“…In our models, we used an unsampled or "ghost" population as the source of gene flow (Excoffier et al, 2013), which simply models a reservoir for some divergent lineages now found in the sampled population (Beerli, 2004;Slatkin, 2005). It can represent a population that separated a long time ago from the sampled population, as in the case of a secondary contact after a period of isolation, like in hybrid zones at the population or species level (Alcala, Jensen, Telenti, & Vuilleumier, 2016;Alcala & Vuilleumier, 2014;Hvala & Wood, 2012;Tine et al, 2014) or after long-distance dispersals between population of distinct ancestries (Fortes-Lima et al, 2018;Sedghifar, Brandvain, Ralph, & Coop, 2015;Verdu et al, 2014). Interestingly, if the source population is actually sampled, the joint SFS for the source and the target populations will reveal in the target population an excess of rare or event quite frequent derived alleles, for small and large immigration rates, respectively (Supporting Information 12), as previously reported in population or species having recently reconnected (Alcala et al, 2016;Alcala & Vuilleumier, 2014;Fraïsse et al, 2018;Tellier et al, 2011;Tine et al, 2014).…”