Skip to main content

Phenological shifts in out-migrating juvenile Pacific salmon

Cite this dataset

Wilson, Samantha M. et al. (2023). Phenological shifts in out-migrating juvenile Pacific salmon [Dataset]. Dryad.


This dataset contains the rates of change in juvenile migration timing of 66 Pacific salmon populations ranging from Oregon to Alaska, the rates of change in timing of the initiation of spring phytoplankton bloom in nearby coastal regions, and covariates. Species represented in this dataset include coho salmon (Oncorhynchus kisutch), sockeye salmon (O. nerka), Chinook salmon (O. tshawytscha), pink salmon (O. gorbuscha), chum salmon (O. keta), and steelhead trout (O. mykiss). Rates of change in salmon phenology includes rate of change in peak date, and rate of change in range (number of days between 25th and 75th quantile of migration).


This dataset represents the rates of change in juvenile salmon outmigration phenology. Rates of change were determined using the 'phenomix' R package from either daily outmigration counts, or mark-recapture corrected abundance estimates. In general, juvenile salmon were intercepted during their migration to the ocean using a full fence weir, partial fence weir, rotary screw trap, incline plane trap, floating trap, or seine. Individuals were identified to species, counted, and released. Mark recapture expanded estimates were only used when more than three stratified time estimates within a year were made. All rates of change were calculated from datasets with at least 20 years of outmigration timing data, with at least 15 sampling days per year. The longest dataset began in 1951. Rates of change in phenology were determined for the full dataset and for a subset of years spanning 1999 - 2019 (to match with availability of chlorophyll-a concentration data).

In addition to rates of change in juvenile salmon outmigration phenology, this dataset contains rates of change in the corresponding coastal ocean spring phytoplankton bloom timing. Bloom timing estimates were derived from level-3 processed daily global composites (9 km x 9 km) of surface chlorophyll-a concentration from two satellites, Sea-viewing Wide Field-of-view Sensor (SeaWiFS; 1999 - 2010) and the Moderate Resolution Imaging Spectroradiometer (MODIS-Aqua; 2003 - 2019) from the Goddard Space Flight Center. 2x2 degree grid cells along the coastal ocean between 42 – 60oN, 161.5 – 124.5oW were subset from global composites. Sequential 8-day chlorophyll-a concentration estimates from Jan 1 to Aug 1 for 20 years spanning 1999 – 2019 for each grid cell were generated. The day of year of initiation of the coastal spring phytoplankton bloom was determined to be the mid-date of the first 8-day chlorophyll-a concentration estimate that exceeded 5% of the median chlorophyll-a concentration estimate for that year. Rate of change in bloom date was determined using a linear regression of year and bloom day of year. Each population of salmon was linked to a rate of change in initiation of the spring phytoplankton bloom by both grid cell and river outlet. Thus, the coastal ocean grid cell corresponded to areas where juvenile salmon would enter the ocean. 

Rates of change in seasonal air temperature and precipitation were determined for each population. Seasonal air temperature and precipitation were estimated using ClimateNA (v.5.21) and the latitude, longitude, and elevation where the juvenile salmon were intercepted, for each year salmon were counted, up to 2013 (limit of ClimateNA v.5.21). The rate of change in seasonal air temperature and precipitation were determined for each salmon population using a weighted linear regression of year and seasonal air temperature or precipitation value for that year, weighted by the inverse variance in the estimate of air temperature or precipitation.