So we’re working on getting the methods ready for the Real Food Campaign lab, and one of those methods is mineralizable carbon. We’re following a method similar to the Solvita “Burst” type method (24 hour incubation after adding water) but instead using a CO2 sensor rather than a solvita type test (cheaper, easier).
Here’s the problem - you get a different ‘burst’ based on the soil type and amount of water you add. Ideally, you want to add 50% of the water holding capacity of the soil (abbreviated AWC). Problem is, different soil types and amounts of organic matter impact the water holding capacity.
So, we have a few options:
- Do what solvita does and just add a fixed amount of water, and to hell with the error.
- If we know the soil type (and better yet the organic matter), we can use this formula http://www.weap21.org/WebHelp/Mabia_Alg_AWC.htm (scroll to the bottom) to estimate the AWC and therefore the amount of water to add. We will be measuring Total Organic Carbon, which probably relates to organic matter but isn’t it.
- We could manually measure AWC on each sample that comes in.
- We could check the range of possible water additions (maybe 3g - 9g max) from all soil types (or at least 99% of them), and choose one which is >50% for all soil types but less than 100% for all soil types… does that exist? What is the biological impact of 25% AWC versus 90% AWC versus 50% AWC? My guess is here is a reasonable cutoff there where activity peters off.
- Perhaps we just use the 3 day? That way the activity has hopefully topped out and isn’t going up any more, at which point the moisture level is less relevant?
… Given that we’re trying to keep costs down and keep methods simple, (3) above seems like a tough sell. (1) is ok but I’m not sure how much error it’ll produce. I love (2) but we’d need to know the GPS location with some accuracy (not impossible) or we’d need to determine soil type in the lab quickly.
My starting points for next steps -->
- Perhaps we can simplify things by putting values in rough boxes… like “sandy”, “sandy-loamy”, “loamy”, “loamy-clay” and “clay” is good enough? We could probably do that in the lab. That may get a big quality improvement without a lot of effort.
- Also, could we guess at or are there standard calculations to get organic matter from total organic carbon? My guess that would improve our accuracy, and since we’re doing total organic carbon on each sample, that’s a data point we’ll have anyway.
@snapp @DanT @kanedan29 Ideas? Thoughts? Crushing realizations? Fantastic solutions? I’ll take it all
I don’t agree with going with 1 (what solvita does – this does not work, might as well use a dartboard)
I think trying 2 would be good – what do you think Xinyi??
I mean 1 in the original post!
There is a standard calc of C from SOM and vv
However, SOM by loss on ignition is going to be a different measure (correlated with of course) dry combustion C…
Thanks for this research, I hv learnt some things from it.
I like option 5 for what it’s worth. If a consistent peak can be achieved across soil types, this seems like a strong option and easy to standardize across soil types.
Down side is @DanT says that some soils can actually begin to dip back down after a time… actually, Dan could you follow up here with our most recent thinking on this, with references if you got them?
The key to short term incubations for mineralizable Carbon is the flush of CO2 released following the rewetting of dried soil. This flush is going to be strongly affected by the amount of water added. Add too much and the soil biological activity is reduced by anaerobic conditions, add too little the the moisture level is not optimum for maximum biological activity.
So, it is true that 3 days after rewetting the soil the rate of CO2 evolution may be the same, but the peak from the initial flush of activity will be different.
Check out this paper: https://pdfs.semanticscholar.org/94d8/0ace84da21e43ed38486631be4e044cf5b90.pdf
Thanks for that. Riffing on the 5th option still.
If the soils are being dried then the user could do a before drying and after drying weight to determine the amount of water in the sample. Then to re-wet the sample they could add the same amount back in by weighing it out. This could account for major differences in soil texture and would ride on an assumption that people take their sample from a generally moist soil (not dry and wet, somewhere in-between). Sample depth could be a factor here, perhaps focusing on 3-6 inches which would be more consistent in terms of moisture levels throughout the sample. Thoughts?
Could the rewet soil moisture be standardized by weighing soils before after drying? Say a soil came in at 50% moisture. The user weighted 20 grams sampled from the field and after drying it the soil weighted 10 grams. The user would know to add 10 grams of water back to get it to 50% moisture but say 25% moisture is a chosen optimum for rewetting soils then the user could calculate an addition at 5 grams of water back.
This method would not account for the holding capacity of soils though. Could the rewet % depend on soil CEC as a rough proxy for texture (CEC is a common measure used on soil tests and remains pretty consistent year to year)? So maybe with a cec of 7, which would be a very sandy soil, the user adds 25% of the soil weight back in water, a clay soil might have a cec of 20, and the user adds 50% of the soil dry weight in water. Soil moisture values will vary depending on soil texture. A sandy soil will reach 100% saturation with less water than a clay soil. So by using CEC or another (even rough) measure of soil texture, a somewhat consistent soil moisture level for rewetting could be achieved across a variety of soils using CEC data that most farmers have.
Found one study that gives a significant 56% regression correlation between CEC and soil water holding content (not capacity) - capacity might give a higher correlation:
Great minds @harrid Last I remember our discussion (@DanT and @snapp) we also noted the correlation between CEC and soil water holding, and apparently there is an API or calculation for any given soil for the CEC. So the latest plan I remember was to auto-calculate CEC @ 50% exactly as you suggest!
That being said, all of these things are noisy - 56% correlation isn’t exactly good as you point out - but it’s better than just adding a fixed amount for everything. Also, even with the location to location noise, for a single soil sample or single farm tracking data over time, it’s a pretty useful and simple tool.
Dan or Sieg do you remember the source of the CEC --> water holding capacity calculator thingy (hopefully I’m not just making that up, I’m pretty sure it was discussed)?