Estimating effective population size trajectories from time-series Identity-by-Descent (IBD) segments

Abstract

Long, identical haplotypes shared between pairs of individuals, known as identity-by-descent (IBD) segments, result from recently shared co-ancestry. Various methods have been developed to utilize IBD sharing for demographic inference in contemporary DNA data. Recent methodological advances have enabled the screening for IBD in ancient DNA (aDNA) data, making demographic inference based on IBD also possible for aDNA. However, aDNA data typically have varying sampling times, but most demographic inference methods designed for modern data assume that sampling is contemporaneous. Here, we present TTNE (Time-Transect Ne), which models time-transect sampling to improve inference of recent effective population size trajectories. Using simulations, we show that utilizing IBD sharing in time series has increased resolution to infer recent fluctuations in effective population sizes compared to methods that only use contemporaneous samples. Finally, we developed an approach for estimating and modeling IBD detection errors in empirical IBD analysis. To showcase the practical utility of TTNE, we applied it to two time transects of ancient genomes, individuals associated with the Corded Ware Culture (CWC) and Medieval England. In both cases, we found evidence of a growing population, a signal consistent with archaeological records.