Applications of Bayesian skyline plots and approximate Bayesian computation for population demography: A case study of Northern Fur Seals (Callorhinus ursinus)
Villanea, Fernando Alberto
MetadataShow full item record
The demographic history of a population can be reconstructed from its genetic composition, as populations carry signatures of their evolutionary pasts in their genomes. However, the application of population genetics to infer evolution histories across space and time has illustrated the difficulty of reconstructing demographic history based on contemporary genetic information alone. Contemporary genetic data record the sum of all evolutionary events to date, but are strongly influenced by the most recent, significant events in the history of a population. This obscures the effects of older events and makes them difficult to understand. The analysis of ancient DNA is a powerful approach for investigating the historical dynamics of populations that have undergone recent significant shifts in population size. The strength of this approach lies in the ability to directly observe genetic information from populations before, after, or during critical events in their evolutionary history. In this study, we focus on the particular difficulty of revealing ancient demographic events if the population has undergone a more recent population shift. We do this by exploring the demographics of Alaskan northern fur seals (Callorhinus ursinus) through the Holocene. Northern fur seal populations were hunted almost to extinction in the 1800s, severely reducing contemporary genetic diversity. Here, we study alternative complex demographic models using simulated data and test the discriminative power of limited sample sizes of ancient DNA data sets. We focused on the ability of Bayesian skyline plots (BSP) and approximate Bayesian computation (ABC) analysis based on various small ancient DNA data sets to correctly report population events that took place in the Holocene. Our results indicate two population expansions which took place during the Holocene, and the ABC analysis was supported by a statistical power analysis. In addition, we tested the ability of ABC to accurately report the correct population history from simulated datasets, and caution against the use of ABC without an accompanying power analysis, as some population histories are more difficult to accurately capture with small ancient DNA data sets.