Optimizing sampling effort within a systematic design for estimating abundant escapement of sockeye salmon (Oncorhynchus nerka) in their natal river

The upstream migration of adult sockeye salmon (Oncorhynchus nerka) in the Horsefly River was monitored by a DIDSON imaging sonar during the dominant stock-cycle year 2005 using a systematic 20-min h−1 sampling scheme. We used a subset of these data collected between 16 and 29 September to investigate whether this sampling protocol was justified based on temporal variation in the salmon migration data. During post-processing, the 20-min sequence was split into two 10-min periods and the number of migrating salmon was counted separately. Cross- and autocorrelation analysis showed that estimates from the first and second 10-min samples were similar (r = 0.65) and variation between them (i.e., within the hour) was random, supporting the conclusion that systematic-hourly sampling is a defensible sampling design for acoustic enumeration when temporal variation in fish migration is unknown a priori. Using a simple benefit–cost model (statistical reliability of point estimates of salmon escapement–sampling effort), we recommend a minimum sampling effort of 10-min h−1 and a maximum effort of 20-min h−1 for projects using a systematic sampling scheme in which the goal is to estimate total upstream salmon escapement. An alternative sampling approach targets high-passage events such as diurnal peaks or periods when total daily upstream escapement exceeds 25 000 fish d−1, for increased sampling effort while reducing sampling effort during low-passage periods. This design will improve the statistical reliability of the resulting point estimates of upstream escapement relative to that achievable with a systematic effort with no overall change in total sampling effort over the course of the migration period.