Comparative phylogeny and the host shift scenario. If we accept current phylogenies, then a pollinating wasp of an ancestor of Chlamydodorae and Platyphyllae must have shifted to an ancestor of Cyathistipulae and Crassicostae, an event required to explain the discrepancy between ITS+ETS (left) and pollinator (right) phylogenies. However, under the hypothesis of a wasp suddenly changing host (grey arrows), there is no cytoplasm transfer and the Ficus lineage resulting from the hybridization necessary bears a chloroplast related to the Caulocarpae and Galoglychia lineages (middle bottom), a feature which is not observed in the cytoplasm phylogeny. On the contrary, if a wasp uses two hosts for a prolonged period of time (black arrows), repeated backcrosses may allow the introduction of a chloroplast originating in the Chlamydodorae plus Platyphyllae lineage inside the ancestor of Cyathistipulae and Crassicostae (middle top), as observed in the data set. The set of phylogenetic trees therefore support the role of an intermediate stage in which a pollinating wasp has consistently used two host species.