Frameless—finding and refining a sampling frame for surveying recreational fisheries: lessons from estimating Swedish harvest of western Baltic cod

To achieve sustainable fisheries, advice to management should be based on reliable science and unbiased data. Attaining quality data (i.e. precise and unbiased) on recreational fishing can be challenging, particularly when prior knowledge of the sector is limited and a proper sample frame of recreational fishers or vessels does not exist. In this study, a registry of access points was constructed for the Swedish south-west coast and used as a spatial sample frame in determining both effort and catches of the private boat fishery. Sampling dates, times for sampling, and access points visited were selected using probabilistic methods, ensuring unbiased results. The final multi-stage sampling design involved multiple strata, clusters, and probability selection methods and enabled first-time estimation of Swedish recreational landings of western Baltic cod by private boats to be used in stock assessment. Concurrent data collection covering aspects such as boat counts at access points, provided additional information on e.g. activity patterns. That additional information opens possibilities to refine the design of the original survey and optimize the sampling effort towards different goals, such as other fished resources. In this paper, we reflect on the challenges that limitations in initial information poses to the design and deployment of a new recreational fisheries survey. We suggest ways, whereby indirect sampling frames can be developed from initially incomplete or limited information to access the fishers and their catch. Our experience shows that, despite initial frame and knowledge limitations, full probabilistic methods are worth considering in data limited scenarios and that the design-based point estimates and variances they provide on recreational fishing effort and catches are useful in guiding initial management and the next steps of survey improvement.