Species-Specific Lipid Responses to Frozen Feed Storage Duration in Farmed Seahorses: A Preliminary Lipidomic Study

Abstract

Frozen mysid shrimp is the primary feed for farmed medicinal seahorses, yet no standardized storage duration exists, and whether storage-related feed changes affect the lipid-based medicinal quality of seahorses remains unknown. This study used targeted lipidomics to compare 98 lipid species across six lipid classes in three farmed seahorse species—Hippocampus erectus, H. abdominalis, and H. trimaculatus—fed mysid shrimp frozen for either 2 or 10 months. Prolonged storage did not cause a uniform decline in lipid quality; instead, each species exhibited a distinct response pattern. H. erectus showed compensatory increases in fatty acids, whereas phospholipids predominantly decreased; H. abdominalis exhibited broad attenuation across multiple lipid classes; and H. trimaculatus displayed an intermediate, class-dependent pattern. Notably, oxidized sterol derivatives declined consistently across all three species despite divergent fatty acid and phospholipid responses, suggesting that oxidized sterol derivatives may serve as candidate cross-species indicators of storage-related feed changes, although validation is required. Acylcarnitine profiles remained stable in H. erectus but decreased in the other two species, suggesting possible differences in lipid utilization among species. These preliminary results suggest that frozen-storage guidelines may need to be evaluated on a species-specific basis in future studies, with sterol monitoring as a candidate indicator for further evaluation.

1. Introduction

Seahorses (genus Hippocampus), often hailed as the “ginseng of the sea,” are among the most valued animal-derived traditional Chinese medicines, with a documented clinical history dating back to the Shennong Bencao Jing (Divine Farmer’s Materia Medica). In traditional Chinese medicine theory, their core function is to “warm and tonify kidney yang,” and modern pharmacological studies have corroborated these claims by demonstrating anti-fatigue, anti-inflammatory, antioxidant, anti-tumor, and antibacterial activities [1,2]. Seahorses contain diverse bioactive components—including lipids, amino acids, bioactive peptides, glycoproteins, polysaccharides, and trace elements (zinc, selenium, etc.)—that exert multiple physiological regulatory effects [3,4,5,6,7]. Among these, the lipid fraction—particularly steroid compounds (e.g., cholesterol derivatives serving as sex hormone precursors), glycerophospholipids (e.g., phosphatidylcholine with membrane-stabilizing and anti-inflammatory properties), and n-3/n-6 polyunsaturated fatty acids (PUFA) such as eicosapentaenoic acid (EPA), docosahexaenoic acid (DHA), and arachidonic acid (ARA)—is recognized as the core material basis for the “kidney-tonifying and yang-strengthening” effects [3,4,5,6]. Consequently, the content and compositional ratio of these lipid species may contribute to quality grading and are relevant to the pharmacological properties of seahorse as a medicinal material, making them candidate chemical biomarkers for quality evaluation [7].

Since the 1990s, compounded pressures from habitat degradation, marine pollution, and chronic overfishing have caused a precipitous decline in global wild seahorse populations [8]. In 2004, all Hippocampus species were listed under CITES Appendix II, subjecting them to international trade quotas. In stark contrast to dwindling wild resources, the demand from the traditional Chinese medicine market has continued to grow rapidly, with domestic consumption in China increasing rapidly in recent years. This intensifying supply–demand imbalance has made large-scale artificial breeding not merely an economic choice but an inevitable pathway to secure medicinal material supply while alleviating pressure on wild stocks. Since the breakthrough in commercial-scale cultivation around 2015, China’s seahorse aquaculture industry has progressively shaped a diversified farming landscape dominated by the lined seahorse (H. erectus) and the big-belly seahorse (H. abdominalis), with the three-spot seahorse (H. trimaculatus) and the spotted seahorse (H. kuda) as secondary species [9]. Today, farmed seahorses have become the predominant source for the domestic medicinal market, placing quality control during cultivation at the forefront of efforts to improve product consistency.

Among the numerous technical bottlenecks constraining large-scale standardized seahorse aquaculture, feed quality control is particularly critical. Feed quality is also a critical determinant of growth, nutrient utilization, and flesh composition in cultured fish, with differences among commercial feeds producing measurable variation in both feed characteristics and animal performance [10]. Because seahorses are natural live-prey feeders and a mature artificial formulated feed system has yet to be developed, frozen mysid shrimp remains the primary long-term feed for pre-commercial and commercial-grade seahorses (body height > 8 cm) [9,11]. However, frozen storage is not a neutral preservation method—it induces complex, time-dependent changes in nutritional composition and lipid oxidation status, rather than following a simplistic “longer storage equals poorer quality” linear relationship [12]. Critically, no unified production protocol or frozen-storage duration standard currently exists in China for mysid feed. Key technical parameters—such as maximum safe storage time and appropriate temperature/humidity conditions—remain largely undefined, resulting in inconsistent feed quality across farming batches and constituting a hidden bottleneck that undermines both production stability and product quality consistency [13].

Existing research on frozen aquatic feeds has focused primarily on the deterioration mechanisms of the feed itself or on overt phenotypic responses of farmed animals, such as growth rate and survival following ingestion [14,15,16]. A deeper and more ecologically relevant question, however, has received surprisingly little attention: whether the frozen storage duration of feed can be transmitted through the food chain to affect the synthesis and accumulation of bioactive compounds—particularly lipid-based pharmacologically active substances—within the farmed animals themselves [17]. This gap is especially pronounced in the context of medicinal marine organism aquaculture. Importantly, our preliminary aquaculture observations have already provided initial evidence that feeding mysid shrimp of different frozen storage durations produces significant phenotypic differences in seahorses. Individuals fed freshly frozen mysids consistently outperformed those fed long-term frozen mysids across multiple health indicators, including feeding activity, body surface coloration, growth rate, survival, plasma immune parameters, and gut microbiota diversity [18]. These observations suggest that storage-related feed changes are biologically meaningful to the seahorses, but the molecular consequences for their medicinal lipid profiles remain entirely unexplored.

In recent years, targeted lipidomics has enabled simultaneous qualitative and quantitative analysis of hundreds of lipid molecular species across multiple major classes—including glycerophospholipids (GP), glycerolipids (GL), sphingolipids (SP), sterol lipids (ST), and fatty acyls (FA)—with high throughput and high sensitivity. Recent feeding studies have likewise demonstrated that dietary formulation can substantially modify tissue fatty acid composition and lipid nutritional indices in cultured fish, highlighting the sensitivity of fish lipid profiles to dietary inputs [19]. This technology offers a resolution far exceeding that of traditional “total lipid” or “fatty acid profile” measurements, and has been preliminarily applied to characterize lipid diversity and nutritional quality differences among seahorse species, indicating potential for medicinal quality evaluation [3]. In light of this, we propose the following hypothesis: the frozen storage duration of mysid feed alters the bioavailability of nutritional substrates and introduces varying levels of oxidized lipids, which, upon ingestion, modulate the synthesis and accumulation of lipid-based pharmacologically active components in the seahorse, ultimately producing distinct compositional profiles.

To test this hypothesis and to address the critical knowledge gap concerning how feed storage effects propagate to bioactive compounds within farmed medicinal animals, this study focuses on the three predominantly farmed seahorse species in China—H. erectus, H. abdominalis, and H. trimaculatus. Using targeted lipidomics, we systematically compared the lipid profiles of these three species after feeding on mysid shrimp subjected to short-term (2-month) versus long-term (10-month) frozen storage. Our specific aims were (1) to clarify, at the molecular level, the intrinsic relationship between mysid frozen-storage duration and the lipid-based medicinal quality of farmed seahorses; (2) to identify lipid classes or individual lipid species that consistently reflect storage-related quality changes across species; and (3) to provide preliminary insights that may inform future studies on species-specific storage-duration standards for frozen mysid feed, thereby supporting a comprehensive quality control system spanning the entire “from aquaculture to medicinal use” chain.

2. Materials and Methods

2.1. Experimental Design and Feeding Trial

Three farmed seahorse species—Hippocampus erectus (lined seahorse, 11–12 cm body height), H. abdominalis (big-belly seahorse, 12–13 cm), and H. trimaculatus (three-spot seahorse, 11–12 cm)—were obtained from commercial seahorse farms in Rizhao, Shandong Province (H. erectus and H. abdominalis) and Dongshan, Fujian Province (H. trimaculatus), respectively. Upon arrival at the Ninghai Experimental Base, East China Sea Fisheries Research Institute, Zhejiang Province, the seahorses were acclimatized for one week in plastic rearing tanks without feeding. During acclimatization, seawater salinity was maintained at 20‰ with continuous aeration. Rearing temperatures were maintained at 24–25 °C for H. erectus and H. trimaculatus using 500 W heaters, and at 16–17 °C for H. abdominalis using 100 W heaters (initial seawater temperature was 11–12 °C), following the species-specific optimal temperature ranges for commercial cultivation.

After acclimatization, 40 healthy individuals of each species with no external injuries and normal activity were selected and randomly divided into two dietary groups (n = 20 per group), each housed in a separate rearing tank. Group A (the 10-month storage group) was fed mysid shrimp frozen-stored for 10 months, whereas Group B (the 2-month storage group) was fed mysid shrimp frozen-stored for 2 months. The formal feeding trial lasted for 1 month (30 days). All six treatment groups (three species × two storage durations) started the formal feeding trial simultaneously after acclimation. Including the 1-week acclimation period, the total experimental period from arrival at the experimental base to final sampling was approximately 37 days (about 5 weeks). The preparation and storage of mysid shrimp are detailed in Section 2.2. Seahorses were fed three times daily at 07:00, 11:00, and 16:00. Feed was supplied to apparent satiation; two hours after feeding, residual feed and feces were removed by siphoning, and complete water exchange was performed every other day. Water temperature and quality parameters were maintained at the same levels as during acclimatization throughout the feeding trial. For each seahorse species, there was one tank per storage treatment. Therefore, the tank was the experimental unit, and the pooled samples described in Section 2.3 were analytical subsamples rather than independent biological replicates.

2.2. Mysid Shrimp Preparation and Storage Durations

Mysid shrimp (Neomysis sp.) were collected from brackish ponds (salinity 12‰) in Haiwan Town, Fengxian District, Shanghai, China, during nighttime hours using trawling gear. The mysids used for the two storage treatments were obtained from the same original brackish pond system but were collected on two separate occasions. The first collection was assigned to the 10-month storage group, and the second collection was assigned to the 2-month storage group. After each collection, live specimens were transported overnight via a live-aquatic-animal carrier vehicle to the Ninghai Experimental Base. Upon arrival, mysids were immediately pooled, drained of excess moisture, and rinsed several times with boiled and cooled seawater. They were then portioned into small containers at a mysid-to-water ratio of 7:3 (w/w) and stored at −20 °C until use. The fatty acid composition and lipid oxidation status of the mysid shrimp subjected to identical storage protocols were comprehensively characterized in our previous study [8]. In that study, ultra-performance liquid chromatography–tandem mass spectrometry (UPLC-MS/MS) was employed to quantify 52 free fatty acids, along with thiobarbituric acid reactive substances (TBARS), in fresh, 2-month-frozen, and 10-month-frozen mysid shrimp samples. These data are used here as supporting background information. However, because the two feed groups in the present trial were collected at different times, the previous characterization should not be interpreted as evidence that the two feeds originated from the same collection batch. Given that the formal feeding trial lasted for 1 month, the frozen storage duration of the two mysid collections increased synchronously with the experimental progression. For Group A, the mysid feed had been frozen for 9.5 months at the beginning of the feeding trial and 10.5 months at the end. For ease of description, the frozen storage duration of the mysid feed provided to Group A was averaged to 10 months; therefore, this group is uniformly designated as the 10-month storage group. Similarly, for Group B, the feed had been frozen for 1.5 months at the beginning and 2.5 months at the end of the trial, with an average of 2 months; this group is uniformly designated as the 2-month storage group. It should be noted that the 2-month and 10-month feeds were collected at different times from the same original brackish pond system. Therefore, storage duration was confounded with collection time and possible batch-to-batch variation in initial mysid composition. Consequently, the terms “2-month storage group” and “10-month storage group” are used as operational labels for the two feed treatments, and the observed differences cannot be attributed solely to frozen storage duration.

2.3. Sample Collection and Preparation

At the end of the feeding trial, nine seahorses were randomly selected from each species-by-storage treatment group and anesthetized in an ice–water mixture for 5 min. The specimens were thoroughly rinsed with ultrapure water and dried in an oven at 60 °C to constant weight. The nine dried seahorses from each group were divided into three pooled samples, with three individuals per pool, yielding three pooled samples per species-by-storage group. Each pooled sample was ground to a fine powder using an ultra-fine grinder, passed through a 60-mesh stainless-steel sieve, collected in sterile centrifuge tubes, and stored at −80 °C until lipidomics analysis. Because each species-by-storage treatment was housed in a single rearing tank, the three pooled samples per group were not independent biological replicates. The experimental unit was the tank, and the pooled samples served as analytical replicates for lipidomic measurements. The 60 °C drying procedure was applied uniformly to all samples; however, it may have influenced oxidation-sensitive lipids and should be considered when interpreting the absolute abundance of oxidized lipid species. The targeted lipidomics analysis covered six major lipid classes: glycerophospholipids (GP), glycerolipids (GL), sphingolipids (SP), acylcarnitines (AC), sterol lipids (ST), and fatty acyls (FA).

2.4. Lipid Extraction

Lipid extraction was performed using the butanol–methanol (1:1, v/v) single-phase method [20], which is a variant of the chloroform-free BUME extraction protocol validated for efficient recovery of sterols, glycerolipids, glycerophospholipids, and sphingolipids from animal tissues [21] as well as for adipose-rich tissues [22]. Frozen samples were thawed at 4 °C and thoroughly mixed by vortexing. An aliquot of 20 mg from each sample was transferred to a centrifuge tube, followed by the addition of 190 μL of butanol–methanol (1:1, v/v) containing 10 mM ammonium formate and 10 μL of internal standard solution. The internal standard mixture contained deuterated lipid standards representing each target lipid class (including PC (14:0/14:0)-d4, PE (14:0/14:0)-d4, and cholesterol-d7) at known concentrations. The mixture was vortexed for 30 s and sonicated at room temperature for 1 h. The samples were then centrifuged at 13,000× g for 10 min, and the supernatant was collected for subsequent LC–MS/MS analysis. All extraction procedures were performed on ice to minimize lipid degradation and oxidation.

2.5. LC-MS/MS Analysis

Chromatographic separation was performed using a Nexera X2 LC-30AD ultra-high-performance liquid chromatography (UHPLC) system (Shimadzu, Japan, purchased in Shanghai, China) equipped with an ACQUITY UPLC BEH HILIC column (130 Å, 1.7 μm, 2.1 × 100 mm; Waters, USA, purchased in Shanghai, China). The HILIC-based separation and targeted MS/MS detection followed the same platform previously validated for simultaneous profiling of complex lipids and polar metabolites [20]. Mobile phase A consisted of 50% acetonitrile containing 10 mM ammonium acetate (pH 8.0), and mobile phase B consisted of 100% acetonitrile. The gradient elution program was as follows: 0–0.1 min, 90% B; 0.1–5.0 min, 90–65% B; 5.0–5.3 min, 65–0% B; 5.3–7.2 min, 0% B; 7.2–7.4 min, 0–90% B; and 7.4–10.0 min, 90% B. The column temperature was maintained at 35 °C, the flow rate at 300 μL/min, and the injection volume was 6 μL.

Mass spectrometric analysis was performed using a Triple Quad 6500+ mass spectrometer (AB SCIEX, USA, purchased in Shanghai, China) in both positive- and negative-ion modes. The ion spray voltage was set to +5500 V and −4500 V in positive- and negative-ion modes, respectively. The source temperature was 400 °C, with ion source gas 1 and gas 2 set at 55 and 60 psi, respectively, curtain gas at 35 psi, and collision gas at 8. The declustering potential (DP) and collision energy (CE) were optimized for individual lipid species.

Chromatographic peak areas were extracted and integrated using Multiquant software (version 3.0.3; SCIEX, USA). Lipid identification and quantification were performed by comparing retention times with lipid standards, using the internal standard method. Sample concentration (X, ng·mg⁻¹) was calculated as:

$$X={\displaystyle \frac{R·M_{IS}}{M}}$$

where R is the peak-area ratio of the analyte to the corresponding internal standard, MIS is the amount of internal standard added (ng), and M is the mass of sample used for extraction (mg).

2.6. Quality Control and Data Processing

Quality control (QC) samples were prepared by pooling equal volumes of supernatant from all extraction samples. QC samples were injected periodically throughout the analytical run (every 10 injections) to monitor instrument performance and data stability. Lipid species with a coefficient of variation (CV) > 30% in QC samples were excluded from subsequent analysis [23]. A total of 2105 lipid species were initially detected; after QC filtering, 98 species with QC-RSD < 30% were retained for statistical analysis. All quantified lipid data were normalized by internal standard recovery and sample dry weight prior to statistical analysis.

2.7. Statistical Analysis

All statistical analyses were performed using IBM SPSS Statistics 27 (IBM Corp., Armonk, NY, USA). Data are presented as mean ± standard deviation (SD). Normality of data distribution was assessed using the Shapiro–Wilk test, and homogeneity of variances was confirmed using Levene’s test. A two-way analysis of variance (ANOVA) was performed to evaluate the effects of seahorse species, mysid storage group (2-month vs. 10-month), and their interaction on the abundance of individual lipid species. When significant differences were detected, post hoc pairwise comparisons were performed using Tukey’s honestly significant difference (HSD) test. To control for false discovery rates arising from multiple comparisons across 98 lipid species, the Benjamini–Hochberg procedure was applied at a false discovery rate (FDR) of 0.05 as previously recommended for lipidomics and metabolomics data [24]. For visualization purposes, heatmap data were Z-score normalized before hierarchical clustering. However, because each species-by-storage treatment had only one rearing tank, tank effects were completely confounded with treatment effects. The three pooled samples per group were analytical subsamples from the same tank and were not independent biological replicates. Therefore, the ANOVA results should be interpreted as exploratory and descriptive rather than as formal tests of storage-duration effects at the experimental-unit level. Future studies with replicated tanks per treatment are required to establish robust treatment effects. Differences were considered statistically significant at p < 0.05 after FDR correction, but these results should be interpreted with caution because of the limited experimental replication.

3. Results

A total of 2105 lipid species were initially detected in cultured seahorse powder. After quality control filtering based on QC-RSD < 30%, 98 lipid species were retained for subsequent analysis. These 98 species comprised six lipid classes: fatty acids (FA, 8 species), glycerolipids (GL, 16 species), glycerophospholipids (GP, 38 species), sphingolipids (SP, 25 species), sterol lipids (ST, 5 species), and acylcarnitines (AC, 6 species). Quantitative data for all 98 species are systematically presented in the following sections.

3.1. Fatty Acid Composition and Oxidative Status of the Mysid Feed

Table S1 (based on data from [8]) presents the fatty acid composition and TBARS values of the 2-month and 10-month frozen mysid feed treatments used in this trial. These data are provided as background information; because the two feeds were collected at different times, they should not be interpreted as originating from the same collection batch. The TBARS values of the same feed samples are presented in Table S2. TBARS remained low in fresh (0.122 mg/100 g) and 2-month frozen mysids (0.225 mg/100 g), with no significant difference between these two groups, but increased sharply after 10 months of storage (1.247 mg/100 g, p < 0.01). This indicates that severe lipid oxidation occurred mainly during the late storage stage. Relative to fresh mysids, both frozen feed treatments showed substantial free fatty acid losses, but the pattern of change differed markedly between them. Rather than continuing to decline from 2 to 10 months, several free fatty acid classes rose sharply during this interval: C18:2 and C18:3 showed the largest gains (+153% and +388%, respectively), followed by C20:4 (+64%) and C20:3 (+77%) (p < 0.05 for all), while C14:0 nearly tripled. Free EPA, DHA, C18:1, C16:0, and C18:0 remained statistically unchanged between the two frozen time points.

3.2. Heatmap Analysis of Lipid Abundance Patterns

Quality control samples were injected every 10 injections throughout the analytical run, and their PCA score plots showed tight clustering (Supplementary Figure S1), indicating analytical repeatability of the pooled samples. Only lipid species with QC-RSD < 30% were retained for subsequent analysis, reducing the initial 2105 features to 98 reliable species. Hierarchical clustering heatmap analysis (Figure 1) was used for exploratory visualization of the relative abundance patterns of the 98 retained lipid species after Z-score normalization. The three analytical replicates within each group generally clustered together. Group-associated patterns were observed across multiple lipid classes. In H. trimaculatus, a cluster comprising several TG, DG, SM, and HexCer species showed relatively higher standardized abundances in the 10-month group than in the 2-month group. In H. erectus, differences between storage groups were observed for a cluster containing several fatty acids, whereas multiple glycerophospholipids and sphingolipids displayed contrasting abundance patterns. In H. abdominalis, differences between storage groups were observed for several glycerophospholipids, acylcarnitines, glycerolipids, and sphingolipids. However, because each species-by-storage treatment had only one tank, these heatmap patterns should be interpreted as exploratory and descriptive rather than as independent evidence of treatment effects.

Figure 1. Heatmap of the 98 lipid species retained for focused analysis. Colors represent Z-score-normalized relative abundances, ranging from low (blue) to high (red). A1–C2 denote the six species-by-storage groups, with three pooled samples per group. Hierarchical clustering of lipid species is shown on the left.

3.3. Class-Specific PCA Reveals Differential Responses Across Lipid Categories

To explore storage-associated lipidomic patterns while reducing the influence of interspecific variation, PCA was performed separately for each species (colored by storage duration), in addition to a global PCA across all species (colored by species). The following sections present the PCA results and quantitative data for each of the six lipid classes.

3.4. Storage-Associated Alterations in Six Major Lipid Classes

3.4.1. Fatty Acids

PCA of fatty acid profiles revealed distinct separation patterns both among species and between storage durations within each species (Figure 2a,b). The global PCA across all three species showed clear species-specific clustering (Figure 2a), indicating that the three seahorse species possess intrinsically different fatty acid compositions. When analyzed separately by species, PCA score plots colored by storage duration showed that H. trimaculatus exhibited the most pronounced separation between 2-month and 10-month storage groups (Figure 2b), whereas H. erectus showed moderate separation (Figure 2c), and H. abdominalis displayed the weakest separation (Figure 2d). This species-dependent pattern may indicate differences in fatty acid profiles among the three seahorse species under the two feed treatments. However, because rearing temperature and feed collection time differed among treatments, these patterns cannot be attributed solely to storage duration.

Figure 2. PCA score plots of fatty acids (FA, 8 species). (a) Global PCA with samples colored by species (Ht, He, Ha). (b) Species-specific PCA for H. trimaculatus colored by storage duration. (c) Species-specific PCA for H. erectus colored by storage duration. (d) Species-specific PCA for H. abdominalis colored by storage duration.

Quantitative analysis of the pooled samples (Table 1) showed similar descriptive patterns. In H. trimaculatus, the C18 PUFAs FA (18:2) and FA (18:3) increased significantly in the 10-month group (6.2-fold and 2.9-fold, respectively), whereas the oxidized FA 24:0; O decreased significantly. No significant changes were observed for the long-chain PUFAs EPA, DPA, or DHA. In H. erectus, all eight measured fatty acids increased significantly in the 10-month group, with particularly pronounced elevations in FA (18:2) (5.3-fold), FA (18:3) (17.4-fold), and DHA (4.0-fold). By contrast, H. abdominalis showed significant decreases in EPA, DPA, DHA, the saturated FA (10:0) and FA (11:0), and 2-hydroxytetracosanoic acid after 10-month storage, while FA (18:2) and FA (18:3) remained unchanged. Arachidonic acid (FA 20:4) showed a species-dependent pattern. In H. erectus, FA (20:4) increased significantly in the 10-month group, whereas no significant changes were detected in H. trimaculatus or H. abdominalis. These differences should be interpreted cautiously because of the limited replication and confounding factors discussed in Section 4.5. In the present pooled samples, the long-chain PUFAs EPA, DPA, and DHA showed different patterns among the three species: they remained unchanged in H. trimaculatus, increased in H. erectus, and decreased in H. abdominalis in the 10-month group. These comparisons should be interpreted with caution because H. abdominalis was reared at a lower temperature and the two feeds were collected at different times.

Table 1. Fatty acid contents in pooled samples of three seahorse species fed mysid shrimp with different frozen storage durations.

Lipid Species	H. trimaculatus		H. erectus		H. abdominalis
	2-Month Storage	10-Month Storage	2-Month Storage	10-Month Storage	2-Month Storage	10-Month Storage
	ng·mg −1
FA (18:2)	34.7916 ± 3.0786 Ab	215.8514 ± 16.3517 Aa	76.5250 ± 5.5514 Bb	402.8557 ± 52.4061 Ba	210.2613 ± 60.9087 Ba	184.9469 ± 23.3157 Ba
FA (18:3)	5.094 ± 0.6009 Bb	14.7506 ± 0.3307 Ba	2.9066 ± 0.2935 Ab	50.4554 ± 10.1795 Aa	12.3198 ± 0.8471 Ba	11.3264 ± 1.0221 Ba
FA (20:4)	1.0834 ± 1.0767 Aa	2.565 ± 0.8074 Aa	0.2542 ± 0.4403 Ab	2.7218 ± 0.1769 Aa	0.9284 ± 0.8153 Aa	1.6356 ± 0.3944 Aa
FA (20:5)	1.5479 ± 0.1395 Aa	1.8497 ± 0.2342 Aa	1.0529 ± 0.0859 Ab	3.1129 ± 0.3218 Aa	2.9158 ± 0.9107 Aa	1.2171 ± 0.0587 Ab
FA (22:5)	1.2185 ± 0.2685 Ba	1.3171 ± 0.2160 Ba	0.4421 ± 0.1641 Ab	4.1145 ± 0.6993 Aa	2.1905 ± 0.7829 Ba	0.7791 ± 0.1274 Bb
FA (22:6)	3.9792 ± 0.4429 Ba	2.6034 ± 0.1566 Ba	2.1292 ± 0.0894 Ab	8.4115 ± 2.0198 Aa	5.8281 ± 1.9773 ABa	2.5310 ± 0.1721 ABb
FA (10:0)	5.3426 ± 0.3441 Ba	6.2631 ± 0.5121 Ba	4.5365 ± 0.7156 Ab	12.6498 ± 1.6975 Aa	11.0232 ± 2.9167 Aa	4.6491 ± 0.6184 Ab
FA (11:0)	0.1295 ± 0.0057 Ba	0.0276 ± 0.0194 Ba	0.0128 ± 0.0047 Ab	0.7774 ± 0.1928 Aa	0.2718 ± 0.1933 Ba	0.0132 ± 0.0106 Bb
FA 24:0; O_2-hydroxy-tetracosanoic acid	0.8783 ± 0.0469 Aa	0.2728 ± 0.0300 Ab	0.1227 ± 0.0316 Bb	0.7528 ± 0.0472 Ba	0.5357 ± 0.2385 Ca	0.0693 ± 0.0045 Cb

Data are presented as mean ± SD. Different lowercase letters (a, b) indicate differences between the 2-month and 10-month storage groups within the same seahorse species; different uppercase letters (A, B, C) indicate significant differences among seahorse species within the same storage group (p < 0.05). Values sharing the same letter are not significantly different.

3.4.2. Glycerolipids

PCA of glycerolipid profiles revealed that interspecific differences were less pronounced than for fatty acids, while species-specific separation between storage durations remained evident (Figure 3a–d). The global PCA showed partial overlap among species (Figure 3a). When analyzed separately, H. trimaculatus again showed the clearest separation between storage groups (Figure 3b), whereas H. erectus and H. abdominalis displayed weaker separation (Figure 3c,d).

Figure 3. PCA score plots of glycerolipids (GL, 16 species). (a) Global PCA colored by species. (b) H. trimaculatus colored by storage duration. (c) H. erectus colored by storage duration. (d) H. abdominalis colored by storage duration.

Quantitative analysis of the pooled samples (Table 2) showed similar descriptive patterns. In H. trimaculatus, the majority of measured glycerolipids—including DG (12:0_20:4), DG (12:0_20:5), all eight TG species, and all three MG species—were significantly higher in the 10-month group than in the 2-month group. By contrast, H. erectus showed either no significant changes or significant decreases (e.g., DG (12:0_18:2), DG (12:0_22:6), and MG 14:1) in the 10-month group. H. abdominalis displayed a mixed pattern: DG (12:0_22:6), MG 22:5, and MG 22:6 increased, while several TG species (TG44:2-FA18:2 through TG48:4-FA18:2) decreased significantly in the 10-month group. Among PUFA-containing glycerolipids, the ARA-containing DG (12:0_20:4) and EPA-containing DG (12:0_20:5) increased significantly only in H. trimaculatus, with no significant changes in the other two species. In contrast, the DHA-containing DG (12:0_22:6) remained unchanged in H. trimaculatus, decreased in H. erectus, and increased in H. abdominalis.

Table 2. Glycerolipid contents in three seahorse species fed mysid shrimp with different frozen storage durations.

Lipid Species	H. trimaculatus		H. erectus		H. abdominalis
	2-Month Storage	10-Month Storage	2-Month Storage	10-Month Storage	2-Month Storage	10-Month Storage
	ng·mg −1
DG (12:0_18:2)	29.5397 ± 5.3903 ABa	26.8384 ± 7.5325 ABa	30.6898 ± 3.6844 Ba	10.6349 ± 2.1332 Bb	30.9448 ± 6.7067 Aa	28.4741 ± 8.674 Aa
DG (12:0_20:4)	39.7007 ± 4.1908 Ab	70.9997 ± 21.9643 Aa	22.4561 ± 4.8653 Ca	12.2657 ± 1.5233 Ca	40.6924 ± 6.2358 Ba	39.6514 ± 6.4273 Ba
DG (12:0_20:5)	247.4383 ± 66.4247 Ab	481.3660 ± 91.8575 Aa	156.8127 ± 13.6751 Ca	119.6647 ± 26.0148 Ca	217.1728 ± 21.1978 Ba	297.1442 ± 31.1432 Ba
DG (12:0_22:6)	147.8100 ± 30.2893 Ba	118.3707 ± 21.6919 Ba	79.7401 ± 18.9785 Ca	28.5349 ± 4.3909 Cb	216.3168 ± 13.6284 Ab	254.3605 ± 25.3438 Aa
DG (12:1_22:5)	39.4450 ± 7.5938 Aa	56.2852 ± 15.0345 Aa	37.0584 ± 4.8263 Ba	21.1782 ± 0.8042 Ba	46.5629 ± 18.5320 ABa	33.7963 ± 3.3121 ABa
TG44:2-FA18:2	0.6928 ± 0.1427 Ab	2.3835 ± 0.2226 Aa	0.1642 ± 0.0435 Ca	0.1359 ± 0.0049 Ca	0.7031 ± 0.3882 Ba	0.2601 ± 0.0548 Bb
TG46:2-FA18:2	2.7997 ± 0.3673 Ab	15.6080 ± 1.8955 Aa	1.4957 ± 0.2341 Ca	1.1601 ± 0.1361 Ca	4.1742 ± 1.7129 Ba	1.6269 ± 0.3202 Bb
TG46:3-FA18:2	0.5581 ± 0.0797 Ab	2.0205 ± 0.4175 Aa	0.4120 ± 0.0916 Ca	0.2341 ± 0.0295 Ca	0.9322 ± 0.3958 Ba	0.5891 ± 0.1046 Ba
TG46:3-FA18:3	1.3511 ± 0.2028 Ab	6.2171 ± 0.7508 Aa	0.9900 ± 0.0495 Ca	0.6346 ± 0.0542 Ca	2.1002 ± 0.6899 Ba	0.8780 ± 0.1816 Bb
TG48:2-FA18:2	5.7585 ± 0.7336 Ab	59.0105 ± 2.9766 Aa	5.6001 ± 0.8111 Ca	6.2290 ± 0.6400 Ca	18.0620 ± 6.7184 Ba	7.7035 ± 2.0090 Bb
TG48:3-FA18:2	2.1406 ± 0.3364 Ab	22.4167 ± 1.5325 Aa	3.7688 ± 0.6615 Ca	3.1328 ± 0.4487 Ca	9.8691 ± 3.3560 Ba	5.7904 ± 1.6986 Bb
TG48:3-FA18:3	2.5848 ± 0.3806 Ab	24.6873 ± 2.9915 Aa	3.7027 ± 0.5253 Ca	3.3134 ± 0.2467 Ca	8.4023 ± 2.6610 Ba	3.6373 ± 0.8021 Ba
TG48:4-FA18:2	0.3691 ± 0.0972 Ab	2.2119 ± 0.2601 Aa	0.3412 ± 0.0819 Ca	0.2796 ± 0.0508 CaF	1.1585 ± 0.4301 Ba	0.7344 ± 0.1443 Bb
MG 22:5	40.9244 ± 2.9024 Bb	50.4887 ± 4.2730 Ba	15.8633 ± 1.0734 Ca	16.4991 ± 2.8097 Ca	41.5906 ± 5.4064 Ab	68.2715 ± 3.4523 Aa
MG 22:6	94.1383 ± 20.6241 Bb	136.2605 ± 19.1098 Ba	66.1834 ± 3.9946 Ca	38.6110 ± 3.7333 Ca	238.0856 ± 25.7633 Ab	361.9636 ± 13.3009 Aa
MG 14:1	79.0592 ± 7.8529 Ab	131.4449 ± 6.8146 Aa	78.2106 ± 3.4754 Ba	48.8000 ± 11.8149 Bb	96.5779 ± 5.3013 Aa	105.6493 ± 3.8551 Aa

Data are presented as mean ± SD. Different lowercase letters (a, b) indicate differences between the 2-month and 10-month storage groups within the same seahorse species; different uppercase letters (A, B, C) indicate significant differences among seahorse species within the same storage group (p < 0.05). Values sharing the same letter are not significantly different.

3.4.3. Glycerophospholipids

Glycerophospholipids exhibited the most complex and species-divergent responses (Figure 4a–d; Table 3). The global PCA showed clear species separation (Figure 4a). Species-specific PCA revealed that H. trimaculatus displayed moderate separation between storage groups (Figure 4b), whereas H. erectus showed weak separation (Figure 4c), and H. abdominalis exhibited intermediate separation (Figure 4d).

Figure 4. PCA score plots of glycerophospholipids (GP, 38 species). (a) Global PCA colored by species. (b) H. trimaculatus colored by storage duration. (c) H. erectus colored by storage duration. (d) H. abdominalis colored by storage duration.

Table 3. Glycerophospholipid contents in three seahorse species fed mysid shrimp with different frozen storage durations.

Lipid Species	H. trimaculatus		H. erectus		H. abdominalis
	2-Month Storage	10-Month Storage	2-Month Storage	10-Month Storage	2-Month Storage	10-Month Storage
	ng·mg −1
PG (14:0_22:5)	0.0008 ± 0.0014 Aa	0.0016 ± 0.0004 Aa	0.0012 ± 0.000038 Aa	0.000023 ± 0.0000 Aa	0.0016 ± 0.0014 Aa	0.0022 ± 0.0001 Aa
PG (14:0_22:6)	0.0021 ± 0.0019 Ba	0.0022 ± 0.0005 Ba	0.0016 ± 0.0012 Ba	0.000023 ± 0.0000 Ba	0.0054 ± 0.0004 Aa	0.0034 ± 0.0006 Ab
PG (14:1_22:5)	0.0056 ± 0.0009 Ba	0.0078 ± 0.0005 Ba	0.0035 ± 0.0004 Bb	0.0087 ± 0.0024 Ba	0.0176 ± 0.0020 Ab	0.0234 ± 0.0034 Aa
PG (16:0_18:2)	0.0467 ± 0.0054 Ab	0.1638 ± 0.0176 Aa	0.1438 ± 0.0178 Ca	0.0290 ± 0.0393 Cb	0.0976 ± 0.0182 Ba	0.0921 ± 0.0033 Ba
PC (16:1_18:2)	2.0644 ± 0.1033 Cb	7.4902 ± 1.2525 Ca	8.6603 ± 0.0697 Ba	5.3762 ± 0.1218 Bb	14.1761 ± 2.0141 Aa	12.7707 ± 0.9145 Aa
PC (18:0_18:3)	1.2115 ± 0.3247 Cb	4.4841 ± 0.8508 Ca	7.4494 ± 0.9197 Ba	2.9956 ± 0.5262 Bb	13.5193 ± 1.0367 Aa	7.1611 ± 1.1640 Aa
PC (18:1_18:3)	1.2846 ± 0.3652 Aa	3.4103 ± 0.4062 Aa	9.7589 ± 0.4680 Bb	2.7869 ± 0.3467 Ba	15.8307 ± 2.0684 Cb	9.0257 ± 1.6344 Ca
PC (18:2_16:0)	15.3880 ± 2.902 Cb	32.6197 ± 2.4774 Ca	54.6247 ± 6.0019 Ba	31.6274 ± 2.0157 Bb	61.0916 ± 9.9317 Aa	44.6882 ± 3.7211 Ab
PC (18:2_16:1)	1.0505 ± 0.4519 Bb	5.1049 ± 0.9728 Ba	5.3474 ± 0.0743 Ba	2.7764 ± 0.5590 Bb	8.6306 ± 1.6180 Aa	7.1452 ± 1.6299 Ab
PC (20:0_18:2)	0.1790 ± 0.1586 Bb	0.8096 ± 0.5640 Ba	2.2831 ± 0.0970 Aa	0.2261 ± 0.1488 Ab	1.6453 ± 0.185 Aa	0.5411 ± 0.3364 Ab
PC (20:1_18:2)	0.2708 ± 0.1932 Bb	1.0268 ± 0.2059 Ba	3.1717 ± 0.5534 Aa	0.2769 ± 0.0814 Ab	2.8367 ± 0.4451 Aa	0.6623 ± 0.2373 Ab
PE (14:0_20:5)	0.000023 ± 0.0000 Bb	0.0269 ± 0.0072 Ba	0.0370 ± 0.0165 Ba	0.0100 ± 0.0097 Bb	0.0398 ± 0.0153 Aa	0.0407 ± 0.0107 Aa
PE (16:0_18:2)	0.4681 ± 0.0498 Cb	0.6529 ± 0.0482 Ca	1.1630 ± 0.0455 Ba	0.7650 ± 0.0931 Bb	1.0881 ± 0.1233 Aa	1.2335 ± 0.1395 Aa
PE (16:0_20:4)	0.5547 ± 0.0119 Ab	1.6471 ± 0.1988 Aa	1.6489 ± 0.0249 Aa	0.8051 ± 0.0863 Ab	0.9853 ± 0.1496 Ab	1.3091 ± 0.0477 Aa
PA (18:0_18:2)	0.0585 ± 0.0233 Ab	0.1740 ± 0.0341 Aa	0.1310 ± 0.0234 Aa	0.0640 ± 0.0449 Aa	0.1249 ± 0.0795 Aa	0.1741 ± 0.0145 Aa
PA (18:0_22:6)	0.1957 ± 0.0188 Bb	0.5874 ± 0.0647 Ba	0.4727 ± 0.0303 Ab	0.1250 ± 0.0190 Aa	0.3392 ± 0.0776 Ba	0.3164 ± 0.0382 Ba
PA (20:0_18:3)	0.1241 ± 0.0248 Aa	0.4864 ± 0.0520 Aa	0.9665 ± 0.1197 Ab	0.0379 ± 0.0165 Aa	1.0456 ± 0.0621 Aa	1.1696 ± 0.0473 Ab
PA (20:0_22:5)	0.1917 ± 0.0768 Ba	1.0789 ± 0.0427 Ba	0.4938 ± 0.0742 Ab	0.2689 ± 0.0389 Aa	0.3643 ± 0.0403 Ba	0.4534 ± 0.0567 Bb
PI (14:1_18:2)	0.0002 ± 0.0004 Aa	0.0008 ± 0.0006 Aa	0.0011 ± 0.0010 Aa	0.000023 ± 0.0000 Aa	0.0015 ± 0.0013 Aa	0.0005 ± 0.0002 Aa
PI (16:0_20:4)	1.7609 ± 0.2859 Ab	1.0204 ± 0.0890 Aa	1.4431 ± 0.1118 Aa	0.7041 ± 0.6175 Ab	1.7967 ± 0.1563 Aa	0.8041 ± 0.0658 Ab
PI (16:0_20:5)	0.6731 ± 0.1184 Ca	0.3307 ± 0.0351 Cb	0.9513 ± 0.0621 Ba	0.2355 ± 0.1774 Bb	1.4943 ± 0.1193 Aa	0.4825 ± 0.0657 Ab
PI (16:0_22:5)	1.4383 ± 0.2814 Aa	1.5325 ± 0.1300 Aa	0.7527 ± 0.0413 Ca	0.8308 ± 0.6685 Ca	1.7135 ± 0.1981 Ba	0.4538 ± 0.0765 Bb
PI (16:0_22:6)	2.5256 ± 0.3955 Ba	1.0095 ± 0.1608 Bb	1.4333 ± 0.1511 Ca	0.8166 ± 0.6434 Ca	4.7174 ± 0.4486 Aa	0.9086 ± 0.0798 Ab
PI (16:1_18:2)	0.0257 ± 0.0039 Ba	0.0335 ± 0.0019 Ba	0.0172 ± 0.0017 Ca	0.0182 ± 0.0169 Ca	0.0740 ± 0.0099 Aa	0.0340 ± 0.0066 Ab
PI (16:1_18:3)	0.0313 ± 0.0276 Ba	0.0266 ± 0.0018 Ba	0.0290 ± 0.0040 Ca	0.0176 ± 0.0150 Ca	0.0846 ± 0.0074 Aa	0.0284 ± 0.0045 Ab
PS (14:1_22:5)	0.4628 ± 0.0622 Ba	1.8587 ± 0.4889 Ba	0.2098 ± 0.0603 Ab	3.8954 ± 2.0303 Aa	0.1375 ± 0.0216 Ca	0.2733 ± 0.0765 Ca
PS (16:1_20:4)	0.1800 ± 0.0446 Aa	0.3240 ± 0.0481 Aa	0.0469 ± 0.0208 ABb	0.3998 ± 0.2982 ABa	0.0175 ± 0.0025 Ba	0.0544 ± 0.0134 Ba
PS (16:1_22:5)	1.1165 ± 0.1592 Aa	0.4216 ± 0.0566 Ab	0.1509 ± 0.0717 Bb	0.4357 ± 0.2384 Ba	0.0296 ± 0.0184 Ca	0.0345 ± 0.0114 Ca
PS (18:1_22:5)	0.000023 ± 0.0000 Ba	0.0212 ± 0.0142 Ba	0.0270 ± 0.0212 ABa	0.0283 ± 0.0264 ABa	0.0402 ± 0.0027 Aa	0.0427 ± 0.0078 Aa
LPC (18:3)	1.6709 ±0.0712 Ca	3.1237 ± 0.4401 Ca	18.7032 ± 4.3125 Aa	4.2313 ± 0.6921 Ab	3.6786 ± 0.7159 Bb	12.5966 ± 0.3672 Ba
LPC (18:0)	11.1388 ± 0.4235 Aa	1.5179 ± 1.3161 Ab	7.5371 ± 1.7258 Aa	2.7310 ± 0.8381 Ab	3.9415 ± 0.2782 Ba	0.000023 ± 0.0000 Bb
LPC (22:0)	1.2851 ± 0.1422 Aa	1.1107 ± 0.0095 Aa	1.9772 ± 0.4495 Aa	0.7254 ± 0.0829 Ab	0.8567 ± 0.0383 Ba	0.7101 ± 0.0858 Ba
LPC (22:1)	0.8581 ± 0.0489 Bb	1.1134 ± 0.0486 Ba	1.5328 ± 0.2527 Aa	0.7830 ± 0.0637 Ab	0.6182 ± 0.0521 Ca	0.6142 ± 0.0640 Ca
LPC (22:2)	1.5545 ± 0.0862 Ab	3.1676 ± 0.3120 Aa	2.8396 ± 0.6879 Aa	1.7018 ± 0.2085 Ab	1.1786 ± 0.0500 Ba	1.2684 ± 0.0191 Ba
LPC (22:3)	2.1822 ±0.0512 Ca	4.3718 ± 0.3173 Ca	15.7106 ± 3.4128 Aa	7.8347 ± 1.8051 Ab	2.2613 ± 0.2129 Bb	9.9824 ± 0.8579 Ba
LPE (12:0)	0.5892 ± 0.6521 Ba	0.6577 ± 0.1638 Ba	0.0978 ± 0.1693 Ba	0.4823 ± 0.8353 Ba	2.7357 ± 0.6938 Aa	0.7421 ± 0.2291 Ab
LPE (18:0)	443.9972 ± 106.1854 Aa	305.9560 ± 70.8759 Aa	74.5145 ± 7.7182 Bb	246.7177 ± 152.749 Ba	214.3349 ± 32.6876 Ba	112.6112 ± 8.0223 Ba
LPE (22:1)	10.5182 ± 4.308 Aa	5.6870 ± 1.5785 Ab	2.7396 ± 0.2185 Ba	3.3368 ± 2.7473 Ba	6.8134 ± 1.3732 Ba	1.1590 ± 0.5690 Bb

Data are presented as mean ± SD. Different lowercase letters (a, b) indicate differences between the 2-month and 10-month storage groups within the same seahorse species; different uppercase letters (A, B, C) indicate significant differences among seahorse species within the same storage group (p < 0.05). Values sharing the same letter are not significantly different.

Quantitative analysis (Table 3) revealed the most striking contrast in phosphatidylcholine (PC) species: six of the seven measured PCs increased in H. trimaculatus, whereas all seven decreased in H. erectus, and four decreased in H. abdominalis. Because H. trimaculatus and H. erectus were reared at the same temperature, the opposite direction of PC changes may suggest possible species differences in membrane lipid remodeling. However, this interpretation remains speculative and requires functional validation. Among 20:4-containing glycerophospholipids, PI (16:0_20:4) increased in H. trimaculatus but decreased in both H. erectus and H. abdominalis; PE (16:0_20:4) increased in H. trimaculatus and H. abdominalis but decreased in H. erectus. PA (18:0_22:6) increased in both H. trimaculatus and H. erectus while remaining unchanged in H. abdominalis.

3.4.4. Sphingolipids

Sphingolipid PCA revealed distinct species separation in the global analysis (Figure 5a). Species-specific analysis showed that H. trimaculatus exhibited the strongest separation between storage groups (Figure 5b), whereas H. erectus and H. abdominalis displayed weaker separation (Figure 5c,d).

Figure 5. PCA score plots of sphingolipids (SP, 25 species). (a) Global PCA colored by species. (b) H. trimaculatus colored by storage duration. (c) H. erectus colored by storage duration. (d) H. abdominalis colored by storage duration.

Quantitative analysis (Table 4) corroborated these findings. In H. trimaculatus, Cer d16:1/18:2, Cer d16:1/18:3, three HexCer species, and all eight SM species were higher in the 10-month group, whereas five DhCer species decreased. In H. erectus, no sphingolipid increased significantly; decreases were observed for multiple Cer, HexCer, DhCer, and seven of eight SM species. In H. abdominalis, decreases occurred in several Cer, HexCer, six DhCer, and three SM species, while SM 31:0; O2 and SM 31:1; O2 increased. At the class level, sphingomyelin (SM) showed the clearest contrast: all eight SMs increased in H. trimaculatus, seven decreased in H. erectus, and H. abdominalis displayed mixed responses (three decreased, two increased). The SM responses generally paralleled the PC responses within each species.

Table 4. Sphingolipid contents in three seahorse species fed mysid shrimp with different frozen storage durations.

Lipid Species	H. trimaculatus		H. erectus		H. abdominalis
	2-Month Storage	10-Month Storage	2-Month Storage	10-Month Storage	2-Month Storage	10-Month Storage
	ng·mg −1
Cer d16:0/20:4	0.0163 ± 0.0056 Aa	0.0161 ± 0.0124 Aa	0.0163 ± 0.0037 Aa	0.0077 ± 0.0012 Aa	0.0125 ± 0.0049 Aa	0.0128 ± 0.0035 Aa
Cer d16:0/20:5	0.0194 ± 0.0108 Aa	0.0226 ± 0.0071 Aa	0.0198 ± 0.0061 Aa	0.0065 ± 0.0031 Ab	0.0104 ± 0.0046 Aa	0.0129 ± 0.0070 Aa
Cer d16:0/22:5	0.0373 ± 0.0049 Aa	0.0391 ± 0.0178 Aa	0.0353 ± 0.0047 Ba	0.0109 ± 0.0056 Bb	0.0296 ± 0.0042 Ba	0.0143 ± 0.0021 Bb
Cer d16:1/18:2	0.0861 ± 0.0235 Ab	0.3249 ± 0.0410 Aa	0.0958 ± 0.0152 Ca	0.0950 ± 0.0116 Ca	0.1324 ± 0.0520 Ba	0.1643 ± 0.0319 Ba
Cer d16:1/18:3	0.0456 ± 0.0099 Bb	0.0806 ± 0.0161 Ba	0.0602 ± 0.0112 Ca	0.0263 ± 0.0029 Cb	0.0914 ± 0.0096 Aa	0.0967 ± 0.0180 Aa
HexCer d18:2/12:0	0.0520 ± 0.0039 Aa	0.0622 ± 0.0117 Aa	0.0516 ± 0.0066 Ba	0.0256 ± 0.0068 Bb	0.0690 ± 0.0058 Aa	0.0416 ± 0.0067 Ab
HexCer d18:2/14:0	0.0283 ± 0.0256 Ab	0.0709 ± 0.0091 Aa	0.0483 ± 0.0061 Ba	0.0177 ± 0.0034 Bb	0.0480 ± 0.0112 ABa	0.0317 ± 0.0060 ABa
HexCer d18:2/22:0	0.0850 ± 0.0182 Ba	0.1116 ± 0.0281 Ba	0.1593 ± 0.0183 Ba	0.0681 ± 0.0171 Bb	0.1792 ± 0.0302 Aa	0.1794 ± 0.0310 Aa
HexCer d18:2/24:0	0.0345 ± 0.0092 Ab	0.0941 ± 0.0133 Aa	0.0372 ± 0.0323 Ba	0.00002 ± 0.00002 Ba	0.0265 ± 0.0459 ABa	0.0572 ± 0.0170 ABa
HexCer d18:2/24:1	0.0293 ± 0.0283 Bb	0.0961 ± 0.0035 Ba	0.0747 ± 0.0244 Ba	0.0347 ± 0.0127 Bb	0.1665 ± 0.0087 Aa	0.1143 ± 0.0203 Ab
DhCer d18:0/14:0	0.4221 ± 0.0408 Ba	0.5000 ± 0.0560 Ba	0.7784 ± 0.0315 Aa	0.7969 ± 0.1020 Aa	0.3974 ± 0.0590 Ca	0.2277 ± 0.0250 Cb
DhCer d18:0/16:0	0.5215 ± 0.0392 Ba	0.5366 ± 0.0747 Ba	0.7509 ± 0.0721 Aa	0.7658 ± 0.0897 Aa	0.4347 ± 0.0571 Ca	0.2545 ± 0.0384 Cb
DhCer d18:0/18:0	0.5113 ± 0.0333 Ba	0.3199 ± 0.0252 Bb	0.6706 ± 0.0444 Aa	0.5735 ± 0.0906 Aa	0.4750 ± 0.0926 Ba	0.3002 ± 0.0397 Bb
DhCer d18:0/20:0	0.2090 ± 0.0305 Ba	0.1219 ± 0.0058 Bb	0.2687 ± 0.0196 Aa	0.1475 ± 0.0143 Ab	0.1516 ± 0.0311 Ca	0.1015 ± 0.0173 Cb
DhCer d18:0/20:1	0.1802 ± 0.0236 Ba	0.1183 ± 0.0134 Bb	0.2411 ± 0.0158 Aa	0.2002 ± 0.0361 Aa	0.1604 ± 0.0337 Ba	0.1037 ± 0.0032 Bb
DhCer d18:0/22:0	0.5732 ± 0.0304 Ba	0.3354 ± 0.0265 Bb	1.5741 ± 0.0546 Aa	1.0105 ± 0.1865 Ab	0.4428 ± 0.0805 Ba	0.3772 ± 0.0238 Ba
DhCer d18:0/22:1	0.0999 ± 0.0219 Ba	0.0366 ± 0.0105 Bb	0.1257 ± 0.0048 Aa	0.0736 ± 0.0156 Ab	0.0843 ± 0.0266 Ba	0.0430 ± 0.0058 Bb
SM 28:0; O2	0.0532 ± 0.0061 Ab	0.1378 ± 0.0174 Aa	0.1407 ± 0.0155 Ba	0.0284 ± 0.0047 Bb	0.0846 ± 0.0035 Ca	0.0706 ± 0.0048 Ca
SM 28:1; O2	0.0554 ± 0.0075 Bb	0.2135 ± 0.0318 Ba	0.3217 ± 0.0304 Aa	0.0364 ± 0.0070 Ab	0.1070 ± 0.0101 Ca	0.1198 ± 0.0034 Ca
SM 29:0; O2	0.0499 ± 0.0055 Ab	0.0963 ± 0.0081 Aa	0.0608 ± 0.0042 Ca	0.0220 ± 0.0007 Cb	0.0777 ± 0.0039 Ba	0.0542 ± 0.0038 Bb
SM 30:0; O2	0.4254 ± 0.0339 Ab	1.1489 ± 0.0926 Aa	1.1766 ± 0.0654 Ba	0.3894 ± 0.0400 Bb	0.7714 ± 0.0430 Ca	0.6853 ± 0.0345 Ca
SM 30:1; O2	0.0100 ± 0.0014 Ab	0.0191 ± 0.0021 Aa	0.0054 ± 0.0047 Ca	0.0012 ± 0.0020 Ca	0.0099 ± 0.0012 Ba	0.0017 ± 0.0029 Bb
SM 30:2; O2	0.0294 ± 0.0031 Ab	0.0523 ± 0.0051 Aa	0.0437 ± 0.0036 Ca	0.0139 ± 0.0009 Cb	0.0423 ± 0.0021 Ba	0.0279 ± 0.0010 Bb
SM 31:0; O2	0.3077 ± 0.0262 Bb	0.4983 ± 0.0170 Ba	0.4909 ± 0.0287 Ca	0.3023 ± 0.0016 Cb	0.4500 ± 0.0179 Ab	0.5009 ± 0.0132 Aa
SM 31:1; O2	0.6704 ± 0.0365 Cb	0.9314 ± 0.0634 Ca	1.0472 ± 0.0362 Ba	0.6491 ± 0.0455 Bb	0.9036 ± 0.0364 Ab	1.1315 ± 0.0539 Aa

Data are presented as mean ± SD. Different lowercase letters (a, b) indicate differences between the 2-month and 10-month storage groups within the same seahorse species; different uppercase letters (A, B, C) indicate significant differences among seahorse species within the same storage group (p < 0.05). Values sharing the same letter are not significantly different.

3.4.5. Sterol Lipids

Sterol lipids showed the most consistent storage-associated decline across three species in the present dataset (Figure 6a–d; Table 5). The global PCA revealed species separation (Figure 6a). In species-specific PCA, across the three species showed some between storage groups (Figure 6b–d), with H. trimaculatus showing the most pronounced separation.

Figure 6. PCA score plots of sterol lipids (ST, 5 species). (a) Global PCA colored by species. (b) H. trimaculatus colored by storage duration. (c) H. erectus colored by storage duration. (d) H. abdominalis colored by storage duration.

Table 5. Sterol lipid contents in three seahorse species fed mysid shrimp with different frozen storage durations.

Lipid Species	H. trimaculatus		H. erectus		H. abdominalis
	2-Month	10-Month	2-Month	10-Month	2-Month	10-Month
	ng·mg −1
ST 27:0; O2_6alpha-hydroxycholestanol	164.3079 ± 32.7094 Aa	80.8774 ± 5.0872 Ab	68.6451 ± 10.0653 Ba	88.1318 ± 25.9885 Ba	84.9876 ± 11.9043 Ca	62.1677 ± 12.7187 Ca
ST 27:1; O2_5beta,6beta-epoxy-cholesterol	98.6310 ± 9.5042 Ba	57.6371 ± 7.4273 Bb	61.0130 ± 3.9245 Ca	30.3305 ± 7.2175 Cb	117.0280 ± 7.2485 Aa	112.7453 ± 14.0089 Aa
ST 27:2; O2_24,25-epoxy-cholesterol	809.2961 ± 192.7767 Aa	529.4543 ± 14.0732 Ab	468.8204 ± 102.9131 Ba	366.3292 ± 8.2604 Ba	935.1201 ± 196.7766 Aa	578.3601 ± 74.2418 Ab
ST 27:2; O_3-ketocholesterol	960.6153 ± 108.9029 Aa	618.2656 ± 130.0592 Ab	600.2219 ± 187.3225 Ba	485.6408 ± 126.1386 Ba	765.9083 ± 66.6899 ABa	479.3596 ± 72.2610 ABb
ST 29:1; O_23-methylene-24-methylcholestanol	343.1579 ± 42.4054 Aa	233.3449 ± 47.0892 Ab	292.9137 ± 15.0259 Aa	189.7704 ± 19.5898 Ab	255.4818 ± 20.5900 Aa	232.9255 ±36.1583 Aa

Data are presented as mean ± SD. Different lowercase letters (a, b) indicate differences between the 2-month and 10-month storage groups within the same seahorse species; different uppercase letters (A, B, C) indicate significant differences among seahorse species within the same storage group (p < 0.05). Values sharing the same letter are not significantly different.

Quantitative analysis of the pooled samples (Table 5) showed that in H. trimaculatus, all five measured sterols—ST 27:0; O2, ST 27:1; O2, ST 27:2; O2, ST 27:2; O, and ST 29:1; O—decreased in the 10-month group. In H. erectus, ST 27:1; O2 and ST 29:1; O decreased, while the other three remained unchanged. In H. abdominalis, ST 27:2; O2 and ST 27:2; O decreased, with no significant changes in the remaining three. Notably, the first four sterols resolved by targeted analysis were oxidized cholesterol derivatives (e.g., epoxycholesterol and ketocholesterol), rather than native cholesterol. Their consistent decline across the three species in this dataset suggests that oxidized sterol derivatives may be candidate cross-species indicators of storage-associated feed changes. However, because only two storage durations were tested and no direct quantitative relationship between feed and tissue sterols was established, these compounds cannot yet be considered validated biomarkers of feed deterioration. A reduction in tissue sterol derivatives may reflect differences in dietary availability, absorption, metabolism, degradation, or tissue deposition rather than feed deterioration alone.

3.4.6. Acylcarnitines

Acylcarnitines showed the smallest magnitude of change among the six lipid classes examined (Figure 7a–d; Table 6). The global PCA showed moderate species separation (Figure 7a). Species-specific PCA revealed weak or no separation between storage groups across the three species (Figure 7b–d).

Figure 7. PCA score plots of acylcarnitines (AC, 6 species). (a) Global PCA colored by species. (b) H. trimaculatus colored by storage duration. (c) H. erectus colored by storage duration. (d) H. abdominalis colored by storage duration.

Table 6. Acylcarnitine contents in three seahorse species fed mysid shrimp with different frozen storage durations.

Lipid Species	H. trimaculatus		H. erectus		H. abdominalis
	2-Month Storage	10-Month Storage	2-Month Storage	10-Month Storage	2-Month Storage	10-Month Storage
	ng·mg −1
Acyl-Carnitine (22:6)	0.4808 ± 0.1732 Ca	0.4987 ± 0.0749 Ca	4.1217 ± 0.6889 Aa	2.7034 ± 1.1255 Aa	2.8275 ± 0.4538 Ba	1.7030 ± 0.2733 Ba
Acyl-Carnitine (2:0)	109.1885 ± 8.2847 Ca	71.8370 ± 1.7217 Ca	182.1343 ± 2.8933 Ba	100.6788 ± 15.3898 Ba	233.8784 ± 45.2849 Aa	176.7734 ± 9.1575 Aa
Acyl-Carnitine (3:0)	26.2034 ± 2.6076 Ca	18.1192 ± 0.2239 Ca	60.1737 ± 0.1405 Ba	34.1293 ± 5.9970 Ba	69.3545 ± 12.4463 Aa	51.4564 ± 2.1266 Ab
Acyl-Carnitine (4:0)	14.7599 ± 1.0639 Ca	5.2723 ± 0.1272 Ca	13.6994 ± 0.7556 Ba	16.6901 ± 4.7010 Ba	24.8973 ± 6.6718 Aa	18.3493 ± 1.6007 Ab
Acyl-Carnitine (4:0) OH	22.0284 ± 2.6988 Ba	10.2421 ± 0.3772 Bb	14.5347 ± 0.0565 Ba	15.3723 ± 5.2759 Ba	49.9681 ± 11.4053 Aa	25.1814 ± 2.0315 Ab
Acyl-Carnitine (5:0)	6.9075 ± 0.9509 Ba	2.7545 ± 0.1471 Bb	12.1866 ± 0.5495 Aa	10.8121 ± 1.9749 Aa	9.6332 ± 2.2947 Aa	11.9148 ± 0.8628 Aa

Data are presented as mean ± SD. Different lowercase letters (a, b) indicate differences between the 2-month and 10-month storage groups within the same seahorse species; different uppercase letters (A, B, C) indicate significant differences among seahorse species within the same storage group (p < 0.05). Values sharing the same letter are not significantly different.

Quantitative analysis of the pooled samples (Table 6) showed that in H. trimaculatus, Acyl-Carnitine (4:0) OH and Acyl-Carnitine (5:0) decreased in the 10-month group, while the other four acylcarnitines remained unchanged. In H. erectus, none of the six acylcarnitines differed between storage groups. In H. abdominalis, Acyl-Carnitine (3:0), Acyl-Carnitine (4:0), and Acyl-Carnitine (4:0) OH decreased, while the remaining three showed no change. These patterns may indicate altered acylcarnitine metabolism under prolonged feed storage. However, acylcarnitine profiles are only indirect indicators and do not provide direct evidence of mitochondrial fatty acid β-oxidation capacity without functional measurements. Therefore, the present results should be interpreted as exploratory.

4. Discussion

This study tested the hypothesis that frozen storage duration of mysid feed can be transmitted through the food chain to influence the lipid-based medicinal quality of farmed seahorses, and that this influence is not a uniform decline but rather a species-specific metabolic response. Our targeted lipidomics data are consistent with this hypothesis, but should be interpreted cautiously. They reveal three main patterns that support this hypothesis and reveal three principal findings. First, prolonged frozen storage did not cause a simple, linear deterioration of lipid profiles; instead, each species exhibited a distinct response pattern. Second, among all lipid classes examined, oxidized sterol derivatives showed the most consistent decline across species, making them a more promising candidate indicator of feed storage deterioration than total fatty acids or phospholipids. Third, the marked interspecific differences in sensitivity to prolonged storage suggest that frozen storage duration limits should be species-specific rather than uniform.

4.1. Species-Associated Lipid Patterns and Possible Metabolic Differences

The present dataset revealed species-associated, rather than uniform, lipidomic patterns in response to the two feed treatments. In H. erectus, fatty acids generally increased while many glycerophospholipids decreased; in H. abdominalis, multiple lipid classes—including PUFAs, sterols, phospholipids, and acylcarnitines—showed broad declines; and H. trimaculatus displayed an intermediate pattern, with membrane and energy-storage lipids mostly increasing while sterols decreased. These patterns are based on pooled samples from a single tank per treatment and should therefore be regarded as exploratory. In addition, the two feed treatments were collected at different times, and H. abdominalis was reared at a lower temperature than the other two species. Consequently, the observed differences cannot be attributed solely to species-specific responses.

Nevertheless, because H. erectus and H. trimaculatus were maintained under identical thermal conditions (24–25 °C) yet frequently showed opposite directions of change—most notably in phosphatidylcholine (PC) and sphingomyelin (SM)—temperature alone may not fully explain the divergent patterns. This observation may point to possible intrinsic differences in lipid metabolic capacity among species, although functional validation is required. Teleost fishes vary substantially in their ability to retain, remodel, and endogenously synthesize long-chain PUFA via fatty acyl desaturase (Fads2) and elongase (Elovl) activities, and this capacity has been shown to differ even among closely related species [25,26]. Dietary lipid composition can also markedly influence tissue fatty acid profiles and broader lipid-related physiological responses in cultured fish [27]. Based on these considerations, we hypothesize that H. erectus may have a relatively greater capacity for PUFA conservation or de novo synthesis under the present conditions, whereas H. abdominalis may depend more directly on dietary supply and therefore be more vulnerable to PUFA loss. However, this interpretation remains speculative and requires direct measurements of enzyme activity, gene expression, and functional lipid metabolism.

The coordinated changes in PC and SM across species provide further clues. PC and SM are metabolically coupled through the sphingomyelin synthase (SMS) reaction, in which PC serves as the phosphocholine donor for SM synthesis, generating diacylglycerol (DAG) as a byproduct [28]. In H. trimaculatus, both PC and SM increased in parallel; in H. erectus, both decreased in parallel. This concordance may suggest coordinated regulation of a shared membrane-lipid synthesis pathway, rather than coincidental changes in two independent lipid classes. The fact that this pattern occurred between two species reared at identical temperatures supports the possibility of intrinsic species differences in membrane lipid remodeling, potentially mediated by phospholipases and acyltransferases [29]. However, because only two storage durations and one tank per treatment were used, these mechanistic inferences should be treated as hypotheses for future testing. Recent lipidomic research in Atlantic salmon has likewise demonstrated that dietary lipid sources can modify tissue phospholipid composition together with the expression of genes involved in lipid metabolism, highlighting the close interaction between dietary lipid supply and metabolic regulation [30].

The feed data help contextualize these tissue patterns. Free C18:2 and C18:3 in the mysids’ feed increased between 2 and 10 months of storage, as microbial and endogenous lipases gradually released these fatty acids from structural lipids [8]. By 10 months, the diet contained a larger pool of readily absorbable C18 polyunsaturated fatty acids, and these were the two fatty acids that increased in H. erectus and H. trimaculatus. This suggests the apparent increase in host fatty acids may largely reflect direct dietary transfer of lipolysis products, rather than enhanced endogenous synthesis. H. abdominalis is an exception: despite receiving the same feed treatment, it showed no increase in FA (18:2) or FA (18:3), which may point to reduced assimilation or faster disposal of the extra substrate in this species. However, because the two feeds were collected at different times, these comparisons should be interpreted cautiously.

The picture is different for long-chain n-3 PUFAs. Free EPA and DHA levels did not differ significantly between the 2- and 10-month feed treatments [8], so dietary supply of these two key fatty acids remained essentially constant. Yet EPA, DPA, and DHA increased in H. erectus, remained unchanged in H. trimaculatus, and decreased in H. abdominalis. Since the dietary input was similar across species, this divergence may reflect species-associated differences in retention, β-oxidation, membrane remodeling, or sensitivity to oxidative stress from the aged feed. This is a notable pattern in the present dataset, suggesting a possible host-driven response, but it does not establish causality and may be influenced by temperature and tank-level effects. Functional measurements of β-oxidation, enzyme activity, and oxidative stress markers are needed to test this hypothesis.

Overall, these findings suggest that the three seahorse species may differ in their lipid metabolic responses to the two feed treatments, but the present experimental design cannot separate species effects from temperature, feed collection time, and tank-level confounding. Future studies with replicated tanks, common thermal regimes, multiple storage time points, and functional assays are needed to determine whether these species-associated patterns reflect intrinsic metabolic differences or are driven by husbandry and feed-batch variation.

4.2. Sterol Lipids: Candidate Cross-Species Indicators of Storage-Associated Changes

An important consideration is that the 2-month frozen feed was not equivalent to fresh feed. Data from [8] show that C18:2, C18:3, C20:4, and C20:5 decreased substantially from fresh to 2 months, but were relatively stable or rebounded from 2 to 10 months. Thus, the present 2-month vs. 10-month comparison captures later-stage changes rather than the initial deterioration phase. The early loss of free fatty acids, together with the absence of a significant TBARS increase at 2 months, suggests that early deterioration was primarily associated with freeze-induced cell lysis, lipolysis, and leaching of water-soluble nutrients, rather than lipid oxidation. In contrast, the marked increase in TBARS and TVB-N at 10 months indicates that late-stage deterioration was dominated by lipid oxidation and microbial spoilage. Among the six lipid classes examined, sterol lipids showed the most consistent storage-associated decline across the three seahorse species in the present dataset. In H. trimaculatus, all five measured sterols decreased significantly in the 10-month group; in H. erectus and H. abdominalis, two of the five sterols decreased significantly in each species. Notably, the sterols resolved by targeted analysis were predominantly oxidized cholesterol derivatives—such as epoxycholesterol, ketocholesterol, and hydroxycholesterol—rather than native cholesterol. This cross-species consistency contrasts with the divergent responses observed for fatty acids, glycerolipids, and glycerophospholipids, and suggests that oxidized sterol derivatives may be useful candidate indicators of storage-associated feed changes.

We hypothesize that this pattern is largely related to the dietary origin of these compounds. Unlike fatty acids, which can be endogenously synthesized or remodeled, oxysterols in animal tissues are thought to be primarily derived from dietary cholesterol oxidation products [31,32]. Their tissue levels may therefore reflect, at least in part, the input from the feed. During prolonged frozen storage, cholesterol in mysid shrimp can undergo progressive oxidation, and the resulting oxysterols are themselves unstable secondary oxidation products that may undergo further degradation over time [33]. Consequently, the decline in tissue oxysterol levels with prolonged feed storage does not necessarily indicate a reduction in overall oxidative damage. Instead, it may reflect a shift in the oxidation-product profile—from primary oxysterols toward more degraded species that were not captured by our targeted panel. This interpretation is consistent with previous reports that oxysterol composition in frozen seafood changes dynamically with storage duration, with certain oxysterol species declining as storage progresses [33].

However, several alternative explanations must be considered. A reduction in tissue sterol derivatives may reflect differences in dietary availability, intestinal absorption, metabolic conversion, degradation, or tissue deposition rather than feed deterioration alone. Because the present study did not quantify oxysterols directly in the feed and did not measure their absorption or turnover in seahorse tissues, a direct quantitative relationship between feed sterol oxidation and tissue sterol profiles cannot be established. In addition, only two storage time points were tested, which does not define a temporal trajectory or a threshold at which lipid quality deteriorates. The 60 °C drying procedure used during sample preparation, although applied uniformly to all samples, may also have influenced the absolute abundance of oxidation-sensitive sterols and therefore complicates their interpretation as biomarkers of feed storage. Finally, because each species-by-storage treatment had only one tank, these sterol patterns should be considered exploratory and may be affected by tank-level pseudoreplication and confounding by feed collection time and rearing temperature.

These findings may have implications for medicinal quality assessment. Sterol lipids serve as precursor substrates for sex hormone synthesis and are considered relevant to the traditional pharmacological action of seahorses in “warming the kidney and strengthening yang” [34]. A decline in oxidized sterol derivatives could therefore be relevant to the medicinal quality of farmed seahorses. However, no pharmacological activity, bioavailability, or functional health-related endpoint was measured in this study. Consequently, changes in lipid composition should not be automatically interpreted as changes in medicinal efficacy or quality. The link between sterol profiles and medicinal quality remains inferential and requires direct pharmacological validation.

Despite these limitations, the cross-species consistency of the sterol decline makes these compounds promising candidates for further investigation. Because sterol changes were less divergent among species than fatty acid or phospholipid changes, they may be less confounded by species-specific metabolic compensation. However, before sterol monitoring can be recommended as a practical quality control tool, future studies must directly quantify oxysterols in stored feed, establish their dose–response relationship with tissue levels, evaluate absorption and metabolism, and validate the association with functional or pharmacological endpoints. With such validation, sterol lipid profiles—particularly oxidized cholesterol derivatives—could become a useful component of feed quality assessment in seahorse aquaculture.

4.3. Acylcarnitines Profiles: Possible Indications of Altered Lipid Utilization

Among the six lipid classes examined, acylcarnitines showed the smallest magnitude of change in the present dataset. The global PCA showed moderate species separation, whereas species-specific PCA revealed weak or no separation between storage groups across the three seahorse species. Quantitative analysis of the pooled samples showed that in H. trimaculatus, Acyl-Carnitine (4:0) OH and Acyl-Carnitine (5:0) decreased in the 10-month group, while the other four acylcarnitines remained unchanged. In H. erectus, none of the six acylcarnitines differed significantly between storage groups. In H. abdominalis, Acyl-Carnitine (3:0), Acyl-Carnitine (4:0), and Acyl-Carnitine (4:0) OH decreased, while the remaining three showed no significant change. These patterns may indicate differences in how the three seahorse species regulate lipid utilization under the two feed treatments, but they should be interpreted as exploratory.

Acylcarnitines are key intermediates of the carnitine shuttle system, which transports acyl groups into mitochondria for β-oxidation, and their profiles are commonly used as surrogate biomarkers of mitochondrial fatty acid oxidation flux [35]. However, acylcarnitine profiles are indirect indicators and do not provide direct evidence of mitochondrial fatty acid oxidation capacity without functional measurements of β-oxidation or mitochondrial activity. Therefore, the relative stability of acylcarnitine profiles in H. erectus—where no significant changes were detected—may be consistent with a possible preservation of core energy metabolism under the present conditions, but it cannot be interpreted as proof that mitochondrial function was maintained. Similarly, the decreases in several acylcarnitine species in H. trimaculatus and H. abdominalis may suggest that these two species were less able to maintain acylcarnitine homeostasis under prolonged feed storage, but this interpretation remains speculative.

The role of dietary lipid composition in energy metabolism is supported by recent work showing that dietary oleic acid can alter lipid deposition and energy utilization in Atlantic salmon, demonstrating that dietary fatty acid composition can influence pathways related to lipid storage and oxidation [30]. This is consistent with the possibility that the aged mysid feed may have influenced lipid utilization in the seahorses. However, because the two feed treatments were collected at different times and H. abdominalis was reared at a different temperature, the observed species-associated acylcarnitine patterns cannot be attributed solely to species differences or to storage duration. In addition, the present study did not measure carnitine palmitoyltransferase activity, β-oxidation flux, mitochondrial respiration, or expression of genes involved in fatty acid oxidation. Therefore, the acylcarnitine changes reported here should be regarded as possible indications of altered lipid utilization rather than as direct evidence of differences in energy metabolic resilience. Other general limitations are detailed in Section 4.5.

Future studies should combine targeted acylcarnitine profiling with functional measurements of mitochondrial fatty acid oxidation, enzyme activity, and gene expression in replicated tank designs with common thermal regimes and multiple storage time points. Such an approach would help determine whether the acylcarnitine patterns observed here reflect meaningful species-specific differences in lipid utilization or are primarily driven by confounding husbandry and feed-batch factors. At present, the acylcarnitine data provide exploratory evidence that the three seahorse species may differ in their metabolic responses to the two feed treatments, but they do not establish a causal link between frozen feed storage and mitochondrial fatty acid oxidation capacity.

4.4. Implications for Future Quality Monitoring and Research

The present findings may have implications for the management of frozen mysid feed in seahorse aquaculture, but they should be regarded as preliminary. Currently, no unified frozen-storage duration standard exists for mysid feed in China, and farming enterprises often rely on empirical practices with considerable batch-to-batch variability. Our exploratory results suggest that a one-size-fits-all storage limit may not be optimal for all seahorse species. However, species-specific storage limits cannot be established from the present design because each treatment had only one tank, the two feeds were collected at different times, and H. abdominalis was reared at a different temperature from the other two species. Therefore, the following considerations are intended to guide future research rather than to provide operational recommendations.

Under the present experimental conditions, H. abdominalis exhibited the broadest lipid changes, including declines in sterols, PUFAs, phospholipids, and several acylcarnitines. This broad attenuation may be related to its distinct lower-temperature rearing environment (16–17 °C versus 24–25 °C for the other two species), which could affect membrane lipid composition and oxidative stability [36,37,38]. Because temperature effects cannot be separated from species effects in this study, the greater sensitivity of H. abdominalis should be considered preliminary and requires validation under a common thermal regime. In contrast, H. erectus showed relatively stable energy-related and PUFA profiles, but its sterol precursors and most phospholipids still decreased, indicating that feed-associated changes were not absent in this species. H. trimaculatus occupied an intermediate position. These species-associated patterns are interesting, but they should not yet be translated into species-specific storage guidelines.

Quality monitoring should not rely solely on total lipid content or fatty acid profiles, because these measures can be confounded by species-specific metabolic compensation and by differences in feed intake, absorption, and tissue turnover. The cross-species consistency of the sterol decline observed here suggests that oxidized sterol derivatives may be promising candidate indicators for further evaluation. However, before sterol monitoring can be recommended as a practical quality control tool, several steps are needed. First, oxysterols should be quantified directly in stored feed at multiple time points to establish their formation and degradation kinetics. Second, the relationship between feed oxysterol levels and tissue sterol profiles should be validated through controlled feeding trials with replicated tanks. Third, the absorption, metabolism, and deposition of these compounds in seahorse tissues should be characterized. Fourth, the association between sterol changes and functional or pharmacological endpoints should be tested. Only after such validation could sterol lipid profiles—particularly oxidized cholesterol derivatives—be considered a reliable component of feed quality assessment.

Future studies should also incorporate replicated tanks per treatment, common thermal regimes across species, multiple storage time points, and different storage temperatures. A time-series design would help define the trajectory of lipid deterioration and identify potential thresholds. Combining targeted lipidomics with functional measurements of mitochondrial fatty acid oxidation, enzyme activity, gene expression, and pharmacological activity would further clarify whether the observed lipidomic changes have biological or medicinal relevance. In addition, the 60 °C drying procedure used for sample preparation, although standard for seahorse medicinal material processing, may influence oxidation-sensitive lipids; future work should evaluate alternative drying methods or include appropriate controls. Until such data are available, the present results should be interpreted as exploratory evidence that frozen feed storage is associated with species-associated lipid changes, rather than as a basis for establishing storage standards or quality control protocols.

Other general limitations of this study are detailed in Section 4.5.

4.5. Limitations and Future Directions

Several limitations of this study should be acknowledged. First, each species × storage-duration treatment was housed in a single rearing tank, and the three pooled samples per group were derived from fish within the same tank. Therefore, the experimental unit was the tank, not the pooled sample, and the statistical comparisons may be affected by tank-level pseudoreplication. The ANOVA results should be interpreted as exploratory and descriptive rather than as formal tests of storage-duration effects at the experimental-unit level. Second, the 2-month and 10-month feeds were collected at different times from the same original brackish pond system. Therefore, storage duration was confounded with collection time and possible batch-to-batch variation in initial mysid composition. Consequently, the observed lipidomic differences cannot be attributed solely to frozen storage duration. Third, H. abdominalis was reared at 16–17 °C, whereas H. erectus and H. trimaculatus were reared at 24–25 °C. Because temperature can strongly influence lipid metabolism, membrane lipid remodeling, fatty acid utilization, and energy metabolism, interspecific differences cannot be fully separated from temperature effects. In particular, the conclusion that H. abdominalis is the most sensitive species should be considered preliminary because this species was maintained under a different thermal regime. Fourth, only two storage time points (approximately 2 and 10 months) at a single storage temperature (−20 °C) were examined. This design does not establish a temporal trajectory or a threshold at which lipid quality deteriorates. A time-series design incorporating multiple intermediate time points and different storage temperatures would provide a more comprehensive picture. Fifth, the 2-month frozen feed was not a fresh control. According to our previous study [8], major losses of free fatty acids occurred between fresh and 2 months, and TBARS did not increase significantly until 10 months. Therefore, the present 2-month vs. 10-month design captures late-stage storage changes and does not define the early deterioration trajectory. Sixth, no functional measurements of mitochondrial β-oxidation or mitochondrial activity were performed. Acylcarnitine profiles are only indirect indicators and should not be interpreted as direct evidence of mitochondrial fatty acid oxidation capacity. In addition, growth performance (survival and weight gain) was not systematically assessed in this study because the primary objective was lipidomic profiling. Future studies should integrate growth, survival, and reproductive endpoints to link feed lipid changes to seahorse performance. Seventh, no pharmacological activity, bioavailability, or functional health-related endpoint was measured. Therefore, changes in lipid composition should not be automatically interpreted as changes in medicinal efficacy or quality. Eighth, the 60 °C drying procedure used for sample preparation, although standard for seahorse medicinal material processing and applied uniformly to all samples, may have introduced additional oxidation of labile lipids. This could influence the absolute abundance of oxidation-sensitive compounds and complicate their interpretation as biomarkers of feed storage. Ninth, the proposed sterol markers were not directly validated against feed oxidation indices or tissue absorption and metabolism. In addition, oxysterols were not directly measured in the feed, so the proposed sterol indicators require direct validation against feed oxysterol contents. A direct quantitative relationship between changes in the stored feed and subsequent changes in seahorse tissues was not established. Finally, the PCA and heatmap analyses are useful for describing overall lipidomic patterns, but group separation should not be interpreted as independent evidence of treatment effects in the presence of limited biological replication and tank-level pseudoreplication. Future studies should use replicated tanks per treatment, common thermal regimes, multiple storage time points, direct feed–tissue validation, and functional assays of lipid metabolism and pharmacological activity to confirm and extend the present findings.

5. Conclusions

This study provides preliminary evidence that frozen storage duration of mysid feed is associated with species-associated, rather than uniform, changes in the lipid profiles of farmed seahorses. Oxidized sterol derivatives showed the most consistent decline across the three species and may be candidate indicators for further evaluation. However, because each treatment had only one tank, the two feeds were collected at different times, and H. abdominalis was reared at a different temperature, species-specific storage limits cannot be established. These findings should be considered exploratory and hypothesis-generating. Future studies with replicated tanks, common thermal regimes, multiple storage time points, and direct feed–tissue validation are needed to determine whether sterol monitoring can serve as a practical quality control tool.