Introduction history of Drosophila subobscura in the New World: a microsatellite‐based survey using ABC methods

Abstract

Drosophila subobscura is a Palearctic species that was first observed in South and North America in the early 1980s, and that rapidly invaded broad latitudinal ranges on both continents. To trace the source and history of this invasion, we obtained genotypic data on nine microsatellite loci from two South American, two North American and five European populations of D. subobscura. We analysed these data with traditional statistics as well as with an approximate Bayesian computation (ABC) framework. ABC methods yielded the strongest support for the scenario involving a serial introduction with founder events from Europe into South America, and then from South America into North America. Stable effective population size of the source population was very large (around one million individuals), and the propagule size was notably smaller for the introduction into South America (i.e. high bottleneck severity index with only a few effective founders) but considerably larger for the subsequent introduction into North America (i.e. low bottleneck severity index with around 100-150 effective founders). Finally, the Mediterranean region of Europe (and most likely Barcelona from the localities so far analysed) is proposed as the source of the New World flies, based on mean individual assignment statistics.

Characterization of simple sequence repeats (SSRs) from Phlebotomus papatasi (Diptera: Psychodidae) expressed sequence tags (ESTs)

“…Although, SSR motifs with more than five repeats were considered here, shorter SSRs were identified. The maximum length achieved was 16 repeats; this is consistent with studies that revealed shorter SSRs in Drosophila [26,27]. …”

Inferring introduction routes of invasive species using approximate Bayesian computation on microsatellite data

“…Genetic variation was summarized within and between populations, using the following statistics: the F ST values between pairs of populations (Weir and Cockerham, 1984) and the mean individual assignment likelihoods of individuals collected in population i and assigned to population j (L i-j , Pascual et al, 2007), the mean number of alleles per locus, the expected heterozygosity (H, Nei, 1987) and the mean variance of the absolute allelic size (V, Estoup et al, 2004) computed for each population (A, H, V) or for each pair of populations (that is by pairwise pooling of population samples) (A2P, H2P and V2P). The default set of statistics (hereafter referred to as 'default stat') was A2P, H2P, V2P, F ST values and L i-j and hence 'Miller' and 'Beaumont stat' to select the best set of summary statistics.…”

“…Such flat posterior probabilities should, therefore, be interpreted with caution, as they may indicate errors in parameter priors and/or model specification. The comparison between observed summary statistics and simulated statistic distributions may also be used to detect such errors (Pascual et al, 2007). In practice, one could advise ABC users to (i) choose broad support of priors (that is region of the prior with positive probabilities) to ensure that it includes the 'true values' of parameters, (ii) be sceptical about prior distributions and/or model specifications when obtaining low maximal probabilities with large 95% CI and (iii) compare the results obtained with and without the unsampled and the inverse-serial scenarios to be able to detect error in scenario specifications.…”

“…Approximate Bayesian computation (ABC) based on summary statistics (Beaumont et al, 2002) does not require likelihood computation, but only requires the ability to simulate datasets and makes it possible to handle large datasets, such as data for hundreds of individuals genotyped at tens of microsatellite loci. This method was recently successfully used to estimate the demographic parameters of various complex scenarios (Fagundes et al, 2007;Pascual et al, 2007) and for model selection (Miller et al, 2005;Pascual et al, 2007;Cornuet et al, 2008).…”

“…We considered microsatellite-like datasets, as this category of highly variable molecular markers have proved successful for addressing questions relating to invasive species (for example Estoup et al, 2004;Miller et al, 2005;Fagundes et al, 2007;Pascual et al, 2007). We evaluated this method in the simple, but crucial case of two invasive populations strongly suspected to have originated from the same source population.…”