Evaluation of juvenile salmonid behavior near a prototype weir box at Cowlitz Falls Dam, Washington, 2013

Abstract

Collection of juvenile salmonids at Cowlitz Falls Dam is a critical part of the effort to restore salmon in the upper Cowlitz River because the majority of fish that are not collected at the dam pass downstream and enter a large reservoir where they become landlocked and lost to the anadromous fish population. However, the juvenile fish collection system at Cowlitz Falls Dam has failed to achieve annual collection goals since it first began operating in 1996. Since that time, numerous modifications to the fish collection system have been made and several prototype collection structures have been developed and tested, but these efforts have not substantially increased juvenile fish collection. Studies have shown that juvenile steelhead (Oncorhynchus mykiss), coho salmon (Oncorhynchus kisutch), and Chinook salmon (Oncorhynchus tshawytscha) tend to locate the collection entrances effectively, but many of these fish are not collected and eventually pass the dam through turbines or spillways. Tacoma Power developed a prototype weir box in 2009 to increase capture rates of juvenile salmonids at the collection entrances, and this device proved to be successful at retaining those fish that entered the weir. However, because of safety concerns at the dam, the weir box could not be deployed near a spillway gate where the prototype was tested, so the device was altered and re-deployed at a different location, where it was evaluated during 2013. The U.S. Geological Survey conducted an evaluation using radiotelemetry to monitor fish behavior near the weir box and collection flumes. The evaluation was conducted during April–June 2013. Juvenile steelhead and coho salmon (45 per species) were tagged with a radio transmitter and passive integrated transponder (PIT) tag, and released upstream of the dam. All tagged fish moved downstream and entered the forebay of Cowlitz Falls Dam. Median travel times from the release site to the forebay were 0.8 d for steelhead and 1.2 d for coho salmon. Most fish spent several days in the dam forebay; median forebay residence times were 4.4 d for juvenile steelhead and 5.7 d for juvenile coho salmon. A new radio transmitter model was used during the study period. The transmitter had low detection probabilities on underwater antennas located within the collection system, which prevented us from reporting performance metrics (discovery efficiency, entrance efficiency, retention efficiency) that are traditionally used to evaluate fish collection systems. Most tagged steelhead (98 percent) and coho salmon (84 percent) were detected near the weir box or collection flume entrances during the study period; 39 percent of tagged steelhead and 55 percent of tagged coho salmon were detected at both entrances. Sixty-three percent of the tagged steelhead that were detected at both entrances were first detected at the weir box, compared to 52 percent of the coho salmon. Twelve steelhead and 15 coho salmon detected inside the weir box eventually left the device and were collected in collection flumes or passed the dam. Overall, collection rates were relatively high during the study period. Sixty-five percent of the steelhead and 80 percent of the coho salmon were collected during the study, and most of the remaining fish passed the dam and entered the tailrace (24 percent of steelhead; 13 percent of coho salmon). The remaining 11 percent of steelhead and 7 percent of coho salmon did not pass the dam while their transmitters were operating. We were able to confirm collection of tagged fish at the fish facility using three approaches: (1) detection of radio transmitters in study fish; (2) detection of PIT-tags in study fish; (3) observation of study fish by staff at the fish facility. Data from all three methods were used to develop a multistate mark-recapture model that estimated detection probabilities for the various monitoring methods. These estimates then were used to describe the percent of tagged fish that were collected through the weir box and collection flumes. Detection probabilities of PIT-tag antennas in the collection flumes were 0.895 for juvenile steelhead and 0.881 for juvenile coho salmon, although radiotelemetry detection probabilities were 0.654 and 0.646 for the two species, respectively. The multistate model estimates showed that all steelhead and most coho salmon (94.5 percent) that were collected at the dam entered the collection system through the flumes rather than through the weir box. None of the tagged steelhead and only 5.5 percent of the tagged coho salmon were collected through the weir box. These data show that juvenile steelhead and coho salmon collection rates were much higher through the collection flumes than through the weir box. Low detection probabilities of tagged fish in the fish collection system resulted in uncertainty for some aspects of our evaluation. Missing detection records within the collection system for fish that were known to have been collected resulted in four tagged steelhead and seven tagged coho salmon being removed from the dataset, which was used to assess discovery rates of the weir box and collection flumes. However, the multistate model allowed us to provide unbiased estimates of the percentage of tagged fish that were collected through each route, and these data showed that few fish were collected through the weir box. Overall, the fish collection system performed reasonably well in collecting juvenile steelhead and coho salmon during the 2013 collection season. Fish collection efficiency estimates from the Washington Department of Fish and Wildlife showed that steelhead collection efficiency was slightly higher than the 10-year average (46 percent compared to 42 percent), whereas coho salmon collection efficiency was more than twice as high as the 10-year average (63 percent compared to 30 percent). However, the performance of the weir box was poor because most fish were collected through the collection flumes.