search

SCBI-ForestGEO / McGregor_climate-sensitivity-variation

repository for linking the climate sensitity of tree growth (derived from cores) to functional traits
0 stars 0 forks source link

include species-specific Rt plot? #100

Closed teixeirak closed 4 years ago

teixeirak commented 4 years ago

I actually really like this figure and think it a good candidate for the main manuscript (as Fig. 2), which in general could use some more figures.

image

I'd either leave it as is, or potentially break out by drought, color-coded as in Fig. 1. What do you think, @mcgregorian1?

mcgregorian1 commented 4 years ago

This would work!

I think there are two ways this could go. Which do you prefer?

One is like this: image

while the other breaks the drought years into groups: image

mcgregorian1 commented 4 years ago

Personally I think the second one may be better? It emphasizes the comparison of the droughts as opposed to the species themselves.

teixeirak commented 4 years ago

Nice! Definitely the first.

teixeirak commented 4 years ago

Oops, just saw the second message. I’d still go with the first. I think it’s easier to read because it’s more compact.

mcgregorian1 commented 4 years ago

Ok! Sounds good, I'll change the axis labels and upload a more final one.

mcgregorian1 commented 4 years ago

Given this, do we want the NEON figure to be Fig. 3?

teixeirak commented 4 years ago

Given this, do we want the NEON figure to be Fig. 3?

Yes, at least for now. R1 suggested moving that to the SI, and we still haven't made that determination.

teixeirak commented 4 years ago

@mcgregorian1, a couple small things related to this:

teixeirak commented 4 years ago
mcgregorian1 commented 4 years ago
  • [x] On a somewhat related note, R1 asked for a test of significance of traits across species. Could you please run that as well?

Do you mean doing an anova just looking at TLP and PLA then?

mcgregorian1 commented 4 years ago

@mcgregorian1, a couple small things related to this:

  • [x] Please pull out the mean $Rt$ by species across all droughts-- at least for LITU and FAGR.

The table looks like this. I have saved it as a csv in the publications folder.

image

teixeirak commented 4 years ago

Does this match what's plotted in the figure near the top of this issue? It looks like these values are higher--perhaps because outliers haven't been discarded?

teixeirak commented 4 years ago

*

Do you mean doing an anova just looking at TLP and PLA then?

All, although I'm not yet sure if we'll show these results.

mcgregorian1 commented 4 years ago

Does this match what's plotted in the figure near the top of this issue? It looks like these values are higher--perhaps because outliers haven't been discarded?

The outliers in these plots are general outliers, not the ones that were removed before doing the analysis. And yes this is the same data, with the mean Rt per species with all droughts combined.

teixeirak commented 4 years ago

Okay, thanks!

mcgregorian1 commented 4 years ago

I ran the anova for the differences of Rt across sp. I'm not sure of the best way to present the results, especially as what's significant (<0.05, the black horizontal line in the plot) is different across years. Maybe instead of a figure we just say xx% of combinations in y drought scenario were significantly different?

I'll get to the anova between trait values later.

image

teixeirak commented 4 years ago

We could stick this figure in the SI, if you'd like.

mcgregorian1 commented 4 years ago

We could; I was assuming it would be going there. I still think there's a better way of representing what's significant and what's not, but it would take some more time for me to play around with it.

I can still get the raw %s later tonight for each year if you want - that would be easy to do as the data is already there.

teixeirak commented 4 years ago

A better way might be to show this sort of plot, although I'm not sure how you do it for all droughts (maybe show just the combined model?):

image

mcgregorian1 commented 4 years ago

True. I think it comes back to what we're trying to present and how. It seems the original figure is fine if we're emphasizing that the majority of the combinations are not significantly different from each other.

I just realized I still haven't done this same test between the different traits. I will have to do that later tonight

teixeirak commented 4 years ago

I don't think there's any particular message to emphasize with that figures; the idea would be just to provide thorough info in case readers are curious about significance of differences among species..

teixeirak commented 4 years ago

By the way, I added this as a temporary figure to the SI before sending out to coauthors.

mcgregorian1 commented 4 years ago
  • [ ] On a somewhat related note, R1 asked for a test of significance of traits across species. Could you please run that as well?

Looking at the data again, I'm not sure an anova can be done in the same way as we did for species vs Rt. At the very least I think we would be able to do this for height and TWI, because we have a random assortment of values across each of the species. For TLP and PLA, we have one value per species, whereas for position_all we have only counts. I suppose we could do anova for position vs Rt?

In total it seems we can only do the following:

I also re-read R1's comment and it seems they mentioned traits across species as an example for the main target of getting more figures in the paper, not because they were more interested in seeing this specific thing.

What do you think? image

teixeirak commented 4 years ago

From my phone, that looks like a start of a very useful graph. The main modification is that we do have three samples of the traits for each species. Those are in the hydraulic traits repo. I believe you have access, right?

mcgregorian1 commented 4 years ago

Using that data, I can get the following. I took out the year labels because the raw TLP and PLA don't have years associated with them, so I figured it would be a more cohesive figure if I removed year from height and TWI as well (especially as they were mostly similar).

Thus, we can do anova on these, but they would need to be done separately per trait (excluding crown position). Do you want to move forward with this? I don't know how much time I'll have tomorrow but I can focus on this on Wednesday.

image

teixeirak commented 4 years ago

I like that plot; let's include it in the SI. Please fix up the canopy position plot so that it's easier to read: darker and bigger symbols on a white background.

Let's run the ANOVA (separately for each independent variable) and add the letters to the plot.

mcgregorian1 commented 4 years ago

Ok sounds good. I will get to this tonight.

mcgregorian1 commented 4 years ago

I think this works better image

teixeirak commented 4 years ago

Yes, thanks! FYI, I already saved a copy of your older figure in the figures/publications folders, and have that pulled into the SI file.

mcgregorian1 commented 4 years ago
  • [x] On a somewhat related note, R1 asked for a test of significance of traits across species. Could you please run that as well?

I've run the significance tests via anova and TukeyHSD. To explain the groupings: "Species that are assigned the same letter are not significantly different from each other with regard to the tested variable". I have saved this as "Figure 5" but I wasn't sure what number we wanted this.

image

teixeirak commented 4 years ago

Looks great; thanks!

I think we want this as an SI figure. I currently have it as S3. Note that I recently changed some of the figure numbers (when I deleted previous fig. S3) without fixing the numbers on the file names or in the main text.

teixeirak commented 4 years ago

Species that are assigned the same letter are not significantly different from each other with regard to the tested variable"

Thanks; I included this in the caption.