Estimating time-dependent rates of speciation and extinction from dated phylogenetic trees of extant species (timetrees), and determining how and why they vary, is key to understanding how ecological and evolutionary processes shape biodiversity. Due to an increasing availability of phylogenetic trees, a growing number of process-based methods relying on the birth-death model have been developed in the last decade to address a variety of questions in macroevolution. However, this methodological progress has regularly been criticised such that one may wonder how reliable the estimations of speciation and extinction rates are. In particular, using lineages-through-time (LTT) plots, a recent study (Louca & Pennell, 2020) has shown that there are an infinite number of equally likely diversification scenarios that can generate any timetree. This has led to questioning whether or not diversification rates should be estimated at all. Here we summarize, clarify, and highlight technical considerations on recent findings regarding the capacity of models to disentangle diversification histories. Using simulations we illustrate the characteristics of newly-proposed "pulled rates" and their utility. We recognize that the recent findings are a step forward in understanding the behavior of macroevolutionary modelling, but they in no way suggest we should abandon diversification modelling altogether. On the contrary, the study of macroevolution using phylogenetic trees has never been more exciting and promising than today. We still face important limitations in regard to data availability and methodological shortcomings, but by acknowledging them we can better target our joint efforts as a scientific community.

Fig2.R allows for the simulation of examples of unidentifiability in a simple dataset using an MCMC chain. MCMC.Rdata contains the associated files for an MCMC simulation run.

Fig3.R and Fig4.R details the simulations of phylogenetic trees, LTT plots under scenarios of constant diversification rates and a single increase in speciation rate respectively. Corresponding simulation data can be found in constantRates.Rdata and rateShift.Rdata.

Fig5.R provides code for simulating three different commonly-investigated scenarios of diversification rate change over time, and a final more complex and unrealistic scenario.