Reconstructing evolution at the community level: A case study on Mediterranean amphibians.

Reconstructing reliable timescales for species evolution is an important and indispensable goal of modern biogeography. However, many factors influence the estimation of divergence times, and uncertainty in the inferred time trees remains a major issue that is often insufficiently acknowledged. We here focus on a fundamental problem of time tree analysis: the combination of slow-evolving (nuclear DNA) and fast-evolving (mitochondrial DNA) markers in a single time tree. Both markers differ in their suitability to infer divergences at different time scales (the 'genome-timescale-dilemma'). However, strategies to infer shallow and deep divergences in a single time tree have rarely been compared empirically. Using Mediterranean amphibians as model system that is exceptional in its geographic and taxonomic completeness of available genetic information, we analyze 202 lineages of western Palearctic amphibians across the entire Mediterranean region. We compiled data of four nuclear and five mitochondrial genes and used twelve fossil calibration points widely acknowledged for amphibian evolution. We reconstruct time trees for an extensive lineage-level data set and compare the performances of the different trees: the first tree is based on primary fossil calibration and mitochondrial DNA, while the second tree is based on a combination of primary fossil and on secondary calibrations taken from a nuclear tree using mitochondrial DNA (two-step protocol). Focusing on a set of nodes that are most likely explained by vicariance, we statistically compare the reconstructed alternative time trees by applying a biogeographical plausibility test. Our two-step protocol outperformed the alternative approach in terms of spatial and temporal plausibility. It allows us to infer scenarios for Mediterranean amphibian evolution in eight geographic provinces. We identified several tectonic and climatic events explaining the majority of Mediterranean amphibian divergences, with Plio-Pleistocene climatic fluctuations being the dominant driver for intrageneric evolution. However, often more than one event could be invoked for a specific split. We give recommendations for the use of secondary calibrations in future molecular clock analyses at the community level.