Speciation Uncertainty as the Dominant Driver of Arsenic Risk Characterization in Tuna: A Probabilistic Sensitivity Analysis

Arsenic (As) is common in marine fish, predominantly as arsenobetaine and other organic forms of negligible toxicological concern. Inorganic arsenic, the species of regulatory relevance, typically constitutes only a minor fraction of total arsenic in pelagic fish such as tuna, yet speciation data are frequently unavailable in occurrence monitoring programmes, forcing risk assessors to rely on assumed inorganic fractions. The magnitude of uncertainty introduced by this assumption, and its influence on risk conclusions, remains poorly characterised. This study quantified the impact of speciation uncertainty on probabilistic inorganic arsenic risk characterisation in Portuguese and Spanish adult tuna consumers. Total arsenic concentrations were determined by ICP-MS/OES in 37 yellowfin tuna samples collected at the Port of Vigo and fitted to a log-normal distribution. Consumption data from IAN-AF 2015 (Portugal) and ENALIA2 2016 (Spain) were modelled as gamma distributions fitted by percentile matching to national dietary survey summary statistics. Monte Carlo simulation (n = 10,000 iterations) was performed independently under three inorganic arsenic fraction assumptions: 100% (fully conservative), 10%, and 5%, the latter two reflecting values reported in the literature for marine fish. Under the fully conservative scenario, median margins of exposure (MOE) against the upper BMDL₀₁ were 21 and 37 for Portugal and Spain respectively, with no iterations exceeding the low-concern threshold of MOE = 10,000. Even under the most realistic assumption of 5%, median MOEs of 432 and 742 remained far below 10,000, with fewer than 2% of simulated exposures classified as low concern in either population. Speciation assumption drives substantial variation in absolute MOE values but does not alter the fundamental risk conclusion: inorganic arsenic exposure from tuna consumption warrants concern across the full consumer distribution regardless of the fraction applied.