Ambiguous branch support comes from two major sources: indifference, i.e. lack of differentiating signal, and conflict, i.e. parts of the data prefer a different topological alternative. When doing phylogenetics, especially with (often inevitable) suboptimal data providing non-trivial signals as it was the case here, we may distinguish between these two sources using e.g. bootstrap consensus networks: quick-to-generate by-products of the standard tree-inference pipeline as outlined in this post at Genealogical World of Phylogenetic Networks using one of the data matrices provided with this study.