Search

ABSTRACT Estuarine habitats serve as critical feeding and nursery grounds for many aquatic species and support fisheries. However, monitoring these complex ecosystems using conventional trawling methods is destructive, costly, and labor‐intensive. This study compared trawling and a multi‐marker environmental DNA (eDNA) metabarcoding approach to monitor marine vertebrate and crustacean communities in an estuarine environment in subtropical Hong Kong. We analyzed 16 bottom trawl samples and the eDNA from 32 two‐liter water samples using primer sets specific to fishes and mammals (MiFish‐U, 12S‐V5, and Berry‐Fish) and decapod crustaceans (MiDeca). We found that the eDNA approach detected more pelagic and demersal fishes (237 vs. 106 in trawling) and elasmobranchs (6 vs. 3) than trawling. The eDNA approach was also more effective than trawling in detecting threatened vertebrates (16 vs. 4), including the Indo‐Pacific Finless Porpoise and the critically endangered Large Yellow Croaker. Among the detected fish at species level, 70 species were detected by both approaches, 32 species were detected by trawling only, and 142 species were detected by the eDNA approach only. Regarding crustaceans, the eDNA approach detected slightly fewer decapods (61 vs. 77) and stomatopods (5 vs. 8) than trawl surveys. However, the eDNA approach could be enhanced through the development of suitable decapod‐specific primers and the expansion of the local reference database. In addition, multivariate analyses of the eDNA data revealed spatial patterns of fish and crustacean assemblages that might be associated with sediment loading, oxygen, and nutrient levels. Furthermore, there was a positive correlation between eDNA read counts and trawl catch, but their correlation coefficient was low. We conclude that eDNA metabarcoding can provide high‐resolution detection of species, composition, and unravel community–environment relationships in estuarine ecosystems. Overall, integrating the non‐destructive eDNA approach can complement the conventional trawling method for better‐informed sustainable fishery management and conservation.

Due to the increasing implementation of smart cities, crowd flow prediction has become a crucial aspect in various fields such as transportation management and public risk assessment. However, accurately predicting crowd flow can be influenced by various complex factors, including local and long-range spatial dependencies, as well as long-term temporal dependencies. Many works have attempted to model long-range spatial dependence relying on stacked convolutional layers, failing to directly consider it from a global perspective. In this article, we propose a novel recurrent convolutional network based on grid correlation, named RCNGC. In this model, the correlation between different grids based on crowd flow pattern is identified first. Then, a deep network following an encoding-prediction framework is proposed. Specifically, a novel GC-ConvLSTM module based on the grid correlation is utilized to capture long-range dependence in both temporal and spatial dimensions in encoding phase while ConvLSTM is leveraged in prediction phase. We conduct extensive experiments on a dataset in Futian, China. The results demonstrate that the proposed RCNGC is superior to baseline methods in multi-step prediction, indicating that identifying the correlation between grids preliminarily can improve the precision of prediction.

In order to solve the problem of high labor intensity of a traditional pontoon lotus root machine with manual pushing operation, a riding pontoon digging lotus root machine was proposed in this paper. The working principle of the digging machine was introduced, the key components were designed, and the buoyancy of the pontoon was calculated and analyzed. The performance of the digging machine was tested in a lotus root pond, the lotus root digging rate was 76%, and hydraulic flushing generated by the digging machine can disperse the soil within 30 cm under water. In addition, the best flushing effect was achieved when the number of working nozzles was controlled to three. Finally, according to the test results, the optimization measures of the hydraulic lotus root digging machine were proposed, and the CFD-DEM coupled simulation model was used to simulate and analyze the soil disturbance effect of hydraulic flushing upward from the bottom of the lotus root. The results showed the relationship between nozzle diameter, angle and hydraulic flushing depth. When the nozzle angle was fixed, the soil disturbance effect decreased with increasing flushing depth and increased with decreasing nozzle diameter. When the nozzle depth was fixed, both the diameter and the angle had a significant effect on the soil disturbance effect. When the nozzle diameter was fixed, the interaction of angle and depth had no significant effect on the soil disturbance effect.