Dating species divergences using rocks and clocks

Absolute times of species divergences are fundamental to biology, providing a means of calibrating organic evolution to geological time. Methods of establishing evolutionary timescales have evolved with new data and philosophies, developing from data interpreted in light of perceived trends, through more or less formalised approaches to combining stratigraphic age and comparative morphology, to the general abandonment of time as a factor in topology estimation, except where fossil records have been deemed sufficiently complete that phylogenies can be built from sequences of ancestor-descendent relations of morphospecies in the fossil record. Otherwise, palaeontologists have generally only sought to resolve relative relationships and calibrate them post hoc to geological time based on minimum age constraints that incomplete fossil records provide.

The molecular clock hypothesis provides a means of supplementing such age constraints, by providing age estimates for clades that lack any fossil representation.  Data on fossil species are nevertheless essential to disentangle times and rates of molecular evolution to generate estimates of absolute clade ages.  Integration of fossil or other data in molecular clock dating analysis is a very active area of research, with recent Bayesian methods attempting to use more data than ever from the fossil record.  Models of morphological character evolution allow fossil species to be included directly into molecular clock analyses, instead of indirectly informing clade ages.  Diversification models have been developed that control for variation in the fossilization process, that account for uncertainties in fossil assignment, and that incorporate fossil ages with their associated uncertainty.  However, those direct methods, without constraining node ages on the phylogeny, may be very sensitive to the prior or the assumed branching process, and, without the knowledge of the true divergence times, the veracity of these methods may be hard to assess. More than ever, it is important that molecular clock methods to accommodate the reality of molecular evolution and to assess the impact of the multiple factors that may influence the resolution of genetic distances in sequences into estimates of absolute times and rates. Given the limited amount of information in the fossil record, there may be hard limits to the precision in divergence time estimation that is achievable through addition of sequence data alone and it is important that such limits are considered in the deployment of this increasingly powerful tool.

The fossil record provides the primary source of data for calibrating the origin of clades. Although minimum ages of clades are given by the oldest preserved fossil, these underestimate the true age, which must be bracketed by probabilistic methods based on multiple fossil occurrences. Although most of these methods assume uniform preservation rates, such an assumption is unsupported over geological timescales. On geologically long timescales (>10 Myr), the origin and cessation of sedimentary basins, and long-term variations in tectonic subsidence, eustatic sea level, and sedimentation rate control the availability of depositional facies that preserve the environments in which species lived. The loss of doomed sediments, those with a low probability of preservation, imparts a secular trend to fossil preservation. As a result, the fossil record is spatially and temporally nonuniform. Models of fossil preservation should reflect this non-uniformity by using spatially and temporally partitioned empirical estimates of fossil preservation or by using indirect proxies of fossil preservation. Geologically realistic models of preservation will provide more reliable estimates of the origination of clades.

The stratigraphic record is a remarkable yet incomplete archive of the evolution of life and environments on this planet over the past 4.567 billion years.  The low preservation potential of sedimentary successions, due to both the variable creation of accommodation space and long-term tectonic cycles that result in fragmentation, modification and recycling of sedimentary sequences, is responsible for the imperfect nature of this unique archive.  Interrogation of the stratigraphic record, as a means to assess linkages between geological, biological and environmental processes requires an independent framework within which geological, palaeontological and palaeo-environmental data can be objectively integrated.  For over a century the decay of radionuclides to their stable daughter nuclides, with geologically significant half-lives, has been exploited to obtain absolute dates from rocks and minerals, underpins our current understanding of the quantification of absolute geological time and provides the temporal framework within which we can integrate disparate geological archives.  Modern geochronology has evolved to the point where certain rocks and minerals can be dated with a range of radio-isotopic dating methods (e.g., U-Pb, Re-Os, K-Ar) with accuracy approaching the 0.1% level, and can be integrated with other proxies for time in the stratigraphic record, including magnetic field reversals and orbitally influenced cycles. These methods underpin the ongoing numerical calibration of the geological timescale and permit disparate geological archives to be objectively compared.  The nature of the stratigraphic record is inherently limiting and geochronology increasingly plays a role in demarking areas of both certainty and uncertainty in how the stratigraphic archive can be translated into a high-fidelity record of Earth’s evolution.

Molecular and fossil data are observations from the same speciation and extinction process. However, when dating (time-calibrating) phylogenies, we typically assume different models for the generation of the extant species phylogeny (obtained from the molecular data) and for the fossil data. While extant species phylogenies might be modelled through a speciation and extinction process, fossil occurrence times are modelled through arbitrary calibration densities. I will discuss our new approach assuming a speciation and extinction model simultaneously for both data types, together with a model for molecular and morphological evolution. We show in a simulation study the improvement in accuracy of dating phylogenies, while avoiding arbitrary calibration densities. The approach is available within Beast v2.0 and MrBayes.

Total-evidence dating allows evolutionary biologists to incorporate a wide range of sources of dating information (and uncertainty) into a unified statistical analysis. One might expect this to improve the agreement between rocks and clocks, but this is not necessarily the case. We explore the reasons for such discordance using a mammalian dataset with rich molecular, morphological and fossil information. Despite almost 37 kb of sequence data, there is still dramatic elasticity in divergence time estimates. Much of the uncertainty is apparently due to rampant convergence in the morphological data, causing a conflict between morphology and molecules that has a tendency to push estimated divergence times back in time under uninformative or incorrect priors. However, once tip sampling is modeled correctly, we show that informative prior assumptions that penalize unsampled “ghost” lineages tend to give quite close agreement between rocks and clocks. Such informative priors can involve assumptions of high (initial) net diversification, a low extinction rate, or a high fossilization rate; any of these assumptions alone, or any combination of them, produces highly congruent divergence time estimates with a minimal gap between rocks and clocks. The morphological plasticity makes it challenging to place the fossils correctly in the tree but the fossils nevertheless have a stabilizing influence on divergence time estimates and significantly increase the precision of those estimates. We conclude that total-evidence dating is not a panacea for dating problems, but it does give neontologists and paleontologists a common platform for understanding and addressing conflicts between different sources of dating information.

Estimating divergence times, testing hypotheses about patterns of species diversification and analyzing morphological evolution are classically considered as separate research questions. However, many connections would deserve to be made between these various topics in macro-evolutionary sciences. Over the last years, several attempts at integrating some of these various aspects of macro-evolutionary sciences have been made, using hierarchical modeling approaches. Here, I will present a Bayesian framework, which extends previous developments, combining features of the molecular comparative method (Lartillot and Poujol, 2011, Mol Biol Evol 28:729) and of the total evidence dating paradigm (Ronquist et al, 2012, Syst Biol 61:973), in particular by considering fossils as tips. Taking as an input nucleotide sequence data for extant taxa, discrete and continuous characters for both extant and extinct taxa, as well as information about fossil ages, the program jointly estimates divergence times, patterns of morphological evolution and correlations between rates of molecular and morphological evolution. An application of the method to placental mammals will be presented, revealing extensive correlations between life-history, morphological and molecular evolution, while at the same time emphasizing the importance of the assumptions about species diversification for molecular dating. Altogether, this Bayesian approach represents a further step toward a complete integration between the classical comparative method, diversification studies and molecular dating.

Recent advances have allowed for both morphological fossil evidence and molecular sequences to be integrated into a single combined inference of divergence dates under the rule of Bayesian probability. In particular the fossilized birth-death tree prior and the Lewis-MK model of discrete morphological change allow for the estimation of both divergence times and the phylogenetic relationships between fossil and extant taxa. We exploit this statistical framework to investigate the internal consistency of these models by estimating the phylogenetic age of each fossil in turn, within a number of well-characterized data sets of fossil and extant species. We find that we can accurately estimate the age of individual fossils based only on phylogenetic evidence. For example, in two of the data sets we analyze the phylogenetic age of a fossil species is on average only 2 My from the midpoint age of the geological strata from which it was excavated. The high level of internal consistency found in our analyses provides strong evidence that the Bayesian statistical model employed is a good fit for both the geological and morphological data, and provides striking evidence from real data that the framework used can accurately model the evolution of discrete morphological traits coded from fossil and extant taxa. We anticipate that this approach will have diverse applications beyond divergence time dating, including dating fossils that are temporally unconstrained, testing the "morphological clock", and evaluating model assumptions when controversial phylogenetic hypotheses are obtained based on combined divergence dating analyses.