“…(3) linked selection, background selection, selection at non-targeted loci and selection from standing variation (Schluter and Conte, 2009;Charlesworth, 2012;Cutter and Payseur, 2013;Burri et al, 2015;Christe et al, 2017); (4) demographic and evolutionary histories, including ancient introgression and introgression from other sister species (Fraïsse et al, 2014Nadeau et al, 2016;Ma et al, 2018); (5) gene density (Nordborg et al, 2005;Yeaman, 2013); or (6) the genetic architecture of adaptive and/or barrier loci (Le Corre and Flaxman et al, 2014;Conte et al, 2015;Yeaman, 2015). GEA models have evolved with the aim to reduce false-positives from diverse sources of confounding, such as patterns of isolation-by-distance (IBD), alignment of environmental and genetic gradients, or cryptic relatedness (Frichot et al, 2013;Rellstab et al, 2015;de Villemereuil et al, 2015) and they probably outperform GS models in identifying environmentally-driven local adaptation (De Mita et al, 2013;Payseur and Rieseberg, 2016).…”