Science & Environment

About the data: How we analyzed salmon and steelhead survival

By Irena Hwang (ProPublica)
May 24, 2022 12 p.m.
The salmon viewing area at the Bonneville Lock and Dam, August 2021.

The salmon viewing area at the Bonneville Lock and Dam, August 2021.

Kristyna Wentz-Graff / OPB

Oregon Public Broadcasting and ProPublica obtained data from Columbia Basin Research at the University of Washington describing fish in several salmon and steelhead trout populations that were embedded with electronic tags. Tagged fish can be detected by special technology, often at dams, and tag data can provide a window into fish migration and survival. Our approach took a basin-wide view of the hatchery program to create a meaningful, accessible and representative picture of hatchery efforts to support vulnerable salmon and steelhead populations in the region.


Related: The US has spent more than $2B on a plan to save salmon. The fish are vanishing anyway.

We focused our analysis on eight fish populations, all of them Columbia River Basin stocks that are highly vulnerable and monitored by the National Oceanic and Atmospheric Administration with the goal of restoring populations to healthy and harvestable levels. Focusing on fish populations that originated in the middle and upper Columbia and Snake rivers had two benefits: First, these were areas that were significantly impacted by the building of hydropower dams in the basin, and second, these were regions for which data was available. The upper Columbia River spring Chinook and Snake River sockeye populations are listed as endangered under the 1973 Endangered Species Act. The Snake River fall Chinook, Snake River spring/summer Chinook, Snake River steelhead and upper Columbia River steelhead populations are listed as threatened. The Mid-Columbia Coho Restoration Program includes all coho released in the Wenatchee and Methow basins, and the population was considered by NOAA for designation as threatened or endangered, as were upper Columbia River summer/fall Chinook; though these populations ultimately were not listed, they are still monitored by Columbia Basin Research and NOAA. These eight monitored populations are supported by more than 30 artificial propagation programs along the Columbia River and its tributaries.


The University of Washington’s Columbia Basin Research center provides data about these eight populations at any one of three federal dams: Bonneville Dam, Lower Granite Dam and McNary Dam. The tags, called passive integrated transponders, help generate data including the number of tagged juveniles released each year on their outbound journey downriver and the number of fish from each release year that were later detected as adults returning upriver from the ocean. Comparing these two quantities taken at Bonneville Dam, the nearest dam to the ocean on the Columbia River, provides an estimate of how many fish survived the ocean.

We calculated survival rates across two time periods: 2008-2013 and 2014-2018. During the first period, coastal conditions and climate conditions in the Pacific Ocean were particularly favorable for salmon and steelhead trout. Conditions changed around 2014, the beginning of our most recent span of data. Though Chinook, coho, sockeye and steelhead trout all mature at different times and follow distinct migration patterns, the majority of adult fish from these species return to freshwater to spawn after four years in the ocean, which is why we ended our analysis with the 2018 population: Any juveniles released after that may not yet have had time to return as adults, so that was the most recent population for which data was reliable.

We compared survival rates to benchmarks established by the Northwest Power and Conservation Council. In 2003, the council set a goal of 4%, on average, of all juvenile salmon who headed for the ocean would return to fresh water as adults, although it allowed for a range of between 2% and 6% annually. According to the council, these rates should be sufficient to ensure the recovery of the salmon species that are listed as endangered and to help reach the council’s goal of 5 million total salmon and steelhead returning to the Columbia and its tributaries each year. We used this approach for a couple key reasons. While individual hatcheries assess their programs using a variety of measures, we found that these assessments aren’t standardized and very few people are looking at the overall success of the hatchery system; the 4% benchmark allows us to look at the health of the system as a whole. We also listened to the advice of experts in looking at the data across multiple years, because the results from any given year are too volatile to be meaningful.

In 2008-2013, only two of the eight populations we examined had average returns exceeding 4%: mid-Columbia coho and Snake River fall Chinook. In 2014-2018, none of the populations had average returns exceeding 4%. For the statistically minded, some further notes: To characterize the uncertainty in average survival rates for the two time periods, we ran bootstrapping experiments on the data using 1,000 trials within each time period to calculate 95th percentile confidence intervals around the bootstrap mean. The confidence intervals for three additional populations (Snake River spring/summer Chinook, upper Columbia River summer/fall Chinook and upper Columbia River steelhead) included the 4% recovery goal between 2008 and 2013. Between 2014 and 2018, the confidence interval for one population, the mid-Columbia coho, included the 4% minimum threshold.

It is important to note that fish biologists evaluate salmon and trout using a variety of performance indicators and metrics, with ocean survival being only one of them. Other metrics include the total number of juvenile fish released by hatcheries, the total number of adult fish returning to fresh water, the proportion of adult returns that started as hatchery juveniles, the state of local habitats, and even fish genetics. However, many of these measures are complex and difficult to compare across a variety of fish management practices, geographies and fish populations. By contrast, estimates of survival are readily available and offer a relatively holistic picture of how a population is doing. They also allow us to see the return on investment of the resources that have been allocated to hatcheries programs — a crucial measure given the limited amount of money available for this effort.

This estimate of ocean survival has some caveats. Only a portion of the salmon released each year are tagged. Furthermore, only a fraction of juvenile salmon survive the journey from release sites far upstream of Bonneville to the dam, and not all adult salmon that make it to Bonneville on the return trip will survive the full freshwater journey back through the hydropower system. That means the estimates drawn from these numbers are generous — the highest that they’ll be on the journey upriver. Nevertheless, these ratios of adult to juvenile tag detection can be considered an index that reflects the trends in populations that can be compared across species, migration patterns and release sites.