I have decided not sending the paper again to reviewers in order to speed up the process, but a few important modifications are still necessary before the paper could be recommended.

Regarding the biogeographic models: I found satisfying that the results of the unconstrained model are presented together with the results in which jump dispersal was forced to occur between Europe and the Cape region. Although I am still sceptical with this assumption, I agree this approach could be valid if we assume that widespread ancestral reconstructions are biologically unlikely. I regret, however, the authors don’t provide a more detailed justification of this assumption; if they argue in the introduction that ancestral continuous (widespread) distributions fragmented by vicariance probably explain disjoint patterns of (arid-adapted) African plant groups (e.g., Sanmartín et al., 2010; Pokorny et al., 2015; Bellstedt et al., 2012), why in this particular case this biogeographic scenario (an ancestral widespread distribution) does not make biological sense? I imagine it has something to do with the temperate adaptations of the genus and the fact that temperate conditions were never continuous in Africa, but this needs to be clarified (maybe in Appendix 3 or in the main text?).

Reviewer 3 was concerned for the potential correlation of the studied variables (i.e. geographic distances and niche similarity). In this new version, the authors evaluated their correlation and found what is generally considered a moderate to strong value of correlation (Kendall’s R = -0.64), but was this result taken into account at all? It has strong implications, as one of the main questions of the study (i.e. the relative importance of niche vs. distance in dispersal patterns) cannot be answered. I am convinced this limitation does not affect the main conclusions of the study since the preferred model was the combined “niche/distance” model. In addition, this is an interesting result “per se”. However, in light of this, I have the impression that comparisons of the “geographic distance” and the “niche similarity” models do not make sense anymore because these variables are not independent. Only comparisons of the null and the combined “niche/distance” model might remain informative. I leave to the authors the decision on whether they want to exclude these models from the model comparison table or not, but this limitation needs to be taken into account and the results/discussion sections modified accordingly.

Again, I hope you will find these last comments helpful and look forward to read the final version of the manuscript!

Best Andrea

This paper has improved from the last version and the authors have taken into consideration and answered many of the concerns raised in the initial review. Reviewer 1 believes it is in good shape for recommendation after a few minor revisions. Reviewer 2 did not see the original, but only the revised version and agrees with Rev 1 that the paper is of general interest. However, Rev 2 has some substantive comments that need to be dealt with before the paper can be accepted for recommendation. The most critical comments centre on the niche modelling approach and how this affects your results. Rev 2 has a number of other critical but supportive and constructive comments that will make the paper more useful for the biogeographic community.

The authors’ response to my critique of their use of the DEC+J models is satisfying. I agree that this ms is not the place to go into the detail of this argumentation, although I appreciate they acknowledge the controversy and provide the DEC results. Just as a note of caution, the recent paper of Ree & Sanmartin (2018) shows that DEC and DEC+J models are not directly comparable using statistical methods such as AIC that assume probabilistic equivalency of events. Therefore, I recommend privileging arguments based on empirical (biological, geographic) considerations rather than AIC scores. AIC comparisons might otherwise work to compare DEC (and DEC+J) models among them.

I regret however, the authors have not fully taken into account my comments about the adjacency matrix. I apologize if this is because I was not clear enough. The “Cape to Cairo” hypothesis (dispersal from Europe to the Cape, followed by migrations to the Drakensberg; Figure 1), could not be properly tested if the adjacency matrix does not allow ancestral connections between Europe and the Cape regions (0 values on this connection). By setting 0 values you are specifying that these two areas were not connected in the past (an ancestral distribution in Europe and the Cape is disallowed). Hence, although in theory dispersal is still allowed, in practice this constraint forces species to follow another route to disperse between these areas or forces jump dispersal. In my opinion this is a too strong assumption when the “Cape to Cairo” is the hypothesis to test, specially given the possibility that before the Miocene aridification of northern Africa, Erica was distributed in regions where it does not occur today. This limitation seems to apply only to DEC+J and not to DEC analyses where the adjacency matrix was not implemented. I suggest testing again the hypotheses without this constriction, which will make the results more solid.

I also have a few minor points: - The biogeographic reconstructions on Appendix 13 don’t have any legend. You need to explain the symbols on the trees, the colours, the abbreviations or the analyses that have produced these figures. In addition, the size of the pie charts on the trees is way too big and needs to be reduced. - The unconstrained biogeographic DEC and DEC+J reconstructions need to be presented the Appendix. - Please, clarify if the “Cape to Cairo” model fits the data better than “Southerly stepping stone” model. Discussion lines 285-287 state: “Cape to Cairo” and “Drakensberg melting-pot” mostly fit the data better than “Southerly stepping stone”. Meanwhile, in the results: “Under DEC+J given the best tree, the Drakensberg melting pot, geographic distance, and southerly stepping stone models revealed the lowest AIC; under DEC the Drakensberg melting pot model alone scored best, but with higher AIC ”. Moreover, in Table 1, the CtoC model does not even appear among the best models when considering the best tree.

I hope that the authors will find these comments helpful to improve their work.

Yours sincerely,

Andrea S. Meseguer

All the reviewers now chimed in with their opinions and agreed on the interests of the manuscript. I commend the authors for all the work they have done, and I echo the reviewers. I think this study represents a nice piece of work investigating the factors and processes mediating dispersal of plant clades from Europe to Africa. I very much appreciated the thoroughness of the analyses and the clarity of the text.

I recommend the authors undertake a thorough revision based on the constructive comments of the reviewers, taking particular care to address the reviewerʼs methodological concerns. I found particularly interesting the criticism of one reviewer regarding the conceptual problem of proposing hypotheses that are not mutually exclusive (or framed at different levels), as well as all methodological suggestions proposed by the three reviewers.

I am, however, a bit more cautious than the reviewers regarding the results of this study. In agreement with one reviewer, I regret the choice of biogeographic models. The DEC+J model has been shown at best to not be directly comparable to the DEC model, and at worst to present some statistical problems (Ree & Sanmartin 2018). I would recommend excluding this model from the comparisons and if the authors still wanting to present it, to do it on supplementary material. I am also suspicious with the DEC+* model of Massana & al. (2015). This model has been published as a preprint with no reviewer assessment on their quality/performance. I thus suggest excluding this model from the comparisons as well.

In addition, I have concerns on the validity of the results. In addition to the different disperal matrices, you implemented an adjacent area matrix to constraint the maximum number of areas allowed as ancestral states. I agree on this procedure to decrease model uncertainty, for example as you did by using the “maximum number of areas” command. However, I think the constrictions on the adjacent matrix you implemented are problematic and could have affected the results: 1) my apologies if I’m wrong, but I think that by using this adjacency matrix you force stepping stone dispersal to occur, while this is one of the main hypotheses you want to test. For example, in the matrix on Appendix 3 you are impeding dispersal between Europe and the Cape region, the Drakensberg and Madagascar, while only allowing dispersal through tropical Africa. It is thus not surprising that you found support for the “Drakensberg melting pot” steeping stone scenario in comparison with other long distance dispersal models. 2) In addition, by implementing this matrix you automatically disallow disjoint distributions at ancestral nodes, decreasing the likelihood of extinction to occur.

An important part of the manuscript focuses on whether colonization was mediated by niche changes or occurred across similar habitats. On this regard one reviewer had concerns about the areas used for comparison. I agree with him that differences between study areas could dissapear when compared with other regions in Africa or Europe where Erica does not occur. I additionally regret you didn’t differentiate the Northern Hemisphere Mediterranean region from other Northern Hemisphere regions. My guess is that the low climatic similarity between southern African and Europe might most likely apply to the Eurosiberian region, but not to the Mediterranean one. I think this differentiation is important to test if long distance dispersals involved niche shifts.

Concerning model comparisons, in addition to AIC scores (e.g. on Appendix 8), I would like to see the differences in deltaAIC values, akaike weights or any other metric that allows to evaluate model improvements and perform model choice. Generally, it is the differences between the likelihoods or AICs that matter, not their absolute values. That is, the larger difference in AIC indicates stronger evidence for one model over the other (Burnham and Anderson, 2002). Delta (AIC differences) within 0-2 has a substantial support for a suboptimal model; delta within 4-7 considerably less support and delta greater than 10 essentially no support.

To conclude, apart from the nexus file I would suggest the author to include a figure with the most likely biogeographic reconstruction, maybe on SI.

I leave this, and the reviewer’s comments, for the authors to consider as they revise and improve the paper. I hope the authors will find that many of these will be helpful in improving the manuscript