EcoClimLab / ForestGEO-tree-rings

Repository for analysis of tree-ring data from 10 globally distributed forests (Anderson-Teixeira et al., in press, Global Change Biology)
2 stars 2 forks source link

DBH x climate interactions table or figure #107

Closed teixeirak closed 3 years ago

teixeirak commented 3 years ago

would it be easy to make plots of the significant climate x DBH interactions? I'm thinking plots like the climate response plots, but with separate lines for min and max DBH.

Originally posted by @teixeirak in https://github.com/EcoClimLab/ForestGEO-climate-sensitivity/issues/106#issuecomment-755333726

teixeirak commented 3 years ago

@ValentineHerr , I'm moving this to a new issue.

I think we need either a figure or table summarizing significant climate x DBH interactions. I tend to prefer figures, but a table would work.

For a figure, I'd pick out just a few interesting examples.

teixeirak commented 3 years ago

If it's tricky, don't worry. A table will also be good.

ValentineHerr commented 3 years ago

I am getting there.... image

Need to run errands, will try to finish later today.

teixeirak commented 3 years ago

wow, that's cool! More dramatic than I expected.

teixeirak commented 3 years ago

Hmmm... LITU is a cool example, and interactions are in the expected direction (based on McGregor et al), but the interaction term is not significant. If your script can easily/automatically generate these plots for all species, that would be really cool to see. If not, here are a few that I think would be cool examples, where the interaction is significant:

-PIPU and POTR at CB, which have sig interactions for both T and P. PIFL and PIEN at CB have sig interactions for T. Coefficients are all consistent, and all in the expected direction for a drought effect. -TOCI at HKK also has sig interactions fro both T and P.

ValentineHerr commented 3 years ago

I can generate for all sites but have to re-run eveyrthing which takes a while.

I ran it for LD and CB to see if it was running without errors.

The sites that have interactions can be filled in a plot like the one below ( What response variable do you want to see?). But it might be a bit hard to see...

image

I'll have individual plots like this: image

We can pick the best ones to show when we have them all.

teixeirak commented 3 years ago

The summary plot may be handy for reference, but I think it would be too busy. I'd prefer to pick out just a couple interesting examples, so maybe just make the individual plots for now?

teixeirak commented 3 years ago

'll have individual plots like this: image

I'm confused by this one. Where are the max DBH plots for ABBI, PIEN, and PIFL?

ValentineHerr commented 3 years ago

I'm confused by this one. Where are the max DBH plots for ABBI, PIEN, and PIFL?

I think (hope!) that is an old version of what I did, when I forgot to adjust y-scale. They must higher than 0.008. It should be fixed when I re-run that site.

teixeirak commented 3 years ago

What response variable do you want to see?

As to this, probably RW. BAI would be the other possible metric. I don't think we'd want to show ∆AGB.

(Sorry, I'm super distracted by the news right now.)

ValentineHerr commented 3 years ago

(Note: allodb has been updated, which will change the agb a bit. For now I am rerunning all sites with old version agb, to make sure my latest changes with dbh x clim are not messing up the rest. Hopefully it is done by tomorrow and then I can do a final big run (with detrending, early years, old years, with Year, CO2 etc... with latest version of allodb over the weekend).

ValentineHerr commented 3 years ago

I pushed all interaction plots. You can see then all in plot per response variable here, here and here

teixeirak commented 3 years ago

Thanks! These are great. I'm reviewing with the intention of picking a limited number of interesting examples.

teixeirak commented 3 years ago

Here are 3 where there are sig interactions for both T and PPT variables, all consistent with drought effect, and where there's not a steep difference in RW. between small and large trees .

I'm tending to favor one species per plot (otherwise too busy), and am thinking of a 6-panel plot with these examples. I have to run now but will confirm later.

teixeirak commented 3 years ago

@ValentineHerr, sorry for the long pause. Let's try the figure described above., using RW as the growth metric.

Note that I'm planning to include the all interaction plots for RW and BAI in the SI. (Not including ∆AGB because there's such a huge difference between large and small trees.)

ValentineHerr commented 3 years ago

@teixeirak, is this what you had in mind? DO you want the legend for min and max dbh?

image

teixeirak commented 3 years ago

This looks great; thank you! Yes, please put a legend saying "min DBH" and "max DBH". I like to have the figures be mostly self-explanatory without reading the legend.

teixeirak commented 3 years ago

Also, if you can fit the species name in the title, that would be great.

ValentineHerr commented 3 years ago

I pushed updated version

teixeirak commented 3 years ago

looks great; thanks!