We quantified the time each species spent off the ground, i.e., its degree of arbore-ality, by observing individual lizards in their natural habitat. Observations were done on the same populations as those sampled for the locomotor trials. Whenever we observed an animal, we noted the substrate type (ground, leaf litter, grass, trunk, branch, or boulder or stone wall) and dimensions (height and width) of the perch it was on at first sight. An individual was subsequently scored as being "on the ground" if the substrate type was ground, leaf litter, or grass; perch height was less than 10 cm; and perch width was greater than 400 cm. We subsequently approached the lizard and noted in which direction it fled (up, down, or horizontally).
As an estimate of "degree of arboreality," we used the ratio of number of observations the individual lizards were not sitting on the ground against the total number of observations for that species. As an estimate of "proportion of observations fleeing upward," we used the ratio of number of observations fleeing upward against the total number of observations for that species.
Each individual was only observed once. Except for A. garmani, only observations on males were included in the analyses. Since we were unable to observe any undisturbed A. garmani male in the field, we used observations on females instead.
Because the values for the variables under study were not normally distributed, muscle mass-specific power output during running and jumping and all morphological variables were logarithmically (log10) transformed and the ecological variables were transformed by taking their arcsine  before statistical analyses.
To compare muscle mass-specific power output during running to muscle mass-specific power output during jumping for all 10 species, we performed a two-way ANOVA with the locomotor mode (i.e., running or jumping) and species as the factors. Two species, A. carolinensis and A. valencienni, were further analyzed in detail for two reasons: (1) They showed marked differences in muscle mass-specific power output during running and jumping (see Section 12.3), and (2) movies of both jumping and running were available, thus allowing the quantification of the two-dimensional joint angles. For just A. carolinensis and A. valencienni, we repeated the two-way ANOVA as described above. Since the analysis showed a significant locomotor mode-species interaction effect and species effect (see Section 12.3 for details), we performed one-way ANOVA (with species as factor) on the three joint angles for each locomotor mode separately.
We followed the procedure described below to test whether muscle mass-specific power output during running and/or jumping were intercorrelated and whether the variation in muscle mass-specific power output was explained by the variation in ecology (i.e., degree of arboreality and proportion of time escaping upward). In the latter case, we only used the maximal muscle mass-specific power output for each species, regardless of whether it was attained running or jumping. We refer to this variable as "maximal muscle mass-specific power output."
Because species share parts of their evolutionary history, they cannot be regarded as independent data points in statistical analyses [30-32]. Various methods and computer programs have been developed over the years, however, in which phylo-genetic relationships among different species are taken into account in statistical analyses [30-33]. In this study, we used the independent contrast approach [30,31].
We calculated the standardized independent contrasts using the PDTREE program  on the transformed means per species of muscle mass-specific power output during running and jumping, maximal muscle mass-specific power output
FIGURE 12.2 Phylogenetic tree of the 10 Anolis species used in this study. Relationships are based on mitochondrial DNA data by Harmon L.J. et al., Science, 301, 961, 2003. Branch lengths are not to scale.
snout-vent length (SVL), degree of arboreality, and proportion of time escaping upward. We subsequently performed two multiple regression analyses (backward method). In the first regression, we used the contrasts in muscle mass-specific power output during running as the dependent variable and the contrasts in muscle mass-specific power output during jumping as the independent variable. In the second regression, we used the contrasts in maximal muscle mass-specific power output as the dependent variable, and the contrasts in SVL, degree of arboreality, and proportion of time escaping upward as the independent variables. All the regressions were forced through the origin .
The independent contrast method requires information on the topology and branch lengths of the phylogenetic tree. The phylogeny of the 10 Anolis species under study here is based on a phylogenetic analysis of a much larger number of anole species by Harmon et al. , using mitochondrial DNA sequences (Figure 12.2). Branch lengths are available upon request from L. Harmon. Moreover, checks of branch lengths, using the diagnostics in the PDTREE program, did not show any significant correlation between the absolute values of the standardized contrasts and their standard deviations . Recently, the phylogenetic relationships among Anolis lizards have been reexamined, resulting in minor changes . Because, however, branch lengths have not been made available for this phylogeny, we preferred to use the phylogeny data of Harmon et al. .
Was this article helpful?