Trade and foreign fishing mediate global marine nutrient supply

Significance The world produces enough food to nourish the global population, but inequitable distribution of food means many people remain at risk for undernutrition. Attainment of Sustainable Development Goal 2 relies on greater attention to distribution processes that match food qualities with dietary deficiencies. We explore this in the context of fisheries. Foreign fishing and international trade divert nutrients caught in marine fisheries from nutrient-insecure toward nutrient-secure nations. Where nutrient-insecure countries do benefit from foreign fishing and trade, there tends to be high vulnerability to future changes in nutrient flows arising from changes to foreign fishing and trade. This research highlights the need for greater transparency around distribution of fish and for nutrition security to be considered more centrally in development of trade agreements.

Taxonomic resolution: The resolution of any taxonomic identity assigned to a fish or commodity, i.e. species, genus, family or a lower resolution taxonomic group. Transshipment: Fish caught by one vessel is transferred to another vessel for transport to port.

Supplementary Methods
Catch data. We used the catch database of Watson and Tidd (1). This data source is strongly correlated (Spearman rho = 0.95) to landings data from the Sea Around Us project (2), another commonly used global catch database based on country-by-country reconstructions (Fig. S11) (3). Our database of choice was also closely linked to official reporting by the UN's FAO and its associated regional organizations. However, these reported landings can underestimate total landings for a variety of reasons, which include reporting failures and illegal activity. Spatial mapping of the catch uses a process of matching the distribution of reported taxa, the access right and fishing patterns of fleets and the reported areas from a wide range of publicly available sources. This mapping was improved by using fine scale regional reporting such as that provided by tuna Regional Fisheries Management Organizations (RFMOs) and observed patterns from satellite data of Global Fishing Watch's (GFW) vessel Automatic Identification System (AIS)based data (4).
Trade data. We used the trade data of Watson, Nichols, Lam and Sumaila (5). This database uniquely links fisheries capture data (described above) to seafood exports and imports reported by FAO (6). Flows of fish in the global seafood trade are then traced in tonnes by matching commodity groups to taxa and trading partners for seafood from UN's annual Comtrade data (1988-2015) (7). Where no information on trade was available, WTO's primary trading partner data were used (8).
Interpolation of nutrient data. Where catch and trade information were provided at taxonomic resolutions below family, and therefore nutrient concentrations were not available from the hierarchical model, the mass of nutrients was estimated using a four-tier interpolation approach. This process is described below using the example of Spain sourcing from Namibia: 1. We used the median nutrient concentration weighted by tonnage for all taxa of known nutrient value caught in Namibia by Spain (for catch data interpolation) or exported from Namibia to Spain (for trade data interpolation). 2. If the above data were not available, we used the median nutrient concentration weighted by tonnage for all taxa of known nutrient value caught in Namibia by all countries (for catch data interpolation) or exported from Namibia to all countries (for trade data interpolation). 3. Where these data were not available, for trade data interpolation, we used the estimated median nutrient concentration weighted by taxa caught in Namibia by all countries. 4. Finally, if the above data were not available, we used the estimated median nutrient concentration weighted by tonnage for all taxa of known nutrient value caught in all countries (interpolation of catch data) or exported from all countries (interpolation of trade data). The percentage of tonnage interpolated varied widely among nations (Dataset S2); the mean percentage across all nations was 24% for domestic catch data, 10% for foreign fishing data, and 25% for trade data. Sensitivity Analyses. The resolution of the catch and trade data varied widely among nations (Dataset S2). The hierarchical model was only used to estimate nutrient concentrations for those taxa identified to family, genus or species level. The nutrient content of lower resolution taxa was estimated using interpolation. To understand the potential impact of this interpolation on the vulnerability calculations we recalculated national vulnerability to changes in fishery-derived nutrient supplies in two ways: 1) only including those countries where at least 75% of the caught and traded tonnage was identified to family, genus or species level; and 2) only including data identified to family, genus or species level. We then explored the relationships between these different vulnerability estimates.
When the calculations only included countries with >75% of caught and traded tonnage identified to family, genus or species level, the estimated vulnerabilities were all slightly higher for these nations compared with their vulnerabilities when all nations were included ( Fig. S12A-B). When only data identified to family, genus or species level were used, the estimated vulnerabilities showed more variation but followed a similar, bimodal distribution to the vulnerability estimates including all the data (Fig. S12C-D).

Fig. S1
. Annual flow and yield of fishery-derived nutrients due to foreign fishing, highlighting the nutrient status of countries dependent on fish for food (>10% of animal source protein from fish). (A-D) Flow of calcium (mg), iron (mg), zinc (mg), and vitamin A (mcg) due to foreign fishing for nations grouped by prevalence of inadequate intake of respective nutrients in source countries and foreign fleet (sink) countries. (E-H) Top panels show yields (log10 percentage of RNI capita -1 day -1 ) extracted from source nation's EEZs grouped by source nations' nutrient intake. Middle panels show yields caught by foreign fleets in EEZs and the high seas, grouped by foreign fleet (sink) nations' nutrient intake. Bottom panels only include catches from foreign EEZs. Prevalence of inadequate nutrient intake categories: V.Hi (dark shading) -countries with >50% of population with inadequate intake of respective nutrient.; Hi -25<%≤50; Med -10<%≤25; Lo -5<%≤10; V.Lo (light shading) -≤5%; ND (grey shading) -no deficiency data available for these nations; LFS (white shading) -<10% of animal sourced protein from fish; HS (black shading) -nutrients sourced from high seas. Flow of calcium (mg), iron (mg), zinc (mg), and vitamin A (mcg) due to foreign fishing for nations grouped by prevalence of inadequate intake of respective nutrients in source countries and foreign fleet (sink) countries. (E-H) Top panels show yields (log10 percentage of RNI capita -1 day -1 ) extracted from source nation's EEZs grouped by source nations' nutrient intake. Middle panels show yields caught by foreign fleets in EEZs and the high seas, grouped by foreign fleet (sink) nations' nutrient intake. Bottom panels only include catches from foreign EEZs. Prevalence of inadequate nutrient intake categories: V.Hi (dark shading) -countries with >50% of population with inadequate intake of respective nutrient.; Hi -25<%≤50; Med -10<%≤25; Lo -5<%≤10; V.Lo (light shading) -≤5%; ND (grey shading) -no deficiency data available for these nations; HS (black shading) -nutrients sourced from high seas. , iron (mg), zinc (mg), and vitamin A (mcg) due to trade for nations grouped by prevalence of inadequate intake of respective nutrients in source (exporter) and sink (importer) countries. (E-H) Top panels shows yields (log10 percentage of RNI capita -1 day -1 ) exported from source nation, grouped by source nations' nutrient intake. Bottom panels show yields imported by sink countries, grouped by sink nations' nutrient intake. Prevalence of inadequate nutrient intake categories: V.Hi (darkest shading) -countries with >50% of population with inadequate intake of respective nutrient.; Hi -25<%≤50; Med -10<%≤25; Lo -5<%≤10; V.Lo (lightest shading) -≤5%.

Fig. S4.
Annual flow and yield of fishery-derived nutrients due to international trade, highlighting the nutrient status of countries dependent on fish for food (>10% of animal sourced protein from fish). (A-D) Flow of calcium (mg), iron (mg), zinc (mg), and vitamin A (mcg) due to trade for nations grouped by prevalence of inadequate intake of respective nutrients in source (exporter) and sink (importer) countries. (E-H) Top panels shows yields (log10 percentage of RNI capita -1 day -1 ) exported from source nation, grouped by source nations' nutrient intake. Bottom panels show yields imported by sink countries, grouped by sink nations' nutrient intake. Prevalence of inadequate nutrient intake categories: V.Hi (darkest shading)countries with >50% of population with inadequate intake of respective nutrient.; Hi -25<%≤50; Med -10<%≤25; Lo -5<%≤10; V.Lo (lightest shading) -≤5%; LFS (white shading) -<10% of animal sourced protein from fish.  Conceptual framework used to estimate vulnerability to changes in supply of fishery-derived nutrients due to foreign fishing and trade. Framework details constituent exposure, sensitivity and adaptive capacity indices and scaling. Metrics in blue and grey text were scaled from 0 to 1. For those metrics shown in blue text, values were reversed for rescaling to ensure high values equalled high exposure, sensitivity or adaptive capacity. Weightings of imports or foreign fishing used in exposure metric account for imports/foreign fishing relative to total nutrient supply from imports, foreign fishing and domestic catch. See Table S4 for more information.  S7. National exposure and sensitivity to changes in fisheries-derived nutrient supply due to changes in foreign fishing and trade. Exposure to changes in (A) foreign fishing and (B) imports. Sensitivity due to (C) dependence on fish for animal-sourced protein and (D) prevalence of inadequate nutrient intake. Scales in A and B only go up to 0.5 as these metrics were weighted by relative contribution of foreign fishing or imports to national nutrient flows. Catches flowing to vessels using flags of convenience were not included in the exposure estimates.

Fig. S8.
Conceptual framework used to estimate vulnerability to changes in supply of fishery-derived nutrients due to foreign fishing, trade and climate change. Framework details constituent exposure, sensitivity and adaptive capacity indices and scaling. Metrics in blue and grey text were scaled from 0 to 1. For those metrics shown in blue text, values were reversed for rescaling to ensure high values equalled high exposure, sensitivity or adaptive capacity. Weightings of imports or foreign fishing used in exposure metric account for imports/foreign fishing relative to total nutrient supply from imports, foreign fishing and domestic catch. See Table S4 for more information.      Used as an indicator of socio-economic status.
UNDP (16) Proportion of animal sourced protein from fish Proportion of animal-sourced protein from marine fish that is consumed in a country.
Used as indicator of reliance on fish for food. Included in the sensitivity dimension of the vulnerability framework.

FAO (17)
Prevalence of inadequate micronutrient intake index (PIMII) Mean level of inadequate nutrient intake across 14 micronutrients: calcium, copper, iron, folate, magnesium, niacin, phosphorus, riboflavin, thiamin, vitamin A, vitamin B12, vitamin B6, vitamin C, and zinc. Data is in the form of % of the population.
Used to understand the level of micronutrient deficiency risk in different countries for a wide range of nutrients, not just those prevalent in fish. Included in the sensitivity dimension of the vulnerability framework.
Beal et al (15) Adequacy of daily energy supply Provides the dietary energy supply (calories) as a percentage of energy required.
Indicator of hunger, but as it was strongly correlated with PIMII it was not used in the vulnerability framework.

FAO (17)
Index of political stability and absence of violence Measures perceptions of the likelihood that a government will be destabilized or overthrown by unconstitutional or violent means. Values from -2.97 to + 1.53 Used as an indicator of the stability of a country. Included in the adaptive capacity dimension of the vulnerability framework.

FAO (17)
Health care expenditure as a percentage of GDP Percentage of the GDP spent on health care. Used as an indicator of the capacity of a nation to address nutrient deficiencies. Included in the adaptive capacity dimension of the vulnerability framework.
UNDP (16) Status as a food importer Three tier classification of high net food importer, net food importer, non-net food importer.
Used as an indicator of the capacity of a nation to cope with changes in the flow of fishery-derived nutrients. Included in the adaptive capacity dimension of the vulnerability framework.
UNCTAD (20) Predicted change in catch due to climate change 2010 to 2050 Mean log relative change in catch 2010 to 2050 under RCP 8.5. Values from -0.44 to +0.11.
Used as an indicator of potential exposure to climate change impacts on nutrient supplies from domestic fisheries. Included in the exposure dimension of the vulnerability framework.