María José Tulli, Instituto de Herpetología, Fundación Miguel Lillo, Miguel Lillo 251 (4000), San Miguel de Tucumán, Argentina. Tel.: +54 381 4230056; fax: +54 381 4330868; e-mail: email@example.com
The central tenet of ecomorphological theory holds that different ecological requirements lead to different organismal designs (morphology). Here, we studied the relationships between performance (interlocking grasping) and forelimb morphological traits in species of lizards that exploit different structural habitats in a phylogenetic context. The performance (measured by the maximum force of clinging to substrate) was measured on different substrate types. After phylogenetically informed analyses, we found that arboreal and saxicolous species showed stronger resistance to mechanical traction in all substrates when compared to generalists and sand dweller lizards. These species showed a positive relationship between forelimb dimensions (humerus length and length of claw of toe 5) and maximum force exerted, on the contrary, hand width, claw height (CH) of digits III and IV and claw length of toe 4 showed a negative relationship. In addition, we observed a partial positive correlation between CH and maximal cling force on rough surfaces, but not on smooth surfaces.
Many animal species, such as squirrels, woodpeckers and lizards, are able to climb vertical surfaces thanks to specialized features such as claws or tails. Lizards have evolved specializations (digital pads, claws) for moving effectively on heterogeneous surfaces, such as, small or large tree branches, boulders or rocks. To achieve a vertical position, the animals need to be in equilibrium by pulling their fore feet toward the substrate, while its hind feet push on the substrate. The friction between the animal and the bark must be enough to counter the upward force acting on the claws. The ability to balance over a support requires an animal maintain its centre-of-mass in line with its support. There are two basic means by which an animal can accomplish this: either it can achieve an interlocking surface with the substrate to generate a new nonvertical contact surface between itself and its support, or it can develop an adhesive or suction force between its body and the contact surface (Cartmill, 1985).
Many lizards cling by performing an interlocking grip on rocks (Biewener, 2003). This involves the development of claws that may penetrate into the surface (or grip into small cracks) of the substrate creating a new nearly perpendicular contact surface to the gripping adductor force for the digits (Biewener, 2003). This aspect was studied in several lizard species, mainly aiming to explain how arboreal lizards perform it (Irschick et al., 1996; Zani, 2000; Goodman et al., 2008). From these studies, pad area in pad-bearing lizard species (Irschick et al., 1996) and claw height (CH) and toe length (Zani, 2000) were considered the most important factors determining climbing ability. In a recent study, Tulli et al. (2009) observed that claws of climbing species (arboreal and saxicolous) are tall and sharply curved, whereas terrestrial species have long and gently curved claws. Similar associations between morphology and ecology were found in mammals. For example, arboreal squirrels have sharp and abruptly curved claws at the tip, whereas their terrestrial relatives have blunter, smoothly curved claws (Cartmill, 1985). Similar claw differences distinguish tree-climbing birds from their closest nonclimbing relatives (Bock & Miller, 1959).
Liolaemus is one of the most diverse and species-rich lizard genera in the world, comprising more than 200 species (Abdala & Quinteros, 2008; Quinteros et al., 2008), and together with Phymaturus (with more than 25 species described) and Ctenoblepharys constitute the clade Liolaemini (Schulte et al., 2003). These species use a variety of habitats and range from generalized ground dwellers to highly specialized forms that live exclusively on rock boulders or associated with loose sand (Schulte et al., 2004; Pincheira-Donoso et al., 2009; Tulli et al., 2009). This ecological variety and the knowledge of their phylogenetic relationships make these lizards an ideal system to explore the proximate relationships between morphology and function in an ecological context.
In this work, we study the interspecific variation in functional capability (performance), morphology and habitat in Liolaemus species. An historical perspective allows us to test the expected evolutionary relationship between form and function. Our main questions are the following: (i) what is the degree of phylogenetic signal of morphological features and performance in this group of lizards? (ii) what is the relationship between performance (clinging ability) and morphology? and (iii) do animals occupying different habitats differ in morphology and performance?
In addition to the Liolaemus species considered, we include seven species of Phymaturus, Diplolaemus sexcintus and Leiosaurus belli for comparison. These were selected as they occupy very distinct and restricted habitats allowing us to test the generality of the results obtained for Liolaemus across other iguanians.
Materials and methods
Twenty-nine Liolaemus species were collected during summer months of 2007–2009; together with seven Phymaturus species, D. sexcintus and L. belli (see Appendix S1 for species list). Specimens were captured during normal activity periods either by hand or by noose. Lizards were placed in cloth bags and transported to the laboratory in Bariloche, Argentina. The lizards were housed in glass terraria of (120 cm × 60 cm × 40 cm), divided in five 0.12-m lanes each. Lizards were fed ad libitum with live crickets or flower buds (depending on their diet type) and watered twice daily and were maintained in a climate-controlled room at 22 °C (±1.4 °C). Three 150 W infrared light bulbs at one end of the terraria provided light and heat. Temperature within the terrarium ranged from 44 °C to room temperature, allowing lizards to choose their preferred temperature. Voucher specimens of the species studied were deposited at the Herpetology Collection of the Fundación Miguel Lillo, Tucumán, Argentina (Appendix S1).
We measured performance as the maximum grasping force needed to detach lizards from the surface in each trial. We test substrate dependence by measuring grasping force when pulling the lizard across three different substrate types: smooth rock, rough rock and tree bark (Fig. 1). Once lizards reach their preferred body temperature, we tied a harness to their hips which was attached to a dynamometer (accuracy: 0.1 g). Each lizard was placed on each of the selected substrates and allowed to grasp the substrate with both hands. Then, the lizard was dragged horizontally at a constant speed (lizards were not jerked) and at the maximum, force was measured at the moment when the lizard not increase the exerted force. On each surface, every lizard underwent three trials, qualified as good or bad (Bauwens et al., 1995; Losos et al., 2002). We included only those trials qualified as good, during which lizards exerted a maximum effort by extending both forelimbs. Only one investigator (FBC) conducted these trials to ensure consistency. Three trials, including at least three pull-offs each, were recorded for each animal on each surface (N = 221, X = 5.81, SD = 2.43) (Appendix S2). We used the best of these three attempts for further analyses.
Before the experiments, all specimens were weighed with an Ohaus electronic balance (accuracy, 0.01 g). After experimental trials, specimens were euthanized with an overdose of Pentothal. Lizards were fixed with 10% formalin and preserved in 75% ethanol. Body dimensions were measured before preservation, with a digital caliper (Mitutoyo CD-15B; ±0.01 mm) as follows: snout-vent length (SVL); maximum body width (BW); body length (BL) measured from the base of the neck until the vent; and inter limb length (ILL) which is the distance between the fore and hind limbs; forelimbs: humerus length (HL); radius length (RL); dorsum of the hand length (DHL); dorsum of the hand width (DHW); length of digits of the hand not including the claws; and longest toes length of the foot (d3, d4, and d5) not including the claws. Following the same protocol in Tulli et al. (2009), we also calculated CH, claw length (CL) and claw curvature (CC) corresponding to digits III and IV of the hand, and 3, 4 and 5 of the foot. These toes were selected because they are the longest and presumably functionally the most important in push the lizard body on the substrate when lizards climb (Vrcibradic & Rocha, 1996; Teixeira-Filho et al., 2001). Only adult specimens of both sexes were measured and used in experimental trials. Means and SDs for morphological traits and force exerted are presented in Appendix S2.
Phylogeny and phylogenetic signal
Because a complete phylogenetic tree for all the species studied here is not available, we used a composite tree topology based on Lobo (2001, 2005); Lobo & Quinteros (2005); Abdala (2007) and Frost et al. (2001) for Leiosaurine species. We arbitrarily set the branch lengths to unity as divergence times among the different species are unclear because of the diverse nature of the original phylogenetic analyses. We ran all analyses using constant branch lengths and branch lengths transformed using an Ornstein-Uhlenbeck (OU) model of evolution. In the latter transformation, we set the d-value equal to 0.2, thus creating a topology with less structure and more star-like, or equal to 0.8, resulting in a more hierarchical topology (Blomberg et al., 2003). Because species show different degrees of relatedness as a result of common evolutionary history, they cannot be regarded as independent data (Felsenstein, 1985; Harvey & Pagel, 1991). However, although it is necessary to conduct analyses within a phylogenetic framework, whether hypotheses based on phylogenetic or conventional statistics will be considered more informative depends on whether data show evidence of significant phylogenetic signal (Blomberg et al., 2003; Garland et al., 2005; Goodman, 2006). SVL was introduced in PHYSIG after log10 transformation. All other morphological traits (i.e. CL, CH, forelimb length as well as its segments) were size-corrected (we used the residuals of log10 morphometric variable vs. log10 SVL) following the method suggested by Blomberg et al. (2003). Performance variables were corrected using log10 body mass (Blomberg et al., 2003). We used body mass as the independent variable because some of the species are heavier than other Liolaemini species at the same SVL as the case of stocky Phymaturus species compared to the more slender Liolaemus species Cruz et al., 2009, in press). We calculated the K-statistic as an estimate of phylogenetic signal relative to Brownian motion evolution considering the topology and branch lengths (Blomberg et al., 2003). K equal to 1 indicates that a trait shows an expected amount of phylogenetic signal among close relatives, whereas K-values < 1 indicate a trait is less similar among close relatives and therefore has less phylogenetic signal than expected. A K > 1 suggests there is more phylogenetic signal than expected. We tested for phylogenetic signal in all traits by a randomization test (PHYSIG, Blomberg et al., 2003).
Conventional statistical and independent contrast analyses
We ran multiple regression analyses, as the dependent variable we used the residuals of log 10 of performance on each surface on the log10 body mass. Residuals of the log10 of mean morphological variables calculated on the log10 mean SVL were used as independent variables. Many of the residuals of the morphological variables were considered redundant in the multiple regression analyses because of colinearity and were not used in the analyses. We ran multiple regressions for both conventional statistics as well as independent contrasts (IC) analyses (Felsenstein, 1985, 1988). For further analyses, IC of all morphological variables were regressed on the IC of SVL (forced through the origin; see Garland et al., 1992), except for CC (size independent) and calculated the residuals. In the case of performance variables, each IC was regressed to IC of body mass; we used body mass because for several species at the same SVL body mass was quite different, besides strength may be more related to body mass rather than SVL. We used the topology showed in Fig. 2 and mesquite program v 2.72 (Maddison & Maddison, 2009) to calculate IC.
For each test, we recorded the peak grasp force for each individual and then calculated the mean value for each species for each surface. This variable was included in subsequent analyses. The performance variables were corrected by log10 to test whether species analysed exerted different forces on each of the different substrates. Then, we used a univariate analysis of covariance (ancova), using the mass as covariate and habitat use as categorical variable. For phylogenetically based ancova, an empirical null distribution of F-statistics taking into account the phylogeny was generated using PDSIMUL and analysed with pdanova (Garland et al., 1993). For each variable, we ran 1000 simulations using a speciational model of evolution. The speciational model sets all branch lengths to one in PDSIMUL (Martins & Garland, 1991). The means and variances of the simulations were set to the means and variances of the original data.
K-values were not higher than one in any of the variables studied, including SVL (Table 2) which is an unexpected result in light of the results of Blomberg et al. (2003) for several taxa. However, randomization tests showed that CH presents significant P-values for all fingers and toes, indicating the presence of a pattern among closely related species (Blomberg et al., 2003). Branch length may influence the degree of phylogenetic signal as well as k-statistics, indicating some features may be sensitive to model specification in branch length transformations. K-values after OU transformation equal to 0.2 indicate a tendency for a better fit of star-like structure for our data set, contrary to a stabilizing selection model expected for high K-values for OU transformation higher than 0.8. However, these values are intermediate and preclude us to make stronger statements.
Table 2. Summary of analyses calculating phylogenetic signal and using constant BL and Ornstein-Uhlenbeck (OU) transformed branch lengths.
Significant values (P < 0.05) and presence of phylogenetic signal (k > 1) are indicated by bolded values.
All traits were log-transformed, and effects of SVL were removed prior to analysis as described in Blomberg et al., 2003.
The informative features under conventional analyses were CH of digits III and IV and CL of digit III and toe 3 (Table 3a). CH of digit III and CL of toes 3 were positively correlated with performance for the three contact surfaces substrates, whereas CL of digit III and CH of digit IV were negatively correlated with performance (Table 3b). When multiple regression of IC residuals of morphological variables were used as independent variables and grasping forces used as dependent variables for the three type of surfaces under experimentation (Table 3a), results were different. Partial correlations show six morphological variables were informative in all experimental trails: CL of toe 4, CH and length of toe 5, HL, interlimb length and hand width (Table 3b). HL and CL of toe 5 correlated positively with interlocking grasp force, whereas CL toe 4, CH of toe 5, interlimb length and hand width correlated negatively with performance on the three surfaces under study (Table 3a). Interestingly, CH DIII and CH DIV correlated positively and strongly with performance on rough rock and tree bark surfaces, respectively. But no manus features correlated positively with performance on smooth rock surface (Table 3a).
Table 3. (a) Results of multiple regression of residuals of morphological variables were used as independent variables and grasping forces used as dependent variables for the three type of surfaces under experimentation. (b) Results for multiple regressions of the residuals independent contrasts (IC) of performance as dependent variable and residuals of IC of morphological data.
BW, body width; CH, claw height; CC, claw curvature; CL, claw length; DHL, dorsum of the hand length’ DHW, dorsum of the hand width; HL, humerus length; ILL, inter limb length; RL, radius length.
All traits were log-transformed, and effects of snout-vent length and mass were removed prior to analysis. Bold values indicated the results on significant models and significant independent variables and their partial correlations (partial r) are given. Significant values (P < 0.05) are indicated by bolded values.
Residual cling smooth
Residual cling rough
Residual tree bark
Residual IC smooth rock
Residual IC rough rock
Residual IC tree bark
Ecological groups and performance
We found significant differences between ecological groups (arboreal, saxicolous, arenicolous and generalized terrestrial) in performance (body mass residual strength to detach lizards from surface). The significant differences we found in both for conventional (Trad) and for phylogenetically informed analysis of covariance (Phy) (Table 4). Arboreal and saxicolous species needed more force to be detached from all substrate surfaces (Fig. 3). Residuals of the force exerted by arboreal and saxicolous lizards were positive, whereas residuals of force exerted by arenicolous and generalized terrestrial were negative (Fig. 3). Our data show that arenicolous species exerted the weakest force whereas arboreal species exerted the maximum force relative to body weight (Fig. 3). These results are consistent with the expected hypothesis of higher clinging ability for those species occurring on vertical surfaces, such as rocky boulders and trees.
Table 4. Conventional analysis of covariance and phylogenetic ancova simulation results of the force exerted on four habitat types (arboreal, saxicolous, arenicolous and ground dwellers). The analysis shows the significant differences among the four groups. F and P-values are for conventional non-phylogenetic ancova and phylogenetic ancova with mass as the covariate and habitat type as factor.
Significant differences among habitat groups are shown in bold.
Bold values are significant at α = 0.05.
An assumption of performance studies is that performance measures reflect ecologically relevant aspects of the organisms under study (Losos, 1990a–c, 2009; Melville & Swain, 2000; Goodman et al., 2008). It is interesting that many other studies (Jaksic et al., 1980; Zaaf & Van Damme, 2001; Schulte et al., 2004) failed to find evolutionary relationships between some of these components (habitat use, morphology or performance). Thus, a study comparing gecko species from two habitats (climbing vs. terrestrial) that require very different modes of locomotion identified no difference in limb length, suggesting habitat-induced shifts in limb morphology are not inevitable (Zaaf & Van Damme, 2001). In general, Liolaemini lizards exhibited a clear relationship between morphology and performance considering the type of habitat they explore. For example, lizards that exploit vertical surfaces (arboreal and saxicolous) exerted greater force than terrestrial groups such as generalist and arenicolous species, suggesting that those traits considered in this study are relevant to performance.
Phylogenetic signal analyses often show importance of phylogenetic structure and branch lengths (Blomberg et al., 2003). Our results showed that for most variables (gross morphology, claws and performance) K-statistics were lower than one, exhibiting less signal than expected given the topology and branch lengths, indicating a deviation from Brownian motion because of adaptation or measuring error. Interestingly, K statistic values lower than one seem to be a common aspect in analyses that include Liolaemini lizards (Vanhooydonck et al., 2010; Cruz et al., in press). Our results are similar to Kohlsdorf et al. (2008), who studied tropidurine lizards. These authors obtained a phylogenetic signal lower than one in all the traits analysed, including SVL. Thus, these results indicate the traits examined are evolutionary labile. It has been shown that hand morphology and even SVL are evolutionary labile in tropidurine and Liolaemini lizards (Kohlsdorf et al., 2008 and this study); in contrast with other studies on these same phylogenetic groups (for example Jaksic et al., 1980; Schulte et al., 2004; Tulli et al., 2009). Apparently, species sampling and features considered have an important impact of the phylogenetic signal, at least for these iguanian lizards.
There was disagreement between conventional and phylogenetically informed multiple regression analyses. It is true that both analyses show a main effect for the several morphological features studied, but partial correlations were significant for different features depending the nature of the analysis. Interestingly, CH showed to be important when tree bark and rough rock were used in the experimental trails, even under phylogenetically based analyses (Table 3b), but did not show a correlation with clinging ability on smooth surface. Disagreement between our conventional and phylogenetically informed analysis with regard to CH may be related to the higher k-statistic values observed for this feature and that hierarchic structure (after OU transformations) showed to be affected in the different analyses. Therefore, CH may be clustered in different clades that use similar habitats (Vanhooydonck et al., 1999). Clinging by interlocking is the simplest and most familiar of the nonfrictional grip mechanisms. It is exemplified by any animal that clings to a support (Cartmill, 1974) mainly rough surfaces (such a tree bark or a rough rock) because both offer a more complex structure where claws can get a better grip by gripping adductor force of their digits (Biewener, 2003). This adduction force depends on the claws curvature and depth to which they may penetrate (Cartmill, 1974). Thus, the type of surface in terms of roughness is a relevant performance factor to consider, which also explains the significant correlation between the IC of each digit CH and CL. In addition, claw morphology of digit III and IV was positively correlated with performance on rough surfaces. This result is consistent with our expectations, because lizards use their hand claws to exert the interlocking grasp. The importance of digit III and IV in locomotion have been highlighted by Teixeira-Filho et al. (2001), who stressed that digit IV is mainly used as support and digit III seems to be the one driving clinging. With regard to CL of toe 3, the common location of toe 3 during clinging (Fig. 1) allows us to infer that it is being used as support, that is having a longer claw would increase the stability of the lizard during performance.
Biomechanical predictions posit that lizards that have some degree of vertical use of habitat tend to have shorter limbs (relative to SVL) and shorter and heavy distal limb segments, to bring the centre-of-mass closer to the substrate and broaden the plane of support (Van Damme et al., 1997). Most of the saxicolous Liolaemini live in volcanic and basaltic tablelands in Patagonian Argentina (Lobo & Quinteros, 2005). These plateaus are rocky with deep crevices that are used as permanent or occasional refuges by Phymaturus or Liolaemus species, where these lizards are frequently observed in a vertical position. Species can often be found living in sympatry and often live beneath rocks. The challenge that this particular milieu represents for the lizards is solved by having similar features, because all of them have not only longer forelimb, but also higher claws, and shorter interlimb length, traits also shared with arboreal species. Finally, both ecological groups exhibit significantly better clinging performance than generalists and sand-dweller Liolaemini species.
In Caribbean Anolis lizards, longer hind limbs and shorter forelimbs are associated with evolution of performance traits, such as, clinging, running and jumping (Losos, 1990a–c; but see Irschick et al., 2005). On the contrary, saxicolous scincid lizards use longer forelimbs to increase clinging ability using an increased base of support and frictional force, improving contact on irregular surfaces (Goodman, 2006; Goodman et al., 2008). Our analysis of forelimb length in a phylogenetic context showed that arboreal and saxicolous species present a positive relationship between the interlocking grasping force and proximal segment of limb length (HL). Thus, arboreal and saxicolous species have a longer humerus, in agreement with Goodman (2006). It is possible that the controversies in these results are explained by considering that the relevance of limb size is small compared to other adaptations. It has been argued that the presence of ‘key adaptations’ may free organisms from some of the constraints imposed by the environment (Zaaf & Van Damme, 2001). It is possible that environment may have constrained claw morphology and conversely limb size show a relaxed response in different lineages.
Having a narrower hand is shown here to be important and positively related to strength. Liolaemini species which are harder to detach live on vertical surfaces and tend to have narrow hands (Table 3b). The hands of most lizards show a tendinous palmar structure with a large palmar sesamoid, and in lizards with this condition, the hand moves as a single rigid functional unit (Abdala et al., 2009). The presence of a large sesamoid bone in the tendinous flexor plate may prevent buckling of this tendon (Abdala et al., 2009) and tend to maintain a flattened hand. It seems that to maintain the hand flat is not as important for climbing. In ground dwelling lizards, the rigid unit of the hand may be beneficial during terrestrial over-ground locomotion as it allows the animal to effectively use its entire hand as a pivot (Abdala et al., 2009). Remarkably, Grizante et al. (2010) found that tropidurid species with narrower foot soles were associated with a frequent use of rocks and branches. Likewise, they found that species that utilize trunks exhibit longer femora. The similarity of our results and those from Grizante et al. (2010), with respect to the feet in relation to the hands, suggest that both fore and hind limbs may be functionally constrained in tropidurine and Liolaemini lizards.
A strong correlation was found between shorter bodies, as expressed by a lower interlimb length, and clinging force. This result is consistent with observations made on scincids by Goodman et al. (2008). For these lizards, shorter bodies are associated with saxicolous habits. It is possible that interlimb length is a trait related to sexual dimorphism (Cei et al., 2003; Goodman, 2006), where a relatively longer distance between fore and hind limbs concurs with a larger volume for egg bearing; however the relationship with performance remains unclear for Liolaemini.
In conclusion, our data suggest that a relationship among morphology, performance and habitat use is present in Liolaemini lizards. Several traits were adaptive with arboreal and saxicolous Liolaemini exerting more strength than other Liolaemini that exploit horizontal substrates. Further study on the kinematics and biomechanics of these morphologically conservative lizards are needed to better understand the complex relationship between their morphology, physiology and environment.
We are very grateful to C. Abdala for his invaluable help in the field and suggestions for this study, and to S. Quinteros, J. C. Stazonelli, G. Scrocchi for help in the field. We appreciate the comments of D. Irschick and J. A. Schulte on early versions of the manuscript. M.J.T. was supported by a Consejo Nacional de Investigaciones Científicas y Tecnológicas (CONICET) fellowship. Grants CONICET PIP 112-200801-00225; PIP-CONICET 6347 to V.A. and 6287 to F.B.C. and PICT 01205 to FBC partially supported this work.