enhanced treatment of vertical profiles

[teixeirak]

@mcgregorian1,

In part inspired by an NSF grant proposal that Neil and I (and others) just submitted, I think we could enrich the treatment of the vertical profile. I've become aware of more literature that I'd like to add, and this would also highlight the canopy position data/ give it a strong role despite being removed from the models (#112).

Please see my latest version of Table 1, which I'll push after I submit this issue. The new predictions there are not exactly novel, but good to document in the context of this study.

It would probably be good to do statistical tests of the significance of the height profile differences shown in Fig.3, which should be easy. Then, I would re-consider putting current Fig. 3 as either number 1 or 2, mirroring the order in Table 1.

I need to give this some more careful, focused attention tomorrow AM, but want to get the conversation started.

[teixeirak]

Note that I also think this change would help to compensate for the interesting introduction/ discussion content that is lost when we drop CP from consideration in the main model.

[mcgregorian1]

To clarify, you mean doing an anova between each vertical variable? Or do you mean anova for each variable for each height? (as in, 10m vs 20m vs 30m)

[teixeirak]

The test of interest would be how each varies with height. For simplicity, I'd test for differences just between the top and bottom of the height profile. Maybe we could annotate the plot with color-coded astricks indicating significant differences between top and bottom of profile (see screenshot for example of what I mean).

Let's also put letters indicating significance groupings in crown position by height.

[teixeirak]

On a related note, I've recently revisited your old version of this figure with biological temperature for a grant proposal and review paper on vertical temperature gradients led by current intern Nidhi Vinod (we'll probably be in touch about this later).

After reading more on thermal regulation by leaves, I think the findings make more sense; leaves regulate temperature pretty close to that of the air unless they don't have enough water for evaporative cooling, and then they really heat up. If you want to reinstate that panel, I feel that we could explain/ discuss it.

[mcgregorian1]

I could add it back in but that means we would have a 5-panel figure. I don't think that's a problem necessarily but if we had it out already then potentially let's keep it at that unless there's a strong reason (i.e. adding more to the analysis) for putting it back in?

On Fri, Jul 17, 2020 at 9:24 AM Kristina Anderson-Teixeira < notifications@github.com> wrote:

On a related note, I've recently revisited your old version of this figure with biological temperature for a grant proposal and review paper on vertical temperature gradients led by current intern Nidhi Vinod (we'll probably be in touch about this later). [image: image] https://user-images.githubusercontent.com/6355854/87790478-9b5ecc80-c80e-11ea-87fb-328d9c74edd3.png After reading more on thermal regulation by leaves, I think the findings make more sense; leaves regulate temperature pretty close to that of the air unless they don't have enough water for evaporative cooling, and then they really heat up. If you want to reinstate that panel, I feel that we could explain/ discuss it.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/SCBI-ForestGEO/McGregor_climate-sensitivity-variation/issues/115#issuecomment-660105451, or unsubscribe https://github.com/notifications/unsubscribe-auth/AJNRBEKBKDJHQRIEOPI56BTR4BGJ5ANCNFSM4O5AF2WA .

--

Ian McGregor

Ph.D. Student | Center for Geospatial Analytics

He/Him/His

College of Natural Resources

Jordan Hall 4120 | Campus Box 7106

North Carolina State University

2800 Faucette Dr.

Raleigh, NC 27695 USA imcgreg@ncsu.edu | 714-864-1005 | geospatial.ncsu.edu

[teixeirak]

Yeah, probably just leave it out.

[teixeirak]

A minor cosmetic thing: Please write out the canopy position names if possible. I'm thinking those should be removed from table 2.

[mcgregorian1]

Question for this - in the plot, I split up min and max by taking the max value for each height for each day. Then, what's plotted is the mean of those values (same for min).

I can run a significance test on the maxes (top vs bottom) and the mins (top vs bottom) and put the asterisks as you suggested. But I'm wondering if it would be better to do a significance test on all values (top vs bottom) instead?

I will continue doing what you initially suggested for now.

[teixeirak]

Both would be relevant, and give us different info. I'd stick with the mins and maxes separate.

[mcgregorian1]

ok sounds good! sorry it's taking so long; I had to fix some stuff with the neon data

[mcgregorian1]

I've made the following. For the comparison between the top and bottom of the height profiles, I tested to see whether the data for each sub-sub group was normal (shapiro). If it was, then I ran a t-test, otherwise I did a wilcox. The asterisk denotes whether either of those was significant.

For the boxplots, I did an anova. Ultimately they're all significantly different from each other (The c-d group barely made it). I wasn't sure how you wanted to denote that, so I just put ABCD as being the different "groups" of comparison.

[mcgregorian1]

This is saved now as Fig 2

[teixeirak]

Perfect, thanks!