This study addressed an interesting question of the effects of multi-level processes on the geographic distribution of forest woody plant sexual diversity, which helps to advance the understandings of the underlying drivers of geographical pattern of forest plant diversity. The manuscript is overall clearly written. However, at the moment I still have a couple of concerns and some minor comments, listed as below,   

Major concerns:



Comment 1: According to my understanding, "region" in the text means continent-level landmass (Lines 123-124) and its effects on shaping plot-level plant sexual diversities were important (Lines 161-162), while latitude effects were unimportant when other factors' effects were controlled (Lines 155-157). It would be better not to group the regions as either tropical regions or temperate regions, which is confusing.

Response 1: Thank you for this comment. In the analysis, we grouped regions based on continent-level landmass. In the abstract, we referred to tropical and temperate regions in the context of previous debates (Lines 52-55). This might have led the referee to believe we grouped the regions as either tropical regions or temperate regions. In the revision, we rephrased the results in the abstract regarding regions as follows:

“Our results showed that plants were more likely to be dioecious than hermaphroditic in Oceania and Tropical Asia, but were more likely to be monoecious than dioecious in Europe and North America compared with Tropical Africa.”

We have updated this result in the Abstract (Line 14-16).

Comment 2: As the authors stated in the Introduction, if (a) dioecious is more likely associated with dry or poor resource habitats (Lines 61-56), and (b) areas with younger species age is expected to have a higher incidence of dioecy (Lines 55-57), then an interactive effect likely exists between (a) and (b) on uplifting the incidence of dioecy. Have the authors considered of including this interactive term in the regression models? 

Response 2: Thanks for the suggestion. In the revision, we have added the interactive term between mean species age and mean annual precipitation in the models (Table 1), which showed marginally significant effects on the proportion of plant sexual systems. We added the results in the revisions (Line 177-179).

Minor comments:



Comment 3: Lines 14-16   I am not sure this finding (not significant when the effects of climate etc. were controlled as shown in the Results at Lines 155-157) supports Baker's law. Lack of dioecy in temperate regions can be a result of biotic and/or abiotic filters, apart from the expected low colonization capacity of dioecious plants. 

Response 3: Agreed. We have rephrased the text to highlight the role of region factors rather than Baker’s Law. For example, in the abstract, we concluded that “Our results showed that plants were more likely to be dioecious than hermaphroditic in Oceania and Tropical Asia, but were more likely to be monoecious than dioecious in Europe and North America compared with Tropical Africa.” (Lines 14-16).

Comment 4: Lines 40-41   Mathematical inference? What is the underlying biological/ecological meaning?

Response 4: Thanks a lot for this valuable comment. According to this suggestion, in the revision, we have rephrased the sentence as:

“The proportion of hermaphroditism might be higher in tropical forests than in temperate forests as precipitation and the proportion of biotic pollination decrease with latitude.”

Comment 5: Line 43   replace "showed" with "show"?

Response 5: Corrected.

Comment 6: Lines 53-55   I am not sure this expectation holds, if the effects of abiotic and biotic filters have not been ruled out first. Besides, I think the authors meant that dioecious plant species would have lower relative incidence in temperate than tropical regions compared with other sexual systems plant species, due to the expected low colonization rates of dioecious plants in temperature regions. 

Response 6: We intended to mean dioecious species would have lower relative incidence in temperate than other sexual systems supposing other conditions are the same (Lines 53-57). We revised the sentence as:

“Supposing abiotic and biotic conditions are equal, dioecious plant species would have lower relative incidence in temperate regions, e.g., Europe and North America than in tropical regions, e.g., Tropical Asia and Oceania, compared with other sexual systems, due to the expected low colonization rates of dioecious plants in temperate regions.”

Comment 7: Line 81   what does "PET" stand for?

Response 7: PET stands for annual potential evapotranspiration. We removed PET here as it was not retained in the final mode.

Comment 8: Lines 85   should be "forest dynamics plots"?

Response 8: Yes.

Comment 9: Lines 112, 132   replace "number of trees" with "number of stems"?

Response 9: We used number of trees.

Comment 10: Lines 127-134  A common format is that R package name be in italics and R function name in quotation marks, please check and be consistent in the text.

Response 10: Addressed.

Comment 11: Line 180   add a "be" between "to" and "hermaphrodite".  replace "Fig. 5, S4" with "Fig. 5, S5"?

Response 11: Corrected.

Comment 12: Line 181  "Fig. S5" should be replaced with "Fig. S4 "?

Response 12: We updated the figure in the appendix and revised it in the main text accordingly.

Comment 13: Lines 196-199   These two findings contradict each other. Why low colonization capacity is important in one case but not important in another? 

Please elaborate more on this.

Response 13: In the revision, we found island effects were also important after including the interactive term (between species age and precipitation) and plot characteristics (number of species and number of trees). Plants were more likely to be monoecious than dioecious in island communities (Table 1), which did not contradict the effects of region (Lines 203-204).

Comment 14: Lines 221-222   How do these results support Baker's law?

Response 14: We appreciate your question. It was a mistake. We revised the sentence as “These results suggested evolutionary processes diminish the effect of Baker’s law” (Lines 224-225).

Comment 15: Lines 224-226 It is not clear what regional processes or climate factors, can the authors give an example?

Response 15: The effects of age could depend on precipitation as you suggested (Lines 227-229). We revised the sentence as follows:

“The effect of species age on plant sexual systems was different at regional and global scales (Wang et al., 2020), which could be explained by the interactive effects between species age and precipitation across different regions.”

Comment 16: Line 229   replace "less" with "more".

Response 16: Corrected.

I thank the authors for their interesting study. I have enjoyed reading the manuscript, which is mostly presented clearly and concisely, and I find the emergence of biogeographic patterns in plant reproductive traits quite exciting, personally. I have nevertheless a few general points that I think need the author's attention:

Comment 1: I am a bit unsure about the chosen modelling strategy. At the very least, some more details should be given (in the appendix), like the correlation between the different variables, and the results of the VIF analyses. I also have some doubts that fitting “regions” is the most meaningful variable to fit. I wonder if it wouldn’t make more sense, at least as an alternative approach, to model this variable as a random factor in a mixed model.

Response 1: Thank you for the suggestions. We added tables for the pairwise correlation among climate variables and VIF analyses in the appendix. We agree that region could be treated as a random effect if we were not interested in the effect of region. However, region is a major factor we are interested with as in many similar global biogeographic studies. We need to include that variable. It was retained by model selection procedure as the most important variable.

Comment 2: Another aspect that I think is somewhat problematic, is the apparent overfitting of the models. As can be seen in all the graphs of the logistic models, the blue line fits almost individual data points, which is also expressed in the relatively high variances explained in the different models. I think that this “better” fit clearly comes at the cost of interpretability. I suspect that also here the region variable plays a major role, which again makes me wonder if it wouldn't be better to fit it as a random factor. Another point in this context is that you present certain variables as not significant in the models, but then still visualize their relationships with sexual system in the presented graphs (e.g. latitude etc.).

Response 2: The region variable is of major interest in global biogeographic study and it has been constantly shown to have significant effects on species richness and functional traits (Ricklefs and He 2016, Zhang et al. 2016). That is also the case for our study. Our model selection processes also retained region as the most important variable in affecting the global distribution of plant sexual systems (Table 1, S2). We moved the graph with latitude to the appendix because it has been used in previous studies (Baker and Cox 1984). We further excluded other graphs with variables as not significant in the models.

Comment 3: In some places in the discussion I have problems following what you are trying to say, and you can find those in my detailed comments below.

Response 3: We revised the discussion to make it clearer. Please see our explanations below.

Detailed comments:

Comment 4: Line 15: Throughout the manuscript, you are constantly confusing the use of adjectives and nouns, when referring to the sexual systems. Please make sure that you use dioecious/dioecy, monoecious/monoecy, hermaphroditic/hermaphroditism appropriately.

Response 4: We have checked and corrected the usage throughout the manuscript.

Comment 5: Line 112, “plot characters”: These seem to me two important potentially confounding variables, and I don't understand why they have not been kept in the models (no matter their VIF). In addition, I would also think species richness in the plots could be another important covariate. 

Response 5: Thanks for this suggestion. In the revision, we kept the plot characters (number of species, number of trees) in the final models (Table 1).

Comment 6: Line 128f: I don't understand how you can estimate lambda for a discrete trait, as lambda is based on a Brownian motion model of trait evolution, which is models continuous traits. Could you please comment and specify how that works for discrete traits? Could that also be the reason that you get either 0 or very high values for lambda?

Response 6: We used the “fitDiscrete” function in the R package geiger for discrete traits, which could fit various likelihood models for discrete character evolution (https://cran.r-project.org/web/packages/geiger/index.html). As the effects of lambda were not significant, we did not include it as a variable in the model.

Comment 7: Line 136ff: This selection procedure based on variance inflation, while maybe statistically valid, is somewhat elusive. First, it would be good to see the results of this in the appendix, also to get a better overview of what variables were used and which were excluded, and also how they were all correlated with each other.  I would also argue that it could be important to keep some potentially confounding variables (i.e. the plot characteristics plus plot species richness) in the model, no matter their VIF. Otherwise, what is the point of having those variables in the first place?

Response 7: Thanks for this suggestion. We added the table of VIF in the appendix and included the two plot characteristics in the model (Table 1, S2).

Comment 8: Line 156f: I would be more careful in how you express that here. Latitude is clearly reflected in the "region" variable, so it is not that it is unimportant, it is just that in the model all the explanatory power is "taken away" by the region variable, but that region variable still expresses latitude as well. See also my earlier general comment about the use of region as a variable.

Response 8: We agree that the region variable, to a good degree, expresses latitude, but is more than the latitude. For example, temperate Asia, Europe, and North America were coded as different regions though they have similar latitudes. Neotropics and tropical Asia were also coded as two different regions. Anyway, we rewrote the sentences describing the effects of latitude but kept the region variable because it contains more information than latitude and it represents the major hypothesis we aimed to test. We revised the sentences as follows:

“However, when the effects of region, plot characteristics, and climate factors were considered, the latitude variable was not retained in the final model, as it was closely correlated with the region and environmental factors (Table 1, S2).”.

Comment 9: Line 162, “plot characteristics”: I can't see the plot characteristics in these tables. In general, since you never show what the “full” model was that you tested, it is not possible to understand from these table what variables where included and which excluded, as you seem to present only the significant variables. As mentioned earlier, it would be good if you could provide more details on that (see my comments about VIF and correlation between variables).

Response 9: We added the tables in the appendix (Table S1, S2) and included the plot characteristics in the model (Table 1).

Comment 10: Line 174: This table does not show anything about lambda. Again, it would be helpful to see the correlations between variables presented somewhere. Also, if it wasn't significant in the models, why did you decide to include plots in the appendix? Also, I am sceptical about the calculation of lambda for these traits. First because of the aforementioned doubt that lambda is meaningful for discrete variables. Second, because as can be clearly seen in the graphs you provide in the appendix, the lambda is either very high, or zero, which does not seem very meaningful, certainly not for including lambda as an explanatory variable.

Response 10: To clarify, we did not include lambda as an explanatory variable and removed the plots in the revision. See our above responses for other questions.

Comment 11: Line 194-195: I don’t understand how that relates to the previous sentence. You say that dioecy is disadvantageous for long distance colonisation, but what has that to do with tropical regions? Why should they disfavour species with long distance colonisation ability? Also, the end of the sentence doesn't make sense.

Response 11: We revised the sentence as: As temperate plant communities have been assembled from tropical flora (Qian and Ricklefs 2016), tropical and the nearby regions where dioecy originated are expected to have higher incidences of dioecy than temperate regions (Bawa 1980; Renner and Ricklefs, 1995; Sakai and Weller, 1999; Renner, 2014).

Comment 12: Line 197, “which supports…”: Again, I don't follow the logic here. Please try to express yourself clearer.

Response 12: We revised the sentence as: Our study found a high proportion of dioecy in Oceania and tropical Asia (Table 1, S3), which supports the deduction of Baker’s law on the dispersal limitation effect on dioecy.

Comment 13: Line 246: Unclear what processes you mean here, and it results in this phrase sounding quite vague.

Response 13: We revised the sentences as: Regional processes such as long-distance dispersal, evolution of sexual systems after colonization, and local climate, e.g., precipitation, together could shape the global distribution of plant sexual systems.