compare trends with elevation

[lizzieinvancouver]

@buniwuuu It would be very cool to have a plot of growth versus elevation and see how the trends look across species (and papers). Please:

1. Go through the papers in issue #18 and also Neil's email and quickly enter what might have useful data in dataoverview tab of this file
2. Figure out what data seems most common; I think it will be growth (width or RWI or similar) versus elevation; but review enough papers to check this.
3. Scrape data from the papers on growth versus (most likely) elevation.
4. Write an R script to plot the data altogether but color code by datasetID.

Let me know how it goes this week! Thank you!

[buniwuuu]

@lizzieinvancouver dario2015 has sample statistics for chronologies used. He defined mean of sensitivity being "the relative difference from one tree ring to the next (i.e. the year to year variation)". jbi12462-sup-0001-appendixs1.docx

Also do we want basal leaf area data? https://besjournals.onlinelibrary.wiley.com/doi/full/10.1111/1365-2745.13603

For king2013, there's only line plots like this and they didn't provide raw data...

Do we want height or height increment data? oleksyn1998

[buniwuuu]

zhu2018 has tree ring index for multiple years, should I scrape every year's data, take the average, or take only the most recent year's data? jgrg21148-sup-0002-2017jg004292-ds01.docx

[lizzieinvancouver]

@buniwuuu Sorry I went silent there! I was traveling the last two days and didn't know there would be no internet .... anyway....

1. For dario2015 -- we cannot use mean sensitivity based on its definition so no need to scrape that.
2. Also skip basal leaf area -- as we're focused on wood growth here.
3. king2013 -- ahh! Yeah, a lot of tree ring people do this type of plot. We could try scrape it but it would be messy does not seem worth it -- so skip it.
4. for okeksyn 1998 let's do height increment.
5. We only want the zhu2018 data we can match to elevation or similar ... if they provide the average for each site that would be fine; scraping all the data would work too, but we'll end up plotting the mean and variance in the end for the figure we're thinking of.

I am around today if I can answer other questions!

[lizzieinvancouver]

@buniwuuu Looks like you have scraped a lot of data! Let me know if you want help working up a quick R plot to look at the results.

[buniwuuu]

grephon_plot.pdf

@lizzieinvancouver I made a R plot, but some people used altitude and I wasn't sure if they're the same as elevation in this case so I plotted them separately. Also oleskyn1998 data has cm/yr as the growth value unit so we might need to process the data somehow.

[lizzieinvancouver]

@buniwuuu Amazing!

Altitude usually is used for planes -- as it's the value above some fixed point. Often people say it interchangeably with elevation (which should be above sea level). Did the altitude papers mention any fixed point? If not, we can assume they meant elevation.

[lizzieinvancouver]

@buniwuuu I pushed some code to try correcting Oleksyn, but it comes out REALLY small (you can run the code yourself to see) ... so now i have two queries:

1. Were these grown together in one location (a common garden)?
2. You scraped from Fig 2, right? I notice in the methods they write:

To compare growth rates between seed stands average stand height and d.b.h. were divided by stand age to calculate height (m per year) and d.b.h. (cm per year ) increments.

Do you think that's what they are plotting? If so, I think this is a WEIRD metric, but need to think more and discuss with colleagues.

[lizzieinvancouver]

@buniwuuu One more query!

3. Could you find data from deSauvage 2022? It was in the git issue, but the link was funny.

[buniwuuu]

@lizzieinvancouver the link in git issue didn't work! I just read through it now and couldn't find raw data.. We can potentially scrape the graphs but the regression lines could be a problem?

[buniwuuu]

@lizzieinvancouver for oleskyn, I think they plotted the increments as the height/dbh divided by stand age. The trees they examined were both from both native and common garden environments.

[lizzieinvancouver]

@buniwuuu If they (de Sauvage and colleagues) don't provide the data outside of the graphs then it seems too hard to do, let's skip it (I just checked and there's no data sharing or data publication statement either, which is disappointing).

[lizzieinvancouver]

@buniwuuu For the Zhu2018 (this paper, I think their elevation co-varies with latitude (based on Table 1) or am I wrong?

[buniwuuu]

@lizzieinvancouver Oh right it is! Should we not include it?

[buniwuuu]

@lizzieinvancouver also I checked the code you pushed and I think the adjustment for Oleskyn should be *10 instead of /10 because we are converting cm to mm?

[buniwuuu]

@lizzieinvancouver I read through Oleskyn again and the data I scraped was from transects along the elevation gradient, not the common garden or nursery (those data are in figure 3 and table 3)!

[lizzieinvancouver]

also I checked the code you pushed and I think the adjustment for Oleskyn should be *10 instead of /10 because we are converting cm to mm?

OMG -- yes! You're right, thank you! It looks so much more normal now.

[lizzieinvancouver]

[Zhu2018] Oh right it is! Should we not include it?

@buniwuuu No, it's good that we scraped it, we should just think through how to present it then.

[lizzieinvancouver]

@buniwuuu Your figure is great -- are the error bars showing the scraped error?

[buniwuuu]

@lizzieinvancouver sorry for the late reply! I completely miss the email. Yes, the error bars are showing the scraped error.

[lizzieinvancouver]

@buniwuuu That's great. Thank you!

[lizzieinvancouver]

@AileneKane If you have time to update the figures at all, please do! It would be great to add error to the lit review one.

[AileneKane]

Updated figures are here: https://github.com/lizzieinvancouver/grephon/tree/main/analyses/growthxelevationetc/figures and I will try to remake with annual growth and growing season length

[lizzieinvancouver]

@buniwuuu Remember this long ago issue? I could use help again on it. Could you:

1. Make a much nicer version of the figure of growth x elevation near the end of this code. Make sure:

• [ ] The lm lines that show uncertainty are nicer colors (use viridis package) and only extend as far as the data for each species
• [ ] Italicize the species names in the legends and try to move to inside of plot
• [ ] Specify units for the axes and update axes names (e.g., Elevation (m) )
• [ ] Save it as a PDF to folder

2. Add methods text (here is fine, or can add txt file to docs folder or such) that include how many papers you reviewed today, how you scraped the data (image J?), a short list of the main reasons we rejected papers (could not get or could not use data).

Thank you!

[lizzieinvancouver]

@buniwuuu Update -- if you want to put methods text in the repo, I suggest you put it as a plain text or markdown file in notes and call it something like growthxelevationlitreview_methods.txt

Thanks again!

[buniwuuu]

@lizzieinvancouver does this plot look good?

[lizzieinvancouver]

@buniwuuu Is that the right data? The elevations are all REALLY low ... can you double check? I had made a draft and this is how it looked see here ... did I use the wrong data? I think we want only the studies we agreed on before and only for elevation. As you work up the methods, I think what we agreed should be clear so work on that let me know. I'd also like to include the species names in italics and the refs in () after if easy ....

[buniwuuu]

@lizzieinvancouver my bad! I used the latitude chunk of the script... Here's the updated figure! I will work on the methods today.

[buniwuuu]

@lizzieinvancouver I pushed a draft for the methods here. Also, could you remind me why we didn't end up including Oleksyn? I saw that your plot has Oleksyn data but in the script it was excluded.

[lizzieinvancouver]

Also, could you remind me why we didn't end up including Oleksyn? I saw that your plot has Oleksyn data but in the script it was excluded.

@buniwuuu Great question! I was not sure, so I went back through this issue and the code. It looks like I updated the code to remove oleksyn on 29 Aug 2023 while I thought the conversion of oleksyn was wrong but then on 1 Sept 2023 we have this:

also I checked the code you pushed and I think the adjustment for Oleskyn should be *10 instead of /10 because we are converting cm to mm?

OMG -- yes! You're right, thank you! It looks so much more normal now.

And I updated the code to change the conversion, but failed to add it back in. So I think it was a mistake to exclude it and we should add it back in. I checked the code history (via the blame feature, which is a lifesaver!)

Thank you for catching this -- good work!

I need a few more tweaks to the figure. Could you:

• [x] Switch to use + theme_classic()
• [x] Remove the `L.' from Fagus legend
• [x] Make the top legend: Species (reference)
• [x] Make the references look like they are formatted, so instead of calvin2016 it's probably Calvin et al. 2016

I pushed edits to the methods to make them read more like typical language in lab (we prefer active voice 90% of the time). Also, please always use the spelled out month for dates! So 15/Apr/2024 not 15/4/2024 -- this avoids future confusion.

• [ ] Please check my updated methods are correct
• [x] Finally! Do we have a reference list of the papers we looked at somewhere? If not, please make one using Google Scholar and bibtex and upload the .bib file to the repo. (Basically just make a list of refs in google scholar then export it to bibtex and upload the resulting .bib file).

[lizzieinvancouver]

@buniwuuu A couple more requests I found from co-authors to finalize the figure and methods.

• [ ] Add details on where (geograpgically data come from and what exact measurements were reported; for example are these values across several years per dot, several trees per dot, etc)
• [x] Show 89% intervals in the shading for the lines in ggplot

[buniwuuu]

@lizzieinvancouver I updated the figure! Please let me know if there's anything that needs to be changed.

[buniwuuu]

@lizzieinvancouver While updating the methods, I realized Cavin and Jump's data went across a wide range of latitudes and longitudes (58.5° to 40.8°N, and from 5.8° to 11.8°E)... Do we still want their data to be included?

[lizzieinvancouver]

While updating the methods, I realized Cavin and Jump's data went across a wide range of latitudes and longitudes (58.5° to 40.8°N, and from 5.8° to 11.8°E)... Do we still want their data to be included?

@buniwuuu It depends if the elevation and latitude COVARY. Can you check by plotting one against the other?

[lizzieinvancouver]

Please let me know if there's anything that needs to be changed.

Looks much better, but a couple more tweaks:

• [x] For references the et al. needs to be in italics
• [x] Add () to the Oleksyn ref
• [x] Remove the conf interval legend entirely (I will explain in caption) and remove the 95% interval, we only need 95%.
• [ ] Is Growth (mm) -- the same as Growth (mm of ring width)? If so, replace with that as we need to be explicit.

Thanks @buniwuuu !

[lizzieinvancouver]

Can you check by plotting one against the other?

@buniwuuu I just did this and they do covary for Calvin and Jump -- how annoying! We should delete it from the figure -- I left a note in the code for you. I also did some of the tasks above but could not figure them all out. Can you please do the rest? And ...

• [x] Make the confidence interval shading the same color as the lines (you may have had this and I broke it? Sorry, if so).

[buniwuuu]

@lizzieinvancouver Growth (mm) is ring width except for Oskelyn which uses DBH increments...

[lizzieinvancouver]

@buniwuuu Thank you for the figure! Please let me know when you have had time to finalize the methods (including a bib file).

[buniwuuu]

@lizzieinvancouver I updated the methods and added the bib file to bibtex folder!

[lizzieinvancouver]

I updated the methods and added the bib file to bibtex folder!

Thanks @buniwuuu but the bib file only has three articles, it should have all 18 you reviewed. I especially need the ones mentioned in the methods so I can cite them in the paper. Thanks for updating this with all articles when you can (ideally before end of this week).

[lizzieinvancouver]

@buniwuuu The updated you pushed has 19 articles, not 18 and none of them match to the name Huang or the year 2010.

Could you please carefully review this and fix it? Can you also reply here give me the cite keys (e.g., zhou2022altitudinal) for the three articles we show, and the 6 articles we scraped?

[lizzieinvancouver]

@buniwuuu One more request in going through co-author comments. Could you also add details for the three studies we plotted on how many years of rings and/or DBH they had (give an estimate if you're not sure). And then, did you scrape (and plot) the averages of these values? More details here needed. Thank you!

[buniwuuu]

@lizzieinvancouver On it! Also, upon reviewing I realized I made a mistake for wang2017 SE. The figure was a box plot so the tick marks were just quantiles, not error bars. I fixed both input/classicalrefs_datascraped.csv and data/classicalrefs.xlsx

[buniwuuu]

@lizzieinvancouver Also, for Oleskyn which has 54 sites, would it be better to have a table? Or do you think I should list the information about 54 sites?

[lizzieinvancouver]

Also, for Oleskyn which has 54 sites, would it be better to have a table? Or do you think I should list the information about 54 sites?

@buniwuuu I am just looking for general information that I can write in the text -- so best would be if you could quickly estimate the range or such. We don't want so much detail that we are either re-writing the paper or giving a level of info better found in the data we scraped. Hope that makes sense and thank you!

[buniwuuu]

@lizzieinvancouver got it! I just pushed the edited version. For the references, I did 20 in the classicalrefs.xlsx (14 from issue #18, 5 from Neil's email, and desauvage2022 which was mentioned earlier in this issue but wasn't in the .xlsx (updated to the file just now). I missed Huang2010 for some reason but now they should all be there!

Study IDs: (I'm not sure if you wanted "altitudinal" with every ID, but please let me know if you mean something else)

• scraped and showed:
  • wang2017altitudinal
  • zhou2022altitudinal
  • oleskyn1998altitudinal
• didn't show:
  • huang2010altitudinal
  • cavin2016altitudinal
  • zhu2018altitudinal

[lizzieinvancouver]

@buniwuuu Thank you! I think we're super close, but I need your help to clarify a few things by updating the growthxelevationlitreview_methods file. Can you update to clarify:

1. How many papers scraped and checked. We say 18 checked: 13 excluded, which must be wrong as we also say we scraped 6 (13+6 = 19) but you also now mention 20. Please update all the numbers to be correct!

2. You mention 6 altitudinal transects between 1600m to 2500m for Zhou et al.'s study (2022) but only four points seem to show. Can you explain this? I think maybe the axes on the figure (seegrowthbyelevation_plot.pdf) are wrong and that's the problem as the Wang data looks too few and too low.

3. Confirm (and if correct, update text): we scraped DBH and height increment measurements from this (Oleksyn et al.) study and use DBH for our visualization.

4. For oleksyn1998growth, we say 'The mean of values collected from each population are presented in this study.' But I don't think we show 54 populations. Can you check the text/data/figure?

After this I will finalize the manuscript text for your approval and then we should publish the data together so it's done!

[buniwuuu]

Hi @lizzieinvancouver sorry for the delay!

1. Done!
2. The plotting code for the legend was in the wrong order. It's fixed now!
3. We only scraped DBH increment.
4. The figure (figure 2) only shows 42 points although they got data from 54 populations. I added clarification in the methods.

Please let me know if there's anything else to fix!

[lizzieinvancouver]

@buniwuuu Thanks!

I have now published the dataset: Tree growth x elevation and latitude relationships from the literature https://knb.ecoinformatics.org/view/urn%3Auuid%3A39be03f4-14ff-4c6e-9c42-f3bcfbf5eea8

Could you please review it and see if there are any missing methods or missing info others would need to use the dataset? For example, I think perhaps we also found papers from colleagues who recommended papers ... if so, we should add that. We could also explain the uploaded file more (and did I upload all the files)? I am back so if there are many edits, I can get you logged in on my KNB account and you can update it.

[buniwuuu]

@lizzieinvancouver we did have a couple of papers from colleagues and classicalrefs_datascraped.csv is the only data needed for analysis! (Unless they need the script for plotting/converting units). I can update the methods and push to repo if that's easier too!

[lizzieinvancouver]

@buniwuuu Remind me that we should chat about this when we meet!