LL1-12 digest — Ch12 The precautionary principle and early warnings of chemical contamination of the Great Lakes#

Late lessons from early warnings: the precautionary principle 1896–2000 (EEA, 2001), report pp. 126–134. Sole author: Michael Gilbertson (International Joint Commission). The editors flag him as a career-long researcher and advocate of Great Lakes clean-up (p. 12). His contributor note describes needing causal proof of injury before control as the mark of “inappropriate technologies” (p. 196). The chapter has no panels. Table 12.1 is credited to the EEA, not the author (p. 132).

Core story#

Organochlorines (DDT, dieldrin and related pesticides, PCBs, hexachlorobenzene, mirex) built up in the Great Lakes food web.

Warnings came first from wildlife. Visiting naturalists found soft-shelled cormorant eggs (1963). Ranch mink fed Great Lakes fish died (mid-1960s). Hickey’s students and co-workers documented effects of DDT and dieldrin on herring gull reproduction and mortality (1966), and the first deformed chick (a juvenile bald eagle) was reported in 1968. Systematic deformity studies came only in the early 1970s. Great Lakes bald eagle research began only after most of the population was gone, even though Florida declines had been published in 1952 and 1958 (p. 126). New analytical methods kept revealing contaminants: PCBs after 1966, hexachlorobenzene in 1972, mirex in 1974 (p. 126).

Human studies followed. The first (1974), of adult fishermen, found raised PCB levels but no recognised acute effects (p. 126). The Lake Michigan infant cohort (from 1980) linked prenatal PCB exposure to growth retardation that testing “indicated” was irreversible, memory and attention deficits, and an IQ deficit of more than six points at age 11 (p. 127). Oswego replicated the behavioural, but not the birth-size, effects (p. 129); the reference titles show these were newborn behavioural assessments, as was the later congener-specific finding (Stewart et al., 2000) (p. 134).

Action came through several routes: - UK restrictions (1961, 1964), opposed by Shell (p. 127) - Environmental Defense Fund litigation against a USDA that “tended to align itself” with manufacturers and farmers. After registration moved to the new EPA, this led to decisions “to suspend” DDT (1972) and dieldrin (1973) registrations (p. 128). EPA’s own history calls the DDT action a cancellation, and the dieldrin date looks early (external; unverified) - a Canadian petition (1969) (p. 128) - Monsanto’s 1970 limit on open PCB uses (p. 128) - the 1972 Great Lakes Water Quality Agreement and its 1978 pledge that persistent toxics be “virtually eliminated” (p. 128).

Concentrations fell sharply, then flattened. Levels dropped from about 1975 to the early 1980s, and bald eagle reproduction gradually improved (pp. 128–129). The decline then flattened at levels that still matter toxicologically: PCBs in water are about 100 times the cancer-based human-health criterion (uncited). Sediments, several hundred leaking landfills and air emissions sustain them (pp. 129–131). In the 1980s, officials demanded causal proof before “massive” spending (p. 128). The 1978 “virtually eliminated” pledge was “not properly implemented” (Table 12.1, p. 132).

Remediation costs are itemised, and benefits almost never estimated (pp. 130–131). Examples: - Detroit: about USD 2 billion spent and planned on general sewer and treatment works, with uncertain PCB benefit. - Niagara: failed containment would cost more than USD 19 billion over 100 years. That is the USD 3 billion outlay plus the USD 16 billion cost to society of no action.

Love Canal (1978–80) shows grassroots action forcing relocation over company and official denial (p. 127). The chapter tells only the residents’ side; later health studies were mixed (external).

The author’s lessons (pp. 129–132)#

• There are “extraordinary lengths of time” between introduction, detection, proof and an adequate response (citing Lawless, 1977). From introduction to action is “seldom less than 25 years” (p. 131).
• Scientists face a dilemma between investigating and pleading for action on suspicion.
• Proven causation has still not produced remediation: a “cognitive dissonance” between science and government.
• Several more decades are needed, with harm passed between generations through in-utero exposure.
• Without benefit estimates, support for clean-up will wane.

His distinctive argument is that precaution applies under high uncertainty. Integrated causal inference gives certainty, and so a defensible basis for remediation that makes precaution unnecessary. Complexity framings, by contrast, were “not inconvenient” to those resisting costs (p. 129). He meets O’Brien’s objection that this approach is “not precautionary” by treating that as a strength (p. 129). He recasts remediation as precaution for the next generation of infants (pp. 129–130). He contrasts settled science with complex social, economic and political dilemmas (pp. 131–132). Yet at p. 129 he calls it an open question whether present levels still affect development.

Transferable insights (technology-neutral)#

1. First signals come from indirect sentinels, and targeted research starts late (p. 126). Moderate.
2. What can be warned about is bounded by measurement capability (p. 126). Moderate.
3. Warnings from elsewhere may not be acted on until harm is seen locally (p. 126). Suggestive (the lag is documented; discounting is inferred).
4. Studying the wrong population or endpoint (acute effects in adults rather than developmental effects after prenatal exposure) gives false reassurance (pp. 126–127). Suggestive to moderate.
5. Incumbents demand proof the evidence cannot supply, keep denying after a cause has been traced, and pressure publishers and media (pp. 127–128). Moderate (partly one secondary source; the chapter does not date when the 1963 fish kill was traced to endrin, so “denial after tracing” is the author’s framing).
6. Regulators aligned with the regulated sector stall action; litigation and new institutions break the deadlock (p. 128). Moderate.
7. Precaution that restricts individuals rather than sources needs public legitimacy: the mirex fish ban collapsed (p. 128). Suggestive.
8. Commitments erode with political change (moderate), and demands for proof rise with the cost of the remedy (suggestive; one instance) (pp. 128, 132).
9. Persistent agents give quick early gains, then long legacy tails that cross generations (pp. 129–131). Strong.
10. Proof of harm is not enough; willingness to pay is the binding constraint (pp. 130–131). Moderate for this case; asserted as general.
11. Complexity and simplicity framings both serve interests; whoever controls the frame sets the burden of proof (p. 129). Suggestive.
12. Visible costs against invisible benefits erode support; deferral shifts cost to future generations (pp. 130–131). Moderate.
13. Protective advice creates distributional conflict, including for commercial, sport and tribal fisheries (p. 132). Suggestive.
14. Long-term monitoring with indicator organisms, set up early, makes both success and stalling visible (p. 129). Moderate for this case. (The editors add that monitoring and research can also raise new uncertainties, pp. 172–173.)

Main caveats#

• Participant-advocate author. He cites his own work, gives opponents only one-line treatment, and uses rhetorical intensifiers.
• Overstated certainty. “Proven” causation (p. 131) outran the 2001 evidence. Schantz et al. (2003) described a growing weight of evidence, with confounded congener attribution (external). The cohorts were exposed to contaminated fish, a mixture, so attribution to PCBs is an inference (reference titles, pp. 132–133), and the congener-specific finding the chapter relies on concerns newborn behaviour. The chapter’s own p. 129 leaves open whether present levels still harm development.
• Factual slips and omissions (external checks).
• Love Canal dumping was 1942–53, and the August 1978 federal emergency declaration is omitted.
• The DDT “suspension” (EPA called it a cancellation), the dieldrin date (1973), the Wisconsin “suit” against USDA, and Canada’s 1969 “ban” need checking.
• The TSCA PCB ban and the 1987 and 1995 Great Lakes measures are missing.
• One-sided costs. Almost only remediation costs. Beyond one unexplained “cost to society” of no action at four Niagara sites, there are no costs of harm and no benefits of use. Several figures are uncited.
• Uncalibrated lag figure. “Seldom less than 25 years” (from introduction to action) comes with no case list. The chapter’s own dates suggest the delay lay mainly in detection and accepted proof. Some responses after detection were quick: Monsanto about 4 years after PCBs were identified, and Love Canal relocation about 2.5 years after residents began investigating.
• Table 12.1 (credited to the EEA) says “banned” where the text says “suspend” for US DDT, generalises the eagle improvement to “wildlife”, and omits the Lake Michigan memory and IQ findings (p. 132).
• The editors re-read the chapter. The synthesis presents it as showing research amplifying uncertainty and “paralysis by analysis” (pp. 172–173, 181). The chapter itself argues that causal proof exists and that complexity framings were used strategically. Both agree that proof alone did not produce action (p. 181).

Hindsight. - The “several more decades” forecast looks right: PCBs remain the leading legacy contaminant in predator fish (Zhou et al., 2018). - The “asymptote” claim was too pessimistic: declines continued. - The prediction of waning support was partly wrong: large US programmes followed. - The 2012 Agreement Protocol added “precaution” explicitly but qualified virtual elimination “as appropriate”. - PCBs became a Chemical of Mutual Concern in 2016, and IARC classified them as Group 1 carcinogens in 2013.