CRABS: Congruent rate analyses in birth-death scenarios

Höhna, Sebastian; Kopperud, Bjorn T. und Magee, Andrew F. (2022): CRABS: Congruent rate analyses in birth-death scenarios. In: Methods in Ecology and Evolution, Bd. 13, Nr. 12: S. 2709-2718

Abstract

Diversification rates inferred from phylogenies are not identifiable if the rates are allowed to vary freely over time. There are infinitely many combinations of speciation and extinction rate functions that have the exact same likelihood score for a given phylogeny, building a congruence class. The specific shape and characteristics of such congruence classes have not yet been studied. Whether speciation and extinction rate functions within a congruence class share common features is also not known. Prior hypotheses typically render diversification rates identifiable, but the results then depend on these a priori hypotheses and assumptions. To test the robustness of diversification results to these a priori hypotheses, we use two different approaches to explore congruence classes: (i) constructing congruent models under alternative hypotheses and (ii) sampling alternative rate function within the congruence class. Our methods are implemented in the open-source R package CRABS (). CRABS provides a flexible approach to explore the congruence class and provides summaries of rate functions within a congruence class. The summaries can highlight common trends, that is, increasing, flat or decreasing rates. Although there are infinitely many equally likely diversification rate functions, these can share common features. CRABS can be used to assess if diversification rate patterns are robust despite the non-identifiability of the birth-death model. In our example, we clearly identify three phases of diversification rate changes that are common among all models which we sampled from the congruence class. Thus, congruence classes are not necessarily a problem for studying historical patterns of biodiversity from phylogenies.

• Dokumententyp: Zeitschriftenartikel
• Fakultät: Geowissenschaften > Department für Geo- und Umweltwissenschaften
• Themengebiete: 500 Naturwissenschaften und Mathematik > 550 Geowissenschaften, Geologie
• ISSN: 2041-210X
• Sprache: Englisch
• DFG: Gefördert durch die Deutsche Forschungsgemeinschaft (DFG) - 491502892