Origin of the Eumetazoa: testing ecological predictions of molecular clocks against the Proterozoic fossil record

Abstract

Molecular clocks have the potential to shed light on the timing of early metazoan divergences, but differing algorithms and calibration points yield conspicuously discordant results. We argue here that competing molecular clock hypotheses should be testable in the fossil record, on the principle that fundamentally new grades of animal organization will have ecosystem-wide impacts. Using a set of seven nuclear-encoded protein sequences, we demonstrate the paraphyly of Porifera and calculate sponge/eumetazoan and cnidarian/bilaterian divergence times by using both distance [minimum evolution (ME)] and maximum likelihood (ML) molecular clocks; ME brackets the appearance of Eumetazoa between 634 and 604 Ma, whereas ML suggests it was between 867 and 748 Ma. Significantly, the ME, but not the ML, estimate is coincident with a major regime change in the Proterozoic acritarch record, including: (i) disappearance of low-diversity, evolutionarily static, pre-Ediacaran acanthomorphs; (ii) radiation of the high-diversity, short-lived Doushantuo-Pertatataka microbiota; and (iii) an order-of-magnitude increase in evolutionary turnover rate. We interpret this turnover as a consequence of the novel ecological challenges accompanying the evolution of the eumetazoan nervous system and gut. Thus, the more readily preserved microfossil record provides positive evidence for the absence of pre-Ediacaran eumetazoans and strongly supports the veracity, and therefore more general application, of the ME molecular clock.

Fig. 1.

ML analysis of a concatenated sequence of enolase and glyceraldehyde-3-phosphate dehydrogenase (668 amino acids) from six different arthropods by using the polychaete annelid Nereis as the outgroup. The branches leading to each of the six different arthropod taxa are similar, despite a Triassic divergence for the two dipertans and Ordovician and Silurian divergences for the chelicerates and myriapods, respectively. Hence, calibrating a molecular clock analysis to the myriapod and/or chelicerate divergence will result in spurious overestimates of divergence times between deuterostomes and protostomes, given the similar rate of molecular evolution between dipterans, echinoderms, and molluscs (5).

Fig. 2.

Maximum parsimony analysis of a total-evidence data set consisting of 2,039 amino acids derived from the seven different housekeeping genes used in Fig. 2, 228 amino acids from the cytochrome oxidase I gene, 1,747 nucleotides from the 18S rDNA gene, and 150 morphological characters coded for the genus where possible, for 24 metazoan taxa by using two fungal and two plant taxa as outgroups (see Appendices 1–3, which are published as supporting information on the PNAS web site). Purple taxa are the ambulacrarians (echinoderms + hemichordates), blue taxa are the spiralian protostomes (molluscs, annelids, and nemerteans), green taxa are the ecdysozoans (insects and priapulid), orange taxa are the cnidarians, and red taxa are the sponges. This is one of two trees at 10,765 steps (number of parsimony informative characters, 1,889); consistency index = 0.54; retention index = 0.52; rescaled consistency index = 0.28. Bootstrap values are derived from 1,000 replicates. Note that Porifera is paraphyletic, with calcisponges more closely related to eumetazoans than to demosponges. Because of the clear homology between the water-canal systems of calcisponges and silicisponges (15, 16), sessile microsuspension feeding must have evolved sometime before the last common ancestor of metazoans (. 14) and lost sometime before the last common ancestor of eumetazoans (indicated by the red line).

Fig. 3.

Cladogram derived from ME analysis of the seven concatenated protein sequences from 29 metazoan taxa by using a fungus and a plant as outgroups. Color coding is the same as in Fig. 2. Numbers in black boxes are calibration points, and numbers in white boxes are molecular clock estimates derived from r8s v. 1.5; ME estimate is on the top and the ML estimate is on the bottom (see the key). Bootstrap values (1,000 replications) are given to the left of the boxes. Note that eumetazoan apomorphies (e.g., the gut) arose between 634 and 604 Ma according to ME but 826–748 Ma according to ML.

Fig. 4.

Acanthomorphic acritarchs from before (A and B) and after (C and D) the Marinoan turnover. (A) Tappania sp. from the 850-Ma Wynniatt Fm, Northwestern Canada (39), but also known from ≈1,450-Ma (38) strata, giving it a ≈600-myr age range. (B) Trachyhystrichosphaera aimica from the 850-Ma Wynniatt Fm (39), but also known from 1,000-Ma (40) and post-Sturtian (43) strata, giving it a ≈350-myr age range. (C) Meghystrichosphaeridum chadianesis from the Doushantuo Fm, Southern China; age range, <55 my. [Image courtesy of Shuhai Xiao (Virginia Polytechnic Institute and State University, Blacksburg, VA).] (D) Unidentified Lower Cambrian acanthomorph, which, as a class, have a mean age range of 7.7 my (41); note the fundamentally smaller dimensions relative to the Precambrian acanthomorphs, indicative of a planktic habit. (Scale bar in C applies to all images.)