Systems approach to quantify the global omega-3 fatty acid cycle

Long-chain omega-3 fatty acids—eicosapentaenoic and docosahexaenoic acids—are essential components of human diets and some aqua and animal feeds, but they are sourced from finite marine fisheries, and are in short supply and deficient in large parts of the world. We use quantitative systems analysis to model the current global eicosapentaenoic acid/docosahexaenoic acid cycle and identify options for increasing supply. Opportunities lie in increased by-product utilization and food waste prevention. However, economic, resource, cultural and technical challenges need to be overcome. Omega-3 fatty acids are important for the human diet and for some aqua and animal feeds. This study reports a supply gap, and using quantitative systems analysis identifies targets for increasing efficiency in the global omega-3 cycle.

evaluating their combined effects. Here, we use a systems approach and quantify the global EPA and DHA cycle to: (1) provide a comprehensive problem description to improve overall resource efficiency; and (2) identify system-wide opportunities and challenges for meeting the human EPA/DHA demand. Thereafter, we aim to inform decision-makers on the current EPA/DHA status, its drivers and the most effective intervention options at a global level.
We find that between net primary production and higher predators, approximately 90% of EPA/DHA is lost via respiration, defecation and deaths, indicating that large trophic losses occur up the food chain (Fig. 1). The zooplankton and phytoplankton stocks are of comparable sizes (approximately 40 Mt EPA + DHA), with no net yearly addition to stock. Caught wild seafood accounts for 0.04% of the EPA/DHA produced via net primary production. Approximately half of harvested marine EPA/DHA is managed through fish meal and fish oil production (primarily for aquaculture consumption; Fig. 2a) and half is reserved for direct human consumption.
Despite aquaculture being a major consumer of EPA/DHA, it is also a major producer via non-fed species, such as molluscs and carp, which accumulate EPA/DHA from the environment and/or endogenous production through the elongation of shorter-chained fatty acids. Freshwater fish are better at elongation compared with marine fish due to unique enzymes and desaturase genes that allow for EPA/DHA synthesis 8 . In contrast, fed high-trophic salmonid species: (1) consume a high proportion of aquaculture's use of fish meal and fish oil (58 and 22%, respectively, in 2015); (2) have EPA/DHA retention rates varying from 30 to 75%; and (3) are inefficient at fatty acid elongation 9 , but also supply EPA/DHA through a farmed product based on an otherwise under-utilized wild fish resource.
We find that the supply of EPA/DHA for human consumption is 420 kt yr −1 , or 149 mg EPA + DHA per capita daily, representing 30% of global demand. Therefore, we confirm the supply gap identified by Tocher 2 but find it to be over 50% larger than previous estimates suggest. Significant losses occur due to unavoidable and avoidable food waste (114 and 105 kt yr −1 of EPA + DHA, respectively) and unutilized fish-processing by-products (53 kt yr −1 of EPA + DHA), with the largest losses in Asia (Fig. 2b).
While many options exist to fill the EPA/DHA gap, each has associated challenges. Aquaculture's strategic use of fish meal and fish oil in feed at key life stages can: (1) influence the EPA/DHA utilization efficiency by farmed fish; and (2) optimize the benefits of marine ingredients from a fish and human health perspective (for example, finishing diets to increase EPA/DHA towards harvest time 10 ). Fish stock recovery could increase long-term fish yields and the EPA/DHA supply (albeit with probable short-term decreases) 4   The purple dot denotes net endogenous EPA/DHA production by fish. Mass balance inconsistencies are due to rounding errors and uncertainty. All flows in process 6 were calculated independently, and the remaining mass balance inconsistency is <1% of total flows in this process. Net endogenous production in the ocean system is not visualized. DOM, dissolved organic matter; FM&O, fish meal and oil; FO, fish oil; NPP, net primary production; PP, phytoplankton; ZP, zooplankton. reproductive success 11 . With the krill harvesting rate (~300,000 tonnes of biomass in 2018) being below the catch limit of 5.6 million tonnes annually, as defined by the Commission for the Conservation of Antarctic Marine Living Resources, increasing krill catch for use as feed could substantially increase the EPA/DHA supply 12 . However, Antarctic krill harvesting operations face challenges related to geography and costs, and effective stock management is imperative to ensure sustainable harvesting levels. Trophic losses could be avoided (and supply increased) by: (1) consuming EPA/DHA from a lower trophic level (for example, seaweeds, krill and bivalve molluscs); (2) increasing non-fed fish farming; and/or (3) diverting more wild catch to human consumption through direct consumption or oil supplementation produced from these species. However, for this to prove effective, the digestibility, bioavailability and efficacy of EPA/DHA in these products need to be understood (for example, the bioavailability of fatty acids in fish oil is lower than in fish 13 ), and although the nutraceutical market is strong, the wild fish market depends on factors including, among others, catch quality, acceptance and temporal challenges (that is, the seasonal surplus of fish catch that cannot be absorbed by the market 14 ). In addition, logistical challenges exist for the distribution to populations that are EPA/DHA deficient 3 .
Improved by-product utilization and food waste avoidance can substantially increase the supply of EPA/DHA while reducing waste. Processing by-products can be used for fish meal and fish oil production for aquafeed and/or human consumption, provided the regulatory frameworks are followed 15 . However, a major challenge is collection and processing, as by-products are often geographically dispersed. For example, Asia-where most of the by-product potential is concentrated (Fig. 2b)-has a culture of buying fish whole and disposing of by-products at the household level 16 . Centralized fish processing is needed to recover by-products in this region, but would require a substantial cultural shift in the way fish is consumed. Food waste prevention is also an effective means for increasing supply, as it avoids the unnecessary use of EPA/DHA to produce food that is wasted 17 .
Future options to produce EPA/DHA include large-scale production of natural and genetically modified microalgae, microbacteria and higher plants. However, current technologies and concerns about genetically modified material limit the volume of supply, their cost-effectiveness and widespread penetration into the market 18 , although regulatory challenges related to genetically modified feed use are primarily constrained to Europe 19 .

Methods
We used a multi-layer material flow analysis framework to quantify the stocks and flows of EPA/DHA throughout our defined system. The 'mother' layer contains the biomass system (tonnes of wet weight per year) and the 'child' layer includes the sum of EPA and DHA balance (tonnes of EPA + DHA per year). From a mass balance standpoint, quantifying the EPA/DHA content of biological organisms is a methodological challenge due to: (1) marine and freshwater species storing EPA/DHA within their lipids and, thus, metabolizing them as an energy source; and (2) organisms endogenously producing EPA/DHA through the elongation of α-linolenic acid (18:3n − 3) at various rates depending on, among others, the species, time of the year and habitat 20 . Therefore, unlike substances (that is, chemical elements), EPA/DHA can be created or destroyed, which limits mass balance conservation when modelling and makes it necessary to consider production and destruction. Preliminary estimates have shown endogenous EPA/DHA production to contribute little to the EPA/DHA supply from farmed fish (that is, EPA/DHA consumed by aquaculture equals the EPA/DHA contents of the produced fish) 21 . However, for certain species, endogenous EPA/DHA production can be potentially significant, especially for bivalve molluscs and carp 21 . Therefore, we accounted for this by calculating the net EPA/DHA production of each biological process for which EPA/DHA can be created/destroyed. We assumed that processes that mechanically transform the flows (that is, fish processing) do not affect the EPA/DHA content of the biomass. We defined the system to include the natural and anthropogenic stocks and flows of EPA/DHA. Freshwater ecosystem food chains were not considered due to their minor role relative to the marine ecosystem and limited data availability; however, we included the EPA/DHA contained in freshwater fish capture and freshwater aquaculture. In addition, we did not consider natural export from marine to terrestrial ecosystems (for example, due to the consumption of drifted algae by lizards, birds and other terrestrial animals), as preliminary estimates (24 kt yr −1 of EPA + DHA) have shown this to be insignificant relative to the overall marine food web 22 .
Primary data were sourced from scientific publications, reports, statistics and industry data from the International Marine Ingredients Organization (IFFO). Ocean carbon flows were based on Stock et al. 23 and represent a 20-year average . The long time frame minimized the uncertainty related to yearly variations in primary production due to, for example, El Nino events 24 . Capture data were based primarily on the Food and Agriculture Organization dataset FishStat, and include an average between 2009 and 2013 to normalize yearly variations. Due to the large number of species, we only accounted for the top 20 fish, cephalopod and crustacean species caught and farmed in each geographical region. EPA/DHA calculations were performed at the species level. However, we accounted for all wild and farmed bivalve molluscs and plants.
Overall, we accounted for over 90% of fishery and aquaculture production. Avoidable food waste was defined to include all edible food that was wasted at the household level. Unavoidable food waste included the remaining inedible fraction, such as peels, shells and bones. Further information regarding the methods can be found in the Supplementary Information. Reporting Summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
This work used data collected from a variety of sources-both proprietary and freely available. See the references in the Supplementary Information for data specification. All figures are based on this collected dataset, and geographically aggregated data (in more refined detail than the source data) will be made available on request from the corresponding author. Source data for Figs. 1 and 2 are provided with the paper.

Statistics
For all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section.

n/a Confirmed
The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly The statistical test(s) used AND whether they are one-or two-sided Only common tests should be described solely by name; describe more complex techniques in the Methods section.
A description of all covariates tested A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient) AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals) For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted

Software and code
Policy information about availability of computer code Data collection Primary data are sourced from scientific publications, reports, statistics and industry data from the International Marine Ingredients Organization (IFFO). Data and their associated sources are detailed in the supplementary information and will be made available upon request.

Data analysis
The data was analysed using a MatLab script. This script performs mass balance calculations and are detailed extensively in the supplementary information. The script can be made available upon request.
For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors/reviewers. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Research guidelines for submitting code & software for further information.

Data
Policy information about availability of data All manuscripts must include a data availability statement. This statement should provide the following information, where applicable: -Accession codes, unique identifiers, or web links for publicly available datasets -A list of figures that have associated raw data -A description of any restrictions on data availability The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.

nature research | reporting summary
October 2018 Field-specific reporting Please select the one below that is the best fit for your research. If you are not sure, read the appropriate sections before making your selection. All studies must disclose on these points even when the disclosure is negative.

Study description
The study is quantitative, using the method 'material flow analysis'. This method is based on conservation of mass principles and relies on statistics to quantify mass balances.

Research sample
Primary data are sourced from scientific publications, reports, statistics and industry data from the International Marine Ingredients Organization (IFFO).

Sampling strategy
No sampling was performed in this analysis.

Data collection
The data was collected via online platforms.

Timing
Data was collected from statistics to represent a 20 year average for oceanic flows and a 5 year average for anthropogenic flows. The long time frame was used to minimize the uncertainty related to yearly variations in primary production due to, e.g., El Nino events.

Data exclusions
No data was excluded