A Stratified Random Transect Design for Acoustic Surveys of Fish Stocks

Abstract

The differences between design-based and model-based inference are examined briefly and some recent applications of model-based procedures in fishery surveys are discussed. It is shown that valid estimates of mean and variance can be obtained by taking a stratified random sample, where the primary sampling units in each stratum are parallel transects randomly-spaced within certain non-critical limits. The strata are defined on the basis of the expected fish density or the expected variance in fish density within the region. Formulae are developed for the optimum allocation of transects to strata according to the density or variance in density within each stratum. To illustrate that such designs are entirely feasible, an example is given of a stratified random acoustic survey of South African anchovy (Engraulis capensis), in which the strata were defined on the basis of anchovy distribution encountered on previous surveys of the region. Through a two-phase approach, the transect allocation was adjusted during the survey for differences between the expected and observed distributions. It is demonstrated that in this survey, stratification, with uniform allocation of effort between strata, would have increased precision by an estimated factor of 2.4 compared with an unstratified random sample, while optimising effort within strata gave a further potential increase of 1.4. The coefficient of variation actually achieved was 16%