Search Results

Inference of phylogenetic networks - the evolutionary histories of species involving speciation as well as reticulation events - has proved to be an extremely challenging problem even for smaller datasets easily tackled by supertree inference methods. An effective way to boost the scalability of distance-based supertree methods originates from the Pareto (for clusters) property, which is a highly desirable property for phylogenetic consensus methods. In particular, one can employ strict consensus merger algorithms to boost the scalability and accuracy of supertree methods satisfying Pareto; cf. SuperFine. In this work, we establish a Pareto-like property for phylogenetic networks. Then we consider the recently introduced RF-Net method that heuristically solves the so-called RF-Network problem and which was demonstrated to be an efficient and effective tool for the inference of hybridization and reassortment networks. As our main result, we provide a constructive proof (entailing an explicit refinement algorithm) that the Pareto property applies to the RF-Network problem when the solution space is restricted to the popular class of tree-child networks. This result implies that strict consensus merger strategies, similar to SuperFine, can be directly applied to boost both accuracy and scalability of RF-Net significantly. Finally, we further investigate the optimum solutions to the RF-Network problem; in particular, we describe structural properties of all optimum (tree-child) RF-networks in relation to strict consensus clusters of the input trees.