At the Commission for the Conservation of Antarctic Marine Living Resources, which had 27 members with a vote as of 2026-09-18, Article XII of the convention decides matters of substance by consensus and treats the question of whether a matter is one of substance as itself a matter of substance — so a conservation measure that has to be agreed again each year needs the same agreement to renew that a new one needs to adopt

What is happening?

One decision rule governs an ocean area of 35,716,100 square kilometres, about 10 percent of the world ocean, and that rule has two settings rather than one.

The Convention on the Conservation of Antarctic Marine Living Resources was signed on 1980-05-20 and entered into force on 1982-04-07. The body it created had 27 members with a vote and 10 acceding states without one as of 2026-09-18. Article XII sets out how that body decides. Decisions on matters of substance are taken by consensus. Every other matter is decided by a simple majority of the members present and voting. And the article adds a third sentence that folds the first two together: the question of whether a matter is one of substance shall itself be treated as a matter of substance.

On the protection side, two areas exist and four proposals do not. A protected area over the southern shelf of the South Orkney Islands was established in 2009 at 94,000 square kilometres, the first such area on the high seas. A second, over the Ross Sea region, was agreed on 2016-10-28, entered into force on 2017-12-01, covers more than 1.5 million square kilometres with about 1.1 million of them closed to commercial fishing, and runs to 2052. Four further proposals stand behind them. One covers East Antarctica, tabled in 2012 at about 1 million square kilometres. One covers the Weddell Sea, tabled by Germany in 2018 at more than 2 million square kilometres and split into two phases in 2019 when Norway joined as a co-proposer. One covers the Antarctic Peninsula, put to the Scientific Committee in 2017 and to the Commission in 2018. At the forty-fourth meeting, held 2025-10-20 to 2025-11-01, none of the four obtained consensus, and the peninsula proposal was reported as having failed for a seventh consecutive year.

On the fishing side the same rule ran in the other direction. A conservation measure distributing the krill catch across the four subareas of Area 48 had been in force for about fifteen years, and secondary accounts of it describe a requirement that every member agree each year for it to continue. That agreement was not obtained and the measure expired on 2024-11-30. In the season that followed, the catch limit for Area 48 as a whole, 620,000 tonnes, was reached for the first time and the fishery closed on 2025-08-01. Subarea 48.1, along the Antarctic Peninsula, took 357,961 tonnes and 57.5 percent of the season total. In the next season the same limit was reached a second consecutive time and the fishery closed roughly three months before the season would otherwise have ended, on a statement issued 2026-08-14.

Whose problem is this?

The body that would have to adopt the measure is composed of the same governments that have not adopted it, and the body that would have to renew the lapsed one is the same body again. Nothing outside it can perform either act in its place.

Where does this problem end?

The boundary matters here because the rule and the outcome are two different things, and a member that withholds agreement is using a procedure the membership wrote for itself. A reader who thinks the four proposals should be adopted and a reader who thinks they should not can both agree on the dates and the counts set out below.

What is the state now, and what should it be?

Now

What it should be

Two of the standards here belong to the Commission rather than to anybody outside it, and both are reported through secondary sources rather than read in the original. The first is the trigger level of 620,000 tonnes for Area 48, which the Commission set in 1991 and which is about 11 percent of the precautionary catch limit of 5.61 million tonnes that its own assessment produced. The second is the lapsed measure itself: for about fifteen years the Commission required that the catch be spread across four subareas rather than taken from one, which is a target state it wrote down, applied, and then stopped applying without deciding to stop.

Needs a new measurementthe coverage target: no source opened here carries a figure, a share or a date that the Commission has set itself for how much of the Convention Area should be under protection, or by when. A coalition of environmental organisations states that the four pending proposals would together cover 26 percent of the Southern Ocean, and that is a description of the proposals rather than a standard the Commission adopted. This document does not substitute the one for the other.

How big is it?

The quantities the opened sources carry are areas, tonnages and counts of member governments. None of them is a count of persons.

Area. The Convention Area is 35,716,100 square kilometres, about 10 percent of the world ocean. Two protected areas stand inside it: 94,000 square kilometres over the southern shelf of the South Orkney Islands from 2009, and more than 1.5 million square kilometres over the Ross Sea region from 2017, of which about 1.1 million are closed to commercial fishing. The four pending proposals are described at about 1 million square kilometres for East Antarctica and more than 2 million for the Weddell Sea. No source opened here gives an area for the Antarctic Peninsula proposal.

Catch. The precautionary catch limit for Area 48 is 5.61 million tonnes and the trigger level that actually binds is 620,000 tonnes, about 11 percent of it. In the 2025 season that trigger was reached for the first time and the fishery closed on 2025-08-01, with 357,961 tonnes and 57.5 percent of the total taken in subarea 48.1 and 259,640 tonnes and 41.7 percent in subarea 48.2. A magazine feature covering the same season records subarea 48.1 above 57 percent and 118 percent higher than the previous season. A statement of 2026-08-14 records the trigger reached a second consecutive time, the total more than 80 percent above the level of two years earlier, and the peninsula subarea up 117 percent. The two growth figures are attached to different seasons in their own sources and no opened source states whether they rest on the same comparison.

No affected population is derivable, and the reason is structural. The units these sources count are square kilometres of ocean, tonnes of catch and member governments. The nearest thing to a count of living things is a report that chinstrap penguin numbers fell by about 30 percent in recent years and that Antarctic fur seal numbers and pup survival are falling, and that report is a commentary in a news outlet rather than a population study. No baseline year, no survey area and no method were established for it in this round, so it is carried as a description rather than as a quantity. Converting the area of the Convention Area into people would mean assigning a population to an ocean that nobody lives in, which no opened source does and which would measure something other than this problem.

Under what conditions does it arise?

1. Doing nothing is the outcome that requires no agreement. Adopting a protected area needs every member present to accept the same decision at the same meeting. Not adopting it needs nothing at all, because it is what the meeting produces when the first decision does not carry. A proposal can therefore be tabled in 2012 and still be pending in 2025 without any member ever having to defend a decision, because no decision was taken.

2. Renewal sits on the same side of the rule as adoption. A measure that has to be agreed again each year is a measure that has to clear the consensus threshold each year. The direction of the failure then reverses: where a new area fails to appear, an existing rule disappears. The members who wanted the distribution rule kept had no procedure available to them once agreement was not reached, in the same way that the members who want the four areas adopted have none.

3. The classifying question is classified by the rule it would classify. Article XII decides matters of substance by consensus and everything else by simple majority, and then treats the question of which kind a given matter is as itself a matter of substance. Nothing in the procedure turns a repeated failure to agree into a decision of any other kind, so the measure that lapses and the measure that is never adopted arrive at the same place.

4. The cost of the lapse landed in one place rather than four. The purpose of spreading the catch across four subareas was to prevent it concentrating on any one of them. With the basin-wide limit alone in force, more than half the season total was taken from the subarea along the Antarctic Peninsula in the first season after the lapse, and the season ended early because a single number was reached rather than four.

5. Withholding agreement carries no procedural consequence, because the rule attaches none. Nothing in Article XII records a withheld agreement, escalates it, or refers the question to any other body. Doing so is an exercise of the rule as written rather than a departure from it, and that is why it can be repeated at meeting after meeting without limit.

What has been tried?

Two directions have been pursued at once and neither has closed the gap. One is to add protection, which has produced four proposals that have been tabled and retabled without a decision against them and without a decision for them. The other is to keep the protection already in place, which has produced two single-year extensions and then an expiry date that arrived with nothing behind it.

What was found?

Why is it still unsolved?

Coordination failure — the rule that has to produce a new measure is the same rule that has to produce the renewal of an old one, and neither direction has a fallback when agreement is not reached.

The symmetry is the whole of it. A protected area is adopted only if no member withholds agreement, and a measure that has to be agreed again each year continues only if no member withholds agreement. Those two sentences describe one requirement, and the requirement is silent about which side of it the status quo sits on. So the same procedure that has left four proposals pending since 2012, 2018 and 2018 also let a rule that had governed the krill catch for about fifteen years reach its expiry date and stop. In the first case failing to agree preserves the sea as it was. In the second case failing to agree changes it. Nobody chose the difference and no member had to argue for it, because in both cases what happened is what happens when a meeting ends without a decision.

The second part is that Article XII closes the exit it appears to leave open. The article provides a simple majority for matters that are not matters of substance, which reads like a route around a deadlock until the third sentence is read: whether a matter is one of substance is itself a matter of substance. A member that withholds agreement on the underlying question can withhold it on the classifying question by the same act. No source opened here identifies any organ with authority to decide otherwise, which means the procedural question of how to break a deadlock is answered by the deadlock.

The third part is that the failure produces no event. A proposal that is not adopted is simply tabled again the following year, and the record shows a meeting that met. A measure that expires does so on a date set years earlier, so nothing is announced on the day it stops. The catch concentration that followed in subarea 48.1 was visible only a season later, in a tonnage figure published by an industry association, and the early closure that marked it looked from the outside like a fishery working as designed. A gap that produces no headline produces no deadline either.

What observation would mean it is solved?

Candidates — (a) the Commission adopts at least one of the four pending protected areas, with a stated entry into force (b) a spatial distribution rule for the krill catch is back in force, whether the lapsed measure renewed or a replacement adopted (c) the Commission acquires a procedure that produces a decision when consensus is not reached, so that neither adoption nor renewal depends on unanimity alone.

(a) alone is weaker than it looks. Adopting one of four leaves three, and the Ross Sea area shows that a single adoption says nothing about the next one, having taken four years while the East Antarctic proposal tabled in the same year is still pending. An adopted area can also carry boundaries, exemptions or a review date that leave the fishing pattern where it was, and it can expire in its turn under the same rule that let the distribution measure expire.

(b) alone can move for the wrong reason. The concentration in subarea 48.1 would also fall if the krill moved, if the fleet went elsewhere, or if a season ran short for reasons of weather or price. A distribution rule back in force would be the mechanism working, but a fall in the concentration figure on its own would not distinguish the mechanism from the season it came from.

(c) is the strongest and the least likely. Changing how the Commission decides is itself a matter of substance under Article XII, so the procedure would have to be adopted by the procedure it is meant to replace. That is not an argument that it cannot happen, because the members wrote the article and can write another. It is the reason the first two observations can both come true while the condition that produced the problem stays exactly as it is, which is why the three have to be read together.

What is it connected to?

Fills with researchconsensus requirements at other regional fisheries management organisations and at other bodies of the Antarctic Treaty system, the treatment of lapsing conservation measures in other treaties that require periodic renewal, the relationship between the protected area proposals and the agreement on biodiversity beyond national jurisdiction, and the dependence of krill predator populations on the distribution of the catch rather than its total. Relation type and evidence grade were not established in this round.

What these sources do not say

• Which members withheld agreement. No source opened here that was published by the Commission records how any member acted on any of the four proposals or on the renewal of the distribution measure, and none records a tally of the meeting.
• When the distribution measure was adopted. Secondary accounts describe it as having been in force for about fifteen years, which places its adoption around 2009 by subtraction from the expiry date, and no opened source names the year or the measure text. The record of the two single-year extensions rests on search summaries rather than on an opened page.
• The exact date the fishery closed in 2025. Two opened sources give 2025-08-01 and a search summary that was not opened gives 2025-08-12, the latter explaining an overrun of about 4,900 tonnes by the interval between catch reports. No opened source reconciles the two dates.
• Whether the two growth figures describe the same comparison. One source gives 118 percent for subarea 48.1 against the previous season and another gives 117 percent in a statement about a later season. The seasons named are different and neither source states its base, so the two cannot be treated as one figure.
• The area of the Antarctic Peninsula proposal. No source opened here gives it. The figures circulating for it differ and the brief page that would settle them refused automated retrieval.
• When the Weddell Sea proposal was first put. One opened page dates it to 2018 and an unopened search summary places a first version in 2016 within a European Union framework. No opened source states which of the two is the date it reached the Commission.
• What the Scientific Committee has completed on a management approach for the krill fishery. The paper that would carry it is a binary file whose contents were not read in this round, and no other opened source describes its state.
• The economic size of the fishery. No opened source gives revenue, employment or product volumes for krill harvesting, so the cost side of any decision about it is absent from this document.
• Whether the declines reported in penguin and fur seal numbers are attributable to the catch. The commentary that reports them does not establish that link, no monitoring data was opened, and the question of what caused them is not answered here.

See the evidence

No document of the Commission was read beyond its public pages. The convention text page carrying Article XII, the membership page, the convention area page, the convention history page, the Commission page and the marine protected areas page were opened directly, and between them they supply the decision rule, the count of 27 members, the area of 35,716,100 square kilometres, the dates of signature and entry into force, and both established protected areas with their dates. A United States federal fisheries agency page and the Antarctic agency of one proposing state were also opened directly and supply the Ross Sea dimensions and the East Antarctic proposal. Everything on the krill side is secondary. The expiry date of 2024-11-30 rests on a magazine feature and a news commentary that agree; the catch figures rest on a press release of an industry association; the second consecutive closure rests on a statement by a coalition of environmental organisations. Two of those three are parties to the debate they describe, on opposite sides of it, and they are carried here for the figures they publish rather than for the conclusions they draw. Where sources disagree the disagreement is left visible rather than resolved — the closure date is given as 2025-08-01 by two opened sources against 2025-08-12 in an unopened search summary, the concentration in subarea 48.1 is 357,961 tonnes in one place and 359,266 in another, the consecutive failure count is seven years in one account and eight meetings in another, and the first year of the Weddell Sea proposal is 2018 on the opened page and 2016 in an unopened one. Three rows carry no URL — two pages that would have settled parts of this could not be read, and one claim rests on a synthesis of search results whose individual pages were not opened. Rather than substituting a likelier source for any of them, those rows are left with the URL column empty and the reason in place. This is a Path A output, a research-based definition, so observation_refs is empty and provenance_mode: press-derived.

People affected

Estimated range Not derivable

The reason and what is missing are listed under “What is missing” below