Fossils, Molecules, Divergence Times, and the Origin of Salamandroidea

COMMENTARY

Fossils, molecules, divergence times, and the origin of Salamandroidea

Jason S. Anderson1 Department of Comparative Biology and Experimental Medicine, University of Calgary, Calgary, AB, Canada T2N 4N1

e paleontologists can some- a classic synapomorphy of Lissamphibia times suffer from envy when (14), the tooth cusp (usually bicuspid) is W we look at the sheer amount separated from the base (or pedicel) by of data that a molecular a zone of poor mineralization (15). Non- systematist can bring to bear on a question pedicellate teeth are also seen in stem of evolutionary relationship between spe- group salamanders such as Karaurus and cies. Our fossils are often fragmentary, and Kokartus (16), and Gao and Shubin (12) the fossil record can be stingy, especially argue that pedicely might be a condition failing to produce small, fragile taxa such derived within salamanders, thus ques- as amphibians over periods of tens of tioning its status as a synapomorphy for millions of years. Nevertheless, paleontol- the group as a whole. Although this in- ogists have exclusive access to some in- ference is certainly possible, pedicely and formation from deep time, primarily the bicuspidality are ontogenetically variable combination of primitive and derived conditions within an individual sala- states that can tell the story of character mander’s life history (17). Furthermore, transformation within a lineage, assuming when one considers the most probable the taxon is sufficiently well represented in lissamphibian outgroup, the amphibamid the fossil record. We also used to claim temnospondyls (18), pedicely and bicuspi- exclusive access to information on the dality have been reported, in some cases timing of events of evolution, but because with a morphology nearly indistinguish- of the development of molecular clock able from what is seen in modern am- techniques we now have to share this phibians. Regardless, this is an intriguing stage. Molecular divergence estimates are specimen that has much to teach us about highly sensitive to the number, location, Fig. 1. Stratigraphic column from the mid-Me- salamander evolution. and topological distribution of calibration sozoic to recent, showing the impact of the new It is the 40-Myr age range extension that – points (1 4), and fossils, being constraints fossil salamander published in PNAS (12). The blue has the most exciting implications. A on the younger end of cladogenesis, nec- bar represents the estimate of salamandroid di- number of estimates for the divergence of essarily underestimate divergence timing. vergence from Yang and Rannala (1), and the red bar represents that from Zhang et al. (6). “A” is cryptobranchoids and salamandroids have The result of these sets of potential biases been made in the past several years (5–9). is a frequent mismatch in the ages esti- the stratigraphic placement of the stem salamander Kokartus; “B1” the age of Chunerpeton used These estimates vary, but in general they mated by using either set of data. This in Reisz and Müller (3) and Roelants (9); “B2” the have become progressively younger as the discrepancy has certainly been evident age of Chunerpeton used by Zhang et al. (6); “C” use of multiple calibrations from the fossil with respect to the evolution of extant the new fossil salamandroid; and “D” the pre- record (3, 4, 19) has become more com- amphibians (frogs, salamanders, and cae- viously oldest known salamandroid Valdotriton. monplace. However, these estimates are cilians; collectively Lissamphibia), with The new salamandroid extends their known range still much older than suggested by a direct molecular estimates being much older (as by 40 Myr (dashed red line), placing it older than reading of the fossil record. One recent much as 100 Myr) than the fossil record an estimate made from the fossil record (6). The attempt to infer divergence by using just suggests (4–11). However, a study in impact of changing hypotheses of taxonomic or stratigraphic placement is also demonstrated; the fossils (10) estimated this split to be at PNAS (12) shows that rapprochement different age interpretations for Chunerpeton has approximately 143 Mya, which is more within at least salamander phylogeny may a 17-Myr impact on the divergence estimate as than 30 Myr more recent than the youn- be within grasp. read from the fossil record (dashed line). gest of the molecular estimates, and 80 Salamanders are the morphologically most generalized of the three extant Myr younger than the average of the groups, not having the locomotor special- The specimen is not only the oldest molecular estimates. izations of their sister taxa frogs (i.e., salamandroid, but it is placed lowest on This discrepancy results from a few jumping) or caecilians (i.e., burrowing). the stem, preserving an interesting mosaic issues with the fossil data that skew this There is an important divergence near the of derived and primitive features. It retains estimate toward younger ages, and biases gills and lateral line canal grooves, hall- in the molecular data that skew estimates base of the salamander phylogenetic tree fi (Fig. 1) between the cryptobranchoids marks of aquatic, larval life, but it is clearly in the older direction. The rst, and (Cryptobranchidae and Hynobiidae), and adult from several osteological indicators, probably greatest, effect with respect to salamandroids (all other salamanders, ex- demonstrating it was neotenic like many this particular fossil-based estimate is the cepting Sirenidae, whose placement re- modern salamanders. A laterally com- incompleteness of the fossil record for mains controversial). Until the PNAS pressed and fin-bearing tail reinforces this lissamphibians. There are enormous study (12), the oldest known salamandroid interpretation. It retains several bones was from the Cretaceous of Spain (13). that, in extant salamanders, are either Gao and Shubin’s new species extends this highly reduced or lost entirely, which is Author contributions: J.S.A. wrote the paper. range by 40 Myr, with important impli- of great interest to specialists. Most sig- The author declares no conflict of interest. cations for our understanding of sala- nificantly, it bears monocuspid, non- See companion article on page 5767. mander evolution. pedicellate teeth. In pedicellate teeth, 1E-mail: [email protected].

www.pnas.org/cgi/doi/10.1073/pnas.1202491109 PNAS | April 10, 2012 | vol. 109 | no. 15 | 5557–5558 Downloaded by guest on September 26, 2021 periods of nonrepresentation during which and Shubin’s new salamander extends the (i.e., the clade must have diverged by this we know there must have been fossils. known range of salamandroids by 40 Myr, point). They then took the suggestion of Approximately 80 Myr span between Tri- or 17 Myr older than the time that had using maxima (10), but instead of using adobatrachus fi (the rst frog) and the stem been estimated by the fossils alone. This these as absolute or hard constraints, used Karaurus Kokartus salamanders and . Sim- starkly underscores the caution we must them as “soft” constraints by creating ilarly, there are approximately 55 Myr take when making statements about evo- a distribution of possible ages for these between Triadobatrachus and the first lutionary trends from a direct reading of events. The result of their method, de- caecilian Eocaecilia, and the next caecilian the fossil record. However, the new fossil fossil does not occur for another 45 Myr salamander is still much younger than pending on the exact parameters used, (20). The fossil record improves as it gets most estimates from molecular clocks. places the basal salamander divergence progressively younger, but not evenly or slightly (3 Myr) to reasonably (13 Myr) even predictably. This incompleteness Gao and Shubin’s new older than this first salamandroid (Fig. 1). makes what is already an underestimate Not using any maximal constraint in- biased toward even younger ages. A sec- salamander extends creased the age of estimation for the di- ond issue is a methodological decision vergence by 20 to 30 Myr, and using made by Marjanovic and Laurin (10) to hard maxima made the estimate 10 Myr “ ” the known range of use hard maximal constraints, i.e., younger than the soft maxima. The pref- making the statement that a clade can salamandroids by 40 Myr. erence for using the soft maxima with- definitely not be older than a particular stood this test from the fossil record well, lower bound (in stratigraphic terms, older is toward the bottom and younger to the This may be a result of a fragmentary and may point the direction for future top of a stacked time series). However, fossil record; fossils await discovery that molecular clock work (1), but further an- we already know that the fossil record is will create more range extensions into alytical work will be necessary to ensure an underestimate of true age of di- deep time, eventually to approach the these results are repeatable and consistent vergence, so those bounds themselves are dates suggested by the molecular clocks. before they become standard practice. just as prone to these same systematic Alternatively, it might point toward sys- However, notwithstanding the caution biases, and most of the molecular studies tematic biases within the chosen calibra- with which fossil data should be used, they have not used maximal constraints as tion method (1) or the molecular data appear to increase the accuracy of esti- a result. Finally, there is a disagreement themselves (22). mates of divergence made by molecular over the age of the earliest crypto- One recent molecular study estimated clock models, and should be incorporated Chunerpeton fi branchoid (21), which un- atimeofdivergencethat ts very well (with appropriate provisos) whenever derscores the need for explicitness when with this new fossil (5). The authors used possible. using stratigraphic and fossil data to con- fossil data in a number of innovative ways. First, they used a number of external strain divergence estimates (4). ACKNOWLEDGMENTS. I thank A. Jaszlics for the This discrepancy is where the new fossil and internal calibration points from the image of Notophthalmus; and Brian Gratwicke for salamandroid fits into the picture. Gao fossil record as hard minimal constraints the image of Cryptobranchus.

1. Yang Z, Rannala B (2006) Bayesian estimation of spe- ians predated the breakup of Pangaea. Am Nat 165: 16. Skutschas P, Martin T (2011) Cranial anatomy of the cies divergence times under a molecular clock using 590–599. stem salamander Kokartus honorarius (Amphibia: Cau- multiple fossil calibrations with soft bounds. Mol Biol 9. Roelants K, et al. (2007) Global patterns of diversifica- data) from the Middle Jurassic of Kyrgyzstan. Zool J Evol 23:212–226. tion in the history of modern amphibians. Proc Natl Linn Soc 161:816–838. 2. Müller J, Reisz RR (2005) Four well-constrained calibra- Acad Sci USA 104:887–892. 17. Tesche M, Greven H (1989) Primary teeth in Anura are tion points from the vertebrate fossil record for molec- 10. Marjanovic D, Laurin M (2007) Fossils, molecules, diver- – nonpedicellate and bladed. J Zoological Syst Evol Res ular clock estimates. Bioessays 27:1069 1075. gence times, and the origin of lissamphibians. Syst Biol 1989:326–329. 3. Reisz RR, Müller J (2004) Molecular timescales and the – 56:369 388. 18. Anderson JS, Reisz RR, Scott D, Fröbisch NB, Sumida SS fossil record: A paleontological perspective. Trends 11. Hedges SB, Kumar S, van Tuinen M (2006) Constraining Genet 20:237–241. (2008) A stem batrachian from the Early Permian of fossil calibrations for molecular clocks. Bioessays 28: 4. Parham JF, et al. (2012) Best practices for justifying Texas and the origin of frogs and salamanders. Nature 770–771. fossil calibrations. Syst Biol 61:346–359. 453:515–518. 12. Gao K-Q, Shubin NH (2012) Late Jurassic salamandroid 5. Zhang P, Wake DB (2009) Higher-level salamander re- 19. Donoghue PCJ, Benton MJ (2007) Rocks and clocks: from western Liaoning, China. Proc Natl Acad Sci USA lationships and divergence dates inferred from com- Calibrating the Tree of Life using fossils and molecules. 109:5767–5772. plete mitochondrial genomes. Mol Phylogenet Evol Trends Ecol Evol 22:424–431. – 13. Evans SE, Milner AR (1996) A metamorphosed sala- 53:492 508. 20. Anderson JS (2008) Focal review: The origin(s) of Mod- mander from the Early Cretaceous of La Hoyas, Spain. 6. Zhang P, Zhou H, Chen Y-Q, Liu Y-F, Qu L-H (2005) – ern Amphibians. Evol Biol 35:231–247. Mitogenomic perspectives on the origin and phylog- Phil Trans R Soc Lond 351:627 646. 21. Gao K-Q, Shubin NH (2003) Earliest known crown- eny of living amphibians. Syst Biol 54:391–400. 14. Parsons TS, Williams EE (1962) The teeth of amphibia group salamanders. Nature 422:424–428. 7. Vieites DR, Min M-S, Wake DB (2007) Rapid diversifica- and their relation to amphibian phylogeny. J Morphol 22. Zheng Y, Peng R, Kuro-o M, Zeng X (2011) Exploring tion and dispersal during periods of global warming by 110:375–389. patterns and extent of bias in estimating divergence plethodontid salamanders. Proc Natl Acad Sci USA 104: 15. Davit-Béal T, Chisaka H, Delgado S, Sire J-Y (2007) Am- 19903–19907. phibian teeth: Current knowledge, unanswered ques- time from mitochondrial DNA sequence data in a par- 8. San Mauro D, Vences M, Alcobendas M, Zardoya R, tions, and some directions for future research. Biol Rev ticular lineage: A case study of salamanders (order Meyer A (2005) Initial diversification of living amphib- Camb Philos Soc 82:49–81. Caudata). Mol Biol Evol 28:2521–2535.

5558 | www.pnas.org/cgi/doi/10.1073/pnas.1202491109 Anderson Downloaded by guest on September 26, 2021