The impact of sampling schemes on demographic inference : an empirical study in two species with different mating systems and demographic histories

• Most species have at least some level of genetic structure. Recent simulation studies have shown that it is important to consider population structure when sampling individuals to infer past population history. This is because different sampling strategies will introduce biases in the site frequency spectrum (SFS), which is generally the basic information used to infer population history. The relevance of the results of these computer simulations for empirical studies, however, remains unclear. In the present study we use DNA sequence datasets collected in two closely related species with very different histories, the selfing species Capsella rubella and its outcrossing relative C. grandiflora, to assess the impact of different sampling strategies on summary statistics and the inference of historical demography. Sampling strategy did affect values of Tajima’s D in both species but differences were not significant, and the general conclusions about demographic history were comparable across sampling schemes even when re-sampled data were analysed with Approximate Bayesian Computation. We explore the effects of sampling scheme when a limited number of loci are analysed, and in models that include population expansion within demes and hierarchical structure. While we observe a less severe effect of sampling than predicted under some simulation models, our results should not be seen as an encouragement to neglect this issue. In general, a good coverage of the natural range, both within and between populations, will be needed to obtain a reliable reconstruction of a species demographic history. 

Ämnesord

NATURVETENSKAP  -- Biologi (hsv//swe)

NATURAL SCIENCES  -- Biological Sciences (hsv//eng)

Nyckelord

population structure

Tajima’s D

frequency spectrum

Capsella

Publikations- och innehållstyp

ref (ämneskategori)

art (ämneskategori)