Sample sizes for length and density estimation of 0+ fish when using point sampling by electrofishing

Standard lengths of cyprinids at several sites in the River Great Ouse, England, were compared between catches taken using a conventional 15 × 3 m micromesh seine (pore diameter 2 mm) and using point sampling by electrofishing (PSE). No statistically significant differences in sites were found in six out of 10 trials. In nine out of 10 trials, the difference between the mean lengths differed by <1 mm. No serial bias was found between PSE and seine netting for cyprinid fishes between 14 and 100 mm SL, although variation between size classes was high, owing to small sample sizes. The coefficient of variation in fish length with size tended to increase with the age of 0+ Rutilus rutilus . The relationship between sample size (n) and variance ( s 2 ) was explained by s 2 =13·9 n −1·24 . Sampling more than 30 fish resulted in little increase in precision. The relationship between the mean catch per site (x̄) and the variance of the mean estimate was log s 2 =1·6039 log x̄ +0·973. The number of samples required to estimate density within a given variance range was: n =9·33 x̄ 1·6–2 CV x̄ −2 . Given the high variance‐mean ratio, great care should be taken when interpreting density data collected using PSE, and 50 point samples is the minimum required for density estimation.