Illustrating phylogenetic placement of fossils using RoguePlots: An example from ichneumonid parasitoid wasps (Hymenoptera, Ichneumonidae) and an extensive morphological matrix

Abstract

The fossil record constitutes the primary source of information about the evolutionary history of extant and extinct groups, and many analyses of macroevolution rely on fossils that are accurately placed within phylogenies. To avoid misinterpretation of the fossil record, especially by non-palaeontologists, the proper assessment and communication of uncertainty in fossil placement is crucial. We here use Bayesian morphological phylogenetics to evaluate the classifications of fossil parasitoid wasps (Hymenoptera, Ichneumonidae) and introduce ‘RoguePlots’ to illustrate placement uncertainty on the phylogeny of extant taxa. Based on an extensive, newly constructed morphological matrix of 222 characters in 24 fossil and 103 extant taxa, we test three different aspects of models of morphological evolution. We find that a model that includes ordered characters, among-character rate variation, and a state-space restricted to observed states achieves the highest marginal likelihoods. The individual RoguePlots reveal large differences in confidence in the placement of the different fossils and allow some refinements to their classification: Polyhelictes bipolarus and Ichninsum appendicrassum are moved from an uncertain subfamily placement to Pimplinae, Plectiscidea lanhami is transferred to Allomacrus in Cylloceriinae ( Allomacrus lanhami, comb. nov.), Lithotorus cressoni is moved from Diplazontinae to Orthocentrinae, and we note uncertainty in the generic placement of Xanthopimpla? messelensis. We discuss potential artefacts that might result in biased posterior probabilities in Bayesian morphological phylogenetic analyses, pertaining to character and taxon sampling, fossilization biases, and model misspecification. Finally, we suggest future directions both in ichneumonid palaeontology, in the modelling of morphological evolution, and in the way Bayesian phylogenetics can improve both assessment and representation of fossil placement uncertainty.