Recent acceleration of human adaptive evolution
- *Department of Anthropology, University of Wisconsin, Madison, WI 53706;
- ‡Department of Algorithm Development and Data Analysis, Affymetrix, Inc., Santa Clara, CA 95051;
- §Department of Anthropology, University of Utah, Salt Lake City, UT 84112; and
- ¶Department of Biological Chemistry and Institute of Genomics and Bioinformatics, University of California, Irvine, CA 92697
-
Contributed by Henry C. Harpending, August 13, 2007 (received for review May 24, 2007)
-
Fig. 1.Age distribution of ascertained selected alleles. Each point represents the number of variants dated to a single 10-generation bin. Fitted curves are the number of ascertained variants predicted by Eq. 2 under a constant population size and constant s̄ = 0.022 for YRI and s̄ = 0.034 for CEU. The distribution drops to zero approaching the present, because all alleles have frequencies >22% today. The 2,965 (YRI) and 2,246 (CEU) selection ages shown have had 509 alleles removed that are likely examples of ongoing balanced selection (SI Appendix). Including these alleles in the analysis does not change the overall conclusion of acceleration of selection.
-
Fig. 2.Historic and prehistoric population size estimates for human populations (SI Appendix). Key features are the larger ancestral African population size and the earlier Neolithic growth in core agricultural areas.
-
Fig. 3.Tip of the iceberg. Both the demographic and constant-rate models can account for the age distribution of ascertained variants (CEU data shown), but they differ greatly in the expected number of variants above the ascertainment frequency (fixed or near-fixed). The demographic model predicts a low long-term substitution rate and few alleles >78%, consistent with the observed data.
Footnotes
- †To whom correspondence may be addressed. E-mail: jhawk{at}wisc.edu, harpend{at}xmission.com, or rmoyzis{at}uci.edu
- © 2007 by The National Academy of Sciences of the USA
