In revision, Janssen et al. much clarified the methods section, updated their result section to reflect methodological changes, and expanded the discussion section. I also really appreciate the authors taking comments and suggestions about their (greatly improved) figures on board, and providing new supplementary material which will be very useful to the readers. Whilst acknowledging all of these positive changes, as well as being convinced by the authors responses to the sampling bias, drought definition, and VPD issues raised before, I still have major methodological concerns about the additional transpiration responses reported in this study.
In my earlier review of the manuscript, I commented that opposite hydraulic strategies were likely being compensated / cancelled at both the leaf- and tree- levels, as the authors chose to analyse all species’ leaf-level transpiration responses at once. In their response to this comment, the authors essentially agreed that high wood density and low wood density species were generating opposite responses when going from the wet to the dry season, therefore resulting in no change in leaf-level transpiration (now called potential transpiration). But in the manuscript, lines 297 - 305, the authors invoke gs and VPD compensating effects as a mechanism to explain sustained potential transpiration. This is very confusing, given that besides displaying varied gs responses to soil moisture decline, different species also display different gs sensitivities to changes in VPD. The reader has to wait for Section 3.3 to see mentions of varied hydraulic strategies. Essentially, lines 297 - 305 lack a clear mention of the fact that the lack of change in potential transpiration is also the product of compensatory gs responses to changes in soil moisture and VPD among the different species!
A further issue with the potential transpiration estimates is that, although the authors now acknowledge that atmospheric VPD is not akin to leaf-to-air VPD, they wrongly assume that because the “meta-analysis deals with relative changes”, then the “overestimation should not have a major impact on the drought-induced percentage changes in potential leaf transpiration”. This in itself assumes that feedbacks from the leaf to the atmosphere, as well as the strength of the coupling relationship between the leaves and the atmosphere above, remains the same from the wet to the dry season; but this is not supported by previous studies, and at the very least warrants introducing more caveats in the methods section.
As a result, the authors also have to be more careful in reporting “WUE” (or potential WUE?) changes between the wet and dry seasons.
The revised methods section now explicits how tree-level transpiration was calculated. I previously thought that the relationship between Jmax and tree-level transpiration shown in Figure S2 had come from another study, but in fact, it seems that it comes from this study? If I understand this properly, then the authors are using a relationship based on Jmax to generate tree-level transpiration data for 17 studies, based on the data from another 17 studies that they themselves have collected. Is this actually adding information, or is this propagating errors / divergence from the fitted line? It would be useful to know whether the 17 data the authors are deriving are for high Jmax (where the data in Figure S2 diverges from the fitted line) or whether they are for low Jmax. Either way, providing uncertainty bounds on these tree-level transpiration estimates would be useful (I think the previous manuscript mentioned a 34% uncertainty, which is non trivial).
Additionally, the editor asked about compounding LAI effects, but I don’t find the authors’ response very convincing on that point. Given that some species flush more leaves than others, and that leaf flushing is not happening at a constant rate between the wet and the dry season, and between the dry season and an episodic drought, then not normalising tree-level transpiration (in Figure S2) by LAI cannot lead to a sensible comparison (put simply, isn’t this like looking at apples in July vs oranges in December?).
In view of the range of uncertainties associated with the leaf- and tree- level transpiration estimates (and given that the authors have now added a measure of crown conductance which can loosely be compared to stomatal conductance when discussing responses across scales), I would suggest:
(1) to completely remove the additional transpiration measures from the meta-analysis;
(2) to remove them from the “bulk analysis” and to only incorporate the leaf-level responses in Figure 4 / the analysis clustered by wood density. In that case, Section 2.2 would also benefit from being reworked to clearly separate Epot and WUE from the other additional measures which seem less problematic.
So in my opinion, the additional derived measures of transpiration are taking away from the study’s scientific soundness, by relying on too many uncheckable assumptions and by largely confounding information. Otherwise, this study represents a timely and very nice piece of work.
Minor comments
--------------------
The flow of text would greatly benefit from shortening any sentence that spans more than 3 lines.
There is a mix of referring to the “dry season” or to a “seasonal drought” in the manuscript. This is somewhat confusing, can the authors make sure the term used is “dry season” throughout?
It occurred to me that using the term “measures” when referring to variables is misleading as it somewhat implies measures of traits or so. Why not simply refer to variables?
L. 23: whilst I appreciate the reformulation, there needs to be a clear explanation of how this could be done in the main body of text, perhaps in the conclusions?
L. 35: typo, replace “forest” by “forests”
L. 46: is it totally fair to say there is no understanding?
L. 60: please move “in the neotropics” to the start of the sentence
L. 61: “plant functioning” should be “plant function”
L. 63: “in atmospheric demand” instead of “of atmospheric demand”
L. 67: typo, missing the ending “e” in severe. Also, please replace “climate warming” by “climate change”
L. 88: the text following this equation would be clearer if, instead, the equation were written as a system of two equations
L. 89: E is still the leaf transpiration rate here
L. 90: VPD appears in the equation vs VPDs on this line
L. 90 - 92: these sentences would be clearer if referring to two different equations, which can be done by rewriting Eqn 1 as a system
L. 93: typo, should be “a drier soil”
L. 93: rephrase “reduced hydraulic conductance of the xylem as a result of xylem embolism” as “of a reduced xylem hydraulic conductance from embolism”?
L. 94: “all things being equal” is unnecessary text
L. 94: “decline in” instead of “decline of”
L. 96: please replace “if” by “whether”
L. 99 - 100: this is nicely phrased!
L. 101: please replace “compared to” by “than”
L. 101: “increase in” instead of “increase of”
L. 116 - 118: this sentences doesn’t logically follow
L. 124: “drought avoidance” instead of “drought avoiding”
L. 133: “cavitation” instead of “dehydratation”?
L. 135: it’s not just midday! “midday” could simply be removed
L. 150 - 151: would make more sense as “We searched Web of Science for literature published between 1979 and 2019, which matches the span of the ERA5…”
L. 152 - 155: and given Figure S2, tree level daily transpiration too?
L. 158: soil matric potential is not the same as soil water potential, but the authors later refer to soil water potential. So which is it?
L. 173 - 175: so it’s not actually midday data? This is really unclear from the sentence.
L. 179 - 181: therefore I’d expect the authors to comment on the fact that the REW metric is not an exact measure of how much drought an ecosystem is really experiencing
L. 188 - 189: it’s not using atmospheric VPD instead of leaf-to-air VPD which disregards the fact that leaves are often decoupled from the atmosphere above in dense tropical canopies, it’s using gs instead of gc that does. Rather, the atmospheric VPD instead of leaf-to-air VPD ignores feedbacks. Those two points are different.
L. 196: missing “rate” after photosynthesis
L. 201: “summed” is unnecessary
L. 203: “Jdaily” is a weird choice, why not use “Edaily”?
L. 204: please specify what “a” equates to
L. 214: so given Eqn 1 we know have E = Jmax * P? So following that logic, Epot could be Jmax * P instead of gs * VPD? Have the authors compared those values at all?
Also, where do the P data come from? I don’t think that information is in the text.
L. 241: “per year” should be moved after “at each site”
L. 242: typo, “has changed” instead of “have changed”
L. 245 - 248: please rephrase this sentence
L. 248 - 249: this sentence is completely unrelated?
L. 251 - 253: this doesn’t logically follow
L. 260 - 262: given that you never get back to this “control” and “treatment” linguo, it’s just confusing, why keep it?
L. 272 - 273: isn’t that a result rather than methods?
L. 279 - 280: why not simply drop these individuals? This is yet a further assumption which introduces uncertainty that’s hard to quantify, and it is unnecessary given the authors would still have data from 738 individuals to analyse.
L. 329: typo, “change” instead of “changed”
L. 340: typo, “correlated” is missing an “r”
L. 342 - 344: except only the years after 2000 are visible on the figure…
L. 372: “even” is unnecessary
L. 388 - 390: please rephrase the sentence
L. 394 - 396: have the authors looked at any literature on tissue porosity? This could be an explanation relating to low wood density + high stomatal control.
L. 398 - 399: which falls from the previous finding
L. 425: and also hydraulic architechture, etc. this statement would benefit from being more nuanced
L. 437: this is not always true, there is also observational evidence of unchanged soil to leaf hydraulic conductance, for a decreased stomatal conductance during drought
L. 447 - 448: why despite? There is no obvious incompatibility between these two things
L. 473 - 475: the authors have to be careful here, this is in ERA5 which is a reanalysis product, so might contain such a drying trend due to intrinsic biases...
L. 492 - 500: this feels like a repeat of one of the results paragraphs and doesn’t actually discuss anything per se
L. 516 - 518: very nice discussion!
L. 527 - 528: which begs the question of why the authors chose to include the first couple months after an episodic drought in the wet season, rather than to exclude them from the analysis altogether...
L. 593: hydraulic properties and architecture
L. 619: should the ref to tropical montane cloud forest be mentioned here considering how montane forests have specifically been excluded from the meta-analysis?
Figure 1: the legend reads “number of episodic drought months recorded per month at each site in the database”, shouldn’t it be “across sites in the database”?
Also, subplots d and e are wrongly labelled b and c on the figure
Figure 2: whilst the legend now explains what the capital letters are, it omits the fact that the numbers below indicate the number of? Consulted studies? Sites?
Figure 3: what is blue and what is red? Same applies to some supplementary figures…
Figure 4: why not match the colors with those used in Figure 3? Also, please consider separating the r2 values and p values reported on the plot by introducing a semi-colon or increasing the space between them (it's currently a bit hard to read) |