Calibrating phylogenies: are we using the right fossils?

Past FMNH REU Projects

Current FMNH REU Webpage

A critical contribution of the fossil record to our understanding of evolution is to provide ages for when taxa and their distinguishing features arose. Few studies broadly integrate information from the fossil record with that for living organisms, and fewer yet take into account the biases inherent to either, leading to conflicting results from the two disciplines. For example, fossil record dates for evolutionary events clash with those extrapolated from molecular phylogenies for many major clades. One possible reason for this may be incongruent scales of resolution between the fossil record and molecular divergence estimates. In order to test how compatible these two sources of age information are, age estimates and variances for evolutionary branching events will be compared to statistical measures of fossil record quality for a diverse sample of clades. These analyses will determine whether confidence intervals on fossil ages of overlap corresponding molecular age estimates, and whether their magnitude is greater or smaller than the variances on the extrapolated molecular ages of these groups. They will also inform us as to whether fossils chosen for calibration tend to come from taxa and ages with relatively good fossil records or not. Results may shed light on causes for discrepancies between the two sources of temporal information and provide criteria for improved choice of fossils for calibration purposes.



Research methods and techniques: The participant will be given an introduction to the theoretical background for methods of measuring completeness of the fossil record and techniques for reconstructing divergence ages from molecular data, as well as instruction on paleontological  methods for analyzing the fossil record (simple gap analysis) and sources for temporal information on fossils (e.g. www.paleodb.org).  The project will cover several taxonomic groups (e.g. birds, mammals, arthropods), and the undergraduate participant will gather information on the fossil record of these groups, run analyses to determine the confidence intervals on their group and its constituent clades, and compare results to the published predictions of molecular phylogenies using appropriate statistics. Candidates with programming/bioinformatics skills are preferred.



Curator/Advisor: Dr. Peter Makovicky, Dept. of Geology


REU Intern: GREGOR-FAUSTO SIEGMUND

Biological Sciences major

University of Chicago

Symposium Presentation Title: Measuring the quality of the avian fossil record to explain the ‘rock-clock’ divide of the avian evolutionary timescale.

Symposium Presentation Abstract: Over the course of the past two decades, a series of studies have posited that modern birds (Aves: Neornithes) diverged in the early to mid-Cretaceous, suggesting that most modern orders were present prior to the Cretaceous-Paleogene (K-Pg) extinction event 65.5 million years ago. Nearly the entire, reliable neornithine fossil record, however, is younger than the K-Pg boundary, a discrepancy that has made establishing an absolute timescale for modern avian origins contentious. Explanations pinning the incongruity in estimates on inadequate molecular rate models or on a poor and an unevenly sampled fossil record abound, yet comprehensive tests of the latter are lacking. To address this issue, I compiled and annotated a database of stratigraphic horizons, completeness of material, and taxonomic assignment for Mesozoic and Paleogene avian fossils. Raw measures of taxonomic diversity through time, of basal and modern birds, show both lower standing diversity and uneven sampling in the Mesozoic. The latter conclusion is likewise supported by the Simple Completeness Metric, a measure of fossil record quality, which suggests that avian orders are far less well-sampled in the Mesozoic than in the Cenozoic.  A further gap analysis, which places confidence intervals on observed fossil horizons, supports this result and suggests that past studies that have calculated confidence intervals for avian orders by assuming a random distribution may have been too assumptive. Sister group relationships have also been unaccounted for in these studies, indicating that there is at least some room to decrease the rock-clock divide. While these metrics point to a level of neornithine diversification in the Cretaceous, discord between molecular and paleontological approaches remains, suggesting that fossil sampling alone does not account for the disagreement in divergence dates.