Introduction

Radionuclides released from the Fukushima Daiichi Nuclear Power Plant (FDNPP) accident in March 2011 entered the coastal and ocean waters east of Japan via multiple pathways: deposition of airborne releases, direct liquid releases, and runoff from land via rivers. These releases increased activity concentrations (the massic activity concentration (unit: Bq kg−1) is referred to here as the activity concentration or the radionuclide concentration) in marine organisms1,2,3. After the accident, several organizations monitored marine fish and other seafood. These results were used, along with restrictions on the production, distribution and consumption of seafood (food restrictions) imposed by Japanese authorities to safeguard the food supply. The monitoring results showed that activity concentrations of cesium (134Cs and 137Cs) initially increased due to the accident, followed by a gradual decrease over the years4,5. By October 2024, the distribution and consumption of all seafood had been lifted6. In general, the accident heightened public concern regarding radionuclide concentrations in seafood7, and this concern persists today, in part due to the discharges of Advanced Liquid Processing System (ALPS) treated water into the sea, which began in August 2023 as part of the FDNPP decommissioning process.

Radiation dose assessments conducted before the FDNPP accident suggested that the typical (background) seafood ingestion dose to a Japanese person was more than three times the global average8, primarily due to higher seafood consumption rates in Japan compared to many other countries9. Pre-accident assessments also showed that the natural radionuclide 210Po contributed substantially to the total seafood ingestion dose. Ota et al. estimated that, before the accident, the annual dose from total food intake for a Japanese person was 0.8 mSv (note that dose values given in this paper represent committed effective dose values from annual food intake), with 91% (0.73 mSv) of this due to 210Po, and 86% of the dose from 210Po coming from fish and shellfish10. Sugiyama et al. similarly found a high contribution from 210Po in fish and shellfish (~70%). In their study, doses from 210Po varied substantially across seven large Japanese cities, ranging from 0.15 to 0.81 mSv (geometric mean (GM): 0.24 mSv)11. This variability is thought to reflect regional differences in seafood consumption, variations in 210Po concentrations in different types of purchased seafood, and whether the seafood was cooked or not. The importance of foods containing 210Po in the Japanese diet has also been highlighted in studies conducted by international organizations12,13.

After the FDNPP accident, radiation doses from accident-derived radionuclides (mainly radioactive cesium) in foods were assessed in several studies. Harada et al. estimated the total food ingestion dose from radioactive cesium for residents in Fukushima Prefecture to be 26 µSv yr−1 (arithmetic mean, AM) using measured activity concentrations from meals prepared for consumption (duplicate portion method) in December 201114. It should be noted that their results are limited to a short period after the accident and are not representative of the broader population.

The Ministry of Health, Labour and Welfare (MHLW) of Japan assesses ingestion doses to residents of Fukushima Prefecture using the market basket method, in which the dose is calculated based on radionuclide concentrations in samples collected from local markets and food consumption rates15,16,17,18,19,20,21,22,23,24,25,26,27. The results indicate that total food ingestion doses to residents of Fukushima Prefecture were estimated at 19 µSv 6 months after the accident, decreasing to ~1 µSv over the next 4 years, with little change thereafter. However, the assessment does not differentiate contributions from different food types or detail the relative contributions of natural versus accident-derived radionuclides.

The duplicate portion method and the market basket method provide relatively accurate estimates of ingestion doses of individuals in the survey. However, they are not suitable for obtaining representative doses for a large population due to the complexity of the procedures. Simulation is an effective approach for obtaining representative ingestion dose values from specific exposure pathways. Several simulations have been conducted for ingestion doses from seafood, with estimated doses ranging from 33 to 3000 µSv28,29,30. Some of these simulations focus on specific groups under hypothetical exposure scenarios. The large variations in the results of these studies highlight the fact that dose estimates can differ substantially depending on the assessment conditions considered, and that there remains a need to clarify the realistic ingestion doses received by the general population in Tohoku Region (northeastern Japan, including Fukushima Prefecture).

In this study, we aim to provide realistic estimates of pre- and post-accident ingestion doses from both anthropogenic and natural radionuclides to adult seafood consumers in the Tohoku Region. We apply a consistent methodology to a comprehensive dataset of measured activity concentrations for all relevant radionuclides and a wide range of seafood representative of the Japanese diet. The large volume of data enables us to consider the effects of seafood restrictions, account for the spatial and temporal variability of the radiological consequences of the FDNPP accident, and incorporate multiple seafood categories. Certain marine organisms are known to concentrate specific radionuclides: bivalves exhibit a higher 210Po bioconcentration factor (activity concentration ratio of biota to seawater) than other marine organisms12,31,32, fish have a higher 137Cs bioconcentration factor than other seafood1, and seaweed is particularly rich in iodine33. Therefore, in this study, seafood was classified into five groups: fish, crustaceans, bivalves and gastropods, cephalopods, and seaweeds.

Our calculation showed that the seafood ingestion dose from accident-derived radionuclides during the first year after the accident was 19 µSv for consumers exposed to the 95th percentile (95th%) dose. When comparing accident-derived and natural radionuclides, only 2% of the radiation dose from seafood consumption in the first year after the accident resulted from accident-derived radionuclides. This finding indicates that, even in the first year after the accident, the impact of FDNPP releases on ingestion doses to general seafood consumers in the Tohoku Region was minimal compared to the typical background dose from natural radionuclides, consistent with findings reported under various specific exposure assumptions28,29,30,34. Our study is important in providing a realistic assessment of seafood ingestion doses, accounting for the large temporal and spatial variations in radionuclide concentrations in seafood and the influence of the complex sequence of food restrictions imposed after the accident.

Results

Dose from accident-derived radionuclides

During the first year after the accident (March 2011 to March 2012), doses from accident-derived radionuclides (90Sr, 110mAg, 131I, 134Cs, 137Cs, and 239,240Pu) were 3.3 µSv for consumers at the 50th% of seafood ingestion dose, 5.8 µSv for the AM, and 19 µSv for consumers at the 95th%. These doses were ~4–6 times higher than those before the FDNPP accident (Mann–Whitney U test, P < 0.05) (Fig. 1 and Table 1). However, from 3 to 10 years after the accident, the dose was not significantly different from pre-accident levels (Mann–Whitney U test, P > 0.05), with 50th% doses of 0.74 µSv before the accident and 0.62 µSv after the accident (Table 1). This indicates that form 3 to 10 post-accident, the impact of the FDNPP accident on seafood ingestion doses became negligible for general seafood consumers in the Tohoku Region.

Fig. 1: Distributions of committed effective ingestion doses per annual seafood intake (34 ± 12 kg yr−1) from accident-derived radionuclides (90Sr, 110mAg, 131I, 134Cs, 137Cs, and 239,240Pu) to adult seafood consumers in the Tohoku region.
figure 1

The horizontal axis is logarithmic.

Full size image
Table 1 Committed effective ingestion dose estimates for adult residents in the Tohoku Region (µSv per annual seafood intake)
Full size table

In the first year after the accident, most of the accident-derived ingestion dose came from 131I, 134Cs, and 137Cs (Fig. 2(i), (iii)). The contributions of 134Cs and 137Cs were 1.2 µSv and 1.1 µSv at the 50th% and 8.2 µSv and 6.5 µSv at the 95th%, respectively, indicating that the effect of 134Cs with a half-life of ~2 years, was nearly as large as 137Cs in the first year. Three to ten years after the accident, the contributions of 134Cs and 137Cs were 17% and 74% (50th%) and 19% and 47% (95th%), respectively, with 137Cs accounting for the majority of the dose.

Fig. 2: Contributions of various radionuclides and seafood groups to committed effective doses to adult seafood consumers in the Tohoku Region from radionuclides associated with FDNPP releases (90Sr, 110mAg, 131I, 134Cs, 137Cs, and 239,240Pu).
figure 2

Left: contribution of each radionuclide (i) and seafood group (ii) to the 50th% dose. Right: contribution of each radionuclide (iii) and seafood group (iv) to the 95th% dose. Biv & Gast: Bivalves and gastropods.

Full size image

Considering seafood groups, fish contributed the most to the accident-derived ingestion dose, ranging from 40 to 90%. Seaweeds also made a substantial contribution, particularly in the first year after the accident (37–47%). Dose contributions from other seafood groups (crustaceans, bivalves and gastropods, and cephalopods) were relatively small (1–13%) (Fig. 2(ii), (iv)). A highly influential factor in these contributions is the consumption rate of each seafood group (e.g., fish accounts for more than 70% of overall seafood consumption in the study area) (Table 2).

Table 2 Consumption rates (g day−1) of fresh and processed seafood among adults in the Tohoku Region
Full size table

For fish, the largest contribution to the 50th% dose came from 137Cs (36%), followed by 131I (34%) and 134Cs (28%) in the first year after the accident. Three to ten years after the accident, most of the contribution came from radiocesium (137Cs 79% and 134Cs 16%). For seaweeds, main contributors in the first year after the accident were 134Cs (47%), 137Cs (30%), and 131I (21%). The smaller contribution of 131I to the dose, despite the high iodine bioconcentration factor in seaweeds, indicates that seaweeds with extremely high 131I concentrations were excluded by food restrictions considered in this study.

To identify the causes of the higher doses received during the first year after the FDNPP accident in detail, seafood consumption rates of groups receiving the 50th% and 95th% doses and activity concentrations in their seafood were compared. The analysis revealed that seaweed consumption rates, 134Cs and 137Cs activity concentrations in fish, and 134Cs and 137Cs activity concentrations in seaweed were significantly higher in the 95th% group (Mann–Whitney U test, P < 0.05).

Contribution of radionuclides and seafood groups to the total seafood ingestion dose

Total seafood ingestion doses including natural radionuclides were 350 µSv for consumers at the 50th%, and 1100 µSv for consumers at the 95th% (Table 1). The contributions of a small number of natural radionuclides far exceeded those of accident-derived radionuclides (Fig. 3(i)). The 210Po contribution to the seafood ingestion dose was the largest (71%) among all radionuclides, consistent with previous studies10,11,30. The contributions of 210Pb and 228Ra were also substantial (8.7% and 8.1%, respectively). The dominance of these natural radionuclides is attributed to the abundance of uranium- and thorium-series nuclides in the marine environment, primarily owing to atmospheric deposition of radon into the ocean, inputs from the Earth’s crust, and inflow from rivers35. Conversely, the contribution of accident-derived radionuclides was approximately 2% in the first year after the accident. This can be compared to baseline conditions before the accident when the same radionuclides, originating from general anthropogenic sources (e.g., global fallout), contributed 0.6% of the dose. The contribution of other anthropogenic radionuclides (3H, 60Co, 99Tc, 129I, and 241Am) was negligible in both pre- and post-accident doses. However, available data for these radionuclides were scarce compared to other radionuclides. In recent years, tritium in the marine environment has gained increased attention following the announcement of plans to release ALPS-treated water from the FDNPP into the ocean36. A study by the International Atomic Energy Agency (IAEA) and other independent studies concluded that the radiological impact of ALPS-treated water releases on people and the environment is negligible37,38,39. Releases began in August 2023, and ongoing monitoring data continue to be reported, providing a basis for updating assessments of potential impacts from these releases relative to natural background levels.

Fig. 3: Contribution of radionuclides and seafood groups to the 95th% committed effective ingestion dose in the first year after the accident.
figure 3

(i) Contribution of each radionuclide. (ii) Contribution of each seafood group.

Full size image

Of the five seafood groups, fish had the highest contribution (62%) to the total dose because of the high consumption of fish in Japan, similar to the case of accident-derived radionuclides (Fig. 3(ii)). The contribution of bivalves and gastropods to the ingestion dose was relatively high (14%) despite their consumption accounting for only 3% (Table 2). This is because 210Po activity in bivalves and gastropods is approximately one order of magnitude higher than in other seafood (Table 3), confirming that bivalves and gastropods tend to concentrate 210Po and are consequently key contributors to ingestion dose40,41,42. Although not as pronounced as in bivalves and gastropods, the relatively high activity of 210Po and 228Ra in crustaceans also resulted in a substantial contribution of 11%.

Table 3 Activity concentrations (Bq kg−1-fresh weight) of natural radionuclides50
Full size table

The range of variation in total seafood ingestion dose for a typical adult consumer (350 µSv) was 240–450 µSv (up to ±30%). Of the many factors contributing to the variation in ingestion doses, higher contributions came from spatio-temporal variabilities of radionuclide concentrations, the inter-individual variability of seafood consumption, and uncertainties of loss effects from cooking. Other lower contributions came from measurement uncertainties of radionuclide concentrations, uncertainties in the treatment of nondetected values, variabilities of random numbers generated by the Monte Carlo method, and uncertainties in ratios used to convert data from dry weight to fresh weight. See “Uncertainties” in Supplementary Discussion for details.

Discussion

Dose assessment for general adult consumers

Comparing the results of different dose assessments is challenging, as the assessment methods (e.g., dietary surveys, simulations) and conditions (e.g., subjects, target nuclides, food types, and origins) often vary substantially. Therefore, relevant results from previous studies are presented here, along with their specific conditions, as a reference for comparison with the findings of this study. We compare here only studies that assess ingestion dose to the general population, rather than specific groups with particular lifestyles that result in higher exposure.

Nakano and Povinec estimated the dose from 134Cs and 137Cs in seafood to the general Japanese population in 2012 to be 1.66 µSv, assuming seafood consumption originating from the open ocean in the Northwest Pacific, relatively far from the FDNPP43. Other ingestion dose assessments for general food (not limited to seafood) have been conducted using activity data from food products sourced both inside and outside the contaminated areas. Murakami and Oki estimated the dose to adults in Fukushima City from 134Cs and 137Cs in general food and drinking water per annual intake to be 28 µSv 4 months after the FDNPP accident and 5.3 µSv 1 year after the accident (Note: the doses from 1 month’s ingestion were reported in the paper, as 2.36 µSv month−1 and 0.44 µSv month−1)44. Takahara & Watanabe estimated doses to general adults in Fukushima Prefecture from accident-derived radionuclides (mainly 134Cs and 137Cs) in general foods in 2011, 2012, and 2013 to be 14.0, 7.5, and 4.6 μSv, respectively45.

In this study, the median dose from 134Cs and 137Cs in seafood to general adults in the Tohoku Region was 2.4 µSv in the first year after the accident and 1.5 µSv in 1–3 years after the accident (Table 1). Compared with previous studies, the results of the present study are of the same order of magnitude but somewhat lower. This is reasonable, as the present study considered only seafood ingestion, and more accurate seafood sourcing data were used here, in which a large portion of the seafood originated from unrestricted fishing areas in the Northwest Pacific, where the impact of the accident was smaller (see “Seafood production and imports of each subarea” in Supplementary Methods).

When compared with exposures from other pathways, doses from terrestrial pathways, such as direct exposure from the ground, were estimated to be orders of magnitude larger than that of our seafood ingestion dose (see “Doses from other exposure pathways” in Supplementary Discussion).

Dose assessment for a representative person

Radiological doses are often reported for a “representative person,” a hypothetical person representing the more highly exposed individuals in the potentially affected population, whose dose can be compared with recommended constraints. Consistent with the approach of the International Commission on Radiological Protection (ICRP)46, the representative person is estimated at the 95th% dose level of the probabilistic results for the potentially affected population of the Tohoku Region. For the first-year exposure following the accident, the contribution to this representative person’s dose from accident-derived radionuclides via seafood ingestion was estimated to be 19 µSv, with reductions in subsequent years (Table 1). Therefore, although there was a slight increase in ingestion dose to the more highly exposed representative person in the Tohoku Region after the accident, it remained orders of magnitude lower than the recommended total dose constraints for members of the public47.

The ICRP has also proposed a deterministic dose assessment method for the representative person. As this method characterizes the more highly exposed group by their habits, it replaces the approach for assessing doses to the “critical group” as in previous recommendations46. Although recent advances in computers and computational codes have enabled probabilistic dose assessments with greater accuracy, the deterministic approach remains common and has been frequently used in recent studies. Povinec and Hirose estimated doses from radiocesium in highly contaminated fish to a hypothetical group assumed to consume greater amounts of fish to be 3000 ± 2000 µSv28. Johansen et al. estimated the dose from accident-derived radionuclides (mainly 134Cs and 137Cs) in 2013 to be 130 µSv to a hypothetical consumer assumed to ignore food restrictions and consume fish from near the FDNPP29. Similarly, a mean dose estimate of 32.6 µSv was made under the assumptions that consumers selected and ate only Pacific bluefin tuna contaminated by the accident and that the intake amount of fish was more than twice the more realistic amount used in this study30.

This study also conducted a preliminary calculation of the ingestion dose using a deterministic approach for comparison. The assessment was assumed to include families of fishermen along the coast of Fukushima Prefecture; however, seafood consumption data for this group is not readily available. Therefore, the 95th% value of seafood consumption for general adults in the Tohoku Region was used. For environmental data, the median radionuclide concentrations of seafood from the “Fukushima coast” subarea were used (Fig. 4). As a result, the ingestion dose during the first year after the accident was estimated to be about 160 µSv per annual intake.

Fig. 4: Definition of areas and sampling locations for 137Cs measurements across all seafood groups in the FAO 61 area after the accident.
figure 4

(i) Map of the main island of Japan. (ii) Map of the coast of Fukushima Prefecture. The FAO 61 area (Pacific, Northwest) was divided into four subareas: red—Fukushima coast, orange—Fukushima offshore, green—Miyagi and Ibaraki, and blue—Rest of the FAO 61. Sampling locations and activity concentrations for 137Cs data are shown as dots. Maps were created using QGIS.

Full size image

These results for hypothetical consumers are higher (in some cases, orders of magnitude higher) than the findings of the present study on the general population using realistic exposures. One of the main reasons for this is that these hypothetical assessments used conservative scenarios, assuming that consumers would be supplied with seafood from restricted areas—contrary to the safety measures in place—and would consume this specific seafood consistently over yearly time periods (Povinec and Hirose used activity concentration data from seafood within 20 km of the FDNPP, whereas Johansen et al. used monitoring data from a station located 3 km from the FDNPP). Deterministic calculations with such hypothetical scenarios provide useful bounding estimates when limited data are available. However, it should be noted that these assessments may result in doses that are substantially larger than the more realistic dose to the representative person calculated using a probabilistic approach.

More detailed information on the references compared here as well as assessments of additional references are provided in “Comparison with other studies on ingestion dose assessment” in the Supplementary Discussion (Fig. S1 and Table S1).

Need for further examinations

This section lists items that may require further examination in the future. For a detailed description of each item, see “Need for further examinations” in the Supplementary Discussion.

  1. 1)

    Since only a few measurements of 131I concentrations were available for this study, the temporal changes in measured 131I/137Cs ratios were used to supplement the 131I data (see “Supplementation of 131I data” in Supplementary Methods). However, various other methods and modeling approaches could be used. For example, a method that estimates 131I concentration using 129I, a long lived isotope, as a tracer has been applied to estimate the distribution of 131I in soil after the FDNPP accident48,49. This method was not used in this study due to limited 129I data in seafood. Other possible methods include estimating seafood 131I concentrations from seawater monitoring data using dynamic modeling or bioconcentration factors.

  2. 2)

    The attenuation of activity due to the delay between the catch and consumption of seafood was not included in the results reported above to allow for comparison with other studies. Such decay is minimal for 134Cs and 137Cs; however, it can be important, especially for the short-lived natural radionuclide 210Po, as addressed in “Delay time between catch and consumption” in the Supplementary Discussion. These analyses suggest that the 95th% total seafood dose reported above (1100 µSv without delay time; Table 1) may be overestimated by ~15–18%, implying a total ingestion dose in the range of 940–980 µSv, with the reduction primarily due to 210Po decay.

  3. 3)

    The consumption data available from the MHLW survey are presented as means and standard deviations (SDs) by region. Therefore, we interpreted these mean values as AMs and performed Monte Carlo calculations assuming normal distributions. The calculated results may differ if seafood consumption does not follow a normal distribution.

Methods

Dose assessment time periods

Large variations in radioactive cesium concentrations in the marine environment around the FDNPP were observed in the years following the accident. In this study, a series of time periods (phases) was defined to enable the determination of summary statistics for radionuclide concentrations as follows:

Phase I: The 10 years before the FDNPP accident (15 March 2001 to 14 March 2011)

Phase II: 0 to 1 month after the accident (15 March to 14 April 2011)

Phase III: 1 to 4 months after the accident (15 April to 14 July 2011)

Phase IV: 4 months to 1 year after the accident (15 July 2011 to 14 March 2012)

Phase V: 1 to 3 years after the accident (15 March 2012 to 14 March 2014)

Phase VI: 3 to 10 years after the accident (15 March 2014 to 14 March 2021).

The start of Phase II was defined as 15 March 2011, when a major release of radioactive materials from the FDNPP occurred. For the post-accident phases (II through VI), the activities of accident-derived and other anthropogenic radionuclides were decay-corrected to the midpoint of each phase (with the exception of 131I; see “131I concentration data” in Supplementary Methods). In “Results” and “Discussion”, the committed effective ingestion dose per annual seafood intake is reported. The shorter Phases II, III, and IV were aggregated to represent the first year after the accident.

Radionuclide concentrations in seafood

This assessment evaluated ingestion doses from radionuclides originating from the FDNPP accident (90Sr, 110mAg, 131I, 134Cs, 137Cs, and 239,240Pu), various other radionuclides released by human activities (3H, 60Co, 99Tc, 129I, and 241Am), and natural radionuclides (210Pb, 212Pb, 210Po, 224Ra, 226Ra, 228Ra, 228Th, 230Th, 232Th, 234U, 235U, and 238U). 40K was not included in this study because it is homeostatically controlled, and 40K ingestion dose remains relatively constant regardless of diet (~165 µSv per year for adults13) (see “Impact of 40K on ingestion doses” in the Supplementary Discussion). Note that 90Sr, 137Cs, and 239,240Pu were present in the marine environment before the FDNPP accident due to human activities (e.g., global fallout); therefore, the post-accident assessment included these background contributions (see “Contribution of radionuclides and seafood groups to the total seafood ingestion dose” in “Results”).

Radionuclide concentrations in five seafood groups (fish, crustaceans, bivalves and gastropods, cephalopods, and seaweeds) were obtained from the Marine Radioactivity Information System (MARIS), a data portal maintained by the IAEA50. MARIS facilitates the search and download of radionuclide measurement results in seawater, marine biota, sediment, and suspended matter, using parameters such as location, radionuclide, and sampling year. All data in MARIS have been compiled from publicly available scientific papers, government reports, and databases. More than 230,000 records from over 260 references were downloaded from MARIS for this study, and the data were screened using methods described in “Data extraction from MARIS” in Supplementary Methods. The dataset after all these processes was provided as Supplementary Data 1.

The activity concentrations in the accident-affected areas were monitored by numerous organizations (e.g., the Fisheries Agency4, Fukushima Prefecture51, the Ministry of the Environment52, and Tokyo Electric Power Company53). A combined dataset shows large spatial variation. To account for this, the data were categorized based on the sample collection locations.

First, the global oceans were divided into two main areas using the Food and Agriculture Organization (FAO) fishing areas32: the FAO 61 (Pacific, Northwest) area, which contains the FDNPP (referred to as “FAO 61”54), and all other areas (referred to as “Others”). The FAO 61 area was further divided into four subareas based on differences in 137Cs concentrations in fish: “Fukushima coast,” “Fukushima offshore,” “Miyagi and Ibaraki,” and “Rest of the FAO 61” (Fig. 4). The Others area was also divided into two subareas: “FAO 27” and “All except FAO 61 & 27.” See “Definition of subareas” in Supplementary Methods for details.

The seafood radionuclide concentration data for the Others area represent the pre-accident “background” for Japan and include all global marine seafood data except for the Northwest Pacific (FAO 61). However, several steps were taken to ensure that this data accurately represented the background for Japan (rather than, for example, Europe). The background was weighted to reflect the sourcing of seafood for Japan, as only 3% comes from the Northeast Atlantic, the fishing area encompassing Western Europe. Additionally, the data prior to the year 2000 were excluded, removing historical measurements of elevated levels from Sellafield discharges into the Irish Sea and 1980s data influenced by fallout from Chernobyl.

When no results were available for accident-derived radionuclides in the FAO 61 area, data from the Others area were used. The seafood supplied to markets in the Tohoku Region can originate from a mix of these areas. Therefore, representative activity concentrations for this study were determined by weight-averaging the data from the subareas based on their respective contributions to the general seafood market supply, considering both production and imports. For more detailed descriptions, see “Seafood production and imports of each subarea” in Supplementary Methods. Radionuclide concentration data for accident-derived radionuclides and other anthropogenic radionuclides in each subarea are provided as Supplementary Data 2.

Restrictions on the distribution and consumption of food

This study considers only seafood available in markets in Japan. After the accident, the Government of Japan imposed restrictions on seafood caught in specific offshore areas of Aomori, Miyagi, Fukushima, and Ibaraki Prefectures. If radionuclide concentrations in seafood exceeded specified levels, then their supply to the market was prohibited. The initial provisional regulation values from the date of the accident to 31 March 2012 were 2000 Bq kg−1 for radioactive iodine, 500 Bq kg−1 for radioactive cesium, 100 Bq kg−1 for uranium, and 10 Bq kg−1 for alpha-emitting radionuclides of plutonium and transuranic elements55,56. The regulation value for radioactive cesium was later changed to 100 Bq kg−1 (effective 1 April 2012)57. Seafood restrictions were defined by species and fishing areas by the Government of Japan, and the list of foods subject to restrictions is available on the MHLW website58.

In this study, radionuclide concentration data corresponding to the species, radionuclides, fishing areas, durations, and activity concentrations regulated by food restrictions described above were excluded. The final activity concentration dataset after all processes described above and in Supplementary Methods (i.e., data extraction, dry/fresh weight correction, treatment of nondetected values, and supplementation of 131I data) is provided as “Supplementary Data”.

After the above compilation and sorting, the GM and geometric standard deviation (GSD) of the activity concentrations of each radionuclide during each phase were calculated and used for the assessment59. Table 4 presents the GM, GSD, and the number of records for accident-derived radionuclides in fish during each phase. Detailed data for all subareas and seafood groups are provided in “Supplementary Data”. Table 3 shows the GM, GSD, and the number of records for natural radionuclides in each seafood group.

Table 4 Activity concentrations (Bq kg−1-fresh weight) of accident-derived radionuclides in fish under food restriction considerations50
Full size table

Seafood consumption

Seafood consumption data for the Tohoku Region were taken from the National Health and Nutrition Survey conducted by the MHLW in 2016, when a large survey was carried out60. Seafood classifications in MARIS and the MHLW survey differ, and correspond as shown in Table 2. Information on seafood origins (FAO 61 and Others) was obtained from Fishery and Aquaculture Statistics9,61 and statistics on the import and export of agricultural, forestry, and fishery products62. Data on the classification of fresh and processed seafood were taken from statistical surveys by the Government of Japan63,64. For these data, a 5-year average (2014-2018) was used. Table 2 also presents the seafood consumption rates calculated using the above information. More details on the method for calculating consumption rates are provided in “Calculation of seafood consumption rates” in Supplementary Methods.

The consumption data in this study represent typical adults in the Tohoku Region. They do not include seafood consumption by infants, children, or adolescents but account for adults who do not eat seafood (a relatively small percentage of the adult population in Japan).

Dose calculation

The committed effective ingestion dose in each Phase \(p\) (\({{{\rm{Dose}}}}_{p}\), Sv) was calculated as follows:

$${{{\rm{Dose}}}}_{p}={\sum}_{a,f,r}\left({{{\rm{Act}}}}_{a,f,p,r}* {{{\rm{Cons}}}}_{a,c,f}* {{{\rm{CF}}}}_{r}* {{{\rm{RF}}}}_{f,r}\right),$$
(1)

where,

  • \({{{\rm{Act}}}}_{a,f,p,r}\): Activity of radionuclide \(r\) in seafood \(f\) caught in area \(a\) during Phase \(p\) (Bq kg−1, Tables 3 and 4);

  • \({{{\rm{Cons}}}}_{a,c,f}\): Consumption of seafood \(f\) in condition \(c\) caught in area \(a\) (g day−1, Table 2);

  • \({{{\rm{CF}}}}_{r}\): Conversion factor from the activity concentration of radionuclide \(r\) to the committed effective ingestion dose for adults (Sv Bq−1)65;

  • \({{{\rm{RF}}}}_{f,r}\): Cooking retention factor of radionuclide \(r\) in seafood \(f\) (See “Cooking effects” in Supplementary Methods).

Subscripts \(a\), \(c\), \(f\), \(p\), and \(r\) denote the catch areas (FAO 61 or Others), the condition of seafood (fresh or processed), the five seafood groups (fish, crustaceans, bivalves and gastropods, cephalopods, or seaweeds), the phases (I-VI), and the 23 radionuclides targeted in this study, respectively. Only adult doses are calculated here.

The dose calculation was performed stochastically using a Monte Carlo method. Ten thousand sets of \({Act}\) and \({Cons}\) values were generated using GSALab66, based on the GMs and GSDs shown in Tables 3 and 4, as well as the means and SDs in Table 2. The calculation was repeated 10,000 times. Details of the assumed distribution are provided in “Assumed distribution shapes for Monte Carlo calculations” in the Supplementary Discussion (Fig. S2).

More detailed methods on data extraction, dry/fresh weight correction, treatment of nondetected values and 131I data, and cooking effects are described in Supplementary Methods.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.