The stakes are considerable. Atlantic salmon ranks among the ten most produced and valuable marine fish globally, with production stable at around two million metric tons since 2000. Norway dominates the market, producing over 1.5 million metric tons in 2023. But cage-based farming is increasingly challenged by climate-driven sea surface temperature increases, which are linked to higher fish mortality, welfare problems, elevated operational costs, revenue losses, and environmental spillback effects. Climate projections indicate global sea surface temperatures will rise by 1 to 3 degrees Celsius over the next century, with the greatest increases expected at higher latitudes—precisely where much of the salmon farming industry is concentrated.

Salmon lice have plagued the industry since its inception in Norway in the late 1960s. Prevention and treatment costs in Norway alone are estimated at €610 million annually. The parasite exerts the greatest economic impact of any aquaculture parasite and also damages wild fish populations, including Atlantic salmon, sea trout, Arctic char, and Atlantic cod. Warmer waters intensify the problem because lice development, larval production, and infectivity are all temperature-dependent; higher temperatures shorten the time from non-infective nauplii to infective copepodites, substantially increasing infection pressure around farms. In Norway, the traffic-light regulatory system can reduce permitted production capacity by 6 percent in areas classified as red due to high lice-related mortality risk for wild salmon smolts.

At the individual fish level, lice attachment and feeding cause epithelial damage, osmotic imbalance, stress responses, immune activation, and reduced condition—all of which suppress feed intake and divert energy away from somatic growth. Delousing treatments compound the problem, as they commonly involve fasting, crowding, pumping, and exposure to freshwater, chemicals, heat, or mechanical forces, each associated with stress, injury, reduced appetite, and short-term biomass loss. Temperature itself has a direct and non-linear effect on salmon growth: feeding, digestion, metabolism, oxygen demand, and feed conversion are all temperature-dependent. Optimal growth for post-smolts and adults in seawater is commonly reported around 13 to 14 degrees Celsius, with performance declining at both extremes.

To identify which modeling frameworks could best handle these interacting pressures, the researchers followed PRISMA guidelines and searched Web of Science and Oria using combinations of salmonid species terms and growth-model keywords. An initial focused search on Atlantic salmon growth models returned 190 papers but yielded only 12 after screening, because most literature concerned the freshwater parr life stage rather than the marine production phase. The search was therefore broadened to include trout, charr, and general salmonid terms alongside specific model names such as Thermal-unit Growth Coefficient, Ewos Growth Index, logistic growth functions, bioenergetic models, Dynamic Energy Budget, Net Energy Balance, and nutritional models. This expanded search produced 712 papers, of which 23 farmed-salmonid studies met the inclusion criteria after 689 were excluded.

Wild salmonid models were systematically excluded from the final synthesis. The authors explain that these models are generally designed for ecological or population-dynamic questions—growth, survival, migration, density dependence—rather than farm-level production forecasting. Their assumptions about food availability, mortality, and management interventions differ fundamentally from aquaculture systems, where feeding is controlled, husbandry interventions are scheduled, and the relevant outputs are production biomass, treatment timing, and harvest outcomes. One additional study was added through targeted citation searching because it directly coupled a temperature-dependent Atlantic salmon growth model with sea-lice control and economic outcomes, bringing the final synthesis to 24 temperature-dependent studies.

The retained models fell into three main classes. Index and phenomenological models—including the Thermal-unit Growth Coefficient, Ewos Growth Index, and logistic growth functions—describe growth statistically without modeling underlying physiological fluxes. They require few inputs, typically temperature or degree-days, initial and final body weight, and growth duration, all routinely available from farm production records. Bioenergetic models represent growth through energy intake, metabolic expenditure, and allocation to somatic growth, storage, or waste, incorporating feed intake, energy density, digestibility, body composition, and temperature-dependent metabolism. Nutritional models focus specifically on how feed composition, digestibility, and nutrient partitioning determine growth, body composition, feed efficiency, and waste production.

Performance varied considerably across model classes. Empirical models often showed strong fit within controlled experiments or region-specific datasets—Thyholdt’s regional logistic growth functions for Norwegian farmed salmon achieved R² values above 0.95—but their transferability depends heavily on local calibration. Notably, temperature effects differed by region: warmer conditions increased growth in Northern and Central Norway but reduced it in Southern Norway, where fish are already closer to upper thermal limits. Mechanistic bioenergetic and nutritional models required substantially more parameters and higher-resolution data but could predict a broader range of outputs, including feed conversion, nutrient deposition, waste production, harvest timing, product quality, and economic consequences. Dynamic Energy Budget models were applied to simulate growth under environmental forcing in several studies, including site-selection analysis in the Yellow Sea and harvest-timing estimation across European systems.

The most striking finding was how rarely biotic stressors are explicitly integrated into growth models. Most retained studies did not represent parasites, bacteria, or viruses as dynamic drivers of growth, feed intake, mortality, or treatment timing. One framework linked deterministic growth models with stochastic host–pathogen dynamics, mortality, and economic loss, but its salmon case concerned viral disease rather than salmon lice. The closest salmon-lice example came from a 2014 study that coupled a Thermal-unit Growth Coefficient-based salmon growth model with empirical lice and treatment profiles, linking treatment events to direct costs, growth delay, added mortality, production loss, and profit in Norwegian farming. However, even this framework used an empirical lice profile rather than a mechanistic lice population model.

The review concludes that no single model class is optimal for all purposes, but for scenario-oriented assessment of warming and salmon lice in Atlantic salmon aquaculture, the most practical near-term framework is a validated, temperature-driven farmed-salmon growth model—locally calibrated with production and environmental data—coupled modularly to a lice-dynamics or infestation-pressure component. Such a framework would require an explicit lice module representing temperature-dependent development, larval production, infection pressure, attachment, and treatment thresholds, together with a stress-effect function translating lice burden and delousing events into reduced growth, appetite suppression, mortality, or delayed harvest. More complex bioenergetic or nutritional components should be added only when feed intake, waste output, nutrient retention, oxygen demand, or product-quality endpoints are central to the research question. As sea temperatures continue to climb and lice pressure intensifies, this modular approach offers the industry a realistic pathway toward quantitative climate adaptation planning.

Subject of Research: Growth modeling for farmed Atlantic salmon aquaculture under ocean warming and salmon-lice pressure

Article Title: Growth models for farmed post-smolt to adult salmonids: a systematic review for Atlantic salmon aquaculture under warming and salmon-lice pressure

Article References: Kleiven, S. K., Abdel-Fattah, D., & Bøhn, T. (2026). Growth models for farmed post-smolt to adult salmonids: a systematic review for Atlantic salmon aquaculture under warming and salmon-lice pressure. Frontiers in Aquaculture, 5, Article 1835152. https://doi.org/10.3389/faquc.2026.1835152

Image Credits: AI Generated

DOI: 10.3389/faquc.2026.1835152

Keywords: Atlantic salmon, aquaculture, salmon lice, growth models, sea surface temperature, climate change, bioenergetic models, Dynamic Energy Budget, Thermal-unit Growth Coefficient, systematic review, fish welfare, Norway