The experiment was carried out in nine mud aquaculture ponds with length, width, and depth of 70, 30, and 1.5 m, respectively, in Dongtan Aquaculture Base at Shanghai Daohong Aquaculture Technology Co., Ltd (31.62° N, 121.40° E) from July to November 2018. To begin the experiment, 6,000 juvenile crabs weighing between 1 and 2g were placed in each aquaculture pond. Alternanthera philoxeroides was planted in the center of the pond, and the surrounding area was a “large rectangle embedded with a small rectangle.” To provide shelter and clean water, A. philoxeroides was transplanted evenly in the circular groove of the pond and accounted for 75% of the experimental pond, and the area of hollow algal species in each pond was strictly controlled, with these species being removed when exceeding this area or supplemented in case of being insufficient. The outside of the pond was surrounded by a 40 cm-high fence to prevent crabs from escaping. The crabs were fed under three different feeding modes: traditional (consisting of wheat, bran, and soybean meal), formulated (Zhejiang Aohua Feed Co., Ltd., Jiaxing, China), and mixed (1:1 mixture of traditional and formulated feeds) feeds. The conventional biochemical composition of the different feeding modes is shown in Table 1. Each feeding mode was applied in three replicates. During the breeding period, the feeding ratio of each feeding mode was consistent, and the amount fed was approximately 1–3% of the total body weight of the crabs, at approximately 16:00 h every day.

A Hach HQd portable water quality analyser (HQd, Hach Water Analysis Instrument Co., Ltd., Shanghai, China) was used to measure the temperature, dissolved oxygen, and pH of the water at 9–10 am in each pond. Water quality changes in the breeding process are shown in Figure 1. During the experiment, 1–2 g of formulated feed was collected, freeze-dried, crushed (particle size < 0.1 mm), and then cryopreserved at −40°C for subsequent analysis. The water sample collected using a water collector was passed through a 100 μm sieve, and the water samples removed from zooplankton were filtered using a glass fiber membrane (GF/F Whatman, Little Chalfont, United Kingdom; 47 mm diameter, 0.7 μm aperture) to obtain POM, which was wrapped in aluminum foil and frozen for subsequent analysis. At the end of the experiment, a fresh A. philoxeroides sample was collected from the crab breeding pond and washed slowly and repeatedly with double distilled water to remove soil on its surface and then dried, freeze-dried, and crushed (particle size < 0.1 mm) before storing it at −20°C for subsequent stable isotope analysis. In October, three male and three female crabs with sound appendages and good vitality in each pond were collected and selected for the experiment, and three samples were combined into one sample according to gender. The collected crabs were anesthetized on ice, and the surface was disinfected with 70% ethanol. The specimens were then dissected, and their stomach contents were extracted and transferred to 2.0 mL labeled cryopreservation tubes. The tubes were immersed into a liquid nitrogen tank and then stored in a refrigerator at −80°C for subsequent 18S rDNA analysis. The three E. sinensis specimens collected in each pond in November were anesthetized on ice, and the surface was disinfected with 70% ethanol. The shell was separated from the body along the side of the shell, and the hepatopancreas, head breastplate, appendages, and gills were removed to get body. Then, the muscle of the crab was freeze-dried, crushed (particle size < 0.1 mm), and cryopreserved at −40°C for subsequent stable isotope analysis.

Twenty male and twenty female crabs were randomly taken from each pond in the middle of each month to understand their growth during the experiment. The crabs were dried with a dry towel and accurately weighed on an electronic balance (accurate to 0.01g). The weight gain rate (WGR) of each juvenile E. sinensis was calculated as follows:

where, Wt and Wn are the average body weight (g) of crabs in months t and n, respectively.

Genomic DNA from the stomach contents of E. sinensis was extracted using the E.Z.N.A. SOIL DNA Kit (Omega Bio-Tek, Norcross, GA, United States). The diluted genomic DNA was used as a template to amplify the target gene with specific primers. The primer sequences were TAREF (5′-CCAGCASCYGCGGTAATTCC-3′) and TAREF (5′-AC TTTCGTTCTTGATYRA-3′). The ABI Geneamp^® 9700 PCR (Applied Biosystems, Foster City, CA, United States) instrument was used for PCR using TransGen AP221-02: TransStart FastPFU DNA polymerase. Using high-efficiency and high-fidelity enzymes for PCR can ensure amplification efficiency and accuracy (Bokulich and Mills, 2013). The PCR amplification was carried out in a total volume of 20 μL containing 4 μL of 5× FastPFU Buffer, 0.8 μL of each primer (5 μmol/L), 0.4 μL of FastPFU Polymerase, 10 ng of template DNA, and 2 μL of 2.5 mmol/L dNTPs. The following cycling program was used: initial denaturation at 95°C for 2 min; 25 cycles of denaturation at 95°C for 30 s, annealing at 55°C for 30 s, and extension at 72°C for 30 s; and a final extension at 72°C for 5 min. The PCR products were detected using 2% agarose gel electrophoresis and quantified using the Quantifluor^TM-ST Blue Fluorescence Quantitative System (Promega, Madison, WI, United States) according to the preliminary quantitative results of electrophoresis. The PCR products were mixed and purified using 2% agarose gel electrophoresis in 1× TAE buffer, followed by gel extraction. PCR products with the main band size between 300 and 400 bp were selected, recovered by gel-elution using the AxyPrep DNA Gel Recovery Kit (Axygen Biosciences, Union City, CA, United States), and eluted with Tris-HCl.

The purified PCR products were quantified on a Qubit 3.0 fluorometer (Life Technologies, Invitrogen, Carlsbad, CA, United States), and all 24 barcoded amplicons were evenly mixed and used to construct an Illumina antithesis library, according to the preparation procedure of the Illumina genomic DNA library. Sequencing was performed on the Illumina MiSeq platform (Shanghai Biozeron Co., Ltd., Shanghai, China) (2 × 250).

The crab samples were small and comprised a mix of three individual samples. After freeze-drying, the processed samples were ground, and screened using an 80-mesh sieve. The stable isotopes of the samples were determined at the Laboratory of Ingestion Ecology, Shanghai Ocean University. The samples were covered with aluminum foil, and the muscle of E. sinensis, POM, A. philoxeroides, and feed samples weighed 1.00, 5.00, 2.00, and 1.50 mg, respectively. A stable isotope mass spectrometer (ISOPRIME100, Isoprime Corporation, Cheadle, United Kingdom) was used to measure the stable isotope (δ¹³C and δ¹⁵N) values of crab samples, international standard material (PBD), and purified atmospheric N₂ as reference standards; δ¹³C and δ¹⁵N were calculated using the following formula:

where, X is ¹³C or ¹⁵N; R_sample is the sample’s isotope ratio of ¹³C/¹²C or ¹⁵N/¹⁴N; and R_standard is the isotope ratio of the standard (Caut et al., 2009). To ensure the accuracy of the experimental results and the stability of the instrument, one standard sample was inserted every ten samples for calibration. The analysis accuracies of the δ¹³C and δ¹⁵N values of the samples were 0.05 and 0.06‰, respectively.

Sequences were classified as operational taxonomic units (OTUs) at 97% sequence similarity level using UPASE (version 7.1), and UCHime was used to identify and delete chimeric sequences (Edgar, 2010) for OTU clustering and species classification analysis based on the valid data, so as to obtain the corresponding species information and abundance distribution. In addition, species composition and alpha diversity were analyzed to obtain the composition and richness information of eukaryotes in river crabs (Bokulich and Mills, 2013; Mueller et al., 2014).

The software program SPSS 26.0 (SPSS, Chicago, IL, United States) was used to analyze the obtained data. One-way analysis of variance was used to compare the differences in δ¹³C or δ¹⁵N in feeding sources under different feeding modes. To determine homogeneity of variance, inverse sine or square root processing was performed on the percentage data. One-way analysis of variance was then applied, using the Tukey’s-b (K) method for multiple comparisons. If homogeneity of variance was not satisfied after data conversion, then multiple comparisons were performed with Games-Howell Parametric tests. P < 0.05 was considered significant. To assess the contribution of different feed sources to the isotopic characteristics of E. sinensis, the stable isotope mixing model “Stable Isotope Analysis in R” (SIAR; Parnell and Jackson, 2013) package (version 4.2) was used in R 3.3.2 (R Development Core Team, 2016). The SIAR model used the δ¹³C and δ¹⁵N values of consumers and their feed sources to estimate the potential contribution of each feed source that consumers prey on. The model was run in R, allowing variability to be included in the stable isotope ratios of predators and potential feed sources (Parnell et al., 2010). The SIAR model was used to identify the stomach contents and obtain stable isotopic values according to the study of Caut et al. (2009), who found that the δ¹⁵N fractionation factor of E. sinensis, i.e., the Δ value, was 2.75 ± 0.01 ‰, and the fractionation factor of δ¹³C was 0.75 ± 0.11 ‰.

Changes in the average weight and WGR of juvenile crabs during culture under three feeding modes are shown in Figure 2. As shown in Figure 2A, the average weight of juvenile E. sinensis in all three feeding modes gradually increased as the crabs matured to the breeding stage. At the end of the breeding stage (November), there was no significant difference in the average weight between the traditional feeding mode (8.43 ± 2.13) and the mixed feeding mode (8.36 ± 2.39) (P > 0.05), which was significantly lower than the average weight of the juvenile crabs in the formulated feeding mode (10.52 ± 2.60) (P < 0.05). As shown in Figure 2B, there is no significant difference in the WGR between the traditional feeding mode (593.23 ± 15.29) and the mixed feeding mode (615.91 ± 38.09) (P > 0.05), and was significantly lower than that in the formulated feeding mode (593.23 ± 15.29) (P < 0.05).

The alpha diversity of the stomach contents of male and female crabs under the three feeding modes is shown in Figure 3. The dilution (Figure 3A) and Shannon–Wiener diversity index curves (Figure 3B), which were based on the number of OTUs, tend to saturation.

The Simpson’s diversity indices of the stomach contents of female crabs under traditional, formulated, and mixed feeding modes were 0.27 ± 0.07, 0.32 ± 0.11, and 0.27 ± 0.03, respectively, indicating no differences between any two groups (Figure 3C; P > 0.05), and the corresponding Simpson’s diversity indices of the stomach contents of male crabs were 0.25 ± 0.14, 0.24 ± 0.09, and 0.57 ± 0.30, respectively, also indicating no differences between any two groups (Figure 3C; P > 0.05). There were no differences in Simpson’s diversity indices between male and female crabs under the same feeding mode (P > 0.05).

The Chao’s species richness indices of the stomach contents of female crabs under traditional, formulated, and mixed feeding modes were 61.67 ± 26.77, 24.33 ± 7.42, and 65.33 ± 20.50, respectively, indicating no differences between any two groups (Figure 3D; P > 0.05). Similarly, the Chao’s species richness indices of the stomach contents of male crabs under traditional, formulated, and mixed feeding modes are 37.00 ± 4.04, 53.00 ± 22.50, and 22.33 ± 8.41, respectively, indicating no differences between any two groups (Figure 3D; P > 0.05). There were no differences in the Chao’s species richness indices between male and female crabs under the same feeding mode (P > 0.05). The abundance of the main potential eukaryotes in the stomach of male and female crabs was not different under the different feeding modes, but the compositions of their stomach contents were different.

The species with the highest OTU abundance under the three feeding modes are shown in Table 2. A total of 34 phyla were identified in the stomach contents of crabs. The number of phyla identified in the stomach contents of crabs reared under traditional, formulated, and mixed feeding modes were 24, 24, and 29, respectively.

No differences (P > 0.05) in the abundance of stomach phyla were observed between crabs reared under the three feeding modes, with Arthropoda showing the highest overall phylum abundance in the stomach contents of crabs reared under the three feeding modes. The content of Arthropoda in the stomach of male (45.68 ± 0.14%) and female (51.48 ± 0.12) crabs was the highest under the traditional feeding mode, but the opposite was true under the formulated feeding mode. The results showed that the contents of Phragmoplastophyta in the stomach of male (23.20 ± 0.15%) and female (18.99 ± 0.16%) crabs was the highest under the formulated feeding mode, and the contents of Diatom in the stomach of male (8.43 ± 0.02%) and female (6.96 ± 0.02%) crabs was the highest under the mixed feeding mode, while the contents of Phragmoplastophyta (in male crabs abundance accounted for 6.80 ± 0.01%; in female crabs abundance accounted for 16.90 ± 0.13%) and Diatom (male crabs abundance accounted for 4.35 ± 0.02%; in female crabs abundance accounted for 3.72 ± 0.03%) in the stomach of crabs were the lowest under the traditional feeding mode.

The class level of abundance of the main potential feed sources of crabs under the different feeding modes is shown in Figure 4, in which Malacostraca has the highest overall abundance in the stomach contents of crabs under the three feeding modes, with the lowest being 71.02% (formulated feeding mode), the highest 61.84% (mixed feeding mode), and an average of 67.44%. The proportions of Malacostraca, Embryophyta, Trebouxiophyceae, Bivalvia, Chlorophyceae, Insecta, and Intramacronucleata decreased successively to 69.45, 9.01, 8.34, 4.34, 3.30, 3.18, and 2.39%, respectively, under the traditional feeding mode. The proportions of Malacostraca, Trebouxiophyceae, Embryophyta, Chlorophyceae, Intramacronucleata, Insecta, and Bivalvia decreased successively to 71.02, 13.39, 8.95, 6.15, 0.48, 0.1, and 0.08%, respectively, under the formulated feeding mode. The proportions of Malacostraca, Embryophyta, Trebouxiophyceae, Chlorophyceae, Intramacronucleata, Insecta, and Bivalvia decreased successively to 61.84, 12.91, 5.45, 3.5, 0.29, 0.1, and 0.09%, respectively, under the mixed feeding mode. The taxa with different feeding patterns at different class levels are the same but have different proportions (Figure 4). Among the three feeding modes, the top three taxa were Malacostraca, Embryophyta, and Trebouxiophyceae.

The δ¹³C and δ¹⁵N values of E. sinensis juvenile crab muscle under three different feeding modes are shown in Table 3. The muscle δ¹³C values (‰) of juvenile crabs under the formulated, mixed, and traditional feeding modes were −24.02 ± 0.72, −24.45 ± 1.22, and −23.31 ± 1.19, respectively, indicating no difference between the three feeding modes (P > 0.05). The muscle δ¹⁵N values (‰) of juvenile crabs in the formulated, mixed, and traditional feeding modes were 3.37 ± 0.64, 3.08 ± 0.37, and 2.13 ± 0.57, respectively, and that of the formulated feeding mode was significantly higher than that of the traditional feeding mode (P < 0.05).

The overall distribution of δ¹³C and δ¹⁵N stable isotopes of the potential feeding sources of crabs under the three feeding modes is shown in Figure 5. The δ¹³C values ranged from −33.90 ± 0.54‰ to −23.19 ± 0.11‰, and the δ¹⁵N values ranged from 0.49 ± 0.02‰ to 5.63 ± 0.01‰. Among them, the δ¹³C variation ranges of the potential feed sources of crabs under the traditional, formulated, and mixed feeding modes were –33.43 ± 0.71‰ to –26.04 ± 0.19‰, –32.98 ± 0.93‰ to –23.19 ± 0.12‰, and –33.90 ± 0.54‰ to –24.47 ± 0.05‰, which changed substantially, like the corresponding δ¹⁵N variation under the three feeding modes (0.49 ± 0.02–5.63 ± 0.01‰, 2.01 ± 0.07–5.63 ± 0.01‰, and 1.09 ± 0.21–5.63 ± 0.01‰, respectively).

The average feed contribution rate in E. sinensis was analyzed using SIAR (Figure 6). Under the traditional feeding mode, the main diet of E. sinensis was a traditional feed, with an average contribution of 71.89%, followed successively by POM and A. philoxeroides, with average contributions of 16.66 and 11.46%, respectively. Under the formulated feeding mode, the main diet of crabs was formulated feed, with an average contribution of 66.48%, followed successively by A. philoxeroides and POM, with average contributions of 17.50% and 16.03%, respectively. Under the mixed feeding mode, the main diet of crabs was a mixed feed, with an average contribution of 76.84%, followed by POM (11.92%) and A. philoxeroides (11.24%), whose average contribution rates were not significantly different.

In this study, 18S rDNA was used to analyze the stomach feed composition of male and female juvenile crabs under traditional, formulated, and mixed feeding modes. The flat curve of Rarefaction index and Shannon–Wiener curve indicated that the sequencing data covering the feeding source species of juvenile crabs and were sufficient to reflect the feed source diversity of crabs under different feeding modes. The Simpson and Chao indexes represented the diversity and abundance of food composition in the stomach of the crabs, respectively. The higher the Simpson index, the lower the diversity index, while the Chao index of abundance showed a contrary trend. The results showed that there was no significant difference among the three feeding modes (P > 0.05), but the terminal body weight and WGR under the formulated feeding mode were significantly higher than those under traditional feeding mode and mixed feeding mode. This result may be due to the traditional, formulated, and mixed feeds made up the majority of the crabs’ diet, so the diversity and abundance of other feed items were different but not significantly so. But the growth of E. sinensis is affected by factors such as feed and culture environment (Shao et al., 2013, 2014), which can reflect the quality of feed (Yang et al., 2014). The formulated feed containing high protein and fat (Yang et al., 2011), which is conducive to the growth of juvenile crabs. However, the traditional feeding mode had fewer nutrients, so the above indexes of juvenile crabs fed with formulated feed alone were significantly higher than those of the other two feeding modes.

The most representative phyla reported for the sampling site (Arthropoda, Phragmoplastophyta, and Diatomea) also showed the highest occurrence in the dietary samples analyzed. The phylum Arthropoda, with the highest proportion, may be composed of arthropods within the feeding space of crabs, including Malacostraca (injured or newly molting crabs) (Li, 2006) and Insecta (aquatic nymphal insects) (Chen et al., 1989). This is different from the results of Jin et al. (2003), who studied the feeding habits of the second instar adult E. sinensis, and can be attributed to the different ecological niches and living environments of the crabs. The animal feed content of Malacostraca and Bivalvia in the stomach of juvenile crabs reared under the traditional feeding mode was the highest among the three feeding modes because the traditional feed is a plant-based feed mainly including bran, wheat, and soybean meal, which are insufficient to meet the nutritional needs of juvenile crabs (Veilleux and de Lafontaine, 2007), thus driving them to consume more animal feed. Under the formulated feeding mode, the plant feed content of Embryophyta and Trebouxiophyceae in the juvenile crab stomach was the highest among the three feeding modes, mainly because the formulated feed is mainly composed of animal feed, and to achieve nutritional balance (Chen et al., 1989), which are entwined and attached to the other feed sources of crabs (Vizzini and Mazzola, 2003). Under the mixed feeding mode, the animal and plant feed contents in the crab stomach were the lowest between the three feeding modes because the mixed feed contained both animal and plant feeds and provided balanced nutrition to crabs (Qian and Zhu, 1999).

Compared with zooplankton, fish, and macroinvertebrates in the near waters of Gouqi Island, the variation in the values of δ¹³C and δ¹⁵N were relatively large (Jiang et al., 2014), indicating that crabs have a wide range of feed sources. Previous studies have shown that the crab’s δ¹³C and δ¹⁵N changes depending on what it eats (Zeng, 2010) and the δ¹³C value of consumers is usually close to that of their feed, the δ¹³C value of individuals does not change much in the transmission of the whole feed chain (Lin, 2013), which is in accordance with the results of this study. Animals reject δ¹⁵N during exudation of substances; therefore, the higher the trophic level, the higher the content of δ¹⁵N (McCutchan et al., 2010). However, the δ¹⁵N values of E. sinensis crabs under traditional (2.13 ± 0.57‰), formulated (3.37 ± 0.64 ‰), and mixed feeding (3.08 ± 0.37 ‰) modes in the present study were lower than those of A. philoxeroides (5.63 ± 0.37‰), and this may be because δ¹⁵N is poorly conserved and easily modified by biogeochemical processes, and an increase in the N content in the environment may increase the N content in plants (Lin, 2013; Yu et al., 2014).

Previous studies have shown that juvenile crabs with better growth performance can be obtained by feeding high quality feed (Wu et al., 2009), the content of fat and ash in formulated feed was higher than that of two other feeding modes indicating that formulated feed had higher energy level, this may be the reason why the growth performance of the crab fed with formulated feed is better. Under the traditional feeding mode, crabs mainly consumed POM as a feed supplement in addition to the traditional feed, which is consistent with the results of Paning (1934) and Rudnick et al. (2000), who reported that the stomach contents of E. sinensis contain a large amount of POM. POM is the substrate of complex biological communities, formed by the decomposition and mineralization of dead animals and plants, humus, and feces by microorganisms (Liu, 1999), and some of its components are the main feed sources for benthic invertebrates (Vizzini and Mazzola, 2003). Under the formulated feeding mode, apart from the formulated feed, the proportion of A. philoxeroides was the highest. In the presence of sufficient animal feed, crabs will still choose A. philoxeroides as a feed supplement (Li et al., 2018). This is because A. philoxeroides is rich in cellulase, which helps to improve the cellulase activity (Yang et al., 2014) and digestive ability of crabs. The mixed feed containing plant- and animal-based feeds had a balanced nutritional ratio (Qian and Zhu, 1999) and high palatability, and the proportion of mixed feed consumed by E. sinensis was the highest, resulting in the lowest feeding proportion of POM and A. philoxeroides between the three feeding modes. But the size, shape and palatability of the two diets were significantly different, and the physical difference of the mixed feeding group made the crab unable to make full use of the feed (Wang S. M. et al., 2017). Therefore, there was no significant difference in the weight change and WGR of juvenile crabs under the mixed feeding mode and the traditional feeding mode. The results of the stable isotope technique were consistent with those of the 18S rDNA analysis. Juvenile crabs feed on both plant- and animal-based feeds in an aquaculture pond, but they are not complete predators and selectively feed on animal or plant feeds as supplements of that which is deficient, in addition to their main feed (Jin, 2003).

Various studies have shown that there are no differences in feed composition between male and female crabs, and they are, therefore, ecological equivalents (Spooner et al., 2007; Young and Elliott, 2020). In the present study, the diversity and abundance of female crabs were generally higher than those of male crabs, and female crabs had a higher rate of feeding on available feed sources in the pond, this may be because female crabs increase their feeding rate during October to November every year to meet their growth and development needs (Chen et al., 1989), while promoting gonad maturation and transformation (Xu et al., 2019). However, no significant differences were observed in Simpson’s diversity and Chao’s abundance indices of male and female crabs under the same feeding mode as well as in the abundance of the three main feeding sources, Arthropoda, Phragmoplastophyta, and Diatomea (Table 2), indicating that male and female crabs showed similar feed composition under the same feeding mode. Cordone et al. (2020) studied the feeding habits of C. maenas and found that different living water depths and a wide range of available feed sources lead to differences in feed composition between male and female crabs, which is different from the results of this study. This may be because the ponds mainly contained artificial feeds for arthropods (including shrimp), mollusks (including snail), cultivated A. philoxeroides, Phragmoplastophyta dominated by Trebouxiophyceae species, and diatoms dominated by Chlorella (Lu et al., 2013). The pond was shallow, and the crabs can eat less feed than wild crabs, but the difference was minor.