Molecular Timetrees Reveal a Cambrian Colonization of Land and a New Scenario for Ecdysozoan Evolution  Omar Rota-Stabelli, Allison C. Daley, Davide Pisani  Current Biology  Volume 23, Issue 5, Pages 392-398 (March 2013) DOI: 10.1016/j.cub.2013.01.026 Copyright © 2013 Elsevier Ltd Terms and Conditions

Current Biology 2013 23, 392-398DOI: (10.1016/j.cub.2013.01.026) Copyright © 2013 Elsevier Ltd Terms and Conditions

Figure 1 Molecular Clock Analyses of Five Data Sets Support an Ediacaran Emergence of Body Plans, a Cambrian Explosion of Arthropods, and an Unexpectedly Early Colonization of Lands Consensus super-tree presenting divergence dates obtained from the application of fossil calibration points (Table S1) to five partially overlapping data sets, with relevant macroecological events shown at the bottom of the tree. Arthropods emerged at the Ediacaran/Cambrian boundary and diversified throughout the Cambrian. Myriapoda, Arachnida, and Hexapoda diversified during the late Cambrian to early Ordovician, likely just after they invaded land. Terrestrializations of Onychophora, Nematoda, and (Eu)tardigrada are more recent and mirror the establishment of vascular plants or forests and the radiation of insects. Colored bars are 95% credibility intervals estimated from each data set independently (see legend), with circles representing the mean divergence date. Terrestrialization events are indicated with yellow stars, where the black arrows extending backward from them indicate ghost lineages. Red and black dotted ovals represent Pancrustacea and Crustacea diversifications respectively. Branch lengths represent time in millions of years from present (mya). Individual chronograms were obtained using the CIR autocorrelated process (LogNormal for RNA), the heterogenous CAT substitution model, and soft bound 10%. Because a supermatrix analysis of the five partially overlapping data sets was not computationally feasible (see Supplemental Information), a supertree (Figure S1A) summarizes our independent molecular clock results (Figure S1B). Current Biology 2013 23, 392-398DOI: (10.1016/j.cub.2013.01.026) Copyright © 2013 Elsevier Ltd Terms and Conditions

Figure 2 Sensitivity Experiments Show that Prior Choice Does Not Effect Major Results Divergence dates were reestimated after varying different priors, and new dates were plotted (on the y axis) against dates obtained under the original set of priors (on the x axis) to explore the properties (slope and R2) of the regression line interpolating the two estimates. An R2 of 1 indicates that estimates are similar, and R2 greater than 1 indicates that the new dates are older than the original prior dates. Results for EST1 data set are reported here, and results for the other four data sets are in Figures S1B and S2 (see also Table 1; Table S2). (A) Posterior estimates are correlated with posterior priors but are not overdriven by them, because analyses performed without the data (under priors only) show that most of the estimates are either older or younger. In addition, estimates are poorly correlated as suggest by low R2 values observed, and priors CI are much wider than when using molecules (see Table S2). (B) Use of a 75% jackknifed set of calibration (mean over four replicates) does not affect estimates. (C) The use of significantly softer bounds shifts basal nodes older and upper nodes younger. (D) The use of significantly harder bounds shifts basal nodes younger and upper nodes older. (E) The exclusion of fossil constraints that were consistently violated during clock estimate does not modify age of the nodes. (F) Exclusion of all maxima fossil bounds from the ingroup allows basal nodes close to the root to become much older, but other dates, particularly those related to terrestrialization, are compatible with original dates. (G) Ninety percent jackknifing of the data set (reduction in number of alignment positions) does not affect date estimates. (H and I) Employment of less fitting GTR evolutionary model and autocorrelated LogNormal clock model significantly affect fast evolving Nematoda and Tardigrada lineages (circled). (J and K) Molecular dating using (J) the topology from the CAT analysis (depicting group A, Mandibulata, and Tardigrada+Nematoda) and (K) using an alternative topology depicting Vericrustacea, Myriochelata, and Panarthropoda. Date estimates for the nodes relevant to our discussion are extremely similar. Only notable difference is the age for the diversification of the nematodes, which is older (but in any case Ordovician) when Panarthropoda is enforced. Current Biology 2013 23, 392-398DOI: (10.1016/j.cub.2013.01.026) Copyright © 2013 Elsevier Ltd Terms and Conditions