How Likely Are The Dates From Nicholson et al.?

Recent posts on Anole Annals evaluated the taxonomic implications of Nicholson et al.’s [1] new systematics, yet their manuscript included similarly bold interpretations of anole biogeography and the chronology of their diversification.  Nicholson et al. claim that a single genus genus concept for anoles can stifle “scientific communication regarding evolutionary events” and used their new multi-genera taxonomy “to propose a bold hypothesis of the biogeographic history of the family within the constraints of the phylogeny inferred here, the latest known fossils, and a paleogeographic interpretation of the deep history of the West Indies, North America, Mesoamerica, and South America.”  My goal today is to address if their phylogenetic dating analysis is capable of delivering on such claims.

Anoles are characterized by a sparse and poorly understood fossil record.  Any attempt to elucidate the evolutionary history and biogeography of anoles depends on neontological data in the form of DNA sequences from extant species.  Nicholson et al. utilized molecular clock dating methods to hypothesize about the temporal history of anoles.  They calibrated their tree with two amber fossils containing lizards identifiable as anoles.  They attribute the first, Anolis dominicanus, to the clade containing A. aliniger, A. chlorocyanus, A. coelestinus, and A. singularis and assign an age of 23 million years before present (mypb) to this clade.  They use the second second fossil, A. electrum, to calibrate the split between A. limifrons and A. zeus to 28 mybp.  With this background in hand, let’s turn to evaluating their results.

For the sake of a critical evaluation, I have centered the remainder of the post around a three questions I would have asked had I been selected to review this paper during the peer review process.

1) Have the authors presented sufficient information to allow for a proper evaluation of their results?
Modern phylogenetic dating analyses incorporate calibration uncertainty and the limits of the data (i.e., the ability of the DNA data to distinguish between alternative rate distributions) in their results.  Results from such analyses are the distribution of hypothesized ages for nodes in the tree.  These analyses do not return single dates, they return credible intervals of dates.  Interpretation of the results and evaluation of biogeographic hypotheses therefore depends on considering these ranges of likely dates.  With regard to the first question, Nicholson et al. present only single dates.  Moreover, it’s not even clear what these dates are.  Are they the minimum credible age?  Are they the mean?  The median?  Already, I’m afraid, it is impossible to critically evaluate what follows in their manuscript because they did not report appropriate summary statistics.

2) Are the calibrations appropriately placed and have the authors addressed potential sources of bias?
The placement of calibrations in a dating analysis dictates the final outcome.  Thus, the obvious question is whether or not the two fossils utilized by Nicholson et al. have been assigned to appropriate positions in the anole phylogeny.  As Jonathan Losos discussed in a related post, it is not immediately evident if the position of the A. electrum calibration is justified.  In fact, there has never been an effort to position A. electrum in the anole tree with cladistic methods.  Nicholson et al. use this fossil to assign an age of 28 mypb to one of the most shallow divergences in the tree.

Improperly positioned calibration points may conflict with one another, but diagnosis of such mistakes is not immediately obvious from a single analysis.  To diagnose potentially conflicting calibrations, other authors advocate the use of a “cross-validation” approach whereby individual calibrations are systematically removed and the analysis re-run without them [3].  If dates strongly deviate between unique combinations of calibrations, then there is reason to believe one or more of the calibrations is unduly influencing the outcome.  Nicholson et al. present no evidence that they evaluated the role of each of their two calibrations in their results.

3) Are the results consistent with related and independent lines of evidence?
Because modern phylogenetic dating analyses incorporate tremendous amounts of uncertainty, it is all the more important to compare the analysis with results from separate lines of evidence.  An obvious first pass is to assess if the estimated clock rates (and hence the dates themselves) are consistent with results from molecular clock analyses of other animal clades.

Nicholson et al. used a fragment of the mitochondrial ND2 subunit neighboring genes in their dating analysis.  This stretch of DNA sequence has a rich history in systematics and the properties of its molecular clock have been explored in a diversity of vertebrate groups.  Although there is no strict rate for ND2 sequence divergence, studies from lizards to birds to fish and frogs typically find that ND2 sequences diverge at a rate of 0.65% per lineage per million years [4].  In other words, if we look at the ND2 sequence analyzed by Nicholson et al. for two species that last shared a common ancestor 1 million years ago, we can hypothesize that these two species will be different at 1.3% of the bases in the sequence.  If we apply this general, back-of-the-napkin calculation to the data analyzed by Nicholson et al., we find that the average divergence across the root of the anole radiation is 24.89% (range: 20.00-30.40%), corresponding to an age of 38.30 mybp (range: 30.77-46.77).  This is a significant deviation from the age of 87 mybp listed by Nicholson et al. for the most basal divergence within the anole radiation (their figure 4a).  Is this discrepancy impossible?  Well, no, but I believe it further calls into question the validity of their results.

If the idea of simply employing the 1.3% rate is too much of an informal goodness-of-fit rubric, we could simply ask if other studies of squamate relationships have recovered similar estimates for the age of anoles.  A recent study of squamate relationships [5], employing multiple fossil calibrations, found that the age of the split between anoles and their corytophanid relatives occurred around 58 mybp (range: 45-75 mybp).

Of course, we could ask if their dates are consistent with paleogeography.  Anoles are predominately tied to terrestrial habitats; there are no sea-faring anoles.  Nicholson et al. estimate the age of Jamaican anoles to be 36 mybp.  This is an interesting finding because  several sources of  Caribbean paleogeography posit that Jamaica sunk beneath the ocean in Eocene times and only re-emerged in the last 10-12 million years [6].  Indeed, the history of emergent landmasses in the Caribbean (and everwhere) is somewhat contentious (primarily, I think, because geologists are not necessarily concerned with such details).  However, I find it a difficult pill to shallow.  Perhaps in the comment section an Anole Annals reader can refer me to specific passages that hypothesize Jamaica has always had emergent land.  We could also evaluate if clades of anoles from the Lesser Antilles have similarly contentious dates, but the authors did not report ages for these clades (note that they do not even present a phylogeny with branch lengths scaled to time anywhere in the paper).

Concluding remarks
The phylogenetic dating analysis conducted by Nicholson et al. does not satisfy the most obvious questions of a critical reviewer.  The authors did not report the appropriate summary statistics necessary to thoroughly evaluate their proposals, nor did they conduct the exploratory analyses necessary to ascertain calibration biases in their results.  Their results are not consistent with additional lines of evidence.  I believe their dating analysis incorporated a considerable bias in utilizing the A. electrum fossil to date a shallow divergence in the tree, which forced their dates to be substantially older than reality.

References Cited
1.    Nicholson K.E., Crother B.I., Guyer C., Savage J.M. 2012 It is time for a new classification of anoles (Squamata: Dactyloidae). Zootaxa 2477, 1-108.
2.    de Oliveira T., Pybus O.G., Rambaut A., Salemi M., Cassol S., Ciccozzi M., Rezza G., Gattinara G.C., D’Arrigo R., Amicosante M., et al. 2006 Molecular epidemiology – HIV-1 and HCV sequences from Libyan outbreak. Nature 444(7121), 836-837. (doi:Doi 10.1038/444836a).
3.    Near T.J., Meylan P.A., Shaffer H.B. 2005 Assessing concordance of fossil calibration points in molecular clock studies: An example using turtles. American Naturalist 165(2), 137-146.
4.    Macey J.R., Schulte J.A., Ananjeva N.B., Larson A., Rastegar-Pouyani N., Shammakov S.M., Papenfuss T.J. 1998 Phylogenetic relationships among agamid lizards of the Laudakia caucasia species group: testing hypotheses of biogeographic fragmentation and an area cladogram for the Iranian Plateau. Mol Phylogenet Evol 10(1), 118-131.
5.    Townsend T.M., Mulcahy D.G., Noonan B.P., Sites J.W., Kuczynski C.A., Wiens J.J., Reeder T.W. 2011 Phylogeny of iguanian lizards inferred from 29 nuclear loci, and a comparison of concatenated and species-tree approaches for an ancient, rapid radiation. Mol Phylogenet Evol 61(2), 363-380. (doi:Doi 10.1016/J.Ympev.2011.07.008).
6.    Graham A. 2003 Geohistory models and cenozoic paleoenvironments of the Caribbean region.  (pp. 378-386, Amer Soc Plant Taxonomists.

• Author
• Recent Posts

Daniel Scantlebury

Latest posts by Daniel Scantlebury (see all)

• How Likely Are The Dates From Nicholson et al.? - October 5, 2012
• Macroevolutionary methods in R workshop in Santa Barbara, CA June 11-15, 2012 - April 19, 2012
• Anolis hendersoni in the Dominican Republic - December 15, 2011