A Fishwheel-ARIS study to evaluate the ARIS length-based discrimination estimates of species composition for the Mission Hydroacoustic monitoring site

Studies and analyses conducted from 2009-15 have demonstrated the spatially-stratified hydroacoustic counts can be combined with species composition data from the fishwheels and Whonnock test fisheries to produce scientifically defensible estimates of the number of sockeye and pink salmon that pass the Mission hydroacoustic site in odd-number years (Robichaud et al. 2010; English et al. 2016). In 2013 and 2015, fish length data derived from images acquired by imaging sonar DIDSON was used in mixture-model discrimination analysis to derive species composition estimates for the near-shore strata monitored by the DIDSON systems at the Mission hydroacoustic site (Grant et al. 2014). Substantial differences have been observed between the DIDSON length-based species composition estimates for near-shore strata and those derived from fishwheel samples (English et al. 2016). Both methods have their strengths and weaknesses. For the fishwheel, species should be identified accurately, but sockeye and pink salmon may not be equally vulnerable to the gear. For the imaging sonar systems, both species should be equally vulnerable to being observed in the image, but the measurement of lengths from the DIDSON images is less precise than manual measurements and the length distributions of sockeye and pink salmon are highly overlapped. To date, the PSC1 estimates of sockeye proportions derived from DIDSON length-based approach have tended to be higher than those derived from the fishwheel samples. Being a next generation of DIDSON, the adoptive resolution imaging sonar (ARIS) can provide a much higher down-range resolution than DIDSON for fish images. Through this project, we will assess the reliability of the ARIS length-based species composition estimates and further explore the potential causes of the discrepancies between ARIS and fishwheel based species proportions.