Underwater video has become an important tool for monitoring reef fish populations worldwide because it is nonextractive and not strongly selective. A variety of approaches have been developed to enumerate fish on videos, but to our knowledge these metrics have not been tested to determine if they are proportional to true abundance. We compared the most commonly used metric, MaxN (i.e., the maximum number of fish in a single frame during the viewing interval), to a newly developed metric, MeanCount (i.e., the mean number of fish observed in a series of snapshots over a viewing interval), using simulations, a laboratory experiment, and an empirical study. MaxN was nonlinearly related to true abundance using all three approaches, providing increasingly dampened estimates of abundance with increasing true abundance (i.e., hyperstability). Therefore, MaxN may result in positively biased indices of abundance for declining fish stocks or negatively biased abundance indices when fish stocks are increasing. Alternatively, MeanCount was generally linearly related to true abundance and its variability was similar to MaxN, suggesting that MeanCount can be useful for indexing abundance of fish in underwater video surveys.
Underwater observations of seal–fishery interactions and the effectiveness of an exclusion device in reducing bycatch in a midwater trawl fishery
