Since natural habitats are disappearing fast, there is an urgent need to collect, characterize, and phenotype banana (Musa spp.) crop wild relatives to identify unique genotypes with specific traits that fill the gaps in our gene banks. We report on a collection mission in Papua New Guinea carried out in 2019. Seed containing bunches were collected from Musa peekelii ssp. angustigemma (N.W.Simmonds) Argent (3), M. schizocarpa N. W. Simmonds (4), M. balbisiana Colla (3), M. acuminata ssp. banksii (F. Muell.) Simmonds (14), M. boman Argent (3), M. ingens Simmonds (2), M. maclayi ssp. maclayi F.Muell. ex Mikl.‐Maclay (1), and M. lolodensis Cheesman (1). This material, together with the seeds collected during a previous mission in 2017, form the basis for the development of a wild banana seed bank. For characterization and phenotyping, we focused on the most ubiquitous indigenous species of Papua New Guinea: M. acuminata ssp. banksii, the ancestor of most edible bananas. We calculated that the median genomic dissimilarity of the M. acuminata ssp. banksii accessions was 4% and that they differed at least 5% from accessions present in the International Transit Centre, the world's largest banana gene bank. High‐throughput phenotyping revealed drought avoidance strategies with significant differences in root/shoot ratio, soil water content sensitivity, and response towards vapor pressure deficit (VPD). We deliver a proof of principle that the wild diversity is not yet fully covered in the gene banks and that wild M. acuminata ssp. banksii populations contain individuals with unique traits, useful for drought tolerance breeding programs.