criteria for excluding NPP measures

[teixeirak]

From Helene:

I was puzzled by some of the criteria for excluding NPP measures. In particular, the text said that measures of NPP that included reproductive structures, NPP lost to herbivory, understory, VOCs, and/or exudates were excluded. Why, and for that matter how? Reproductive structures are usually part of litterfall, so wouldn’t pretty much every measure of NPP include these either implicitly or explicitly, and is this a case of only excluding them when they are mentioned explicitly? And all of these are generally quite small, so does it really make sense to exclude sites that include them? Or does this mean in practice that the analyses still included these sites but without these fractions?

[teixeirak]

@beckybanbury, could you please verify that the info in Table 2, up to woody-stem ANPP, is correct?

[beckybanbury]

@teixeirak in order to keep consistent across C fluxes, I only used NPP_1 and ANPP_1 and ANPP_2 (so that the definition of NPP used includes all the components of ANPP included). This should mean that none of the NPP or ANPP values used include reproductive or understory.

In terms of what Helene has commented; the way that these measures are excluded is just based on the structure of the ForC database; ForC should classify measures based on whether or not they specifically state the components included, yes? And if components aren't specified, the classification would be NPP/ANPP_0? I didn't use measures where components weren't specified to try to standardise as far as possible.

Perhaps methodologies used should be changed to "typical methodologies used" to make that part clearer...

[teixeirak]

Hmmmm.... this may point to some inadequacies in the ForC variable descriptions. In particular, I think Helene is right that a lot of these would include litterfall measures that include leaves, twigs, and reproductive structures. I'll need to look at this issue more and adivise.

One option may be to broaden our criteria to include all ANPP and NPP measures meeting the minimum criteria. That would obviously require re-running results... Would that be too much work?

[teixeirak]

One option may be to broaden our criteria to include all ANPP and NPP measures meeting the minimum criteria. That would obviously require re-running results... Would that be too much work?

@beckybanbury, looking at this a little more carefully, I do think this is likely our best option. Please let me know what you think.

[beckybanbury]

It probably wouldn't be too much work, but the main reason I limited it was to keep consistency between fluxes e.g. for the flux sums. For example, in ForC there is no measure of ANPP that explicitly includes reproductive structures or herbivory, so I was concerned that using all measures of NPP introduced inconsistencies when looking to compare between ANPP and NPP. This is also why I used ANPP_foliage instead of ANPP_litterfall (which can include twigs and reproductive structures), as theoretically (according to the ForC variable definitions) ANPP_2 = ANPP_foliage + ANPP_woody. Of course, these components are likely to be small and so may not have that much influence anyway - perhaps it doesn't matter so much to be so rigorous!

I think this partly comes down to how the ForC variable definitions are interpreted...!

[teixeirak]

in ForC there is no measure of ANPP that explicitly includes reproductive structures or herbivory, so I was concerned that using all measures of NPP introduced inconsistencies when looking to compare between ANPP and NPP.

Unfortunately, it looks like the variable definitions for ANPP were inadequate. Looking at the specific methods listed for ANPP in the methods table, they all specify "litterfall" and not "foliage". I think Helene's right that ANPP would most commonly include reproductive structures, and unfortunately ForC doesn't have the structure to reliably differentiate between ANPP with strictly foliage and with all litterfall. I've changed the ForC variables table accordingly.

[beckybanbury]

Okay! So when I rerun, I should include all definitions of NPP and ANPP, with the exception of NPP_0 and ANPP_0, where variables are undefined?

[teixeirak]

This is also why I used ANPP_foliage instead of ANPP_litterfall (which can include twigs and reproductive structures), as theoretically (according to the ForC variable definitions) ANPP_2 = ANPP_foliage + ANPP_woody. Of course, these components are likely to be small and so may not have that much influence anyway

In this case, I think I'd probably keep foliage separate, although either way would probably be acceptable. This would parallel our philosophy with ANPP_woody_stem, where we keep out branchfall and go with the option where we have more data (ANPP_woody_stem instead of ANPP_woody).

ANPP_repro, averaged across biomes (see C cycle diagrams in ForC repository), is always ≤0.32. Branch fall can be much more signficant.

[teixeirak]

Okay! So when I rerun, I should include all definitions of NPP and ANPP, with the exception of NPP_0 and ANPP_0, where variables are undefined?

Actually, ANPP_0 and NPP_0 should both meet the minimum criteria currently given in Table 2, so I think they could be included.

[beckybanbury]

Which analyses do we want to change this for - everything? Currently Fig. 2 sums are specific to the components each flux definition includes, but should this be changed to include all measures of the flux regardless of components? Similarly, should I include all measures of NPP/ANPP in the allocation tests?

[beckybanbury]

I've just pushed some updated results including all measures of NPP + ANPP

[teixeirak]

Which analyses do we want to change this for - everything? Currently Fig. 2 sums are specific to the components each flux definition includes, but should this be changed to include all measures of the flux regardless of components? Similarly, should I include all measures of NPP/ANPP in the allocation tests?

Sorry for the slow reply on this. It looks like you've already made a decision on this (?). In any case, yes, please update all. In Fig. 2, I think this only affects ANPP=ANPP_woody-stem + ANPP_foliage. In reality, ANPP is often going to include ANPP_littefall (ANPP_foliage+ANPP_repro), so updating will be more realistic.

[teixeirak]

I've just pushed some updated results including all measures of NPP + ANPP

Did anything change significantly?

[teixeirak]

@beckybanbury, I see that you've updated some of the graphs, but it looks like (most of?) the ones in the manuscript are still the previous version. This is just a reminder to do those, in case it's not already on your radar.

[beckybanbury]

@teixeirak all the graphs should be updated, but I'm having trouble knitting the manuscript from my computer - could you update it and then we'll be able to check the graphs have updated.

[beckybanbury]

Thanks, that looks great!