figures/tables manuscript cleanup

[teixeirak]

Figures & Tables

General:

• [x] update everything with change from 1994-66 to 1966.
• [x] deal with issue of .png tables getting labeled as figures (and consequent incorrect numbering of figures). Related to this,
• [x] Becky figured out how to pretty easily make a similar to Table 1, 2, etc in R markdown. May be worth doing here.
• [ ] figure out how to put variable symbols into tables

Fig. droughts-

• [x] add density plot as second panel
• [x] color code drought shading to match density plot
• [x] remove shading from 1964-1965 (shading is for pre-drought period; these years are drought)
• [x] extend x-axis back to 1950

Fig. height-

• [x] remove 'height' label from panels b-d
  • [x] remove panel d? - see issue #57
  • [x] adjust NEON instrument heights slightly -- see .pdf from John Schnebelen.
  • [x] see issue #58

Table 1-

• [x] fill in
• [x] note that it is currently placed incorrectly in the manuscript (in line, cut off)
• [x] minor cleanup and standardization (e.g., consistency with punctuation and capitalization in predictions)
• [ ] check table: errors may have been introduced with new results (see issue #60)
• [ ] when possible, write out variable names so that table makes sense stand-alone (and to avoid symbol formatting problems!)

Table 2-

• [x] doesn't fit, so move number of stems in each canopy position to SI (or format in Excel?)
• [x] code in R markdown
• [x] fix up formatting
• [ ] @mcgregorian1 , update numbers
• [x] add missing definitions, units, etc.
• [ ] put symbols in italics
• [ ] remember that we'll need to fix TLP at proof stage

Table 3-

• [ ] replace values under range_DBH with "min-max"
• [ ] add n cores analyzed
• [ ] replace codes with species names
• [ ] list in descending order of productivity
• [ ] give ANPP_stem for each species
• [x] move units from legend to table
• [ ] add SD on trait values
• [ ] formatting

Table 4- individual predictor tests

• [x] remove variable name from coefficients columns, make extra rows for categorical variables (if this doesn't make sense, I can do it).
• [x] code in R markdown
• [ ] see if we can put dAIC>2 in bold
• [ ] did we use Y in the null models after it came out non-significant? (I think either way is justifiable, but make sure table is correct)

Table 5- best full models

• [x] format in Excel (issue #40)
• [x] code in R markdown
• [x] fix markup surrounding table. It was fine until I added a full legend...

manuscript

General-

• [x] left align text
• [ ] ensure consistent ,clean names/abbreviations to each variable. For ln-transformed variables, put e.g., "ln[H]"
• [ ] review and cite more papers (see #56 )
• [ ] integrate references
• [ ] review New Phyt instructions for authors, identify & work on needed changes.
• [ ] 1.5 line spacing
• [ ] word count: I added table on title page, as required by NP, and did rough assessment. We're currently about on track, but don't want to make it any more wordy.

Abstract-

• [ ] work on abstract (mainly @teixeirak )

Methods-

• [x] work on Analysis section
• [x] add methods for NEON vertical profiles

Results-

• [x] Fill in missing info in the section "Focal droughts and their characteristics"

Discussion-

• [x] adjust outline (Ian, I'm now thinking we should make some modifications to the version we discussed)
• [x] write first draft

supplementary info

General-

• [ ] break out SI as a separate document. Ultimately, provided that it gets accepted at New Phyt, we'll want it in the New Phyt SI template, although that's probably not necessary for the first submission.
• [x] create bark thickness allometry table
• [ ] do a careful assessment of what needs to be in SI, add anything needed

• [ ] spell check! (pasting into word revealed how badly I type when there's no auto-correct/ spell check!)

  table S1-

• [x] replace x with ln[DBH]
• [x] Table S1 currently says the equation for FAGR as 1x. It should be 0x, and it looks like its that way in your code (line 804 here). Please confirm/ change in the SI.

table S2 -

• [ ] would be more intuitive if you put a column for each drought scenario, with an "x" in columns where it comes out in the top. Otherwise the variables make sense

table S3-

• [ ] clean up formatting (e.g., remove "NA"s)
• [ ] maybe add stats from pointer year analysis?

table S4-

table S5-

• [x] add traits/ height table

Fig. map-

• [x] add species legend

[mcgregorian1]

Fig. map-

• [x] can you color code by species?

I can but depending on how we want to present this table, would we need to have a legend with all 12 species? This may take up more room than we want.

e.g.

[teixeirak]

That looks nice. You know, in retrospect it may be better to just put this in the SI. I'm not sure it adds enough to warrant inclusion in the main doc.

[mcgregorian1]

I've now labeled this as Figure S3

[teixeirak]

Fig. map-

• [ ] can you color code by species?

I can but depending on how we want to present this table, would we need to have a legend with all 12 species? This may take up more room than we want.

Back to this question-- I still think it would be good to color code by species. It will be an SI figure, so it doesn't matter how much space it takes.

[mcgregorian1]

Sounds good. I can get to this hopefully soon.

[mcgregorian1]

Questions from me:

Table 1

• I'm assuming if we're having the abbreviations of the variables, that we want the descriptions of those in the table caption?

Table 2

• Actually, would it make more sense for Table 2 to come first, since it describes the variables?
• I was trying to put in the real symbol for TLP (the pi with tlp subscript) and for the life of me I could not figure out how to make it work in markdown. The closest I came is indicated in the version I've just pushed, where we call TLP "T" and then in the non-R-code portion right after it, spell out the equation, which does work.

In addition, pulling straight from github and trying to knit the document, I kept getting error:

! Undefined control sequence.
<argument> ...ithin 2.0 of the best model (dAICc\<
                                                  2).\relax }}{\thepage }{\@...
l.860 ...within 2.0 of the best model (dAICc\<2).}

I tried putting in "\" in places around the code where we have dAICC<2 (for a table), and I tried doing this in other places as well but I could never get this error to go away to actually knit the document fully. I'm not sure what's happening.

[teixeirak]

* Actually, would it make more sense for Table 2 to come first, since it describes the variables?

I have a strong preference for keeping Table 1 first. We can spell out variable names as needed.

[teixeirak]

Regarding the challenges with coding symbols in tables, I will see if I can get help on that today.

[teixeirak]

I've tried but haven't been able to figure this out yet. Apparently there's no easy answer, but it must be a common problem, so I don't want to give up yet.

[teixeirak]

@ValentineHerr, Alyssa thought you'd figured out how to put LaTEX symbols in a table in R markdown (using the kableExtra package). If so, can you please tell us how?

[ValentineHerr]

I would have to google it too. symbols are always tricky and I don't know a solution off the top of my head.

If you still struggle with it after trying for a while, I can try but I need to know what markdown file and what line(s) in the file you are working on so that I can try it out.

[ValentineHerr]

after a quick look it sounds like if you specify the argument col.names in function kable, you should be able to add the symbol.

For example, for the headers of table2, I think you can do this:

kable(table2, booktabs=TRUE, caption = "**Table 3. Overview of analyzed species, their productivity in the plot, numbers and sizes sampled, and traits.** Given are DBH mean and range of cored trees, the number of cores represented by each crown position of each species, and mean hydraulic trait measurements. Units of measurements are in mm (DBH), \\% (PLA), g/m2 (LMA), MPa (TLP), and g/cm3 (WD).", col.names = c("sp", "DBH_mean", "range_DBH", "RP", "PLA", "LMA", "$\\pi_{tlp}$", "WD")) %>% kable_styling(bootstrap_options = "striped", full_width = FALSE)

and it should give you the correct symbol for TLP in the column header.

If it is in a row I would change the value of the cell to "$\\pi_{tlp}$" and then use the function kable.

NOTE: use two backslashes in from of the pi when you are in the table (as opposed to one in the text), I think.

[mcgregorian1]

I was trying this last week to different degrees but was never successful. @ValentineHerr our table 2 was mislabeled in the text (I've changed it now), and our actual table 2 code is below, where we're more concerned with having the specific cell value be the symbol. I've made sure the cell has that symbology but when rendering it does not come out. I think it's because it's already in a table / is not being read directly as latex.

@teixeirak I checked with my advisor (as per your email) and he wasn't sure either. He said he normally uses equations and symbols directly as latex typography instead of within a df or imported table.

I have now asked a question online, so maybe someone will know

```{r Table 2, eval = TRUE, echo=FALSE, warning=FALSE}
library(knitr)
library(kableExtra)
table2 <- read.csv("tables_figures/variables_table.csv", stringsAsFactors = FALSE, check.names = FALSE)

kable(table2, booktabs = TRUE, caption = "**Table 2. Summary of variables**", format = 'latex', escape = TRUE) %>%
  kable_styling(latex_options = c("scale_down", "hold_position"), font_size = 12, protect_latex = TRUE) %>%
  add_header_above(c(" "," "," "," "," "," ", "observed values" = 3, " ")) %>%
  column_spec(4, width = "7cm") %>%
  pack_rows("Dependent variable", 1, 1, latex_gap_space = "1em", colnum = 1, hline_before = FALSE) %>%
  pack_rows("Independent variables", 2, 18, latex_gap_space = ".4em", colnum = 1, hline_before = FALSE) %>%
  pack_rows("tree size", 5, 6, latex_gap_space = ".4em", colnum = 1, hline_before = FALSE, bold = FALSE, italic = TRUE) %>%
  pack_rows("microhabitat", 7, 11, latex_gap_space = ".4em", colnum = 1, hline_before = FALSE, bold = FALSE, italic = TRUE) %>%
  pack_rows("species' traits", 12, 18, latex_gap_space = ".4em", colnum = 1, hline_before = FALSE, bold = FALSE, italic = TRUE) %>%
  kable_styling()

[mcgregorian1]

The person who's responded to my comment on the online forum (linked above) has asked if I can share the document code, since he doesn't know what could be causing the issue. What do you think?

[teixeirak]

Fine with me.

[mcgregorian1]

Currently in the text, you have this written just before the analysis with monthly PET and PRE sourced from Climatic Research Unit high-resolution gridded dataset (CRU TS v.4.01; Harris etal., 2014). However, upon looking up Harris et al 2014 I'm seeing a description of CRU TS 3.10 as seen here. Is Harris the right paper for this?

[mcgregorian1]

Also another question:

in the methods, you have written this just before the description of the drought years (**add description of climate data used in Fig. 1, NEON vertical profiles**)

Currently Fig 1 is the time-series plot made by Valentine. Did you mean you wanted a description of the climate data used for both Fig 1 and NEON vertical profiles, so in other words a description of CRU and NEON methodology? Or did you mean you only wanted a description for one of them?

[teixeirak]

Yes, that's still their most recent pub.

[teixeirak]

Or did you mean you only wanted a description for one of them?

Just describe your methods behind NEON climate analysis in Fig. 2.

[mcgregorian1]

Ok those are both done

[mcgregorian1]

@teixeirak did you want table4 to have the covariate abbreviations? Currently we have LMA, PLA, and WD spelled out.

In addition, where were you thinking of getting the data for this? (from the results)

• "Community-level tree growth responses to these droughts were modest, with modal resistance values of #, #, and # for 1966, 1977, and 1999, respectively (Fig. 1b). In each drought, roughly 30% of the cored trees suffered ≥ 30% growth reductions (R ≤ 0.7): #% in 1966, #2% in 1977, and #% in 1999. Some trees exhibited increased growth: (R > 1.0): #% in 1966, #% in 1977, and #% in 1999".

Also, here you mention n# of trees. Did you mean all trees in the plot that match the focal species? Or did you mean the number of trees that we did the actual analysis on? If the latter, we only have 541 trees that made the final cut to be included in model runs (this equals 1623 individual runs when done for each year).

• We reject H2.2, finding no evidence that taller trees tend to have traits associated with greater drought resistance. In part because of the large sample size (n=# trees--all individuals of our 12 focal species $\ge$ 10 cm in 25.6 ha), there were very significant (p<0.0001) correlations of $H$ with all species' traits (SI Table?).
  • unsure which SI table or other table to support this. I don't remember where that pvalue comes from since currently we're not reporting pvalues in any of the tables.
  • Ah, I found what you were thinking of - it's the file named "tested_traits_height.csv" - but we currently aren't including that. Did you want this changed?

[teixeirak]

@teixeirak did you want table4 to have the covariate abbreviations? Currently we have LMA, PLA, and WD spelled out.

Okay either way--probably leave as is. I definitely wouldn't abbreviate crown position or xylem porosity. I don't think that using abbreviations would save much/any space.

[teixeirak]

In addition, where were you thinking of getting the data for this? (from the results)

* _"Community-level tree growth responses to these droughts were modest, with modal resistance values of #, #, and # for 1966, 1977, and 1999, respectively (Fig. 1b). In each drought, roughly 30% of the cored trees suffered ≥ 30% growth reductions (R ≤ 0.7): #% in 1966, #2% in 1977, and #% in 1999. Some trees exhibited increased growth: (R > 1.0): #% in 1966, #% in 1977, and #% in 1999"_.

You have all this info in the data behind Fig. 1b. It should be an easy calculation.

[teixeirak]

* Ah, I found what you were thinking of - it's the file named "tested_traits_height.csv" - but we currently aren't including that. Did you want this changed?

Yes, I suppose we should include it in the SI. It's a tiny/simple table, but we should provide that info, and wouldn't want to give the coefficients in the main text.

[mcgregorian1]

In addition, where were you thinking of getting the data for this? (from the results)

* _"Community-level tree growth responses to these droughts were modest, with modal resistance values of #, #, and # for 1966, 1977, and 1999, respectively (Fig. 1b). In each drought, roughly 30% of the cored trees suffered ≥ 30% growth reductions (R ≤ 0.7): #% in 1966, #2% in 1977, and #% in 1999. Some trees exhibited increased growth: (R > 1.0): #% in 1966, #% in 1977, and #% in 1999"_.

You have all this info in the data behind Fig. 1b. It should be an easy calculation.

Ok, this has been updated in the text

[teixeirak]

transferred remaining tasks to issue #62