2.1 | Sampling
Two sampling campaigns were conducted in Cuba in 2011 and in 2013 in three different locations (Villa Clara, Ciego de Avila and Matanzas) located between 20 and 300 km apart using the same paired population sampling design (Figure 1, Table 1). Infected banana leaves were collected in a banana plantation pair in each location except in Ciego de Avila, where two pairs were sampled in 2013. The same varieties were collected in all pairs; one plot was planted with a susceptible variety (Macho ¾ belonging to the banana AAB genomic group and the plantain subgroup) and the other with a resistant variety (the tetraploid FHIA 18 belonging to the banana AAAB genomic group). More than 1 000 strains were isolated representing 14 populations with a number of isolates per population ranging between 38 and 135. The six populations sampled in 2011 had already been analyzed to investigate local adaptation to banana quantitative resistance and the underlying genetic basis in P. fijiensis (Dumartinet et al., 2019; Carlier et al., 2021b). To further describe adaptive architecture, in this study, we added eight more populations sampled in 2013 in the same locations. Banana plantations are frequently replanted and different plantation pairs were collected in the two first locations in 2011 and in 2013. In Matanzas, the same plantations were sampled but they were replanted between 2011 and 2013. The data obtained from samples collected in the two years were then considered as spatial replicates but not as time series.