Data

We examined skull morphology in 19 species of South American xenodontine snakes (range 1–5 specimens per species, mean of 4), representing the diversity of diet preferences present in xenodontine snakes. The analyzed specimens were from the following museums: Museu Nacional, Rio de Janeiro, Brazil (MNRJ), Instituto Butantan, São Paulo, Brazil (IBSP), Universidade Federal do Mato Grosso, Cuiabá, Brazil (UFMT), Museu de Zoologia “Prof. Adão José Cardoso”, Unicamp, Campinas, São Paulo, Brazil (ZUEC). The complete list of the analyzed specimens can be found in the Supporting Information, Table A in S1 File. The skulls were dissected from museum specimens and skeletonized by hand. We obtained digital images of the skulls in dorsal and lateral views, using a Canon PowerShot S5 SI digital camera. Images were standardized for skull position, camera lens plane position, and distance between camera lens and skull.

One of us (JK) digitized a set of 20 landmarks in each view (dorsal and lateral) of the skull (Fig 2, for landmarks definitions see list on Table B in S1 File) using tpsDig 2.12 [25]. The landmarks were digitized only on one side of the skull, since our focus was not on the asymmetric component of shape [26]. The landmark data were aligned using Generalized Procrustes Analysis [27; 23]. Then the average shape for each species was calculated and used for subsequent analyses.

Cranial landmarks recorded from South American Xenodontinae snakes. Dorsal view (A), anatomical wire frame of dorsal view (B), lateral view (C), and anatomical wire frame of lateral view (D).

Principal Component Analysis was performed on the average shape for all species Procrustes shape coordinates to visualize the variation among species. We obtained 18 principal axes (PCs) for each view that describes the shape variation. The differences in shape described by each principal axis (PC) were summarized using thin-plate spline deformation grids [28; 29]. The analyses were performed using Geomorph package v.2.1 [29; 30] for R software v.3.1.1 [31], and tpsRelw 1.46 software [25].

Diet preferences were obtained from literature, and are presented as the proportion of each item on the diet (Table 1).