advances.sciencemag.org/cgi/content/full/1/10/e1500743/dc1 The PDF file includes: Supplementary Materials for The burrowing origin of modern snakes Hongyu Yi and Mark A. Norell Published 27 November 2015, Sci. Adv. 1, e1500743 (2015) DOI: 10.1126/sciadv.1500743 Materials and Methods Fig. S1. Vestibular shape of all samples. Fig. S2. Placement of shape variables. Fig. S3. Regression of centroid size to scores of principal component 1. Fig. S4. Distribution of shape variables reconstructed for the hypothetical ancestor of crown snakes. Fig. S5. Principal components analyses of vestibular shape. Fig. S6. Missing data in shape variables. Fig. S7. An alternative phylogeny for all samples. Table S1. Taxon sampling in the three habitat groups. Table S2. MANOVA test of Procrustes coordinates among the three habitat groups. Table S3. Scores of the first two principal components. Table S4. Accuracy of the linear discriminant function analysis. Table S5. Habit predictions for D. patagonica and the hypothetical ancestor of modern snakes. Table S6. Habit predictions for D. patagonica and the hypothetical ancestor of modern snakes. Table S7. Habit predictions for D. patagonica and the hypothetical ancestor of modern snakes. Table S8. CT scanning parameters*. Reference (32)
Materials and Methods: X-ray Computed Tomography (CT) was completed at the American Museum of Natural History. Among the 44 samples, 37 were scanned in a GE v tome x dual-tube CT scanner. We CT scanned Dinilysia patagonica (MACN-RN 1014) at the resolution of 26 μm, and X-ray of 170 kv and 130 ma. Modern samples include dry skeletons and alcohol-preserved specimens, and were scanned at 70-220 kv and 80-200 ma (Table S8). We obtained CT data of seven samples (non-amnh specimens) from the CT lab of the University of Texas, Austin. We built virtual models of the inner ear for all samples; the models in PLY format are deposited in MorphoBank (29) Project 2170. We used six type-2 landmarks (17) and 22 semilandmarks placed along the lateral surface of the vestibule and the medial surface of the lateral semicircular canal (Fig. S2). Vestibule expansion or shrinkage was captured with the curvature of the vestibule, and the distance between the vestibule and lateral semicircular canal. The six type 2 landmarks are: intersection of the vestibule and lateral semicircular canal (c0-0: [curve number]-[landmark number]), intersection of the vestibule and utricle (c0-15), medial intersection of the posterior and lateral semicircular canals (c1-0), intersection of the lateral semicircular canal and lateral ampulla (c1-9), intersection of the anterior and lateral ampullae (s0), and intersection of the anterior and posterior semicircular canals (s1). The semilandmarks are equidistant points along two curves surrounding the vestibule and lateral semicircular canal, automatically placed in Landmark Editor 3.6 (32) (Fig. S2D). The six type-2 landmarks are defined as intersection points between two homologous structures, but they can move along the sutures without additional position
constraint. To increase the consistency in landmark placement, we incorporated a reference plane (Fig. S2B) the plane contains landmarks c0-0, c0-15, and s0, and intersects as much of the medial line of the lateral semicircular canal as possible. In some samples, taxon-specific morphologies can render the position of some landmarks uncertain or inapplicable. We define a landmark as missing when its position cannot be determined due to separation of two anatomical structures that are fused in other samples, or vice versa. For instance, the posterior and lateral semicircular canals are connected in most squamates, but separate in Varanus indicus (Fig. S6). In this case, landmark c1-0 was first placed at an approximate position on the lateral semicircular canal, and later estimated using the position of landmark c1-0 in the average Procrustes shape of all samples. After generalized Procrustes analysis, the dataset contains no missing data of shape coordinates. Generalized Procrustes Analysis aligns the 28 landmarks and semilandmarks of all samples. We used the semilandmark sliding algorithm that minimizes bending energy in the R program Geomorph V 1.1-5 (30). Phylogenetic signal is insignificant in the dataset (p = 0.28, iteration = 10), tested by permuting Procrustes shape coordinates of terminal taxa on the snake cladogram (Fig. 1G). In contrast, signal for function (habitat) is significant. MANOVA test using the Procrustes coordinates returned significant difference among the three habitat groups (p = 0.01, Table S2). Principal component analysis extracts major shape variations as principal components (PCs) in the order of percent total variance explained. The first six PCs explain 87 percent of the total variance, of which 47 percent is explained by PC1. Linear
relationship between centroid size of Procrustes shapes and PC1 score was insignificant in a regression test (p = 0.09 for slope = 0; Fig. S3). Linear discriminant analysis was performed among the three habit groups to describe between-group shape separation. The group memberships of Dinilysia patagonica and the hypothetical ancestor were both assigned as unknown in the input data. The linear discriminant analysis then predicted both taxa to belong to the burrowing group. We used standard statistical software (Minitab 17) for the linear discriminant function analysis. We mapped the shape coordinates on the phylogenies using squared-change parsimony, implemented in MorphoJ (31). For the hypothetical ancestor of crown snakes, shape coordinates of the inner ear were reconstructed with and without Dinilysia patagonica. For both reconstructions, the hypothetical ancestor was predicted as a burrower using principal component analysis (Fig. 3; Fig. S5) and linear discriminant analysis (Table S5 to 7).
Fig. S1. Vestibular shape of all samples. (A) Specimen (1) to (12). (B) Specimen (13) to (24). (C) Specimen (25) to (36). (D) Specimen (37) to (44). Not to scale.
Fig. S2. Placement of shape variables. (A) General anatomy of the snake bony labyrinth represented by Ptyas mucosa. (B) Virtual model of the bony labyrinth and the reference plane. (C) Landmarks s0 and s1, and two surface curves consisting of semilandmarks. (D) Landmarks: c0-0, c0-15, c1-0, and c1-9. Semilandmarks: c0-1 to c0-14, and c1-1 to c1-8. Not to scale.
Fig. S3. Regression of centroid size to scores of principal component 1. Blue squares are aquatic species; black triangles are generalists; red dots are burrowing species. The cross is Dinilysia patagonica.
Fig. S4. Distribution of shape variables reconstructed for the hypothetical ancestor of crown snakes. (A) the hypothetical ancestor reconstructed with Dinilysia patagonica. (B) The hypothetical ancestor reconstructed without Dinilysia patagonica.
Fig. S5. Principal components analyses of vestibular shape. (A) The hypothetical ancestor of crown snakes (45) was reconstructed with Dinilysia patagonica (44). (B) The hypothetical ancestor of crown snakes (45) was reconstructed without Dinilysia patagonica (44). Taxon list: 1. Laticauda colubrina, 2. Aipysurus laevis, 3. Laticauda semifasciata, 4. Acrochordus javanicus, 5. Platecarpus coryphaeus, 6. Hydrophis caerulescens, 7. Heloderma suspectum, 8. Lampropeltis getulus, 9. Anguis fragilis, 10. Varanus indicus, 11. Pareas hamptoni, 12. Lamprophis lineatus, 13. Corallus caninus,
14. Python molurus, 15. Naja naja, 16. Eunectes murinus, 17. Vermicella annulata, 18. Trimeresurus stejnegeri, 19. Rhinobothryum lentiginosum, 20. Echiopsis curta, 21. Ptyas mucosa, 22. Boiga irregularis, 23. Pygopus nigriceps, 24. Chrysopelea ornata, 25. Aprasia pulchella, 26. Cylindrophis maculatus, 27. Uropeltis ceylanica, 28. Anniella pulchra, 29. Typhlops jamaicensis, 30. Eryx colubrinus, 31. Anilius scytale, 32. Loxocemus bicolor, 33. Heterodon platirhinos, 34. Simoselaps bertholdi, 35. Rhinotyphlops caecus, 36. Sonora semiannulata, 37. Exiliboa placata, 38. Achalinus spinalis, 39. Typhlosaurus lineatus, 40. Rhinophis philippinus, 41. Xenopeltis unicolor, 42. Dibamus novaeguineae, 43. Bipes canaliculatus, 44. Dinilysia patagonica, and 45. the hypothetical ancestor of crown snakes.
Fig. S6. Missing data in shape variables. (A) The inner ear of Ptyas mucosa in lateral view; the posterior and lateral semicircular canals are connected. (B) The inner ear of Varanus indicus in lateral view; the posterior and lateral semicircular canals are separate. (C) Close-up view of the lateral semicircular canal in Varanus indicus. Not to scale.
Fig. S7. An alternative phylogeny for all samples. Mosasauria is the sister group to all snakes in this phylogeny adapted from refs No. 7 and 22.
Table S1. Taxon sampling in the three habitat groups. Snakes are listed separately from lizards and amphisbaenians. Daggers denote fossils. AQUATIC GENERALIST BURROWING SNAKES Laticauda colubrina Lampropeltis getulus Cylindrophis maculatus Aipysurus laevis Pareas hamptoni Uropeltis ceylanica Laticauda semifasciata Lamprophis lineatus Typhlops jamaicensis Acrochordus javanicus Corallus caninus Eryx colubrinus Hydrophis caerulescens Python molurus Anilius scytale Naja naja Loxocemus bicolor Eunectes murinus Heterodon platirhinos Vermicella annulata Simoselaps bertholdi Trimeresurus stejnegeri Rhinotyphlops caecus Rhinobothryum lentiginosum Sonora semiannulata Echiopsis curta Exiliboa placata Ptyas mucosa Achalinus spinalis Boiga irregularis Rhinophis philippinus Chrysopelea ornata Xenopeltis unicolor Dinilysia patagonica NON-SNAKE SQUAMATES Platecarpus coryphaeus Heloderma suspectum Aprasia pulchella Anguis fragilis Anniella pulchra Varanus indicus Typhlosaurus lineatus Pygopus nigriceps Dibamus novaeguineae Bipes canaliculatus
Table S2. MANOVA test of Procrustes coordinates among the three habitat groups. df SS.obs MS P.val Habitat groups 2 0.604 0.302 0.01 Total 43 2.106 0.049 NA
Table S3. Scores of the first two principal components. Taxon PC1 (47%) PC2 (17%) 1 Aprasia pulchella -0.0621 0.0003 2 Cylindrophis maculatus -0.1818 0.0264 3 Heloderma suspectum 0.1355-0.0929 4 Lampropeltis getulus 0.0431 0.0529 5 Laticauda colubrina 0.2549 0.1077 6 Uropeltis ceylanica -0.1908-0.0201 7 Aipysurus laevis 0.1678-0.0251 8 Anguis fragilis -0.0682-0.0475 9 Anniella pulchra -0.1309-0.0226 10 Laticauda semifasciata 0.3777 0.0314 11 Varanus indicus 0.0865-0.0657 12 Pareas hamptoni -0.0223-0.0752 13 Lamprophis lineatus -0.0187 0.0753 14 Corallus caninus -0.0566-0.1576 15 Typhlops jamaicensis -0.0360-0.1864 16 Python molurus -0.0982-0.0758 17 Naja naja 0.1541 0.1080 18 Eryx colubrinus -0.0745 0.0182 19 Anilius scytale -0.1915-0.0002 20 Loxocemus bicolor -0.1950 0.0854 21 Heterodon platirhinos 0.1318-0.1306 22 Eunectes murinus -0.0783-0.1279 23 Vermicella annulata 0.0182-0.0281 24 Trimeresurus stejnegeri 0.0921-0.0344 25 Simoselaps bertholdi -0.1152 0.0969 26 Rhinotyphlops caecus 0.0583-0.1534 27 Rhinobothryum lentiginosum 0.0743 0.0199 28 Sonora semiannulata 0.1295-0.0325 29 Exiliboa placata -0.0045-0.0243 30 Echiopsis curta 0.2133 0.0368 31 Acrochordus javanicus 0.0481-0.0422 32 Achalinus spinalis -0.0771-0.1537 33 Ptyas mucosa 0.1276-0.0160 34 Platecarpus coryphaeus 0.2130 0.2191 35 Typhlosaurus lineatus -0.1381 0.0498 36 Rhinophis philippinus -0.2028-0.0215 37 Hydrophis caerulescens 0.2110 0.0373 38 Boiga irregularis 0.1985 0.0205 39 Pygopus nigriceps -0.0681 0.0031 40 Xenopeltis unicolor -0.2348 0.1877 41 Dibamus novaeguineae -0.1660 0.0833
42 Chrysopelea ornata 0.1483 0.0900 43 Bipes canaliculatus -0.1592 0.0978 44 Dinilysia patagonica -0.2265 0.1195 45 Hypothetical ancestor of crown snakes (Reconstructed with Dinilysia patagonica) -0.0865-0.0337
Table S4. Accuracy of the linear discriminant function analysis. Predicted group True group Aquatic Generalists Burrowing Aquatic 4 2 0 Generalists 2 10 4 Burrowing 0 6 15 Total N 6 18 19 N correct 4 10 15 N correct proportion 0.667 0.556 0.789 N = 43 N correct = 29 Proportion correct = 0.674
Table S5. Habit predictions for D. patagonica and the hypothetical ancestor of modern snakes. The hypothetical ancestor was reconstructed with D. patagonica. Observation Predicted Groups From Group Squared Distance Probability Dinilysia Burrowing Aquatic 21.046 0.000 patagonica Generalists 10.092 0.066 Burrowing 4.780 0.934 Hypothetical Burrowing Aquatic 13.506 0.002 ancestor Generalists 3.466 0.297 Burrowing 1.749 0.701
Table S6. Habit predictions for D. patagonica and the hypothetical ancestor of modern snakes. The hypothetical ancestor was reconstructed without D. patagonica). Observation Predicted Groups From Group Squared Distance Probability Dinilysia Burrowing Aquatic 17.119 0.001 patagonica Generalists 8.341 0.069 Burrowing 3.142 0.930 Hypothetical Burrowing Aquatic 12.183 0.004 ancestor Generalists 2.851 0.403 Burrowing 2.077 0.593
Table S7. Habit predictions for D. patagonica and the hypothetical ancestor of modern snakes. The hypothetical ancestor was reconstructed with D. patagonica and an alternative phylogeny in fig. S7. Observation Predicted Groups From Group Squared Distance Probability Dinilysia Burrowing Aquatic 17.130 0.001 patagonica Generalists 8.354 0.069 Burrowing 3.141 0.931 Hypothetical Burrowing Aquatic 8.425 0.010 ancestor Generalists 1.298 0.350 Burrowing 0.092 0.640
Table S8. CT scanning parameters*. Taxon Specimen number Voxel size (x=y, mm) 1 Laticauda colubrina Voxel size (z, mm) Filter Scanning lab AMNH R-28996 0.048 0.048 N AMNH 2 Aipysurus laevis AMNH R-5087 0.047 0.047 0.1 mm Cu** AMNH 3 Laticauda AMNH R-161779 0.044 0.044 0.1 mm AMNH semifasciata Cu 4 Acrochordus AMNH R-92269 0.081 0.081 N AMNH javanicus 5 Platecarpus AMNH FM 1645 0.054 0.054 0.1 mm AMNH coryphaeus Cu 6 Hydrophis AMNH R-86181 0.042 0.042 0.1 mm AMNH caerulescens Cu 7 Heloderma AMNH R-161167 0.082 0.082 N AMNH suspectum 8 Lampropeltis AMNH R-95965 0.025 0.025 0.1 mm AMNH getulus Cu 9 Anguis fragilis AMNH R-21745 0.027 0.027 N AMNH 10 Varanus indicus AMNH R-58389 0.032 0.032 N AMNH 11 Pareas AMNH R-153711 0.016 0.016 N AMNH hamptoni 12 Lamprophis AMNH R-50646 0.040 0.040 N AMNH lineatus 13 Corallus AMNH R-55910 0.039 0.039 0.1 mm AMNH caninus Cu 14 Python molurus TNHC, to be 0.044 0.106 N AMNH accessioned 15 Naja naja FMNH 22468 0.044 0.114 N UTCT 16 Eunectes AMNH R-29349 0.067 0.067 N AMNH murinus 17 Vermicella AMNH R-161747 0.015 0.015 N AMNH annulata 18 Trimeresurus AMNH R-21057 0.040 0.040 N AMNH stejnegeri 19 Rhinobothryum AMNH R-52395 0.028 0.028 N AMNH lentiginosum 20 Echiopsis AMNH R-115397 0.033 0.033 N AMNH curta 21 Ptyas mucosa AMNH R-33243 0.037 0.037 N AMNH 22 Boiga AMNH R-69292 0.021 0.021 0.1 mm AMNH
irregularis Cu 23 Pygopus AMNH R-161427 0.030 0.030 N AMNH nigriceps 24 Chrysopelea AMNH R-161965 0.031 0.031 N AMNH ornata 25 Aprasia AMNH R-99706 0.010 0.010 N AMNH pulchella 26 Cylindrophis AMNH R-126605 0.016 0.016 N AMNH maculatus 27 Uropeltis AMNH R-43344 0.015 0.015 N AMNH ceylanica 28 Anniella AMNH R-12851 0.020 0.020 N AMNH pulchra 29 Typhlops USNM 12378 0.005 0.016 N UTCT jamaicensis 30 Eryx colubrinus FMNH 63117 0.018 0.048 N UTCT 31 Anilius scytale USNM 204078 0.018 0.038 N UTCT 32 Loxocemus FMNH 104800 0.021 0.053 N UTCT bicolor 33 Heterodon FMNH 194529 0.029 0.067 N UTCT platirhinos 34 Simoselaps AMNH R-115427 0.016 0.016 N AMNH bertholdi 35 Rhinotyphlops AMNH R-05844 0.023 0.023 N AMNH caecus 36 Sonora AMNH R-170522 0.016 0.016 N AMNH semiannulata 37 Exiliboa AMNH R-102892 0.014 0.014 N AMNH placata 38 Achalinus AMNH R-34620 0.014 0.014 N AMNH spinalis 39 Typhlosaurus AMNH R-98481 0.018 0.018 N AMNH lineatus 40 Rhinophis AMNH R-24674 0.010 0.010 N AMNH philippinus 41 Xenopeltis AMNH R-161693 0.044 0.044 N AMNH unicolor 42 Dibamus AMNH R-86710 0.006 0.006 N AMNH novaeguineae 43 Bipes AMNH R-113487 0.018 0.018 N AMNH canaliculatus 44 Dinilysia patagonica MACN-RN 1014 0.026 0.026 0.1 mm Cu AMNH
* All AMNH specimens were scanned on site, using loans from the herpetological and paleontological collections of the American Museum of Natural History. The specimen of Dinilysia patagonica is non-amnh, but it was scanned at the American Museum of the Natural History. **For AMNH specimens, all scanning parameters follow the output of a GE v tome x dual-tube CT scanner. Copper filter is generally not recommended for recent specimens. For the recent specimens listed with a 0.1 mm Cu filter, it may have been an operational error: a filter may have been selected for data output when it was not actually in place. In these cases, we recommend scanning with and without a filter for optimal results. Institutional abbreviations: American Museum of Natural History (AMNH), Field Museum of Natural History (FMNH), Museo Argentino de Ciencias Narutales, Argentina (MACN), Texas Natural History Collections, University of Texas at Austin (TNHC), and United States National Museum, Smithsonian Institution (USNM).