Dungeness River Winter Steelhead SONAR-based Escapement Estimates: 2019-2023

This report contains methods and results of SONAR-based escapement estimates for naturally spawning winter steelhead on the Dungeness River from 2019-2023. The Washington Department of Fish and Wildlife (WDFW) deployed a multibeam SONAR unit in the lower Dungeness River during the majority of the steelhead return from either early February or early March through the end of June or July, 2018-2023. Although the SONAR was operated in 2018, we were not able to produce a SONAR-based escapement estimate for 2018 due to frequent milling behavior at the site, and limited data review. We initiated a sub-sampling protocol in 2019 to enable us to review the entire season of imagery, and this was continued in subsequent years.

To exclude similarly sized fish that were not steelhead from our SONAR counts, WDFW conducted weekly species composition sampling in 2021, 2022 and 2023 throughout the periods of SONAR operation. The majority of fish encountered were steelhead and bull trout and these species overlapped in length. To exclude potential bull trout from the SONAR fish counts we fit a binomial generalized additive model (GAM) with a logit link, using splines of fork length and Julian day to predict the probability of a fish being a steelhead. The probability of being a steelhead was greatest during March and April, and least in February and June when fewer steelhead were present.

To fill in the missing data from periods when the SONAR was not operating or the data had not been reviewed, we fit the fish count data with a negative binomial GAM with a log link function using hour of day, discharge, and Julian day-of-year as covariates. For both upstream and downstream moving fish, the model indicates that a greater number of steelhead will move during the night, especially in the late evening, compared to the daytime, and more will move when discharge flows are near 600 cfs. There is also a clear effect of day-of-year, with upstream numbers peaking near the beginning of April, while downstream numbers peak near the beginning of May, with some year-to-year variability in the run-timing curve.

To generate an annual estimate of natural spawning escapement, we subtracted the total estimate of downstream moving fish from the total estimate of upstream moving fish from February 1 - May 15th to prevent double counting milling fish. After May 15, we ignored downstream moving fish and only estimated upstream moving fish, under the assumption that downstream moving fish are steelhead kelts that have spawned and therefore should not be subtracted from our escapement estimates. Total SONAR-based naturally spawning escapement estimates for steelhead ranged from 568 (CI 552 - 587) in 2021 to 1,075 (CI 982 - 1,193) in 2020. We did not estimate the proportion of hatchery-origin spawners (pHOS). The species composition sampling suggests a pHOS of 9.5 - 12% which is similar to hatchery-origin proportions from hook and line sampling in the Dungeness 2014- 2020 (WDFW 2024). The SONAR-based estimates were consistently roughly twice the redd-based estimates.

We learned some important lessons over the first 6 years of SONAR operation on the Dungeness River. Site selection is important and may need to change if fish aren’t actively migrating past the site. For future seasons we recommend deploying the SONAR in January, to capture the start of the steelhead return. We also recommend considering re-designing the species composition sampling to attempt to exclude holding fish, and to investigate how species composition changes with hour of day, since returning steelhead appear to move primarily in the late evening. Finally, we recommend investigating methods to account for kelts that are less simplistic than the methods we used here, and investigating methods for estimating pHOS.