@alex_glgz Both Fig. S3A and B are "neighbour-net splits graphs", a neighbo[u]r-net is a planar (2-dim.) network that can serve as a phylogenetic network in a general sense, being a graph depicting (alternative) phylogenetic relationships. In that respect A is pointless: too noisy input data -> unrepresentative distance matrix = spider-web. B is not a haplotype network but a NNet based on a more meaningful distance matrix.
See also: How to interpret splits graphs: https://phylonetworks.blogspot.com/2012/08/how-to-interpret-splits-graphs.html