Statistical assessment of fish behavior from split-beam hydro-acoustic sampling

Statistical methods are presented for using echo-traces from split-beam hydro-acoustic sampling to quantitatively assess fish behavior in response to a stimulus. The data presented are from a 2002 study designed to assess the response of free-ranging, lake-resident fish, primarily kokanee (Oncorhynchus nerka) and rainbow trout (Oncorhynchus mykiss), to high intensity strobe lights. The study was conducted in the Forebay of Grand Coulee Dam on the Columbia River in Northern Washington State. The lights were deployed immediately upstream from hydroelectric turbine intakes, in a region of approximately 50 m depth, which is exposed to alternate periods of high and low flows over a 24-h power generation cycle. The study design included five down-looking split-beam transducers positioned in a line at incremental distances upstream from the strobe lights, and a 3-level treatment applied in randomized pseudo-replicate blocks. The fundamental sampling units (fish-tracks) were analyzed by transducer position and treatment level using odds-ratios estimated from log-linear models. Fish-track direction of movement was analyzed using circular probability distributions. Both analyses are depicted graphically. Study results suggest the lights and light frame both increased fish activity in the vicinity of the light frame that could be interpreted as attraction. Increased fish activity associated with the strobe lights was most notable at night and during periods of low flow; increased fish activity associated with the strobe light frame was most notable during periods of higher flow. The lights induced notable bimodality in the angular distributions of the fish tracks suggesting further induced changes in behavior associated with the lights. Statistical summaries are presented along with statistical inferences supporting interpretations and conclusions on fish behavior.