Remaining questions and checkpoint

[mcgregorian1]

Hi @teixeirak, these are just some questions I wanted to ask before going to CA

1. Are we officially referring to "dominant", "codominant", etc as crown positions, canopy positions, or canopy classes? I think crown positions would be the least confusing, since for part of the analysis (choosing the pointer years) I refer to the categories as "canopy" and "subcanopy".

2. Do we want to include r2 values in the table showing the coefficients of the best model, given that the r2 value for the best biophysical one, e.g., is only 0.105?

3. I have this under the graphs section in my [next_steps]() doc: "each trait plotted against height (logged values of height), possibly by weighted mean/% (Meakem made code for this when she was here, ask valentine if know anything)". I asked Valentine but she said she didn't know what this referred to. Thinking about it myself more, how were you thinking this graph would look?

   • I can make a combined graph that looks like this:

4. I think the next step is to re-format how the test models should be run (at least for me because I'm a little confused). I know you mentioned we would use the null model as the best model, then if we test for height, for example, we'd take out height from the null. However, it seems the other hypothesis testing should be different from what we currently have (such as the P1.3b about dbh*elev interaction). What do you think? I believe that's the last thing I need to do before just writing and formatting.

Other things

5. I've explained the way to review the html file and the .Rmd files in #30, but let me know if anything is unclear.

6. I have run the model with height, position, and the hydraulic traits (minus p50 and p80 like we talked about). Interesting that position doesn't come out in the top model. Also interesting that basically all the traits are coming out as significant

Here are the coefficients:

[teixeirak]

1. Are we officially referring to "dominant", "codominant", etc as crown positions, canopy positions, or canopy classes? I think crown positions would be the least confusing, since for part of the analysis (choosing the pointer years) I refer to the categories as "canopy" and "subcanopy".

crown position.

[teixeirak]

2\. Do we want to include r2 values in the table showing the coefficients of the best model, given that the r2 value for the best biophysical one, e.g., is only 0.105?

please include, at least for now.

[teixeirak]

3\. I have this under the graphs section in my next_steps doc:
    "each trait plotted against height (logged values of height), possibly by weighted mean/% (Meakem made code for this when she was here, ask valentine if know anything)". I asked Valentine but she said she didn't know what this referred to. Thinking about it myself more, how were you thinking this graph would look?

1. Use census data (whole community), not just the set of trees for which we have cores.
2. Bin the trees in 5m (tentative- can adjust) increments of max height.
3. Assign TLP to each individual based on species mean.
4. Calculate mean and standard deviations of trait values across all trees in each height bin. Ignore NAs (species with no trait measurement), but record what percent of trees have no values. We should define some threshold (75% of trees with data??) below which we don't report/plot values.
5. make plots similar to those you made for the climate variables, but including SD (and add SD on those plots as well).

[teixeirak]

4. I think the next step is to re-format how the test models should be run (at least for me because I'm a little confused). I know you mentioned we would use the null model as the best model, then if we test for height, for example, we'd take out height from the null. However, it seems the other hypothesis testing should be different from what we currently have (such as the P1.3b about dbh*elev interaction). What do you think? I believe that's the last thing I need to do before just writing and formatting.

Previously, we decided that we'd create results tables based on both the best model and a minimal model (just random effects + year, in the case of the biophysical model) and then present whichever we thought would be clearer to communicate. In the case of this interaction term, the null should be dbh (or height) + elev.

[teixeirak]

1. I've explained the way to review the html file and the .Rmd files in #30, but let me know if anything is unclear.

Thanks! Haven't had a chance to look at it yet. I'll let you know if anything is unclear.

[teixeirak]

2\. I have run the model with height, position, and the hydraulic traits (minus p50 and p80 like we talked about). Interesting that position doesn't come out in the top model. Also interesting that basically all the traits are coming out as significant

Interesting! I'll come back to this. But one minor glitch-- it looks like year is back to being a linear effect as opposed to a categorical variable.

[mcgregorian1]

2\. I have run the model with height, position, and the hydraulic traits (minus p50 and p80 like we talked about). Interesting that position doesn't come out in the top model. Also interesting that basically all the traits are coming out as significant

Interesting! I'll come back to this. But one minor glitch-- it looks like year is back to being a linear effect as opposed to a categorical variable.

Whoops, my bad on that. Here are the models again (notice the deltaAIC now), with the correct coefficients.

[teixeirak]

I'm not sure I trust this (biologically); it may be that we have too many traits in there at once and they are interacting in unexpected ways. That's why I want to run them individually (issue #36). For example, we'd typically expect that resistance would increase with WD and increase with LMA, but here they're going in opposite directions.

[teixeirak]

Closing (obsolete).