Artificial substrata increase pond farming density of grass carp (Ctenopharyngodon idella) by increasing the bacteria that participate in nitrogen and phosphorus cycles in pond water

Experimental design

The experiment was conducted in the precise aquaculture base of the Pearl River Fisheries Research Institute. The experimental period was 60 days. Thirty square cement ponds with of 2 × 2 m were employed. An explosion disc was placed in the centre of each pond by continuous aeration 24 h every day, with an approximate airflow of 0. 42 ×10⁻⁴ m³/s. Before the experiment, the ponds were sterilized for 2 h using 0.1% of potassium permanganate. Then, water was added the ponds to a depth of 1 m. The ponds for treatment groups (D0+S2, D20+S2, D30+S2, D40+S2, and D50+S2, based on density, described as follows) contained two artificial substrata (AquaMats). The polypropylene nonwoven that does not easily decompose and is environmentally friendly was used for the artificial substratum. The size of the artificial substratum was 1 × 1 m and its density was described by Azim et al. (2004) and Zhang et al. (2019). The artificial substrata were placed as shown in Fig. 1. The ponds for control groups (D0, D20, D30, D40, and D50) did not contain the artificial substratum. There were three replicates for each group and five densities were tested, including 0, 20, 30, 40, and 50 individuals per pond, for both the control and treatment groups. The initial average body weight of the grass carp was 0.35 ± 0.07 kg. Commodity feed containing 30% crude protein was administered at 9:00 and 16:00 every day, and the daily feed amount was 3% of the total weight of the grass carp in pond. During the experiment, all ponds were aerated by disc microporous aeration. The pond water was changed according to the pH value of the water. Specifically, the pH value was determined once per day and when this value was less than 6.0, 10% of the pond water was exchanged with aerated tap water.

Determination of physical and chemical indexes

Starting from the stocking of grass carp, the dissolved oxygen, pH, and temperature in the water were measured daily using the ProPlus portable multiparameter water quality measurer (YSI, USA). Total nitrogen (TN), total phosphorus (TP), ammonia nitrogen (NH₄⁺-N), nitrate nitrogen (NO₃⁻-N), nitrite nitrogen (NO₂⁻-N), and chemical oxygen demand (COD) were measured at 10:00 once every 7 days. Among them, TN and TP contents were measured using the Kjeldahl method and molybdenum blue colorimetry method, respectively (Lu, 2000). NH₄⁺-N, NO₃⁻-N, and NO₂⁻-N were measured using a Pharo100 multi-parameter water quality analyzer (Merck, Germany), and COD was measured with a COD measurer (HACH, USA).

The total exchange ratio of water (TERW) was calculated as follows: ${\displaystyle \mathrm{TERW}=\mathrm{TV\; EW}∕\mathrm{V\; PWS},}$ where TVRW is the total volume of exchanged water and VPWS is the volume of pond water at the start.

Growth performance evaluation

At the end of the experiment, after 24 h of starvation, the pond water was drained and the body weights of grass carp were determined. The specific growth rate (SGR, %/d) and feed conversion rate (FCR, %) were calculated as follows (Cui & Wootton, 1988): SGR = {[ln(final weight) − ln(initial weight)] ×100}/experimental days. FCR = feed consumption (dry weight)/(final fish weight (wet weight) − initial fish weight (wet weight)).

Microbiota collection, DNA extraction, and sequencing

Each 200-ml pond water sample was collected at the end of the experiment and filtered using a GF/C filter with a 0.22-µm pore size. The filter was cut into fragments and placed in a 50-ml sterile centrifuge tube for DNA extraction. Three grams of each artificial substratum sample were also weighed and added to 200 ml of sterile water. The mixture was then vortexed for 2 h. The mixture was centrifuged at 2,500 × g for 15 min at 4 °C and the suspension was collected for DNA extraction. For this, the suspensions were filtered using GF/C filters with 0.22-µm pore size. The filter was cut into fragments and placed in a 50-ml sterile centrifuge tube for DNA extraction. Bacterial DNA was extracted using a kit for the extraction of bacterial DNA from water (Omega, Norcross, GA, USA). DNA concentration and purity were evaluated with 1% agarose gels. Based on the concentration, DNA was diluted to 1 ng/µl with sterile water for further amplification. The V4 hypervariable region of the 16S rRNA gene was amplified using the 515F and 806R primers with sample-specific barcodes (Yan et al., 2016). PCR was performed in 30-µl reaction volumes with 15 µl of Phusion High-Fidelity PCR Master Mix (New England Biolabs, Ipswich, MA, USA), each primer at 0.2 µM, and approximately 10 ng of template DNA. Thermal cycling conditions were as reported previously (Li et al., 2017). The PCR products were mixed at ratios of equal density and the mixed PCR products were purified using the Gene JET Gel Extraction Kit (Thermo Scientific, Waltham, MA, USA). Sequencing libraries were constructed using the NEB Next Ultra DNA Library Prep Kit for Illumina (New England Biolabs), according to the manufacturer’s recommendations, and index codes were added. Finally, the libraries were sequenced using the Illumina MiSeq platform and 250-bp paired-end reads were generated. MiSeq sequencing was conducted by Novogene Co. (Beijing, China).

The paired-end reads from the raw DNA fragments were merged using FLASH software. The merged tags were assigned to each sample according to the sample-specific barcodes, and the low-quality sequences were detected and removed using QIIME 1.9.0 software according to previous reports (Li et al., 2017; Huang et al., 2018; Ni et al., 2017; Ni et al., 2019). Chimeric sequences were detected and removed using Uchime algorithm (Edgar et al., 2011) embedded to QIIME 1.9.0 software. QIIME 1.9.0 software (Caporaso et al., 2010) and the vegan package (Dixon, 2003) in the R platform (R Core Team, 2014) were used to analyze alpha (within samples) and beta (among samples) diversity. Sequences with ≥97% similarity were assigned to the same operational taxonomic unit (OTU) using UPARSE software (Edgar, 2013). Representative sequences for each OTU were selected and the RDP classifier was used to annotate them with appropriate taxonomic information (Wang et al., 2007). In-house Perl scripts were used to analyse alpha and beta diversity.

All merged DNA sequences have been submitted to the genome sequence archive database (https://bigd.big.ac.cn/) under the accession number CRA001827 (https://bigd.big.ac.cn/gsa/s/Z6ZLD9E9).

Data analysis

Data were recorded and pre-treated using Excel. Data are shown as the mean ± standard error. Principle component analysis (PCoA) was conducted using QIIME 1.7.0, non-parametric multivariate analysis of variance (PERMANOVA; Anderson, 2001) was conducted using R software with the vegan package (Dixon, 2003), and one-way ANOVA and t-tests were conducted using R software with the base packages. Linear discriminant analysis effect size (LEfSe) was conducted using the Galaxy platform as our previous description (Li et al., 2019). P values <0.05 were considered significant.

Artificial substrata significantly increase the pond farming density of grass carp

The water temperatures of ponds were not significantly different during the experiment (one-way ANOVA, F = 0.736, p = 0.676). Dissolved oxygen in the pond water was closely and negatively correlated with farming density. The pH values of ponds without farmed fish ranged from 7.9 to 8.4, which were significantly higher than those of ponds housing farmed fish (from 5.5 to 7.5; one-way ANOVA, p <0.05; Table S1 ). No fish died during the experimental process. Except for those in the pond without grass carp (D0 and D0S), the concentrations of TN and TP in other ponds fluctuated and increased with culture time from the beginning of the experiment, but remained in specific range after the middle of the experiment (Figs. S1 and S2). The increase in TN concentration was mainly due to the significant increase in nitrate concentration in the pond water (Fig. S3), because nitrite and ammonia nitrogen concentrations did not gradually increase during the experimental process, but rather fluctuated within a certain concentration range (Figs. S4 and S5). In most cases, TN and TP in the ponds that contained artificial substrata were lower than those in the control ponds at the later stage of the experiment (Figs. S1 and S2). Similar results were also obtained regarding the concentrations of nitrate, nitrite, and ammonia nitrogen, especially for the high stocking density ponds (D40 and D50; Figs. S3–S5).

With an increase in grass carp farming density, the TERW and FCR of both treatment and control groups increased, whereas the SGR of grass carp decreased gradually (Fig. 2). The SGR of grass carp was significantly increased with the presence of artificial substrata (Paired t-test, t = −3.070, p = 0.011), and the FCR was significantly reduced (Paired t-test, t = 3.045, p = 0.011), which indicated that the introduction of artificial substrata into aquaculture ponds promotes the growth of grass carp. According to the fitting formulas SGR_treatment = 6.1075 (farming density)${}_{\mathrm{treatment}}^{-0.553}$ and SGR_control = 5.3384 (farming density)${}_{\mathrm{control}}^{-0.565}$, the artificial substrata were able to increase stocking density to 10.68 grass carp per pond while maintaining an SGR of 0.8. Thus, the estimated stocking density in the pond could be increased by 0.88 kg/m³ (2.67 individuals/m³).

Artificial substrata significantly increase the bacteria that participate in nitrogen and phosphorus cycles

To determine whether the introduction of artificial substrata could increase the microbial species involved in nitrogen metabolism in pond water, the microbiota community structures of 30 pond water samples (three duplicate samples in five treatment groups and five control groups) and 15 artificial substrata were analysed. In total, 2,557,442 (56,832.04 ± 1,202.30) high-quality sequences were obtained. To exclude the influence of sequencing depth, 28,604 sequences were randomly resampled from each sample for further analysis. A total of 6782 OTUs were obtained based on 97% sequence similarity. These sequences belonged to 51 phyla, except for a few (1.03 ± 0.06%) sequences that could not be classified at the phylum level. Acidobacteria, Actinobacteria, Armatimonadetes, Bacteroidetes, Chlorobi, Chloroflexi, Cyanobacteria, Firmicutes, Fusobacteria, Gemmatimonadetes, Nitrospirae, Planctomycetes, Proteobacteria, Synergistetes, and Verrucomicrobia were the dominant phyla, for which relative abundance was more than 1% in at least one sample (Fig. 3A). The relative abundances of Actinobacteria, Chlamydiae, Fusobacteria, and Planctomycetes in pond water were significantly decreased with the treatment, whereas the relative abundance of Proteobacteria was significantly increased (Fig. 3B). This was closely correlated with the significant enrichment of Proteobacteria induced by the artificial substrata (Fig. 3B).

Except for a few sequences (1.03 ± 0.06%), 928 prokaryotic genera were obtained, of which 320 were dominant genera (their relative abundances were more than 0.1% in at least one sample). PCoA results showed that the introduction of an artificial substratum could not only result in the formation of microbial communities in the artificial substrata that were distinct from those in pond water, but could also significantly change the composition of microbiota in the pond water (PERMANOVA, F = 6.77, p = 0.005). Interestingly, the composition of microbiota in pond water with artificial substrata was more similar to that of the artificial substrata (Fig. 4).

LEfSe based on dominant genera showed that those that were significantly enhanced by the artificial substratum gradually decreased with increasing grass carp stocking densities, mainly because the microbiota on the artificial substrata were closer to those of the pond water (Fig. 4). Significantly enriched bacteria on the artificial substrata mainly comprised Proteobacteria, in which Ideonella, Nordella, Hyphomicrobium, Pseudoduganella, and Chitinimonas were the most significantly enriched (they were significantly enriched with at least two farming densities; Fig. 5). In addition, Perlucidibace, Polynucleobacter, Legionella, Romboutsia, Shewanella, Roseiflexus, Planktothrix, and Limnothrix in the pond water with artificial substrata were significantly enhanced (Fig. 5).