The shift of phosphorus transfers in global fisheries and aquaculture

Phosphorus (P) is essential for life and food production, yet global demand for food has quadrupled P inputs to the biosphere since preindustrial times, leading to concerns over depletion of mineral P reserves, uneven distribution of P, and widespread eutrophication of inland and coastal waters. While most large-scale P budgeting focuses on agriculture and livestock, fisheries and aquaculture represent an overlooked subsector influencing land–aquatic P transfers. Fish harvest moves P from aquatic systems to land–human systems, whereas aquaculture requires external P inputs via feeds and fertilizers, with generally low phosphorus use efficiency (PUE), especially in Asia where most aquaculture occurs. This study asks: how have global fisheries and aquaculture altered P flows between aquatic and land–human systems historically, what is the current net P balance, how do regional patterns contribute, and what efficiency gains are needed to achieve a P-neutral fish production sector by 2050? The purpose is to quantify these net P transfers with a data-driven approach and identify targets for improving P management in fisheries and aquaculture within the broader biogeochemical P cycle and planetary boundaries context.

Prior work has documented substantial anthropogenic P inputs and losses from crop and livestock systems, with significant contributions of manure and fertilizers to freshwater P loading, and emphasized mitigation via recycling and wastewater recovery. Studies have also highlighted aquaculture’s rapid growth, reliance on feeds sourced from wild fish and crop–livestock products, and environmental impacts including nutrient discharges and eutrophication. Reported aquaculture PUE is low (e.g., 8.7–21.2% in China), implying large P losses. Other known pathways transporting P from aquatic to land (e.g., via anadromous fish, seabirds, sea-salt) are relatively small. There is growing recognition to integrate fisheries, aquaculture, and agriculture in sustainability assessments (food, land use, biodiversity, climate), but comprehensive, data-driven quantification of fishery-driven P fluxes has been lacking.

Data-driven global P budget reconstruction for fisheries and aquaculture (1950–2016) with projections to 2050:

• Production data: FAO FishStatJ (v3.04.6) for global fishery production (capture and aquaculture) by species, environment (fresh, brackish, marine), and country (1950–2016), complemented by Sea Around Us reconstructed marine landings to correct underreported captures (scaled country-level FAO capture to landed catches; 2014 scaling applied to 2015–2016). Production expressed as live weight.
• Fish whole-body P concentration database: 175 peer-reviewed studies, 1164 records across 224 species (1088 finfish, 41 crustaceans, 35 mollusks). Converted dry to wet basis using moisture content when necessary. Used species means; if missing, substituted order-level mean; if still missing, used major group mean. Covered >80% of production at order level or below.
• Aquaculture culture-system PUE database: 168 cases from 96 studies across diverse systems (pond, tank, cages, recirculating, flow-through) and countries (e.g., China, India, US, Bangladesh, Vietnam, etc.). PUE defined as harvested fish P divided by external P inputs (feeds + fertilizers) at farm/system level. Separated into six groups: freshwater finfish, marine finfish, freshwater crustaceans, marine crustaceans, freshwater mollusks, marine mollusks.
• P-retention efficiency (PRE) database: 348 controlled feeding experiments (closed/controlled systems) reporting fraction of ingested feed P retained in fish biomass, used as a technological reference/upper bound for potential PUE.
• Computation of P fluxes: Annual harvested P (P-harvest) = sum over species of production (live mass) × whole-body P fraction. Aquaculture P-input computed as P-input = P-harvest / PUE, summing across the six aquaculture groups using group-specific PUE distributions. Net P transfer (P-net) = P-harvest − P-input (positive means net landward transfer, negative means net input to aquatic systems via aquaculture).
• Uncertainty: Monte Carlo (n=1000) sampling of species P concentrations (drawing from species/order/group distributions with rescaled SD by order-level CV to preserve mean) and production (assumed percentile uncertainty bands; main text reports 50% interval scenario) for P-harvest; sampling PUE within each aquaculture group for P-input. Interquartile range (IQR) used to summarize uncertainty due to skewed PUE distributions and limited sample sizes.
• Scenarios to 2050: Baseline assumes stable wild capture at 2005–2014 mean and aquaculture growth to 140 Tg by 2050 (≈2.3× 2010) per Waite et al., with current global average PUE ≈20%; computes resulting P-input, P-harvest, P-net. An idealized neutral-P target scenario determines the PUE required (weighted by production) to achieve P-net ≈ 0 by 2050, and compares to upper ranges of PUE and PRE to assess technical feasibility.

• 2016 global fluxes: Aquaculture P-input ≈ 2.04–2.06 Tg P yr−1; P-harvest ≈ 1.10 Tg P yr−1. Net P transfer (P-net) is negative, indicating net P added to aquatic systems by fish production.
• Historical shift: From 1950 to mid-1980s, wild capture dominated (up to 99% of P-harvest in 1950), yielding positive P-net (landward transfer), peaking at ~0.54 Tg P yr−1 around 1986–1997. Rapid aquaculture expansion since the 1980s (without commensurate PUE gains) drove P-net downward, crossing to negative around 2004.
• 2016 regional contributions: Asia alone contributed a negative P-net of −1.13 Tg P yr−1, while the rest of the world had a small positive P-net of +0.18 Tg P yr−1. P-input first exceeded P-harvest regionally around 1988 (South Asia), ~1990 (East Asia), 2005–2006 (Southeast Asia), and ~2016 in West Asia.
• Drivers by country (1980s vs 2007–2016): Aquaculture P-input increased ~9×. Major contributors to the global increase: China ~60%, India ~10%, Indonesia ~6%, Vietnam ~6%, Bangladesh ~3%, Thailand ~2%. Declines in wild P-harvest in some developed countries (e.g., Japan, Russia, Chile, Denmark, Canada) also reduced P-net.
• Fresh vs marine contributions: Freshwater aquaculture accounts for ~84–94% of aquaculture P-input. Within freshwater, finfish account for 95–100% of P-input; crustacean share rose slightly to ~5.3% by 2010. In marine aquaculture, >90% of P-input went to finfish during 1950–1970, shifting to ~50% crustaceans after 1990.
• Comparison to other P fluxes: Fishery P-input (2.06 Tg P yr−1 in 2016) is substantial but smaller than crop–livestock P loads to aquatic systems (estimates ~4–13.5+ Tg P yr−1 depending on study and inclusions). Historically, fishery-driven landward P transfer (~0.2 Tg P yr−1 in 1950s; peak ~0.54 Tg P yr−1 in 1980s) was the largest known aquatic-to-land pathway, exceeding seabird, anadromous fish, and sea-salt deposition pathways.
• 2050 baseline projection (business-as-usual PUE ~20%): Wild production 98.7 Tg; aquaculture 140 Tg. P-harvest ≈ 1.41 Tg P yr−1; P-input ≈ 3.42 Tg P yr−1; P-net ≈ −2.01 Tg P yr−1 (imbalance roughly doubles relative to 2016).
• Efficiency targets: To reach P-neutrality by 2050, global production-weighted aquaculture PUE must rise from ~20% to ~48%. Technical plausibility indicated by upper PUE values in practice (e.g., China finfish 95th percentile ~44%) and PRE in controlled settings (75th percentile PRE ~52% for finfish; 95th percentile ~78%). Dominance of finfish (≈87% of harvested fish P) implies that achieving ~75th percentile PRE across species could correspond to PUE ≈48%.

The study quantifies, for the first time at global scale, how fisheries and aquaculture have reshaped P flows between aquatic ecosystems and land–human systems. Historically, wild capture provided a significant landward P transfer that could help alleviate terrestrial P scarcity. However, the ascendance of aquaculture—with low PUE and concentrated in Asia—reversed this balance, making the fish production sector a net source of P to aquatic environments, thereby contributing to eutrophication risks and tightening the margin relative to the planetary boundary for P loading, especially regionally in Asia. Freshwater aquaculture dominates P inputs, and finfish are the primary sink of inputs, with increasing contributions from marine crustaceans. Addressing this imbalance requires improving aquaculture PUE via technological and managerial advances: optimized feeds and feeding regimes, enhanced culture systems (e.g., recirculating aquaculture systems, biofloc), and integrated systems (IMTA, IAA) that recycle nutrients across species and sectors. The analysis shows that reaching a global average PUE of ~48% by 2050 is technically feasible when considering upper observed PUEs and PREs, particularly for finfish. Additionally, expanding P recovery and reuse from fish-processing wastes, food waste, human excreta, and wastewater can enhance landward P recycling, reducing net aquatic loading. Integrating fisheries and aquaculture with broader P management across agriculture and waste systems is essential to optimize P use across sectors and regions.

This work provides a data-driven reconstruction of global fishery-driven phosphorus budgets (1950–2016), revealing a historical shift from net landward P transfer under wild capture dominance to net aquatic P loading driven by aquaculture expansion with low PUE. In 2016, aquaculture inputs (~2.06 Tg P yr−1) exceeded P harvested in fish (~1.10 Tg P yr−1), yielding a negative P-net; under business-as-usual, the imbalance could reach ~−2.01 Tg P yr−1 by 2050. Achieving a P-neutral fish production sector by 2050 will require raising global, production-weighted aquaculture PUE from ~20% to ~48%, a target that appears technically achievable based on upper PUE and PRE benchmarks, especially for finfish. Policy and practice should focus on: improving feed formulations and feeding strategies; adopting low-impact and integrated production systems; scaling up P recovery and reuse from fish-processing and broader waste streams; and integrating fisheries, aquaculture, and agriculture into coordinated P management frameworks. Future research should develop spatially explicit P budgets that incorporate trade, quantify the fate of aquaculture P losses and harvested P across supply chains, assess economic and policy levers for PUE improvements, and design multi-sector optimization strategies for P use and recycling.

• Data limitations: FAO production data underrepresent nonindustrial and illegal fisheries; marine capture corrected using Sea Around Us, but uncertainties remain. Aquaculture area, feed composition, and fertilizer inputs are poorly documented globally, particularly for farm-made feeds and manure.
• PUE data coverage and heterogeneity: Culture-system-level PUE database (168 cases) is limited in spatial/temporal coverage, with few repeated measurements over time; no statistically significant temporal trend in PUE could be detected.
• Methodological simplifications: Aquaculture P-input inferred via P-harvest/PUE rather than bottom-up feed/fertilizer inventories; species without P data rely on order/group averages; discarded fish excluded from land–aquatic transfers by assumption.
• Uncertainty treatment: Monte Carlo approach uses assumed percentile uncertainties for production (e.g., 50% interval) and relies on IQR due to skewed PUE distributions; exact uncertainty magnitudes remain uncertain.
• Scope boundaries: Study focuses on net transfers (P-harvest vs P-input) and does not track detailed fates, forms, or locations of P losses (e.g., excreta vs uneaten feed; pond sediments vs receiving waters; processing and transport losses) over time; does not explicitly model riverine retention or eventual ocean delivery.
• Trade and regional flows: International fish trade not incorporated, potentially altering regional P budgets and responsibilities.
• Marine ecosystem feedbacks: Potential contributions of aquaculture P leakage to marine food webs not quantified; likely minor globally but could be important locally.