Temporal Variation in Effective Population Size Across the Holocene: Evidence from Ancient European Genomes

The effective population size (N_e) is a foundational parameter in population genetics, yet it is typically treated as a constant when modeling recent human evolution. Using 17,629 ancient European genomes from the Allen Ancient DNA Resource spanning 0-8000 years before present, we measured drift variance at approximately 890,000 putatively neutral loci across 500-year time bins. Drift variance, corrected for sampling error, varied 3.3-fold between Early Neolithic (maximum) and Medieval (minimum) periods in the pan-European dataset. To control for population structure between geographically distinct samples, we repeated the analysis using only British and Irish samples (n=1,388). The single-population analysis yielded a 4.6-fold variation in drift variance, indicating that the pan-European estimate was conservative. These results demonstrate that N_e cannot be treated as a time-invariant parameter across the Holocene. Calculations of selection thresholds, fixation probabilities, and expected allele frequency trajectories that assume constant N_e are systematically biased, with the direction and magnitude of bias depending on the time period under consideration.