Zhejiang Marine Fisheries Research Institute

Introduction Artificial reefs (ARs) have been widely used to provide ecological and fisheries benefits. Quantitative evaluation of how the numbers and spatial layout of ARs affect fish responses can provide critical insights for management. Methods We used survey data from an area in the Bohai Strait (China) that has ARs deployed and developed species distribution models (SDMs) predicting the occurrence probabilities of three commercially important species ( Charybdis japonica , Sebastes schlegelii , and Hexagrammos otakii ) and one undesirable focal species ( Asterias amurensis and Asterina pectinifera combined). Three versions of SDMs were developed: a Best-fit model based on survey data, and two modified versions (Intermediate and Extreme), that incorporated increasing levels of AR-to-AR interactions, reflecting competitive and connectivity effects. A 24×24 grid with 50-m cells was populated with bottom habitat type (mud, gravel, rubble, boulder, or AR) and key environmental variables affecting occurrence probabilities. Using simulations of 10,000 randomly-generated spatial layouts of added ARs (bottom type switched to AR), we compared predicted occurrence probabilities from the Best-fit, Intermediate, and Extreme SDMs when 5, 10, 25, and 50 ARs were added to the existing ARs. Performance was evaluated using the predicted species-specific occurrence probabilities, with occurrence of the undesirable species treated inversely (i.e., lower occurrence probability indicated higher performance). Results Increasing AR numbers increased the percentage of grid cells supporting good habitat, but saturation and interference effects caused similar values for 25-50 ARs for several species, while reducing variation across spatial layouts. The three desirable species showed similar patterns with the representative layouts categorized into good and bad performing: increasing spread of good ARs on the grid with number of ARs deployed, shift of good ARs from upper to lower triangle, and an increasingly bad-performing central area with 50 ARs. The spatial overlap between high-performing cells for desirable species and elevated occurrence of undesirable Sea star illustrated an inherent trade-off that should be considered in management objectives. Discussion We conclude with a discussion of the need for AR layout-specific evaluation, consideration of AR interaction strength, broader applicability to other systems, potential pathways to expand the analysis (e.g., add hydrodynamic models), and caveats and recommendations for further development.

ABSTRACT Although the learning‐based surrogate model could achieve fast prediction, the model uncertainty in data and predictions remains a significant challenge. To alleviate the uncertainty problems for the surrogate model, a learning‐based uncertainty analysis method for improving the stability of the inverse model of RF devices is proposed. The inherent uncertainty in the inverse model is analyzed by leveraging ensemble adversarial learning, enabling the prediction of confidence intervals for the inverse surrogate model output without changing the original neural network structure. Subsequently, the differential evolution (DE) algorithm is employed to optimize the geometric parameters of the inverse model output within the predicted interval. To validate feasibility and effectiveness of the method, RF hairpin filter is used for the design example via the neural networks. RF filters with multiple center frequencies and bandwidths are simulated, fabricated and measured. Both simulated and measured results demonstrate a notable enhancement, with a 37.22% predicted accuracy improvement in the degree of fitting between the optimized response and the label of the input network, compared to the circuit response (S‐parameter) pre‐optimization. The proposed method exhibits much better performance in terms of the predicted accuracy and computational time compared to direct optimization without inverse model. The learning‐based method could also be applicable to the other RF devices, affirming the practical applicability and robustness of the approach.

Mandarin fish (Siniperca chuatsi) is a highly valued freshwater species in China, owing to its high-quality meat and economic importance. This study comparatively evaluated the effects of an industrialized recirculating aquaculture system (RAS) and traditional pond aquaculture system (TPAS) on the muscle quality and further explored the role of gut microbiota in muscle quality regulation. Our results showed that the RAS resulted in superior textural properties, with meat that was significantly more tender and elastic. The RAS also promoted higher muscle protein and reduced lipid levels. Notably, the RAS elevated sweet-tasting amino acids (Gly and Pro) while suppressing bitter amino acids (His). Electronic nose and GC-iMS analyses revealed distinct flavor compound profiles between the two systems, and the RAS enriched desirable volatiles (esters and alcohols) while suppressing aldehydes (e.g., nonanal and heptanal) associated with off-flavors. Gut microbiota profiling indicated higher diversity and enriched beneficial genera (e.g., Cetobacterium, Lactobacillus) in RAS-treated fish. We found that the Cetobacterium in the RAS group showed a significant positive correlation with sweet amino acids and pleasant flavor substances (such as esters, alcohols), while exhibiting a negative correlation with undesirable flavor precursors (such as certain aldehydes). This finding contributes to the sustainable and high-efficiency advancement of intensive Siniperca chuatsi aquaculture.

Larimichthys crocea, commonly known as large yellow croaker, is a marine species of significant commercial value in East Asia. However, overfishing and long-term artificial breeding have raised concerns about the genetic diversity and adaptive capacity of farmed populations. Understanding the genetic differentiation and environmental adaptation between wild and cultured populations is crucial for conservation efforts and sustainable aquaculture development. Whole-genome resequencing of 195 individuals revealed: (1) Significant genetic divergence between wild and farmed populations (764,656 SNPs), with cultured stocks showing 18-22% lower heterozygosity; (2) 153 selective sweep regions containing 739 candidate genes, including hsp40 (thermal adaptation) and ndufs6 (energy metabolism); (3) Genome-environment association identifying 9,265 loci linked to salinity, temperature, and dissolved oxygen (RDA adj. R²=0.036, P < 0.001); (4) Strong isolation-by-environment patterns (Mantel R²=0.89, P = 0.042) outweighing geographical distance effects. Our results demonstrate that environmental factors rather than geography drive genetic differentiation in L. crocea, with aquaculture populations exhibiting signatures of both artificial selection and reduced adaptive potential. These findings provide actionable insights for conservation management and marker-assisted breeding to maintain genetic diversity in cultured stocks.