Dating lineages: molecular and paleontological approaches to the temporal framework of clades

• Published in: International journal of plant sciences 2004-07, Vol.165 (S4), p.S7-S21
• Main Author: Magallon, S.A
• Format: Article
• Language: English
• Subjects: Autocorrelation; Calibration; Classification; Confidence interval; Data smoothing; Estimation methods; Evolution; Fossils; geographic disjunction; literature reviews; Maximum likelihood estimation; Molecular biology; molecular systematics; Paleobotany; paleontology; Parametric models; Phylogenetics; Phylogeny; plant genetics; plant taxonomy; population genetics
• ISSN: 1058-5893; 1537-5315
• DOI: 10.1086/383336

The recent proliferation of methodological advances in molecular phylogenetic and paleobiological research has resulted in powerful approaches to investigate the temporal framework of lineages. This article is a review of molecular and paleontological methods to estimate ages of clades. Inferring ages of clades is complicated by the nature of the process of molecular substitution and the uncertainties of the paleontological record. Some of the greatest problems associated with molecular methods include the stochastic nature of molecular substitution, the assumption of rate constancy among lineages when such constancy is absent, and the inextricable link between substitution rate and elapsed time on the branches of phylogeny. Molecular methods to estimate ages are ultimately based on the fact that as time elapses, molecular differences accumulate among sequences. Under rate constancy, methods to estimate ages include linear regression of molecular distance on elapsed time and maximum likelihood optimization of node ages under a single rate. Recently developed methods that allow rate heterogeneity are powerful approaches to estimate rates and divergence times under more realistic assumptions. Among‐lineage rate variation is introduced as a compound Poisson process or more frequently is guided by the principle of temporal rate autocorrelation. These methods are based on numerical, semiparametric, and Bayesian parametric approaches, and some allow incorporation of constraints on the ages of nodes (derived, e.g., from fossils), conferring additional realism to age estimates. The paleontological record provides times of first appearances of morphological traits but not of lineage divergences; nevertheless, it represents one of the few sources of absolute information to decouple rates and times in a phylogeny. Analytical methods applied to paleontological data provide an alternative source of information about lineage duration. Stratigraphic confidence intervals that contain the time of origin of a lineage under a known probability are based on the frequency and abundance of fossil finds through the lineage’s fossil record. Tests of postulated lineage durations, derived, for example, from a molecular age estimate, are available under probabilistic or likelihood frameworks. A powerful approach toward achieving more robust inferences about evolutionary rates and timing of lineage divergence lies in the complementary use of molecular‐ and paleontological‐based appro