- Figure S1 - Schematic workflow for adding new datasets to the metatree. Publications must include a machine-readable version of the character-taxon matrix, or this must either be available elsewhere or recreatable from the information within the publication. Redundancy with older data sets is recorded with “Parent” tags in the XML file, and alternate coding strategies are recorded with the “Sibling” tags.
- Figure S2 - Schematic workflow for generation of the metatree. Greater detail for each step is given in the text.
- Figure S3 - Strict consensus of 1,000 MPTs with STR taxa reinserted. Note the overall high resolution compared to formal supertrees (599 of a maximum possible 961 internal nodes assuming a rooted fully bifurcating tree), with the lowest areas of resolution in Sauropodomorpha and Enantiornithes.
- Figure S4 - Figure to demonstrate that the node-dating algorithm of Hedman is not generalisable to a tree-dating algorithm. Outgroup sequences for three nodes (X-Z) on a hypothetical phylogenetic tree of six fossil taxa (A-F) with sampled ages (T1-T6). Note that for the standard (non-conservative) Hedman approach taxon C is re-dated as T5, the same age as the oldest member (B) of its sister clade (AB), biasing the date of node Y towards a minimum possible age (invalidating the standard approach). Whereas under the conservative Hedman approach taxon C is ignored, as it is younger (T6) than the lower bound for Y set by node AB (T5). Under the conservative scenario the sequence of outgroup taxa for nodes X and Y are identical, meaning branch XY would have zero-length (invalidating the conservative approach). Due to these problems in the present paper only nodes X and Z are considered dateable using the Hedman algorithm (conservative approach), whereas node Y would be dated by making random draws from a uniform distribution bounded by separate random draws from the posterior distributions of nodes X and Z.
- Supplementary Information - Additional information describing: 1) Tree construction (a full description of the metatree protocol), 2) Description of the Hedman algorithm (one of the two dating methods applied here), 3) Probabilistic divergence dates (summary values for dates plotted in Figure 1), 4) Clock-based dates for comparison (Sources for dates in lower right of Figure 1), 5) Dryad files (list of additional Dryad files), 6) Supplementary References, and 7) Supplementary Figure Captions