*Morphological Data*

Males were significantly larger than females in *D. agassizi*, *D. princeps*, and *N. townsendi* (Table 2). Most of the morphological variables (SVL, head, and limb measurements) are likely the products of having a larger body size and therefore cannot be characterized as independently evolving traits, all three species unquestionably display sexual size dimorphism, similar to solitary anoles in the Caribbean (Table 3). *Norops medemi* on the other hand displayed no significant differences between sexes with the exception of males having one more row of toe lamellae on average. However, sample sizes for this species were very small (*N* = 5 for each sex). Because of small sample sizes for *D. chocorum* and *N. parvauritus*, and only one female measured for *D. gorgonae*, those three species are omitted from further morphological analysis.

**Table 2.** Average morphological measurements by sex for four species of Pacific Island anoles. (SVL = snout-vent length, # lamelle = # of lamellar rows). *P*-values from pairwise t-tests between the sexes are listed below the means. Standard deviation is included for each mean value in parentheses. Significant values after Bonferroni correction are bolded and italicized.



**Table 3.** Average SVL by sex for solitary species of *Dactyloa* and *Norops*. Standard deviation is provided when available.

Comparative sample sizes from the Caribbean were small for solitary relatives to *D. agassizi* and *N. townsendi* without wide-scale population data available but serve as an exploratory indication of variation within and between species. Both sexes of *D. agassizi* are considerably larger than their Caribbean counterparts, although the magnitude of sexual size dimorphism is similar to *D. luciae* (and also the distantly related *N. concolor*). *Norops townsendi* displayed a lower magnitude of sexual size dimorphism than any other species of solitary *Norops*, although the di fference was still significant (Tables 2 and 3).

The four populations of *N. townsendi* displayed significant di fferences among each other. Two of these populations (Chatham Trail and Playa) showed few di fferences except for Playa individuals being more massive with longer hindlimbs. Mass could be attributed to di fferences in prey availability between the sites, so we also re-analyzed our populations with these two sites combined to treat as a single population (hereafter referred to as the "main" population). In all morphological measurements, Cerro Yglesias individuals were significantly smaller than the main population, while Manuelita individuals were significantly larger with a greater number of toe lamellae (Table 4). The main population occupied perches >1 m higher than either the cerro or the islet, which may provide an explanation for the divergence in morphological evolution among the populations.

Perch height, diameter, type, and position were di fficult to evaluate objectively for *D. agassizi* because the surface of the island consists of bare rock with nothing to perch upon except occasional boulders and pebbles. We observed some individuals perching on boulders, but most were on the ground or on pebbles, rendering a perch height mean of 1.5 m and perch diameter that was not quantifiable (often individuals were on the ground); perch type could only be described as rock. Alternatively, perch height, diameter, type, and position were measurable for *N. townsendi* because Isla Cocos had a variety of available perch heights and types. *Norops townsendi* was observed throughout the island on everything from sand to branches well within the canopy, making averages or generalizations di fficult. Canopy height extended well over 20 m (the upper limit of our measuring abilities) and animals were often observed as high, or within the canopy. The aforementioned lack of samples for two of the Isla Gorgona species, inhibits our ability to evaluate habitat use across the local anole community, especially since the low number (1) of *D. chocorum* and *N. parvauritus* observed is likely due to their preferred niche in the canopy. The forest of Isla Cocos is similar to non-climax lowland rainforest habitats of Central America in having a range of tree species of varying heights and diameters with substantial undercover shrubs providing a variety of perches and niches. However, we found that *D. gorgonae* occupies higher perches than either *D. princeps* or *N. medemi*, while *D. princeps* occupied significantly larger perches than *N. medemi* (Table 5), likely owing to a much larger body size and longer limbs (Table 2). This habitat partitioning aligns with Nicholson et al.'s [21] ecomode classification, where each species occupies a di fferent ecomode: *D. gorgonae* = trunk-crown, *D. princeps* = crown giant, *N. medemi* = trunk ground (Figure 6). *Dactyloa chocorum* and *N. parvauritus* are also both considered crown giants, but *D. princeps* clearly utilizes habitat lower down in the forest.


*Diversity* **2019**, *11*, 141

**Table 4.** (**A**) Average morphological and ecological measurements for four populations of *Norops townsendi* on Isla Cocos. (**B**) Results for pairwise comparisons of all


**Table 5.** Mean values and pairwise comparisons of habitat use in Isla Gorgona anoles. Bold values are statistically significant. \* Denotes that only one individual was observed and is not believed to reflect typical habitat use.

**Figure 6.** The Isla Gorgona anole community depicted by habitat in accordance with ecomode designation by Nicholson et al. [21]. Figure modified from [13], which was modified from [78].
