Using occupancy models to assess the effectiveness of underwater cameras to detect rare stream fishes

Many conservation efforts for freshwater fishes have been undertaken; however, continuing the monitoring of both the distribution and the abundance of species to determine the effectiveness of these actions can be difficult. As species increase in rarity, they are more difficult to detect in the field, making inferences on occupancy less reliable. Conventional sampling methods, such as electrofishing and seining, require the physical handling of rare fishes, which may cause stress and mortality and, consequently, compromise conservation goals and limit monitoring programmes. Non‐invasive surveillance methods, including underwater video, are playing an increasingly important role. In this study, occupancy models were used to estimate the detection probability of underwater cameras as an alternative to the conventional sampling methods for rare stream fishes. Redside dace ( Clinostomus elongatus ), a small minnow listed as Endangered, was used as a model organism for rarity. A total of 69 historical redside dace sites were sampled using three sampling methods to determine the effect of gear type on detecting and identifying the habitat preferences of this rare minnow. On average, using multiple underwater cameras is as effective at detecting a rare minnow as conventional sampling methods (backpack electrofisher and seine) and causes no harm. The detection probability of both underwater cameras and backpack electrofishing were adversely affected by turbidity, whereas seining was positively affected by stream velocity. The probability of occupancy of redside dace is driven by open channels and sediment size, and this provides a strong basis for informing stream restoration projects. The use of multiple underwater cameras over conventional sampling methods is recommended when sampling for rare and endangered minnows in systems with low turbidity.