EVALUATING CONFLICT AND BIAS IN GENETIC APPROACHES TO DATING EVOLUTION: RODENTS AS A CASE STUDY
Ryan W Norris, C. William Kilpatrick, and Arlin Stoltzfus
Dating evolutionary divergences among living organisms has traditionally been based solely on information available in the geologic record. More recently, molecular clock techniques have been developed and are being refined that estimate divergence times based on the substitution rate of genes (the rate at which nucleotides or amino acids change over time). In several instances, these two methods have generated contradictory results. For example, molecular and paleontological approaches have produced extremely different estimates for the timing of evolutionary divergence times among orders of placental mammals and within rodents. Molecular studies routinely suggest a Cretaceous (65.5 Ma to 145.5 Ma) origin for the Rodentia and other orders, but the fossil record shows no indication of any placental mammal that can be assigned to a modern order until the Paleocene (55.8 Ma to 65.5 Ma). Here we evaluate the conflict between the fossil record and molecular data and find a significant correlation between dates estimated by fossils and relative molecular branch lengths, suggesting that molecular data agree with the fossil record regarding divergence times in rodents. Our approach includes a correction for tree hierarchy involving simulating the random appearance of fossils and holds true across different molecular clock techniques. We propose that some molecular clock studies may overestimate divergence times due to a period of accelerated molecular evolution across multiple lineages or due to saturation of data that is not adequately corrected by the evolutionary model. Importantly, these types of situations are predicted to overwhelmingly err in a single direction thus introducing bias into calculations of evolutionary divergence times.