minor fix-up

[teixeirak]

• [x] update Table 1
• [x] remove climate SDs from Table 1
• [x] update results/ stats
• [ ] fill in references
• [x] clarify certain points in the methods with Valentine
• [x] read/cite Mathias & Thomas PNAS

[ValentineHerr]

@teixeirak, FYI, I just pushed a big update after re-running everything. Nothing serious has changed except some curves have changed line type, so some variables gained or lost some significant terms.

[teixeirak]

Thanks! Does this affect just models with ∆AGB, or are there some other changes?

[ValentineHerr]

Somehow the changes are not only on ∆AGB. to be honest, I don't really know what happened. Maybe the new allodb prevented the deletion of some rows that had negative values, or decreasing agb values and that affected the other variables too...

[teixeirak]

@ValentineHerr , I have a few questions/ clarifications for you in the methods section (You can answer here or edit directly in the .Rmd.):

1. For GAM in climwin step, how did you decide how flexible to make the spline? (current line 410)
2. A coauthor requested that we include the GLS model formula (current line 427)
3. In sampling_details.csv, are "n trees all" and "n cores all" before or after exclusion of cores that could not be included?

[ValentineHerr]

Looking into this now

[ValentineHerr]

I think I answered first 2 question directly in the manuscript (which I'll pish soon)

About:

• In sampling_details.csv, are "n trees all" and "n cores all" before or after exclusion of cores that could not be included?

There is so many stages where we can exclude things, it is hard to keep track....

It is after:

• all the cleanup done in the dendro repo (so removed last few years, remove years where <2 dbh represented etc...)
• after years before or after climate records are removed
• after cores with <30 years are removed

but before:

• cores with <10 measurements were removed (this is probably more relevant for analysis of older or younger records only)
• species were removed because they had <7 trees
• any other restrictions specific to interactions or year analysis.

[teixeirak]

@ValentineHerr , thanks!

One more question: How did we handle cases where our reference DBH was taken after coring (I think we have cases of that...)? For context, here's the text that needs to be filled in:

[ValentineHerr]

Sorry I forgot to answer this.

Here is what I have in my script:

1. when year of DBH measurement is in a core measurement year, anchor the dbh there

2. If last year of tree ring data is just one year before the year of DBH measurement - 1 (allowing one year slack since current growing season is sometimes not present in cores) anchor current dbh to that last year of measurement

3. When year of DBH measurement is no more than 10 years later than last year of measurement, calculate what dbh would have been, replacing the n missing tree-ring data by average tree-ring data in the n last years measured.

4. now for any core that still doesn't have an anchored dbh (either because last measurement is > 10 years before year of DBH measurement or because there is no dbh measured), see if we can anchor dbh 0 at pith or estimated young dbh near pith if we can (see 5 and 6)

5. if [we don't have DBH for the tree] OR [last year of tree-ring data < coring year minus 10] AND [core hit center OR we have an estimate of n rings to pith], reconstruct inside-out. Assume that width of missing rings = average of 10 earliest records.

6. Any remaining cases won't have DBH.

[teixeirak]

Got it; thanks!