How real is the Cambrian explosion? In a sense, it wasn't an explosion at all in any commonly understood meaning of the term—it was a relatively rapid apparent diversification of animal phyla over the course of at least tens of millions of years, at a rate that is compatible with unexceptional rates of evolution. Even at the most 'explosive' rate that can be inferred from the observations, this is not an event that challenges evolutionary theory, nor should it give comfort to creationists of any stripe.

However, there are controversies here. One camp holds that the rapid divergence of the metazoan phyla in the Cambrian is real: the different phyla all arose sometime around the boundary, 543 million years ago, and then evolved into the various forms we see now. This interpretation is supported by the fossil record, in which the first recognizable representatives of the phyla are found from roughly the same period.

Another interpretation is that the Cambrian explosion is only apparent: that the divergence occurred well before 543 million years ago, and that there was a long period of undetectable evolution. The major groups of animals separated 600 or perhaps even as much as 700 million years ago, flourished as small wormlike forms that would have fossilized poorly, and what the Cambrian represents is an emergence of larger forms with hard body parts that fossilized well. Some of the molecular data supports an early divergence, and there are known pre-Cambrian trace fossils and fossils—the phosphatized embryos of the Doushantuo formation, about 600 million years old, are a good example.

There are also other ambiguities to be resolved. The relationships of many animal phyla are confusing, and who branched from whom remains to be resolved. In the diagram below, the dashed lines in the tree are the problem: do they branch exactly as shown? How deep in time do those branches go?

The fossil record and evolution of 9 of the 35 currently recognized metazoan phyla suggest that most animal phyla diverged/arose at the beginning of the Cambrian (C) period. The thick lines represent the known ranges of fossils from their first appearance in the fossil record. Thin lines represent the inferred metazoan phylogeny based on fossil data. Dashed lines represent an amalgam of three conservative estimates of the inferred metazoan phylogeny.

Rokas, Krüger, and Carroll have taken an ambitious molecular approach to answer those questions. What they have done is examine 50 genes in each of 17 different species spanning 9 phyla, making a special effort to collect new data from phyla underrepresented in previous work: Porifera, Cnidaria, Platyhelminthes, Mollusca, Annelida, and Priapulida. The goal was to obtain sufficient data to resolve those branches in the animal family tree.

Their results are the kind that are most challenging to present: they were abstract and negative. Despite all their data, an alignment of 12,060 amino acids in proteins from all those phyla, the relationships of many of the taxa remain murky. The diagram below summarizes the tree they found. The numbers at the branches are the results of maximum likelihood/maximum parsimony analyses (big numbers are better, reflecting greater certainty in the validity of the branch point), while branches without numbers were not resolvable statistically.

The lack of resolution in phylogenetic relationships among major metazoan phyla. Values above internodes correspond to support values from ML and MP analyses, respectively. Only internodes with significant support in at least one of the two analyses (ML and MP) or internodes present in majority-rule consensus trees of both analyses are drawn. Analyses were also performed by Bayesian inference. Although certain analyses provided strong support for particular clades, analyses of different subsets of taxa produced significantly different and conflicting results.

Hmmm. All that work, and the branching is still blurry. The investigators evaluated a number of possible problems to see if they were the source of the difficulty. For instance, long branch attraction is a common problem, so they reanalyzed the data, excluding some of the long branched taxa to see if it sharpened up the results. It didn't help.

They speculated that there could be a few "rogue" taxa, whose position in the tree was problematic, so they threw the least stable groups out of the analysis to see if that helped. It didn't.

They didn't think that missing data from some taxa would affect the results, but they reanalyzed, excluding the two phyla with the most incomplete data sets. They were right, it didn't improve their analyses. They tried multiple other approaches to figure out why exactly they couldn't discriminate many of the branch points.

Maybe it's simply a shortcoming of their methods: they don't have the capability to resolve events that occurred several hundred million years ago. To address that, they used their same techniques on a completely different kingdom, the fungi, in which we don't know of any major threshold event comparable to the Cambrian explosion. Those results are diagrammed below; metazoa are in the top half of the tree, while fungal taxa are in the bottom half. The key point is that in most cases, it worked! They could see the relationships of the various fungal taxa with a high degree of statistical significance.

The contrast in phylogenetic resolution between the clades of Metazoa and Fungi. Values above internodes are as in Fig. 1. Eleven out of 13 internodes in the fungal clade are significantly supported by both optimality criteria (ML and MP), whereas only 4 out of 14 internodes in the metazoan clade are significant. Analyses were also performed by Bayesian inference.

One possibility is that, unlike fungal history, the speciation events that produced the metazoan phyla were so tightly compressed that they can't be resolved at this distant remove—in other words, that the Cambrian explosion represented a real burst of macroevolutionary branchiness that occurred in a narrow window of time. They tested this hypothesis with a simulation of another known rapid adaptive radiation, the emergence of the mammalian orders. The mammals diverged 107 million years ago over a span of 42 million years, so the window of time is comparable to that of the Cambrian explosion, only much more recent. They then simulated an additional 500 million years of molecular evolution to produce a data set that could be analyzed, and found again that the passage of that much time would obscure the relationships between the orders.

What they propose, then, is that the lack of resolution is data, and represents a positive signature of an adaptive radiation. They come down on the side of the reality of the Cambrian explosion—it is comparable to other situations in which a lineage rapidly diversifies as it exploits a novel or otherwise empty environment.

An accompanying Perspectives article by Jermiin et al. raises some objections to the analyses that were out of my depth, but in particular mentions that some of the assumptions of homogeneity in rate and site of change in genes are unrealistic, and calls for further detailed study of how genes change over time—they are less pessimistic about the resolvability of the branches, and think there is room for improvement in the study. They suggest that it's a step in the right direction of reconciling paleontological and molecular data, and hint that there are 26 more phyla that haven't yet had a similarly rigorous examination yet…there is much work to be done!

Jermiin LS, Poladian L, Charleston MA (2005) Is the "Big Bang" in animal evolution real? Science 310:1910-1911.

Rokas A, Krüger D, Carroll SB (2005) Animal evolution and the molecular signature of radiations compressed in time. Science 310:1933-1938.