LL2-17 digest — Ch17 Ecosystems and managing the dynamics of change#

Late lessons from early warnings II (EEA 2013), pp. 407–428. Authors: Jacqueline McGlade (then EEA Executive Director; formerly a Canadian DFO fisheries scientist) and Sybille van den Hove (report editorial team). Panel 17.1 is by McGlade alone, first person. Panel 17.2 is by Constança Belchior and Johnny Reker. No dissenting or industry panel.

Core argument#

The chapter’s claim is that we fail to anticipate sudden ecological change (p. 407). Warnings are often present. They are lost because of the following:

How knowledge is produced. Reductionist, equilibrium, “variable-oriented” models are used where configurational, “diversity-oriented” understanding is needed (pp. 408–409).
How it is siloed. Knowledge is split into political, disciplinary and geographic silos (pp. 407, 415).
How institutions filter it. Institutions filter and defend against inconvenient evidence (pp. 413–415, 419).
Poor institutional fit. Institutions are mismatched in scale to the systems they govern (pp. 419–420).

The chapter’s two exemplars:

Charles Keeling’s CO2 monitoring is the model of close, persistent observation (Box 17.1, p. 410).
The collapse of Newfoundland’s northern cod is the model of failure. It is set against Norway’s rapid response on Barents Sea cod (pp. 411–415).

It then surveys planetary boundaries, tipping elements, panarchy and resilience (pp. 416–419), and institutional fit (pp. 419–420). Panel 17.2 covers EU fisheries overriding scientific advice (pp. 421–422).

Key evidence#

Canada: warnings, then override. The warnings were in place long before the fishery closed:

Templeman’s 1960s warning that catch rates could rise as a stock collapsed (p. 413).
Larkin’s 1977 critique of MSY (p. 411).
Inshore fishers’ observations through the 1980s (p. 412).
A 1986 Memorial University review that reached “startlingly different conclusions” from the same data (p. 412).
The 1987 Alverson review: “tragically wrong” figures and a “faulty mathematical model” (p. 413).
DFO’s own December 1988 call to halve the quota (p. 413).

The political record ran the other way:

Ottawa raised the TAC for 1988.
Ministers rejected the 1989 cut because “the scientists had been wrong before”.
Politicians “asked for more evidence”.
The fishery closed only in 1992, blamed on “unusual ecological and environmental conditions” (pp. 413–414).

What went wrong in the assessment (pp. 411, 414):

The stock was treated as one unit.
Recruitment and natural mortality were held constant.
Discards distorted the data.
Offshore landings were used to “tune” the model.
Inshore data were excluded as “messy and often anecdotal”.

The 1990 Harris review found scientists “lulled by false data signals” and emotionally committed to the F0.1 strategy (p. 413). The panel says DFO “now admits that it had evidence as early as 1986”. It cites the Auditor General (1997) for the finding that the model for “most of the key groundfish stocks… consistently overstated their abundance” (p. 414). Only the model finding is clearly the Auditor General’s.

Institutional self-defence: the Fo’c’sle “credibility… not questioned” article, civil-service silence, and uncertainty played down before advisory bodies (pp. 414–415).

Norway. Norway made swift, politically backed cuts, called a “moratorium” in January 1990 (p. 414). It paid more than EUR 50 million to remove fleet capacity, and fishers’ incomes fell by up to 40%. It also funded research on cod–capelin–seal links. By 1992 spawning biomass was the highest in 25 years, a recovery the chapter attributes to the “prompt action” without further analysis (pp. 411, 414). The chapter’s summary: “no evidence of harm is not the same as evidence of no harm” (p. 411).

EU (Panel 17.2) (pp. 421–422):

TACs matched scientific advice for only 8% of stocks between 1987 and 2006.
Since 2003, TACs averaged 47% above advice.
About two-thirds of NE Atlantic TAC stocks lack data for MSY advice.
Member States fail their legal data duties.
The panel also blames “the lack of scientific knowledge and ineffective management options”, not only political override. It concedes that the impacts of TAC–advice mismatches “remain uncertain” (p. 421).

Authors’ lessons (p. 423)#

Close observation.
Diversity-oriented management.
Wider knowledge sources, including lay and local.
Shorter delays between warning and action.
Understanding the full adaptive cycle.
Capacity for transformation, not just adaptation, with participation and transparency about vested interests.

They also recommend:

Agreeing in advance “which diagnostic criteria and metrics will be used to elicit action” on early warnings (p. 423).
Dropping generic tipping-point models in favour of local and regional observation of cascading thresholds (p. 417).
Abandoning the “hubris” of assuming future states are predictable. The chapter likens that assumption to financial “quants” (p. 424).

Panel 17.2 wants scientific advice treated as a “fixed boundary condition” (p. 422).

Main mechanisms#

Masked decline. Indicators produced by the damaging activity itself may hide the decline (hyperstability). Templeman warned of this, and the panel offers it as “one potential explanation” (p. 413); it is not documented in the chapter.
Loyalty to the model. The model becomes the object of commitment.
Evidence rules. Admissibility rules exclude lay signals.
Credibility trap. Earlier expert error is used to discount later warnings.
Delay. Demands for “more evidence”.
Burden of proof. Proof of harm is required before acting.
Legitimacy defence.
Short-term socio-economic override.
Framing. The “tragedy of the commons” framing narrowed the solution to fleet cuts (p. 419).
Uneven costs. Authority built on control of science survives its own failure while communities bear the collapse (p. 419).
Courts decline to weigh science (p. 420).

Transferable insights (technology-neutral)#

#	Insight	Pages	Strength
1	Activity-generated indicators can look fine while the system collapses	413–414	moderate–strong (mechanism offered as “potential explanation”)
2	Commitment to a model or strategy blinds institutions to anomalies	413, 415	strong (in case)
3	Rules on what counts as evidence can exclude the earliest warnings	412, 414	strong in case, moderate general
4	Independent re-analysis by outsiders surfaces warnings the institution misses	412–413	moderate–strong
5	The burden of proof determines the timing of action	411, 413	strong logic, moderate evidence
6	Credibility trap: past error used to dismiss correct warnings	413	moderate
7	“More evidence” demands function as delay	413	moderate–strong
8	Concentrated short-term costs drive repeated override of advice	413, 419–421	strong
9	Legitimacy threats produce reassurance and denial; internal dissent muted	413–415	moderate (dissent part rests mainly on McGlade’s insider testimony)
10	Prompt, backed action plus research and capacity cuts enabled recovery (Norway)	411, 414	suggestive–moderate
11	Framing narrows the solution space	419	moderate
12	Knowledge institutions survive their failures; costs fall on the least powerful	419	suggestive
13	Pre-agreed triggers for action reduce delay	423	asserted (logically strong)
14	Data non-compliance, uncertain advice and easy override form a loop	421	moderate

Main caveats#

Insider advocacy. The chapter is co-written by the EEA’s own head, and its central case is her first-person account. No DFO, industry or government voice appears except as evidence of error.
Weakest sourcing on the exemplars. The Norway contrast rests mainly on a journalist’s book (Michael Harris 1998). “Moratorium”, the claimed ban on spawning-ground fishing and the named minister (whose remark is given only in reported speech) all need checking. Box 17.1 is unsourced and contains errors, including Keeling’s death date (2005, not 2004). The Nunavut “absolute discretion” passage presented as the Minister’s claim may be the court’s statement of law (unverified).
Inconsistencies. The main text dates the cod collapse to 1986 (p. 411). The panel says “in the 1980s” (p. 412) but has the fishery closed in 1992 (p. 414), and Annex 2 gives 1992 (p. 703). The Parsons quote dates a rise in natural mortality to “the early 1980s” for a 1991 decline (p. 413).
The central dichotomy is asserted, not demonstrated. Most cod failures could be fixed within better conventional assessment. Keeling is a poor fit for “diversity-oriented” analysis. The four-concept lens is never applied to the cases. The chapter itself concedes that diversity-oriented analyses “tend not to provide insights into causal generalisations” (p. 409) and that there is no consensus replacement for simple models (p. 415).
Environmental causes are treated asymmetrically. They are accepted for Norway but called “myth-like” for Canada.
Norway acted on a 100-year low. That is prompt response rather than precaution under uncertainty.
Unresolved tension about science. Deference to scientific advice (Panel 17.2) sits beside a critique of science’s entrenched authority (p. 419). Panel 17.2’s own figures also show NE Atlantic improvement despite “continued to worsen”. It claims “sufficient available knowledge” (p. 422) while stressing large data gaps (p. 421).
Things to check against later evidence: northern cod after 2013, Barents Sea cod after 2013, CFP 2013 reform outcomes, planetary-boundary assessments, and IPBES’s use of indigenous and local knowledge.