LL1-12 hindsight check: Great Lakes chemical contamination (Gilbertson), Late lessons from early warnings (EEA, 2001), Ch. 12, pp. 126–134#

Checked 25 September 2026. The check covers what happened between 2001 and September 2026 that bears on the chapter’s claims, evidence, forecasts and recommendations. Page numbers refer to the 2001 report. The chapter’s sole author, Michael Gilbertson, worked for the International Joint Commission (IJC) and was a long-standing advocate of Great Lakes clean-up; that context matters for how firmly some claims are stated.

Overview#

The chapter makes three kinds of claim: a history of early warnings and responses (1960s–1980s); a scientific judgement (prenatal PCB exposure from Great Lakes fish has been “proven” to injure children’s development); and forecasts and policy diagnoses (decades more of remediation; stalled declines; waning support; governments failing the 1978 “virtual elimination” pledge). Twenty-five years on, the record splits cleanly.

Vindicated or strengthened - The long legacy tail (claim 1). PCBs still drive most Great Lakes fish consumption advisories. The State of the Great Lakes 2025 fish-fillet assessment rates the lakes “Fair” overall; PCB levels have fallen 65–94% since the 1970s but remain in the “Fair” (1–4 meals a month) band for Lakes Michigan and Ontario. The chapter’s “several more decades” is on track to be literally true. - The direction of the neurodevelopmental evidence (claim 3). The Oswego cohort replicated the Michigan IQ finding at age 9 (Stewart et al., 2008). Reviews found consistent deficits in executive function, attention and cognition (Boucher et al., 2009; Balalian et al., 2024). PCBs were named among five industrial chemicals recognised as developmental neurotoxicants (Grandjean and Landrigan, 2006). IARC classified PCBs as carcinogenic to humans in 2013, strengthening the basis of the cancer-based water criterion the chapter cites. - Wildlife recovery after the bans (claim 8). The bald eagle was delisted in the lower 48 United States in 2007; shoreline Great Lakes eagles recovered more slowly and still showed PCB-associated productivity effects into the 2000s, consistent with “gradual”. - Virtual elimination unachieved (claim 7). The 2012 Protocol kept virtual elimination and zero discharge but qualified both with “as appropriate”. Only eight Chemicals of Mutual Concern have been designated (2016); decisions on further nominations are still pending in 2026. The IJC calls progress “incremental”.

Overstated, wrong or not borne out - “Proven” causation (p. 131) and “irreversible” growth retardation (p. 127). The later literature is a strong weight-of-evidence case, not proof; heterogeneity across cohorts remains a live objection, and whether today’s lower exposures still harm development is unresolved (a question the chapter itself raises on p. 129). - Congener specificity “determined” (Table 12.1, p. 132). Collinearity among congeners makes this unresolvable with the data used; mechanistic work points to structural classes that do not map neatly onto “highly chlorinated”. - Support “likely to wane” (p. 131). The opposite happened: the Great Lakes Legacy Act (2002), the Great Lakes Restoration Initiative (2010; more than USD 4.8 billion through FY2025), USD 1 billion from the 2021 infrastructure law, repeated congressional protection against proposed cuts (2017, 2025), and a new wave of Area of Concern delistings (12 of 43 by August 2026). - Love Canal (p. 127). Officials did not uniformly deny harm: New York State and the federal government declared emergencies and began relocations in August 1978. Later studies support some reproductive effects (preterm birth, low birth weight) but not the cancer excess the chapter lists.

What the chapter could not see - Recovery is not monotonic. Invasive mussels re-routed PCBs through food webs: dissolved PCBs in Lake Ontario water have rebounded from a late-1990s low, and Lake Erie fish declines slowed through remobilisation (Pagano et al., 2025; Pagano and Garner, 2024). - Some floors are set by current, not legacy, sources. An inadvertent by-product PCB congener (PCB-11) did not decline in Great Lakes air from 2004 to 2015, and hexachlorobenzene has “not changed much in 25 years” (Hites, 2018; Hites et al., 2022). - The legacy stock itself is barely managed in places. US PCB stocks in use fell only about 3% (by mass) after 2006, and at most 30% of countries are on track for the Stockholm Convention’s 2028 PCB goal (Melymuk et al., 2022). - Benefit estimates arrived, but not the kind the chapter asked for. Governments now cite economic returns (USD 3.35 of activity per USD 1 of restoration spending) rather than health benefits.

Weight for the lens. The chapter’s mechanism-level lessons (long legacy tails from persistent agents; wildlife sentinels; long-term monitoring; proof not being sufficient to trigger action; commitments being qualified over time) are well supported by the later record. Its certainty language, its congener claim, its Love Canal narrative and its prediction of waning support should not be cited without correction. The most useful hindsight lesson is one the chapter points towards but does not state: what unlocked remediation was not “proof” or “precaution” as such, but dedicated appropriations tied to bounded, measurable local targets (Areas of Concern and their impairment lists) and an economic-benefits case.

2013 update. Late lessons from early warnings II (EEA Report 1/2013) has no Great Lakes update in Annex 3 (“An update of some case studies”), which covers fisheries, benzene, asbestos, PCBs, halocarbons, DES, antimicrobials, MTBE and hormones. Annex 2 (p. 713) reprints Table 12.1 unchanged, including its errors (“DDT and related pesticides are banned in Canada” in 1969; dieldrin “banned in 1973”), and closes: “Decades later the efforts to clean up the area continue.” Annex 2’s Table A2.1 (after Gee, 2009) gives the case “45+” years of “substantial inaction” and notes “Debates continue about persistent health damaging pollution”. The Annex 3 PCB update (Koppe) adds claims about PCBs and diabetes, obesity and epigenetics, and says denial of PCB toxicity “is still taken place”; it does not address the Great Lakes. I read the 2013 annexes from the EEA-hosted PDF, not the project’s local copy.

Claim 1. It will probably take “several more decades” of remediation before organochlorine concentrations protect human reproduction and development, especially via fish consumption (p. 131)#

Original claim. “The existing evidence concerning the persistence of the organochlorine compounds released into the Great Lakes during the past 60 years indicates that it will probably be several more decades before the necessary remedial actions will have reduced concentrations sufficiently to protect human reproduction and development from chemically induced injury, particularly from consumption of contaminated Great Lakes fish” (p. 131). The chapter frames this as “trans-generational transmission” of contamination and injury (p. 131).

What happened since. - PCBs remain the dominant legacy contaminant in predator fish. Zhou et al. (2018, Journal of Great Lakes Research) analysed the US EPA Great Lakes Fish Monitoring and Surveillance Program (lake trout and walleye, 1999–2014). PCBs made up 70–75% of legacy contaminant mass. The authors concluded that total PCBs “will continue to be of the greatest concern into the future” because they have the largest mass and the second-slowest decline. - Fish advisories are still PCB-driven, but easing. The State of the Great Lakes (SOGL) 2025 “Contaminants in Fish Fillets” sub-indicator (published June 2026 by the US and Canadian governments): - Overall status “Fair”; 10-year (2013–2022) and long-term (1975–2023) trends “Improving”. - “Fish consumption advisories for all of the Great Lakes have typically been driven by” PCBs; mercury is secondary. - PCB levels in the edible portion of five indicator species have fallen 65–94% since the 1970s. - Average PCBs in the latest cycle: Lake Michigan 168 ng/g and Lake Ontario 158 ng/g (both “Fair”, roughly 1–4 meals a month); Superior 68, Huron 76 and Erie 97 ng/g (“Good”, below 100 ng/g, though individual values fall in “Fair”). - Share of Canadian-water advisories driven by PCBs: Lake Ontario 67%, Lake Erie 57%, Lake Huron 36%, Lake Superior 35% (mercury leads in Superior and Huron). - Advice has loosened: for Canadian waters of Lake Ontario, “Do Not Eat” advisories fell from 32.1% to 18.0% over ten years and unrestricted advice rose from 33.1% to 50.3%. - PFAS now account for about 10% of Canadian advisories (SOGL fish-consumption summary page). - The Parties’ own assessment. The 2025 Progress Report of the Parties (January 2026) states that “some chemicals still bioaccumulate in the food web to levels that harm wildlife and create the need for fish consumption advisories to protect human health”. The 2012 Agreement’s General Objective that waters “allow for human consumption of fish and wildlife unrestricted by concerns due to harmful pollutants” is not met. - Declines continue but slowly. Pagano and Garner (2024, Environmental Science & Technology) report legacy contaminant groups down 25.8–97.9% since 2004 in top predator fish, with Lake Michigan sites the most impacted on an age-normalised index. - The legacy stock is not closed off. Melymuk et al. (2022, ES&T) estimate US PCB stocks in use fell only about 3% (by mass) after 2006; the US is not a party to the Stockholm Convention. Ontario and Czechia, by contrast, cut pure PCB stocks by 99% in ten years.

Verdict: held up. Twenty-five years after publication, PCB concentrations in fish still restrict consumption in the two most contaminated lakes, and the Parties still describe levels that harm wildlife. The forecast is on track to be literally correct. The chapter’s specific endpoint (concentrations low enough to protect reproduction and development) is not directly measured by anyone; advisories are the working proxy, and those have eased substantially.

Weight. Strong support for the lesson that persistent agents create legacy tails spanning generations (digest insight 9). A careful user should note the improvement as well as the persistence: most of the reduction in fish PCBs was achieved decades ago, and advice in three of five lakes now falls mostly in the “Good” band.

Claim 2. Declines have in practice become “non-zero asymptotes” at toxicologically significant levels; PCBs in water are about two orders of magnitude above the cancer-based human-health criterion (p. 129)#

Original claim. Concentrations fell markedly from about 1975 to the early 1980s (citing Stow et al., 1999). “While technically the decreases continue at the same logarithmic rates, in practice the curves have become non-zero asymptotes at concentrations that are still of toxicological significance”. “The present concentrations of PCBs in water are about two orders of magnitude higher than the established water quality criterion for the protection of human health based on a cancer risk assessment” (p. 129; no citation).

What happened since. - The criterion. The Great Lakes Water Quality Initiative human cancer criterion for PCBs was revised in 1997 to 6.7 × 10⁻⁶ µg/L, i.e. 6.7 pg/L (US EPA, Federal Register, 12 March 1997). - Water concentrations, 2012–2019. Pagano et al. (2025, Science of the Total Environment) measured total PCBs on US EPA Lake Guardian cruises: - Average 268 pg/L (about 40 times the criterion); range 72 pg/L at Keweenaw Point, Lake Superior (about 11 times) to 834 pg/L at Middle Bass Island, Lake Erie (about 125 times). - Lake Erie fell 44% (2014–2019) and Lake Huron 67% (2012–2017). - Lake Ontario dissolved-phase PCBs “significantly increased, rebounding from a low-point in the late-1990’s“, probably because invasive zebra and quagga mussels diverted PCBs into the benthic food web. - Breakpoint analysis found substantially slower decline rates from the early 1990s in Lakes Superior and Michigan. - Fish. Zhou et al. (2018) found annual declines above 10% in earlier years but under 2% a year after 2010. SOGL 2025 says PCB (and mercury) levels “appear to be improving at slower rates in recent years” in Lakes Superior, Huron and Ontario; Lake Michigan continued to improve through the last decade; Lake Erie fish levels stabilised and in some cases rose in the 2010s (attributed to invasive mussels and round goby) before falling again. Pagano and Garner (2024) attribute slower halving times at Lake Erie sites to remobilisation from extreme weather and past Detroit River remediation. - Air. US EPA’s IADN summary (2017 snapshot): atmospheric PCBs “continue to decrease”, halving about every 15 years. But two sources do not follow the legacy pattern: - PCB-11, an inadvertent by-product of pigment manufacture, showed no significant change at six Great Lakes sites from 2004 to 2015 while Aroclor-type PCBs halved about every 12 years (Hites, 2018, ES&T Letters). - Hexachlorobenzene (whose 1972 discovery in tern eggs the chapter reports, p. 126) and octachlorostyrene concentrations in Great Lakes air “have not changed much in 25 years” (Hites, Bidleman and Venier, 2022, ES&T Letters). - Composite indicator. SOGL 2025 rates Toxic Chemicals “Fair” and “Unchanging to Improving”. Whole-fish sub-indicator trends are “No Trend” for four of five lakes (high variability) and “Improving” for Erie.

Verdict: partly held up. The chapter was right that the steep early declines gave way to much slower change, that concentrations remain toxicologically significant (water PCBs are still one to two orders of magnitude above the cancer-based criterion), and that residual sources sustain them. It was wrong to say the decreases “technically continue at the same logarithmic rates”: the rates themselves slowed (breakpoints in the early 1990s), and in Lake Ontario water they reversed. It was too pessimistic if “asymptote” is read as a floor: fish concentrations kept falling in most lakes, enough to loosen advice substantially.

Weight. Supports a more nuanced version of the legacy-tail lesson: after a source is cut, declines are fast then slow, can reverse when ecosystems change, and bottom out where unregulated or by-product sources continue. The uncited “two orders of magnitude” figure is plausible against the 1997 criterion but should be cited now as “roughly 10 to 125 times the criterion (2012–2019)”.

Claim 3. Prenatal PCB exposure from Great Lakes fish causes irreversible growth retardation, memory and attention deficits, and an IQ deficit of more than 6 points at age 11; the causal relationship “has now been proven” (pp. 127, 131)#

Original claim. In the Lake Michigan cohort (from 1980), the most exposed infants weighed less and had smaller heads (Fein et al., 1984); testing “indicated that the growth retardation was irreversible”; memory and attention were affected; at 11 years the most exposed children had IQ scores “more than six points below the reference group” (Jacobson and Jacobson, 1996) (p. 127). At p. 131: “a causal relationship has now been proven between the injury to health and exposures to persistent toxic substances”; “the scientific aspects are characterised by a high degree of certainty”. But at p. 129 the chapter says “the question remains whether present levels are affecting human development”.

What happened since. - Weight of evidence growing but not conclusive (2003). Schantz, Widholm and Rice (2003, Environmental Health Perspectives 111: 357–376) reviewed cohorts in Taiwan, Michigan, New York (Oswego), the Netherlands, Germany and the Faroe Islands: all reported negative associations between prenatal PCB exposure and cognitive function; only a North Carolina study found none. They described “the weight of evidence for PCB effects on neurodevelopment” as “growing”, not settled. - Dissent. Gilbert Ross of the American Council on Science and Health (2004, Ecotoxicology and Environmental Safety 59: 275–291) argued there is “no reliable evidence” that environmental PCBs cause “intellectual deterioration in children exposed in utero”, and that “little benefit to public health” would come from continued remediation. I found no major assessment body that adopted this view. - Replication of the IQ finding. In the Oswego (Lake Ontario) cohort, Stewart et al. (2008, EHP 116: 1416–1422) found that each 1 ng/g increase in placental PCBs was associated with a three-point drop in Full Scale IQ at age 9 (p = 0.02), roughly a six- to seven-point difference between least and most exposed. Methylmercury was unrelated to IQ, with no PCB–mercury interaction. The authors concluded, together with the Michigan results, that prenatal PCB exposure in the Great Lakes region “is associated with lower IQ”. - Consistent profile. Boucher, Muckle and Bastien (2009, EHP 117: 7–16): the “most consistent effects” across studies are impaired executive functioning, with effects on processing speed, verbal abilities and visual recognition memory also reported by most studies. - Mainstream classification. Grandjean and Landrigan (2006, Lancet 368: 2167–2178) listed PCBs among five industrial chemicals (with lead, methylmercury, arsenic and toluene) recognised as causes of neurodevelopmental disorders and subclinical brain dysfunction. - Heterogeneity objection. Goodman et al. (2010, EHP 118: 727–734) found the prospective cohorts too dissimilar in exposure metrics, outcomes and reporting to support a meaningful meta-analysis, and warned that weight-of-evidence assessment “may be limited”. The study was funded by Cefic-LRI (the European chemical industry council’s research programme), which the authors state had no role in the analysis. - Latest systematic review. Balalian et al. (2024, Environmental Research 252: 118912; 87 studies) found evidence for adverse cognitive development and attention problems in middle childhood; “no or negligible” association with early motor development or ADHD/autism risk; possible stronger effects in boys; and “significant heterogeneity”. They call for work on the lower post-ban exposure levels, i.e. the chapter’s own p. 129 question remains open. - Mechanistic review. Pessah et al. (2019, Acta Neuropathologica 138: 363–387): “most studies reported that prenatal PCBs were related to poorer cognitive function and more behavior problems”. - Growth. A meta-analysis of 12 European birth cohorts (Govarts et al., 2012, EHP 120: 162–170) found birth weight fell by 150 g (95% CI −250 to −50 g) per 1 µg/L increase in cord-serum PCB-153, concluding that low-level PCB exposure “(or correlated exposures)” impairs fetal growth. I found no later evidence establishing that the Michigan growth deficit was irreversible; the Oswego cohort did not replicate the birth-size effect (as the chapter itself notes, p. 129). - Cancer. IARC (Monograph 107; Lauby-Secretan et al., Lancet Oncology, March 2013) classified PCBs and dioxin-like PCBs as carcinogenic to humans (Group 1). This postdates the chapter but strengthens the cancer-risk basis of the water criterion in claim 2.

Verdict: partly held up. The core finding (prenatal PCB exposure is associated with lower cognitive performance and attention, at levels found in Great Lakes fish eaters) was replicated and is now mainstream. The chapter’s language went beyond the evidence: “proven” causation, “irreversible” growth retardation and a clean six-point IQ effect were weight-of-evidence judgements in 2001 and remain so; the heterogeneity objection has not gone away; and whether current, lower exposures still cause harm is unresolved.

Weight. The direction of this evidence can carry weight. The certainty cannot: cite “strong and consistent weight of evidence, contested in its magnitude and generalisability”, not “proof”. The case illustrates a durable pattern: heterogeneous observational evidence lets both advocates and opponents claim the same literature, and systematic-review methods do not by themselves settle the dispute (Goodman et al., 2010, versus Boucher et al., 2009).

Claim 4. The behavioural anomalies are specifically related to prenatal exposure to highly chlorinated PCBs (Stewart et al., 2000); Table 12.1 says this was “determined” in 2000 (pp. 129, 132)#

Original claim. “Subsequent chemical analytical determinations have shown the specific relationship between these behavioural anomalies and prenatal exposures to the highly chlorinated biphenyls (Stewart et al., 2000)” (p. 129). Table 12.1 (credited to the EEA, not the author): “The specific relationship between behavioural anomalies and prenatal exposures to the highly chlorinated biphenyls is determined” (p. 132).

What happened since. - The statistical problem. Schantz et al. (2003) noted that attempts to attribute effects to individual congeners “are hampered by the fact that concentrations of most individual congeners are highly correlated with each other and with total PCBs”, and that many congeners present in human tissue had never been tested for neurotoxicity. Pessah et al. (2019) repeat the point: collinearity makes “disentangling individual congeners difficult”, and most studies use sums dominated by single congeners such as PCB-153. - Supporting data from the same group. Stewart et al. (2008) reported that among the four major congeners (PCBs 118, 138, 153, 180), “the individual associations with IQ were larger among the more highly chlorinated PCBs”. Stewart et al. (2003, EHP 111: 1670–1677) linked cord-blood PCBs to response-inhibition errors, moderated by the size of the corpus callosum, and noted that findings “await replication”. - A different mechanistic axis. Mechanistic research emphasises non-dioxin-like, ortho-substituted congeners acting on calcium signalling through ryanodine receptors (e.g. PCB-95 and PCB-136, which are penta- and hexachlorinated) (Pessah et al., 2019). This does not contradict the Oswego associations, but it does not map onto “highly chlorinated” either. - An alternative reading (my inference). The most highly chlorinated congeners are also the most persistent and bioaccumulative. Their levels in cord blood may therefore be the best available marker of cumulative fish-derived exposure to the whole mixture, rather than the active agents.

Verdict: contested. The Oswego team’s later data are consistent with its 2000 finding, but the specificity claim cannot be established from correlated congener measurements, and no independent body has endorsed it. “Determined” overstates what a single cohort analysis could show.

Weight. Low. Do not cite congener specificity as established. The broader lesson is about exposure markers: when agents occur as correlated mixtures, attributing harm to a sub-component is often beyond what observational data can support.

Claim 5. The time from introducing a technology, product or undertaking to regulatory, judicial or administrative action to reduce exposure is “seldom less than 25 years” (p. 131)#

Original claim. “These various case studies demonstrate that the length of time between the introduction of a new technology, product or undertaking, the discovery of its deleterious effects, and the regulatory, judicial or administrative action to reduce exposures, is seldom less than 25 years” (p. 131), following a reference to “extraordinary lengths of time” (Lawless, 1977).

What happened since and what the record shows. - Against the chapter’s own cases (dates from later primary sources): - DDT: widespread use from the mid-1940s; US EPA cancellation order in 1972 (US EPA): about 27 years. - Aldrin and dieldrin: first produced or used 1948 and 1950; USDA cancelled uses in 1970 but some were reinstated; EPA cancelled crop uses in 1974; termite use ended by voluntary cancellation in 1987 (ATSDR, 2022): 24–39 years. - PCBs: commercial production from 1929; Monsanto’s open-use restriction in 1970 (p. 128): about 41 years. - Love Canal: dumping from 1942 to 1953 (US EPA); emergency action in 1978: 25–36 years. So the figure fits the chapter’s own examples when measured from introduction. - But it conflates two lags. Measured from detection, some responses were quick: Monsanto acted about four years after PCBs were identified in wildlife (1966), and the Love Canal emergency came within months of residents’ 1978 inquiries. The long lags lay mainly in detection and in accepted proof. - The 2013 report’s own tabulation (Table A2.1, after Gee, 2009) measures a different interval (first early warning to effective action) and finds a wide spread: 5–30 years (TBT), 10–17 (BSE), 10–30 (halocarbons), 30–50 (DES), about 100 (PCBs, asbestos), “45+” (Great Lakes). No study I found has calibrated the “25 years” figure across a defined set of cases. - Post-2001 lags in this case’s own domain. - Global PCB phase-out: the Stockholm Convention (in force 2004) requires elimination of PCB use in equipment by 2025 and environmentally sound management of PCB wastes by 2028. At most 30% of countries are on track for 2028 (Melymuk et al., 2022). That is about a century after commercial introduction. - Designating new Chemicals of Mutual Concern under the 2012 Agreement: polycyclic aromatic hydrocarbons became a candidate in 2022, with a recommendation “anticipated in 2026”; recommendations on lead and radionuclides are also expected in 2026 (2025 Progress Report of the Parties). Nominations have taken years just to reach a designation decision.

Verdict: partly held up. As a description of this chapter’s cases, “seldom less than 25 years” from introduction to action is accurate. As a general law it is uncalibrated, it merges detection lag with response lag, and the EEA’s own later tabulation shows wide variation.

Weight. Use the figure illustratively, not as an empirical constant. The better-supported, technology-neutral lesson is the decomposition: for persistent agents, detection capability and the standard of proof set most of the delay, and complete elimination of a legacy stock can take far longer than the first restriction.

Claim 6. Without estimates of the benefits of remediation, “public and political support for these costly schemes is likely to wane” (p. 131)#

Original claim. After itemising Niagara landfill and sediment remediation costs, the chapter says: “There is a need for estimates of the benefits that will accrue from remedial actions at these and at other sites around the Great Lakes, or public and political support for these costly schemes is likely to wane” (p. 131). It notes “few benefits analyses” (p. 131) and closes by asking whether acting on the evidence would be “an apparently impossible financial burden” (p. 132).

What happened since. - Great Lakes Legacy Act (2002). First appropriated 2004, reauthorised 2008. By March 2026: 6.9 million cubic yards of contaminated sediment remediated; 34 projects completed and 6 under way; about USD 1.2 billion committed, of which about USD 510 million came from non-federal sponsors (US EPA). - Great Lakes Restoration Initiative (2010). USD 475 million in FY2010; about USD 300 million a year from FY2011 to FY2019; rising to USD 368 million (FY2024–25) and USD 369 million (FY2026). Over 9,200 projects and more than USD 4.8 billion through FY2025 (glri.us). - Infrastructure law (2021). USD 1 billion for Areas of Concern (AOCs). EPA said in February 2022 this would allow work to be completed at 22 of the 25 remaining US AOCs by the end of 2030 (as reported by the American Presidency Project). - Political tests passed. - In 2017 the administration proposed eliminating GLRI funding (InquireFirst, secondary); Congress kept it at USD 300 million. - The FY2026 request held GLRI at USD 368 million while proposing a roughly 55% cut to EPA overall (Circle of Blue, June 2025, commentary). Congress enacted USD 369 million. - The FY2027 request proposed USD 367.7 million (Wisconsin Public Radio, 15 May 2026). - On 21 July 2026 the Senate Environment and Public Works Committee unanimously advanced a reauthorisation at USD 475 million a year through 2030 (Sen. Slotkin press release). - Canada. A Freshwater Action Plan of CAD 650 million over ten years from 2023, including CAD 420 million for the Great Lakes (Canada Water Agency, August 2026). The Randle Reef project in Hamilton Harbour cost CAD 138.9 million, shared three ways between federal, provincial and local partners; its main dredging stage was completed in 2022 (Water Canada, trade press). The chapter’s range for Hamilton Harbour sediments was CAD 60 million to 1 billion (p. 131). - Delistings accelerated. Of 43 AOCs designated (26 US, 12 Canadian, 5 binational): - United States: eight delisted as of August 2026 (US EPA list: Oswego River, Presque Isle Bay, Deer Lake, White Lake, Lower Menominee River, Ashtabula River, Rochester Embayment and Muskegon Lake, the last in 2025). - Canada: four delisted (Collingwood Harbour 1994, Severn Sound 2003, Wheatley Harbour 2010, Nipigon Bay August 2026), plus two “Areas in Recovery” (Canada Water Agency). - All management actions complete at a further 11 US and 5 Canadian AOCs (2025 Progress Report of the Parties). By October 2025, 133 of 255 beneficial use impairments had been removed (US EPA). The Black River (Ohio) was put forward for delisting in July 2026 (IJC). - Benefit estimates were produced, but mostly economic. A University of Michigan study with the Great Lakes Commission (September 2018) estimated that GLRI spending generates regional economic activity, higher waterfront property values and tourism through 2036; governments now cite “USD 3.35 in economic activity for every USD 1 invested” and estimate CAD 174 million in local benefits from Randle Reef (2025 Progress Report of the Parties). These studies measure economic activity and amenity, not avoided health harm, which the chapter’s Burtraw and Krupnick (1999) reference had in mind.

Verdict: weakened. Support did not wane; it grew and survived hostile budget proposals in 2017 and 2025. The chapter’s underlying mechanism (visible costs need visible benefits to sustain support) is arguably borne out, since benefit estimates became part of the political case. But the benefits that sustained support were economic regeneration, not the health benefits the chapter emphasised, and action came through a new, dedicated funding channel rather than through the 1978 Agreement’s toxic-substance commitments.

Weight. Do not cite the forecast as correct. The durable lesson is conditional: remediation of legacy harm gains durable political support when it is packaged as bounded, local, measurable projects with visible economic returns. Note also the limit: funding has been flat in cash terms since FY2024 and below the authorised USD 475 million.

Claim 7. Governments have not properly implemented the 1978 commitment to virtually eliminate discharges of persistent toxic substances, and remain reluctant to fund remediation even after causation is proven (pp. 128–130, 132)#

Original claim. The 1978 Agreement’s precautionary policy (“the discharge of any or all persistent toxic substances be virtually eliminated”) met 1980s scepticism and demands for causal proof before “massive” spending (p. 128); controls and clean-up of non-point sources fell short of what is “required to protect human health” (p. 129); there is “a powerful cognitive dissonance” between the science and government response (p. 130). Table 12.1: the policy was “not properly implemented” (1978) and there is “reluctance to undertake costly remedial actions even after causal relationship is proven” (2000) (p. 132).

What happened since. - 2012 Protocol. The amended Agreement adds precaution explicitly as a principle (“incorporating the precautionary approach, as set forth in the Rio Declaration”). It keeps virtual elimination and zero discharge but qualifies both: “adopting the principle of virtual elimination for elimination of releases of chemicals of mutual concern, as appropriate” and “adopting the philosophy of zero discharge … as appropriate” (Article 2). Annex 3 repeats “as appropriate”. It also adds an “ecosystem approach” and “adaptive management”. - Chemicals of Mutual Concern. The Parties designated eight in 2016 (PCBs, mercury, PBDEs, HBCD, PFOS, PFOA, long-chain PFCAs, short-chain chlorinated paraffins). None has been added since. PAHs became a candidate in 2022, with a recommendation “anticipated in 2026”; screening for lead and radionuclides is under way, with recommendations anticipated in 2026; sulfates were found in 2025 not to cause transboundary impacts (2025 Progress Report of the Parties). - IJC assessments. - 2020 (Second Triennial Assessment of Progress): the Commission reported hearing that CMC progress “is inadequate”; binational strategies were complete for only three of eight CMCs; and strategies may not reflect “pollution prevention, zero discharge and virtual elimination”. (These are public comments the IJC reports, not its own finding.) - 2023 (Third Assessment): the CMCs are “generally declining”; the Parties are making “incremental progress”, but “much more work is required”; data gaps constrain assessment. The IJC lists precaution, zero discharge and virtual elimination among the Agreement’s key achievements as principles. - PCB stocks. The US had reduced PCB stocks in use by only about 3% (mass) after 2006 and is not a Stockholm Convention party (Melymuk et al., 2022). - Funding of remediation. Contrary to the reluctance diagnosis for the period after 2001, dedicated remediation funding arrived from 2002 onwards (claim 6), though about 24 years after the 1978 pledge and outside the Agreement’s toxic-substance programme.

Verdict: partly held up. Virtual elimination remains unachieved and was formally softened by “as appropriate” in 2012. Designation and strategy work under the 2012 Agreement has been slow, and a large US PCB stock remains. The diagnosis of reluctance to fund remediation does not describe the post-2002 period.

Weight. Supports two technology-neutral lessons: (i) absolute goals in framework agreements tend to be retained as principles but qualified in operational text; (ii) implementation shifts to where progress is measurable and fundable (site clean-ups), while source elimination of dispersed stocks lags. The chapter’s “cognitive dissonance” framing is advocacy; the later record is better described as selective implementation.

Claim 8. The US and Canadian bans of DDT and related pesticides had an “immediate effect” on Great Lakes concentrations, reflected in gradual recovery of bald eagle status and reproduction (p. 128)#

Original claim. Litigation led to “decisions to suspend the registrations of DDT, in 1972, and dieldrin, in 1973”; Canada banned “DDT and related pesticides” after a 1969 petition. “These national decisions had an immediate effect on the concentrations of organochlorine pesticides in the Great Lakes environment that was reflected in gradual improvements in the status and reproduction of bald eagles (Grier, 1982)” (p. 128).

What happened since. - National recovery. US Fish and Wildlife Service: 417 known nesting pairs in the lower 48 states in 1963; at least 9,789 pairs by 2007; removal from the endangered and threatened list announced 28 June 2007; an estimated 316,700 individuals including 71,467 breeding pairs in 2018–2019 surveys. FWS attributes the eggshell failure to DDT and the recovery partly to the 1972 ban. - Great Lakes shorelines lagged. Bowerman et al. (1995, EHP) reported that eagles nesting along Great Lakes shorelines remained impaired while inland populations recovered, with PCBs and dioxin-like compounds the significant hazard. Bowerman et al. (2003, Environmental Toxicology and Chemistry) found nestling blood PCBs and DDE inversely correlated with regional productivity across ten Great Lakes subpopulations (1997–2000). Best et al. (2010, ET&C 29: 1581–1592; 197 eggs, 1986–2000): eggshell thickness had recovered to pre-1946 levels by 2000, but productivity declined significantly above 26 µg/g PCBs. Eakin et al. (2024, Archives of Environmental Contamination and Toxicology; 1999–2013) found PCB and DDE levels above effect thresholds in some nestlings but not significantly predicting productivity. - Contaminant trends. Among legacy contaminants in top predator fish, DDTs had the fastest decline (−13.2% a year, age-normalised, 1999–2014; Zhou et al., 2018). The herring gull egg sub-indicator is “Good” and “Unchanging” in all five lakes (SOGL 2025). - Factual corrections. EPA calls its 1972 DDT action a “cancellation order”, not a suspension (US EPA). For aldrin and dieldrin, EPA cancelled crop uses in 1974, not 1973; termite use continued until voluntary cancellation in 1987 (ATSDR, 2022). I could not verify from a primary source whether Canada’s 1969 action was a ban or a restriction; treat “banned in Canada” (Table 12.1) as unverified.

Verdict: held up. Organochlorine pesticide concentrations did fall quickly after the bans, and eagle reproduction recovered, gradually and later on the Great Lakes shorelines, where PCBs (not DDT) became the limiting factor. The regulatory details need correction.

Weight. Strong support for the lesson that source bans produce measurable ecological recovery when the agent’s main pathway is cut. It also shows a complication the chapter underplays: recovery from one agent can unmask the effect of a co-occurring, more persistent one.

Claim 9. At Love Canal, the company and “local, state and federal government officials” insisted that the leaking chemicals “were not the cause of high rates of birth defects, miscarriages, cancers and other health problems” (p. 127; Table 12.1, p. 132)#

Original claim. Lois Gibbs began investigating in spring 1978 and her association struggled “for more than two years” for relocation, opposed by Occidental Petroleum (owner of Hooker Chemical) and officials who denied the chemicals caused “high rates of birth defects, miscarriages, cancers”. President Carter’s October 1980 Emergency Declaration moved 900 families (p. 127). The chapter says 20,000 tonnes of waste were dumped “during the previous 20 years” (p. 127).

What happened since and what the record shows. - Site facts (US EPA). Hooker dumped over 21,000 tons of hazardous chemicals from 1942 to 1953, not over the 20 years before 1978. FEMA evacuated about 950 of more than 1,050 families. The site was capped and treated, and deleted from the National Priorities List in September 2004; EPA’s fifth five-year review found the remedy still protective. - Official action in 1978 was not simple denial. EPA Assistant Administrator Eckardt Beck’s contemporary account (EPA Journal, January 1979) records that in August 1978 the state health department was investigating “a disturbingly high rate of miscarriages” and birth defects, the governor announced the state would buy affected homes, and President Carter approved emergency aid on 7 August 1978, the first federal emergency funds for other than a natural disaster. By the end of August, 98 families had been evacuated. The chapter omits this first phase; the longer fight was over the wider neighbourhood. - Later health studies (New York State Department of Health cohort). - Cancer: Janerich et al. (1981, Science) found “no evidence for higher cancer rates” from registry data. The follow-up (Gensburg et al., 2009, EHP 117: 1265–1271; 5,052 former residents, 1979–1996) found raised bladder (SIR 1.44; 95% CI 0.91–2.16) and kidney (SIR 1.48; 0.76–2.58) cancer ratios, with intervals including 1; the role of the landfill was “unclear”. - Mortality: raised ratios for acute myocardial infarction and external causes; the landfill’s role “not clear” (Gensburg et al., 2009, EHP 117: 209–216). - Reproduction: Vianna and Polan (1984, Science) found a significant excess of low birth weight in the historic swale area during 1940–1953, when dumping took place. Austin et al. (2011, Environmental Research 111: 693–701) found significantly raised preterm birth for children born on the Love Canal before evacuation (SIR 1.40; 95% CI 1.01–1.90), raised low birth weight among mothers living closest (OR 4.68; 1.24–17.66; very small numbers), and non-significant excesses of malformations in boys and a lower male:female ratio; they urge caution. - Exposure: some chlorobenzenes were 2–14 times higher in the serum of those who had lived closest (Kielb et al., 2010).

Verdict: partly held up. Reproductive harm has some later support, and exposure from residential proximity is documented. The cancer excess has not been shown. The chapter’s account of uniform official denial omits the August 1978 state and federal emergency actions, and it misdates the dumping.

Weight. Use Love Canal as a case of community-led inquiry forcing institutional action and of the long time needed for epidemiology to catch up, not as proof of the full list of harms residents alleged. The chapter tells only the residents’ side (digest caveat), and later evidence partly supports both sides.

Claim 10. Integrated causal inference yields high certainty and makes the precautionary principle unnecessary for legacy contamination; complexity and post-normal framings have “not been inconvenient” to interests resisting remediation (p. 129; cf. the editors’ reading, p. 181)#

Original claim. Methods for integrating evidence (Fox, 1991) produced case studies linking injury to specific chemicals. The precautionary principle applies “in which there is a high degree of uncertainty”; causal inference is “designed to reduce uncertainty … thereby precluding the special need for applying the precautionary principle”. Post-normal science and multi-causal ecological framings have given “singular legitimacy” to complexity, which “has not been inconvenient to those interests reluctant to implement the costly remedial policies”; remedial action could itself be “precautions against the production of another generation of infants prenatally exposed” (pp. 129–130). The digest notes that the report’s editors instead present the case as research amplifying uncertainty and “paralysis by analysis” (pp. 172–173, 181); I did not reread those pages for this check.

What happened since. - Both framings were institutionalised. The 2012 Agreement adopted precaution as a principle and the “ecosystem approach” and “adaptive management” that Gilbertson saw as diversionary (Article 2). The IJC’s 2023 assessment lists all of them among the Agreement’s key achievements. - For persistent chemicals, international policy chose precaution over proof. The Stockholm Convention’s objective is “mindful of the precautionary approach”; in listing new chemicals, “lack of full scientific certainty shall not prevent the proposal from proceeding”, and the Conference of the Parties decides “in a precautionary manner” (Articles 1, 8(7)(a), 8(9)). This is the opposite of the chapter’s view that causal proof should be the basis for action, though it concerns preventing new releases rather than remediating legacy contamination. - Causal inference advanced, but did not end dispute. IARC’s 2013 Group 1 classification and the neurodevelopmental reviews (claim 3) show that evidence integration matured. Yet the heterogeneity objection (Goodman et al., 2010) shows integration itself can be contested, so “high certainty” did not become self-executing. - The author continued the argument. Gilbertson and Watterson (2007, Journal of Public Health Policy 28: 201–215) argued that interest groups had reframed the Agreement as being about “ecosystem integrity for the entire Great Lakes basin” and called this “an unwarranted diversion” given “continuing injury to health and property”. Gilbertson (2009, Journal of Environmental Science and Health C) used cerebral palsy hospitalisation rates in Canadian AOCs as a health index and argued that governments had failed to protect health from mercury, and that the reasons lie in “social, economic, and political contexts”. - What actually unlocked remediation (my inference from claims 6–7). Neither proof nor precaution was the proximate trigger. Funding came through dedicated legislation from 2002, organised around site-by-site impairment lists (beneficial use impairments), measurable delisting targets and economic-benefit arguments.

Verdict: contested. This is an interpretive argument. Later policy did not adopt its central move (proof in place of precaution): precaution was written into both the 2012 Agreement and the Stockholm Convention, alongside the complexity-oriented approaches the author criticised. His diagnosis that proof alone did not produce action is consistent with the later record, and matches the editors’ own conclusion (p. 181, per digest).

Weight. Useful as evidence of a framing contest (whoever controls the frame sets the burden of proof: digest insight 11), not as a settled lesson that causal inference makes precaution unnecessary. The later record suggests a third, more practical lesson: framing disputes mattered less than institutional design (dedicated funds, bounded targets, measurable endpoints).

Minor factual checks#

Implications for the section’s transferable insights#

Mapped to the digest’s numbered insights: - Strengthened. - (9) Persistent agents give quick early gains, then long legacy tails: fish, water and air data through 2025 show slowing declines, local reversals and PCBs still limiting fish consumption (claims 1–2). - (14) Long-term monitoring made success, slowdown and reversal visible (GLFMSP, IADN, herring gull eggs, SOGL); without these series the Lake Ontario rebound and the PCB-11 anomaly would not have been seen. - (4) Studying the wrong endpoint gives false reassurance: developmental outcomes after prenatal exposure proved the sensitive endpoint (claim 3). - Supported with qualification. - (2) Measurement capability bounds what can be warned about: congener-specific methods revealed an unregulated by-product PCB; PFAS surveillance in Great Lakes waters began only in 2008 (IJC, 2023). - (5) Incumbents contest evidence: post-2001 challenges came from an industry-funded methodological critique and an advocacy-group review (claim 3). This is consistent with the insight but does not show bad faith. - (8) Commitments erode: virtual elimination was qualified “as appropriate” in 2012 (claim 7). But funding commitments proved durable (claim 6). - (10) Proof is not enough; willingness to pay binds: money arrived after 2002, but through a new channel and economic framing (claim 6). - (12) Visible costs against invisible benefits erode support: governments answered by making benefits visible, in economic rather than health terms (claim 6). - (13) Protective advice creates distributional conflict: fish advice continues, with GLRI-funded outreach to subsistence and tribal consumers (2025 Progress Report of the Parties). - Weakened or contested. - The “support will wane” forecast (claim 6). - (11) Framing contests: real, but later policy adopted both frames at once (claim 10). - “Seldom less than 25 years”: illustrative only (claim 5). - New lessons from hindsight (not in the chapter). - Recovery from a persistent agent can reverse when the ecosystem changes around it (invasive species remobilising legacy stocks) (claim 2). - Residual floors can be set by continuing inadvertent or by-product sources, not only by legacy reservoirs (claim 2). - Remediation progressed where goals were bounded, local and measurable (AOC impairment lists and delisting), while basin-wide absolute goals (virtual elimination) stalled (claims 6–7). - Removing one agent can unmask a co-occurring, more persistent one (eagles: DDT, then PCBs) (claim 8).

Method and access notes#

• I read the digest and the chapter extract (working/text/chunks/LL1-12.txt) and no other project files. The 2013 annexes were read from the EEA-hosted PDF.
• The session’s web-search allowance ran out partway through; later sources were reached by direct fetches of known URLs (Europe PMC, Crossref, government sites). Some items therefore rest on fewer sources than I would like.
• Not verified in this check (leads for a follow-up):
• Canada’s 1969–1970 DDT action (ban or restriction) and the date when all Canadian registrations ended.
• A reported 2008 controversy over delayed release of a US federal (ATSDR) report on public health implications of hazardous substances in the 26 US Great Lakes AOCs, which draws on the Johnson/Hicks/De Rosa work the chapter cites (p. 130). I could not retrieve a primary source.
• The amount of Occidental’s cost-recovery settlements for Love Canal (the EPA site profile mentions only a 1994 consent decree).
• Pages 288 and 612 of the 2013 report, which its index lists under “Great Lakes, contamination”.
• Detroit wastewater treatment plant PCB loads since 2001 (the chapter’s USD 2 billion example, p. 130).
• Secondary or commentary sources are flagged in the text and list (Circle of Blue, InquireFirst, Water Canada, EHN, American Presidency Project, Wisconsin Public Radio).

Sources#

EEA - EEA (2013). Late lessons from early warnings: science, precaution, innovation, Annexes 1–3 (Annex 2 p. 713, Table A2.1 p. 702; Annex 3 pp. 717–). https://www.eea.europa.eu/en/analysis/publications/late-lessons-2/late-lessons-chapters/late-lessons-ii-annex-1/@@download/file (accessed 25 Sep 2026)

Monitoring and status (fish, water, air) - US and Canadian governments (2026). State of the Great Lakes 2025, sub-indicator “Contaminants in Fish Fillets” (uploaded June 2026). http://stateofgreatlakes.net/wp-content/uploads/2026/06/ContaminantsInFish.pdf - State of the Great Lakes, “Fish Consumption” indicator summary. https://stateofgreatlakes.net/indicators/fish-consumption/ (accessed 25 Sep 2026) - State of the Great Lakes, “Toxic Chemicals” indicator summary. https://stateofgreatlakes.net/indicators/chemicals/ (accessed 25 Sep 2026) - Binational.net (20 Jan 2026). Release of State of the Great Lakes 2025. https://binational.net/2026/01/20/state-of-the-great-lakes-2025/ - Zhou C, Pagano J, Crimmins BA, Hopke PK, Milligan MS, Murphy EW, Holsen TM (2018). PCBs and organochlorine pesticides concentration patterns and trends in top predator fish of Laurentian Great Lakes from 1999 to 2014. J Great Lakes Res 44(4): 716–724. https://pmc.ncbi.nlm.nih.gov/articles/PMC6178843 - Pagano JJ, Garner AJ (2024). Temporal trends of Great Lakes legacy contaminants: ecological and biological considerations applying the age-trend model. Environ Sci Technol 58(5): 2514–2527 (1 Feb 2024). https://pubs.acs.org/doi/10.1021/acs.est.3c09145 - Pagano JJ, Garner AJ, Hopke PK, Crimmins BS, Fernando S, Milligan MS, Holsen TM (2025). PCB concentrations and breakpoint trends across the waters of the Great Lakes by isotope-dilution HRMS. Sci Total Environ 958: 178024 (1 Jan 2025). https://doi.org/10.1016/j.scitotenv.2024.178024 - US EPA (1997). Water quality guidance for the Great Lakes system: revision of PCB criteria. Federal Register 62(48), 12 Mar 1997. https://www.govinfo.gov/content/pkg/FR-1997-03-12/html/97-6215.htm - US EPA. Great Lakes Integrated Atmospheric Deposition Network: trends and changes (January 2017 web snapshot). https://19january2017snapshot.epa.gov/great-lakes-monitoring/great-lakes-integrated-atmospheric-deposition-network-trends-and-changes_.html - Hites RA (2018). Atmospheric concentrations of PCB-11 near the Great Lakes have not decreased since 2004. Environ Sci Technol Lett 5(3): 131–135. https://pubs.acs.org/doi/10.1021/acs.estlett.8b00019 - Hites RA, Bidleman TF, Venier M (2022). Atmospheric concentrations of hexachlorobenzene and octachlorostyrene are uniform across the Great Lakes region and have not changed much in 25 years. Environ Sci Technol Lett 9(8): 660–665 (28 Jul 2022). https://doi.org/10.1021/acs.estlett.2c00444 - Environmental Health News (2 Apr 2018). Most PCBs are decreasing near the Great Lakes, but one’s not (secondary). https://www.ehn.org/one-pcb-remains-in-great-lakes-2554796287.html - Melymuk L, Blumenthal J, Sáňka O, et al. (2022). Persistent problem: global challenges to managing PCBs. Environ Sci Technol 56(12): 9029–9040 (June 2022). https://pubs.acs.org/doi/10.1021/acs.est.2c01204

Health evidence - Schantz SL, Widholm JJ, Rice DC (2003). Effects of PCB exposure on neuropsychological function in children. Environ Health Perspect 111(3): 357–376. https://ehp.niehs.nih.gov/doi/abs/10.1289/ehp.5461 - Ross G (2004). The public health implications of polychlorinated biphenyls (PCBs) in the environment. Ecotoxicol Environ Saf 59(3): 275–291. https://www.researchgate.net/publication/8327294_The_Public_Health_Implications_of_Polychlorinated_Biphenyls_PCBs_in_the_Environment - Stewart P, Fitzgerald S, Reihman J, et al. (2003). Prenatal PCB exposure, the corpus callosum, and response inhibition. Environ Health Perspect 111(13): 1670–1677. https://pmc.ncbi.nlm.nih.gov/articles/PMC1241692/ - Stewart PW, Lonky E, Reihman J, Pagano J, Gump BB, Darvill T (2008). The relationship between prenatal PCB exposure and intelligence (IQ) in 9-year-old children. Environ Health Perspect 116(10): 1416–1422 (Oct 2008). https://pmc.ncbi.nlm.nih.gov/articles/PMC2569105/ - Boucher O, Muckle G, Bastien CH (2009). Prenatal exposure to polychlorinated biphenyls: a neuropsychologic analysis. Environ Health Perspect 117(1): 7–16. https://pmc.ncbi.nlm.nih.gov/articles/PMC2627868/ - Grandjean P, Landrigan PJ (2006). Developmental neurotoxicity of industrial chemicals. Lancet 368: 2167–2178 (16 Dec 2006). https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(06)69665-7/fulltext - Goodman M, Squibb K, Youngstrom E, et al. (2010). Using systematic reviews and meta-analyses to support regulatory decision making for neurotoxicants: lessons learned from a case study of PCBs. Environ Health Perspect 118(6): 727–734 (1 Jun 2010; funded by Cefic-LRI). https://pmc.ncbi.nlm.nih.gov/articles/PMC2898846/ - Govarts E, Nieuwenhuijsen M, Schoeters G, et al. (2012). Birth weight and prenatal exposure to PCBs and DDE: a meta-analysis within 12 European birth cohorts. Environ Health Perspect 120(2): 162–170. Record via Europe PMC: https://europepmc.org/search?query=Govarts%20AND%20%22birth%20weight%22%20AND%20PCB - Lauby-Secretan B, Loomis D, Grosse Y, et al. (2013). Carcinogenicity of polychlorinated biphenyls and polybrominated biphenyls. Lancet Oncol (15 Mar 2013). https://www.thelancet.com/journals/lanonc/article/PIIS1470-2045(13)70104-9/abstract ; IARC Monograph 107 (2015/2016): https://publications.iarc.who.int/131 - Pessah IN, Lein PJ, Seegal RF, Sagiv SK (2019). Neurotoxicity of polychlorinated biphenyls and related organohalogens. Acta Neuropathol 138(3): 363–387. https://pmc.ncbi.nlm.nih.gov/articles/PMC6708608/ - Balalian AA, Stingone JA, Kahn LG, et al. (2024). Perinatal exposure to PCBs and child neurodevelopment: a comprehensive systematic review of outcomes and methodological approaches. Environ Res 252: 118912 (1 Jul 2024). https://doi.org/10.1016/j.envres.2024.118912

Agreements, implementation and assessments - Canada–United States (2012). Great Lakes Water Quality Agreement, as amended by the Protocol of 7 September 2012. https://www.ijc.org/sites/default/files/2018-07/GLWQA_2012.pdf - Binational.net. Annex 3: Chemicals of Mutual Concern (list of 2016 designations). https://binational.net/annexes/a3/ (accessed 25 Sep 2026) - Canada and United States (2026). 2025 Progress Report of the Parties (released 20 Jan 2026). https://binational.net/wp-content/uploads/2026/01/Progress-Report-of-the-Parties-2025.pdf - International Joint Commission (10 Dec 2020). Second Triennial Assessment of Progress on Great Lakes Water Quality. https://www.ijc.org/sites/default/files/2020-12/2020-TAP-Report-online.pdf - International Joint Commission (2023). Third Triennial Assessment of Progress on Great Lakes Water Quality. https://www.ijc.org/sites/default/files/Report%20-%202023%20Third%20Triennial%20Assessment%20of%20Progress%20on%20Great%20Lakes%20Water%20Quality_0.pdf - Stockholm Convention on Persistent Organic Pollutants, text as amended (2025 revision). https://chm.pops.int/Portals/0/download.aspx?e=UNEP-POPS-COP-CONVTEXT-2025.English.pdf - Gilbertson M, Watterson AE (2007). Diversionary reframing of the Great Lakes Water Quality Agreement. J Public Health Policy 28(2): 201–215. https://europepmc.org/article/MED/17585321 - Gilbertson M (2009). Index of congenital Minamata disease in Canadian areas of concern in the Great Lakes. J Environ Sci Health C 27(4): 246–275. https://europepmc.org/article/MED/19953398

Remediation, funding and Areas of Concern - US EPA. Great Lakes Legacy Act (last updated 18 Mar 2026). https://www.epa.gov/great-lakes-aocs/great-lakes-legacy-act - Great Lakes Restoration Initiative. Funding (annual appropriations FY2010–FY2026). https://www.glri.us/funding (accessed 25 Sep 2026) - US EPA. Restoring Great Lakes Areas of Concern (last updated 18 Aug 2026). https://www.epa.gov/great-lakes-aocs/restoring-great-lakes-areas-concern - US EPA. List of Great Lakes AOCs (last updated 31 Aug 2026). https://epa.gov/great-lakes-aocs/list-great-lakes-aocs - US EPA (1 Oct 2025). EPA, EGLE and partners celebrate removal of Muskegon Lake from list of most environmentally degraded areas in the Great Lakes. https://www.epa.gov/newsreleases/epa-egle-and-partners-celebrate-removal-muskegon-lake-list-most-environmentally - Canada Water Agency (28 Aug 2026). Ministers Dabrusin and Hajdu announce successful restoration of Nipigon Bay. https://www.canada.ca/en/canada-water-agency/news/2026/08/ministers-dabrusin-and-hajdu-announce-successful-restoration-of-nipigon-bay.html - International Joint Commission. Areas of Concern (includes July 2026 Black River delisting request). https://www.ijc.org/en/popular-issues/areas-concern (accessed 25 Sep 2026) - American Presidency Project (18 Feb 2022). The Bipartisan Infrastructure Law invests USD 1B in the Great Lakes region (compilation of EPA statements; secondary). https://www.presidency.ucsb.edu/documents/what-they-are-reading-the-states-the-bipartisan-infrastructure-law-invests-1b-the-great - Great Lakes Commission / University of Michigan RSQE (Sep 2018). Assessing the investment: the economic impact of the GLRI. https://www.glc.org/work/glri-econ ; report: https://www.epa.gov/sites/default/files/2019-12/documents/assessing_the_investment_the_economic_impact_of_the_great_lakes_restoration_initiative.pdf - Water Canada (9 Mar 2022). All contaminated sediment removed or capped at Randle Reef, Hamilton Harbour (trade press; secondary). https://www.watercanada.net/contaminated-sediment-removed-randle-reef-hamilton-harbour/ - InquireFirst (2017). Trump’s 97 percent cut for Great Lakes? That was the good news (secondary). https://inquirefirst.org/greatlakes-2/ - Schneider K, Circle of Blue (13 Jun 2025). Trump wants to wreck progress on restoring Great Lakes (commentary; secondary). https://www.circleofblue.org/2025/great-lakes/trump-wants-to-wreck-progress-on-restoring-great-lakes/ - Wisconsin Public Radio (15 May 2026). Steep cuts proposed for EPA draw bipartisan pushback, including from Wisconsin (secondary). https://www.wpr.org/news/steep-cuts-proposed-epa-draw-bipartisan-pushback-including-wisconsin - Sen. Elissa Slotkin (21 Jul 2026). Slotkin and Peters bipartisan bill to extend federal funding and protections for the Great Lakes advances in the Senate. https://www.slotkin.senate.gov/2026/07/21/slotkin-and-peters-bipartisan-bill-to-extend-federal-funding-and-protections-for-the-great-lakes-advances-in-the-senate/

Pesticides and wildlife - US EPA. DDT: a brief history and status. https://www.epa.gov/ingredients-used-pesticide-products/ddt-brief-history-and-status (accessed 25 Sep 2026) - ATSDR (June 2022). Toxicological profile for aldrin/dieldrin. https://www.atsdr.cdc.gov/toxprofiles/tp1.pdf - US Fish and Wildlife Service. Bald eagle (Haliaeetus leucocephalus) species page. https://www.fws.gov/species/bald-eagle-haliaeetus-leucocephalus (accessed 25 Sep 2026) - Bowerman WW, Giesy JP, Best DA, Kramer VJ (1995). Review of factors affecting productivity of Great Lakes bald eagles (title as indexed not re-checked). Environ Health Perspect 103 (Suppl). Record via Europe PMC: https://europepmc.org/search?query=Bowerman%20bald%20eagle%20Great%20Lakes%201995 - Bowerman WW, Best DA, Giesy JP, et al. (2003). Study of PCBs and p,p’-DDE in blood of nestling bald eagles from ten Great Lakes subpopulations, 1997–2000, and regional productivity (title not re-checked). Environ Toxicol Chem. Record via Europe PMC: https://europepmc.org/search?query=Bowerman%20nestling%20bald%20eagles%20PCB%20DDE%202003 - Best DA, Elliott KH, Bowerman WW, et al. (2010). Productivity, embryo and eggshell characteristics, and contaminants in bald eagles from the Great Lakes, USA, 1986 to 2000. Environ Toxicol Chem 29(7): 1581–1592. https://doi.org/10.1002/etc.195 - Eakin CJ, Williams L, Moore J, et al. (2024). Study of contaminant exposure (PCBs, DDE) and productivity in bald eagle nestlings above and below dams on Great Lakes tributaries, 1999–2013 (title not re-checked). Arch Environ Contam Toxicol. Record via Europe PMC: https://europepmc.org/search?query=Eakin%20bald%20eagle%20dams%20Great%20Lakes

Love Canal - US EPA. Love Canal Superfund site: background and cleanup. https://cumulis.epa.gov/supercpad/SiteProfiles/index.cfm?fuseaction=second.cleanup&id=0201290 (accessed 25 Sep 2026) - Beck EC (January 1979). The Love Canal tragedy. EPA Journal. https://www.epa.gov/archive/epa/aboutepa/love-canal-tragedy.html - Janerich DT, Burnett WS, Feck G, et al. (1981). Cancer incidence in the Love Canal area. Science 212: 1404–1407. Record via Europe PMC: https://europepmc.org/search?query=Janerich%20Love%20Canal%201981 - Vianna NJ, Polan AK (1984). Incidence of low birth weight among Love Canal residents. Science 226: 1217–1219. Record via Europe PMC: https://europepmc.org/search?query=Vianna%20Polan%20Love%20Canal - Gensburg LJ, Pantea C, Fitzgerald E, et al. (2009). Mortality among former Love Canal residents. Environ Health Perspect 117: 209–216. https://europepmc.org/article/MED/19270790 - Gensburg LJ, Pantea C, Kielb C, et al. (2009). Cancer incidence among former Love Canal residents. Environ Health Perspect 117: 1265–1271. https://europepmc.org/article/MED/19672407 - Kielb CL, Pantea CI, Gensburg LJ, et al. (2010). Concentrations of selected organochlorines and chlorobenzenes in the serum of former Love Canal residents. Environ Res 110: 220–225. https://europepmc.org/article/MED/20117765 - Austin AA, Fitzgerald EF, Pantea CI, et al. (2011). Reproductive outcomes among former Love Canal residents, Niagara Falls, New York. Environ Res 111: 693–701. https://europepmc.org/article/MED/21555122