Prior to diet preparation, all raw materials were crushed using a grinder and subsequently sieved using a 60-mesh sieve. After sieving, the raw materials were weighed based on the diet formula outlined in Table 1 . Briefly, the weighed raw materials were thoroughly mixed in a plastic zip-lock bag and then transferred to a commercial mixer for mixing and blending, during which fish oil, soya lecithin, and water were added and mixed. After mixing, the ingredients were extruded into long strips using a twin-screw extruder and subsequently transferred to a pelletizer to produce pellets for the trials. Next, diets were steam-conditioned at 60°C, then removed and air-dried until the moisture content of the diet was reduced to approximately 10%. Following packing into plastic ziplock bags, the diets were stored in a refrigerator at -20°C. Different feeding regimes, specifically the 20% fishmeal group, the 10% fishmeal group, and the 10% fishmeal + 0.03% H. pluvialis group, were administered to L. vannamei (NT_C, NT_LFM, NT_LFM_HP, LT_C, LT_LFM, and LT_LFM_HP) at 30°C and 20° water temperature environments.

In Lingshui Li Autonomous County, L. vannamei was used as the test animal for an 8-week culture experiment. Shrimp larvae were temporarily raised on commercial feed for 10 weeks before being used in the culture experiment. A total of 720 shrimp with an average weight of approximately 0.63 g were randomly selected from the temporarily reared shrimp and assigned to 24 cement tanks, each containing 30 shrimp. Sewage pipes were installed at the bottom of each cement tank for effluent discharge, and water pipes were installed in the cement tanks for seawater replenishment. The water was changed uniformly and regularly according to predetermined water quality indices, with 80% of the water replaced at each interval. All seawater used in the aquaculture tanks was sand-filtered, precipitated, sterilized, and filtered prior to use. The cement tanks were fitted with an air tube and air stone to maintain 24-hour continuous aeration.

Feeding rates were initially set at 5% of shrimp body weight and adjusted based on satiation by observing bait residues after feeding. All groups were fed three times daily, and dead shrimp, bait residues, feces, and other bottom waste were aspirated by suction, and dead shrimp were weighed and counted. During the 8-week aquaculture experiment, the temperatures of the water were maintained at 30°C for the normal-temperature groups and 20°C for the low-temperature groups. A chiller and a water recycling system were used to adjust water temperature, which was recorded every 4 hours. Throughout the experiment, natural light cycles were maintained.

During the 8-week culture period, 5 g of feces from each cement tank were collected and stored in a refrigerator at -20°C for apparent digestibility analysis. All shrimp were fasted for 24 hours before sampling. Shrimp in each tank was counted and weighed. The body length, weight, and hepatopancreatic weight of five shrimp from each group were measured and recorded. A total of five shrimp were randomly selected from each group and stored at -20°C for whole shrimp crude nutrient analysis. Frozen hepatopancreas samples harvested from four shrimp were analyzed for enzyme activity and gene expression. Intestinal samples were collected from four shrimp and frozen for intestinal flora analysis. For H&E staining and sectioning, hepatopancreas samples collected from two shrimp were used. For comparison, five live shrimp from each group were boiled under identical conditions.

According to the guidelines of the Association of Official Analytical Chemists (AOAC), moisture, crude protein, and crude lipid content were determined from dried samples. After drying at 105°C, moisture content was determined, following which the samples were crushed. Next, a fully automated Dumas nitrogen tester (N Pro (DT Ar/He Basic), Gerhardt GMBH & CO.KG, Germany) was employed to determine crude protein content from 0.08 g of sample. The crude lipid content was determined using the Soxhlet extraction method on an automatic lipid analyzer (Soxtec System HT6, Tecator, Sweden) with light petroleum reflux extraction on approximately 0.5 g of the sample. Shrimp shell samples were freeze-dried and ground to extract astaxanthin and measure astaxanthin levels. Ionophore atomic emission spectrometry (ICP-AES) was utilized to quantify metal Y-ions in feces, and apparent digestibility was calculated. The methods for extracting astaxanthin in shrimp shells were performed as described in a previous study (50).

After thawing hepatopancreatic tissues, samples were collected and added to PBS solution for grinding. The levels of hepatopancreatic enzyme activity indices were determined by centrifuging at 4000 rpm for 20 min at 4°C and collecting the supernatant. Superoxide dismutase (SOD), total antioxidant capacity (T-AOC), and lipid oxidation (MDA) levels were measured using kits (Nanjing Jiancheng Bioengineering Institute, Nanjing, China).

Real-time quantitative PCR (qRT-PCR) was performed for all gene expression assays in the present study. With the Evo MMLV Reverse transcription reagent kit (Accurate Biology, Hunan, China), total hepatopancreatic RNA was extracted using the TRIzol method, and cDNA was synthesized via reverse transcription. Moreover, qRT-PCR was performed on a Roche real-time quantitative PCR system (LightCycler 480 II, Roche Diagnostics, Basel, Switzerland). Conditions for the reaction were as follows: pre-denaturation at 95°C for 40 amplification cycles (denaturation at 95°C for 5 s, annealing at 60°C for 30 s, and extension at 72°C for 30 s). A melting curve was plotted (95°C for 20 s, 60°C for 20 s, followed by continuous maintenance at 95°C), and the samples were cooled to 4°C. Shrimp primers used in this study are presented in Table 2 . The β-actin gene served as an internal reference gene for gene expression analysis, and relative expression levels were calculated using the 2^-ΔΔct method.

Hepatopancreatic samples were initially fixed in a 4% paraformaldehyde solution for 24 hours and subsequently transferred to a 70% ethanol solution. Prior to sectioning, the samples were subjected to graded dehydration, paraffin embedding, and fixation, following which the sections were sliced into approximately 3.0 μm thick sections using a slicer. Afterward, the sections were subjected to hematoxylin and eosin staining and examined and imaged under a Nikon orthostatic microscope (Eclipse Ni-E, Nikon, Japan). Finally, the resulting images were analyzed utilizing NIS-Elements viewer software (National Institutes of Health, Bethesda, USA).

The genomic DNA of intestinal microorganisms in the sample was extracted and analyzed for purity and concentration using 1% agarose gel electrophoresis. The purified genomic DNA was subsequently dispatched to Shanghai Majorbio Bio‐pharm Technology Co., Ltd (51) for further processing.

Paired-end (PE) reads generated from Illumina sequencing were initially aligned based on their overlapping regions. Subsequently, sequence quality was assessed and filtered, followed by sample differentiation for operational taxonomic unit (OTU) clustering and species taxonomy analysis to facilitate the calculation of various diversity indices. OTU-based diversity index analyses and sequencing depth detection were conducted, while taxonomic information enabled statistical analyses of community structure at different taxonomic levels. Data analyses were conducted utilizing the Meggie BioCloud platform (https://cloud.majorbio.com). Specifically, mothur software was employed to compute alpha diversity metrics such as Chao 1 and Shannon index, while the Wilcoxon rank sum test was utilized to assess variations in alpha diversity among groups (52). Intergroup differences in alpha diversity were evaluated using an algorithm based on the Bray-Curtis distance, coupled with PCoA (Principal Coordinate Analysis) analysis to examine the similarities in microbial community structures across samples. Additionally, the PERMANOVA non-parametric test was utilized in conjunction with LEfSe (Linear Discriminant Analysis Effect Size) analysis (LDA>2, P<0.05) to assess significant variations in microbial community structure among groups, identifying bacterial taxa with differing abundances from phylum to genus level (53).

Parameters were calculated using the following formulae: Initial body weight (IBW, g)=initial total wet weight/initial number of tails; Final body weight (FBW, g)=final total wet weight/final number of tails; Weight gain (WG, %)=100×(final body weight-initial body weight)/initial body weight; Specific growth rate (SGR, %/day)=100×(Ln final mean weight-Ln initial mean weight)/number of days; Food intake (FI, g/shrimp)=total food intake/total number of shrimp; Feed conversion ratio (FCR)=dry diet fed/wet weight gain; Conversion factor (CF, g/cm³)=100×wet weight/(body length)³; Hepatosomatic intake (HSI, %)=100×hepatopancreas weight/wet weight; Survival rate (SR, %)=100×number of terminal surviving tails/number of initial tails; Apparent Digestibility (AD, %)=100× (Y₂O₃ intake-Y₂O₃ output)/Y₂O₃ intake.

Experimental data were expressed as “mean ± standard error”. Two-way ANOVA with Bonferroni multiple comparisons were used to analyze data across treatment groups at the same temperature. Student’s t-test was conducted to analyze data in the same treatment groups at different temperatures. P < 0.05 was considered statistically significant.

Following the conclusion of the 8-week period, the growth performance, feed utilization, and morphometric parameters for the six groups of L. vannamei were individually recorded, as detailed in Table 3 . The weight gain rate and specific growth rate of L. vannamei in the NT_C, NT_LFM, and NT_LFM_HP groups were significantly higher compared to the LT_C LT_LFM and LT_LFM_HP groups (P<0.05). Likewise, the weight gain rate of L. vannamei in group NT_LFM_HP was significantly higher compared to group NT_LFM (P<0.05). In contrast, the weight gain rate of L. vannamei in group NT_LFM_HP was comparable to that in group NT_C. Interestingly, no significant differences were noted in the survival rates of L. vannamei across treatment groups (P>0.05). Conversely, variations in feed efficiency and apparent digestibility were observed between the normal and low-temperature groups (P<0.05).

The moisture, crude protein, and crude lipid content of L. vannamei muscle within each experimental group are outlined in Table 4 . The findings revealed no significant differences in these parameters across the groups (P>0.05).

As displayed in Figure 1 , significant differences were observed in coloration between live and cooked shrimp, with the NT_LFM_HP and LT_LFM_HP groups exhibiting darker pigmentation compared to the NT_LFM and LT_LFM groups. Furthermore, as depicted in Figure 2 , the astaxanthin content in shrimp shells was highest in the NT_LFM_HP and LT_LFM_HP groups.

Figure 3 illustrates the comparison of hepatopancreatic tissue sections across groups. The hepatic tubules within the hepatopancreas of groups NT_C, NT_LFM, NT_LFM_HP, LT_C, LT_LFM, and LT_LFM_HP displayed normal morphology, with normal B cells, R cells, and E cells and the absence of lesions.

The Toll and IMD signaling pathways were identified as pivotal elements of the innate immune response in shrimp, as evidenced by the expression of relevant genes depicted in Figures 6 , 7 .

In the Toll pathway, the relative expression level of toll mRNA was significantly higher in the NT_C and LT_C groups compared to the NT_LFM, NT_LFM_HP, LT_LFM, and LT_LFM_HP groups (P<0.05). The highest relative expression of toll was observed in the NT_C group. At the same time, dorsal mRNA expression levels were significantly higher in the LT_LFM_HP group compared to the other groups (P<0.05), while crustin mRNA expression levels were highest in the LT_C group, and the differences between the groups under different temperature conditions were significant (P<0.05).

In the IMD pathway, the relative expression level of imd mRNA was significantly higher in the NT_LFM_HP and LT_LFM_HP groups compared to the NT_C, NT_LFM, LT_C, and LT_LFM groups (P<0.05). Additionally, the relative expression level of relish was higher in the LT_LFM_HP group compared to the NT_LFM_HP group. The mRNA expression levels of lysc in the LT_LFM_HP group were significantly higher compared to the other five groups (P<0.05), with the lowest expression observed in the NT_LFM group. Specifically, the relative expression of L. vannamei lysc mRNA in the LT_LFM_HP group was markedly higher compared to the remaining groups (P<0.05), with the lowest expression detected in the NT_LFM group.

Sequences were grouped into operational taxonomic units (OTUs) at a 97% similarity threshold. The constructed Venn diagram delineated that the NT_C, NT_LFM, NT_LFM_HP, LT_C, LT_LFM, and LT_LFM_HP groups contained 8, 52, 81, 8, 5, and 10 unique Operational Taxonomic Units (OTUs), respectively. As shown in Figure 8 , significant differences were noted in the microbial community structure due to differences in water temperature and the presence of H. pluvialis. Figure 9 illustrates the analysis of alpha diversity within the shrimp gut microbiome, as determined by sequencing results of the Ace, Chao1, Shannon, and Simpson diversity indices. The findings signaled that the coverage index for each treatment group exceeded 0.998. The beta diversity analysis visualized through PCoA plots ( Figure 10 ) demonstrated the impact of varying water temperatures (normal and low) on variations in the intestinal microbiota of L. vannamei following exposure to H. pluvialis. As anticipated, the composition of the NT_LFM_HP group was significantly different compared to the other groups (P<0.05).

Figure 11 illustrates the distribution of gut bacteria across treatment groups, categorized at the phylum, class, family, and genus levels. Additionally, the relative abundance of species was depicted at each taxonomic level. Figure 11A displays the relative abundance of gut bacterial phyla, primarily composed of Bacteroidota, Proteobacteria, Actinobacteriota, and Verrucomicrobia. Figure 11B displays the relative abundance of gut bacterial classes in L. vannamei, with major bacteria identified as Bacteroidia, Alphaproteobacteria, Gammaproteobacteria, Acidimicrobiia, Verrucomicrobia, and Actinobacteria. Figure 11C displays the relative abundance of gut bacterial families, predominantly composed of Flavobacteraceae, Rhodobacteraceae, Vibrionaceae, and Actinomarinales. Figure 11D displays the relative abundance of gut bacterial genera, including Flavobacteriales, Rhodobacteraceae, Ruegeria, Spongiimonas, Vibrio, Actinomarinales, and Haloferula.