Methods for standardizing CPUE and how to select among them

Abstract

Methods for estimating trends in relative abundance based on catch-per-unit-effort (CPUE) are discussed, including general linear and additive models, habitat-based standardization, neural networks, and regression trees. Methods and criteria for testing among various techniques are presented and recommendations for future research are presented. Good prediction of catch or CPUE does not necessarily infer good estimation of abundance, but is often assumed to estimate a year effect. Tests fall into two categories, either making this assumption, or assuming the dynamics model is correct and finding consistency with ancillary data. Immediate needs and data availability drive current model selection, but required future research in five related areas includes: (1) determining which of current methods for standardizing CPUE are generally applicable, with comparison to other methods; (2) developing tests to determine which methods provide the best index of relative abundance; (3) determining the status of current data and deficiencies, and prioritizing collection for current model application; (4) defining what data should be collected in the future to capture changes in relationships between catch and effort, and to ensure information context and usefulness of long-term data series; and (5) exploring new methods for standardizing CPUE, anticipating changes in requirements of