Geometric morphometric of the R. mexicanus species complex.
No skull views showed sexual dimorphism in shape (dorsal: F = 1.18,p = 0.28; ventral: F = 1.09, p = 0.36) or CS (dorsal: t = 0.25, p = 0.80; ventral: t = 0.93, p = 0.35). For the dorsal view of the skull, there was no significant correlation between CS and shape (F = 0.70; p = 0.55). For the ventral view, the CS-shape correlation was significant (F = 8.39; p = 0.001), with CS explaining only 9.10 % of the shape variation hence the allometric effect was considered relatively weak.
Significant differences were found between the cranial shape (dorsal: F = 3.41, p = 0.004; ventral: F = 8.92, p = 0.001) of the delimited species. For the dorsal view, pairwise comparisons showed significant differences among all putative species (Fig. 5). The mean shape of R. mexicanus clade IIIB differed from R. mexicanus clade IIA in having a relatively longer nasal bone (landmarks 1 and 2), a slightly narrower interorbital region (landmark 5), and the braincase with a tendency to narrow towards the caudal region of the occipital bone. When compared to R. mexicanus clade I, theR. mexicanus clade IIIB showed a narrowing of the anterior-medial region of the skull (landmarks 1-8; 35) although the braincase is slightly wider towards the ends of the parietal-interparietal bones.R. mexicanus clade IIA and R. mexicanus clade I differed in most of the landmarks that characterized the skull shape, having the latter a longer rostrum and a braincase that was wider towards the parietal-interparietal region but narrower towards the curvature of the occipital bone.
In the ventral shape, only the comparison between R. mexicanusclade IIA and R. mexicanus clade IIIB did not differ significantly in skull shape. Differences in the mean shape betweenR. mexicanus clade IIIB and R. mexicanus clade I occurred in the anterior region of the skull; R. mexicanus clade I had a longer incisive foramen, but a shorter palate. The R. mexicanusclade I skull shape showed a notable contraction of the landmarks 22-24, which describe the foramen magnum, with respect to the other putative species. Also, R. mexicanus clade I differed from R. mexicanus clade IIA in that the former had a shorter line of molars and slightly wider basicranium towards the posterior region of the zygomatic bar, but with a tendency to narrow towards the tympanic bullae. No view showed significant skull size differences (CS) between delimited species (dorsal: F = 2.19; p = 0.12; ventral: F = 1.72; p = 0.18).
The CVA showed a partial overlap in the morphospace of all putative species for the dorsal view and considerable discrimination of R. mexicanus clade I for the ventral view along the positive CV1 axis (Fig. 5). The LDA based on skull shape did not assign 100% of the individuals to the corresponding delimited species (Table 3). The dorsal side had a lower correct discrimination rate (65%) than the ventral side (77%). In both views, most of the misclassed individuals belonged to R. mexicanus clade IIIB, while the best classification accuracy was in R. mexicanus clade I, with a majority of individuals correctly assigned.