Seven isonitrogenous (441–456 g kg^-1 crude protein) and isocaloric (21.5–22.0 MJ kg^-1 gross energy) diets were formulated to compare the nutritional value and combination effects of wheat gluten, LT fishmeal, and SPC. The chemical compositions of wheat gluten, FM, and SPC are presented in Table 1. A mixture (GWT) consisting of 77.5% wheat gluten, 20.5% wheat, and 2.0% taurine was produced. The V0, VF1, VF2, and VF3 diets were formulated with GWT gradually replacing 0, 33.3, 66.7, and 50% FM, respectively. Similarly, GWT gradually replaced 33.3, 66.7, and 100% SPC to form the VS1, VS2, and VS3 diets, respectively (Table 2). Based on previous studies (Mai et al., 2006; Li et al., 2012; Hu et al., 2013), the V0 diet could meet the nutrient requirements of Japanese seabass. To maintain the same level of dietary essential amino acids (EAA), the VS1, VS2, and VS3 diets were supplemented with crystalline Lys, Arg, and Thr, and the VF1, VF2, and VF3 diets were supplemented with crystalline Lys, Arg, Thr, and Met. In addition, monocalcium phosphate was also added to the diets, and yttrium oxide (Y₂O₃) was used as an inert marker for digestibility assessment (Austreng et al., 2000; Zhang et al., 2006). The formulation, proximate compositions, and amino acid profiles of the diets are presented in Table 3, 4.

Diets were produced at the Feed Technology Laboratory of the Feed Research Institute, Chinese Academy of Agricultural Sciences in Beijing. All dry ingredients were ground in a hammer mill through a 0.18-mm screen, then mixed, preconditioned, and extruded in a twin-screw extruder (MY56X2A, Muyang, Jiangsu, China) with a 2.0 mm die plate (MY56A 12-03/02 XL 09 11). The goal of the extrusion process was to achieve a bulk density higher than 516 g L⁻¹ prior to drying to facilitate the slow sinking of feed after drying and lipid coating. All extruded pellets were dried to 950 g kg^-1 dry matter at ambient temperature. Fish oil and soy lecithin were coated into pellets using a vacuum coater (ZJB-100). Pellet length and diameter were determined using a digital caliper. Durability and breaking point were estimated for uncoated pellets using a pellet tester (ST-136, Shengtai Instrument Co., Ltd., Jinan, China) and hardness tester (ST-120B, Shengtai Instrument Co., Ltd., Jinan, China), respectively. Water stability and sinking rate were determined according to the methods described by Baeverfjord et al. (2006) and Sørensen et al. (2012), respectively.

A 72-day feeding trial followed by a 10-day digestibility evaluation experiment were conducted in an indoor recirculation system at the Haid Group research station, Seagull Island, Panyu, Guangzhou, China. Freshwater-acclimated Japanese seabass (L. japonicus) juveniles were obtained from a hatchery in Fujian, China and fed a commercial diet for 1 month to adapt to the laboratory condition. Before the experiment, 630 fish with an initial weight of 10.73 ± 0.33 g were fasted for 24 h, anaesthetized with MS-222 (90 mg L⁻¹; Hangzhou Dongbao, Hangzhou, China), batch-weighed, and then randomly assigned to 21 circular tanks, with each diet being assigned triplicates and 30 fish per tank. All fish were fed four meals per day by hand (07:30, 10:30, 13:30, and 16:30 h), with each meal lasting 60 min. Fish were fed 10% in excess based on average feed intake over the previous 3 days. All uneaten feed was immediately sieved and dried to a constant weight at 95°C and reweighed. During the feeding experiment, the water temperature ranged from 28.5°C to 30.0°C, dissolved oxygen exceeded 5.0 mg L⁻¹, the pH value was 6.5–7.0, ammonia nitrogen and nitrite were less than 0.1 mg L⁻¹, and the photoperiod was 12D:12L.

Before the experiment, 30 fish were euthanized by an overdose of MS-222 and stored at −20°C for whole-body composition analyses. At the end of the experiment, fish were anaesthetized with 90 mg L^-1 MS-222 before sampling. Five fish from each tank were individually measured for weight and body length, then dissected to remove the whole viscera; the liver and carcass were weighed separately. Another six fish were randomly sampled from each tank and then kept at −20°C for analysis of whole-body and amino acid compositions. The remaining fish in each tank were assessed for the digestibility evaluation experiment. Faeces were collected by carefully dissecting the abdomen from the last 5 cm of the distal intestine, pooled by tank, and then stored at −20°C.

Initial and final whole-body samples from the same tank were cut into pieces and ground with a meat grinder, then autoclaved (YXQ-LS, Xunbo, Shanghai, China) at 120°C for 30 min, homogenized by a homogenizer (DS-1, Shanghai, China), and oven-dried (Jinghong, Zhejiang, China) at 80°C. Whole-body samples were finely ground with a pestle and mortar until all samples passed through a 0.9 mm screen. Dried whole fish and feed samples were analyzed for dry matter (105°C to constant weight), protein (Kjeldahl’s method), lipids (Soxhlet Extraction System, Jingke, Shanghai, China), amino acids (amino acid analyzer, L-8900, Hitachi, Japan), ash (550°C, overnight), and energy (Phillipson Microbomb Calorimeter, Gentry Instruments Inc., Aiken, SC, United States) based on the previous study (Zhang et al., 2022). Faeces were freeze-dried at −50°C (Labonco Freezon 4.5, Kansas City, United States) and ground with a pestle and mortar. Yttrium oxide in feed and faeces was determined based on the previous study using inductively coupled plasma mass spectrometry (ICP-MS) (Zhang et al., 2012a). Briefly, approximately 50 mg of freeze-dried sample was weighed into a digestion tube, and then 6 ml HNO₃ and 2 ml hydrogen peroxide were added. After microwave digestion, the sample was diluted to a constant volume of 50 ml, aspirated 1 ml diluent, and then re-diluted to 10 ml for on-board testing.

Apparent digestibility (ADAA, %) of individual amino acids was calculated as 100 × (1 - AA_f × AA_d ^-1 × Y_d × Y_f ⁻¹), where AA_f and AA_d represent the concentration of individual amino acids in faeces and the diet and Y_f and Y_d represent the concentration of yttrium in faeces and the diet. Retention of nitrogen and energy (%) was calculated as 100 × (N₁ × FBW - N₀ × IBW) × (N_d × FI)⁻¹, where N₀ and N₁ represent the nutrient or energy in the initial and final whole body and N_d represents the nutrient or energy in the diet. Feed intake (FI, %/d) was quantified by subtracting uneaten feed from the amount of feed on a dry matter basis. The feed conversion ratio (FCR) was calculated as FI × (weight gain, g)⁻¹. The weight gain rate (WGR, %) was calculated as 100 × (FBW–IBW)/IBW, where FBW and IBW represent final body weight and initial body weight, respectively. The hepatosomatic and viscerosomatic indices (%) were expressed as organ weight as a percentage of body weight, and the condition factor (CF, g/cm³) was calculated as 100 × (fish weight, g) × (body length, cm)⁻³.

All data are presented as means ± pooled S.E.M. (n = 3) and assessed by one-way ANOVA (IBM, SPSS Statistics 20.0), with the level of significance set at p < 0.05. Moreover, a follow-up trend analysis using orthogonal polynomial contrasts was performed to determine whether the significant effects were linear and/or quadratic. Quadratic regressions were only presented when the regression coefficient of the second degree component was statistically significant (p < 0.05). Maxima or minima in quadratic regressions were calculated by setting the first derivative of the equation to 0. Significant differences in the ANOVA were ranked by the Pdiff routine under LSMEANS and indicated by different superscript letters ^{a, b, c}.

The processing parameter and pellet physical quality are presented in Table 5. The V0 diet differed from the other diets, and the total amount of process water was limited to 20% to achieve a uniform flow through the extruder. The combined effect of dietary composition and restricted water addition resulted in a lower bulk density, breaking point, and slower sinking rate compared with the other diets. The diametric expansion of all diets ranged from 74% to 88% and increased with increasing dietary wheat gluten. All diets had similar durability, with VS1 being the most water-stable and VF1 the least.

No significant differences were found in FI (Table 6). WGR linearly and significantly decreased from 795% to 627% with increasing dietary GWT. Nitrogen retention significantly decreased from 34.8% to 26.9%, while energy retention decreased from 40.2% to 32.5%. FCR linearly increased from 1.07 to 1.36 as GWT gradually replaced FM. No significant differences were found in WGR or FCR with GWT gradually replacing SPC. The retention of nitrogen and energy linearly decreased with the increase in GWT supplementation.

Moisture, protein, lipid, ash, gross energy in the whole body, and the hepatosomatic index and condition factor were not significantly affected by different dietary GWT supplementation (Tables 7, 8). The viscerosomatic index linearly decreased from 17.3% to 16.1% with the increase of GWT supplementation.

No significant differences were found in the ADCs of Arg, Ile, Met, Thr, Tyr, and Val (Table 9). As GWT gradually replaced FM, the ADC of Cys significantly increased from 77.6% to 86.5%. The ADCs of His, Lys, and total AAs followed a similar pattern and increased with the increase in GWT supplementation. The ADCs of Leu, Phe, and total EAAs increased in a curvilinear mode with increasing proportions of GWT. Except for Cys and Met, the ADCs of essential and total amino acids were not significantly affected by the gradual substitution of SPC by GWT. The ADC of Cys significantly increased from 77.6% to 90.0%, while the ADC of Met significantly increased from 66.8% to 76.7%. The ADCs of Cys and Met were lower in the V0 diet than in the other diets.

The retention of individual and total EAAs significantly decreased in a linear manner (Table 10). The retention of Met was twice or more than the values observed in the other EAAs. The replacement of SPC with GWT had no significant effect on the retention of Arg, His, Ile, Leu, Lys, Phe, Thr, Val, and total EAAs. The retention of Met followed a U-shaped pattern, with the highest value being found in the V0 diet.