Recent progress in resolving the tree of life continues to expose relationships that resist resolution, which drives the search for novel sources of information to solve these difficult phylogenetic problems. A recent example, the presence and absence of microRNA families, has been vigorously promoted as an ideal source of phylogenetic data and has been applied to several perennial phylogenetic problems. The utility of such data for phylogenetic inference hinges critically both on developing stochastic models that provide a reasonable description of the process that give rise to these data, and also on the careful validation of those models in real inference scenarios. Remarkably, however, the statistical behavior and phylogenetic utility of microRNA data have not yet been rigorously characterized. Here we explore the behavior and performance of microRNA presence/absence data under a variety of evolutionary models and reexamine datasets from several previous studies. We find that highly heterogeneous rates of microRNA gain and loss, pervasive secondary loss, and sampling error collectively render microRNA-based inference of phylogeny difficult. Moreover, our reanalyses fundamentally alter the conclusions for four of the five studies that we reexamined. Our results indicate that the capacity of miRNA data to resolve the tree of life has been overstated, and we urge caution in their application and interpretation.
