Supertrees and the Tree of Life: generating a metaphylogeny for a diverse invertebrate family (Insecta:Diptera:Therevidae) using constraint trees and the parsimony ratchet to overcome low taxon overlap

The dipteran family Therevidae (stiletto flies) is cosmopolitan and has been the focus of many taxonomic and phylogenetic studies over the last 25 years. Despite this work, questions remain concerning the relationships between subfamilies, genera and generic groups and membership of those groups. We use the supertree method to produce an inclusive phylogeny for the family Therevidae from 24 phylogenetic studies using matrix representation with parsimony (MRP) analysis. The supertree method, one of the most common approaches to calculating globally inclusive phylogenies from smaller more exclusive analyses, produced the therevid metaphylogeny despite only 34% of the terminal taxa being found in more than one source tree. We describe a method for handling low taxon overlap in supertree analyses, in combination with the parsimony ratchet and constraint tree techniques. The supertree presented here is an overarching phylogenetic hypothesis of the Therevidae, incorporating extensive sampling of major lineages and summarising past phylogenetic work on the family. The inclusive metaphylogeny for 362 therevid taxa robustly retrieves the subfamilies Agapophytinae, Phycinae, Therevinae and Xestomyzinae, and the tribes Cyclotelini and Therevini. The Phycinae and Xestomyzinae form a clade, sister to the remaining Therevidae. The Australasian and South American Taenogera Kröber genus-group is monophyletic and sister to a clade of Therevinae and the Australian endemic Agapophytinae. The Therevinae consists of the Anabarhynchus Macquart genus-group of Australian, South American, New Caledonian and New Zealand taxa as sister to the non-Australasian ‘higher Therevinae’, which contains the tribes Cyclotelini and Therevini. The Therevini includes the Hoplosathe Lyneborg & Zaitzev, Litolinga Irwin & Lyneborg, Baryphora Loew, Pandivirilia Irwin & Lyneborg and Thereva Latreille generic-groups. MRP supertree methods can be used to produce inclusive metaphylogenies in situations where source trees have poor data overlap and low taxon overlap, and are therefore valuable in species-rich groups such as arthropods. These methods may be necessary for constructing the ‘Tree of Life’, representing phylogenetic relationships among the millions of known species. However, our analyses show that in situations of source tree conflict, MRP supertree analyses present only the majority signal. We also show that conflict between source trees can be hidden in MRP supertrees, thus our results emphasise the need to evaluate the resulting clades with reference to the source trees.