Mapping the unjust global impact of harmful sheries subsidies

Harmful sheries subsidies contribute to overshing leading to environmental and societal impacts1. If only sheries within the subsidising nations’ jurisdiction were affected, then unilateral actions might be sucient to help safeguard our ocean and the people reliant upon it. However, just as sh move between jurisdictions2, so too do the subsidised shing eets targeting them3. As such, the impacts and solutions to subsidies-induced overshing are matters of international concern. Mapping that impact is therefore key to understanding these concerns and informing multilateral reform. Here we combine existing datasets4–6 to quantify the amount of harmful sheries subsidies impacting the high seas, domestic and foreign waters, respectively. We estimate that between 24% and 43% of all harmful sheries subsidies impact foreign waters or the high seas. We show that harmful subsidies primarily originate from countries with high-Human Development Index (HDI), strong sheries management capacity and relatively sustainable sh stocks, yet disproportionately impact countries with low-HDI, lower management capacity and more vulnerable stocks. Indeed, over 40% of the harmful subsidies impacting low-HDI countries originate from high-HDI countries. This discrepancy between the source of harmful subsidies and the nations that are ultimately impacted is unsustainable and unjust. Policy-makers from all nations must push for effective multilateral subsidies reform. Prohibiting subsidies to distant-water shing should be prioritised to support equitable and sustainable sheries worldwide.


Introduction
Concerns regarding the subsidisation of the sheries sector are centuries old 7 . Those concerns increased when, in 1992, the Food and Agricultural Organization of the United Nations (UN) rst estimated that the annual amount of sheries subsidies provided by governments globally was likely to be US$ 54 billion 8 .
Subsequently, more comprehensive estimates of the extent and impact of this practice cemented sheries subsidies as a key concern for the conservation and management of marine ecosystems 1,9 and, more recently, for supporting equitable use of marine resources 5,10 . Consequently, sheries subsidies reform is widely regarded as a necessary step towards safeguarding our ocean and the people that rely upon it [11][12][13] . Concerted efforts to achieve such reform, however, have continued, unsuccessfully, for more than twenty years. During this time there have been numerous international commitments to formally discipline sheries subsidies, most notably via the Convention on Biological Diversity, the World Trade Organization (WTO) and the UN Sustainable Development Goals. Yet, international agreement on rules for sheries subsidies remain elusive. Further delay jeopardises the progress needed to achieve sustainable and equitable sheries, which underpin broader efforts to protect biodiversity, reduce poverty, provide nutritious food and secure livelihoods 14,15 .
Although not all sheries subsidies are harmful 16 , the majority of those currently provided (e.g., subsidies for fuel, vessel construction or modernisation) are thought to cause harm 1 . Fisheries subsidies can distort markets 17 and contribute to unfair trade practices 18 , hinder international sheries cooperation 19 , exacerbate inequity by undermining the viability of small-scale producers 20 , lead to higher CO 2 emissions 21 , act as a driver for illegal shing 22 , and contribute to over shing by increasing shing eet capacity 23 . This 'capacity-enhancing' nature of sheries subsidies sets them apart from most other forms of sectoral subsidisation, because increasing shing capacity by its very nature directly increases the risk of additional pressure on the marine ecosystems from which subsidised shing eets are catching sh 16 . Importantly, the global distribution of this impact is largely unknown. Furthermore, much of the data that are disclosed are shrouded by a lack of transparency 24 and con ned to estimates of the amounts of subsidies provided by each country 25 , rather than focusing on the sheries and locations (e.g., domestic versus foreign waters) that these subsidies are impacting.
If harmful sheries subsidies only affected the sustainability and viability of sheries within the national jurisdictions of the subsidising nation, then unilateral actions might be su cient to help safeguard our ocean and the people who rely on it. However, the evidence is clear: sh inhabit and move between multiple jurisdictions including the high seas 2,26 , and shing eets -and their impacts on seafood production-are increasingly transnational 27 and transboundary too 28,29 , largely due to the expansion of subsidised 'distant-water' shing eets 3 . As such, the environmental and societal impacts of shing and the heightened risk of over shing due to the provision of harmful sheries subsidies are a potential source of sheries con ict 30 and a matter of international concern.
However, while sheries science and management divides the world's sh and shing eets into shared and non-shared, domestic and distant-water, respectively, such that the onus for their management lies predominantly either with nation-states or the international community 31 , such distinctions are currently lacking for the impact of harmful sheries subsidies. Estimates now exist of the subsidies provided by almost all of the world's maritime nations 4 and the proportion being directed towards large-scale shing eets that are capable of operating outside of domestic waters 5 , but we are yet to successfully assign the ows and impacts of harmful sheries subsidies in a similar manner. Therefore, building on existing global datasets, we estimate the amount of harmful sheries subsidies that are impacting domestic waters, foreign waters (the Exclusive Economic Zones (EEZs) of other countries), and the high seas. We also estimate the extent to which different regions and countries are contributing to and being impacted by the global subsidisation of marine shing. In doing so we map, for the rst time, the global impact of harmful sheries subsidies.

Results
We nd that of the estimated US$ 22.2 billion of harmful sheries subsidies provided to the world's shing eets in 2018 1 , some US$ 6.0 (± 1.6) billion is likely impacting foreign waters within the EEZs of other countries, and US$ 1.4 (±0.6) billion is impacting the high seas. The remaining US$ 14.9 billion (±4.0) supported domestic shing within the EEZs of the subsidising countries, which consists of both small-scale and large-scale vessels 5 ( Figure 1). Therefore, between 24% and 43% of all harmful subsidies are impacting areas outside of the jurisdiction of the original subsidising country; between 20% and 34% is impacting areas within the jurisdictions of foreign nations, and between 4% and 9% is impacting the high seas.
Such a signi cant proportion of harmful subsidies impacting foreign waters and the high seas means that the provision and subsequent impact of harmful sheries subsidies are not equally distributed across geographies. We nd that Asia, Europe and North America, provide more harmful subsidies to their shing eets than their respective regional ecosystems are impacted by (Figure 2.a). As such, these regions are net subsidy-sources. Conversely, marine ecosystems within Africa and Oceania are net subsidy-sinks-meaning that their waters are impacted by more harmful subsidies than are provided by the countries within those regions (Figure 2.a). Oceania is impacted by more than two times (221%) the amount of harmful subsidies than their constituent countries provide, while Africa is impacted by almost two times (177%) as much. Many of the additional subsidies impacting African waters originate from Asia and Europe; the ows of the impact of subsidies to Africa from these regions are US$ 654.7 (± 188) million and US$ 541.1 (± 119) million, respectively (Figure 2.b).
We also identify individual countries as net subsidy-sources of harmful sheries subsidies (i.e., countries that provide more subsidies than they are impacted by) from the combination of domestic and foreign subsidies (the reverse are net subsidy-sinks). The majority of the largest net subsidy-sources are Asian and European shing countries (Table 1.a; see Table S.3 for all countries). For example, China's EEZ is impacted by US$ 3.5 (± 0.9) billion of harmful subsidies from both domestic and foreign sources, yet China provides US$ 5.9 (± 1.4) billion to their shing eets, meaning that their EEZ is impacted by 60% of the value of harmful sheries subsidies they provide. Spain provides US$ 682.8 (± 130.7) million to their shing eets, but is only impacted by US$ 209.6 (± 36.1) million, meaning that their EEZ is impacted by 31% of the value of harmful subsidies that they provide. Other key net subsidy-sources include Thailand and Taiwan, whose EEZs are impacted by approximately 36% and 40% of the value of harmful sheries subsidies that they provide, respectively.
Conversely, other countries are net subsidy-sinks (Table 1.b; see Table S.3 for all countries). Japan's EEZ is impacted by more than US$ 3.0 (± 0.7) billion, or 144% of the value of harmful sheries subsidies that they provide to their own shing eets (US$ 2.1 ± 0.6 billion). A handful of other high income nations are net subsidy-sinks, including the United Kingdom and the Russian Federation, which are impacted by 552% and 111% of the value of harmful sheries subsidies that they provide, respectively. In these cases the, the difference between the amount of subsidies they provide and the amount they are impacted by are likely due to the presence of signi cant (and often reciprocal) access agreements between neighbouring countries, for example for Japan's EEZ is signi cantly impacted by harmful subsides originating from China, while the United Kingdoms' is signi cantly impacted by harmful subsidies originating from Norway and the Netherlands. Yet, the majority of the largest net subsidy-sink countries are low-HDI countries with much larger disparity between subsidy provision and subsidy impacts, such as Morocco (305%), Indonesia (158%), Guinea (1,016%), Guinea-Bissau (164,336%) and Somalia (33,533%).
Following previous global sheries subsidies studies 1,32 , we present the overall provision and impact of harmful sheries subsidies by clustering countries using various metrics of human developmental status 33 , sheries management capacity 4 , and relative stock status 34 ( Figure 3). These metrics act as de facto indicators of a country's potential resilience against the impacts of harmful subsidies. We nd that while countries with high Human Development Index (HDI) scores provide 83% of the world's harmful subsidies (US$ 17.7 billion), they are impacted by 74% (US$ 15.4 billion) of the global total. Conversely, countries with low HDI scores that provide approximately 17% of the world's harmful subsidies (US$ 3.4 billion), are impacted by more than 26% (US$ 5.7 billion) of the global total (Figure 3.a). Indeed, over 40% of the harmful subsidies impacting low-HDI countries originate from high-HDI countries. We see a similar ow of the impact of harmful subsidies when considering countries clustered by the amount of bene cial subsidies that they provide (Figure 3.b) and the overall status of the sh stocks within their EEZs ( Figure  3.c). Bene cial subsidies refer to government support towards sheries management, enforcement and research, as well as the implementation and maintenance of marine protected area 4 . Countries that provide high levels of bene cial subsidies provide more harmful subsidies than their EEZs are ultimately impacted by. Similarly, countries with relative high stock status also provide more harmful subsidies than their EEZs are ultimately impacted by. Whereas those countries that provide low levels of bene cial subsidies and have relative low stock status are impacted by more harmful subsidies than they provide.

Discussion
Largely due to the global extent of subsidised distant-water shing, the impact of harmful sheries subsidies is being felt by all the world's coastal nations, regardless of the amount of subsidies they provide to their own shing eets. We estimate that between 24% and 43% of all harmful sheries subsidies are impacting either foreign waters (20-34%) or the high seas (4-9%). The damage that harmful subsidies provided to distant-water shing eets cause, by increasing the risk of over shing, is therefore impacting the ecosystems, economies and societies outside of the original subsidising country's national jurisdiction. This demonstrates that harmful sheries subsidies are a matter of international concern and not an issue that can be resolved by unilateral action alone 35 .
Previous studies show that distant-water shing, including that within the high seas, is almost exclusively conducted by a handful of rich nations 3,36 and that the majority of their activity occurs within the EEZs of low-HDI nations 19,37 . However, few, if any, studies have quanti ed the subsidies being channelled towards this activity. We show that there is a disproportionate ow of harmful subsidies from high-HDI countries but impacting low-HDI countries-more than 40% of the harmful subsidies impacting low-HDI countries originate from high-HDI countries. The risk of over shing, therefore, is being disproportionately exported to low-HDI nations which further entrenches global inequities, poverty and malnourishment, and makes the achievement of interconnected UN Sustainable Development Goals less likely 14 . This study corroborates previous ndings 38 that demonstrate that high seas shing, a sub-component of distantwater shing, is heavily subsidised and therefore raises questions regarding the economic viability of high seas shing 39 , and whether this activity would be pro table at all without subsidies 40,41 .
While these disproportionately impacted nations may bene t from fees and clauses included within access agreements, such as clauses that ensure the distant-water shing nation processes a certain amount of catch in the host nation or employs a certain number of local people on board their shing vessels 42 , evidence suggests that only minimal compensation is received and that the terms of these access agreements are often unfavourable for hosts 43 . Therefore, while the host nations contend with the negative ecological, economic and social consequences that harmful sheries subsidies impose, most of the bene t derived from the subsidised foreign shing eets is likely captured by the distant-water shing nation 44 . High-HDI countries use their capital to gain access to resources and subsequent revenues from catches taken from low-HDI countries, and may be less incentivised to sh sustainably because they do not feel the direct impacts of over shing 38 . Another reason to address the increased risk of over shing by subsidized distant-water shing nations is that many developing nations also have aspirations to develop their own sustainable and equitable blue economies that provide livelihoods and economic bene ts, which require sh stocks to be rebuilt and maintained at a sustainable level, and for the bene ts from those sheries to be retained locally 45 . Removing foreign harmful subsidies from their waters would represent a key step towards achieving this goal in the long-term 46 .
In addition, there is a net ow of harmful subsidies originating from countries with stronger sheries management capacity and relatively sustainable sh stocks, towards countries with lower sheries management capacity and relatively poor status sh stocks ( Figure 3). This phenomenon is critical because having limited sheries management and enforcement capabilities results in a lack of regulation and control that might otherwise mitigate the impact of harmful subsidies 12 . Similarly, as the greatest threat from harmful subsidies is the heightened risk of over shing, sh stocks with relatively poor stock status are at the greatest risk. Conversely, vast sums of public money that are spent by many high-HDI countries on bene cial subsidies invariably go towards protecting and rebuilding the sh stocks within their own domestic waters, while the same countries simultaneously use public money to erode the sh stocks, livelihoods, and food security of low-HDI countries 47 . As such, the impacts of and rewards from subsidising shing are not equally shared. This imbalance highlights the structural inequities that are baked into the global practice of providing harmful sheries subsidies, and further exacerbates the risk that harmful sheries subsidies pose towards collective attempts to rebuild and sustain marine biodiversity across our ocean.

Concluding Remarks
This research provides, for the rst time, a snap-shot of the current global ows and impacts of harmful sheries subsidies, taking into consideration the fact that subsidised shing eets operate across multiple national jurisdictions and regions, including the high seas. It highlights the complexity and interconnected nature of distant-water shing and of the provision of harmful sheries subsidies.
Understanding the sources and ultimate destinations of the impact of harmful sheries subsidies through this dataset provides evidence to support multilateral sheries subsidies reform and for measuring the impact of such reform.
Our conclusion is that subsidies provided to shing eets operating outside of the source-countries' EEZs should be prioritised for removal, particularly when they operate in the high seas or the EEZs of low-HDI countries or countries with limited sheries management capacity and/or poor stock status. Such removal would ensure that the onus for managing the impact of any remaining harmful sheries subsidies lies predominantly with the subsidising country themselves. This would begin to redress the unjust global impact of harmful subsidies on the sheries that provide essential food and livelihoods to millions of people, particularly in low-HDI countries 48 . Instead, we have seen some key subsidising regions propose the reintroduction of harmful sheries subsidies that have already been removed 49,50 , in direct contradiction of sheries science, economic theory and insights.
While ongoing negotiations of sheries subsidies rules are necessary and any consensual agreement would represent a signi cant step in the right direction, it is unlikely to address the ongoing and historic inequity and unfairness of the demonstrable transfer of the burden of harmful subsidies to the waters of low-HDI nations. The most impacted regions require signi cant funding to help manage and rebuild their sh stocks in order to effectively undo the damage caused by the continued provision of harmful sheries subsidies to shing eets that operate outside their own waters. However, recent studies show that there is a signi cant gap in current ocean nancing 51 . Moreover, net subsidy-sink countries will need to nd ways of generating alternative funds if they lose the revenues from foreign shing access agreements, which for some countries, are a major contributor to national Gross Domestic Product.
Redirecting the vast sums of public money currently being used to potentially support over shing or to reduce the costs of fuel consumption, for example, towards more equitable coastal development and better sheries management and enforcement 52 would be a step in the right direction.
Main References 17. Kaczynski, W. Joint ventures in sheries between distant-water and developed coastal nations: an economic view. Ocean Management 5, 39-48 (1979).    The approach requires two key overarching assumptions. The rst is that the distribution of the relative impact of harmful sheries subsidies provided to the large-scale shing sector of any country is proportional to the distribution of landed value from that country's large-scale shing sector. This is a reasonable assumption given available information regarding the costs and bene ts of sheries in local versus distant waters and targeting lower-value versus higher-value species, and necessary given the dearth of data at the global scale and the limitations of the data that is available. Our de nition of the harm caused by subsidies is underpinned by the biological sustainability of the sheries. Harm is caused when arti cial reductions to the cost of shing or increases to the revenues from shing lead to overcapacity and subsequently to over shing, reducing opportunities to pro t from sustainably-managed stocks. We therefore argue that it is justi able to estimate the distribution of impact of harmful subsidies based on the value derived from the sh being extracted.
The second assumption is that the negative impact of harmful sheries subsidies is uniform and linearly related to the amount of subsidy being provided. That is, any sheries subsidy that is deemed to be harmful in its nature will have the same degree of impact. Here, we temporarily set aside the status of the sh stocks being shed (e.g., certain harmful subsidies would have a more detrimental effect on sh stocks that are already over shed) in order to allow us to consolidate all harmful subsidies and simplify the calculations at the global scale. While these assumptions will not stand true across all examples of sheries subsidies provision, they are deemed appropriate and necessary for the scale of this analytical study and its central research question. These assumptions do however require clear de nitions of what is considered to be a harmful sheries subsidy and the speci c eet segments (or regions of the ocean) to which subsidies are provided (or impacting).

De ning sheries subsidies
Various research groups and organisations have de ned and categorised sheries subsidies differently e.g. 54,55 . We take as the basis for our de nition the WTOs Agreement on Subsidies and Countervailing Measures, which de nes a subsidy as any direct or indirect nancial contribution by a government or any public body that confers some kind of bene t to the private sector. This includes grants, loans, equity infusions, loan guarantees, scal incentives, the provision of goods or services and the purchase of goods. In addition, we follow Sumaila et al. 32 , who categorises sheries subsidies as either 'harmful', 'bene cial' or 'ambiguous' in their nature, based on the subsidy's likely impact on sh stock sustainability over time. Harmful subsidies, the focus of this study, are broadly de ned as any subsidy that arti cially increases revenue or reduces the costs of shing and include support for vessel construction, renovation and modernisation, tax exemptions, fuel subsidies, port construction and investment in marketing and processing infrastructure 4 .

De ning shing eet sub-sectors and segments
For the purpose of this study, each country's shing eet is divided into two broad sub-sectors-the smallscale shing sub-sector and the large-scale shing sub-sector-and further divided into four eet segments-the domestic small-scale eet, domestic large-scale eet, foreign large-scale eet and high seas shing eet. We use the Sea Around Us de nitions as our starting point for shing eet de nitions 6 .
The small-scale shing sub-sector includes artisanal, subsistence, and small-scale commercial and noncommercial sheries, but excludes recreational shing activities. Some maritime countries provide their own de nition of small-scale shing and these are used where available (see 5 for detail). We make the reasonable assumption that the small-scale shing sub-sector only operates within domestic waters (i.e., in their country's EEZ, <200 NM from shore). There are a few exceptions to this assumption, however, the total amount of sh caught by this sub-sector in non-domestic waters is negligible 6 .
The large-scale shing sub-sector, including industrial and semi-industrial sheries, includes all other shing activities that are not included within the small-scale shing sub-sector de nition. This usually consists of large vessels with xed and/or mobile shing gears operating within a country's EEZ, and also includes all activities taking place outside of a country's own EEZ. All large-scale shing eets are assumed to engage in commercial shing activities. There currently exists no single de nition for different shing eet segments that is applicable across all countries 56 . Our de nition of the large-scale shing eets is divided into three eet segments-although in practice, individual vessels may operate across multiple eet segments and may be agged to a different nation than they originate from: • Domestic large-scale eet. This segment includes the catch of any vessel that is not considered to be small-scale made from within the EEZ of the maritime country under which the vessel is agged; • Foreign large-scale eet. This segment includes the catch of any vessel that is made from the EEZ of another maritime country (excluding overseas territories) other than the country under which the vessel is agged; and • High seas eet. This segment includes the catch of any vessel that is taken from either the high seas or any area beyond national jurisdiction (>200 NM from shore).
Calculating the distribution of the impact of sheries subsidies In order to apportion every dollar of harmful sheries subsidies provided by each maritime country as impacting either their domestic waters, foreign waters (another country's EEZ) or the high seas, we followed the steps outlined here: First, under the assumption that small-scale sheries only operate within their own domestic EEZ, we used estimates for the proportion of harmful sheries subsidies provided to the small-scale and largescale shing sub-sectors for each maritime country provided by Schuhbauer et al. 5 . All harmful subsidies provided to small-scale shing sub-sector were therefore categorized as domestic, while all harmful subsidies allocated to large-scale sheries were further divided into the three spatially discrete eet segments (domestic, foreign and high-seas).
Second, to apportion the percentage of harmful sheries subsidies for the large-scale sheries as impacting either a) domestic waters; b) foreign waters (another country's EEZ); or c) the high seas, for each country, we used the distribution of the landed value of catch 6 broken down by each individual EEZ and the high seas. To do this, we multiply the total amount of harmful sheries subsidies provided by a country to its large-scale sheries by the proportion of landed value that the large-scale shing sub-sector took from a particular region or EEZ, see equations 1 and 2:
For each individual EEZ (and the high seas), we then added the harmful small-scale sheries subsidies from Schuhbauer et al. 5 as domestic subsidies, with the domestic harmful large-scale sheries subsidies, with all the foreign harmful subsidies estimated to be impacting that same EEZ, to complete the analysis.
As these calculations are estimates and not absolute values, the relative quality of the underlying data may affect the robustness of the ndings. In order to re ect this uncertainty in our calculations, we combine data quality scores for the landed value estimates and the harmful sheries subsidies estimates to produce a compound data quality score for each subsidy-EEZ calculation in the dataset. For landed value, we use the weighted average reliability scores taken directly from the Sea Around Us catch data 6 .
Harmful subsidies estimates from Sumaila et al. 4 , however, do not include data quality scores. We therefore revisited the original data for all harmful sheries subsidies and estimated data quality scores by adapting the method used for from the Sea Around Us catch data 6 . The harmful sheries subsidies data are broken down into seven different subsidy types. For each country and each subsidy type there is a record describing whether the estimate was 'reported' (i.e. taken directly from a cited source) or ' modelled' (i.e. calculated using the value transfer model applied in Sumaila et al. 1 ). The quality of the harmful subsidies data for each country was therefore determined based on how many subsidy types had been 'reported' versus 'modelled'. For example, if all data points were 'reported', the quality score applied was 4 (very high data quality), if 'reported' data points were less than 25% of overall data points the quality score was 1 (very low) ( Table 2). We averaged the subsidies data quality scores by shing country with the weighted catch quality scores for each EEZ (and the high seas) to produce a compound score between 1 and 4 for each subsidy-EEZ calculation. The compound quality score for each EEZ was calculated as follows: Where Compound qs EEZ is the quality score for a speci c EEZ, qs Si denotes the subsidy quality score for country i.
The compound score was then converted into con dence intervals to present our nal results in ranges rather than absolute values (Table 2). provided from one country and impacting either the same country or another country (or the high seas). In order to estimate the total domestic impact for each EEZ, we combined the amount provided to a country's small-scale eet with the total estimated to have been provided to the large-scale shing eet shing within its own EEZ. The cumulative impact of harmful subsidies on a single location, such as an individual EEZ, was then calculated by summing the total harmful subsidies impacting that location from all countries, including all subsidies to foreign large-scale and domestic small-and large-scale eets.
Following previous global sheries subsidies studies 1,32 , we present an analysis of countries clustered using the 2017 UN Human Development Index (HDI) as an indicator of the development status, not only economic growth, of a country 33 . Countries were clustered into two groups, high and low, based on the second quartile. We also used the same categorisation approach for clustering countries based on the relative amount of bene cial sheries subsidies they provide, taken from Sumaila et al. 4 , as an indicator of the sheries management capacity of a country. Bene cial subsidies include funding for sheries management programs and services, sheries research and development, and marine protected areas.
Finally, we used the Environmental Performance Index for sh stock status, as an indicator of the relative health of the stocks within a country's EEZ. This indicator uses data from the Sea Around Us 6 to present a percentage of a country's total catch that comes from overexploited or collapsed stocks, considering all sh stocks within a country's EEZs. A score of 100 indicates that none of a country's sh catch come from stocks that are overexploited or collapsed, and a score of 0 indicates worst performance. We clustered all countries into the two groups for each of the three indicators, and presented them using Sankey plots, to understand the ow of harmful subsidies from one group to another.
Data availability  The regional distribution of the provision and impact of harmful sheries subsidies. (a) Total amount of harmful sheries subsidies being provided by each region and the total amount of domestic and foreign harmful sheries subsidies estimated to be cumulatively impacting each region, in US$ billions (See Table S.1 for corresponding data); (b) Inter-regional exchanges of the impact of harmful sheries subsidies, due to their provision to regional distant-water shing eets, in US$ millions (See Table S.2 for corresponding data). Figure 3