Special Feature · Research Report

The Salmon Crisis Hiding in Plain Sight

Every year, Lake Pend Oreille is drawn down eleven and a half feet — partly to help recover endangered salmon hundreds of miles downstream. That is a good and noble goal. We looked at the available information on what’s actually happening to those salmon. What we found should change the conversation.

Correction · August 17, 2026

We have withdrawn one finding from this report.

We reported that juvenile salmon survival through the Lower Monumental–McNary reach collapsed after 2019 while an upstream control reach held steady, and connected that pattern to the contamination record at those dams. We queried the underlying data directly through the University of Washington’s DART system and could not reproduce the result. The reach we identified as impaired returned a higher survival estimate than the control — the opposite of what we published.

The reason is measurement. Reach survival is not counted; it is estimated from PIT‑tag detections. Below Lower Granite Dam, the federal monitoring network detects between one and three percent of tagged fish — 0.7 percent at Lower Monumental, 0.8 percent at Ice Harbor, 3.2 percent at McNary. Estimates built on that few detections carry margins of error of ±30 to ±70 percentage points, wide enough that no comparison between reaches is possible in either direction.

A quotation previously attributed on this page to a researcher at Columbia Basin Research has also been removed. It did not accurately represent his views, and we apologize to him for it.

The rest of the report is unaffected: the comparison across eleven river systems, the four threats common to declining runs with and without dams, the documented contamination record at the eight federal projects, the compounding‑effects model, the cross‑species ESA analysis, the Aleutian monitoring proposal, and the 29 recommended actions.

What the check revealed matters more than what it removed: nobody can presently say what is happening to juvenile salmon in the lower Snake — in either direction. Not this organization, not the dam operators, not the parties to the ongoing litigation, not its critics. Everyone arguing about survival through that stretch of river is arguing from estimates whose margin of error exceeds the range of possible answers. That is a finding about instruments rather than about dams, and it is measured rather than asserted.

Confirmation runs are underway. The workbook below has been corrected, with withdrawn material struck through rather than deleted. If you believe we have this wrong, please tell us — ralph@sandpointwaterfront.com. That is how we caught it.

The Lake Pend Oreille Alliance has spent years documenting the costs of that annual drawdown — to recreation, scenic beauty, lakeside infrastructure, the Spokane Valley Rathdrum Prairie Aquifer, and the ecological function of the lake itself. This report turns the question around. If our water is being released to recover salmon, what does the federal data say is actually killing those salmon? And is the sacrifice working?

The salmon debate has been stuck for decades. Dam advocates point to the ocean. Dam opponents point to the dams. Fisheries agencies produce stock‑specific recovery plans. Bycatch critics point to the trawl fleet. Everyone has a piece of the picture. Nobody is looking at the whole thing at once.

This report does something that hasn’t been done: it puts the pieces in the same room. Across 11 river systems — three dammed, eight undammed, spanning Idaho to Alaska — we asked a simple question. What’s declining everywhere, and what’s declining only where dams are present?

The answer is uncomfortable for everyone.

What’s Different

Four threats appear in every declining system — with or without dams.

Ocean conditions, climate change, commercial bycatch, and predation show up in every collapsing run we examined. Dams add severity — the Columbia River’s cumulative stress score is nearly three times higher than the Kenai River’s — but they aren’t the common denominator. Removing every dam tomorrow wouldn’t stop the decline in Alaska, where there are no dams and runs are collapsing anyway.

But the analysis also shows something that should be uncomfortable for dam operators. Juvenile survival through the dammed Columbia is roughly comparable to survival through the undammed Fraser River over similar distances. The dams aren’t the unique killer everyone assumes. So what is?

The Juvenile Gauntlet

A 750‑mile, 8‑dam obstacle course — and a contamination record nobody has overlaid.

A juvenile salmon born in an Idaho hatchery faces a 750‑mile, 8‑dam obstacle course to reach the ocean. At each dam, it’s pushed through turbines, bypass systems, or over spillways. Between dams, it swims through slackwater reservoirs that are warmer, slower, and more predator‑friendly than the free‑flowing rivers these fish evolved in.

The standard analysis tracks survival through this gauntlet using PIT tags — tiny transponders injected into each fish that are detected at each dam. The data is extraordinary: 25 years of reach‑by‑reach survival estimates for millions of fish, publicly available through the DART system at the University of Washington.

But here’s what nobody has done with that data: overlaid it with the contamination record.

Figure 1
Three contamination sources converge in a single reach

Schematic of dam corridor showing contamination sources

Every one of the eight federal dams on the Snake and Columbia has a documented history of oil spills, PCB contamination, or hazardous waste. The Hanford Nuclear Site — the most contaminated nuclear facility in the Western Hemisphere — sits on 51 miles of Columbia River directly upstream of McNary Dam, with radioactive groundwater still actively seeping into the river as of 2024. The contamination data exists. The survival data exists. They have never been systematically compared.

Ice Harbor Dam had a 1,500‑gallon PCB oil spill in 2012 that exceeded federal water quality standards by 14,000,000 percent. Lower Monumental spilled 1,600 gallons of oil in 2017. Little Goose leaked oil for 90 days before anyone noticed in 2022. Bonneville Dam’s Bradford Island is an EPA Superfund site with the highest PCB concentrations in fish ever recorded in the region.

We extracted actual PIT tag survival data from DART for three Idaho hatchery populations across 25 years, comparing the reach containing Ice Harbor and Lower Monumental dams against an upstream control reach with no major spill events. Then we pulled forebay temperature data for both dams to test whether temperature explains the patterns.

The Finding We Withdrew

An open question — and why we withdrew our answer.

Figure 2 has been removed. It charted a divergence in juvenile survival estimates between the two reaches after 2019. That divergence did not reproduce on direct query of the DART data, and the underlying estimates carry margins of error wider than the effect the chart depicted. Retaining the figure would keep a withdrawn result in circulation.

This reach is unique in the system: it’s the only place where fish pass through dam‑site PCB and oil contamination and then immediately enter the Hanford nuclear plume before reaching the next detection point. Three contamination sources in one reach. No other reach in the system has this combination.

A pull quote has been removed from this position. It attributed to a researcher at the University of Washington’s Columbia Basin Research a statement he did not make and a view he did not hold. His actual contribution was a caution — that reach survival is estimated from PIT‑tag detections, and that a decline in detection cannot presently be distinguished from a decline in survival below Lower Granite. That caution is what led us to re‑examine and withdraw the finding above, and we are grateful for it.

We reported that this established an anomaly specific to the reach. It did not. On direct query the result reversed, and the estimates proved too imprecise to support any comparison between reaches. What the exercise did establish is narrower and, we think, more useful: the federal detection network below Lower Granite cannot presently resolve reach‑level survival at all. The standard salmon debate — dams versus ocean — has been conducted for two decades on data that cannot settle it.

Why It Matters

The blind spot is structural — and the data to close it already exists.

If you work in salmon recovery, you’ve probably seen hundreds of reports, studies, and proposals. Most of them operate within established frameworks: ESA consultations, BiOps, spill operations, hatchery reform, habitat restoration. Those frameworks matter. But they share a blind spot: they treat the river corridor as a passage problem (how to move fish past dams efficiently) rather than a water quality problem (what’s in the water the fish are swimming through).

The workbook is not advocacy. It is an integration of publicly available data that has never been assembled in one place and asked to tell a coherent story. The data speaks for itself. What it says is that the salmon crisis is bigger than dams, more complicated than any single threat, and that the instruments we use to measure it may not be good enough to see what is happening in the middle of the river.

Read the full analysis

21 tabs of sourced findings, raw data extracts, and an action plan organized by timeline and authority — including a correction notice documenting the withdrawn material.

Download the Workbook (.xlsx)
Pacific Salmon Decline Analysis (Enhanced) · Lake Pend Oreille Alliance · May 2026

About this work

This analysis was compiled by the Lake Pend Oreille Alliance, a 501(c)(3) nonprofit based in Sandpoint, Idaho. The reach‑level survival data is drawn from the Columbia Basin DART system (cbr.washington.edu/dart) maintained by the University of Washington. Contamination records are sourced from EPA, U.S. Army Corps of Engineers, and Department of Energy public filings. All data and citations are reproducible from public federal sources.

Corrections, additions, and questions are welcome. The workbook is a working document, and the field benefits from challenge. Contact: info@savependoreille.org

Share →