“…The snmf model implements a fast yet accurate likelihood algorithm (Frichot, Mathieu, Trouillon, Bouchard, & François, 2014), while DAPC is a robust genetic clustering method with no assumption about the underlying population genetic model (Jombart & Ahmed, 2011). Based on previous population genomic studies for other co‐occurring plant species (Carvalho et al., 2019; Lanes et al., 2018; Silva et al., 2020), we tested from 1 to 10 ancestral populations ( k ). In the case of snmf we performed 10 replicate runs for each value of k , choosing the most likely k based on minimized cross‐entropy.…”