Planetary Boundaries as a Balance Sheet

Volume VII — Planetary and Cosmic: What is the largest frame this fits inside?

Nine movements, one balance sheet.

THE PLATE

A woman wrapped in a cream shawl standing on a balcony above a forested valley, mist lying in the folds of the hills. — Plate VII.01The Statement and the Weather.A balance sheet is a claim about a world. The claim is only as good as the audit, and the audit is out of the window.

THE LETTER

This volume asks what the largest frame is, and the honest answer is that the largest frame is already a financial document. It is just a badly drafted one.

Somebody has drawn you the picture. A circle in the middle, nine wedges radiating out, some of them green and some of them pushed out into red. You have seen it on a conference slide and on the cover of a report, and you probably took away one sentence: six of the nine are broken. It is the most reproduced diagram in earth system science and it is, in its intent, entirely serious work by entirely serious people.

It is also a consolidated statement. It takes nine quantities measured in nine incommensurable units — parts per million, Dobson units, megatonnes a year, extinctions per million species-years — normalises each against a proposed limit, and prints them on one face. That is exactly what a balance sheet does, and it is why the picture works: the eye reads nine wedges the way it reads nine line items, as things of the same kind.

They are not things of the same kind, and the framework's own authors say so in the papers. Three of the nine rest on a demonstrated global threshold — a level in a well-mixed global system beyond which the system behaves differently. The other six are, in varying degrees, sums of local limits, and a sum of local limits has no tipping point at all; it has an average.

This chapter does not come to demote the framework. It comes to do to it what an accountant does to a good-faith set of management figures before they go to a board: separate what is measured from what is judged, put each into the statement where it belongs, and refuse to consolidate across the line. The framework deserves that, because after it survives the treatment there are three numbers left standing that nobody can argue with — and one of them is enough.

— The Editors

DISCOVERY

What is already working

Begin with the case the framework has already won, completely, from inside a transgression, and which almost nobody names when they show the diagram.

Stratospheric ozone is the boundary that was crossed and brought back. The control variable is the ozone column, measured in Dobson units. The pre-industrial column is 290 DU. The boundary is a 5 percent reduction from it — 275.5 DU — with the zone of increasing risk running to a 10 percent reduction at 261.0 DU. The Antarctic column went far past both. The Montreal Protocol was signed in 1987, has been ratified by every member state of the United Nations, and has been amended six times, each amendment tightening. The most recent WMO and UNEP assessments find the column recovering and project a return to 1980 values around mid-century. The global mean column now reads 284.6 DU, which is 9.1 DU of headroom — a margin of 3.3 percent above the line.

That is a complete cycle: a threshold identified, a limit set, an instrument built, compliance verified, the variable returned inside the line. It is the only one of the nine to have done it, and it happened because the ozone boundary has the three properties that make a boundary governable. The system is globally well-mixed, so one number genuinely describes it. The control variable is directly measured by satellite, continuously, by more than one agency. And the threshold behaviour is demonstrated in laboratory chemistry, not inferred from a model.

Then the boundary already being run as an account, at the scale where the damage happens. The nitrogen and phosphorus flows are the two boundaries most often shown as a single global bar, and the two places where a global bar is least useful. Around the Baltic, HELCOM has for years run precisely the opposite instrument: the Baltic Sea Action Plan sets maximum allowable inputs of nitrogen and phosphorus for the basin, then allocates national reduction requirements against them, with progress assessed against sub-basin targets. The unit of account is a sea, because a sea is where eutrophication occurs. In New Zealand, the Lake Taupō catchment has operated a capped, tradable nitrogen-discharge allowance regime since 2011 — a hard cap on a catchment, allowances that can be bought and retired, and a fund that has permanently retired a share of them.

Neither scheme has a global number in it anywhere, and that is not a gap. It is the design.

And then the balance sheet itself, already being drafted by finance ministries. In 2021 HM Treasury published The Economics of Biodiversity: The Dasgupta Review — a finance ministry commissioning, and then publishing, an argument that nature is a portfolio of assets and that the fundamental economic problem is a stock-and-flow one. In the same year the United Nations Statistical Commission adopted the System of Environmental-Economic Accounting: Ecosystem Accounting as an international statistical standard, which means ecosystem extent, condition and services now have an agreed accounting treatment sitting alongside the national accounts rather than beside them in a supplement. In 2023 the Taskforce on Nature-related Financial Disclosures published its recommendations, and the first letter of its core method is L, for Locate. Before a firm evaluates anything, it is asked where its assets and its dependencies physically are.

Read those three together and something appreciative and quite specific emerges. The finance world, working on its own problem, independently arrived at the fix the ecologists had been arguing for. Montoya, Donohue and Pimm's objection to a global biodiversity boundary is that biodiversity loss is local and its consequences are local. TNFD's first instruction is to find out where you are. Those are the same sentence, arrived at from opposite ends of the building, and nobody had to persuade anybody.

So the positive core of this chapter is not a hope. It is four working instruments — a protocol, a basin plan, a catchment cap, and a disclosure standard — each of which already does, at its own scale, the thing the consolidated diagram cannot.

THE ARITHMETIC

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

Here is the published table, in the form an accountant would want it: the control variable, the proposed boundary, the far edge of the published uncertainty band, and the most recent reading. Every value is quoted from Steffen et al. (2015) and Richardson et al. (2023). Every ratio after it is computed in lib/verify.py.

  control variable                     boundary     zone edge     reading
  -------------------------------------------------------------------------
  CO2 concentration                    350 ppm       450 ppm      417 ppm
  radiative forcing                    1.0 W/m2      1.5 W/m2     2.91 W/m2
  ozone column                         275.5 DU      261.0 DU     284.6 DU
  aragonite saturation                 2.75          2.408        2.8
  forest cover remaining               75 %          54 %         60 %
  blue water deviation                 10.2 %        —            18.2 %
  green water deviation                11.1 %        —            15.8 %
  nitrogen fixed                       62 Mt N/yr    82 Mt N/yr   190 Mt N/yr
  phosphorus, global                   11 Mt P/yr    100 Mt P/yr  22.6 Mt P/yr
  phosphorus, regional                 6.2 Mt P/yr   11.2 Mt P/yr 17.5 Mt P/yr
  aerosol, interhemispheric            0.1 AOD       0.25 AOD     0.076 AOD
  extinction rate                      10 E/MSY      100 E/MSY    >100 E/MSY
  biosphere intactness                 90 %          30 %         79 %
  novel entities                       0 % untested  —            unquantified

First, the ratios, because they are what the diagram is drawing. Nitrogen is at 3.065 times its boundary. Radiative forcing is at 2.91 times. Regional phosphorus is at 2.823 times, global phosphorus at 2.055 times, blue water at 1.784 times, green water at 1.423 times. Carbon dioxide is at 1.191 times. Forest cover sits at 0.800 of the required fraction — 20.0 percent below the line. Intactness sits at 0.878 — 12.2 percent below. The extinction rate is published as a floor, and against a boundary of 10 E/MSY that floor is more than 10.0 times.

Three of the nine are inside. Ozone has 3.3 percent of headroom. Ocean acidification has 1.8 percent, at a reading of 2.8 against a boundary of 2.75. Aerosols have 24.0 percent, at 0.076 against 0.1.

Second, the criticism, taken at full strength, because it is good criticism.

The economic objection is the oldest and it has two halves. The first is that a boundary is a constraint with no cost function attached. William Nordhaus's tradition — To Slow or Not to Slow in 1991, through to The Climate Casino — insists that a limit without a damage function and a discount rate cannot be allocated against; it tells you a place not to go, and says nothing at all about which of the ten thousand available paths short of it is the one to take. The second half is Thomas Sterner's, and it is subtler and more damaging to the framework's critics than to the framework. Sterner and Persson's An Even Sterner Review shows that when environmental goods become scarce their relative price rises, and that this single correction does much of the work a hard threshold is reaching for — without a cliff, and inside the machinery economics already has. Ted Nordhaus and Michael Shellenberger's Breakthrough Institute review presses the practical version: if a boundary's number can be moved by a literature review, it is not a planetary limit, it is a research finding with a policy costume on.

The ecological objection is sharper still. Montoya, Donohue and Pimm — Implausible Science, Pernicious Policies — argue that the biodiversity boundary is not merely uncertain but ill-posed: there is no evidence for a global threshold in biodiversity loss, the proposed control variables are proxies for proxies, and a global number licenses the trade that ecology most needs to forbid, in which loss here is offset by intactness there. Brook, Ellis and colleagues, asking directly whether the terrestrial biosphere has planetary tipping points, conclude that it very largely does not: terrestrial change is a mosaic of local transitions that sum, and a sum does not tip.

And Blomqvist and colleagues supply the aggregation argument in its cleanest form. It is worth doing their arithmetic, because it is one division.

  global nitrogen fixed                 190 Mt N/yr
  cropland                            1,600 Mha
  ------------------------------------------------
  spread evenly                      118.75 kg N/ha/yr
  concentrated on a fifth of it      593.75 kg N/ha/yr        5.0 x

Both of those worlds report the same 190. One of them has anoxic estuaries, nitrate in the drinking water, and a continent that cannot buy fertiliser. The other has none of those things. The global total is invariant to the distribution, and the distribution is the entire damage function. A number that cannot distinguish between those two worlds is not a boundary. It is a sum.

The same error recurs one aggregate later. The Dasgupta Review cites the familiar figure that humanity uses about 1.6 Earths. Blomqvist and colleagues took the components apart and found that five of the six come out at or below unity, so the whole of the overshoot is carried by the carbon component — a hypothetical forest area computed from emissions, not an observed one. The carbon term is therefore 37.5 percent of the aggregate and 100 percent of the deficit. The headline is the climate problem, restated in hectares, with five measured components riding along contributing nothing to the result.

Third — and this is the movement that does the real work — what does not work.

Several of these boundaries have no defensible global number, and printing them on one face with the ones that do is a false equivalence between a measured threshold and an expert judgement. This is not an argument from suspicion. It is measurable, and there is a clean way to measure it: take the width of each published uncertainty band and divide it by the boundary value it hangs off. That is a like-for-like statement of how much each boundary's authors were willing to commit to, in the units of the boundary itself.

  ozone                                        5.3 %
  aragonite saturation                        12.4 %
  CO2 concentration                           28.6 %
  forest cover                                28.0 %
  nitrogen                                    32.3 %
  radiative forcing                           50.0 %
  freshwater, the 2015 boundary               50.0 %
  phosphorus, regional                        80.6 %
  aerosol, the 2015 regional AOD             100.0 %
  aerosol, the 2023 interhemispheric         150.0 %
  biosphere intactness                        66.7 %
  phosphorus, global                         809.1 %
  extinction rate                            900.0 %

The four climate-and-chemistry bands average 24.1 percent of their own boundary, with a median of 20.5 percent. The two biosphere-integrity bands average 483.3 percent, with a median of the same. That is a ratio of 20.1 times on the means and 23.6 times on the medians — about twenty times either way, which is the point of printing both. A band that is nine times wider than the boundary it surrounds is not an uncertainty estimate. It is a published statement that nobody knows, and it is to the authors' credit that they printed it rather than hiding it. It is to the diagram's discredit that the wedge is drawn the same width as the others.

Here is the second honest negative, and it is the harder one. Novel entities has no measured current value at all. The 2023 control variable is the percentage of synthetic chemicals released to the environment without adequate safety assessment; the boundary is zero; and the transgression is declared on the definitional ground that the percentage is certainly not zero. That may well be correct. It is not a measurement, and a wedge in the red on a diagram cannot say which it is.

And now the cut.

Six of the nine have had their boundary value or their control variable revised since 2009 — 66.7 percent of the framework. Nitrogen's boundary moved from 35 to 62 Mt N/yr, a revision of 77.1 percent, on the strength of de Vries and colleagues' 2013 reassessment; the number changed by more than three quarters in six years and the nitrogen cycle did not. Land-system change was a cropland fraction and became a forest fraction. Freshwater was a cubic-kilometre flow and became a percentage of land area with a hydrological deviation. Biosphere integrity split into two components. Phosphorus split into two. Aerosols changed control variable twice. Novel entities was quantified for the first time in 2022.

The three that have never been revised are climate, ozone and ocean acidification — which are exactly, and without a single exception, the three that rest on a demonstrated global threshold in a well-mixed system. That is not a coincidence and it is not a judgement of ours. It is an identity you can recompute from the framework's own published tables in under a minute, and it means the diagram's most alarming arms are drawn overwhelmingly from the six that keep being redrawn.

Which is the finding, stated as it should be stated: of the six transgressions, one is Grade A evidence and three are Grade C or D — 16.7 percent against 50.0 percent. And that one Grade A transgression is climate, at 1.191 times on concentration and 2.91 times on forcing, measured continuously, with the threshold behaviour modelled to within an inch of its life.

One is enough. The framework does not need the other five to be load-bearing. It needs them to be honestly graded, because a statement that mixes a covenant breach with an unaudited estimate is a statement no auditor will sign — and an unsigned statement moves no capital at all.

DREAM

What becomes ordinary

In the version of this that has already happened, nobody shows the nine-wedge diagram to a board, because the board has something better: a four-part statement that says what it knows, in the register in which it knows it.

The covenant schedule is one page and it carries three lines. Carbon dioxide concentration, ozone column, aragonite saturation — each with its boundary, its current reading, its headroom, its trend, and the clause that says what happens on breach. It reads exactly like the covenant schedule in a loan agreement, because it is one, and the people who read loan agreements read it without translation. Nobody finds this strange. It is simply how a level test is presented.

The asset register is longer and it is organised by place. Forest extent and condition by biome, root-zone soil moisture by basin, land condition by holding. Gross, accumulated impairment, net. A treasurer looking at it can see which specific hectares are appreciating and which are being drawn down, in the same way and on the same page as the plant register. When a site is impaired, the impairment is booked against that site and does not vanish into an average.

The flow account runs monthly and reports throughput at the place the throughput lands. Nitrogen and phosphorus by catchment, against that catchment's own allowable input. There is no global nitrogen figure anywhere in the pack, and nobody misses it, because no decision anyone in the building can take is a decision about global nitrogen. Every decision they can take is about a field, a plant, a supplier or a basin.

And there is a watch list. It is short, it is explicitly not consolidated into any total, and every line on it carries its evidence grade in the same typeface as the number. Novel entities is on it. Biosphere intactness is on it, with its published band printed beside it so that a reader can see at once that the band is wider than the boundary. Nothing on the watch list is ignored and nothing on it is added to anything.

What that pack produces in a meeting is the thing the diagram could never produce: an argument that cannot be dismissed by attacking its weakest line, because the weakest line has already been labelled by the person presenting it. The covenant schedule stands whatever anybody thinks of the watch list. Somebody who wants to fight about extinction rates can fight about extinction rates all afternoon, and the three covenants are still there at the end of it.

DESIGN

The four statements, built

The structure follows from the taxonomy, and the taxonomy is the whole design. Every boundary is one of three things, and the three behave differently enough that putting them on one face is a category error with a price.

A covenant is a level test with a step loss function. You are inside it or you are not. At 2.9 times you are compliant and at 3.1 times the facility is repayable on demand; there is no version of the covenant in which being slightly over costs slightly more. Climate, ozone and ocean acidification behave this way because the underlying system is well-mixed and has a demonstrated threshold. Report them as covenants: boundary, reading, headroom, trend, breach clause.

A stock is an asset with a carrying value and an impairment test. Depletion is real, cumulative and proportional; there is no cliff, and the right response to a 20.0 percent shortfall in forest cover is not an alarm but an impairment charge against the specific assets that are short. Land-system change and freshwater are stocks. Report them on a register, by location and condition, with gross, impairment and net — the treatment IAS 16 already gives to property, plant and equipment, and IAS 41 to living assets.

A flow is a rate through a period, and its damage is local. Nitrogen, phosphorus and aerosols are flows. The global sum of a flow whose damage is threshold-shaped and local is the least informative number available about it, as the 118.75 against 593.75 kilograms per hectare demonstrates. Report them in a flow account, by catchment or airshed, against that place's own allowable input.

And the fourth statement is not a class of boundary. It is the disclosure that holds everything not yet measurable enough to recognise. Biosphere integrity and novel entities go here. The treatment already exists: IAS 37 distinguishes a provision, which is recognised because it can be measured reliably, from a contingent liability, which is disclosed precisely because it cannot. A band of 900.0 percent fails the measurement-reliability test by a distance. Disclosure is not demotion — a contingent liability in a set of accounts is a live, legally consequential statement, and it is read by exactly the people who need to read it.

The governance follows in four moves.

Grade before you consolidate. Every line carries A, B, C or D on the face of the statement, and the grade is set by three tests: is the system well-mixed, is the control variable directly measured, is the threshold demonstrated rather than inferred.
Never sum across grades. This is the one rule the diagram breaks and the one that costs the most. A total that spans a covenant and an expert judgement is not a total.
Downscale before you allocate. A global boundary that has not been divided into places is not yet an instrument. Häyhä and colleagues' work on fair shares is the honest route in, and the honest thing to say about it is that the allocation rule is an ethical choice and should be labelled as one.
Republish the grade, not just the number. When a boundary is revised — and six of the nine have been — the revision belongs in the same line as the value, the way a change in accounting estimate is disclosed rather than quietly applied.

DESTINY

How it holds when nobody is pushing

The failure modes are specific and they are all failures of the grading, not of the science.

Grade inflation. Everything drifts toward A, because A is the grade that gets a meeting. The defence is that the three tests are written down and a named person applies them, and that the grade is auditable against the framework's own published uncertainty bands — which is why the band width is computed rather than asserted.

Grade deflation. The mirror failure, and in practice more common: everything is pushed to D so that nothing has to happen. The defence is the same arithmetic run the other way. Climate's band is 28.6 percent of its boundary, ozone's 5.3 percent, and no honest reading of those puts either in the same class as a band of 900.0 percent. The grades are evidence-bound in both directions or they are worthless in both.

Re-aggregation. Somebody will want one number for the cover. They always do, and they are not being foolish — a cover needs a number. Give them the covenant schedule's worst headroom, which is a real number about a real threshold, and refuse the composite. A composite index across four evidence grades is the diagram again, with a decimal point.

Capture of the grading body. The standard-setter becomes the thing to lobby. Biermann and Kim's appraisal of the framework's governance is the honest map of this terrain, and the structural defences are the ordinary ones: publish the tests, publish the dissent, keep the underlying readings open, and make the revision history a required disclosure.

And the honest thing to say about the whole method: it is slower. A four-part statement takes longer to build and longer to read than one circle, and it will never spread the way the circle spread. That is a real cost and it is worth paying in one place only — the place where money moves. The circle is the better poster. The statement is the better instrument. They are not competing for the same job.

DELIGHT

What it feels like

There is a particular relief in a document that says what it does not know.

Anyone who has sat through a presentation of nine numbers of nine different qualities, all drawn at the same weight, knows the specific fatigue of it — the low-grade suspicion that you are being asked to accept the weakest line in order to get the strongest, and the resentment that follows. A statement that grades itself takes that away entirely. You can put your weight on the A lines and let the D lines be exactly what they are.

And then the pleasure that is better than relief: the moment the covenant schedule goes in front of somebody who has spent twenty years reading covenant schedules, and they do not need any of it explained. They read the headroom, they read the trend, they ask one question about the breach clause, and the conversation is already technical. No translation, no persuasion, no prior agreement about what kind of world this is.

That is what it feels like when a frame the size of a planet finally arrives in a form a person can act on before lunch. It is not smaller. It is just no longer foreign.

OPERATIONALIZE THIS

At the level of finance

The instrument: a nature-linked facility with a segmented covenant schedule.

The market's existing answer is the sustainability-linked loan, and it is worth being precise about why it does not do this job. The market-norm margin ratchet is 25 basis points. On a drawn facility of $400,000,000 that is $1,000,000 a year, against a covenant whose breach makes $400,000,000 repayable. The ratchet is therefore 0.25 percent of the consequence it claims to price — a factor of 400 times out. A ratchet is not a price on a threshold. It is a rounding error wearing one, and the reason so many sustainability-linked loans change no behaviour is visible in that one division.

So the structure segments, and prices each segment the way its physics behaves.

The mechanics.

Segment A — a true covenant. One to three Grade A metrics, level-tested, quarterly, with a defined breach clause and an equity cure. Priced as a covenant is priced: it moves the cost of capital and the facility's availability, not the margin by a quarter of a point. This is the only segment with a step.
Segment B — an impairment test on a register. Grade B stocks, by location, with an agreed condition metric and an annual assurance. The consequence is a borrowing-base adjustment: the register is collateral, and collateral that is impaired reduces availability proportionally. Proportional consequence for a proportional loss.
Segment C — a throughput covenant at the place of effect. Grade C flows, by catchment or airshed, tested against that place's own allowable input. Never a group total.
Segment D — disclosure only. Grade D items are reported and not tested. A covenant on a number with a band of 900.0 percent is unenforceable and both parties know it, so writing one buys nothing and costs credibility.

The balance sheet treatment. Restoration spend that creates or improves a long-lived asset is capitalised under IAS 16 and depreciated over the asset's regenerated life; living assets go to IAS 41 at fair value where a market exists. The Segment A covenant is a loan term, disclosed in the borrowings note with the rest of the covenant schedule — not a separate sustainability disclosure, because putting it there is what makes it optional. Grade D items are IAS 37 contingent disclosure. And the whole register is built once, under the TNFD's Locate step, and then reused: the location data is the expensive part and it is a fixed cost paid once.

The counterparty. A relationship lender or a development finance institution, not a syndicate, for the first facility. Segment A needs a lender who will actually negotiate a breach clause rather than paper one, and that is a two-party conversation. Take it to a syndicate on the second facility, when the schedule has a compliance history.

The number that decides it.

   avoided impairment  +  margin benefit
   ---------------------------------------  >  1
     verification  +  restoration spend

Worked, at the scale most mid-cap borrowers will recognise: avoided impairment of $2,400,000 a year and a margin benefit of $1,000,000 against verification and location data at $340,000 and a restoration programme at $1,250,000. That returns 213.8 percent — a net $1,810,000 a year — and it clears before a single tonne of anything is restored, because most of what the spend buys is the documented right not to impair assets that were never impaired in the first place. That is the sentence to put on the front page, and it is an accounting argument rather than an environmental one.

The honest commercial negative. Segment A is the only segment that will be genuinely difficult to negotiate, because a real breach clause on a real threshold is a real risk transfer and lenders price real risk transfers. Expect the first lender to want Segment A as a reporting covenant rather than a financial one. That is a reasonable place to start and an unreasonable place to stop; the term to fight for is the second-facility step-up, where Segment A becomes financial once a compliance history exists.

The first ninety days.

Day	Action	Artifact
1–15	Grade every metric already in your reporting against the three tests	The graded register
16–30	Locate: every material asset and dependency, by site	The location layer
31–45	Build the covenant schedule for Grade A only	The one-page covenant schedule
46–60	Build the asset register and the flow account by place	Two statements
61–75	Agree the assurance scope and the verifier	Signed assurance scope
76–90	Take the four statements to one lender	The segmented term sheet

APPRECIATIVE QUESTIONS

Twelve, for a room

Discovery — what is already working

The Montreal Protocol took a transgressed boundary and brought it back inside the line. What did it have — in measurement, in instrument, in politics — that made that possible, and where else do we already have all three?
Where in our own reporting do we already report a genuine level test correctly, and who insisted on it?
Which of the places we operate in already has a basin, catchment or airshed plan with a real allocation in it — and what would it take to report against theirs rather than inventing ours?

Dream — what becomes possible

If our board pack carried a one-page covenant schedule with three lines on it, what would the first meeting about it actually be about?
Imagine our asset register with condition alongside carrying value, by site. Which site would surprise us most, and in which direction?
If every number we published carried its evidence grade in the same typeface as the number, what would change about which numbers we chose to publish?

Design — what we build

Which of our metrics is a covenant, which is a stock and which is a flow — and where have we been managing one as though it were another?
Where are we currently summing across places, and what would we see if we stopped?
What is the widest uncertainty band we currently publish, and what would it take to say so on the same line as the number?

Destiny — how it holds

What would have to be true for this grading to survive the first person who wants a single number for the cover?
Who outside this organisation would we want auditing our grades, and what would we have to publish for them to be able to?
What is the first sign we would see that grades were drifting — in either direction — and who would notice it first?

WORKS CITED

Rockström, J., Steffen, W., Noone, K., et al. (2009). "A safe operating space for humanity." Nature, 461, 472–475.

Rockström, J., Steffen, W., Noone, K., et al. (2009). "Planetary boundaries: exploring the safe operating space for humanity." Ecology and Society, 14(2), 32.

Steffen, W., Richardson, K., Rockström, J., et al. (2015). "Planetary boundaries: Guiding human development on a changing planet." Science, 347(6223), 1259855.

Richardson, K., Steffen, W., Lucht, W., et al. (2023). "Earth beyond six of nine planetary boundaries." Science Advances, 9(37), eadh2458.

Persson, L., Carney Almroth, B. M., Collins, C. D., et al. (2022). "Outside the Safe Operating Space of the Planetary Boundary for Novel Entities." Environmental Science & Technology, 56(3), 1510–1521.

Wang-Erlandsson, L., Tobian, A., van der Ent, R. J., et al. (2022). "A planetary boundary for green water." Nature Reviews Earth & Environment, 3, 380–392.

de Vries, W., Kros, J., Kroeze, C. and Seitzinger, S. P. (2013). "Assessing planetary and regional nitrogen boundaries related to food security and adverse environmental impacts." Current Opinion in Environmental Sustainability, 5(3–4), 392–402.

Carpenter, S. R. and Bennett, E. M. (2011). "Reconsideration of the planetary boundary for phosphorus." Environmental Research Letters, 6, 014009.

Nordhaus, W. D. (1991). "To Slow or Not to Slow: The Economics of the Greenhouse Effect." The Economic Journal, 101(407), 920–937.

Nordhaus, W. D. (2013). The Climate Casino: Risk, Uncertainty, and Economics for a Warming World. Yale University Press.

Sterner, T. and Persson, U. M. (2008). "An Even Sterner Review: Introducing Relative Prices into the Discounting Debate." Review of Environmental Economics and Policy, 2(1), 61–76.

Nordhaus, T., Shellenberger, M. and Blomqvist, L. (2012). The Planetary Boundaries Hypothesis: A Review of the Evidence. Breakthrough Institute, Oakland.

Blomqvist, L., Brook, B. W., Ellis, E. C., Kareiva, P. M., Nordhaus, T. and Shellenberger, M. (2013). "Does the Shoe Fit? Real versus Imagined Ecological Footprints." PLoS Biology, 11(11), e1001700.

Brook, B. W., Ellis, E. C., Perring, M. P., Mackay, A. W. and Blomqvist, L. (2013). "Does the terrestrial biosphere have planetary tipping points?" Trends in Ecology & Evolution, 28(7), 396–401.

Montoya, J. M., Donohue, I. and Pimm, S. L. (2018). "Planetary Boundaries for Biodiversity: Implausible Science, Pernicious Policies." Trends in Ecology & Evolution, 33(2), 71–73.

Rockström, J., Richardson, K., Steffen, W. and Mace, G. (2018). "Planetary Boundaries: Separating Fact from Fiction. A Response to Montoya et al." Trends in Ecology & Evolution, 33(4), 233–234.

Mace, G. M., Reyers, B., Alkemade, R., et al. (2014). "Approaches to defining a planetary boundary for biodiversity." Global Environmental Change, 28, 289–297.

Lewis, S. L. (2012). "We must set planetary boundaries wisely." Nature, 485, 417.

Schlesinger, W. H. (2009). "Planetary boundaries: Thresholds risk prolonged degradation." Nature Reports Climate Change, 3, 112–113.

Biermann, F. and Kim, R. E. (2020). "The Boundaries of the Planetary Boundary Framework: A Critical Appraisal of Approaches to Define a 'Safe Operating Space' for Humanity." Annual Review of Environment and Resources, 45, 497–521.

Häyhä, T., Lucas, P. L., van Vuuren, D. P., Cornell, S. E. and Hoff, H. (2016). "From Planetary Boundaries to national fair shares of the global safe operating space." Global Environmental Change, 40, 60–72.

Lenton, T. M., Held, H., Kriegler, E., et al. (2008). "Tipping elements in the Earth's climate system." Proceedings of the National Academy of Sciences, 105(6), 1786–1793.

Armstrong McKay, D. I., Staal, A., Abrams, J. F., et al. (2022). "Exceeding 1.5°C global warming could trigger multiple climate tipping points." Science, 377(6611), eabn7950.

Dasgupta, P. (2021). The Economics of Biodiversity: The Dasgupta Review. HM Treasury, London.

Raworth, K. (2017). Doughnut Economics: Seven Ways to Think Like a 21st-Century Economist. Random House Business.

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

United Nations et al. (2021). System of Environmental-Economic Accounting — Ecosystem Accounting (SEEA EA). Adopted by the UN Statistical Commission, March 2021.

Taskforce on Nature-related Financial Disclosures (2023). Recommendations of the Taskforce on Nature-related Financial Disclosures. TNFD, September 2023.

World Meteorological Organization and UN Environment Programme. Scientific Assessment of Ozone Depletion. Successive assessments, most recently 2022.

United Nations Environment Programme (1987, and subsequent amendments). Montreal Protocol on Substances that Deplete the Ozone Layer.

HELCOM. Baltic Sea Action Plan (2007; updated 2021), including the nutrient reduction scheme and country-allocated maximum allowable inputs.

Waikato Regional Council. Lake Taupō catchment nitrogen management regime (operative 2011), and the Lake Taupō Protection Trust.

IFRS Foundation. IAS 16 Property, Plant and Equipment · IAS 37 Provisions, Contingent Liabilities and Contingent Assets · IAS 41 Agriculture.

Note on figures. Every boundary value, uncertainty band and 2023 reading in this chapter is quoted from Steffen et al. (2015) and Richardson et al. (2023) and is labelled as an input in lib/verify/VII_01.py. Every ratio, band width, count and share is computed in that module and reproducible there. The facility, register, ratchet and restoration sums in Operationalize This are illustrative and are labelled as such in the same module.