Search

Abstract The effect of dried porcine soluble (DPS) supplementation on replacement of dietary fish meal with soy protein concentrate (SPC) for large yellow croaker, Larimichthys crocea, was assessed in a feeding trial conducted in inshore net pens. Five isoproteic and isolipidic diets were tested. A basal diet (BD) contained 400 g/kg fish meal, and SPC was used as an alternative ingredient to replace 40% (F24) and 60% (F16) of the fish meal in BD at an equal protein basis. DPS was added at 5 g/kg in F24 (FD24) and F16 (FD16). Fish weighing 33.6 ± 7.4 g were fed with the test diets in 1.5 × 1.5 × 2 m net pens for 9 weeks. The fish meal replacement level significantly affected weight gain, feed conversion ratio, phosphorus retention efficiency, body phosphorus content, and ratio of fish meal consumption to fish production (RCP), while the DPS supplementation level significantly affected weight gain. The weight gain of fish fed FD24 was higher than that of fish fed F16, while no significant difference was found in weight gain between fish fed BD and fish fed FD24, F24, FD16 and F16. The feed cost, RCP and phosphorus waste of fish fed BD were higher than those of fish fed FD24, F24, FD16 and F16. Fish fed FD24 exhibited the best production performance, and the production performance of fish fed F24 and FD16 was close to that of fish fed BD. The present study indicates that the fish meal level in large yellow croaker diet could be lowered to 240 g/kg with SPC alone as a fish meal substitute, and could be further reduced to 160 g/kg by supplementation of 5 g/kg DPS. Therefore, DPS supplementation can benefit to increase fish meal replacement with SPC in large yellow croaker diet.

Abstract At present, large yellow croaker, Larimichthys crocea, a carnivorous fish ranked fifth in aquaculture production among marine finfishes in the world, still relies on raw fish diet because formulated feed alone is not able to sustain rapid growth of the fish. The present study evaluated growth, feed utilization efficiency, body composition and antioxidant capacity of large yellow croaker fed diets containing four protein (430, 460, 490 and 520 g/kg) and three lipid (60, 90 and 120 g/kg) levels. Fish weighing 11.4 ± 2.8 g were fed with the test diets in net pens for 8 weeks. Final body weight and nitrogen retention efficiency (NRE) were dependent on dietary protein and lipid levels and their interaction. Weight gain, body protein content, serum catalase (CAT) activity and liver glutathione peroxidase (GSH‐Px) activity were dependent on dietary protein and lipid levels. Phosphorus retention efficiency was dependent on dietary protein level and interaction between protein and lipid levels. Hepatosomatic index, body lipid and energy contents, activities of superoxide dismutase (SOD) and GSH‐Px in serum, and CAT activity and malondialdehyde content in liver were dependent on dietary lipid level. The final body weight and weight gain of fish fed at 490 g/kg protein level did not significantly differ from those fed at 520 g/kg protein level. The final body weight, weight gain and NRE of fish fed at 90 g/kg lipid level were higher than those fed at 60 g/kg lipid level, but did not significantly differ from those fed at 120 g/kg lipid level. The maximum weight gain occurred in fish fed the diets either containing 520 g/kg protein or 90 g/kg lipid. The body lipid content, as well as SOD and GSH‐Px activities in serum, were higher in fish fed the diets with 90 and 120 g/kg lipid compared to fish fed the diets with 60 g/kg lipid. The present study reveals that the optimal dietary protein and lipid levels for large yellow croaker reared in net pens were 490–520 g/kg and 90 g/kg, respectively.

Recent research suggests that in vitro neural networks created from dissociated neurons may be used for computing and performing machine learning tasks. To develop a better artificial intelligent system, a hybrid bio-silicon computer is worth exploring, but its performance is still inferior to that of a silicon-based computer. One reason may be that a living neural network has many intrinsic properties, such as random network connectivity, high network sparsity, and large neural and synaptic variability. These properties may lead to new design considerations, and existing algorithms need to be adjusted for living neural network implementation. This work investigates the impact of neural variations and random connections on inference with learning algorithms. A two-layer hybrid bio-silicon platform is constructed and a five-step design method is proposed for the fast development of living neural network algorithms. Neural variations and dynamics are verified by fitting model parameters with biological experimental results. Random connections are generated under different connection probabilities to vary network sparsity. A multi-layer perceptron algorithm is tested with biological constraints on the MNIST dataset. The results show that a reasonable inference accuracy can be achieved despite the presence of neural variations and random network connections. A new adaptive pre-processing technique is proposed to ensure good learning accuracy with different living neural network sparsity.