Agriculture as the Model Case

Volume IV — Production and Regeneration

Nine movements, one hectare.

THE PLATE

A watercolour of two trees side by side, one green and one turning gold, their roots in the same earth. — Plate IV.03The Core, at Eleven in the Morning.Every argument in this chapter is about that line: how far down it goes, how long it took to get there, and how quickly it can be put back.

THE LETTER

Agriculture is the model case for regenerative production because it is the only large industry whose principal asset is alive, whose depreciation is visible in a hand, and whose accounts can be made to run backwards. A factory wears out. A soil does not have to.

That is why this chapter is the empirical centre of the volume. It is also why it is the easiest place in the whole edition to be caught overstating, and being caught here is expensive: a teacher who finds one inflated soil-carbon figure will disbelieve the other seventy-six chapters, and will be right to.

So this chapter is written to a stricter rule than the others. Every number below appears in lib/verify/IV_03.py with a label saying whether it is a published measurement, a declared modelling assumption, or a calculation from the two. You can read the module in ten minutes. The labels exist because an assumption dressed as a measurement is the single most common way an argument like this one goes wrong, and you should not have to take anybody's word for which is which.

What you will find, if you read to the end, is that the case for regenerative agriculture is strong, and that it is not strong for the reasons it is usually sold on. Soil carbon saturates. It reverses faster than it accumulates. It is expensive to measure relative to what it is worth. The yield gap in some systems is real and persistent. Every one of those is true, and the case survives all four — because the money is not in the carbon. The money is in the input bill, the price premium, the drought year, and the fact that an appreciating asset should not be depreciating on a schedule that assumes it is falling.

This is the chapter to hand to the person who does not believe you.

— The Editors

DISCOVERY

What is already working

Agriculture keeps better long-run records than any other industry on earth, because somebody in the nineteenth century had the patience to leave a field alone and write down what happened. That patience is the field's great gift: we are not reasoning from models here. We are reading from experiments that have been running for up to a hundred and eighty years.

Broadbalk, Rothamsted, Hertfordshire — established 1843. The longest continuously running agricultural experiment in the world. One of its plots has received thirty-five tonnes of farmyard manure per hectare every year since the start. Its topsoil organic carbon has risen from roughly 28 to roughly 85 tonnes of carbon per hectare — a tripling, held for the better part of two centuries, on a plot that has grown wheat the whole time. The unmanured plot beside it has sat flat. Whatever else is arguable, this is not: soil carbon can be built, and it can be held for a hundred and eighty years by people who simply kept doing the thing.

The Rodale Institute Farming Systems Trial, Kutztown, Pennsylvania — established 1981. Three systems side by side: a conventional corn–soybean system, an organic system with animal manure, and an organic system relying on legumes. Over the trial's first twenty-two years, mean corn yields came out at 7,450 kg/ha conventional, 7,400 kg/ha organic-animal and 7,170 kg/ha organic-legume. That is 0.7 percent and 3.8 percent below conventional — inside the trial's own year-to-year variation. The organic systems did it with about 30 percent less fossil energy. And in the drought years the organic systems did not merely hold; the 30-year report puts their advantage at around 31 percent, which is what happens when a soil with more organic matter holds more water.

The DOK trial, Therwil, Switzerland — established 1978. Twenty-one years of side-by-side comparison, reported in Science in 2002. Organic yields averaged 20 percent below conventional. Wheat was only 10 percent below; potatoes were 34 to 42 percent below. And the inputs: fertiliser and energy down 34 to 53 percent, pesticide down 97 percent. Hold those two facts against each other and the interesting number falls out — output per unit of input rose by somewhere between 21 and 70 percent. The DOK trial is not a story about producing more. It is a story about producing nearly as much with half the purchased world.

Marsden Farm, Iowa State University. The case that ought to be more famous than it is, because it has no premium in it and no certification. Three cropping systems: the standard two-year corn–soybean rotation, a three-year rotation, and a four-year rotation with a small grain and a legume forage. Corn yielded 4 percent higher in the three-year and 9 percent higher in the four-year rotation. Soybean yielded 9 and 12 percent higher. Synthetic nitrogen fell by 86 percent, herbicide by 88 percent, and freshwater toxicity by about two hundred-fold. Profitability held. More yield, less input, no premium, no label — obtained purely by lengthening the rotation, which is to say by putting time back into a system that had been asked to run without it.

Brown's Ranch, Bismarck, North Dakota. Gabe Brown's own account, in Dirt to Soil, of taking soil organic matter on a working farm from around 1.7 percent to over 6 percent across two decades of no-till, cover cropping and integrated grazing. It is a farm record rather than a replicated trial, and it should be read as one — but it is a farm record kept by somebody who was selling the output, which is a different kind of evidence and not a worse one.

Five cases, one pattern, and the pattern is worth stating precisely because the rest of the chapter depends on it. In every one, the gain came from putting back a biological process that had been replaced by a purchased input — manure for fertility, a legume for nitrogen, a longer rotation for weed and pest control, roots in the ground for structure and water. Nothing here is mystical. It is substitution, run in the direction nobody in the twentieth century was paid to run it.

THE ARITHMETIC

What works, what does not, and where the line sits

First, the size of the prize, honestly.

The world's soils hold about 700 Gt of carbon in the top 30 cm and about 1,500 Gt in the top metre. Human land use has removed something like 133 Gt of it over twelve thousand years. So there is a hole, the hole is enormous, and something once lived in it. That is the true part of the story, and it is the part the advocacy gets right.

The 4-per-1000 initiative proposes raising soil carbon by 0.4 percent a year. Do the arithmetic and the claim divides cleanly into a true half and an unsupported one:

  0.004 × 1,500 Gt C (top metre)   =  6.0 Gt C/yr
  0.004 ×   700 Gt C (top 30 cm)   =  2.8 Gt C/yr
  atmospheric growth, 2.4 ppm × 2.124 Gt C/ppm  =  5.10 Gt C/yr

The top-30-cm target would cover 55 percent of the annual atmospheric increase. That is the true half, and it is genuinely large.

Now the unsupported half. Measured sequestration rates on working cropland, from the syntheses rather than the brochures: West and Post's analysis of 67 long-term experiments gives 0.57 ± 0.14 t C/ha/yr for a switch from conventional tillage to no-till, and 0.20 ± 0.12 t C/ha/yr for increasing rotation complexity — note that the second interval's lower bound crosses zero. Minasny and colleagues put the broadly achievable range at 0.2 to 0.5 t C/ha/yr. Lal's cropland range is 0.05 to 0.76. Take 0.30 t C/ha/yr as a central figure for global cropland, which sits inside all three:

  1.6 Gha cropland × 0.30 t C/ha/yr  =  0.48 Gt C/yr
    →  17 % of the 2.8 Gt target
    →  9.4 % of the annual atmospheric increase
    →  1.76 Gt CO₂/yr, or 4.8 % of global fossil CO₂

Four point eight percent. That is what the world's cropland can actually contribute, and it is worth having — it is larger than most national emissions totals — but it is not a climate solution and a chapter that says it is has spent its credibility on the wrong sentence.

Second, the honest negative that governs everything: soil carbon saturates.

Take Broadbalk and fit the standard first-order model, C(t) = C∞ − (C∞ − C₀)e^(−kt), through 28 t C/ha in 1843 and 85 t C/ha today, assuming an equilibrium of 90. That gives k = 0.01399/yr, a half-life on the remaining gap of 49.6 years, and this:

  year    0     stock 28.0 t C/ha     rate 0.867 t C/ha/yr
  year   20     stock 43.1            rate 0.656
  year   50     stock 59.2            rate 0.431
  year  100     stock 74.7            rate 0.214
  year  180     stock 85.0            rate 0.070

The sink does not close; it decays. And here is the number that should be printed on the front of every soil-carbon prospectus: 4 per 1000 of Broadbalk's current stock would be 0.34 t C/ha/yr, and the plot is currently delivering 0.070 — 21 percent of the rate the target asks of every soil on earth, on the most heavily manured arable soil on record, after a hundred and eighty years of thirty-five tonnes of manure a year. Assume a higher equilibrium and the present rate rises to 0.106 or 0.131; it does not approach 0.34 under any assumption the data will carry.

Third: it reverses, and it reverses faster than it builds.

Davidson and Ackerman's synthesis of cultivation studies found 20 to 40 percent of soil organic carbon lost when previously untilled soils are brought into cultivation, centred near 30 percent, most of it inside about twenty years. Apply that to Broadbalk:

  85.0 t C/ha × 30 %  =  25.5 t C/ha lost
  25.5 as a share of the 57 t C/ha built over 180 years  =  45 %
  loss rate 25.5 / 20  =  1.275 t C/ha/yr
  build rate  57 / 180  =  0.317 t C/ha/yr
  →  carbon leaves 4.0 × faster than it arrives

Nearly half of a hundred and eighty years of accumulation, gone in twenty, at four times the speed it was laid down. Which means a soil carbon credit is not a sale. It is a tenancy, and it must be priced as one. Under a 1 percent annual reversal hazard, only 36.6 percent of a credit's face value survives a century; at 2 percent, 13.3 percent does. A ten-year vintage carries a 9.6 percent reversal probability, which is why the instrument in the last movement holds a 22 percent buffer rather than a token one.

Fourth: measuring it costs more than it is worth — at the wrong scale.

This is the negative nobody prices and it decides more soil-carbon projects than agronomy does. To detect a change you need to see a small signal against a large, spatially variable stock. Take a 50 t C/ha stock, a within-field spatial coefficient of variation of 15 percent, and a four-year contract at 0.30 t C/ha/yr — so you are hunting for 1.2 t C/ha against a standard deviation of 7.5.

  n  =  2 (z₀.₉₇₅ + z₀.₈₀)² σ² / Δ²
     =  2 × 7.8490 × 7.50² / 1.20²
     =  614 cores per group

Six hundred and fourteen. Twice, for before and after, at $25 a sample, is $30,700 — for one stratum. The four years bought 4.40 t CO₂e per hectare. On a 65-hectare field that is 286 tonnes, worth $8,580 at $30. The verification cost $107.34 per tonne. The measurement costs three and a half times the carbon.

Two things fix it, and both are design decisions rather than discoveries. Sample the same points twice: the variance that matters becomes the variance of the change, not of the stock, which Smith and colleagues put nearer 6 percent. Then n = 7.8490 × 3.00² / 1.20² = 50 points, re-sampled, for $2,500 — twelve times cheaper, and the break-even area falls from 232.6 hectares to 18.9.

And then the cut. Look again at what the sample count depends on: variance and effect size. Not area. The cost of knowing scales with the number of soil types, not the number of hectares. Five strata across a 3,000-hectare co-operative cost $12,500 to verify properly — $4.17 a hectare against $132 a hectare of carbon, netting $383,500 across the membership over four years. The same soil, the same practice and the same carbon are worth −$77.34 a tonne on one farm and +$22.45 a tonne across a co-operative.

The binding constraint on soil carbon is not agronomy. It is aggregation. That is not a finance detail bolted onto an agronomic argument. It is the argument: this carbon exists at co-operative scale and does not exist at farm scale, and the entity that is missing is not a technology but a counterparty.

Fifth: the yield gap is real, and it is smaller than its reputation and larger than its advocates.

Three meta-analyses, independently constructed. Seufert and colleagues: 25 percent lower on average, 13 percent under best organic practice, 34 percent against the strongest conventional comparison. De Ponti and colleagues: 20 percent, across 362 paired comparisons. Ponisio and colleagues: 19.2 ± 3.7 percent, falling to 8 or 9 percent where rotations and multi-cropping are used. Mean of the four central estimates, DOK included: 21.1 percent.

The land arithmetic is the part that must not be dodged. At a 19.2 percent gap you need 23.8 percent more land for the same output; at 25 percent, 33.3 percent more; with diversification at 9 percent, 9.9 percent more. A yield gap is a land-use claim and it should be stated as one. The response is not to deny the gap. It is Marsden and Ponisio: the gap is largely a property of the substitution — swapping an input out and putting nothing structural back — rather than of the ecology. Lengthen the rotation and most of it closes, in a system that needs no certificate and no premium.

DREAM

What becomes ordinary

In the version of this that has already happened, a soil test is a financial statement and everybody in the chain treats it as one.

The farm's balance sheet carries two land lines. One is the land at cost, as it always was. The second is the soil carbon stock, in tonnes per hectare, measured on a stratified grid every four years by the co-operative, and it moves. When it rises, the farm's borrowing base rises with it, because the lender has learned what the agronomists have known for a century: a soil with more organic matter holds more water, needs less nitrogen, and does not fail in the dry year. The drought premium — the 31 percent the Farming Systems Trial measures in the hard seasons — is priced into the loan rather than discovered afterwards in the default.

The co-operative runs the measurement, because that is the scale at which measurement is affordable and every member knows it. Five strata, fifty paired points each, re-sampled on a fixed rotation. The cost is four dollars a hectare and nobody argues about it, in the way nobody argues about the cost of the scale in the grain elevator.

Transition is financed the way any capital project is financed. A farmer converting four hundred hectares knows the number before she starts — $324,000 of working capital across three years, and she knows it because the figure is published and her neighbours' actual results are in it. The bridge is drawn down monthly, repaid from the premium, and the lender's covenant is written against the soil test rather than against the tractor.

Rotation length is a covenant term. A four-year rotation is a lower-risk asset than a two-year one, and the insurance market has noticed, so the premium schedule says so. Farmers are not being asked to be virtuous; they are being quoted a better rate for a better asset.

And the argument has stopped being about whether any of this is real, because the long trials are in the textbook and every agronomy graduate can compute a saturation curve and a minimum detectable difference before they are allowed to sell anybody a carbon contract. The change is not that people became persuaded. It is that the numbers became cheap enough to check.

DESIGN

The structure that gets there

Four pieces, in order, and the order matters because each one funds the next.

One — the rotation, before anything else. Marsden is the cheapest move in this chapter and the only one that requires no counterparty, no certificate and no premium. Lengthen the rotation from two years to three or four with a small grain and a legume forage. The measured result is 4 to 9 percent more corn, 9 to 12 percent more soybean, 86 percent less synthetic nitrogen and 88 percent less herbicide, at comparable profit. Do this whether or not anything else in this chapter ever happens.

Two — the paired grid, before any commitment is made. Fifty georeferenced points per stratum, cored to 30 cm, bulk density measured, samples archived. Archiving is the part people skip and it is the part that pays: an archived baseline sample can be re-run against a future method, which means your baseline survives a change in the protocol. Without it, a methodology revision destroys your evidence and there is no appeal.

Three — the aggregation, because it is the constraint. The measuring entity must sit above the farm. A co-operative, a grain buyer, a catchment body, a lender's agricultural book — the legal form matters much less than the area under one stratification. Below roughly 230 hectares per stratum on an unpaired design, or 19 on a paired one, verified soil carbon is a net cost. Above it, it is income. This single threshold decides whether the market exists.

Four — the transition bridge, which is the only place real money is needed. The working capital hole is three years deep and it is computable in advance:

  transition year 1   revenue 1,600 × 0.70 = 1,120   margin 100   shortfall 350
  transition year 2   revenue 1,600 × 0.75 = 1,200   margin 180   shortfall 270
  transition year 3   revenue 1,600 × 0.80 = 1,280   margin 260   shortfall 190
                                              the hole  =  810 USD/ha

Governance, briefly, because it is where these fail. The measuring entity must not be the selling entity — the co-operative that verifies the carbon should not also be taking a commission on its sale, and if it must, the verifier is independently contracted and the contract is published to members. Ostrom's design principles are the right checklist here: clear boundaries, rules matched to local conditions, collective choice by those affected, graduated sanctions, and the monitoring done by people accountable to the members rather than to the buyer. A soil-carbon co-operative is a commons institution wearing a carbon contract, and it will fail in exactly the ways commons institutions fail.

DESTINY

How it holds when nobody is pushing

Three things make this self-sustaining, and each corresponds to one of the honest negatives above, which is not an accident.

Saturation is answered by staging. Because the rate decays, a farm should be converting new ground rather than pushing an already-rich soil for another increment. A co-operative with members at different stages has a flat aggregate sequestration profile even though every individual field's rate is falling. The portfolio holds what the parcel cannot.

Reversal is answered by the buffer and the covenant. Twenty-two percent of issued credits held back, ten-year vintages, re-verification on a fixed cycle, and a reversal clause that cancels from the buffer rather than pursuing the farmer. A farmer who fears a clawback in a bad year will till, and then you have lost the carbon and the member.

Measurement cost is answered by the grid outliving the project. Once the paired points exist and the archive exists, the marginal cost of the next verification is the sampling round alone. The expensive part is the first one, which is precisely why it should be paid for by an entity that will still be there in twenty years.

Now, plainly, where this fails. It fails when the premium compresses: at the modelled cost structure the break-even organic premium is 13.7 percent, and a market that oversupplies can go through that. It fails when the baseline is set in a wet year and the verification lands in a dry one, and nobody agreed a weather normalisation in advance. It fails when the co-operative's measurement arm is paid on volume sold. It fails when a farm sells ten-year carbon and changes hands in year four with nothing running with the land. And it fails, most commonly, when somebody sizes the project on gross soil carbon rather than on the difference between the practice and the counterfactual — at Rodale that difference is 0.439 t C/ha/yr, not the 0.634 the organic-animal system shows on its own, and the difference is the only figure a carbon contract may honestly pay for.

DELIGHT

What it feels like

The pleasure here is tactile and it arrives early. It is the first spring you push a spade into ground you have been working differently for three years and it goes in like the soil has decided to cooperate. It is the smell — geosmin, the actinomycete note, the thing everybody means by earth — coming off a handful that used to smell of nothing.

Then a quieter one, in the fourth August, when it does not rain for five weeks and your field stays green a fortnight longer than the field over the fence. You do not have to say anything about it. Everybody driving past can see it, and about half of them will find a reason to stop.

And the best of them is the worms. You stop counting carbon for a moment and start counting worms, because they are countable and they are visibly delighted and they were not there before. There is a particular feeling in finding that something you did on purpose has been taken up by creatures with no opinion about your business case. It is the most direct evidence any producer ever gets that the thing they made is alive.

OPERATIONALIZE THIS

At the level of finance

The instrument: a transition bridge facility, issued by a co-operative that also owns the measurement.

Two revenue streams repay it, and the second one is deliberately small.

The mechanics, for a 400-hectare conversion.

Facility size. The computed transition hole plus 15 percent contingency: $810/ha × 1.15 = $931.50/ha, or $372,600 on 400 hectares. Drawn monthly against certified operating expense, not advanced in a lump.
Term. Three transition years plus repayment. At a 75 percent sweep of the realised premium, $236.50 × 0.75 = $177.37/ha/yr, plus $24.70/ha/yr of net carbon: $202.07/ha/yr, which repays $931.50 at 8 percent in 5.98 years — year six of operation, a nine-year term, with a hard review at year four when the first verified soil result lands.
Security. The premium contract and the growing crop. Never the land. A facility secured on land converts a good year for the lender into a catastrophe for the borrower, and it will not be signed by the farmers you most want.
Covenants. Rotation length, not practice prescription. Soil sampling access. No tillage of designated strata without notice. Assignment of the facility to any successor owner — the covenant runs with the land or the carbon is not real.
Coverage. $202.07 against interest of $931.50 × 8% = $74.52 is a 2.71× cover on interest alone in a mature year, which is what a credit committee wants to see before it reads anything else.

The balance-sheet treatment, which is the whole argument in one entry. The soil carbon stock is a land improvement with an appreciating carrying value. Most charts of accounts cannot express that, so use the one they can: capitalise the transition cost as a land improvement and set the depreciation schedule against the measured life of the improvement rather than a default. An asset whose productive capacity is rising should not be depreciating on a schedule that assumes it is falling. Book the issued carbon as inventory at lower of cost or market, with the 22 percent buffer as a contra-account, and disclose the reversal hazard. Talk to the auditors in month one; this is a conversation about useful economic life, which they have every year.

The counterparty. The co-operative, not the farm, and not a broker. It holds the stratification, the archive and the buffer, and it contracts the verifier independently. If no such entity exists in your area, building one is the project — and it is a smaller project than it sounds, because the first version is five strata, a corer, a laboratory account and a spreadsheet.

The first ninety days.

Day	Action	Artifact
1–15	Lengthen the rotation on one block. No counterparty required	The cropping plan
16–30	Stratify the holding by soil series; count the strata	The stratification map
31–45	Establish 50 paired georeferenced points per stratum; archive splits	The signed baseline
46–60	Compute the transition hole for your own cost structure	The working-capital note
61–75	Identify the aggregating entity; test the break-even area against it	Facility term sheet
76–90	Draw the first tranche against a signed premium offtake	The first drawdown

The number that decides it. One figure, on the front page:

   premium gain + net carbon income        236.50 + 24.70
  ----------------------------------  =  ------------------  =  28.0 %
       facility size per hectare               931.50

Against an 8 percent cost of capital that clears by 20 points. And the honest companion number, which belongs on the same page because leaving it off is how these deals lose their reputation: the carbon contributes $74.09 of the $810/ha hole — 9.1 percent. For carbon alone to fund a transition, CO₂ would have to reach $315.90 a tonne. It will not this decade.

So state it in the right register. The premium and the input bill finance the transition. The carbon finances the measurement — and the measurement is what makes the premium defensible, the covenant enforceable and the loan bankable. That is a smaller claim than the market makes and it is one you can sign.

APPRECIATIVE QUESTIONS

Twelve, for a room

Discovery — what is already working

Which field or block on this holding performs better than its soil map says it should, and what has been done to it that was never written down?
Think of the last dry year. Where did the crop hold on longest, and what do we know about that ground that we have never measured?
Which of our rotations is longest, who lengthened it, and what did they notice first?

Dream — what becomes possible

If our soil test were read at the bank with the same seriousness as our machinery schedule, what would we start measuring next month?
Imagine the co-operative's annual report carrying one soil figure on the cover. Which figure would we be proud to put there, and why that one?
If every acre under this entity were verified at four dollars a hectare, what would we do with the certainty that we cannot do now?

Design — what we build

What is the smallest area we could stratify and baseline this season, and who already owns a corer?
Which of our neighbours would we want inside the same stratification, and what would make them say yes?
Where would we put the buffer, and who should hold it so that everyone believes it is really there?

Destiny — how it holds

What would have to be true for this measurement grid to still be sampled in twenty years, by people none of us have met?
Who, in a bad year, would be tempted to till a designated stratum — and what would we want them to be offered instead of a clawback?
When this succeeds and the premium narrows, what will still be worth doing, and how would we know that now?

WORKS CITED

Batjes, N. H. (1996). "Total Carbon and Nitrogen in the Soils of the World." European Journal of Soil Science, 47(2), 151–163.

Sanderman, J., Hengl, T. and Fiske, G. J. (2017). "Soil Carbon Debt of 12,000 Years of Human Land Use." Proceedings of the National Academy of Sciences, 114(36), 9575–9580.

Lal, R. (2004). "Soil Carbon Sequestration Impacts on Global Climate Change and Food Security." Science, 304(5677), 1623–1627.

West, T. O. and Post, W. M. (2002). "Soil Organic Carbon Sequestration Rates by Tillage and Crop Rotation: A Global Data Analysis." Soil Science Society of America Journal, 66(6), 1930–1946.

Minasny, B., Malone, B. P., McBratney, A. B., et al. (2017). "Soil Carbon 4 per mille." Geoderma, 292, 59–86.

Poulton, P., Johnston, J., Macdonald, A., White, R. and Powlson, D. (2018). "Major Limitations to Achieving '4 per 1000' Increases in Soil Organic Carbon Stock in Temperate Regions: Evidence from Long-Term Experiments at Rothamsted Research, United Kingdom." Global Change Biology, 24(6), 2563–2584.

Powlson, D. S., Stirling, C. M., Jat, M. L., et al. (2014). "Limited Potential of No-Till Agriculture for Climate Change Mitigation." Nature Climate Change, 4, 678–683.

Six, J., Ogle, S. M., Breidt, F. J., Conant, R. T., Mosier, A. R. and Paustian, K. (2004). "The Potential to Mitigate Global Warming with No-Tillage Management Is Only Realized When Practised in the Long Term." Global Change Biology, 10(2), 155–160.

Davidson, E. A. and Ackerman, I. L. (1993). "Changes in Soil Carbon Inventories Following Cultivation of Previously Untilled Soils." Biogeochemistry, 20(3), 161–193.

Conant, R. T., Easter, M., Paustian, K., Swan, A. and Williams, S. (2007). "Impacts of Periodic Tillage on Soil C Stocks: A Synthesis." Soil and Tillage Research, 95(1–2), 1–10.

Smith, P., Soussana, J.-F., Angers, D., et al. (2020). "How to Measure, Report and Verify Soil Carbon Change to Realize the Potential of Soil Carbon Sequestration for Atmospheric Greenhouse Gas Removal." Global Change Biology, 26(1), 219–241.

Paustian, K., Lehmann, J., Ogle, S., Reay, D., Robertson, G. P. and Smith, P. (2016). "Climate-Smart Soils." Nature, 532, 49–57.

Mäder, P., Fließbach, A., Dubois, D., Gunst, L., Fried, P. and Niggli, U. (2002). "Soil Fertility and Biodiversity in Organic Farming." Science, 296(5573), 1694–1697.

Pimentel, D., Hepperly, P., Hanson, J., Douds, D. and Seidel, R. (2005). "Environmental, Energetic, and Economic Comparisons of Organic and Conventional Farming Systems." BioScience, 55(7), 573–582.

Rodale Institute (2011). The Farming Systems Trial: Celebrating 30 Years. Kutztown, Pennsylvania.

Seufert, V., Ramankutty, N. and Foley, J. A. (2012). "Comparing the Yields of Organic and Conventional Agriculture." Nature, 485, 229–232.

de Ponti, T., Rijk, B. and van Ittersum, M. K. (2012). "The Crop Yield Gap Between Organic and Conventional Agriculture." Agricultural Systems, 108, 1–9.

Ponisio, L. C., M'Gonigle, L. K., Mace, K. C., Palomino, J., de Valpine, P. and Kremen, C. (2015). "Diversification Practices Reduce Organic to Conventional Yield Gap." Proceedings of the Royal Society B, 282(1799), 20141396.

Crowder, D. W. and Reganold, J. P. (2015). "Financial Competitiveness of Organic Agriculture on a Global Scale." Proceedings of the National Academy of Sciences, 112(24), 7611–7616.

Reganold, J. P. and Wachter, J. M. (2016). "Organic Agriculture in the Twenty-First Century." Nature Plants, 2, 15221.

Davis, A. S., Hill, J. D., Chase, C. A., Johanns, A. M. and Liebman, M. (2012). "Increasing Cropping System Diversity Balances Productivity, Profitability and Environmental Health." PLoS ONE, 7(10), e47149.

Brown, G. (2018). Dirt to Soil: One Family's Journey into Regenerative Agriculture. Chelsea Green.

Ostrom, E. (1990). Governing the Commons: The Evolution of Institutions for Collective Action. Cambridge University Press.

Rothamsted Research. The Electronic Rothamsted Archive (e-RA): The Broadbalk Long-Term Experiment, established 1843. Harpenden.

United States Department of Agriculture, National Organic Program. 7 CFR §205.202(b) — Land Requirements.

Friedlingstein, P., et al. Global Carbon Budget (annual). Earth System Science Data.

Cooperrider, D. L. and Whitney, D. (2005). Appreciative Inquiry: A Positive Revolution in Change. Berrett-Koehler.

Note on figures. Every number in this chapter and its apparatus is computed in lib/verify/IV_03.py, where each line is labelled [src] for a published measurement, [asm] for a declared modelling assumption, or [calc] for a calculation from the two. Run python3 lib/verify.py IV.03. Where a figure is an assumption, the range it sits inside is printed beside it; where a fit depends on an assumed equilibrium, the sensitivity to that assumption is printed too.