METHODS

We constructed an index of spawner and total abundances at the regional scale using the best available data. Details of specific data sources for each region and species are outlined in the Technical Documentation and raw data are publicly available in the associated GitHub repository. We often relied on data vetted by the Pacific Salmon Commission and used to assess international obligations under the Pacific Salmon Treaty. Specifically, these data come from the appendices of reports by the Yukon River Joint Technical Committee, Transboundary Technical Committee, Northern Boundary Technical Committee, and Chinook Technical Committee. The Fraser River Panel provides data on pink and sockeye abundance directly via their Annual Report Application. Publication of these appendices may take years and more up-to-date versions were provided to Pacific Salmon Foundation staff directly by Fisheries and Oceans Canada (DFO) upon request.  

For Haida Gwaii Chinook, we used new data from monitoring by the Haida Nation, which enabled us to report on the current state for the first time in this year’s report. In the Nass region, we drew on information from the draft 2025 Post-Season Report from the Nisga’a Lisims Government - Fisheries & Wildlife Department^†. Fraser coho and chum salmon data were provided directly to Pacific Salmon Foundation staff by DFO upon request. The Fraser coho estimates of spawner and total abundances are based on Interior Fraser coho that spawn upstream of Hells Gate, British Columbia (including five Conservation Units: Fraser Canyon, Interior/Middle Fraser, Lower Thompson, South Thompson, and North Thompson). Columbia Chinook and sockeye data were provided by DFO and the Syilx Okanagan Nation Alliance. Steelhead spawner abundances in the Skeena and Fraser are publicly shared by the Province of British Columbia, and steelhead surveys in the Vancouver Island regions were shared directly with PSF by the Province. Steelhead spawner abundance in the Canadian portion of the Columbia is reported by the Okanagan Basin Monitoring and Evaluation Program.  

For regions and species that did not have reliable estimates of aggregate abundance at the regional scale, we constructed an index of spawner abundance from stream spawner surveys reported in DFO’s New Salmon Escapement Database System (NuSEDS) up to 2024. Prior to analysis, the data were checked for quality issues including duplicated or conflicting records and missing stream coordinates. We accounted for variable survey effort using a proportional infilling approach to impute missing surveys based on the average decadal contributions of each stream to the aggregate. We then expanded the imputed dataset to account for smaller “non-indicator” streams. The details of this expansion approach are outlined in the Technical Documentation.  

In total we compiled spawner abundance for 51 regional salmon populations (i.e. unique region-species combinations), with an average time series length of 58 years (range 20 to 133 years). The temporal currency of data varied among sources, with most regions and species having spawner abundance up to 2025 (n = 33) or 2024 (n = 16). Fewer regions and species had estimates of total abundance because information on which rivers (and regions) salmon are destined for when caught in the ocean is not always available, making it difficult to assign catch to a region. We compiled total abundance for 21 regional salmon populations, with an average time series length of 45 years (range 25 to 133 years). Most total abundance time series were current to 2025 (n = 19).  

We smoothed the raw time series of spawner and total abundances using a right-aligned running geometric mean over the generation length. This smoothing reduces the influence of dominant years for species with cyclic dynamics (e.g. Fraser sockeye) and produces an index of abundance that is less sensitive to stochastic interannual variability that is common in salmon population dynamics^†. The generation length is based on the dominant life-history type for each species in a particular region. To facilitate comparison among species that are naturally very different in their absolute abundance, we transformed the smoothed time series into a per cent anomaly from the geometric mean abundance among all years for that species and region.  

We summarised the current state as the per cent anomaly for the most recent year of the smoothed time series (i.e. the per cent difference between the geometric mean spawners over the most recent generation and the geometric mean spawners over all years). For species with long generation lengths (e.g. six years for Yukon Chinook), the current state may not reflect recent, dramatic changes in abundance because the generational average lags behind. Where there are noteworthy changes in abundance in specific years that are not obvious from the smoothed time series, we may refer to the raw “annual spawner count” estimates.  

We summarised trends as the average annual per cent change in abundance, calculated from the slope of a linear regression model fitted to the log-transformed, smoothed time series. We show both short-term trends based on the slope over the most recent three generations only and long-term trends based on the slope over all available years for each species in a given region.