There are many potential impacts of NW discharge, as nuclear contaminants can affect marine ecosystems and human health. On August 24, 2023, the Japanese government officially launched the discharge of NW into the sea, and its impact has received widespread attention from all walks of life (Liang et al., 2024). The National Oceanographic Laboratory Association opposed the plan, citing “the lack of sufficient and accurate scientific data to support Japan’s safety claims” (Normile, 2023). Greenpeace is concerned that Japan’s “Advanced Liquid Processing System (ALPS)” will have difficulty completely treating radioactive substances such as strontium 90, carbon 14, cesium 137, iodine-129, and cobalt-60 in NW, which “easily accumulate in the food chain” and may harm human health over time due to their cumulative intake¹ (Chang et al., 2024). At present, this event has been studied by many scholars, some of them discussing the potential economic impacts of Japan’s NW discharge from the perspectives of regional economies and economic losses (Kenta Tanaka and Shunsuke Managi, 2016, Wu et al., 2023). Some scholars discuss the violation of international environmental law and nuclear safety related laws by NW discharge from a judicial perspective (Chang et al., 2024; Meng and Wang, 2023), and discuss how interested countries and groups can effectively defend their rights through legal channels (Chen and Xu, 2022; Fu and Li, 2024).

However, few scholars have explored the trade impact of Japanese NW discharge; a secondary effect stemming from the direct impacts on the marine environment and marine life, which in turn affects the production and trade of aquatic products in various countries. Dadakas and Tatsi (2021) used a gravity model and Poisson Pseudo Maximum Likelihood estimator to assess impact of the Fukushima accident on agricultural trade flows. However, the study overlooked the impact of pollutants on other countries, and their temporal diffusion which will directly affect the production of aquatic products. Nuclear pollutants may reach the northeast Pacific Ocean within one year ² , and will cover almost the entire North Pacific in more than three years, and cover the entire Pacific Ocean in ten years (Liu et al., 2022). Although the impact of nuclear pollutants on fishery production is still uncertain, after the accident at the Fukushima Daiichi Nuclear Power Plant (FDNPP) in Japan on March 11, 2011, scientists tracked and studied fish in the nearby waters. From April 2011 to April 2012, more than 40% of fish were found to have radiation levels exceeding Japanese regulatory limits (Buesseler, 2012). Five years later, 0.05% of samples in the area exceeded this limit (Wada et al., 2016). There is therefore a factual basis for assuming a decrease in AQ production due to NW release. Any triggering of relevant international agreements, such as the United Nations Convention on the Law of the Sea could have significant implications for the management of marine pollution and international trade for Japan and other coastal countries.

NW discharge has an impact on AQ trade not only through the potential for radiation contamination levels exceeding regulatory thresholds, but also through consumers’ preferences over affected or potentially affected aquatic products and individual countries’ efforts to protect their consumers. According to the empirical research of McKendree et al. (2013), 30% of survey respondents reduced their consumption of seafood after the Fukushima nuclear power plant accident, and more than 50% of respondents believed that the nuclear disaster had posed a risk to consumer health in Asia. Many governments and citizen groups took strong measures based on these concerns. The General Administration of Customs of China enacted a complete suspension of Japanese AQ imports to the Chinese Mainland. Hong Kong and Macao have banned imports of AQs from 10 prefectures in Japan ³ . Malaysia implements Level 4 inspection on high-risk food imported from Japan ⁴ . The 18-nation Pacific Islands Forum has long expressed concerns about the possible impact of NW on the fishing industry (McCurry, 2023). In addition, citizens in both Japan and South Korea, as well as South Korean lawmakers, have strongly protested these releases ⁵ (Li and Li, 2022). According to a Gallup poll conducted in June 2023, 78% of South Korean respondents said they were very or somewhat concerned about the possible damage to marine life (McCurry, 2023). Although there is no exact data on the scientific impact of NW, these negative public opinions will affect the consumption of AQs (Wang et al., 2022), which in turn will affect the trade pattern of AQ. Wang et al. (2022) found that public opinion on Japan’s actions affect not only the trade volume of Japanese AQs, but also volumes of both imports and exports of Japan’s trading partners. Combined with the current status of China’s AQ industry and foreign trade competitiveness, the discharge of NW will promote China’s AQ exports (Hai, 2021). The economic impact also varies across countries, with negative impacts on Japan and the United States contrasted against increased demand for exports benefitting China, India, Indonesia, as examples (Wu et al., 2023). However, compared with AQ, terrestrial animal protein production systems have lower efficiency, exacerbating land use tension (Napier et al., 2020), threatening nutritional security, and expanding greenhouse gas emissions (Zhang et al., 2023). A few studies believe the discharge of NW will have multiple negative impacts, including the potential to shift demand towards terrestrial animal protein production systems with lower efficiency and exacerbating land use tensions (Napier et al., 2020), threatening nutritional security, and expanding greenhouse gas emissions (Zhang et al., 2023). Assessing ST and LT production impacts on AQ trade can provide a clearer understanding of potential impact of Japan’s discharge of NW on major trade partners.

Scholars are increasingly applying social network methods to study structural changes in the global trade network after large shocks (Gutiérrez-Moya et al., 2021; Kosztyán et al., 2024). Understanding the characteristics of the current global AQ trade pattern is the foundation for grasping the impact of nuclear waste discharge. Although a broad range of methodologies to model trade exist, including agent-based or dynamic system approaches, we employed the maximum entropy method to understand the trade impact without making too many assumptions. Agent-based modelling is challenged by difficulties in model construction, verification, validation, computation, and generalization of findings (An et al., 2021). Complex networks (CN) are a very effective tool for studying international trade flows and the stability of trade network structure (Allard et al., 2017; Serrano and Boguná, 2003). Gephart and Pace (2015) used CN methods to analyze the structure and evolution of AQ trade from 1994 to 2012, including quantifying indicators of AQ globalization, changes in bilateral trade flows, changes in centrality, and comparing AQ trade networks with agricultural and industrial trade networks. This provides a reference for further research in this article, which expands the period to 2000-2022, updates the latest trade status of nodes, analyzes the evolution characteristics of network structure since the 21st century, and analyzes the characteristics of AQ trade in response to shocks. The discharge of NW has brought new shocks to the trade relations of AQ trading countries. For reconstructing the trade relations between countries after changes in supply, the maximum entropy method can accurately provide the most uniform and least risk-predicted probability distribution, which is also the most objective allocation for participating countries (Wang et al., 2021). The maximum entropy method analyzes system risks by reconstructing unknown network connections (Ikeda and Watanabe, 2017), and trade across participating countries is assessed utilizing a uniform distribution to minimize the subjective assumptions around trade linkages after changes in supply distributions (Wang et al., 2021). Recent studies that integrate maximum-entropy and econometric approaches in gravity models, providing a more nuanced understanding of trade networks (Di Vece et al., 2022). Here, reallocation is assessed by constructing a network using the degrees and strengths of participating AQ trading countries with minimal assumptions on unknown information. The reallocation model uses trade weights to evenly distribute risk across trading partners, achieving the minimum global supply risk.

This article constructs an AQ trade distribution model based on the global AQ trade network and the principle of maximum entropy. It simulates the possible trade reallocation among countries under the ST and LT impacts of NW, helping trading countries understand potential impacts from Japan’s discharge and formulate corresponding trade strategies.

This study analyzes the world AQ trade system using data from the United Nations Trade Database from 2000 to 2022. AQs include two categories: primary products and finished products. According to the Harmonized System classification method in the United Nations Trade Database, AQs are defined as follows: 0301 (Fish; live), 0302 (Fish; fresh or chilled, excluding fish fillets and other fish meat of heading 0304), 0303 (Fish; frozen, excluding fish fillets and other fish meat of heading 0304), 0304 (Fish fillets and other fish meat (whether or not minced); fresh, chilled or frozen), 0305 (Fish; dried, salted or in brine; smoked fish, whether or not cooked before, or during the smoking process), 0306 (Crustaceans; in shell or not; in shell, steamed or boiled, whether or not chilled), 0307 (Molluscs; whether in shell or not; smoked molluscs), 0308 (Aquatic invertebrates other than crustaceans and molluscs; smoked aquatic invertebrates other than crustaceans and molluscs), 0309 (Flours, meals and pellets of fish, crustaceans, molluscs and other aquatic invertebrates, fit for human consumption), 1603 (Extracts and juices of meat, fish or crustaceans, molluscs or other aquatic invertebrates), 1604 (Prepared or preserved fish; caviar and caviar substitutes prepared from fish eggs), and 1605 (Crustaceans, molluscs and other aquatic invertebrates). The data mainly focuses on import data, which measures trade volume based on the cost, insurance, and freight price. However, when import data is missing, export data is used as a supplement. To ensure network connectivity while basing simulations on the broadest definition of global trade, the analyses removed isolated nodes: that is, countries with no trade connections. The annual number of countries or regions is 236-242, covering almost all countries. To link the analysis to the impact of nuclear waste, this study obtains data on total aquatic production, marine catch, and marine aquaculture from the FAO Fisheries and Aquaculture FishStat database for each country.

The basic structure of the global aquatic product trade network (AQTN) model can be abstracted as the connections between various economic participants. Each node in the network represents a country or region, and network links are weighted in a directed manner, representing the import (export) amount between countries. In this article, the trade amount (in US $) is used as the trade flow, and the link has direction and weight. The direction represents the import and export relationship, and the weight represents the import and export amount of aquatic products. In the model matrix, rows represent exporting countries and columns represent importing countries. Here, W is defined as a weight matrix, where the elements Wij represent the export value of aquatic products from country i to country j. An adjacency matrix A is constructed by setting A_ij = 1 if W_ij >0 and A_ij =0 if W_ij =0.

The analysis results show that from the perspective of unweighted export and entry of AQs trade network, the Chinese Mainland has the largest number of export contact countries in 2022, with 182 export countries or regions, followed by France, the United Kingdom, the Netherlands and India, while Hong Kong and Macao have 52 and 15 export destinations respectively. The United States has the most diverse sources of imports, with up to 157 sources, followed by Canada and France. Hong Kong ranks fourth in the world in terms of the number of countries that import AQs, reaching 133, while Macau and the Chinese Mainland have made progress at 97 and 93, respectively, ranking 21^st and 26^th globally. This indicates that Hong Kong and Macau are important pivot points for the diversified development of China’s AQ imports. The data from the weighted trade network of AQs shows that China, Norway, Chile, India, Canada, Spain, the United States, the Netherlands, Sweden, and Denmark are the top ten global exporters of AQ, with the largest importing country being the United States, followed by China, Japan, and Spain, as shown in Figure 2 .

As presented in Table 4 , the betweenness centrality ranks highest for the Unted States, Canada, France, the Netherlands, and Germany while the Chinese Mainland and Hong Kong rank eight and fifteenth, respectively. These countries have a large amount of information flow in the global AQ trade network, and they are more likely to play the role of trade adjustment intermediary countries when trade changes. For reference, the United States’ highest ranked betweenness centrality is three times larger than Singapore, which ranks 12th.

Regarding the supply and consumption status of AQ, the United States, Chinese Mainland, Japan and other regions are essential “consumers” of imports. They are not only big importers, but also the supply capacity of their import source countries is strong, and consumer demand is satisfied through robust supply linkages. Canada, Chile, India, Chinese Mainland and other regions are important exporters. Although Canada and Chile rank fifth and third in the global export share, they are in a special position because of their relationship with the United States. Although China’s export volume of AQs is large, it ranks third due to its exports are through relatively weak linkages to other countries. It confirms that weighted HITS authority and hub values can reflect the fact that the structure of the world trade network influences a node.The authoritative center value of the United States, at 0.243, is nearly ten times that of Germany. Although both countries are important consumer countries in the top ten globally, their importance and status differ significantly. Compared to others, the difference in central values among the top fifteen hub nodes is relatively small, indicating a greater disparity in the importance of consumption among each node.

Since the 21st century, ties between countries engaged in global AQ trade have been growing: from 8245 in 2000 to 10237 in 2018. After the outbreak of the COVID-19, the trade links of AQs have dropped sharply to 9724. After the outbreak of the Russia-Ukraine conflict in 2022, the trade links have again dropped significantly to 8424. It can be seen that the impact of the COVID-19 plus the Russia-Ukraine conflict has had a significant impact on the trade links of AQs (Jagtap et al., 2022). The results of the network tightness index also support the growing tightness of the trade network of AQs, but under the dual impact of COVID-19 and the Russia-Ukraine conflict, the links between nodes have become loose. From the perspective of the development of a network clustering coefficient, based solely on trade connections, global AQ trade continues to strengthen regional trade. However, considering the weight of trade volume, this trend was broken after 2018. The reversal of this development model may have been affected by the trade friction between China and the United States, and global AQ trade has shifted towards a direction of continuous globalization, which will continue to strengthen in 2021 and 2022. Therefore, to a certain extent, the trade network of AQs can be considered globalized when subjected to external shocks, when the trade clustering coefficient degree loosens (Table 5).

To highlight the changes in major trading countries and major trading relationships, this article retains the 2022 AQ trade network of the top 95% of trade flows, which includes 132 countries. In order to more accurately measure trade changes, the parameters β in Equation 12 are adjusted before simulation to obtain the trade flow between countries that are closest to the true trade under maximum entropy. This reduces the variation in transaction volume caused by the original maximum entropy of trade. Therefore, after calculating the maximum entropy value of the original data, the trade flow after maximizing entropy is used as the original data for analysis of transaction redistribution under different scenarios. Through regression analysis and comparison of bilateral trade relationships and β models, when β is 0.01 (with 10,000 iterations), the entropy maximization redistribution of trade relationships is closest to the true trade value ( Figure 3 ).

Therefore, from a global perspective, based on the previously constructed AQ redistribution model, we can obtain the optimal redistribution strategy for global AQ trade after any country experiences supply risks or inter-state changes in trade levels. This article takes the practical impact of China’s comprehensive suspension of imports of Japanese AQs, as well as the scenario impact of the proportion of marine AQs in the AQs of major trading countries on supply, correspondingly reducing exports, and further analyzing the short- and long-term effects on the production of AQs in various countries as an example, to analyze how to redistribute global AQs to reduce losses caused by risks (Chang et al., 2024).