.Females were considered sexually mature if they had oviductal eggs or enlarged vitellogenic follicles.Males were so considered if presenting enlarged testes and convoluted epididymides indicative of breeding activity.The snout-vent length of the smallest female and male presenting these characteristics respectively were used to estimate minimum size at sexual maturity.
All linear morphological variables, as well as dimensions of prey used by the lizards, were log 10transformed to minimize scale effects.To increase the number of available data, tail length was estimated using a linear regression model for those lizards with broken or regenerated tails.
Body size was defined as a size variable (isometric) following the procedure of Somers (1986).First, scores of a size eigenvector (isometric), defined a priori as p -0.5 , were obtained by multiplying matrix n versus p of the log 10 -transformed data, where n is the number of observations and p is the isometric eigenvector (Jolicoeur, 1963;Somers, 1986).To remove the size effect from the morphological variables, we calculated residuals from the regressions between log 10 -transformed morphological variables and body size.We performed a principal component analysis (PCA) on the size-adjusted morphological variables to examine similarities or differences in morphology between species.To verify if the differences were significant, we performed a multivariate analysis of variance (MANOVA) with the first five PCA factors.We compared body mass between species and sexes with a two-factor ANOVA on the residuals from the body size versus log 10 mass regression.We compared body size between species and sexes with a two-factor ANOVA.Only sexually mature lizards were used in these analyses.
Interspecific differences in microhabitat use at first sighting and after collector's approach (safe site), weather conditions, and lizard's light exposure were tested with Chi-square test.Significance of distribution differences for daily activity patterns and diet composition (by volume and by prey number) were verified using the Kolmogorov-Smirnov test.Divergence between the two species in vertical use of microhabitat (height above ground measured in cm) was tested with a one-way analysis of variance (ANOVA).Species differences in length and width of the largest prey in the stomachs were tested with MANOVA.Lizard's body, substrate, air temperatures at 1 cm above the substrate and air + 1.5 m aboveground were also compared with MANOVA.For each species, the relationship between the lizard's body temperature and the environmental temperatures cited above were tested by a multiple regression (Zar, 1998).
We opened lizard stomachs and contents were carefully removed and spread on a petri dish.Prey items were counted, identified to the lowest taxonomic category possible (usually order), and we measured length and width to 0.01 mm with Mitutoyo  digital calipers.The prey volume was estimated using the spheroid volume formula: We used Simpson's diversity measure (Simpson, 1949) to estimate niche breadth: where p is the proportional utilization of prey item i and n is the number of resource categories.B varies from 1 (exclusive use of a single resource type) to n (even use of all resource types).
Niche overlaps were calculated with the symmetrical overlap formula from Pianka (1973): where symbols are the same as above but j and k represent lizard species.Values approaching 0 indicate no similarity in resource use whereas values approaching 1 indicate highly similar resources.
To investigate the relationship between prey dimensions and morphological head measurements, we performed a canonical correlation analysis (Tabachnick & Fidell, 1996) between two variable groups (maximum prey length and width versus head length, width, height, and jaw length) for each species.The significance level used in the hypothesis tests was 5%.Means are given + 1 SD.All lizards were deposited in the Herpetological Collection of the University of Brasília (CHUNB).
Principal components analysis (PCA) on sizeadjusted morphological data suggested morphological differences among species (Table 2 and Fig. 1).A MANOVA applied to the first five PCA factors showed significant differences in body shape between the species (Wilks' Lambda = 0.5832; p < 0.001, N = 328).However, the first two components accumulated less than 60% of the explained variance.Positive values in the first principal component were related with larger body heights, body widths, and head heights, and inversely with smaller extremities (tail, hindleg, and foreleg lengths).Positive values in the second principal component were related with larger dimensions of the head and inversely smaller body height and width.Tropidurus itambere is more robust than T. oreadicus, having smaller tail, hindleg, and foreleg lengths.
Body and environmental temperatures were similar for both species (Wilk's lambda = 0.9863; p = 0.1892; Table 5).However, the two species showed poor, but significant differences in the relationship between body and environmental temperatures.Body temperatures of Tropidurus itambere were mostly associated with substrate temperature (R 2 = 0.316; F 1,212 = 14.59; p < 0.001).Body temperatures of T. oreadicus were significantly associated with air temperature 1 cm above the substrate (R 2 = 0.071; F 1,223 = 17.15; p < 0.001).

DISCUSSION
Morphological differences in lizard limbs and body shape are usually associated with general use of space, whereas morphological differences in head characteristics in particular suggest diet divergence.Relatively longer limbs increase the possibility of escaping predators by running, or by climbing to higher perches, while more voluminous heads permit the ingestion of larger prey (Pianka, 1969;Pianka & Parker, 1972;Moermond, 1979;Pounds, 1988;Losos, 1990;Vitt & Caldwell, 1993;Milles, 1994;Vitt et al., 2000).Like other "sit-and-wait" lizards, body shape and ecology are well related among Tropidurus species (Kohlsdorf et al., 2001).However, when populations of the same species were compared, morphology is variable between habitats (Vitt et al., 1997).
Tropidurus itambere and T. oreadicus belong to a taxonomic group that uses essentially open habitats and the ecomorphological similarity observed in the outcrops was expected (Harvey & Gutberlet, 2000).Size differences between the two Tropidurus species studied at Pirenopolis are relatively smaller than the differences found between sympatric Tropidurus species in more heterogeneous habitats (Araujo, 1987;Colli et al., 1992;Vitt, 1991Vitt, , 1993;;Vitt & Goldberg, 1983;Vitt & Carvalho, 1995).However, the relatively larger, flatter and more elongated body of T. oreadicus may provide additional advantages when disputing space in the outcrops.Finally, Tropidurus itambere is a smaller, more robust, and proportionally heavier species than T. oreadicus.
High predation pressure can prevent competition among lizard populations by reducing lizard numbers.Vitt et al. (1996) suggested that predation risk in open areas, such as rocky outcrops, can be minimized by reducing the amount of time necessary for basking in the sun, due to the high temperatures on rocky surfaces exposed to direct sunlight.Predation risk for the two Tropidurus at Pirenópolis is high, since the most frequently chosen microhabitat (rocky surface) is largely exposed to predator attack.In Pirenópolis both species are cryptic in the outcrops and their camouflageis similar.Several birds and snakes species that eat lizards were recorded in the study area.Compared with the "good colonist" T. itambere, the elongated and flattened body of T. oreadicus may increase the area exposed to direct sunlight, allowing rapid thermoregulation under high insulation.This characteristic also increases crevice use for refuge in the rocks (safe sites), reducing predation risks and overheating.Crevice use as refuge has already been recorded for T. oreadicus (Vitt, 1993), T. semitaeniatus (Vitt & Goldberg, 1983), and T. hispidus (Vitt & Carvalho, 1995).Sites with great amounts of resources, like thermoregulation perches and crevices for hiding, may be battled over by territorial Tropidurus lizards.Huey & Pianka (1981) showed that the kind of prey consumed by lizards is related with their foraging mode: "sit-and-wait" lizards eat active and mobile preys and "widely foraging" lizards eat more sedentary ones.In the present work, both active (Formicidae, Orthoptera) and sedentary preys (termites and insect larvae) were important items in the diet of both Tropidurus species.Tropidurus itambere and T. oreadicus are probably less selective and consume a larger range of prey during the dry season.
In addition, at this time their daily activity pattern was unimodal.This was not surprising since seasonal differences in diet and daily activity pattern are expected for these species (Van Sluys, 1992, 1995;Rocha & Bergallo, 1990).

TABLE 1 Morphological characteristics summary of sexually mature males and females for Tropidurus itambere and T. oreadicus from Goiás, Brazil. Fig. 1 -
Plot of unrotated factor scores for PCI and PCII on size-adjusted morphological variables for Tropidurus itambere and T. oreadicus from Goiás, Brazil.Factor I was related to larger body heights, body width, and head height, and inversely with smaller extremities (tail, hindleg, and foreleg lengths).Factor II was related with larger dimensions of the head and inversely smaller body height and width.