No phylogenomic support for a Cenozoic origin of the "living fossil" Isoetes

Abstract

Premise: The isoetalean lineage has a rich fossil record that extends to the Devonian, but the age of the living clade is unclear. Recent results indicate that it is young, from the Cenozoic, whereas earlier work based on less data from a denser taxon sampling yielded Mesozoic median ages.

Methods: We analyzed node ages in Isoetes using two genomic data sets (plastome and nuclear ribosomal cistron), three clock models implemented in MrBayes (ILN, WN, and TK02 models), and a conservative approach to calibration.

Results: While topological results were consistently resolved in Isoetes estimated crown group ages range from the latest Paleozoic (mid-Permian) to the Mesozoic depending on data type and clock model. The oldest estimates were retrieved using the autocorrelated TK02 clock model. An (early) Cenozoic age was only obtained under one specific condition (plastome data analyzed with the uncorrelated ILN clock model). That same plastome data set also yielded the oldest (mid-Permian) age estimate when analyzed with the autocorrelated TK02 clock model. Adding the highly divergent, recently established sister species Isoetes wormaldii to the data set approximately doubled the average median node depth to the Isoetes crown group.

Conclusions: There is no consistent support for a Cenozoic origin of the living clade Isoetes. We obtained seemingly well-founded, yet strongly deviating results depending on data type and clock model. The single most important future improvement is probably to add calibration points, which requires an improved understanding of the isoetalean fossil record or alternative bases for calibration.

Keywords: Bayesian inference; Isoetaceae; Isoetales; Mesozoic; Thorne and Kishino model (TK02); clock model; independent lognormal model (ILN); lycopods; node ages; white noise model (WN).