Molecular dating and eucalypts: reply to Ladiges and Udovicic

Abstract

Our paper (Crisp et al. 2004) referred to by Ladiges and Udovicic (2005) is a broad review of the mid-Cenozoic radiation of the Australian sclerophyll flora from the perspective of molecular phylogenies. We presented new chronograms of four taxa (eucalypts,Banksia, Casuarinaceae and Fabaceae) with the intention of making broad comparisons among taxa and across climatic events. We found that eucalypts showed an exponential radiation that apparently continued through the mid-Cenozoic period of climatic change, when the other sclerophyll taxa examined also were radiating. However, eucalypts were not the primary focus of our study and we used a published dataset with limited sampling within closely related species groups, such as the tropical boxes. It was not necessary to add taxa to this sample because only a broad view of deeper divergences was required. Therefore, estimates of more recent divergences (last 10 million years) would have been very unreliable. Our study used one region of DNA (ITS) because this was the only dataset available with broad and balanced sampling across the eucalypts. Calibration was problematic because no well-dated fossil that can be placed accurately on a tree is known. Therefore, we used a presumed vicariance event between Arillastrum in New Caledonia and eucalypts in Australia, although use of a single calibration point may be prone to large errors (Graur and Martin 2004). We presented a table comparing dates estimated for certain nodes from our chronogram and from Ladiges et al. (2003), because the latter paper used the same calibration point. This comparison was intended to indicate the uncertainty of estimates from both sources, and the need for further research. We did not express a preference for either set of estimates. Specific dates for divergences within eucalypts were beside the main focus of our paper, and therefore these comparisons were not discussed in the text. Ladiges and Udovicic (2005) state that our divergence date estimates for the eucalypts ‘vary several fold compared with estimates based on other data’ but they present no evidence to support this statement, except for a single point of comparison with respect to E. deglupta (below). As observed by Ladiges and Udovicic, we did not discuss topological differences between our chronogram of eucalypts (Crisp et al. 2004, fig. 3) and trees in other studies cited by Ladiges and Udovicic. Taxonomic details were irrelevant to our main questions which, with respect to eucalypts, were: (1) whether molecular phylogenies showed differences among taxa in their pattern of radiation during the Cenozoic, as predicted by the fossil record; (2) whether the major climatic and geological events of the Cenozoic were reflected in the phylogenies of taxa present during the changes; (3) whether taxa in the same biome exhibit the same macro-evolutionary patterns; and (4) whether radiations were monophyletic within biomes or showed reticulation across biomes. When we estimated dates on nodes, we took topological uncertainty into account, as we explained on page 1555 and showed by the grey bars on the chronograms. We found topological variation among four phylogenies of eucalypts estimated from the ITS dataset (Steane et al. 2002) using combinations of methods (maximum likelihood and Bayesian inference), data (ambiguous sites included or not) and topological constraints (see below). Nodes that differed among trees included monophyly of Eucalyptus s. str., monophyly of the eucalypts s.l. (Eucalyptus plus Corymbia, Angophora, Stockwellia, Allosyncarpia and Eucalyptopsis), and the relationship of Stockwellia, Allosyncarpia and Eucalyptopsis (discussed by Ladiges and Udovicic). Ladiges and Udovicic suggest that our date for the divergence of Eucalyptus deglupta from its Australian sister taxon is too old. We accept that this is probably true and it should not be surprising given the limitations of our study stated above and the many sources of error. For example, there may have been smoothing artefacts because, although the ME2 sequence of E. deglupta was placed as sister to the rest of E. deglupta, it had a near zero-length terminal branch. The detailed hypothesis of relationships within the tropical boxes presented by Ladiges and Udovicic is interesting but was not needed for themain questions addressed in our paper. We did not discuss the relationships between E. deglupta and other taxa.