Calibrations, dating, and checkpoints
We have assembled a set of 108 calibrations matched to the Phase I ROADIES and ASTRAL trees. Dated trees inferred using MD-CAT are available for ROADIES v0.0.0 and ASTRAL v0.1.1. Future trees with additional taxon-sampling will also be dated using Bayesian methods in mcmctree.
The primary outcomes of this effort are:
- A genomic timescale for vertebrate evolution based on the VGP Phase I guide trees
- A harmonized database of fossil calibrations available to researchers for unlocking the temporal dimension of genome evolution
- An expanded reach and broader integration of the VGP across biological subfields by merging molecular and paleontological approaches
Coordinator: Alex Pyron (rpyron@gwu.edu) – GWU
Contributors:
Sandra Álvarez-Carretero (s.alvarez-carretero@ucl.ac.uk) – UCL
Dahiana Arcila (dkarcila@ucsd.edu) – UCSD
Robert Asher (r.asher@zoo.cam.ac.uk) – Cambridge
Ricardo Betancur (rbetancur@ucsd.edu) – UCSD
Ed Braun (ebraun68@ufl.edu) – UFL
Chase Brownstein (chase.brownstein@yale.edu) – Yale
Santiago Claramunt (sclaramu@uno.edu) – UNO
Phillip Donoghue (Phil.Donoghue@bristol.ac.uk) – Bristol
Daniel Field (djf70@cam.ac.uk) – Cambridge
Lucas Freitas (lucas.freitas@bristol.ac.uk) – Bristol
Delson Hays (dhhays@ucsd.edu) – UCSD
Jon Hughes (hughesjonjames@gmail.com) – ASU
Matt Jones (Matthew.F.Jones@asu.edu) – ASU
Laura Mulvey (laura.mulvey@qmul.ac.uk) – QMUL
Tom Near (thomas.near@yale.edu) – Yale
Davide Pisani (Davide.Pisani@bristol.ac.uk) – Bristol
Mario dos Reis (m.dosreisbarros@qmul.ac.uk) – QMUL
Mark Scherz (mark.scherz@gmail.com) – SNM/UCPH
Tiago Simões (simoes@princeton.edu) – Princeton
Josefin Stiller (josefin.stiller@bio.ku.dk) – UCPH
Nate Upham (nathan.upham@asu.edu) – ASU
We used the VGP Phase I guide trees to place nodal constraints on a reference topology. Our primary aim was to generate well-constrained ranges for a variety of nodes across the tree, to produce a robustly calibrated timescale for vertebrate evolution. These calibrations are congruent with the most up-to-date discoveries in the fossil record. This resulted in a harmonized collection of min & max ages scattered and spaced across the tree. Given the increasing prevalence of fossil-tip-dated phylogenies, much of the previous ambiguity and controversy over the placement and applicability of many fossils was partially alleviated by reference to recent total-evidence results. Generally speaking, we sought the earliest-known crown-group fossils for major clades, to place minimum ages on the MRCA of extant taxa within those lineages. For a set of major named nodes across the backbone of the tree, we also compiled existing consensus estimates for the origin of various groups from the literature, to compare congruence and divergence from previous generations of studies.
We use the VGP Phase I guide phylogenies, ROADIES v0.0.0 and ASTRAL v0.1.1 per-species, each with its own APE-style node numbers from R to facilitate objective reference to nodes. CAL identifiers refer to the master calibration list. Minimal documentation for each constraint is:
- CAL identifier, lineage, and tree-specific node numbers
- Minimum and maximum ages in Ma: e.g., 234.5–456.2
- Associated fossil material and rationale
- Citations and references
We explicitly modeled our calibrations to meet best practices for justifying constraints with the following minimal information:
- Museum numbers of specimen(s) that demonstrate all the relevant characters and provenance data should be listed. Referrals of additional specimens to the focal taxon should be justified.
- An apomorphy-based diagnosis of the specimen(s) or an explicit, up-to-date, phylogenetic analysis that includes the specimen(s) should be referenced.
- Explicit statements on the reconciliation of morphological and molecular data sets should be given.
- The locality and stratigraphic level (to the best of current knowledge) from which the calibrating fossil(s) was/were collected should be specified.
- Reference to a published radioisotopic age and/or numeric timescale and details of numeric age selection should be given.
For some nodes such as the root, calibrations primarily refer to stratigraphic ranges based on initial taxon appearances and expert-assessed consensus for the upper bounds. An example is the root, which likely dates to the mid-Ediacaran ~570Ma. The impact of fossils is more indirect in these instances. Only a few nodes should receive calibrations like this, mostly towards the root and in the outgroups.
Many nodes are calibrated by well-characterized fossils that are widely considered to represent the earliest stem- or crown-group members of a clade based on diagnostic taxonomic characteristics or morphological phylogenetic analyses, but which have not been included in total-evidence analyses for fossil-tip dating. Such fossils provide strong minimum ages, with maximum bounds set according to principles such as the stratigraphic range or next-oldest constrained divergence – an extensive literature exists for this question.
Finally, the ideal type of calibration derives from a recent total-evidence phylogenetic analysis that includes the fossil of interest and therefore provides an objective reference for its placement close to the relevant node in our tree, along with a median age estimate and confidence interval on the stem and crown ages. Recent total-evidence analyses provide such estimates for many major clades.
For a small number of fossils, controversy and disagreement over their phylogenetic placement precluded unambiguous assignment to crown groups in our analyses. Consequently, we will assess the robustness of our analyses by conducting additional runs with and without these constraints to evaluate their impact on estimated node-ages. Since the mcmctree analyses use independent clock rates across branches, we hypothesize that the impacts of these controversial calibrations should be locally restricted.
Additionally, we derived the empirical distribution of dates and ages for a set of clades from TimeTree and additional surveys of the recent literature, giving a broadly sampled distribution from combined previous studies. This empirical distribution was used to generate likelihood checkpoints, employed as post-hoc probabilistic calibration checks. This allowed us to quantify congruence with and deviation from existing estimates on a per-node basis. See the checkpoint overview for the comparisons and their source data.
