Keeping in fraxinus nigra

[mcgregorian1]

Hi @teixeirak

In my next steps document, I put down something we talked about for possibly removing frni from analyses. I have 11 subcanopy cores and 1 canopy core (together now), and I'm forgetting why we were thinking of excluding frni.

In either case, here are the 10 best full model runs and the coefficients of the top models with and without frni:

Biophysical

without frni

with frni

leaf traits

without frni

with frni

[mcgregorian1]

One thing that I noticed in this is that keeping in or taking out frni only applies to the biophysical model runs, since we only have trait data for "cagl", "litu", "qual", "qupr", "quru", "quve"

[teixeirak]

One thing that I noticed in this is that keeping in or taking out frni only applies to the biophysical model runs, since we only have trait data for "cagl", "litu", "qual", "qupr", "quru", "quve"

Those are the only species with p50 and p80, but we have TLP and the other traits for the top 12 most productive species (all but FRNI). As including vs excluding has minimal impact on the results, it doesn't really matter what you do, but presentation will be a bit more straightforward if we drop it. For example,

without: "we considered the top 12 most productive species in our plot"

with: "we considered the top 12 most productive species in our plot, plus F. nigra (19th most productive)" (and that raises the question why not the ones in between)

and

without: "we ran a biophysical model and a traits model"

with: "we ran a biophysical model on all species and a traits model on the species for which we had traits data"

---

My opinion is that we should drop it.

[teixeirak]

Off-topic for this issue, but in the model above, it looks like canopy position has reverted to the binary category. (I assume that's not what you want.)

[mcgregorian1]

Off-topic for this issue, but in the model above, it looks like canopy position has reverted to the binary category. (I assume that's not what you want.)

Whoops, you're right, I had had it switched from when we were talking about potentially moving back to the binary version

[mcgregorian1]

One thing that I noticed in this is that keeping in or taking out frni only applies to the biophysical model runs, since we only have trait data for "cagl", "litu", "qual", "qupr", "quru", "quve"

Those are the only species with p50 and p80, but we have TLP and the other traits for the top 12 most productive species (all but FRNI). As including vs excluding has minimal impact on the results, it doesn't really matter what you do, but presentation will be a bit more straightforward if we drop it.

Ok that sounds fine; I can take it out.

One thing to note is that I can't run models with any missing data. Because we have p50 and p80 for only those 6 big species, I'd have to either drop those traits in order to get the other species with TLP and the other traits, or I include those in as a complementary analysis thing

[teixeirak]

Since p50 and p80 aren't coming out in the top model, we don't have much of a dilemma. We can present an analysis similar to the one shown above (probably just in SI material), cite it as evidence that we're not missing anything when we drop p50 and p80 from the full model, and run the model with all species.

[mcgregorian1]

Ok, sounds good. I'll make another .md of this and have it be part of the manuscript document