The study complied with the guidelines of the Norwegian Regulation on Animal Experimentation and was approved by the National Animal Research Authority in Norway (FOTS ID 28416).

The experiment was carried out at the Department of Biological Sciences, University of Bergen, Norway, between January and April 2022. Atlantic salmon post-smolts of approximately 200 g were PIT tagged and randomly distributed into nine tanks (volume ca 600 L) of 62 fish per tank. The tanks were supplied with flow-through seawater maintained at a constant temperature of 10°C ± 0.00°C, 28‰ salinity, oxygen saturation above 90% and LD12:12 light regime (from 07:00 to 19:00). After a 3-week acclimation period, the water temperature was gradually adjusted over 72 h to establish three experimental temperature groups (three tanks/treatment): 8°C, 12°C and 15°C. Water temperature, salinity, and oxygen saturation were monitored daily and maintained at constant levels for 8 weeks (Supplementary Table S1). Fish were fed a 3 mm commercial pelleted diet (NOFIMA, Norway) once daily using automatic conveyor-belt feeders, operating for 2 hours, from 8:00 to 10:00. The LD12:12 light schedule was used to signal the upcoming feeding event, with lights switched 1 hour before feeding. Feed was offered in excess, and the daily ration administrated to each experimental group was calculated based on estimated fish biomass, assuming a daily growth rate of 1.2% and a feed conversion ratio (FCR) of 1.2 during the first 4 weeks and 1.4 during the final 4 weeks. The daily amount of feed was weighed with a two-digit scale and placed on the conveyor belt feed the day before to minimise any disturbance before feeding. Uneaten feed was collected and weighed at the end of each meal to quantify the actual feed intake per tank.

Multiple samplings were conducted, with five primary time points to assess fish size and growth: one prior to temperature exposure (week 0), and 2, 4, 6 weeks thereafter. Each sampling was started precisely at the conclusion of the two hour-feeding time, and five fish per tank were randomly collected and euthanized with an overdose of tricaine (200 mg/L NaCO3-buffered Finquel®vet). Fork length and weight were recorded, and liver, and gonads were weighed using a precision electronic balance (VWR International AS, Norway) with an accuracy of three significant digits.

On week 8, an additional 10 fish per tank were sampled at 2, 4, 8, 12, 16, 20, 24 h post-feeding (hpf) to assess the short-term impact of temperature on physiological mechanisms involved in regulation of appetite. To minimize the handling stress associated with frequent sampling, fish from each of the three triplicate experimental tanks were lightly anaesthetised (60–80 mg/L NaCO3-buffered Finquel ®vet Argent Chemical Laboratories, Redmond, United States), combined in a 200L holding tank, and randomly redistributed within their original replicate tanks 2 weeks prior the samplings. On the day of sampling, each tank replicate was designated for specific time points: replicate one for 2, 12, and 24 hpf; replicate two for 4 and 16 hpf; and replicate three for 8 and 20 hpf. Each tank thus served as the representative unit for its assigned sampling intervals. Fish were euthanized with an overdose of tricaine (200 mg/L¹ NaCO3-buffered Finquel®vet). Blood was collected from the caudal vein using 2 mL syringes (BD Plastipak, VWR, Norway) with 23G heparinized needles (Microlance TM3, VWR, Norway) and centrifugated at 5000 rpm for 3 min. The plasma was transferred into 0.6 mL PCR tubes (Axygen, Glendale, Arizona), kept on dry ice, and stored at −80°C until analysis. Fish fork length and weight were measured, and liver, gonads and gall bladder weighed using a precision electronic balance (VWR International AS, Norway) with an accuracy of three significant digits. The gastrointestinal tract was collected and divided into stomach (ST), midgut (MG), and hindgut (HG) by using surgical clamps to ensure that there was no loss or transfer of feed and digesta between these compartments (Kalananthan et al., 2020b). The content from each section was meticulously transferred into pre-weighed plastic bags. The wet content was weighed immediately after each sampling, whereas the dry weight was determined after incubating the samples at 74°C for 24 h to ensure full dehydration. A portion of stomach tissue was collected from each fish, cleaned in 1X phosphate buffered saline (VWR, Oslo, Norway) and stored in tubes with RNA later (Invitrogen, Carlsbad, CA, United States). Whole brains were collected from each fish and immediately placed into tubes containing RNA later. Both stomach and brain samples were refrigerated for 24 h and then transferred to −80°C until further analysis.

After the completion of the multiple samplings, the left-over fish from each experimental group were collected, and their body weight and fork length were recorded as additional data for the final growth and biomass assessment of the experimental groups.

The dry content from each GIT section was quantified to assess the level of GIT fullness in relation to the experimental condition, and to evaluate the GIT evacuation dynamics after a meal: G I T s e c t i o n c o n t e n t   % = S T d w or M G d w or H G d w   W − S T w w + M G   w w + H G w w * 100 where STdw, MGdw and HGdw is the dry weight (g) of the stomach, midgut, and hindgut content respectively. While, W is the fish body weight (g), and STww, MGww and HGww is the wet content (g) of the stomach, midgut, and hindgut.

The gall bladder fulness index (GBI) was calculated as follow: G B I % = W g b   W   * 100 where Wgb is the gall bladder weight (g) and W is the fish body weight (g).

Plasma metabolite levels were analysed using a Pentra C400 clinical chemistry analyzer (HORIBA, Japan). The parameters of interest were quantified using spectrophotometry and the following reagents: ABX Pentra Glucose HK CP kit for glucose, the ABX Pentra Lactic Acid reagent for lactic acid, the ABX Pentra Triglycerides CP kit for triglycerides, and the ABX Pentra Cholesterol CP kit for cholesterol. A minimum of 150 µL of plasma per sample was required in the device to enable simultaneous analysis of all parameters. All assays were calibrated and quality-controlled according to instructions provided by the manufacturer.

Plasma level of ghrelin was quantified by a sandwich Enzyme Linked Immunosorbent Assay (ELISA) (MyBiosource, San Diego, United States) according to the manufacturer’s instructions. Samples were run in duplicates, and the optical density of each well was determined immediately by using a microplate reader set to 450 nm (Tecan, Männedorf, Switzerland). The optical density of each plate was read three consecutive times, and each sample ghrelin concentration was extrapolated from the standard curve by using Akima curve fitting.

Statistical analyses and graphics were performed in R (R Core Team, 2024) using a multi-model inference approach (Burnham and Anderson, 2022). A series of models were fitted including a range of predictor variables and then calculated the second order Akaike information criterion (AIC_c) assuming lower AIC_c indicates better fit. Based on ranked AIC_c, the Akaike evidence weights (w) was calculated for individual ranked models. Finally, the model with the highest evidence weight (i.e., the lowest AIC_c) was considered the “best” within the family. Whenever the best fitting model was determined, individual beta coefficients and effect size values were analysed to understand the major effects. Additionally, the hypotheses that the model parameters significantly differ from zero was also tested. The linear models were fitted using standard or generalized least squares depending on the residual error structure. When significant serial correlation of residuals was detected (violating the assumption of independence for errors) a generalized least squares based on restricted maximum likelihood was used, adding an n-order autoregressive ARMA(n) process (Faraway, 2009; Fox, 2016). The optimal n was determined by fitting models with a range of ns and selecting the model characterized by lowest AIC_c as well as by comparing incremental models using analysis of variance (Chambers and Hastie, 1997; Fox, 2016). This was computed using the R ‘gls’ function from the ‘nlme’ package (Pinheiro et al., 2025). For nonlinear estimation, a standard ‘nls’ function was used (Ritz and Streibig, 2008). Model comparison based on Akaike evidence weights was done using the ‘MuMIn’ R package (Barton, 2023).

For all tests, p < 0.05 was considered significant (∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001). All data are presented as mean ± SEM; otherwise, it is stated. More details of the data analysis and notation can be found in the Supplementary Material.

Temperature significantly affected feed intake across all three experimental groups during the 8-week trial, as previously reported in other studies on Atlantic salmon (Handeland et al., 2008; Elliott and Elliott, 2010; Kullgren et al., 2013). An initial decrease in FI and FIR was observed in response to exposure to the experimental temperatures. Nevertheless, the 12°C and 15°C groups appeared to adapt more rapidly than the 8°C group to the new temperature condition, as evidenced by the earlier and steeper increase in FI and FIR the following weeks. A transient decrease in feed intake was also observed in the days that followed the monitoring sampling, indicative of stress from handling (Madaro et al., 2015; Lai et al., 2023). The impact of stress appeared to be temperature-dependent, with fish in the 8°C group being more affected and exhibiting lower FI and FIR than those in the 12°C and 15°C groups. The differences in FI and FIR among the experimental groups were also reflected in feeding efficiency, as indicated by a FCR of 0.96 for the 8°C group, compared to FCRs of 0.77 and 0.79 for the 12°C and 15°C groups, respectively.

The differences in feed intake observed across the experimental temperatures were reflected in the growth performance of the fish during the 8-week trial. No significant differences in body length increase were detected between groups; nevertheless, body weight, K, SGR and RGR increased with increasing temperatures, with the highest peak registered in the 12°C group. A similar bell-shaped growth response to temperature has been reported in previous studies on Atlantic salmon post-smolts, with the optimal growth temperature varying depending on fish body size (Handeland, 2000; Handeland et al., 2003; 2008; Kullgren et al., 2013; Fraser et al., 2025). For instance, a reduction in growth and feed intake was observed at 14°C for 150–300 g post-smolts and above 12.8°C for 70–150 g post smolt under natural light conditions over a 12-week trial (Handeland et al., 2008; Kullgren et al., 2013). Conversely, a temperature threshold adversely affecting growth was identified above 10.5°C in 1.2 kg post-smolts under continuous light, although feed intake data were not collected in that study (Fraser et al., 2025).

A negative correlation was observed between HSI and temperature. The HSI is commonly used as an indicator of energy status in fish (Sheridan, 1994; Chellappa et al., 1995), and its decline with increasing temperature may reflect adjustments in energy mobilisation to support the increased metabolic demands associated with elevated temperatures. In addition, the GSI analysis revealed the presence of maturing males throughout the trial, with a higher incidence in the 15°C group. Maturation is a biological process that alters fish metabolism and redirects energy allocation toward the reproductive system, and temperature has been identified as one of the triggering keys in Atlantic salmon (Fjelldal et al., 2011, Crouse et al., 2022; Pino-Martinez et al., 2023; Fraser et al., 2025).

The fact that the fish in this study were fed in excess suggests that the observed reduction of feed consumption was voluntary. The direct correlation between increased feed intake and elevated temperature can be attributed to the greater energy demands required to support the higher metabolic rate associated with elevated temperatures (Volkoff and Rønnestad, 2020; Jutfelt et al., 2021). Conversely, the declined feed intake and growth observed above the optimal temperature in ectotherms is attributed to a reallocation of energy priorities, a phenomenon known as aerobic protection. In this process, ectothermic organisms reduce their postprandial residual aerobic scope by decreasing meal sizes, thereby modulating the peak specific dynamic action (SDA) response during periods of elevated temperatures (Jutfelt et al., 2021). This adaptive response ultimately results in reduced growth. This phenomenon is supported by several studies, including the tropical clownfish Amphiprion ocellaris (Pham et al., 2021; 2022).

It is noteworthy that the findings of this study are based on feeding regimen involving a single daily meal provided to satiety. Consequently, it cannot be excluded the possibility that a different response to the increased energy demand may have occurred if multiple meals have been offered throughout the day. However, in the study conducted by Handeland et al. (2008), a reduction in feed intake and growth in fish between 170–300 g was observed at temperatures exceeding 14°C, despite feeding twice daily for 2 hours per session. Similarly, in the study by Fraser et al. (2025), fish were fed continuously throughout the day, yet a decline in growth was recorded at temperatures exceeding 10.5°C.

Temperature significantly affected the feed transit rate in the GIT, and the emptying and refilling dynamics of the gall bladder during the 24 h post feeding. The identified positive link between feed intake and temperature was evidenced by higher stomach content and evacuation rate at higher temperatures, as also shown by previous studies on Atlantic salmon (Sweka et al., 2004; Handeland et al., 2008). In addition, although the fish at 8°C had the lowest stomach content, they also exhibited the slowest gastric evacuation rate, likely attributable to reduced rates of feed processing and stomach motility at lower temperatures. This is consistent with previous literature, emphasising that water temperature exerts a greater influence than stomach content on the gastric evacuation rate of ectothermic animals (Kawaguchi et al., 2007; Handeland et al., 2008; Jobling, 2011). At 24 hpf all three groups had no residual stomach content, consistent with the findings of Valen et al. (2011) who reported full gastric evacuation within the same time frame in 45 g Atlantic salmon at 8°C . In contrast, Handeland et al. (2008) observed stomach evacuation time of 48–72 h in fish weighing between 70 and 300 g at water temperatures ranging from 6°C to 18°C The disparities in gastric dynamics among the experimental groups were also evident in the delayed evacuation rates in both the midgut and hindgut at 8°C, in contrast to the groups at 12°C and 15°C. The lowest temperature group demonstrated consistently higher midgut fullness, reflecting the slower stomach emptying and the slower digestion, absorption and motility dynamics in the midgut and hindgut. Consistent with this tendency, the analysis of the GBI revealed both a higher bile content and faster refilling of the gallbladder at higher temperatures, with 15°C exhibiting the most pronounced and rapid response through the 24 hpf.

Elevated temperatures were associated with an increased and accelerated post-prandial metabolic response, as evidenced by higher plasma levels of glucose, lactate, cholesterol, and triglycerides, likely reflecting enhanced feed processing and digestion activity. A delayed homeostatic pattern of glucose relative to triglycerides was observed, underscoring the preference of using lipids over carbohydrates as energy source in salmonids (Polakof et al., 2008; 2011; 2012; Librán-Pérez et al., 2012). Lactate showed a similar, albeit delayed, response pattern to glucose, indicating a possible metabolic connection. The rise of glucose levels in carnivores fish such as rainbow trout (Oncorhynchus mykiss) is hypothesised to be mediated and buffered by the immediate conversion of glucose into lactate followed by stimulation of pyruvate metabolism through the Krebs cycle (Polakof and Soengas, 2008; Polakof et al., 2010; Otero-Rodiño et al., 2015). However, the available published data on this mechanism remain limited. Overall, a temperature-dependent effect was distinctly contingent upon the levels of all metabolites and the development of time. At 8°C, the changes in plasma levels were slower and more delayed compared to those observed at 12°C and 15°C.

Temperature effects were also observed for Ghrl levels in plasma. The postprandial response was flat at low temperature; while fish at 12°C and 15°C had a notable reduction in Ghrl plasma levels from two hpf, with the most pronounced decrease in Ghrl levels observed at 24 hpf in the 15°C group. These results do not align with the usual elevation in plasma Ghrl levels prior to the initiation of next meal, as frequently reported in mammals and some teleost species (Higgins et al., 2007; Abdalla, 2015; Volkoff, 2016). Conversely, our data correlate with studies on the species-specific action of ghrelin in salmonids, where ghrelin has been suggested to function as an anorexigenic peptide and inhibit feed intake (Jönsson, 2013). Indeed, our results demonstrated a postprandial downregulation of Ghrl plasma levels in the fish groups at 12°C and 15°C, which displayed higher feed intake and gastrointestinal content relative to the fish maintained at 8°C. However, the Ghrl response to temperature and feed intake on Atlantic salmon are inconsistent. Similarly, Ghrl levels were found to be lower in Atlantic salmon at 19°C that showed lower feed intake, growth, and feed utilization compared a similar group at 14°C (Hevrøy et al., 2012). No changes in plasma Ghrl levels were observed in Atlantic salmon kept at either 8°C, 12°C or 18°C, despite a negative correlation between temperature and feed intake, FCR and growth (Kullgren et al., 2013). Vikeså et al. (2015) reported no changes in Ghrl plasma levels in fish with comparable feed intake and FCR at 13°C and 19°C. These discrepancies in Ghrl responses might be attributed to differences in feeding regimes, experimental condition, fish developmental stage and sampling time. In addition, ghrelin’s role in hunger regulation in Atlantic salmon may not strictly coincide with meal onset, but its action may be more closely associated with the fish energy balance, highlighting a complex interaction between hormonal regulation and energy homeostasis. Indeed, in a recent study, Ghrl action was linked to stimulation of lipid accumulation as energy storage, where a downregulation of plasma Ghrl levels was observed in Atlantic salmon exposed to an intermittent fasting regime for over 4 weeks (Lai et al., 2025).

A general concave-shaped response was observed in the mRNA levels of ghrl1, ghrl2, and mboat, the enzyme responsible for ghrelin activation. Transcript abundance decreased immediately after feeding and gradually returned towards initial levels by 24 hpf. An interaction between temperature and postprandial time was observed for ghrl1, with the lowest levels registered in the group exposed to 15°C. A temperature effect on ghrelin mRNA levels was observed in a previous study on Atlantic salmon, where lower levels of ghrl1 were registered in fish kept at 19°C compared to a similar group at 14°C (Hevrøy et al., 2012). However, ghrelin mRNA levels were measured only four hpf, making direct comparison with our postprandial dynamics difficult. Most studies on ghrelin in Atlantic salmon have primarily focused on its response over days and weeks after feeding. For instance, ghrl1 mRNA levels increased (Murashita et al., 2009), decreased (Hevrøy et al., 2011; Kalananthan et al., 2023) or showed no response (Del Vecchio et al., 2021) following fasting periods. Additional studies showed that postprandial expression of ghrelin in the stomach can vary significantly depending on the teleost species. A decrease in mRNA ghrelin was observed two hpf in Atlantic cod (Gadus morhua) (Xu and Volkoff, 2009), and 3h after feeding in goldfish (Carassius auratus) (Unniappan et al., 2004) and gibel carp (C. auratus gibelio) (Zhou et al., 2016). While, up and down postprandial dynamics were observed over 72hpf in the spotted sea bass (Lateolabrax maculatus) (Yu et al., 2020). In contrast, no postprandial changes were observed over 24 hpf for tilapia (Oreochromis mossambicus) (Fox et al., 2009).

Notably, our data showed a convex postprandial pattern of ghrelin mRNA expression levels, in contrast to the concave temporal dynamics observed for Ghrl in plasma. For instance, when plasma Ghrl is at its peak, ghrelin mRNA levels are at their lowest point, suggesting post-translational or post-transcriptional regulations, or regulatory mechanism wherein the expression of ghrelin is inhibited at elevated plasma ghrelin levels. Conversely, a decline in plasma ghrelin levels likely stimulates a rise in ghrelin mRNA expression to restore hormone levels. Similarly, Hevrøy and colleagues (2011) demonstrated an opposite response of ghrelin mRNA and Ghrl plasma levels post-feeding in Atlantic salmon. Contrary to these findings, a postprandial decrease in serum Ghrl was concomitant with a decrease in intestinal ghrelin mRNA levels in goldfish (Unniappan et al., 2004). While no changes were observed in both ghrelin mRNA and protein levels in tilapia (Fox et al., 2009). Taken together, the available data on ghrelin show that our understanding of ghrelin physiology in fish remains incomplete, highlighting the need for further functional studies. The many discrepancies in ghrelin responses (both mRNA and protein levels) to nutritional status, time, and environmental conditions highlight the complex regulatory mechanisms governing appetite and energy balance in teleosts. In particular, strong attention needs to be paid when comparing different studies, as variations in experimental conditions and fish status can lead to differing results. Further functional studies are required to elucidate the temporal and spatial dynamics of ghrelin expression and its interaction with other hormonal and metabolic pathways.

While postprandial changes were evident in expression of all hypothalamic genes analysed, temperature effects was exclusively observed in the mRNA levels of npya1, npya2, pomca1 and pomca2. Temperature had a negative effect on the expression dynamics of npya and pomca paralogues, with higher levels in fish at 8°C. Both npya1 and npya2 showed a significant increase post-feeding in all experimental groups, with peak levels registered at 24hpf, consistent with their orexigenic role related to meal anticipation in mammals and teleosts (Volkoff et al., 2005; Rønnestad et al., 2017). An exception was observed in fish at 8°C, where npya1 showed a marginal decline at 24 hpf, but levels remained higher than at 0 hpf, still indicating a meal anticipation response but at diminished levels relative to the 12°C and 15°C groups and aligning with lower feed intake in this group. In contrast, pomca1 and pomca2 exhibited an increase in mRNA levels up to 12hpf, followed by a general drop down to 24 hpf in all groups, aligning with its anorexigenic role in the regulation of appetite (Volkoff et al., 2005; Rønnestad et al., 2017). Temperature was also found to affect the absolute abundance of the npya and pomca paralogues, where their expression higher at the lowest temperature at 8°C. These results also correspond with the absolute expression and postprandial dynamics of ghrl1 in the stomach, where the lowest levels were registered in the group exposed to 15°C. The higher gene expression response observed at 8°C may indicate a metabolic adaptation in fish, where a higher mRNA response possibly is needed to enable reduced food intake in a voracious fish as salmonids, and sustain a lower metabolic rate and energy expenditure for food processing at low temperatures. No temperature effects were observed for agrp1 and cart2b, but a post-feeding increase was noted in all experimental groups. However, the mRNA temporal changes for agrp1 and cart2b were less pronounced compared to those for the npya and pomca paralogues, underscoring the major role of npya and pomca in regulating post-feeding satiety in this study.

The incidence of maturing fish was positively correlated with warmer temperatures. Early sexual development has become an increasing concern in Norwegian salmon farming, with elevated temperatures identified as one of the key triggering factors (Crouse et al., 2022; Pino Martinez et al., 2023; Pino-Martinez et al., 2024; Fraser et al., 2025). Although sexual maturation was not the main objective of this study, the data were utilized to explore possible correlations with hunger and appetite controlling mechanisms. The findings are consistent with existing literature, indicating that maturing fish tend to be smaller and shorter than non-maturing counterparts, but exhibit a higher condition factor (Pino Martinez et al., 2023; Fraser et al., 2025). Maturing fish had lower content in the stomach, midgut and hindgut, indicating that they consumed less feed compared to non-maturing fish maintained at the same water temperature. The results are consistent with the physiological changes that occur during salmon maturation, which are characterized by reduced feed intake. This reduction is thought to be linked with allocation of endogenous factors such as fat reserves, gonad development, and overall condition factor (Kadri et al., 1995; Taranger et al., 2010; Hvas et al., 2024). Data on blood chemistry highlighted the metabolic adjustments typical of reproductive development, with lower levels of glucose and higher levels of triglycerides and cholesterol in maturing fish. Notably, lipids are the primary energy source in teleost, and triglycerides plays a key role in regulating the initiation of reproductive maturation (Jenkins et al., 2019; Fraser et al., 2025).

The data from this study showed no significant differences in the expression of ghrl and mboat genes between maturing and non-maturing fish. However, a downregulation of hypothalamic genes in the maturing fish was observed, which highlights adjustments in the hypothalamic appetite key factors during the maturing phase. Further studies with larger sample sizes are needed to validate these findings and provide a more comprehensive understanding of how the onset of maturation—a key life-history decision in the iteroparous salmon—is modulated through a complex interplay between environmental factors and physiological regulatory mechanisms.