Skip to Main Content
PTAGIS logo link back to home page

Delayed Emigration of Juvenile Steelhead and Its Impact on Survival Estimates

Lora Tennant1, Ryan Kinzer2, Jim Harbeck1, Peter Cleary3

  1. Nez Perce Tribe, 500 N Main Street, Joseph, OR 97846
  2. Nez Perce Tribe, McCall Fisheries Office, McCall, ID 83638
  3. Nez Perce Tribe, Orofino Field Office, Orofino, ID 83544

The Nez Perce Tribe operates five rotary screw traps within the Snake River watershed to capture emigrating juvenile salmon and steelhead. Fishes are enumerated, measured, and PIT-tagged before they are released. An objective of these operations is to estimate cohort survival from the trap sites to Lower Granite Dam. Although juvenile Chinook Salmon emigration timing is often predictable, juvenile steelhead emigration timing is not. We have noticed not all steelhead tagged at Tribal rotary screw traps actively emigrated to the ocean during the migration year (July – June) in which they were tagged. The proportion of juvenile steelhead that exhibit delayed emigration varies by trap location and migration year. It is well documented that steelhead have a variable and flexible freshwater life history and may spend 1 – 5 years in freshwater before downstream migration to the ocean. The propensity of emigrants to delay emigration post-PIT tagging might be associated with multiple interrelated factors such as the location of capture within the respective subbasins (e.g., headwaters, lower mainstem), age of the juvenile, availability of habitat and food, or numerous other abiotic and biotic factors that influence early life-history strategy. Using our current methodology of estimating emigrant survival, delayed emigration would contribute to an underestimate of emigrant survival for a given migration year. We are exploring changes to our survival estimation methods in an effort to remedy our biased emigrant estimates. Changes we are pursuing include: 1) using a software developed by the University of Washington called Basin TribPit to estimate emigrant survival and, 2) developing our own machine learning models using previously collected data to inform predictions on whether an individual will emigrate the same year as tagged or delay emigration, then using these predictions to estimate emigrant survival for the current migration year.