final clean-up tasks

[teixeirak]

Here are some things to do near the end of revisions:

• [x] redo word, etc counts in table at start of manuscript
• [x] full pass through reviewer responses, including copying in text where needed
• [x] redo list of SI materials at end of manuscript
• [x] check that figure captions match below figures and in separate list
• [x] check citation of correct figs/ tables (I've tried to keep this updated as I've made changes, but needs to be checked)
• [x] finish letter to Editor
• [x] format R packages citations in appendix, and make sure Ian is the last one to run (for citations).
  • Add agricolae and visreg if not already mentioned in text

[mcgregorian1]

Also

• [ ] worth noting that the hydraulic traits (both species level and individual level) are still in the private EcoClimLab repo. If we do change this to be public, I will need to update my code for the website it pulls from. Currently I have to update it when I run the code every time because it uses a different token.

[teixeirak]

Yes, I still need to deal with that. Not needed until the paper is accepted, though.

[mcgregorian1]

My eyes are failing me right now but I added text for some instances (up to line 265 in the .Rmd file) for response to reviewers.

Quick question - are these references added as PDFs? I didn't get this warning message when I compiled the full manuscript; did you only put them in the response?

[teixeirak]

Ack, sorry... I put them in wrong (year before first word of title) on the first round and forgot to fix. Hope you didn't lose too much time on that.

[mcgregorian1]

You're good! This still rendered and just gave me warnings

[mcgregorian1]

I have fixed the references and pushed the edits

[mcgregorian1]

I'm going to put remaining questions for reviewer comments here.

First, do I need to say more than what I do here?

Only assessing models with AIC is not enough to prove sound relationships in multidimensional and complex data-models. LL ratio tests or deviance-anova tests can be used in the case of nested models with LMM or GLMM respectively. Along the text: dAIC is generally referred as $\Delta$AIC, please change accordingly.

We have run deviance-anova tests and can be seen in Figs. 3 and S4.

We have gone through the manuscript / tables and figures and updated where it said "dAICc" to be $\Delta$AICc.

[mcgregorian1]

You say here Such patterns are not shown in some relevant textbooks: Gordan Bonan's 2016 Ecological Climatology and Norman & Campbell's Environmental Biophysics [@campbell_introduction_1998].

We don't currently use Bonan 2016 book in the manuscript. Did you want to put this somewhere?

[mcgregorian1]

RE this comment - I think you have a better grasp of the correct language for responding to this.

L234-237: Not very clear how these models were constructed. Why did you fit each predictor independently to test its importance as opposed to standardizing them and having them together in a multiple regression and then compare their coefficients? And why was height retained in all models? Testing one predictor at a time might be sensible if you have lots of collinearity between predictors. But it could also mean you miss important conditional effects (e.g., the effect of variable x only emerges when you first account for the effect of variable y).

(just need to write a response here-- no further changes needed to manuscript)

[mcgregorian1]

Ok, aside from these three questions I have, the response to reviews should be good to go. I will push in a second

[teixeirak]

First, do I need to say more than what I do here?

Only assessing models with AIC is not enough to prove sound relationships in multidimensional and complex data-models. LL ratio tests or deviance-anova tests can be used in the case of nested models with LMM or GLMM respectively. Along the text: dAIC is generally referred as $\Delta$AIC, please change accordingly.

We have run deviance-anova tests and can be seen in Figs. 3 and S4.

(Sorry, had a meeting...)

I actually forgot about this comment. I do think it would be good to say more here. You previously ran the ANOVA tests and found that they gave similar ranking...

[mcgregorian1]

How about:

"We implemented the suggestions by running both LL ratio tests and anova tests on the top models. They both gave similar ranking to dAICC, thus we continued with our original route. However, to give a more comprehensive view of the covariates, we now show significant differences between species compared to Rt (for each drought scenario; Fig. 3), as well as significant differences between species for each individual trait tested in the multivariate models (Fig. S4)."

[teixeirak]

Sounds great!

[mcgregorian1]

Cool! Just my last two comments above, then, and it's ready for a final lookover from you. Let me know when you do and I can switch over to giving a final read-over of the manuscript

[teixeirak]

Okay, let's switch. I just pushed the manuscript, so you can give that a final pass. There are a few references needed in the final paragraph, most of which can be covered by papers we cite earlier. I'll get those later.

[mcgregorian1]

Aside from the final references, I wanted to flag these:

"All data, code, and results are available through the SCBI-ForestGEO organization on GitHub (https://github.com/SCBI-ForestGEO: SCBI-ForestGEO-Data and McGregor_climate-sensitivity-variation repositories), with static versions corresponding to data and analyses presented here archived in Zenodo (DOIs: 10.5281/zenodo.3604993 and [TBD], respectively."

Finally, one question that Josh (?) brought up in the lab meeting was the fact that we measured TLP in 2018 (and all traits I guess), and presumed they did not change going back in time. Are we safe to assume that and, if so or not, I feel like this should be in the text correct?

[mcgregorian1]

I also added in a quick source for visreg in the manuscript just before the beginning of the Results, and kudos to the agricolae package (automating the anova significance groupings) in the Results itself. I think that's ok.

[mcgregorian1]

Otherwise I think it's good on my end! I just need to make sure I am the last one who knits everything before submitting I think.

[teixeirak]

"All data, code, and results are available through the SCBI-ForestGEO organization on GitHub (https://github.com/SCBI-ForestGEO: SCBI-ForestGEO-Data and McGregorclimate-sensitivity-variation repositories), with static versions corresponding to data and analyses presented here archived in Zenodo (DOIs: 10.5281/zenodo.3604993 and [TBD]_, respectively."

We'll take care of that after acceptance (issue #79 ).

[teixeirak]

Finally, one question that Josh (?) brought up in the lab meeting was the fact that we measured TLP in 2018 (and all traits I guess), and presumed they did not change going back in time. Are we safe to assume that and, if so or not, I feel like this should be in the text correct?

The assumption is that interspecific variation is greater than intraspecific variation, variation with height, variation through time, etc. I don't think we need to change anything.

[teixeirak]

Otherwise I think it's good on my end! I just need to make sure I am the last one who knits everything before submitting I think.

I think this problem is solved. I need to take a lunch break now, and will work on final check/ submission after that!

[mcgregorian1]

Otherwise I think it's good on my end! I just need to make sure I am the last one who knits everything before submitting I think.

I think this problem is solved. I need to take a lunch break now, and will work on final check/ submission after that!

Oh that's right! Never mind then. I'll take lunch as well