The characterization of seafood mislabeling: A global meta-analysis

Abstract

Abstract With the advent of DNA forensics, research on seafood fraud has increased drastically. The documentation of mislabeling has raised concern over the identity, value, and safety of seafood. However, the general characterization of mislabeling is limited. We conduct a Bayesian meta-analysis to estimate global mislabeling rates and their uncertainty across several factors. While the effort to document mislabeling is impressive, it is highly skewed toward certain taxa and geographies. For most products, including all invertebrates, there is insufficient data to produce useful estimates. For others, the uncertainty of estimates has been underappreciated. Mislabeling is commonly characterized by study-level means. Doing so often overestimates mislabeling, masks important product information, and is of limited utility—particularly given that studies often lack adequate sampling designs for parameter estimation. At the global level, overall mislabeling rates do not differ statistically across supply chain locations, product forms, or countries. Product-level estimates are the most informative. The majority of products, for which there is sufficient data, have mislabeling estimates lower than commonly reported. The most credible average mislabeling rate at the product-level is 8% (95% HDI: 4–14%). Importantly, some products have high estimates, which should be priorities for research and interventions. Estimates must be combined with other data in order to understand the extent and potential consequences of mislabeling, which is likely to vary drastically by product. Our meta-analysis, which can be updated with new data, provides a foundation for prioritizing research to inform programs and policies to reduce seafood fraud.