CLOSING THE GAP BETWEEN PALEONTOLOGICAL AND NEONTOLOGICAL SPECIATION AND EXTINCTION RATE ESTIMATES

Extant species and fossils are samples of the same underlying macroevolutionary process. However, a large discrepancy between diversification rates estimated from fossils and phylogenies has been widely documented for empirical datasets. In particular, phylogenetic estimates tend to be much lower than those inferred from the fossil record. This has led some critics to question whether phylogenetic methods can ever produce reliable estimates of diversification rates, despite theoretical work to the contrary, while others have pointed to incomplete sampling and biases in the data. Since biases in the fossil record or empirical phylogenies alone cannot be used to explain the magnitude of the observed discrepancies, a satisfactory theoretical explanation is lacking. Here, we explore a previously underappreciated phenomenon: the impact of different speciation modes (budding, bifurcating and anagenic) on diversification rates obtained using alternative data sources. We present an extended birth-death process that unifies different speciation modes, and assess the implications of this modeling framework using extensive simulations and nine empirical datasets. We show that the extended birth-death model has several important and predictable implications for estimates of diversification rates obtained using fossil versus phylogenetic data. For instance, even under perfect circumstances, i.e. complete sampling, paleontological and phylogenetic rates will only ever be equal if all speciation has occurred via budding. We develop a model-testing framework to establish empirical support for the extended birth-death process and validate the approach using simulations. These analyses highlight scenarios in which support for the model diminishes, including non-uniform sampling and diversification rate variation. However, we find support for the model in six empirical datasets. We also demonstrate that support for the model is negatively correlated with the degree of rate variation. Speciation and extinction rates are key parameters in macroevolutionary models, therefore defining and quantifying diversification rate parameters has wide-reaching implications for studies in this field, and for the integration of neontological and paleontological data.