Lukoschek et al., Supplementary Material A.
Evaluating Fossil Calibrations
Cross-validations and empirical coverage
Nuclear DNA identified the crown placements of the Naja and Laticauda fossils as outliers using both the cross-validation analyses and empirical scaling factors; however, this was not the case for the mitochondrial data. We accounted for these differences in terms of the effects of mitochondrial saturation, however, it is also important to evaluate these results in terms of the fossils themselves.
The Naja fossils comprise three extinct species with characters that distinguish Asian and African Naja (Szyndlar and Rage, 1990). However, these fossils also have primitive characters very rare among living cobras (Szyndlar and Rage, 1990, pg 398) suggesting that they belong to the stem Naja. When used to constrain the divergence between extant African and Asian Naja species (crown Naja - node 9) (Kelly et al., 2009; Wuster et al., 2007; Wuster et al., 2008), these fossils overestimated dates for other fossil-calibrated nodes (Fig. 2a) and were identified as outliers by empirical scaling factors (Table 2). However, our alternative placement for dating the Naja-Bungarus divergence (node 8), based on previous evidence of the close relationships between Naja and Bungarus (Slowinski and Keogh, 2000; Wuster et al., 2007), consistently produced much younger divergence dates (Fig. 2) and had intermediate empirical coverage (Table 2). Naja is paraphyletic with Boulengerina and Paranaja and its relationships with other cobra genera are poorly resolved (Slowinski and Keogh, 2000; Wuster et al., 2007); thus, its sister group is difficult to identify as is the most appropriate placement of this fossil on the tree. However, the Naja fossils include well-preserved skull elements with well-defined morphological characters (Szyndlar and Rage, 1990); thus, the best nodal placement might be identified from cladistic analysis of extinct and extant taxa (Doyle and Donoghue, 1993). The relative completeness of these fossils also means they could be used in a Bayesian approach that incorporates morphological data from fossils and extant species (Lee et al., 2009) to evaluate the effects of alternative nodal placements on estimated divergence times.
An elapid fossil from the Australian late Oligocene/early Miocene (20-23 MA) also consistently overestimated dates at other fossil calibrated nodes when used to calibrate the crown hydrophiines (Fig. 2a - node 7) and was identified as an outlier by empirical scaling factors (Table 2). This juvenile vertebra was described as being nested within Laticauda rather than within or basal to any other elapid clade (Scanlon et al., 2003); however, there have been calls for taxonomic and/or stratigraphic revisions of this fossil (J. Scanlon pers. comm.), partly in response to younger molecular dates obtained for the divergence between Laticauda and the remaining hydrophiines (Sanders and Lee, 2008). Our analyses clearly are unable to resolve either the taxonomic affinities or correct stratigraphy of this fossil. Nonetheless, the alternative placement of this fossil to constrain the crown elapids (node 6) tended to produce much younger estimates of other fossil dates (Fig. 2a), though it had amongst the highest empirical scaling factors (Table 2). Proteroglyphous fangs very similar to those of modern elapids first appear in the fossil record in Germany 20-23 MA (Kuch et al., 2006) suggesting that earlier constraints probably might be more realistic for the crown elapids.
The nuclear DNA cross-validations indicated that constraining the stem and crown natricine-colubrine clades (nodes 4 & 5 respectively) at 36 (35-45) My (Alfaro et al., 2008; Guicking et al., 2006) overestimated dates at other fossil calibrated nodes (Fig. 3a). This result is not surprising given that the natricine-colubrine divergence is much shallower than the crown Colubroidea (node 3) in the nuclear gene tree (Figs. 1a), yet their calibrations overlapped to greater or lesser extents (Table 1). Indeed, the colubrine-natricine constraint was almost identical to the set A constraint for the MRCA of all advanced snakes (Table 1 - node 2). This discrepancy is not confined to our study: Alfaro et al., (2008) used 35-55 MA to constrain the tree-root (crown colubroids) and 35-45 MA for the deeply nested natricine-colubrine divergence (see their Fig. 2). These overlapping constraints reflect the fact that the oldest known colubrine fossil, a vertebra assigned to Nebraskophis from the Late Eocene in North America (Parmley and Holman, 2003), occurs simultaneously with the oldest undoubted colubrid fossil from the late Eocene in Thailand (34-37 MA) (Rage et al., 1992), while earliest natricine fossil (Natrix mlynarskii) appears in Europe soon after in the early Oligocene (32-34 MA), where it co-occurs with colubrine (Coluber cadurci) fossils (Rage, 1988). The nodal age for the crown Colubroidea has been inferred from the oldest undoubted colubrid fossil (see below), while the colubrine-natricine divergence has been inferred from the oldest natricine fossil. However molecular phylogenetic appraisals invariably infer nested positions for the colubrines and natricines among the viperids, elapids, atractaspids, and other colubrid subfamilies that comprise the Colubroidea (Kelly et al., 2009; Lawson et al., 2005; Vidal et al., 2007; Yan et al., 2008; Zaher et al., 2009).
How can these discrepancies be resolved? Firstly, Rage (1988, pg. 467) questioned the taxonomic affinities of the natricine fossil in its description. Apart from this fossil, the next appearance of natricine morphology in the fossil record does not occur until the early Miocene (20-23 MA), when several natricine species appear (Rage and Auge, 1993; Ivanov, 2001). The earliest ‘natricine’ fossil may, therefore, be so deeply buried in the stem lineage as to be irrelevant for dating the divergence between modern colubrines and natricines. Secondly, molecular phylogenies do not resolve the natricines and colubrines as sister taxa. Instead, basal divergences among extant colubrid clades comprise one or more poorly resolved polytomies (Lawson et al., 2005; Vidal et al., 2007; Kelly et al., 2009) and a sister-group relationship among colubrines and natricines is only recovered if other clades in the polytomy are not sampled (as in our study and also in Alfaro et al., 2008). Thus, even if the Oligocene fossil is a natricine, its appropriate placement is deeper in the tree. Finally, the first appearance of an undoubted colubrid probably does not represent the earliest divergences among the Colubroidea. Molecular phylogenetic hypotheses indicate that other clades (e.g. viperids, homalopsids, pareatids) diverged earlier in the colubroid radiation (Lawson et al., 2005; Vidal et al., 2007; Alfaro et al., 2008; Wuster et al., 2008); however, the earliest known fossils from these clades date to the early Miocene (Rage and Auge, 1993; Szyndlar and Rage, 1999; Ivanov, 2001).
Multi-calibration Bayesian evaluation of node 2 and 3 calibration sets
The alternative calibration for nodes 2 and 3 were evaluated by comparing the prior and posterior distributions from multi-calibration Bayesian analyses (Fig. 3). The lognormal priors we used model the probability distribution of the actual emergence dates of a clade using the fossil date to inform the hard minimum bound for the youngest possible age, specifying a mean age somewhat older than the fossil, and have a soft maximum bound that allows for the clade to be considerably older than the fossil record (Ho and Phillips, 2009; Yang and Rannala, 2006). It is important to note that the maximum bound of a lognormal prior accounts for a relatively small proportion of the total prior probability and thus has a much smaller effect on the posterior distribution than the log-normal mean or mode, which accounts for a much larger proportion of the prior probability and thus exerts a far stronger constraint (see Fig. 2f in Ho and Phillips 2009). Moreover, the hard minimum bound gives zero probability to the nodal age actually being younger than the oldest fossil known (Ho and Phillips, 2009). As such, posterior distributions that are younger than their lognormal calibration priors strongly suggest that their respective calibrations were too old.
Comparing the set B posterior distributions with their respective priors suggests that the set B calibrations are too old (Fig. 3). In particular, the estimated maximum 95% HPDs are tens of millions of years younger than their respective calibration priors, with the mean and minimum bound for the crown colubroids (node 3) also younger than its calibration prior (Fig. 3). By contrast, the set A posterior distribution for the crown caenophidians (node 2) was much older than its prior (in fact they barely overlapped) indicating that the set A calibration was too young (Fig. 3). The young set A constraint of 38 (34-48) My for crown caenophidians (node 2) (Kelly et al., 2009; Sanders and Lee, 2008) was based on the first appearance of undisputed colubroids (37-39 My) in the fossil record (Head et al., 2005). However, these same fossils also were used to constrain the minimum bound of the crown colubroids (set B - node 3) at 40 My (Wuster et al., 2008), with an maximum bound of 95 My based on possible colubroid fossils from the Cenomanian (93-96 MA) (Rage and Werner, 1999). They were then further used to extrapolate even older dates of 57 (47-140) MA for the origins of the Caenophidia (set B – node 2) (Wuster et al., 2008). Ignoring the contentious taxonomic affinities of the oldest Cenomanian colubroid fossils (Head et al., 2005; Sanders and Lee, 2008), the first primitive colubroids to appear in the fossil record almost certainly belong to the stem colubroids; thus, are inappropriate for dating the crown colubroid radiation (node 3) (Doyle and Donoghue, 1993; Magallon and Sanderson, 2001).
The set C estimated mean and minimum HPD for caenophidian origins (node 2) also were younger than its calibration prior of 65 (63-80) MA (Noonan and Chippindale, 2006a; Noonan and Chippindale, 2006b). This constraint was based on the first Nigerophis fossil dated at 56-65 My (Rage, 1984). However, the relevance of this fossil for dating caenophidian origins relies on the Nigeropheidae belonging to the Acrochordoidea (McDowell, 1987; Rage, 1987) but morphological similarities of this fossil to several disparate groups renders its taxonomic affinities uncertain (Rage, 1984, pg. 71). The maximum bound of this calibration ignores an indeterminate colubroid fossil dated at 49-56 My (Rage et al., 2003) that suggests the Caenophidia started diverging in the early to mid-Eocene. (Head et al., 2005) suggested that this ‘colubroid’ fossil might be an acrochordid: nonetheless, even if reassigned the fossil remains relevant for dating caenophidian origins (acrochordoid-colubroid divergence - node 2).
Alfaro, M. E., D. R. Karns, H. H. Voris, C. D. Brock, and B. L. Stuart. 2008. Phylogeny, evolutionary history, and biogeography of Oriental-Australian rear-fanged water snakes (Colubroidea: Homalopsidae) inferred from mitochondrial and nuclear DNA sequences. Mol Phylogenet Evol 46:576-593.
Doyle, J. A., and M. J. Donoghue. 1993. Phylogenies and angiosperm diversification. Paleobiology 19:141-167.
Guicking, D., R. Lawson, U. Joger, and M. Wink. 2006. Evolution and phylogeny of the genus Natrix (Serpentes: Colubridae). Biol J Linn Soc 87:127-143.
Head, J., P. A. Holroyd, J. H. Hutchison, and R. L. Ciochon. 2005. First report of snakes (Serpentes) from the late middle Eocene Pondaung formation, Myanmar. J Vertebr Paleontol 25:246-250.
Ho, S. Y. W., and M. J. Phillips. 2009. Accounting for calibration uncertaintly in phylogenetic estimation of evolutionary divergence times. Syst Biol 58:367-380.
Kelly, C. M. R., N. P. Barker, M. H. Villet, and D. G. Broadley. 2009. Phylogeny, biogeography and classification of the superfamily Elapoidea: a rapid radiation in the late Eocence. Cladistics 25:38-63.
Kuch, U., J. Muller, C. Modden, and D. Mebs. 2006. Snake fangs from the lower Miocene of Germany; evolutionary stability of perfect weapons. Naturewissenschaften 93:84-87.
Lawson, R., J. B. Slowinski, B. I. Crother, and F. R. Burbrink. 2005. Phylogeny of the Colubroidea (Serpentes): New evidence from mitochondrial and nuclear genes. Mol Phylogenet Evol 37:581-601.
Lee, M. S. Y., P. G. Oliver, and M. N. Hutchison. 2009. Phylogenetic uncertainty and molecular clock calibrations: a case study of legless lizards (Pygopodidae, Gekkota). Mol Phylogenet Evol 50:661-666.
Magallon, S., and M. J. Sanderson. 2001. Absolute diversification rates in angiosperm clades. Evolution 55:1762-1780.
McDowell, J. R. 1987. Systematics. Pages 3-50 in Snakes - ecology and evolutionary biology (R. A. Seigel, J. T. Collins, and S. S. Novak, eds.). Macmillan, New York.
Noonan, B. P., and P. T. Chippindale. 2006a. Dispersal and vicariance: the complex evolutionary history of boid snakes. Mol Phylogenet Evol 40:347-358.
Noonan, B. P., and P. T. Chippindale. 2006b. Vicariant origin of Malagasy reptiles supports late Cretaceous Antarctic land bridge. The American Naturalist 168:730-741.
Parmley, D., and J. A. Holman. 2003. Nebraskophis HOLMAN from the Late Eocene of Georgia (USA), the oldest known North American colubrid snake. Acta Zool Cracov 46:1-8.
Rage, J. C. 1984. Serpentes. Pages 1-79 in Encyclopedia of Paleoherpetology Gustav Fischer Verlag, Stuttgart.
Rage, J. C. 1987. Fossil history. Pages 51-76 in Snakes - ecology and evolutionary biology (R. A. Seigel, J. T. Collins, and S. S. Novak, eds.). Macmillan, New York.
Rage, J. C., S. Bajpai, J. G. M. Thewissen, and B. N. Tiwari. 2003. Early Eocene snakes from Kutch, Western India, with a review of the Palaeophiidae. Geodiversitas 25:
Rage, J. C., E. Buffetaut, H. Buffetaut-Tong, Y. Chaimanee, S. Ducrocq, J. J. Jaeger, and V. Suteethorn. 1992. A colubrid in the late Eocene of Thailand: The oldest known Colubridae (Reptilia, Serpentes). Comptes Rendus de l'Académie des sciences 314:1085-1089.
Rage, J. C., and C. Werner. 1999. Mid-Cretaceous (Cenomanian) snakes from Wadi Abu Hashim, Sudan: The earliest snake assemblage. Palaeontol Afr 35:85-110.
Sanders, K. L., and M. Y. L. Lee. 2008. Molecular evidence for a rapid late-Miocene radiation of Australasian venomous snakes (Elapidae: Colubroidea). Mol Phylogenet Evol 46:1180-1188.
Scanlon, J. D., M. S. Y. Lee, and M. Archer. 2003. Mid-Tertiary elapid snakes (Squamata, Colubroidea) from Riversleigh, northern Australia: early steps in a continent-wide adaptive radiation. Geobios 36:573-601.
Slowinski, J. B., and J. S. Keogh. 2000. Phylogenetic relationships of elapid snakes based on cytochrome b mtDNA sequences. Mol Phylogenet Evol 15:157-164.
Szyndlar, Z., and J. C. Rage. 1990. West Palearctic cobras of the genus Naja (Serpentes: Elapidae): interrelationships among extinct and extant species. Amphib-Reptilia 11:385-400.
Vidal, N., A. S. Delmas, P. David, C. Cruaud, A. Couloux, and S. B. Hedges. 2007. The phylogeny and classification of caenophidian snakes inferred from seven nuclear protein-coding genes. C R Biol 330:182-187.
Wuster, W., S. Crookes, I. Ineich, Y. Mane, C. E. Pook, J.-F. Trape, and D. G. Broadley. 2007. The phylogeny of cobras inferred from mitochondrial DNA sequences: evolution of venom spitting and the phylogeography of the African spitting cobras (Serpentes: Elapidae: Naja nigricollis complex). Mol Phylogenet Evol 45:437-453.
Wuster, W., L. Peppin, C. E. Pook, and D. E. Walker. 2008. A nesting of vipers: phylogeny and historical biogeography of the Viperidae (Squamata: Serpentes). Mol Phylogenet Evol 49:445-459.
Yan, J., H. Li, and K. Zhou. 2008. Evolution of the mitochondrial genome in snakes: gene rearrangements and phylogenetic relationships. BMC Genet 9:569.
Yang, Z., and B. Rannala. 2006. Bayesian estimation of species divergence times under a molecular clock using multiple fossil calibrations with soft bounds. Mol Biol Evol 23:212-226.
Zaher, H., F. G. Grazziotin, J. E. Cadle, R. W. Murphy, J. C. de Moura-Leite, and S. L. Bonatto. 2009. Molecular phylogeny of advanced snakes (Serpentes, Caenophidai) with an emphasis on South American Xenodontines: a revised classification and descriptions of new taxa. Papeis Avulsos de Zoologia 49:115-153.