Predicting fish abundance using single-pass removal sampling
Three-pass removal data for juvenile rainbow trout (Oncorhynchus mykiss) along bank areas of the Henrys Fork of the Snake River, Idaho, were used to construct a mean capture probability (MCP) model to predict abundance from single-pass catch data. We evaluated the MCP model by simulation. The precision of the MCP model was poor when predicting abundance within a specific bank unit. MCP model prediction intervals were about 7.5 times greater than three-pass removal intervals. However, the MCP model performed about the same as three-pass removal for predicting total abundance in a river section from multiple bank samples. We evaluated how the MCP model can be used to improve precision of total abundance estimates. Reallocating effort to sample 150 bank units by single-pass removal rather than 50 bank units by three-pass removal resulted in a 48% increase in prediction interval precision for a simulated population of 10 000 fish. Precision also increased when allocating effort to sampling more bank units of smaller length versus fewer bank units of longer length. Sampling 1500 m of bank as one hundred 15-m bank units increased precision by about 28% versus sampling fifty 30-m bank units and by about 50% versus sampling twenty-five 60-m bank units.