On Eating Contaminated Ducks

This isn’t pearl-clutching, or telling you to eat fewer ducks, or do anything at all. I’m just a curious person who wanted to look into this more after a recent blurb in the Delta Waterfowl magazine. Despite the alarming raw numbers, the average duck you eat probably isn’t all that much worse than other meat on your table.

Based on Dayan et al. (2025), Science of the Total Environment; the New Mexico Environment Department’s 2025 Holloman Lake PFAS Ecology Report; and two state waterfowl breast-muscle studies — Eifert, Gorski, Magee & Strom (2025), Environments (Green Bay, WI) and LaSharr & Carstensen (2024), MN DNR (Minnesota). Reading time ~18 min. Every chart on this page is live — toggle, filter, and poke at the numbers as you read.

For the past few years I’ve paid increasing attention to duck activity in marginal spaces around my home — wastewater plants, landfill waste ponds, and similar environments — and have wondered exactly what they were eating, how that may affect their contamination rates & how much of that contamination makes its way onto my plate. Fortunately, a number of researchers across the country have answered that question for me and the results, while surprising, might suggest something much larger about the health risks of our current food system than perhaps anything else.

The gist on ducks: it’s usually about where the duck was shot. For the overwhelming majority of birds — pulled out of natural marshes and ag areas — waterfowl is a low-risk protein that could have contamination further reduced by skinning and cooking. But a small number of birds are significantly contaminated from the surrounding environment, and for those, no amount of trimming, cooking, or species-picking will reduce the contamination levels to a safe threshold for human consumption.

Regime 1The baseline bird

In January 2025, a team led by Cornell’s Wildlife Health Lab published the best modern baseline we have. They collected 108 wild ducks and geese from hunters across New York, Pennsylvania, Connecticut, and New Jersey and ran the breast meat of five species — Canada goose, wood duck, American black duck, green-winged teal, and mallard — through a battery of tests for mercury, dioxins, PCBs, banned pesticides, and PFAS. Then they ran two EPA-standard risk assessments: a non-cancer hazard index (anything under 1.0 is considered safe) and an excess lifetime cancer risk, on raw meat with the skin and fat left on — the most contaminated way you could prepare a bird.

Three things fall out of their data, and they’re the practical core of everyday duck hunting.

Species is the biggest lever. Canada geese and wood ducks — grazers and perching ducks that spend less time nose-down in contaminated sediment — carried the lowest burdens and are clean enough to eat weekly. Black ducks and green-winged teal are about a once-a-month bird. Mallards were the dirtiest of the five, closer to twice a month at most. My assumption is that this is due to differences in diet and thus contaminant uptake rates. You can see the whole study as one grid — every species against every eating frequency, shaded by regulatory band — in the master grid just below. Flip to any hotspot and the story changes — all of those birds are wildly contaminated.

Preparation is the second lever, and it’s nearly free. The contaminants that drive the cancer numbers — dioxins, PCBs, legacy pesticides — are fat-soluble. They ride in the skin and yellow subcutaneous fat, not the lean red muscle. Skinning and trimming removes roughly half of them; cooking so the fat renders and drips removes about a third; doing both lands around a two-thirds reduction, which effectively doubles how many meals a month you can safely eat. (Mercury is the exception — it binds to muscle protein, so trimming and cooking don’t touch it — but mercury doesn’t drive cancer risk.) The preparation chart turns this into meals-per-month you actually gain.

Keep perspective. Put a once-a-month serving of each bird on the same ruler as everything else people eat, and the ducks are unremarkable. Every one of them sits below a daily glass of wine or a daily strip of bacon, and the cleanest land near a monthly serving of store-bought salmon — see the risk ladder. Even eating the dirtiest of these birds every single day for life raises your lifetime cancer chance by under half a percent, against the ~40% baseline every human already carries. And the cancer risk is shared three ways: two legacy pollutants — dioxins and PCBs — plus PFOS, once you apply the slope factor the study couldn’t (more on that just below). For the cleanest birds, goose and wood duck, PFOS is actually the single biggest contributor. The catch is that trimming and cooking strip the fat-soluble dioxins and PCBs but not PFOS, which rides in the lean muscle and stays put. That composition is the drivers chart.

The fear, defused. The scariest way to state a risk is as a multiplier — but to double your lifetime cancer risk, an exposure would need a multiplier of 2.0, and a lifetime of eating even the dirtiest of these birds every single day comes to about 1.005. Picture a stadium of 10,000 people: about 4,000 develop cancer no matter what they eat. At an everyday-for-life pace the cleanest species add roughly one extra case, the dirtiest a dozen or two — dwarfed by the ~120–130 that a daily drink or daily bacon habit piles on; at a realistic once-a-month pace the birds’ extra cases nearly vanish. Fill the stadium yourself.

If every duck lived in Regime 1, the charts below would be the end of the article, and the cheat sheet — safe meals per month, by species and preparation — would be all you needed to print. Read them, play with them — then comes the twist.

Everyday foods that outrank a duck

The ladder puts a monthly duck against a few anchors; here it is as a plain list. Each item below is a common food or drink that mainstream cancer science treats as riskier than eating an ordinary (Regime 1) duck once a month — with one honest catch: these are things people eat every day, while the duck is monthly. The odds are added lifetime cancer risk on the same “1 in N” scale, and — exactly as the disclaimer up top warns — they’re organ-specific conversions (colorectal for meat; several sites for alcohol), not one universal cancer number.

To put waterfowl on the exact same daily-for-life basis as the foods above, here is the whole span — from the cleanest ordinary bird you could eat every day to the single worst bird from the country’s hottest PFAS site:

For scale: a once-a-month ordinary duck runs from about 1 in 130,000 (Canada goose) to 1 in 12,000 (mallard) on this exact scale — so even the dirtiest Regime 1 bird, eaten monthly, sits hundreds to thousands of times below a daily bacon, drink, or red-meat habit (and those figures are skin-on and raw; trim and cook and the duck improves further). None of this makes bacon a crisis — it makes the point that a clean wild duck is one of the lower-risk proteins on the table. The one place this flips is a PFAS hotspot, which is the rest of this piece.

The other axis: what “non-cancer risk” actually is

Cancer gets the headlines, but for these birds it’s usually not the binding constraint. The study ran a second, separate assessment — the hazard index — and it’s the one that actually sets the “safe meals per month” in the cheat sheet below.

Here’s the idea. For each contaminant, regulators publish a reference dose: the amount you could eat every day for a lifetime with no expected harm. Divide your actual intake by that reference dose and you get a hazard quotient; add the quotients across every contaminant in the meat and you get the hazard index (HI). An HI under 1.0 means you’re inside the safety envelope; over 1.0 means you’ve crossed it. Crucially, this is a margin-of-safety flag, not a probability — an HI of 5 doesn’t mean a 5-in-something chance of anything; it means five times the intake considered acceptable, which is a signal to eat less, not a diagnosis.

What are the actual health effects behind that number? They’re specific, and they’re mostly not cancer — which is exactly why the two axes don’t collapse into one:

• Methylmercury — the nervous system. Developmental neurotoxicity is the driving concern: it crosses the placenta and affects fetal and early-childhood brain development, which is why advisories single out people who are or may become pregnant, and young children. In adults, high chronic intake brings neurological effects. Mercury binds to muscle protein, so trimming and cooking do not remove it — and in Western birds it’s often the effect that sets the limit.
• PFOS & other PFAS — immune, liver, hormonal, developmental. EPA’s critical effect is suppression of the immune response (measurably lower antibody production after childhood vaccines). Add elevated cholesterol, liver-enzyme and thyroid changes, kidney and testicular signals, and reduced infant birth weight. The reference dose is extraordinarily low (PFOS: 1×10⁻⁷ mg/kg/day), so even modest concentrations push the hazard index up — and because PFOS rides in lean muscle, preparation can’t touch it.
• Dioxins & dioxin-like PCBs — developmental, reproductive, endocrine, and immune effects (chloracne at high doses); scored together as toxic-equivalents (TCDD-TEQ). These are fat-soluble, so skinning, trimming, and cooking genuinely reduce them.
• Legacy organochlorine pesticides — liver, neurological, and endocrine effects; also fat-soluble and reducible by preparation.

Two practical takeaways. First, because the hazard index reaches 1.0 before the cancer risk reaches EPA’s action threshold for these birds, the non-cancer number is what governs the cheat sheet — so the “safe meals” figures already bake in mercury and PFOS, not just cancer. Second, the people most affected by the non-cancer endpoints — pregnant and nursing women, and children — should treat every number here as a ceiling to stay well under, not a target.

The plot twistA slope factor the study couldn’t use

Buried in the Dayan paper’s limitations is a sentence that turns out to matter enormously:

“our estimation of cancer risk from PFAS may be especially low, as [cancer slope factor] values were not available for any PFAS except PFOA at the time of writing… should [health guidance values] become available after this writing… our PFAS data may warrant updated analysis.”

First, a quick disambiguation, because it’s the crux of the whole thing. PFAS is the family — thousands of “forever chemicals.” PFOS is its single most notorious member. They aren’t two different things being compared; PFOS is a PFAS, and in waterfowl it’s the one that matters. The Dayan team detected PFOS in 95.3% of their New York birds — the most of any PFAS they measured — while PFOA, the one compound that happened to have a cancer slope factor back then, turned up in only about a quarter. The New Mexico report tells the same story from the other extreme: across its birds, PFOS is the dominant congener. So both studies very much “said something about PFOS” — it’s the needle every measurement lands on.

To turn a concentration into a cancer number you need a slope factor — a dose-to-risk conversion — and when the paper was written, PFOS had none; only PFOA did. So the compound that dominates the birds drove the study’s non-cancer hazard but contributed exactly zero to its cancer estimate. That’s the gap the authors flagged.

In April 2024, the EPA finalized exactly the missing piece: an oral cancer slope factor for PFOS (39.5 per mg/kg/day) and a reference dose (1×10⁻⁷ mg/kg/day). The authors told us what to do when that happened — so let’s do it twice: first to their own ordinary birds, then to the place that makes it unmissable.

Start with the ordinary birds. Dayan reports PFOS right there in Table 2 — in breast muscle, the part you actually eat — averaging 3.08 ng/g across all 107 birds, ranging by species from about 1.0 in geese and wood ducks to 6.0 in mallards. Run those through the EPA slope factor with the study’s identical intake model (a 227-gram meal, an 80-kilogram adult, a lifetime of eating) and the “rounding error” evaporates: at a twice-a-month habit PFOS adds roughly a 1-in-44,000 lifetime cancer risk for an average bird — larger than the entire dioxin-and-PCB total for the cleanest species. PFOS doesn’t just belong in the background cancer math; for goose and wood duck it’s the single biggest piece of it, and because it sits in muscle rather than fat it’s the one driver trimming and cooking can’t reduce — the driver flip below splits each bird’s cancer risk into the part you can trim away and the locked-in PFOS you can’t. That alone reframes Regime 1. Now watch what the same slope factor does somewhere PFOS isn’t measured in single digits.

Regime 2The hotspot

In the desert outside Alamogordo, New Mexico, sits Holloman Lake — a body of water fed for decades by runoff from firefighting foam (AFFF) at the adjacent Air Force base. In 2025 the New Mexico Environment Department published an ecological survey of the lake, and the waterfowl numbers are unlike anything in the Dayan study. Not a little higher. Thousands of times higher.

Start with raw concentration, before any risk math. On the contamination gradient below, grocery chicken, pork, and beef form a near-zero floor around a quarter to half a nanogram of PFOS per gram — 95–99% of samples are outright non-detect (USDA’s national residue program, 2019–2023). Locally caught wild fish sit a step up, a few to ~16 ng/g. New Mexico’s statewide waterfowl — birds nowhere near the lake — already run 100 to 1,000 ng/g. And the Holloman birds themselves? Green-winged teal liver at 20,000 ng/g. American wigeon at 38,000. That’s four to five orders of magnitude above the grocery floor.

Now the risk math the original study couldn’t do. Modeled on the same excess-lifetime-cancer axis as the ordinary birds, a Holloman bird is off the top of the chart — see the hotspot dropped onto the same ruler below:

• Eating a Holloman green-winged teal just once a year, for life, works out to roughly 1 in 810 added lifetime cancer risk — about 40 times the risk of eating an ordinary teal every month, and squarely in the EPA’s “unacceptable” band all by itself.
• At once a month, the number blows past 1 in 100 — so far past the “action” threshold that the linear model behind it stops being a literal probability and simply reads do not eat.
• The non-cancer hazard index at any regular pace runs into the thousands (a hazard index over 1 is the concern threshold). That, more than any cancer figure, is what puts these birds off-limits.

One assumption, stated plainly, because it’s the load-bearing one. Holloman’s headline numbers are liver, and hunters eat breast. PFOS concentrates in liver, blood, and kidney — not fat or muscle — so breast meat runs lower. I modeled muscle at 0.2× the measured liver level (a stated assumption; a 0.1–0.3× range is computed too), and I let PFOS carry the cancer number because it’s the one that’s measured, dominant, and now has a slope factor. Even at the optimistic end of that range, the birds are far past any threshold. And of note: because PFOS lives in the lean meat rather than the fat, the trim-and-cook trick that can reduce contam in Regime 1 does essentially nothing here.

And it isn’t just a modeling assumption — someone measured the muscle. A thousand miles away in Green Bay, Wisconsin, state biologists did the experiment I had to approximate: they sampled PFOS directly in mallard breast muscle — the exact cut a hunter eats — from 2022 through 2024 (the work Sean Strom of the Wisconsin DNR walked through on a 2025 Delta Waterfowl podcast, and published as Eifert, Gorski, Magee & Strom, 2025). At an unimpacted reference site the birds were essentially clean — only two had any detectable PFOS at all. In Green Bay, 67% to over 90% of mallards were contaminated, running up to about 480 ng/g in the lower bay. That was high enough that in September 2025 Wisconsin issued its own waterfowl advisory: do not eat mallards from lower Green Bay, and one meal a month for mallards (one a week for wood ducks) in the upper bay. Different flyway (Mississippi, not Central), different source (industrial legacy, not a single foam pit), and — crucially — muscle, not liver: hard confirmation that hotspot PFOS reaches the meat on the plate, and that no amount of trimming rescues you. Those Green Bay points, and Wisconsin’s 10- and 40-ng/g muscle advisory thresholds, are marked on the contamination gradient — and the background Northeastern ducks sit just below the one-meal-a-month line, which is exactly why they’re a Regime 1 story.

Minnesota found the same thing — plus a twist about species. A Minnesota DNR pilot (LaSharr & Carstensen, 2024) sampled mallards and Canada geese at a long-contaminated lake (Lake Elmo, downstream of decades of PFAS manufacturing waste) and at a clean control. Control mallard breast muscle was essentially spotless — a median of 0.4 ng/g PFOS. At Lake Elmo, mallard breast muscle ran a median 948 ng/g (up to 1,400), 360 to 7,000 times the control. But the Canada geese at the very same lake sat far lower — a median 31 ng/g. Diet is why: mallards dabble in contaminated water and sediment, geese mostly graze grass. It’s the whole thesis in one lake — the place, the species, and the tissue seem to matter the most in regard to contam rates. And there’s a bonus buried in that paper: it reports the Holloman birds’ directly-measured muscle — mean 1,903 ng/g against a liver mean of 9,154 (Witt 2024), a ratio of ~0.21 that lands almost exactly on the 0.2× breast-to-liver figure I had to assume above. The load-bearing assumption checks out against real data.

That reversal is the single most important idea in this piece, and it has its own picture just below: how much of the risk you can trim away. For a Northeastern study bird, about three-quarters of the contaminant-cancer risk is fat-soluble — the dioxins and PCBs that skinning, trimming, and cooking cut into — leaving PFOS as the locked-in remainder. For a hotspot bird there is nothing to trim away: the cancer risk is essentially all PFOS, and it’s sitting in the part of the animal you actually eat.

Why it’s the place, not the duck

None of this is really about ducks; it’s about their habitat. It’s a story about a few specific pieces of ground being poisoned, and migratory birds acting as the messenger. Holloman is closed to hunting — but it’s a migratory stopover, so those same birds are legally harvested hundreds of miles away, carrying their body burden with them. New Mexico’s report explicitly calls for expanded statewide testing of waterfowl and large game.

And this isn’t hypothetical or confined to one lake. Wildlife agencies have already issued wild-game “do not eat” advisories in state after state — see the advisory tracker below — and they trace to two recurring point sources:

• Military firefighting foam (AFFF): Holloman Lake (NM); Oscoda Township, Michigan, home to the first “do not eat” deer advisory in the U.S. (2018); Marinette, Wisconsin.
• PFAS-laden sludge spread as farm fertilizer: Fairfield, Maine, where a ~25-square-mile deer advisory traces to biosolid fertilizer (now banned in the state).
• Industrial legacy in the water: the Green Bay, Wisconsin duck advisory above — the one hotspot on this list where the contaminant was measured straight from the breast muscle, not the liver.

Michigan and New Hampshire go further and warn against eating the organs — liver and kidney — of any wild game, precisely because that’s where PFAS concentrates.

So which regime is your duck in?

For most hunters, most of the time, the honest answer is Regime 1: your birds came from ordinary wetlands, and the cheat sheet is the whole game — pick a species, pick a preparation, eat within the number, skin and trim and cook, and enjoy one of the more honest proteins you can put on a plate. Grocery meat, if you ever want the comparison, is a genuinely near-zero PFAS floor.

Treat a bird as Regime 2 — and check your state’s advisories before eating it at all — if it came from:

• Water near a military base, airport, or fire-training area (AFFF sources).
• Land downstream of industrial discharge or fields fertilized with municipal biosolids/sludge.
• Any water under an existing PFAS advisory for fish or game.

When in doubt in a suspect area, two rules travel well: skip the organs entirely (PFAS loves liver and kidney), and remember that trimming and cooking — your best friends in Regime 1 — do not rescue you from PFOS in Regime 2.

Every flyway has its own hotspots. Holloman is in the Central Flyway, but this isn’t regional bad luck — it’s infrastructure. North America drains into four great migration corridors — the Pacific, Central, Mississippi, and Atlantic flyways — and every one is dotted with the same point source that poisoned Holloman: military airfields that trained for decades with AFFF firefighting foam. The Pentagon is now assessing 723 installations nationwide for PFAS; 352 have public coordinates, and mapped by flyway they number in the hundreds in each corridor — roughly 147 in the Pacific, 99 in the Central, 166 in the Mississippi, and 241 in the Atlantic (the remainder in Alaska, Hawaii, and the territories). The Environmental Working Group separately maps ~720 military AFFF sites among more than 9,700 contaminated U.S. locations. And a migrating bird doesn’t read fence lines.

A note for the Pacific Flyway. Both of these studies are Eastern and Southwestern; no one has published an equivalent hunter-focused study for the ~60% of the flyway’s birds that winter in California’s Central Valley and Bay-Delta. Out West, mercury is the wildcard — a Gold Rush legacy that the region’s flooded rice fields and managed wetlands turn into methylmercury, driving the non-cancer advisory (and untouched by trimming). PFAS in Western waterfowl are essentially unstudied but turn up in Bay fish, so assume presence and watch for point sources exactly as above. See the flyway forecast below and your state’s OEHHA advisories.

The bottom line — or what I’ll be doing

• Most ducks are a which-species-how-often question, and a good answer. Lean on what are likely clean species, skin-trim-cook, and eat within the cheat sheet.
• A few ducks are a where-was-it-shot question, and the answer is no. Point-source PFAS — foam or sludge — produces birds thousands of times over the line, where species and preparation are irrelevant.
• Tissue is the tell. Legacy contaminants hide in fat (trim them away); PFOS hides in lean meat and organs (you can’t). Skip the giblets from any suspect bird.
• Know your water. The most protective thing you can do isn’t picking a species — it’s knowing whether the marsh you hunt sits downstream of a foam site, an outfall, or a sludge-fertilized field.