Phosphorus in Salmonid Aquaculture: Sources, Requirements, and System-Level Implications

Abstract

This review provides a comprehensive synthesis of phosphorus (P) dynamics relevant to salmonid aquaculture, with a particular focus on Atlantic salmon. It explores the global P cycle, the chemical forms of P in aquatic systems, and the implications of P sourcing, processing, and availability in aquafeeds. The review distinguishes between digestibility and availability of P, summarizes requirement studies, and evaluates the contribution of marine, animal, vegetable, and inorganic sources to dietary P. It also examines how aquaculture system design, particularly recirculating aquaculture systems (RASs), influences P accumulation and emissions. By integrating nutritional, physiological, and environmental perspectives, this review offers a uniquely holistic view of P efficiency and sustainability in salmonid aquaculture.

1. Introduction

Phosphorus (P) is a critical nutrient in aquaculture, indispensable for maintaining essential physiological and biochemical processes in fish, including somatic growth, bone development, and metabolic regulation, factors that are fundamental to the production of aquatic protein for human consumption [1,2]. As the global demand for farmed salmon increases, so does the industry’s reliance on P-rich aquafeeds. Norway, producing approximately 50% of the worlds Atlantic salmon (Salmo salar), which equated to 5.2 million metric tons in 2023 [3,4], is seeing aquaculture P use approach that of agriculture [5].

In 2020, the Norwegian agricultural sector consumed approximately 20,700 tons of P as fertilizers [6], while the aquacultural sector had an estimated consumption of roughly 18,700 tons of P as food and feed additives, based on a reported feed consumption of 1,989,100 tons [7] and an average P content in fish feed of 0.94% [8]. Since 2020, the aquafeed volumes has surpassed 2 million tons annually [9,10], with further growth expected especially in technological-intensive farming solutions (TIFs) on land. Den Norske Bank (DNB) predicts a growth of 700,000 tons in production facilities on land alone by the year 2035 [11], implying a proportional P demand growth.

A simulated scenario from 2015 warned that, without improved intra- and cross-sector P recycling, the aquaculture consumption could surpass agriculture fourfold in P imports and losses by 2050 [5]. This is concerning due to the high eutrophication power of P [12,13]. Consequently, P is tightly regulated in the global aquaculture industry, where, e.g., countries in the European Union (EU) subject farmers to strict guidelines in terms of effluent regulations on what and how much can be emitted [14,15,16]. Yet, a 2025 report found that 44% of land based Norwegian salmon farmers exceeded their licensed emission of, amongst others, P [17].

While several studies have explored the origins and dynamics of P [1,12,18,19], and more recent reviews have focused on physiological mechanisms for uptake and excretion and strategies to mitigate the environmental impact of P from aquaculture [20,21,22], a recently published report in Norway highlights the need to contextualize this knowledge specifically for intensive Atlantic salmon production systems.

This article presents a narrative review of P dynamics in salmonid aquaculture, with a particular focus on Norway as the leading salmonid-producing country. A narrative review approach was chosen due to its flexibility in synthesizing diverse sources of information, including empirical studies, technical reports, and regulatory frameworks, across multiple disciplines such as nutrition, physiology, and aquaculture engineering. This method allows for an integrative and contextualized understanding of P use and sustainability. Narrative reviews are particularly well-suited for addressing complex, interdisciplinary topics where quantitative meta-analysis is not feasible [23,24]. The literature was selected based on its relevance to nutritional, physiological, chemical, and system-level aspects of P use. The aim was to clarify key P concepts, evaluate different P sources in salmonid aquafeeds and their bioavailability, highlight methods for estimating requirements, and explore how traditional and emerging system designs influence P emissions from the industry.

2. The Chemical Element P and Its Cycle

P occurs inorganically in most geological formations and soils, primarily as deposits of the mineral apatite, commonly known as phosphate rock [19]. In terrestrial systems, P resides in three main pools: bedrock, soil, and living organisms [25]. The phosphate rock, or phosphorite, contains P in various forms of apatite, namely hydroxyapatite (Ca₅(PO₄)₃OH), fluorapatite (Ca₅(PO₄)₃F), and chlorapatite (Ca₁₀(PO₄)₆(Cl)₂) [26]. Due to its high chemical reactivity, P does not occur in its elemental form in nature but instead found as phosphate, either organically or inorganically bound [27]. The relative proportion of organic P (Po) and inorganic P (Pi) in soils and sediments are influenced by various environmental factors, including pH, redox conditions, and biological activity.

The inorganic fraction can be further divided into two groups based on the environmental pH: orthophosphates and condensed phosphates. Orthophosphates, also known as reactive phosphates, include phosphate (PO₄³⁻), hydrogen phosphate (HPO₄²⁻), dihydrogen phosphate (H₂PO₄⁻), and phosphoric acid (H₃PO₄). Condensed phosphates are P-based molecules that contain more than one atom of P combined with oxygen and hydrogen atoms [28,29]. In aqueous solutions, water breaks one or more chemical bonds through hydrolysis, transforming the condensed phosphates into orthophosphates, which are the simplest form of the phosphate family referred to as monophosphate [29].

Environmental pH plays a crucial role in determining the predominant forms of Pi (Figure 1) [30,31]. At acidic conditions, phosphoric acid remains mostly in its undissociated form H₃PO₄. As pH increases to mildly acidic/neutral conditions, H₃PO₄ loses a proton (H⁺) to form H₂PO₄⁻. With a further increase in pH, H₂PO₄⁻ loses another H⁺ to become HPO₄²⁻, before losing its final proton at basic conditions to form PO₄³⁻.

Pi in rocks is generally not accessible for direct consumption by animals. It becomes accessible for plants through geological processes such as tectonic uplifting and weathering. Tectonic uplifting exposes new rock surfaces, while physical disintegration (e.g., freezing, thawing, root growth) and chemical dissolution (e.g., reactions with water and acids) releases Pi into the soil [25]. Plants absorb this Pi and incorporate it into organic molecules, converting it into Po.

Po are formed through biosynthesis, e.g., when an organic molecule like a hydrocarbon or alkyl group replaces a hydrogen atom in phosphoric acid, resulting in organophosphate compounds found in key biomolecules like DNA and RNA [32]. These organophosphates are synthesized during biological processes, and their phosphate bonds, such as those in adenosine triphosphate, play a role in energy transfer, as in the hydrolysis of adenosine triphosphate to adenosine diphosphate [32,33,34]. Through the food chain, P is transferred from herbivores to carnivores, and both Pi and Po are released back into the soil trough excretion and decomposition of animals or plants. This microbial recycling is an essential part of the P cycle, as it converts Po back into Pi. This occurs through the hydrolysis of Po by microorganisms, a crucial process that releases orthophosphate, which can then be assimilated by autotrophs [34]. The microbial activity in the soils and sediments therefore strongly influences the concentration and chemical form of the P incorporated into the geological record [35].

Phosphate is transported from terrestrial to aquatic environments primarily through erosion and chemical weathering, with river runoff acting as a major conduit [25]. In addition to natural processes, human activities such as agriculture, industrial wastewater discharge, and land-based aquaculture significantly increase P loading in soils, rivers, and oceans [12,13]. Once in aquatic systems, P enters as both Po and Pi. The Pi fraction includes orthophosphate, pyrophosphate, and polyphosphates, while the Po fraction consists mainly of organophosphates and phosphodiesters [12].

In surface water, Pi, primarily in the form of orthophosphate, is absorbed by aquatic plants and becomes part of their tissues [25,36,37]. This Po then moves up the food chain and is eventually released back into the environment through excretion or decomposition of dead organisms, just as in the terrestrial P cycle [37]. Phytoplankton can also die and lyse without being consumed, resulting in the release of both Po and Pi into the environment [38]. Some of the Po is recycled and converted back into Pi, which is then returned to the water column [36]. The recycling occurs when aquatic bacteria secrete phosphatases that break down Po into Pi, while P-solubilizing microorganisms produce organic acids that help solubilize Pi compounds, making them accessible to aquatic plants [36,39]. The remaining Po sinks to the bottom and settles in the sediments, where a proportion of it may be released back into the water column by benthic flux, while the rest is incorporated into the bedrock through sedimentation and diagenesis, and re-enters the P cycle through tectonic uplifting [25]. Upwelling and downwelling processes transport P through different water layers, bringing P closer to the surface and making it available for aquatic plants and phytoplankton again [25,40].

In most natural water bodies, including production water for salmon, the average pH is typically in the range of 6.5–8.5 [41,42], thus providing predominant forms of H₂PO₄⁻ and HPO₄²⁻ [19]. The total phosphorus (TP) concentration comprise Po and Pi [36,37]. While Pi in natural water bodies seldom exceeds 0.1 mg L⁻¹, TP can be as high as 0.5 mg L⁻¹ [19]. However, in anaerobic zones where the solubility of iron phosphate increases, e.g., sediment pore water and hypolimnetic water of eutrophic lakes, the P concentration can be above 1 mg L⁻¹ [19].

3. Phosphorus Terminology Used in Aquaculture

Understanding P dynamics in aquaculture requires clarity in terminology, yet the literature often uses overlapping or inconsistent terms. This can lead to confusion when comparing studies or interpreting environmental monitoring data. To address this, the following section outlines the four operational categories of P commonly used in aquatic sciences [36] (Figure 2), and how these relate to terms more frequently used in aquaculture and environmental monitoring. Clarifying these distinctions is essential for interpreting P measurements and managing emissions effectively.

• Soluble Reactive Phosphorus (SRP)

• Definition: This is the fraction of P that is dissolved in water and readily available for biological uptake. It reacts with molybdate to form a detectable compound.
• Importance: SRP is considered immediately bioavailable and can be quickly utilized by aquatic plants and algae.

2.

Soluble Unreactive Phosphorus (SUP)

• Definition: This includes dissolved phosphorus that does not react with molybdate. It is often Po, such as that found in DNA, RNA, and phospholipids.
• Importance: SUP is not immediately available for biological uptake but can become available over time through biological or chemical processes.

3.

Particulate Reactive Phosphorus (PRP)

• Definition: This is P that is attached to particles in water and can react with molybdate. It includes P bound to sediments or organic matter.
• Importance: PRP can be a significant source of P in aquatic systems, especially when particles are resuspended into the water column.

4.

Particulate Unreactive Phosphorus (PUP)

• Definition: This includes P that is attached to particles but does not react with molybdate. It is often in a form that is not readily available for biological uptake.
• Importance: PUP represents a more stable form of P that can be stored in sediments and may become available under certain conditions.

For simplicity, when addressed in aquaculture and environmental discussions, the terminology used thereafter is dissolved inorganic phosphorus (DIP), of which SRP is a subset; dissolved organic phosphorus (DOP), corresponding to SUP; particulate inorganic phosphorus (PIP), which is a part of PRP; and finally, particulate organic phosphorus (POP), associated with PUP [43,44]. In aquaculture production and environmental lake monitoring, POP is associated to the residue of feces and fish feed in the water column, while DOP represents particles smaller than 0.2 µm [44,45], and DIP is mainly represented as PO₄³⁻ [46].

4. The Global Phosphorus Market and Processing Industry

The concept of P scarcity arose in the 19th century, when major breakthroughs in the understanding and use of P as a fertilizer for agriculture were made [47]. Flash forward to the 21st century, when securing the sustainable use of P has become one of the world’s major challenges [48]. Phosphate rock minerals are the most significant global source of P, which is mined out of large open-pit operations in various regions of the world, with the exception of some underground mines [49]. Mine reserves add up to approximately 74 billion tons; however, the global resource of phosphate rock adds up to more than 300 billion tons, with considerable phosphate stocks identified on the continental shelves and seamounts in the Atlantic and the Pacific Oceans [50] (

Table 1. World production and reserves of phosphate rock in 2022–2023 (in thousand metric tons).

Mine Production
Country	2022	2023 e	Reserves
United States	e 19,800	20,000	1,000,000
Algeria	e 1800	1800	2,200,000
Australia	e 2500	2500	1 1,100,000
Brazil	e 6200	5300	1,600,000
China 2	e 93,000	90,000	3,800,000
Egypt	e 5000	4800	2,800,000
Finland	923	950	1,000,000
India	e 1740	1500	31,000
Israel	2170	2500	60,000
Jordan	11,300	12,000	1,000,000
Kazakhstan	e 1500	2000	260,000
Mexico	442	500	30,000
Morocco	39,000	35,000	50,000,000
Peru	4200	4200	210,000
Russia	e 14,000	14,000	2,400,000
Saudi Arabia	e 9000	8500	1,400,000
Senegal	e 2600	2500	50,000
South Africa	1990	1600	1,500,000
Syria	e 1100	800	250,000
Togo	e 1500	1500	30,000
Tunisia	3560	3600	2,500,000
Turkey	e 900	800	71,000
Uzbekistan	e 900	900	100,000
Vietnam	e 2000	2000	30,000
Other countries	750	800	800,000
World total (rounded)	228,000	220,000	74,000,000

^e Estimated. ¹ For Australia, Joint Ore Reserves Committee-complaint or equivalent reserves were 120 million tons. ² Production data for large mines only, as reported by the National Bureau of Statistics of China.

). From the total production of 220 million tons in 2023, China is the world’s largest producer of phosphate rock (41%), followed by Morocco (with the Western Sahara mines) (16%), USA (9%), and Russia (6%) [51].

The global phosphorus rock marked size is valued at USD 22.30 billion and a majority of mined phosphate rock is transformed into fertilizers, representing over 75%, in terms of revenue in 2021 [52]. Food and feed additives represent approximately 10%, and the remaining revenue is represented as industrial and other areas [52]. While only 10% of the phosphate rock is in feed and food industry, it represents a business of approximately USD 2 billion, and there is expected lucrative growth in the years to come [52]. Even though there is no immediate shortage of phosphate rock, a peak production has been predicted for the 21 century, with some predictions anticipating to reach a peak between 2035 and 2045 [18,53].

Before it is ready to be utilized as a product, the mined phosphate rock is acidified using sulfuric acid, producing phosphoric acid (H₃PO₄) as the key intermediate in P-based products and calcium sulfate (gypsum) as a by-product [54,55,56]:

$$C{a}_3{\left(P{O}_4\right)}_2+3H_{2}S{O}_4+H_{2}O\to 3CaS{O}_4-2H_{2}O+2H_{3}P{O}_4$$

$$PhosphateRock+SulphuricAcid+Water\to Gypsum\left(hydrated\right)+PhosphoricAcid$$

The wet process produces phosphoric acid, which is suitable for use as a fertilizer [55,57,58]. However, a more concentrated and purer product is required for feed additive application [55]. This higher purity is achieved through a thermal process where P is sprayed into a furnace and burnt in air with the addition of steam [54,57,59]. From a burner tower, the product passes directly into an hydration tower where the gaseous P oxide is recycled into H₃PO₄ [57]:

$$P_2{O}_5+3H_{2}O\to 2H_{3}P{O}_4$$

$$Phosphoruspentoxide+Water\to PhosphoricAcid$$

From the phosphoric acid base product, it is possible to produce different inorganic feed phosphates [54,55]. The most commonly used for animals, based on their constant composition, availability, and low impurities, are ammonium, calcium, magnesium, and sodium phosphates [60]. The different phosphates can be used in their mono-, di-, or tribasic forms depending on their hydrogenation status. Some common feed additives include monocalcium phosphate (Ca(H₂PO₄)₂), dicalcium phosphate (CaHPO₄), monoammonium phosphate ([NH₄][H₂PO₄]), and monosodium phosphate (NaH₂PO₄). The industrial processing of P is illustrated in Figure 3.

Crystalline solid phosphorus pentoxide (P₂O₅), formed by burning elemental P in dry air, is used industrially as a desiccant and catalyst [54]. More importantly, in the phosphate industry, P₂O₅ is a standard measure to express P content in various products, including fertilizers and phosphate rock [54,61]. This standardization supports accurate mass balance and profitability assessments.

The stoichiometric relationship between P, phosphoric acid, and P₂O₅ is as follows:

$$P_2{O}_5+{3H}_{2}O\to 2H_3{PO}_4$$

$$P_4+{5O}_2\to 2P_2{O}_5$$

A standard purified phosphoric acid solution containing 85% H₃PO₄ is equivalent to 61.5% P₂O₅ or 26.9% P [54].

5. Methods for Determining Phosphorus Requirements

5.1. Principles and Practices in Phosphorus Nutrition

Due to its essential role, P is classified as a micromineral or microelement in the context of fish farming, indicating that it is required in gram quantities [1]. This classification distinguishes P along with other macrominerals, such as calcium (Ca), magnesium (Mg), sodium (Na), potassium (K), and chloride (Cl), from microminerals and ultra-trace elements, which are required in much smaller amounts—typically ranging from 1 to 100 mg/kg and from 50 to 1000 μg/kg of the diet, respectively.

5.1.1. Differentiating Between Digestible and Available P

Before P is added to an aquafeed, the digestible and/or available levels of the raw material for the fish species in question must be determined. Digestibility and availability are two distinct physiological concepts, both used to assess nutrient absorption from feed. However, the misuse of these terminologies is currently a source of confusion in the literature relating to P.

Hua and Bureau highlighted that most data on P availability (aP) of feed ingredients for salmon nutrition actually refer to P apparent digestibility [62]. Misinterpreting these terminologies can result in excessive P inclusion in feeds, exceeding the fish’s biological requirement. This can unnecessarily increase feed cost, as P is an expensive raw material, and lead to excessive P waste production and subsequent detrimental effects on the environment from aquaculture facilities, e.g., eutrophication [63,64]. Clearly differentiating between digestibility and availability is thus crucial for sustainable P use.

Digestibility expresses the degree of which macro- and micro-nutrients are digested and absorbed by the gastrointestinal tract [65]. It is calculated as the difference between a nutrient’s level in the feed and the feces, where the feces are composed of undigested feed and endogenous losses from the intestine. Digestibility can be expressed as apparent digestibility (Ad) and the true digestibility (Td).

Td accounts for endogenous losses and therefore accounts for the absorbed and used nutrients by the body [66]. The true digestibility coefficient (TdC) is calculated as follow:

$$\mathrm{T}\mathrm{d}\mathrm{C}(\%)={\displaystyle \frac{\mathrm{Dietary}\mathrm{nnutrients}\mathrm{cconsumed}-\left(\mathrm{dietary}\mathrm{nutrients}\mathrm{excreted}\mathrm{in}\mathrm{faeces}-\mathrm{endogenous}\mathrm{losses}\right)}{\mathit{Dietary}\mathit{nutrients}\mathit{consumed}}}\times 100$$

Ad only reflects the proportion of a nutrient absorbed by the gastrointestinal tract, without accounting for endogenous losses. In aquatic environments, measuring nutrient digestibility in fish is more challenging than in terrestrial animals, as feces starts leaking nutrients immediately after excretion, gasses dissolve, and urine collection is difficult. Because of these challenges, estimations on the apparent digestibility coefficient (AdC) is preferred over the TdC for studies in fish.

The AdC is calculated using an inert marker added to the feed [67,68]:

$$AdC(\%)=100-100\left[\left({\displaystyle \frac{I_D}{I_F}}\right)\times \left({\displaystyle \frac{N_F}{N_D}}\right)\right]$$

where $I_D$ represents the inert marker in the diet, $I_F$ represents the inert marker in the feces, $N_D$ represents the actual nutrient in the diet and $N_F$ represents the actual nutrients in feces.

However, digestibility only considers the feed as an input and does not capture the additional mineral absorption from water [1,69,70]. Therefore, the concept of dietary fecal excretion of minerals has become more common [70,71,72]. This metric represents the percentage of ingested minerals that are excreted through feces, calculated as 100-AdC (%) of each mineral [73,74,75].

For P, the overall digestible P (dP) depends on the TP content of the diet, where higher TP levels reduces digestibility [62]. Digestibility can also be affected by the chemical form in which P is present, feed-processing parameters, the content and chemical form of phytate P in vegetable ingredients, water temperature, water chemistry, grinding level of feed ingredients, physiological state of the animal, fish species, previous P status in fish, and interactions with other minerals in the diet [62,76].

When the term digestibility is used to describe a feed or a raw material, it addresses the fish’s ability to digest and absorb nutrients in their digestive system. However, it will not address how much of the TP in the raw material had the potential to be absorbed. The term bioavailability, biological availability or simply availability may be better suited as it refers to the proportion of an ingested nutrient that is made available for an intended mode of action e.g., the formation of bone [77,78]. Availability can be determined either through qualitative measurements, where a specific response unit in the fish is measured, or a quantitative measurement where the total retention is measured [79,80,81,82]. Unlike digestibility studies, availability studies are not affected by dietary TP levels. The variables measured in these studies reliably reflect the true value of the tested source after digestion and absorption in the gastrointestinal tract.

5.1.2. Conducting Requirement Studies

P requirements in fish are typically determined through empirical feeding trials that measure biological responses, such as growth, bone mineralization, or whole-body P content, across graded dietary P levels [1,81,83]. These studies often use either qualitative indicators (e.g., bone ash, blood Pi levels) or quantitative metrics (e.g., retention efficiency, whole-body analysis) to estimate the point at which additional dietary P no longer yields physiological benefits [79,84].

In salmonids, both growth and bone mineral content have proven to be sensitive and reliable endpoints for assessing P utilization [80,85]. However, bone-based methods primarily reflect relative aP and may underestimate TP requirements, as they do not account for P used in soft tissues or metabolic processes [1,2,86,87]. Consequently, retention studies and whole-body analyses are increasingly favored for their ability to capture total P utilization, despite being more technically demanding.

An alternative to empirical trials is the factorial modeling approach, which estimates P requirements based on theoretical equations incorporating growth rate, endogenous losses, feed efficiency, and dietary availability. Originally proposed by Pfeffer and Pieper (1979) [88] and later refined by Shearer (1991, 1994) [89,90], this method allows for predictive modeling of nutrient needs under varying production conditions [88,89,90]. Although validated against empirical data in salmonids [80], its adoption remains limited, likely due to the complexity of parameter estimation and the scarcity of species-specific data.

Overall, while empirical methods remain the gold standard for establishing P requirements, factorial models offer a promising complementary tool, particularly for scenario testing and precision nutrition. Future research should aim to integrate both approaches, leveraging the strengths of each to refine dietary recommendations and reduce environmental P losses.

5.2. Modeling Phosphorus Availability in Feed

In the recent literature, there is unfortunately a lack of estimation models for aP. Hua and Bureau (2006) [62] developed a model for estimation of dP in salmonid fish feed using a multiple regression approach. In their paper, the relationship between the content of dP and dietary levels of P chemical compounds was examined. Hua and Bureau’s model determines the dietary availability of the element (A-factor) in Shearer’s model. The model itself provides a good foundation for further development in terms of availability.

P compounds in different ingredients and diets are characterized into five broad chemical categories: bone P, phytate P, organic P (non-bone P and non-phytate P), ca monobasic/Na/k/Pi supplement, and Ca dibasic Pi supplement. dP is considered a dependent variable in the model, whereas the five broad chemical categories in addition to exogenous phytase are all independent variables [62]:

$$\begin{gathered} P=0.68boneP+0.84Po+0.89Camonobasix,{\displaystyle \frac{NA}{K}}Pi+0.64Cadibasic+0.51{\displaystyle \frac{pytase}{phytate}}-0.02{\left({\displaystyle \frac{\mathrm{p}\mathrm{y}\mathrm{t}\mathrm{a}\mathrm{s}\mathrm{e}}{\mathrm{p}\mathrm{h}\mathrm{y}\mathrm{t}\mathrm{a}\mathrm{t}\mathrm{e}}}\right)}^2 \\ -0.03{\left(boneP\right)}^2-0.14boneP\times Camonobasix,{\displaystyle \frac{Na}{K}}Pi(p<0.0001,R^2=0.96) \end{gathered}$$

The units for all variables are set in g kg⁻¹ except for the phytase/phytate ratio, where the unit is 100 FTU phytase/g phytate. FTU is a measurement unit where 1 FTU is defined as the quantity of enzyme required to liberate 1 μmol Pi min⁻¹, at pH 5.5, from an excess of 15 μmol L⁻¹ sodium phytate at 37 °C kg⁻¹ feed [91].

However, as also stated in the article from Hua and Bureau (2006) [62], the model does not incorporate factors such as endogenous losses and feed conversion rates (FCRs), and because of this, the margin of errors might be considerable. If the aP requirement is determined at higher biological feed conversion rates (bFCRs), such as 1.0, but the farmers achieve a lower bFCR of 0.6, the amount of aP retained by the fish would be lower than their actual requirement [92]. This hypothesis has not been tested but a review from 2018 [93] analyzed metadata from P requirement studies in fish, which revealed a pattern where decreasing FCR showed an increase in dietary P requirement per unit weight gain.

5.3. Statistical Method

The multiple regression analysis used in the model from Hua and Bureau (2006) [62] is fundamentally the same as the analysis of covariance (ANCOVA), since covariance can rely both on the dependent variable and the independent variable in each analysis. ANCOVA does reduce the error term from the straight-line analysis of variance (ANOVA), but it is necessary to assume that regressions be linear, that the slopes of the regressions are the same, and that the error term of variance must be equal, with which proximate composition data might not always be correct. In a 1994 study [90], Karl D. Shearer recommended allometric analysis over ANOVA and ANCOVA because of its ability to account for the effect of size on composition when there is a large group of individuals examined with different sizes, which often is the case in the commercial aquaculture production of salmonids. Shearer explained the method based on a 1936 study by Huxley [94] and a 1968 study of Laird [95], by using the log weight of a proximate component that is regressed against log fish weight, creating a relationship that is linear within a life cycle stage:

$$\left(b=1\right)=Bothcomponentsincreaseequally$$

$$\left(0<b<1\right)=Bodyweightincreasefasterthancomponent$$

$$\left(b>1\right)=Thecomponentmakeupanincreasedproportionofthefishweight$$

$$\left(b=0\right)=Thecomponentdoesnotincrease$$

$$\left(b<0\right)=Theabsoluteamountofthecomponentdecreases$$

Allometric analysis should be performed at the end of an experiment or trial on measurements from each treatment, and regression equations should be compared using an appropriate multiple range test. This procedure would help the assessment of P transferal from feed to fish and predictably modify feed carcass composition because both exogenous (diet and environment) and endogenous factors (genetics) affecting the proximate composition of the fish have been accounted for. Shearer also recommends that the proximate composition of fish should be reported on a wet weight basis to eliminate the possibility that change in one component affects the relative amount of the others. This recommendation is based on a study on rainbow trout (Oncorhynchus mykiss) from 1978 by Papoutsoglou and Papaparaskeva [96].

6. Dietary Requirements and Commercial Diet Composition

P requirements in salmonids vary throughout their life cycle, influenced by growth stage, size, and physiological status. Whole-body TP (Wet weight basis, WW) content is relatively stable during early development but fluctuates during hatching, juvenile growth, smoltification, and maturation [90,93,97]. For example, TP content is typically in the range of 0.46–0.59% WW in juveniles [80,81,82], declines to 0.52 to 0.40% WW in smolts [70,73,98,99], and stabilizes at 0.35–0.41% WW in adult salmon [69,99,100]. Triploid salmon often exhibit lower vertebral ash content than diploids, indicating altered P retention and potentially higher dietary needs [101].

Despite these physiological variations, commercial salmonid diets often exceed the actual P requirements of the fish. This discrepancy is partly due to the lower aP in plant-based ingredients [102,103,104] and variability in fishmeal quality [19,76,105,106].

Table 2. Requirement of phosphorus in diet for Atlantic salmon and rainbow trout [80] compared against commercial feed content of TP (%) for differently sized salmonids.

Requirement (g/kg) of Diet	Species	Fish Weight (g)	Method of Dose–Response Analysis	Reference
10–11	Atlantic salmon	1.4	Regression	[80]
6	Atlantic salmon	6.5	ANOVA	[85]
6	Atlantic salmon	57	ANOVA	[107]
7–8	Rainbow trout	1.2	Broken line	[108]
3.4–5.4	Rainbow trout	35	ANOVA	[109]
2.4–5.9	Rainbow trout	50	Regression	[110]
Commercial salmonid diets
16	Salmonids	Starter		[111]
14	Salmonids	Fingerling		[111]
13–14	Salmonids	Grower		[111,112]

summarizes the estimated dietary P requirements for different salmonid species and life stages, alongside typical TP levels in commercial feeds.

The primary concern regarding P deficiency in salmonids in poor mineralization and increased prevalence of skeletal deformities. The bones and scales of salmonids consist of hydroxyapatite embedded in a collagen matrix, and a good skeletal condition depends on sufficient Ca and P intake [113]. Insufficient intake or impaired uptake of P can result in poor mineralization, and several studies have connected dietary P deficiencies in freshwater-reared salmon (1.3–60 g) over a certain period to impact deformities of slaughter-sized fish [93,114,115,116,117]. However, deformities in Atlantic salmon, which affect the vertebral centra but not intervertebral joints, have the potential to fully recover in the seawater period (724–4519 g kg⁻¹) [118,119,120].

To ensure proper mineralization and bone strength, the balance of Ca to P in the bone matrix along with indicators such as bone density, mechanical strength, and presence of deformities can be evaluated. For salmonids, the mass ratio of Ca:P has been shown to vary with life stages and environmental factors due to changes in the relative amount of bone, skin, and soft tissue as the fish grows [82,93,98,121]. For Atlantic salmon, the Ca:P ratio changes from 0.2–0.4 at first feeding (>1 g) to around 1 in the pre smolt stage (64 g) [93,98]. A decrease to 0.6 occurs during and after smoltification (125 g) before stabilizing at 0.7–0.8 in adults (1–4 kg) [93,98].

Other P deficiency signs for several fish species include increases in liver or body fat, reduced blood phosphate levels, and poor mineralization of scales [1]. For young fish, scales appear to be the most sensitive indicator of P deficiency [122,123]. There is also evidence that P deficiencies can trigger resorption from readily available P in scales of zebrafish (Danio rerio) [124], similar to that observed in Atlantic salmon during upstream spawning [125]. Additionally, triploid Atlantic salmon are more prone to develop vertebral deformities than diploid salmon when fed diets containing aP levels at or below the estimated requirement for Atlantic salmon fry to 3 kg (8 g kg⁻¹) [119,126].

7. Sources of Phosphorus in Aquafeed

7.1. Marine and Animal Ingredients

Fish bones have long been used as a mineral source in the production of animal and fish feed due to their natural richness in minerals like Ca and P [1,106]. Fishmeal is produced by boiling the whole body of fresh fish (e.g., herring, blue whiting, menhaden) [127,128], followed by pressing and drying to remove the oil [106]. The extracted fishmeal is rich in ash (17–25%), which consists primarily of minerals of which Ca and P are the major constituents [106].

In aquafeeds for carnivorous fish like salmonids, fishmeal is vital primarily due to its high protein content [106,129,130], balanced amino acid profile [106], and significant P content, typically ranging from 0.9 to 1.5 TP kg⁻¹ [69]. However, the dP from fishmeal varies greatly, in the range of 20–60% [76,105,126]. Marine by-products and trimmings are increasingly used in feed production and generally offer aP levels comparable to those of fishmeal, e.g., blood and poultry meal with aP levels of 81% compared to menhaden meal with 87% aP [127,131].

While the use of marine by-products and trimmings offer a sustainable mineral source, the presence of insoluble complexes like hydroxyapatite (Ca₁₀(PO₄)₆(OH)₂) can reduce P digestibility [126,128,132,133]. The hydrolysis of fish bones, such as from herring or blue whiting, improves P solubility and bioavailability by converting these complexes into more accessible forms [78,128]. This hydrolyzed P is particularly effective in fish above 15 g, as smaller fish may struggle to utilize it efficiently [134].

7.2. Vegetable Ingredients

Vegetable feed ingredients, such as cereals, legumes, nuts, and oilseeds, contain 60% to 80% of their TP bound in the form of phytate (IP6), an antinutrient that serves as the storage form of P in plants [135]. Most vegetable feed ingredients contain IP6 in the range of 5–30 g kg⁻¹ [136]. The unique structure of IP6, with six negatively charged phosphate groups, facilitates strong binding with di- and trivalent anions [137] and possibly to larger structures such as proteins and starch, negatively affecting the availability of the nutrient [138,139].

Monogastric fish, including salmonids, lack sufficient levels of endogenous phytate-degrading enzymes, such as acidic and alkaline phosphatase, to cleave the phosphate bond in the IP6 molecule and access the P [135,140]. IP6 also chelates with positively charged cations (Ca²⁺, Mg²⁺, Zn²⁺, and Fe²⁺), with an especially high affinity for Zn, in addition to lowering the digestibility of protein and amino acids [1,141,142].

Generally, the aP in phytate is estimated to be 0% for salmonids [62], because of the low activity of phytases in untreated raw ingredients (

Table 3. Estimated available phosphorus levels in common feed ingredients for fish, calculated from Kumar et al. (2011) [102,103,104].

Ingredients	aP Without Pre-Treatment (g kg−1)	aP Without Pre-Treatment (%)
Maize	0.35	15
Maize gluten	0.80	16
Corn	0.80	32
Gross defatted corn germ and bran	2.40	36
Fine defatted corn germ and bran	4.30	36
Hominy meal	0.70	10
Rice bran	1.68	10
Rice	0.40	33
Rice broken	0.45	53
Rice polishing	4.40	28
Wheat bran	2.60	24
Wheat by-products	1.02	13
Wheat	0.88	29
Sorghum	0.51	17
Barley	0.87	32
Oats	0.33	14
Oats, dehulled	1.10	49
Groundnut meal	1.40	23
Palm oil meal	2.20	43
Soybeans, whole	2.47	45
Soybean meal	2.13	32
Coconut meal	1.90	44
Cotton seed, whole	1.80	30
Cotton seed meal	2.25	20
Sunflower meal	1.57	17
Rapeseed meal	4.80	41
Canola meal	2.07	24
Peas	1.78	52
Faba beans	1.58	39
L. albus, whole	1.98	44
L. angustifolius, whole	1.50	48
L. angustifolius, dehulled	1.91	50

) [102,103,104,143].

The processing of vegetable ingredients to enhance protein quality or reduce antinutrients can inadvertently lower aP levels. On such example is soy protein concentrate (SPC), which is used in diets for salmon due to its protein profile [73]. When SPC is produced from soy beans, it can drastically increase the concentrations of IP6 up to 80 g kg⁻¹ [144], resulting in a significantly lower aP level in the final ingredient. This can be a severe issue since a salmonid SPC diet results in lower retention rates and higher losses of P compared to a fishmeal-based diet [145].

While microbial phytases can be added to improve aP, its effectiveness is limited by the extrusion process used in aquafeed production. High temperatures (90–140 °C) and pressure (39 to 56 bar) during extrusion denature both endogenous and added phytase, significantly reducing its activity [146,147,148,149].

To overcome this, several strategies have been developed to increase aP, including milling, fermentation, germination, and enzymatic pre-treatment [103,150]. For example, pre-incubating SPC at 40–45 °C preserved phytase activity and improved P and Mg availability in Atlantic salmon [151]. In rainbow trout, microbial phytase treatment increased dP from 25% to 57% and utilization of P from 17% to 49% [152]. Fungal phytase supplementation has also raised aP from 6% to 64% in plant-based diets [153]. Particularly for salmonids, these methods of increasing aP are preferred, since traditional phytase coating has shown to be ineffective for cold water species such as salmonids reared at water temperatures of 8 °C.

However, when phytase is added to diets that already meet the salmonid P requirements, it may increase the dissolved and suspended fraction of P in the effluent, complicating environmental management [153]. Moreover, current trial designs often cannot distinguish between P leached from feces, uneaten feed, or excreted as DIP in urine [154]. Notably, P not bound to phytate in plant ingredients is generally highly available, with ADC exceeding 84% [62].

7.3. Inorganic Phosphorus Supplements

Despite the P content in marine, animal, and vegetable ingredients, additional Pi sources are necessary to ensure adequate levels of aP in fish diets [155]. Common Pi sources in commercial feed formulations include monocalcium phosphate (MCP), dicalcium phosphate (DCP), ammonium phosphate (MAP), and monosodium phosphate (MSP) [54,55]. A 2018 trial demonstrated that monobasic phosphate had the highest availability, followed by di- and tri-basic phosphates, plausibly due to higher solubility or mineral interactions [156].

A 1997 study determined P retention from fish bone meal, MCP, DCP, and MSP in Atlantic salmon (5.6 to 10 g) using semi-purified casein-based diets deficient in TP (0.42%) supplemented with different inorganic P sources. The trial showed high P retention for DCP (86%) and MCP (91%), intermediate retention for fish bone meal (51%), and the lowest retention for MSP (13.1%) [81]. The results suggest that mineral interactions and solubility affect P retention and should be considered when predicting availability from Pi sources.

8. Phosphorus in Salmonids

8.1. Absorption, Digestion, Storage, and Loss

Salmonids primarily absorb P through their gastrointestinal tract [1,157]. The low pH in the stomach helps decalcify bone tissue and scales, by removing the Ca ions, softening the bone and scales, and making minerals easier to absorb [158].

During digestion, hydrochloric acid (HCl) activates pepsinogen to pepsin, which mixes with food before transport to the middle intestine [158]. Gastric acidity (pH 1–5) is crucial for mineral bioavailability, affecting absorption in the pyloric caeca and intestine [157,159]. A study on rainbow trout showed that the pyloric caeca is responsible for 89% of Pi absorption, mostly diffusive and unregulated [160]. In the intestine, the pH increases to 8-9 before dropping to neutral 7 in the hindgut, with absorption occurring actively through cells or passively via the paracellular pathway, aided by the NaPi-IIb transporter [1,161]. Under these alkaline conditions, Pi uptake is primarily mediated by the sodium-dependent NaPi-IIb transporter, with Pit1 and Pit2 acting as auxiliary phosphate sensors [20]. P absorbed in the intestine is transported in the bloodstream, potentially accumulating in the blood. Pi transporters in intestinal cells and paracellular pathways move Pi into the plasma [162]. Hemoglobin in fish is sensitive to Po, with 2,3-diphosphate binding to hemoglobin to regulate oxygen delivery [1,163]. The blood serves as a P pool, exchanging P with the main storage categories in the salmonid body.

• Bones and scales

Bones and scales are the largest storage sites of P in the fish, where it is deposited as tricalcium phosphate Ca₃(PO₄)₂ and further crystallized into Ca₁₀(PO₄)₆(OH)₂ that is deposited in the organic matrix during the mineralization process [78]. Ca₁₀(PO₄)₆(OH)₂ is also present in the calcified portion of fish scales in rainbow trout and a range of teleost fish [122,164].

2.

Soft tissue (heart, liver, muscles, and kidneys)

P not bound in bone includes inorganic phosphate salts, i.e., orthophosphates and phospholipids [87]. Phospholipids are essential constituents of the cellular membranes and typically contain a phosphate, two fatty acids tails, and glycerol [86].

Plasma Pi diffuses into the urine via the kidneys’ glomerular filter and is reabsorbed by Pi transporters in the renal proximal tubule [162]. Unreabsorbed Pi is excreted in the urine, while indigestible dietary P (uaP) is excreted in the feces. Urinary P excretion accounts for most DP emissions, while fecal P represents the majority of PP waste in fish culture operations [62,162].

Uptake and balance mechanisms in fish are illustrated as follows (Figure 4).

A study by McDaniel et al. (2005) [166] found that the oxygen level significantly influences P absorption in the intestines of rainbow trout. Specifically, increasing dissolved oxygen from 6 to 10 mg/L enhanced both the absorption and utilization efficiency of dietary P. In fish fed diets containing 0.7% and 1.0% TP, this increase in oxygen shifted effluent P from the fecal to the soluble fraction in both groups [166]. The study also reported improvements in the feed efficiency ratio (FER) and specific growth rate (SGR) with higher oxygen levels, regardless of the dietary TP content. This suggests that elevated oxygen enhances nutrient utilization, likely due to its role in the oxidative breakdown of dietary components [167]. However, the group fed 1.0% TP also showed higher levels of soluble P in the effluent [166].

8.2. Hormone, Vitamin, and Mineral Interactions

The serum levels of P in the blood regulate the amount of P absorbed through feed and the surrounding water, as well as reabsorption in the kidneys and mineralization in bone [1]. Given the low concentrations of P in aquatic environment, extruded feeds have become the primary source of P for farmed salmonids. Thus, while serum P directly regulates P metabolism, dietary composition plays a central role by determining the availability of P and influencing serum P levels [78]. For example, when the dietary TP concentration is elevated, the absorption and deposition of P from feed into the fish are reduced, and vice-versa [1,168,169,170].

Interactions among minerals play a significant role in regulating P levels within the fish’s body. For example, an elevated dietary intake of P can lead to a reduction in the absorption of Zn [69]. The Ca to P ratio is also crucial for ensuring optimal P absorption; if P levels are adequate but Ca-deficient, the absorption of P would be adversely affected [1]. Furthermore, minerals can interact indirectly with one another. A study on rats showed that diets with sufficient amounts of dietary Ca increased Ca absorption when dietary Mg increased, which subsequently influenced P absorption [171].

While the diet serves as the primary regulator of P metabolism in salmon and fish in general, several hormones play an important role in maintaining P balance in fish [165]. Several studies on zebrafish (Danio rerio) show that stanniocalcin (STC) tends to increase Pi blood levels, while parathyroid hormone (PTH) and fibroblast growth factor 23 (FGF23) lower Pi blood levels [165]. Additionally, deficiency studies in rats suggests that the hormone prolactin influences organs involved in Ca regulation, thus indirectly affecting the regulation of serum P levels [172].

STC is an homodimeric glycoprotein hormone produced by the corpuscles of Stannius, which are special endocrine organs in fish’s kidney. In both freshwater- and seawater-reared salmon, STC plays a similar integral role in Ca²⁺ and phosphate homeostasis [173]. When extracellular Ca²⁺ levels increase, it promotes the synthesis and release of STC in the blood, which reduces gill and gut Ca²⁺ transport and renal phosphate excretion to restore the basal level of Ca in the blood (~2 mM Ca²⁺) [173]. Recent findings show that this Ca²⁺-stimulated STC release is mediated by a calcium-sensing receptor (CaR) in salmonids, homologous to the mammalian CaR [174]. The use of CaR mimetics confirmed this mechanism, suggesting that CaR plays a broader role in mineral homeostasis by linking extracellular Ca²⁺ levels to the hormonal control of both Ca and P balance.

The PTH family is a group of structurally related peptides involved in bone mineral homeostasis and various developmental processes in vertebrates [175]. In teleost fish, the synthesis of the parathyroid related protein (PTHrP) is distributed amongst several tissues, with two important ones being the pituitary gland and the corpuscles of Stannius [176,177]. The PTHrP is essential for the Ca balance as it mobilizes Ca from bones and scales and regulates Ca uptake [176]. PTH is therefore subsequently active in regulating P, in addition to stimulating vitamin D endocrine production, which in turn can potentially affect P absorption [1].

Dietary vitamin D3 (cholecalciferol) is absorbed in the intestine and is enzymatically converted in the liver to 25-hydroxyvitamin D3 (25(OH)D₃), before being converted to its active form 1,25-Dihydroxyvitamin D (1,25(OH)₂D₃) in the kidneys [178,179]. PTH stimulates the production of pre-vitamin D3 to cholecalciferol [179], and an increase in circulating 1,25(OH)₂D stimulates the intestinal absorption of both Ca and P [1]. While some studies indicate species-dependent responses to vitamin D metabolites in mobilizing P for freshwater eel (Anguilla Anguilla) [180] and common carp (Cyprinus carpio) [181], other studies show no specific link in studies on rainbow trout [182,183]. However, a strong correlation between gastrointestinal disorders, vitamin D system malfunctions, and fish bone diseases has been confirmed [184,185], emphasizing the critical link between vitamin D and bone-forming minerals like Ca, P, and Mg.

The FGF23, which is a hormone referred to as phosphatauric peptide, is mainly secreted by the bone-forming cells osteocytes and osteoblasts, and it is an important hormone in regulating serum phosphate levels through two separate mechanisms [186]. With the first mechanism, FGF23 can reduce the expression and/or insertion of the sodium-phosphate transporters (type IIa and IIc) within renal proximal tubular membranes, allowing more phosphate to be excreted. Through the second mechanism, FGF23 inhibits renal 1(OH)ase expression and promotes 24-hydroxylase, which reduces 1,25(OH)₂D levels and subsequently reduce the amount of phosphate that is absorbed from the gut and bone [186]. In rats, FGF23 can also inhibit PTH mRNA synthesis [187], but it is uncertain if these mechanisms are the same in salmon.

9. Relevance for Farming Practices

The majority of the salmon and trout produced in Norway today is produced in traditional open net pens, located in coastal regions of the country [188,189]. While their permittable net barrier is advantageous to production, it is not secured for escapees, parasites, bacteria, diseases, feces, feed, and pollutants to enter pens or spread to the outer environment [189]. This is a problem in terms of the effective utilization of P and P circularity, since feed and feces constitute the main P effluent from aquaculture facilities. Although these challenges are shared across major salmonid-producing countries [190,191], Norway’s dominant role in global production and its advanced technological infrastructure make it a particularly relevant case for examining P dynamics in detail.

For the average entire Norwegian salmon production, independent mass balance studies carried out between 2010 and 2020 show an apparent P retention of 27% (2010), 29% (2012), 18% (2016), and 25% (2020), meaning that significant amounts of 73, 71, 82, and 75% P, respectively, were excreted as waste by salmon [8,99,192]. With such a high effluent of P originating from salmon-farming facilities, the European Union (EU) subjects farmers to strict guidelines in terms of effluent regulation [14,15,16,193]. While continuous measurements of waste released from the aquaculture industry are not feasible, modelling tools have become the preferred manageable monitoring system [194].

$$\mathrm{Gross}\mathrm{Waste}\mathrm{P}=\left(\mathrm{Feed}\mathrm{given}\times {\displaystyle \frac{\mathrm{P}\mathrm{content}\mathrm{of}\mathrm{feed}}{100}}\right)-\left({\displaystyle \frac{\mathit{Biomass}\mathit{of}\mathit{fish}\times Pcontentinfish}{100}}\right)$$

To help maintain the rate of production while minimizing the emissions generated during aquacultural operations, it is expected that more of the future production will be placed in TIF systems [11]. The enclosed rearing environment of TIF solutions provides a barrier that reduces the risk of fish escaping and limits the contact between cultivated fish and parasites [195,196]. One such TIF solution is the recirculating aquaculture system (RAS), which has the ability to reduce makeup water volume requirements and purify nutrient-rich waste water through a combination of mechanical, biological, and chemical water treatment steps [197,198]. This allows for partial or full reuse of water within the fish culturing system [199,200], resulting in a reduced nutritional discharge compared to flow-through systems (FTSs).

The different TIF solutions can be categorized according to their respective cumulative feed burden (CFB) [201]. The CFB is the ratio between the ingested feed by fish and the replacement of water volume (kg feed/m³ make-up water), where the intensification of waste treatment required increases along with the CFB.

In accordance with CFB, systems are defined as follows: FTS (<0.04 kg m³), re-use (0.04–1 kg m³), RAS (1–5 kg m³), and fully recirculating aquaculture system (FREA) (>5 kg m³) [202,203] (Figure 5). Additionally, a fifth TIF solution is the hybrid flow-through system (HFT), which has the ability to shift between traditional flow-through technology and re-use [11].

When CFB reaches 1+ kg m³ or higher, as in RAS and FREA, the biofilter treatment of the production water is required to remove the accumulating nitrogenous compounds. The primary nitrogenous waste products are ammonia (NH₃), ammonium (NH₄), and urea (CO(NH₂)₂), which are mainly part of the dissolved waste [205,206,207].

In RAS with low exchange rates or a high CFB, the concentrations of nitrate (NO₃⁻) can exceed 200 mg L⁻¹, whereas systems with a low CFB typically do not exceed 50 mg L⁻¹ [208]. Similarly, RAS with low water exchange rates can lead to a substantial accumulation of minerals in the rearing water [209,210,211,212].

The process of nitrogen conversion through biofiltration has been extensively studied, and the optimization of factors such as oxygen, alkalinity, and suspended solid concentrations and temperature control are known to affect the nitrification efficiency [213,214,215]. Given the intensified production and stricter regulation on P, a corresponding overview of P management in relation to the technical system is required to avoid unnecessary P waste production. Such information is today limited.

To avoid potential negative effects on salmon produced within RAS and FREA, the recommended P concentration in production water is around 3 mg L⁻¹ [211]. This recommendation is based on a review of several studies [167,216,217,218,219,220,221,222,223,224]. In contrast, for FTS productions, the P concentrations will typically range around the natural concentration in water, which usually does not exceed 0.5 mg L⁻¹ [19]. Increasing concentrations to levels exceeding 3 mg L⁻¹ is thus only possible in low water exchange systems, as it requires a substantial degree of recirculation [209,210,211,212].

With the increased accumulation that follows lower water exchange, salmonids like rainbow trout have been shown to cover their Mg requirement of 330–600 mg kg⁻¹ through the water, provided the concentration is at least 46 mg L⁻¹ [1]. It can therefore be speculated if the salmon can meet its requirement of P, purely through water when the CFB increases.

By estimating water concentrations of P up against drinking rates for seawater-adapted salmon (4–6.4 mL kg h⁻¹ [225,226]), and assuming a simplified example where a salmon reaches 1 kg in approximately 70 days, it can be calculated that the salmon would need to drink 6.72–10.75 L of the production water. If the average production water in a RAS contains 3 mg P L⁻¹, the salmon can only cover 20–32 mg kg⁻¹ growth, which is insignificant compared to the dietary requirement of 6000–10,000 mg kg⁻¹ [1]. To fully cover the P requirement of the fish, the water concentration would need to be in the range of 893–930 mg P L⁻¹.

While RAS and FREA systems are generally better suited for P removal through water treatment compared to FTS and HFT systems, all four primarily rely on mechanical filtration as the main treatment method [22,227]. Mechanical filters are effective at removing particulate-bound P, which constitutes approximately 52% of the total P excreted by salmonids. However, around 24% of the excreted P is in the form of DIP, which cannot be effectively captured through mechanical filtration [228]. Given these limitations, it is important to consider both the technical and economic aspect of P removal strategies in aquaculture systems.

• Economic Considerations and Available Methods for Phosphorus Removal

To capture the fraction of soluble P compounds in effluents from salmonid aquaculture, chemical precipitation methods, such as the addition of aluminum, ferric chloride, or calcium ions, can be employed to remove P from the water [229]. Although chemical treatment is technically compatible with RAS and FREA and may involve a lower upfront investment than in FTS and HFT systems, the high capital and operational cost still represent a significant economic burden for most aquaculture operations [230].

An alternative or complementary strategy is the use of biological P removal through integrated trophic aquaculture (IMTA) systems. In these systems, extractive species such as blue mussels (Mytilus edulis) and macroalgae (Saccharina latissimia) can assimilate dissolved and particulate P from salmonid farm effluents. While large-scale mitigation effects remain under investigation, early studies suggest that these species can contribute to nutrient uptake and reduce environmental loading when appropriately scaled and positioned [231]. Some studies also report increased biomass yields and modeled scenarios where a price-premium IMTA-produced salmonid significantly improves profitability, suggesting potential consumer willingness to pay more for sustainably farmed products [231,232]. However, broader adaptation remains limited due to high capital and maintenance costs, operational complexity, skepticism stemming from a lack of commercial-scale evidence, and regulatory barriers such as insufficient government support and inflexible regulation-permitting frameworks [231,232].

Consequently, a more broadly effective and potentially more economically viable strategy than implementing chemical or biological treatment can be to reduce TP in the diet while increasing the proportion of aP [21]. This dietary approach not only reduces the need for costly water treatment infrastructure but also enhances the efficiency of mechanical filtration by promoting the binding of P in fecal particles [112]. Minimizing TP while optimizing aP is also a suitable strategy to reduce P emissions from open-pen production systems, where P treatment is not applicable due to technical aspects or spatial constraints, which limits the scalability of IMTA [231]. However, these dietary strategies must be applied carefully to avoid nutrient deficiencies that could compromise fish health and performance. Despite their potential, there is currently limited information on the cost-effectiveness and return on investment of dietary approaches aimed at reducing TP while maintaining aP, especially in comparison to chemical and biological treatment methods. This represents an important area for future research.

10. Conclusions and Perspectives

Unlike previous reviews, this work offers a system-level synthesis that connects P bioavailability, physiological regulation, and aquaculture system design, providing a comprehensive framework for improving P efficiency in salmonid farming. This review explored the multifaceted role of P in salmonid aquaculture, emphasizing the need to balance nutritional adequacy with environmental responsibility. By integrating insights from P chemistry, feed formulation, fish physiology, and system design, it highlights the complexity of optimizing P use across the production cycle.

A key takeaway is the importance of distinguishing between dP and aP, particularly when evaluating feed ingredients and estimating requirements. While marine and animal-derived sources remain reliable, the growing reliance on plant-based ingredients necessitates strategies to overcome phytate-bound P limitations.

Physiological mechanisms, including intestinal transporters and hormonal regulators, play a central role in P homeostasis and should be considered in both dietary formulation and system-level management. Moreover, the transition toward land-based systems like RAS introduces new challenges in P retention and effluent control, underscoring the need for integrated solutions that align feed strategies with system capabilities.

Looking ahead, improving P efficiency will require better availability models, refined requirement estimates under commercial conditions, and innovations in both feed and system design. Aligning these efforts will be essential to support sustainable growth in salmonid aquaculture.