Coverage matters: identifying and mitigating sampling frame issues in recreational fishing surveys

Abstract Surveys play an integral part in monitoring and maintaining sustainable recreational fisheries. For any probabilistic survey, the selection of a sampling frame (e.g., list of individuals or fishers) is an important decision because it influences the ability to provide unbiased estimates of recreational catch and effort. Undercoverage occurs when units of the target population (i.e., the population of interest) are missing from the frame population. This error can undermine the reliability of research advice generated from survey estimates. In this review, we: (i) define six sampling frame configurations that are commonly applied in probabilistic recreational fishing surveys; (ii) synthesise how coverage errors associated with each configuration have been addressed for marine recreational fisheries globally; (iii) outline approaches to identify and correct for coverage errors; and (iv) recommend how to future-proof coverage issues. In our six case studies, multiple types of undercoverage were identified and addressed to varying extents, depending on the characteristics of each fishery and type of sampling frame used. Generalised list frames (particularly phone lists) are arguably the most prone to undercoverage error. To assist in future-proofing surveys, we recommend: (1) considering coverage error during survey planning; (2) designing pilot surveys or scheduling concurrent surveys to evaluate and/or correct for potential bias; (3) recognising that coverage error often changes through time; (4) using technological or multi-frame approaches to mitigate coverage error; (5) considering model-based survey tools to correct for undercoverage; and (6) documenting the sampling frame and potential sources of coverage error in publications. These recommendations extend to inland recreational fisheries, commercial fishing surveys and fisheries-independent surveys. Graphical abstract