Study area

Fieldwork was conducted on islands in Chamela Bay on the Mexican coast of the Pacific Ocean, as well as nearby mainland sites in the Municipality of La Huerta, Jalisco, Mexico (Comisión Nacional de Áreas Naturales Protegidas [2008]). We sampled A. nebulosus on San Pancho Island (SPI; 19° 32′ 0.45″ N, 105° 05′ 2.61″ W; Figure 1) and at the Biological Field Station Chamela (BFSCH), Universidad Nacional Autónoma de México (UNAM) (19° 31′ 31.8″ N, 105° 03′ 44.5″ O; Figure 1). Studies of A. lineattissima were conducted on Cocinas Island (CI; 19° 32′ 8.07″ N, 105° 06.5′ 8.40″ O; Figure 1) and the mainland town of Xametla (19° 32′ 5.15″ N, 105° 04′ 9.53″ O; Figure 1).

The BFSCH population is located about 1,800 m from Chamela Bay while Xametla is located about 500 m from the coast (Figure 1). The dominant vegetation types of BFSCH are tropical dry forest, patches of deciduous forest, and desert scrub (Trejo-Vázquez [1988]). In Xametla, the major vegetation types are tropical dry forest and coastal dunes (Trejo-Vázquez [1988]). SPI is located 500 m from the mainland and has a total area of 2.4 ha, and its dominant vegetation types are tropical dry forest and desert scrub. CI is about 3 km from the mainland, has a total area of 26.2 ha, and is primarily covered by tropical dry forest and coastal dunes (Trejo-Vázquez [1988]).

Fieldwork

We conducted fieldwork during each of six time periods, three during the dry season, and three during the rainy season. Each sample period lasted 16 days, with 3 to 4 days of fieldwork at each individual site. Surveys during the dry season occurred in December 2011 and March and December 2012; rainy season surveys took place in June, August, and October 2012. In order to observe the status of lizard populations, we established three study plots of 30 × 60 m² at each site. Each plot was arranged parallel to others at the same site and separated from adjacent plots by 10 m. Using the mark-recapture method (Ramírez-Bautista [1995]), we obtained information on body size and morphometric characteristics for individuals of both A. nebulosus and A. lineattissima. At each population, we marked individuals by toe clipping technique, and the following linear measurements were recorded to ±1 mm on adult lizards in accordance with Ramírez-Bautista et al. ([2013]): snout-vent length (SVL), head length (HL), head width (HW), arm length (AL), forearm length (FOL), femur length (FL), tibia length (TL), and pelvic girdle length (PGL).

Data analysis

All morphometric comparisons between populations of each species and sex were performed only on adult lizards. Measures of body size and morphometric variables (SVL, HL, HW, AL, FOL, FL, TL, and PGL) were analyzed by sex and population (island vs. mainland); as not all morphometric variables exhibited a normal distribution (Zar [1999]), we used non-parametric Mann-Whitney U tests. Furthermore, to test if sexual dimorphism exists between males and females from both populations and species, we performed a Mann-Whitney U test to compare SVL, HL, HW, AL, FOL, FL, TL, and PGL between sexes (Ramírez-Bautista and Pavón [2009]).

To identify an allometric pattern in lizard form (e.g., dimensions of the head and limbs) with respect to SVL of both sexes and populations of each species, we used multivariate allometric coefficients obtained from the loadings of variance-covariance from principal component analysis (PCA). This analysis was performed as suggested by Jolicoeur ([1963]) and Mora et al. ([2003]), which is based on testing whether the variation explained by principal component 1 (PC1; SVL in this case) is greater than the other components or morphometric variables, because the analysis expresses the relationship between coefficients of each variable with respect to PC1. Depending on the value of the allometric coefficient of each combination (SVL vs. morphometric character), we determined whether there was an isometric or allometric relationship between morphometric variables and body size. Isometry was considered when the allometric coefficient represented values equal to 1, and allometry if the coefficient differed from 1, either positive (>1) or negative (<1; Huxley and Teissier [1936]). According to PCA and its variance-covariance matrix, we identified those variables that explain most of the variation in the shape (dimensions of the head and limbs) of both males and females from each species and population. We excluded the SVL of PC1 from hypothesis testing, since our goal was to isolate the overall effect of body size (SVL) and represent the components that show changes in body shape (Mora et al. [2003]), factors that should be subject to natural selection. To meet criteria of normality and homogeneity of variances prior to each analysis, all data were transformed to log₁₀ to fit a normal distribution (Zar [1999]). Statistical analyses were performed using Past v. 1.81 (Hammer et al.[2008]) and StatView IV (Abacus Concepts [1992]). Means are given ±1 S.E. unless otherwise indicated and with a significance level of 0.05.

Body size

PCA confirms the pattern observed in the morphometric variation between populations of males and females of A. nebulosus and A. lineattissima (Figures 2 and 3). In males of A. nebulosus from both SPI and BFSCH populations, PC1 and PC2 explained 85.0% and 5.55%, respectively, of the variation. Both components indicate that HL, AL, FOL, FL, TL, and PGL are highly correlated variables that provide the most morphometric variation (Table 3). For females, PC1 and PC2 explained 54.7% and 11.9% of the variation, respectively, and revealed that between populations, AL, FL, TL, FOL and PGL explain this variation (Table 3). These patterns are shown in Figure 2, in which males and females from SPI are distributed primarily on the right where the highest values for each component are located, and those from the mainland on the left.

For males of A. lineattissima from both CI and Xametla, PC1 and PC2 explain 76.0% and 7.54% of the variation, respectively; HW, AL, FL, TL, and PGL showed the greatest morphometric variation between populations. In contrast, for female A. lineattissima, PC1 and PC2 explained 57.5% and 17.5% of the total variation, respectively; this variation was explained only by AL, FOL, FL, TL and PGL (Table 3). Morphometric variation observed in both males and females of A. lineattissima from both populations is represented in Figure 3. Males from CI are clustered to the right of both components, whereas males from the mainland (Xametla) are clustered to the left, indicating that males with smaller morphometric dimensions correspond to those from the mainland population. For females from both populations, overlapping is extensive, indicating little morphometric difference.

Allometry

In A. lineattissima, 39% of positive allometric relationships was observed (Table 4). The AL and FOL showed positive allometry with respect to female body size from the island population (CI), whereas FOL, TL, and PGL exhibited positive allometry with respect to SVL in females from the mainland (Xametla). For males from CI, positive allometric relationships were found for HW, FOL, and PGL with respect to SVL, whereas for males from the mainland, positive allometric relationships were found for HW, FOL and PGL; AL and TL showed a marginally significant response (Table 4). Overall, 61% of the allometric relationships in A. lineattissima in both sexes and populations were negative, revealing only one isometric relationship in males from the mainland, the relationship between FL and SVL (Table 4).

Sexual dimorphism

Body size

Results of this study showed that both males and females of A. nebulosus from the island population (SPI) were larger in SVL and in other morphometric characteristics than those from the mainland population (BFSCH). A similar pattern was found in males of A. lineattissima, but not in females, which exhibited similar morphological characteristics at both populations. The latter pattern in female A. lineattissima can be explained in that the female growth rate of A. lineattissima is similar at both CI and Xametla, and females reach sexual maturity at a similar body size (Table 2); a similar pattern has been observed in other lizard species with a broad distribution (Losos et al. [2003]). Another explanation could be that the mortality rate in both populations of A. lineattissima is higher in females than in males and that females are responding to this pressure by growing quickly and reaching sexual maturity at a smaller SVL, a strategy observed in other vertebrate groups (Stearns and Koella [1986]).

On the other hand, larger body sizes in both males and females of A. nebulosus and in males of A. lineattissima from island populations as compared to mainland populations could be explained in that the insular populations have evolved to attain a large size as a function of the absence of mainland predators, an hypothesis that has been tested in other lizards from island and mainland habitats (Andrews [1979]; Lister and García [1992]). The absence from SPI of golden-orb weaver spiders (Nephila), known predators of A. nebulosus, as well as various avian and snake predators of A. lineattissima from CI (Ramírez-Bautista and Vitt [1997]; Ramírez-Bautista et al. [2000]) could be a factor influencing the evolution of larger body sizes. By reducing the number of predators, lizards can eliminate time and energy costs associated with escape behavior and therefore direct a greater amount of energy toward growth (Tinkle and Ballinger [1972]; Sinervo et al. [1991]). Another explanation could be that lizards with a large body size have a higher survival rate during periodic food shortages because they are able to store more energy as fat bodies (Michaud and Echternacht [1995]). This idea is also related to the hypothesis that lizards of a large body size (by energy storage) can better withstand catastrophic weather events such as hurricanes and storms over longer periods, events that are generally more intense on islands than on the mainland (Whittaker and Fernández-Palacios [2007]).

Allometry

Allometric analysis of males and females of both populations of A. nebulosus and A. lineattissima revealed a positive or negative trend in growth of the morphometric characters with respect to body size. In lizards, both types of allometry are considered to be adaptive responses to the selective pressures of the environment, such as predation, diet, and microhabitat (Vitt and Congdon [1978]; Meiri [2010]; Feldman and Meiri [2013]).

In both populations of A. nebulosus, the allometric relationships of HL and HW, along with anatomical structures that provide agility, such as FL and TL, were negative, which could indicate that in both populations, individuals exhibit limited movement, primarily reduced to foraging, or escape from predators. This suggests that A. nebulosus is highly sedentary, a strategy that favors optimal energy expenditure, unlike in other groups such as whiptail lizards, which have large claws, sturdy limbs, and a long tail to escape in an agile way from their predators (Aguilar-Moreno et al. [2010]). Studies of species in the genus Anolis have documented that microhabitat use and/or perch types (e.g., shrubs, brush, trunk-ground, trunk-rock, branches, and large rocks) are correlated with different ectomorphs that vary in body size, length of fingers, claws, and number of digital lamellae (Butler and Losos [2002]). It has also been proposed that selection of microhabitats by lizards is associated with different types of allometric growth of the locomotory morphometric structures; which could explain differences in A. nebulosus populations from the island and mainland (Vitt and Cooper [1985]).

On the other hand, in females of some species of Anolis, the relationships between SVL and PGL may have a stable (stable = positive) allometric pattern (Michaud and Echternacht [1995]; Butler and Losos [2002]); for example, in A. nebulosus, this morphometric characteristic (PGL) revealed a high allometric growth rate in relation to body size. In some species of Anolis, the pelvic girdle is positively correlated with egg size or volume and with the quantity and quality of food resources in the environment (Michaud and Echternacht [1995]). For example, in A. carolinensis, there is a positive allometric relationship between SVL and aperture of the pelvic girdle; in other words, at greater SVL, there is a greater amplitude of the pelvic girdle. In the northernmost populations, the environment is subject to relatively low temperatures for extended periods of time; such populations have relatively large SVL, perhaps to allow for more storage of energy as fat bodies to help survive harsh winters (see Michaud and Echternacht [1995]).

In A. lineattissima, most coefficients of allometry were positive for HW, FOL, AL, and TL, which could indicate that the anterior and posterior limbs are physiologically and/or ecologically more functional than those of A. nebulosus, enabling greater speed to escape successfully from predators and also to better explore the litter layer and ground during foraging, a behavior that has been observed in most whiptail lizards (Aguilar-Moreno et al. [2010]; Mata-Silva et al. [2013]). On the other hand, a positive allometric relationship in HW with respect to SVL in males from both populations could be related to sexual dimorphism (Aguilar-Moreno et al. [2010]). Males exhibited larger heads and jaws than females, which could be explained by male-male combat and/or divergence in prey selection (e.g., males and females may specialize on prey of different sizes).

Sexual dimorphism

According to Andrews ([1979]), sexual dimorphism in lizard species inhabiting islands is more common and conspicuous than in lizard species from mainland habitats. This asymmetry has been called ‘ecological relaxation’ and has been associated with the decreased interspecific competition and low intensity of predation characteristic of islands (Stone et al. [2003]). Ecological relaxation has been observed in some lizard species of the genus Anolis and Microlophus inhabiting in some islands of the Greater Antilles and the Galapagos; in these species, males are larger than females (Butler and Losos [2002]; Stone et al. [2003]). In this study, we found that males of A. nebulosus and A. lineattissima, from SPI and CI, respectively, were larger than conspecific females in all morphometric variables measured (Table 5). In contrast, males of A. lineattissima from the mainland were larger than females only in HL, HW, FOL, and TL. We observed sexual dimorphism just in HL for A. nebulosus from the mainland.

Sexual dimorphism of species from the island and mainland could be explained under the assumption that if resources (e.g., space and food) are limited and male populations are locally dense, then there may be intrasexual competition; lizards of large body sizes are better competitors for these resources. The hypothesis of niche divergence (Schoener [1967]; Hierlihy et al. [2013]) suggests that a degree of overlap in diet between males and females results in a strong competition for food; thus, a divergence in eating habits may be indicative of sexual dimorphism (Ramírez-Bautista and Pavón [2009]; Aguilar-Moreno et al. [2010]; Hierlihy et al. [2013]). In addition, if the sex ratio is skewed toward males, then agonistic interactions between males for access to females during reproductive season should be intense; thus, sexual dimorphism would be explained by sexual selection (Hierlihy et al. [2013]). Consequently, a large body size and head are morphometric attributes that could generate higher fitness in males, a pattern observed in gonochoristic Aspidoscelis species and in most species of Anolis from island and mainland populations (Anderson and Vitt [1990]; Losos et al. [2003]; Aguilar-Moreno et al. [2010]). Finally, if body size at sexual maturity is larger in males than in females, then males should have a higher growth rate and are faster to defend territory and reproduce successfully, which would explain why males of A. nebulosus and A. lineattissima from islands are larger and have higher growth rates than males and females from mainland populations (Andrews [1976]).

The absence of sexual dimorphism in A. nebulosus from mainland populations could be due to food resources are sufficiently abundant in the environment that competition is limited. This pattern has been observed in other species of Anolis from the Greater Antilles (Losos et al. [2003]) and in species of the genus Sceloporus (from Central Mexico; Ramírez-Bautista and Pavón [2009]; Ramírez-Bautista et al. [2013]).

In general, lizards from insular environments are larger than lizards from conspecific mainland populations. However, we still need to evaluate the role of other factors in generating differences between island and mainland populations, such as distance between islands and mainland populations, variation in reproductive characteristics, and differences in feeding habits among populations.