The origin of snakes: revealing the ecology, behavior, and evolutionary history of early snakes using genomics, phenomics, and the fossil record

Abstract

Background: The highly derived morphology and astounding diversity of snakes has long inspired debate regarding the ecological and evolutionary origin of both the snake total-group (Pan-Serpentes) and crown snakes (Serpentes). Although speculation abounds on the ecology, behavior, and provenance of the earliest snakes, a rigorous, clade-wide analysis of snake origins has yet to be attempted, in part due to a dearth of adequate paleontological data on early stem snakes. Here, we present the first comprehensive analytical reconstruction of the ancestor of crown snakes and the ancestor of the snake total-group, as inferred using multiple methods of ancestral state reconstruction. We use a combined-data approach that includes new information from the fossil record on extinct crown snakes, new data on the anatomy of the stem snakes Najash rionegrina, Dinilysia patagonica, and Coniophis precedens, and a deeper understanding of the distribution of phenotypic apomorphies among the major clades of fossil and Recent snakes. Additionally, we infer time-calibrated phylogenies using both new 'tip-dating' and traditional node-based approaches, providing new insights on temporal patterns in the early evolutionary history of snakes.

Results: Comprehensive ancestral state reconstructions reveal that both the ancestor of crown snakes and the ancestor of total-group snakes were nocturnal, widely foraging, non-constricting stealth hunters. They likely consumed soft-bodied vertebrate and invertebrate prey that was subequal to head size, and occupied terrestrial settings in warm, well-watered, and well-vegetated environments. The snake total-group - approximated by the Coniophis node - is inferred to have originated on land during the middle Early Cretaceous (~128.5 Ma), with the crown-group following about 20 million years later, during the Albian stage. Our inferred divergence dates provide strong evidence for a major radiation of henophidian snake diversity in the wake of the Cretaceous-Paleogene (K-Pg) mass extinction, clarifying the pattern and timing of the extant snake radiation. Although the snake crown-group most likely arose on the supercontinent of Gondwana, our results suggest the possibility that the snake total-group originated on Laurasia.

Conclusions: Our study provides new insights into when, where, and how snakes originated, and presents the most complete picture of the early evolution of snakes to date. More broadly, we demonstrate the striking influence of including fossils and phenotypic data in combined analyses aimed at both phylogenetic topology inference and ancestral state reconstruction.

Figure 1

Bayesian phylogenetic tree inferred from phenotypic dataset. Fifty-percent majority rule consensus tree from Bayesian analysis of the state-partitioned phenotypic dataset (766 characters) under the Mkv model [74] in MrBayes. Node values are Bayesian posterior probability support values; only values above 90% are shown. Scale bar represents substitutions/site. Colored boxes indicate major clades. Fossil taxa are marked with a dagger (†).

Figure 2

Bayesian phylogenetic tree inferred from genetic dataset. Fifty-percent majority rule consensus tree from Bayesian analysis of the gene-partitioned genetic dataset (21 nuclear loci and one mitochondrial locus) in MrBayes. Node values are Bayesian posterior probability support values; only values above 90% are shown. Scale bar represents substitutions/site. Colored boxes indicate major clades. Colored lines indicate major clades from traditional taxonomies that do not resolve as monophyletic groups in this topology.

Figure 3

Bayesian phylogenetic tree inferred from combined genetic and phenotypic dataset, unconstrained. Fifty-percent majority rule consensus tree from Bayesian analysis of the unconstrained combined genetic and phenotypic datasets (with corresponding partition schemes) in MrBayes. Node values are Bayesian posterior probability support values; only values above 90% are shown. Scale bar represents substitutions/site. Colored boxes indicate major clades. Colored lines indicate major clades from traditional taxonomies that do not resolve as monophyletic groups in this topology. Fossil taxa are marked with a dagger (†). Grayed taxon names indicate extant species that are included on the basis of phenotypic data only.

Figure 4

Bayesian phylogenetic tree inferred from combined genetic and phenotypic dataset, constrained. Fifty-percent majority rule consensus tree from Bayesian analysis of the combined genetic and phenotypic datasets, with topology constraints implemented as described in the text, as estimated in MrBayes. Node values are Bayesian posterior probability support values; only values above 90% are shown. Scale bar represents substitutions/site. Colored boxes indicate major clades. Fossil taxa are marked with a dagger (†). Grayed taxon names indicate extant species that are included on the basis of phenotypic data only.

Figure 5

Phylogenetic tree inferred using parsimony using phenotypic data. Fifty-percent majority rule bootstrap consensus tree from heuristic searches under the parsimony framework using the complete phenotypic dataset. Node values are bootstrap probabilities; only those above 75% are shown. Colored boxes indicate major clades. Colored lines indicate major clades from traditional taxonomies that do not resolve as monophyletic groups in this topology. Fossil taxa are marked with a dagger (†).

Figure 6

Phylogenetic tree inferred using parsimony using the combined dataset. Fifty-percent majority rule bootstrap consensus tree from heuristic searches under the parsimony framework using the combined (phenotypic + genetic) dataset. Node values are bootstrap probabilities; only those above 75% are shown. Colored boxes indicate major clades. Colored lines indicate major clades from traditional taxonomies that do not resolve as monophyletic groups in this topology. Fossil taxa are marked with a dagger (†).

Figure 7

Ancestral state reconstruction of diel activity pattern. Bayesian SIMMAP ancestral state reconstruction using the constrained tree for the history of the ‘Diel Activity Pattern’ character. Nocturnality is inferred to be ancestral for snakes. The grey box marks the clade Colubroidea, within which diurnal habits re-evolved. Note that SIMMAP estimates the most likely states for tip taxa that are coded as missing or polymorphic – as such, some of the tip states exhibited in this figure are inferred tip states, not coded tip states (e.g., the fossil taxa were coded as missing data; the states they exhibit here are states inferred by the SIMMAP method). Fossil taxa are marked with a dagger (†).

Figure 8

Divergence time tree inferred using the constrained topology. Divergence times inferred using the constrained tree in BEAST. Major crown clades are named, along with two extinct clades (Simoliophiidae and Madtsoiidae). The red line separating the Mesozoic and Cenozoic eras marks the Cretaceous-Paleogene (K-Pg) boundary at 66 Ma. Timescale is in millions of years. Circled numbers and green stars correspond to calibration dates outlined in Additional file 12. Colored boxes indicate major clades. Fossil taxa are marked with a dagger (†). Grayed taxa names indicate extant species that are included on the basis of phenotypic data only.

Figure 9

Reconstruction of the ancestral crown-group snake, based on this study. Artwork by Julius Csotonyi.