Rocks for clocks

Though the fossil record has traditionally provided the timescale for evolutionary history, this role has been usurped completely in recent years by the molecular clock. The molecular clock rests on a few deceptively simple assumptions: (i) mutations accrue within genetic sequences as a result of random copy errors, (ii) the amount of differences in genes of different organisms is a function of the time since they shared an ancestor, (iii) if the age of that ancestor can be determined from the fossil record, the rate at which mutations have accrued can also be determined and, by extrapolation, (iv) the rate can be used to date the times of evolutionary divergence between other lineages. However, the fossil record provides the only viable means of calibrating the molecular clock to time and, thus, the fact that fossil estimates only approximate the timing of evolutionary splits between lineages must be overcome if molecular clock analyses are to be an effective and reliable means of establishing an evolutionary timescale. Surprisingly, the problem of fossil calibrations has been completely ignored until recently. However, the two main molecular clock software packages now provide a means of expressing the uncertainty in calibrations as prior assumptions manifest as probability envelopes that vary with time and may be adapted to individual calibrations. However, they have rarely been implemented for want of evidence on which to base them. The aim of the proposal is to explore the codification of such prior probabilities based on the only appropriate data - knowledge of the biases that control the stratigraphic distribution of fossils. It is generally thought that fossils are randomly distributed within rock sequences. However, it has long been known that the distribution of fossils is tightly controlled by the distribution of rocks representative of the sediments in which the fossil organisms lived. Indeed, the distribution of fossils can be predicted from the distribution of rocks to which they are associated. Thus, it is possible to distinguish between an absence of lineage-representative fossils because (a) there are no suitable rock sequences to sample, (b) suitable fossilization conditions did not obtain (because relatives are also not fossilized), and (c) the lineage had not yet evolved - from which a maximum date constraint may be established for lineage divergence. With minimum (oldest fossil record of the lineage in question) and maximum constraints established, an intervening probability density may also be established that describes how the probability of lineage divergence varies with the availability of suitable rock for sampling. It is our ultimate aim to establish and implement such priors in circumstances where fossil distributions are controlled by subtle variations in sedimentary environments. However, in this proposal we aim to explore the codification and implementation of such prior probabilities in a molecular clock analysis of a broad scale evolutionary event - the establishment of the principal lineages of animals, also known as the Cambrian Explosion. This example is chosen because the priors may be established simply, on variation in the areal extent of marine rock through the Cryogenian-Phanerozoic interval in which these lineages diverged one from another. Most of the data are in place - geological data have been gathered for the Phanerozoic, but we will gather data for the Cryogenian and Ediacaran, and the performance of priors based on these raw data, versus simpler proxies, will be determined. The project will also serve as a training exercise in which the researcher, a palaeontologist, will be trained to perform the molecular clock analyses during the earliest stages of the project, and he will transfer these skills to the other palaeontological staff and students in Bristol, providing them with exotic skills that will allow them to address age-old palaeontological problems anew.