I enjoyed reading the manuscript by Herrera-Alsina et al. on “Accounting for extinction dynamics unifies the geological and biological histories of southeast Asia”. I am not a specialist on southeast Asian biogeography, but I’m familiar with the general questions and approaches.

I appreciate that the authors explored how extinction (low, intermediate, and high) affects biogeographical reconstructions. I see this primarily as an analytical exercise, that is worthwhile doing, because it remains generally unclear how extinction affects phylogenies and the inferred processes (e.g., diversification, trait evolution, biogeographical reconstructions). What I’m less sure about, is whether it can provide answers to the primary question/conclusion of the study, i.e., whether it convincingly shows a potential true match between biological and geological histories. To me, this would require a more quantitative approach, e.g. using fossil data, rather than a modelling approach (or in addition to, perhaps). Discussing this, and providing more lines of evidence (e.g. from fossils) in the discussion for the different clades, could help.

Related to this, the study lacked clear hypotheses/predictions, which made it difficult to follow the analytical approach and results/discussion. For example, was the broader ancestral range (i.e., multiple ancestral areas, and which ones?) expected, given our knowledge on the geology of the region? The issue here is, that this would be expected purely based on the modelling approach (i.e., more extinction means by nature more uncertainty in ancestral ranges), so it would be complicated to disentangle the true history from the neutral model expectation. Perhaps the only question/hypothesis that can really be addressed with the approach and data is that extinction affects model inferences. However, setting up a few clear expectations to evaluate the match in biological and geological histories, would be helpful. For example, to show that intermediate/high extinction leads to more consistency between biological and geological histories. It would help to introduce the expected consistencies (e.g. which regions, the timing, connectivity/dispersal, processes of vicariance/in situ speciation, etc.) more clearly. More implicit is the expectation that the distribution is older than previously inferred, but a clear and explicit expectation linked to this could help structure the paper as well.

L 141 Results. Please briefly explain how extinction was modelled. This is essential for the further interpretation and understanding of the results (without having to read the methods).

The Results can generally be better structured, e.g. along the line of a few expectations. Now it reads as a sum-up of differences between extinction models and clades, but it is unclear what was expected, and what was perhaps surprising. This would also help the discussion. (speciation is mentioned frequently here as well, I am not sure what was expected here, in terms of vicariance vs. in-situ speciation. This needs more introduction and an expectation linked to it, ideally).

Table 1: This is useful, but it would be easier to read as a figure. E.g. with bubbles in size relative to the estimated parameter? So that it is easier to see differences between extinction regimes and clades?

Figure 2: why relative times, why not absolute times? I think it would be interesting to see the congruence or disparity between clades… this would also more closely link with the geological history, which is the same for all clades (and perhaps could be visualized somehow on this same time axis..?)

Methods: please give some more details on the used phylogenies (e.g. sampling, dating, …) and which areas were used for the biogeographical reconstructions (was this the same for each clade? did this follow the original study? could species occupy multiple areas, was there a limit to the number of areas occupied for any ancestral node? etc.). Did you use a single or multiple phylogenies?

Data: I understand that data was not fully provided, only raw code. I understand this - to some extent - but it does not help with the reproducibility. I would recommend the authors to ask permission to also upload the data, so that analyses could be repeated.