In this study, the effects of different stocking densities (1–10 individuals in 144 L; 6.9–69.4 fish m-3, 0.09–0.92 kg m-3) on swimming activity, vertical positioning and inter-individual spatial relationships in juvenile tilapia (Oreochromis niloticus) were examined by automated object detection and multi-object tracking based on the YOLO deep learning algorithm. Swimming distance, speed, mean depth, layer use and nearest neighbour distance were obtained from three replicates and evaluated by repeated measures analysis of variance with replicate as a block effect, Duncan’s multiple range test and linear regression. Swimming distance rose from 60.5 ± 23.4 m with one fish to 99.7 ± 33.2 m with two and then plateaued (127.6 ± 11.1 m with ten; F9,18 = 4.46; p = 0.0034), and swimming speed behaved likewise (3.86 ± 1.41 to 8.52 ± 0.73 m min⁻¹; F9,18 = 4.87; p = 0.0021); in both, only the single-fish level differed from the groups. Mean depth fell from 20.4 ± 5.4 to 12.5 ± 3.5 cm (F9,18 = 3.70; p = 0.0088) and upper layer use rose from 39.9 ± 18.5% to 65.9 ± 15.1% (F9,18 = 3.83; p = 0.0074), both levelling off at about seven individuals. Nearest neighbour distance fell from 37.27 ± 4.39 to 11.45 ± 0.55 cm (F8,16 = 56.44; p < 0.0001; R² = 0.846); compared with a random spatial arrangement it matched expectation up to four fish but fell to 0.66–0.69 of it from seven onwards (p ≤ 0.014), so the decrease reflects active aggregation and not geometric crowding alone. Over short observation windows and at biomass densities far below production scale, an increasing number of individuals in a fixed volume thus alters both individual movement and social spatial organisation in yavru tilapia, inter-individual distance being the most responsive measure.
The family Callionymidae comprises a taxonomically challenging group of benthic marine fishes, mainly due to pronounced sexual dimorphism, ontogenetic variation, and the limited reliability of some diagnostic morphological characters. In the present study, six dragonet species occurring along the Turkish coasts, Callionymus filamentosus, C. lyra, C. pusillus, C. risso, Synchiropus phaeton, and S. sechellensis, were examined using morphological and molecular approaches. Specimens were collected from the Black Sea, Sea of Marmara, Aegean Sea, and Levantine Basin. Morphometric and meristic characters, coloration patterns, sexual dimorphism, and preopercular spine morphology were evaluated. In order to complete morphological identification and provide an independent molecular assessment of species differentiation, mitochondrial COI sequences were analysed using DNA barcoding and phylogenetic approaches. A Principal Component Analysis (PCA) based on 14 morphometric ratios from all 35 individual specimens was used to evaluate morphological variation among the examined taxa. The first two principal components accounted for 55.0% of the total variation, with the two Synchiropus species showing comparatively distinct distributions in morphometric space, whereas greater overlap was observed among the four Callionymus species. COI sequences were obtained from 22 individuals. Intraspecific genetic divergence ranged from 0 to 1.1%, whereas interspecific distances ranged from 11.9% to 24.7%. Maximum Likelihood analysis recovered distinct clusters corresponding to the morphologically identified species, while the position of C. filamentosus sequences indicated that phylogenetic relationships within the examined callionymids warrant further investigation. Overall, the combined morphological and molecular results support the complementary use of traditional diagnostic characters and DNA barcoding for the identification of Mediterranean dragonets. The present study also contributes additional COI barcode data for poorly represented Mediterranean Callionymidae and provides baseline information for future systematic and phylogenetic studies.
This study investigates the impact of aquaculture production on economic growth across the top twenty aquaculture-producing countries for the 2001–2024 period. Accounting for cross-sectional dependence and slope heterogeneity, the study employs second-generation CIPS panel unit root tests, panel VAR Granger causality analysis, and fixed-effects estimation with Driscoll–Kraay robust standard errors. A dynamic panel specification is also employed to assess the robustness of the findings. A dynamic panel specification is also applied for robustness. The empirical findings confirm a strong Granger causality running from aquaculture to economic growth. Panel estimates indicate that a 1% increase in aquaculture output is associated with about 0.14% higher per-capita GDP, while CPI is also positively and significantly associated with GDP. These associations remain robust in the dynamic specification. Overall, the findings highlight aquaculture's vital macroeconomic role and emphasize the need to integrate circular economy principles into sustainable growth strategies.
This study characterized the fatty acid profiles and concentrations of total arsenic (As), total mercury (Hg), cadmium (Cd), and lead (Pb) in visceral by-products from ten commercially important marine fish species representing eleven sample groups, including wild and cultured European seabass. The highest eicosapentaenoic acid (EPA) proportions were observed in Sardina pilchardus (8.75%), Scomber japonicus (8.18%), and Engraulis encrasicolus (8.09%), whereas docosahexaenoic acid (DHA) was highest in S. japonicus (24.36%), E. encrasicolus (20.43%), Scorpaena porcus (20.38%), and S. pilchardus (20.24%). Element concentrations ranged from 0.802 to 7.852 mg/kg for total As, 0.006 to 0.127 mg/kg for total Hg, 0.002 to 0.313 mg/kg for Cd, and 0.007 to 0.247 mg/kg for Pb. For each fish group, the available visceral material was analyzed as a bulk mixed-viscera sample, and analytical determinations were conducted in triplicate; these triplicates represent analytical rather than independent biological replicates. Several visceral samples contained high relative proportions of EPA and DHA within the total fatty acid methyl ester (FAME) profile, indicating compositional potential for further lipid-recovery studies. However, total lipid yield, contaminant speciation, batch-to-batch variability, and matrix-specific safety considerations require further evaluation before food or nutraceutical applications can be considered.
TroPhish investigated how well freshwater fish feeding data could be integrated with Fish-base's species, ecology, diet, and nutrient component tables. It also explored how open datasets downloaded from Fish-base could be used to model fish feeding groups, feeding diversity, and trophic levels. For the analyses, 54,750 raw feeding records and 4,571 sample-level specified "fish ID" units were used as the basis. Fish-base matching was successfully completed for 88.6% of samples and 84.2% of species. This research led to the addition of "SpecCode" and "FamCode" to 4,051 samples, feeding-based trophic levels to 3,280 samples, and nutrient data to 3,440 samples for analysis. In the "Species Group K-Fold" validation, the Random Forest model was applied and enhanced with Fish-base datasets. In the feeding group classification, the trophic identification scenario achieved 0.515% accuracy, 0.397% balanced accuracy, and a 0.390% macro F1 score. In the training data, the model's generalizability to uncharacterized families was poor. The low performance in the "Family Code Group K-Fold" validation of the same scenario is evidenced by this. While the relational power of the feeding breadth regression was found to be limited (maximum R² = 0.092), the trophic level regression based on fish-based data yielded realistic findings in the taxon-trait scenario (R² = 0.403). The results suggest that the misleading signal related to fish feeding habitat is amplified by the inclusion of TroPhish and fish- based data. Model outputs are not an automated feeding classification system that produces definitive results. They should be considered as descriptive findings generated from research-driven modeling and publicly available data.
