The effect of ancient DNA damage on inferences of demographic histories

TY - JOUR

T1 - The effect of ancient DNA damage on inferences of demographic histories

AU - Axelsson, Erik

AU - Willerslev, Eske

AU - Gilbert, Marcus Thomas Pius

AU - Nielsen, Rasmus

N1 - Keywords: Animals; Bayes Theorem; Bison; DNA Damage; Evolution; Evolution, Molecular; Genetic Variation; Genetics, Population; Models, Genetic; Phylogeny; Polymerase Chain Reaction; Population Density; Sequence Analysis, DNA

PY - 2008

Y1 - 2008

N2 - The field of ancient DNA (aDNA) is casting new light on many evolutionary questions. However, problems associated with the postmortem instability of DNA may complicate the interpretation of aDNA data. For example, in population genetic studies, the inclusion of damaged DNA may inflate estimates of diversity. In this paper, we examine the effect of DNA damage on population genetic estimates of ancestral population size. We simulate data using standard coalescent simulations that include postmortem damage and show that estimates of effective population sizes are inflated around, or right after, the sampling time of the ancestral DNA sequences. This bias leads to estimates of increasing, and then decreasing, population sizes, as observed in several recently published studies. We reanalyze a recently published data set of DNA sequences from the Bison (Bison bison/Bison priscus) and show that the signal for a change in effective population size in this data set vanishes once the effects of putative damage are removed. Our results suggest that population genetic analyses of aDNA sequences, which do not accurately account for damage, should be interpreted with great caution.

AB - The field of ancient DNA (aDNA) is casting new light on many evolutionary questions. However, problems associated with the postmortem instability of DNA may complicate the interpretation of aDNA data. For example, in population genetic studies, the inclusion of damaged DNA may inflate estimates of diversity. In this paper, we examine the effect of DNA damage on population genetic estimates of ancestral population size. We simulate data using standard coalescent simulations that include postmortem damage and show that estimates of effective population sizes are inflated around, or right after, the sampling time of the ancestral DNA sequences. This bias leads to estimates of increasing, and then decreasing, population sizes, as observed in several recently published studies. We reanalyze a recently published data set of DNA sequences from the Bison (Bison bison/Bison priscus) and show that the signal for a change in effective population size in this data set vanishes once the effects of putative damage are removed. Our results suggest that population genetic analyses of aDNA sequences, which do not accurately account for damage, should be interpreted with great caution.

U2 - 10.1093/molbev/msn163

DO - 10.1093/molbev/msn163

M3 - Journal article

C2 - 18653730

SN - 0737-4038

VL - 25

SP - 2181

EP - 2187

JO - Molecular Biology and Evolution

JF - Molecular Biology and Evolution

IS - 10

ER -