2.7 | Demographic inference using fastsimcoal2 and
δaδi
We used the fastsimcoal2 (Excoffier, Dupanloup, Huerta-Sánchez, Sousa,
& Foll, 2013) approach to deduce the recent demographic history ofParascaris spp. populations. We chose only SNPs located in
intergenic regions to avoid the influence of SNPs under selection(Zhou
et al., 2018). We used δaδi (Gutenkunst, Hernandez, Williamson, &
Bustamante, 2009) to investigate alternative demographic scenarios for
the species complex. In the absence of historical evidence, we
hypothesized that there may or may not be any form of gene flow between
roundworm populations. In order to get the best model, we first
simulated a total of ten δaδi 3D models, including one simple model,
three simultaneous splitting models, one ancient migration model, one
simultaneous splitting variations model, one admixed (“Hybrid”) origin
model and three divergences with gene flow variations models. When
fitting demographic models, we perform multiple runs (100 rounds) and
ensure that final optimizations are converging on a similar
log-likelihood score. To estimate demographic parameters, the
derivative-based BFGS algorithm was used to optimize the composite
log-likelihood. All models and scripts are available athttps://github.com/dportik/dadi_pipeline.