Utility and Validation of Day and Night Snorkel Counts for Estimating Bull Trout Abundance in First- to Third-Order Streams

Abstract

Despite the widespread use of underwater observation to census stream-dwelling fishes, the accuracy of snorkeling methods has rarely been validated. We evaluated the efficiency of day and night snorkel counts for estimating the abundance of bull trout Salvelinus confluentus in 215 sites within first- to third-order streams. We used a dual-gear approach that applied multiple-pass electrofishing catch data adjusted for capture efficiency to estimate true or baseline fish abundance. Our multiple-pass electrofishing capture efficiency models were based on a prior study and used recapture data for known numbers of individually marked fish. Snorkeling efficiency was estimated by comparing day and night snorkel counts with the baseline. We also evaluated the influence of fish size and stream habitat features on snorkeling efficiency. Bull trout snorkeling efficiency was higher at night (mean = 33.2%) than during the day (mean = 12.5%). Beta-binomial regression indicated that bull trout day and night snorkeling efficiencies were positively related to fish size and negatively related to stream width and habitat characteristics. Day snorkeling efficiency also was positively influenced by water temperature and nonlinearly related to underwater visibility, whereas night snorkeling efficiency was nonlinearly related to water temperature and pool abundance. Although bull trout were our target species, day and night snorkeling efficiencies combined for rainbow trout Oncorhynchus mykiss and subspecies of cutthroat trout O. clarkii averaged 32.3% and 18.0%, respectively. Our ability to detect and accurately count fish underwater was influenced by fish size, species, time of day, and stream habitat characteristics. Although snorkeling is versatile and has many advantages over other sampling methods, the use of raw snorkel counts unadjusted for the effects of these biases will result in biased conclusions. We recommend that biologists adjust underwater count data to minimize the effect of such biases. We illustrate how to apply sampling efficiency models to validate snorkel counts.