Ancestry-specific recent effective population size in the Americas

Abstract

Populations change in size over time due to factors such as population growth, migration, bottleneck events, natural disasters, and disease. The historical effective size of a population affects the power and resolution of genetic association studies. For admixed populations, it is not only the overall effective population size that is of interest, but also the effective sizes of the component ancestral populations. We use identity by descent and local ancestry inferred from genome-wide genetic data to estimate overall and ancestry-specific effective population size during the past hundred generations for nine admixed American populations from the Hispanic Community Health Study/Study of Latinos, and for African-American and European-American populations from two US cities. In these populations, the estimated pre-admixture effective sizes of the ancestral populations vary by sampled population, suggesting that the ancestors of different sampled populations were drawn from different sub-populations. In addition, we estimate that overall effective population sizes dropped substantially in the generations immediately after the commencement of European and African immigration, reaching a minimum around 12 generations ago, but rebounded within a small number of generations afterwards. Of the populations that we considered, the population of individuals originating from Puerto Rico has the smallest bottleneck size of one thousand, while the Pittsburgh African-American population has the largest bottleneck size of two hundred thousand.

Fig 1. Estimated ancestry-specific effective population size in simulated data.

Analysis of 500 simulated individuals from a three-way admixed population. Each column is one of the three simulated ancestries. The y-axes show ancestry-specific effective population size (N_e), plotted on a log scale. The x-axes show generations before present. The dashed lines show simulated effective population sizes. The solid black lines show estimated ancestry-specific effective population sizes, and the gray regions show 95% bootstrap confidence intervals.

Fig 2. Estimated ancestry-specific effective population size in HCHS/SOL data.

The y-axes show ancestry-specific effective population size (N_e), plotted on a log scale. The x-axes show generations before the present. The lines show estimated ancestry-specific effective population sizes, and the colored regions show 95% bootstrap confidence intervals.

Fig 3. Ancestry-specific effective population size for selected populations.

The y-axes show ancestry-specific effective population size (N_e), plotted on a log scale. The x-axes show generations before present. The lines show estimated ancestry-specific effective population sizes, and the colored regions show 95% bootstrap confidence intervals. Each plot shows a different ancestral component. HCHS/SOL populations are included if the sample size multiplied by the average genome-wide ancestry proportion for the given ancestry in that population is at least 100. African ancestry for African American (AA) in Memphis, and European ancestry for European American (EA) in Memphis are included for comparison.

Fig 4. Estimated effective population size in two US cities.

The y-axes show ancestry-specific effective population size (N_e), plotted on a log scale. The x-axes show generations before present. The solid lines show estimated effective population sizes, and the colored regions show 95% bootstrap confidence intervals. Overall effective sizes are shown for the African American (AA) and European American (EA) populations, as well as ancestry-specific effective sizes for African and European ancestry in the African-American populations.