Intestinal AW of L. waleckii and alkaline water samples were collected from Dali Nor-Lake (116°40′13″ E, 43°24′37″ N), Inner Mongolia, China (Figure 1), which were designated as DLG 1-3 and DLS 1-3, respectively. The living alkaline water conditions were water temperature of 24°C−25°C, pH of 9.4–9.6, the dissolved oxygen level of 8.86–9.34 mg/L, and alkalinity of 52.8–53.6 mmol/L. In addition, intestinal FW of L. waleckii and freshwater samples labeled as TRG 1-3 and TRS 1-3 respectfully were taken from the Tangwang River (129°44′24″ E, 46°41′15″ N) belonging to the Songhuajiang River tributaries, Heilongjiang Province, China (Figure 1). The living freshwater conditions were water temperature of 23°C−25°C, pH of 7.3–7.9, the dissolved oxygen level of 7.56–10.25 mg/L, and alkalinity of 0.5–0.8 mmol/L. Due to the influence of feeding habits and water temperature of L. waleckii, all the samples were collected in August when the bait was most abundant.

The water samples were collected using an aquatic microbial collection method (Li et al., 2017). Microbiological specimens were collected at 0.22 μm filter membrane via vacuum filtration, and the filter membrane containing the whole water sample was maintained in 15 ml sterilized centrifuge tubes, froze in liquid nitrogen, and then stored at −80°C for use. Sixty healthy adult fish (average 398.56 ± 13.58 g) taken from both locations above were quickly brought back to the sterile laboratory after being anesthetized and placed on a sterile operating table. The fishes' surface was sterilized with 75% ethanol, and the intestine was aseptically dissected. Each intestine sample is mixed with 10 healthy fishes intestines. The intestine was put into a 1.5 ml sterile centrifuge tube, froze in liquid nitrogen, and then stored at −80°C for use (Fan et al., 2019).

The pure Soil DNA Mini Kit (Magen, Guangzhou, China) was used for microbial DNA extraction as per the instructions of the manufacturer. The proposed DNA concentration and purity were determined using a NanoDrop 8000 spectrophotometer (NanoDrop Technologies, Wilmington, USA). The conditions for PCR amplification of the V3–V4 region of the 16S rRNA were as follows: 94°C for 2 min, 98°C for 10 s, 62°C for 30 s, 68°C for 30 s for 30 cycles, and 68°C for 5 min. The primers used were 341 F (CCTACGGGNGGCWGCAG) and 806 R (GGACTACHVGGGTATCTAAT) (Guo et al., 2017), which were subjected to three PCR replicates as counterparts. The PCR amplification system followed by ReverTra Ace-α-™ (TOYOBO, Japan) contained the following: 5 μl 10 × KOD buffer, 5 μl 2 mM dNTPs, 3 μl 25 mM MgSO₄, forward and reverse primers in 1.5 μl each (10 μM), 1 μl KOD polymerase, 100 ng DNA template, and appropriate volume of ddH₂O obtaining a total reaction volume of 50 μl. All the PCR products were detected using 2% agarose gels containing ethidium bromide and were purified with the AxyPrep DNA Gel Extraction Kit (Axygen Biosciences, Union City, CA, USA), and quantified using QuantiFluor™-ST (Promega, USA). Purified PCR amplicons were pooled in equimolar and paired-end sequenced (PE250) using an Illumina MiSeq platform (Illumina, San Diego, CA, USA) according to the instructions of the manufacturer. The raw sequencing reads of all samples were deposited to the NCBI database under BioProject accession number: PRJNA810745.

Sequencing libraries were generated using the TruSeq® DNA PCR-Free Sample Preparation Kit (Illumina, USA) following the recommendations of the manufacturer, and index codes were added. Next, the library quality was assessed on a Qubit@ 2.0 Fluorometer (Thermo Scientific) and Agilent Bioanalyzer 2100 system. Finally, the library was sequenced on an Illumina NovaSeq platform, and 250 bp paired-end reads were generated.

The process comprises five main steps: data split, sequence assembly, filtration, chimera removal, and gene function prediction. Briefly, raw fastq sequence files produced in the 18.0 version were demultiplexed, quality-filtered by Trimmomatic, and concatenated by FLASH (Mago and Salzberg, 2011). Cluster sequences with similarities higher than 97% were assigned to the same operational taxonomic units (OTUs) (Caporaso et al., 2010). The most abundant sequence in each OTU was aligned with the Greengene database (DeSantis et al., 2006) to obtain taxonomic information on the OTUs. Using UPARSE (version 7.1 http://drive5.com/uparse/), OTUs were grouped with a 97% similarity criterion, and chimeric sequences were identified and deleted using UCHIME. Based on the Silva16S rRNA database (SSU123), the taxonomy of each 16S rRNA gene sequence was assessed by the RDP Classifier algorithm (http://rdp.cme.msu.edu/) against the Silva 16S rRNA database (SSU115) with a confidence level of 70% (Quast et al., 2012; Amato et al., 2013).

Phylogenetic Investigation of Communities by Reconstruction of Unobserved States PICRUSt (http://picrust.github.io/picrust/) is a biocomponent analysis package that aligns microbial community richness and databases by predicting microbial functions and metabolic pathways based on 16S rRNA high-throughput sequencing results. The useful Kyoto Encyclopedia of Genes and Genomes community information via the corresponding OTU table after aligning the sequencing data through the Greengenes database 13.8, and then homogenizing each out abundance (Loudon et al., 2014).

Based on the characteristics of the amplified regions, small fragment libraries were built for single-end sequencing utilizing the single-end sequencing method based on the IonS5TMXL sequencing platform, according to the principle of 16S rDNA amplicon sequencing technology. Furthermore, the t-test method for independent samples was used to compare differences in microbial diversity and relative abundance of bacteria in the analyzed intestines of AW, FW, and water samples. Principal coordinates analysis and analysis of similarity based on distances of unweighted UniFrac, Jaccard, and Bray–Curtis were performed using the R language vegan package (Conway et al., 2017) to compare their groups of samples in terms of differences and significance detection, respectively, in the composition structure of microbial communities.

For all samples, 16S rRNA gene V3–V4 regions of bacteria were sequenced to profile the microbiota of the intestines of AW and FW in alkaline water and freshwater, respectively, using the Illumina MiSeq platform. The initial data quality and chimeric filtering delivered 883,017 high-quality sequencing reads from 12 samples split into four groups (three samples per group) followed by 10 fishes per sample (Table 1 and Supplementary Table S1). In Figure 2, the richness and diversity of bacterial species are split into four groups: ACE, Chao, Shannon, and Simpson indices related to OTU level. The Shannon and Simpson indices are commonly used to quantify biodiversity. The richness of each sample was determined using the Chao index and ACE indices (Fan et al., 2019). The estimators of ACE ranged from 677.729 to 1224.136, Chao ranged from 727.238 to 1226.519, Shannon ranged from 5.390 to 7.079, and Simpson ranged from 0.930 to 0.989 (Table 1). These were summarized with a significant difference (P < 0.05) from each sample at a genetic distance of 3%. The results revealed that the microbial richness and diversity were significantly reduced in the alkaline water group compared to the freshwater group. The Good's coverage within each sample to calculate sequence thoroughness was an average of 99.3325%, indicating that the sequences discovered could represent the majority of microorganisms in each group. The data for the t-test can be found in Appendix S1.

The major phylum and genus relative abundance in intestinal bacteria of all samples from alkaline water and freshwater groups are summarized in Figure 3A, and any known sequences' groups are labeled as “others.” All the samples followed abundant phylum in descending order: Proteobacteria (30.07%) > Actinobacteria (17.52%) > Oxyphotobacteria (16.25%) > Bacteroidetes (13.40%) > Planctomycetes (5.88%) > others (4.54%) > Verrucomicrobia (4.42%).

Operational taxonomic unit (OTU) abundance was normalized using a standard sequence number corresponding to the sample with the most miniature sequences. Next, alpha and beta diversity analyses were performed based on the normalized output data. The Venn diagram in Figure 3B was developed to identify prominent OTUs provided in the four groups above to explore the significant microbiome in all samples between intestines and water types in alkaline water and freshwater environments. There were 284 OTUs shared among DLG, DLS, TRG, and TRS (n = 3), representing 4.64% of the total reads; sequencing by the number of OTUs was TRS > DLG > TRG > DLS, and the high-quality sequences were grouped into 1,467 OTUs at 97% similarity level (Figure 3B). Among them, 601 OTUs in the alkaline water group (DLG and DLS) were identified based on a similarity of <730 OTUs inside the freshwater group. Whereas 1,024 OTUs in intestinal samples of L. waleckii (DLG and TRG) were defined totally (specialized OTUs were, respectively, 1,182 and 896), based on a similarity more remarkable than the total 443 OTUs in alkaline water and freshwater (DLS and TRS) had specialized OTUs were, respectively, 428 and 789.

The main strains in the four groups (DLG, DLS, TRG, and TRS) are shown in Figure 4A. According to the relative abundance of strains, four groups are arranged as follows from rich to poor: (a) For DLG (Figure 4B), Planctomycetes (19.68%) > Actinobacteria (17.78%) > Oxyphotobacteria (17.47%) > Proteobacteria (16.85%) > Firmicutes (10.57%) > Verrucomicrobia (9.89%) > Bacteroidetes (3.68%) > others (2.53%) > Deinococcus-Thermus (0.92%) > Tenericutes (0.62%) > Armatimonadetes (0.01%); (b) for DLS (Figure 4C) accounting for more than 10%, Actinobacteria (28.23%) > Bacteroidetes (25.58%) > Oxyphotobacteria (21.40%) > Proteobacteria (16.39%); (c) for TRG (Figure 4D), proteobacteria occupied an absolute advantage of 63.89%, Firmicutes account for 20.50% DLG, and the total proportion of other bacteria is 15.61%; (d) for TRS (Figure 4E), the proportion of main fungi was relatively average, and four kinds of bacteria accounting for more than 20% were dominant: Oxyphotobacteria (24.98%) > Proteobacteria (23.15%) > Actinobacteria (22.14%) > Bacteroidetes (20.85%); and (e) the clustering results of the four groups of samples revealed little difference between each group, (Figures 4F,G), indicating that the detection results are highly correlated and reliable.

Several researchers have applied a multinomial logit model to analyze the forecasting accuracy of indicator populations and discovered that they would have highly accurate results at the bacteroideae level (Xiong et al., 2014). Following this method, the indicative evaluation in this study was used to filter the variations in the bacterial genus (Figure 5) and phylum (Figure 6), reflecting the survival status of AW and FW in alkaline water and freshwater, respectively. We chose communities with indicator values > 0.90. We screened 15 microbial families to increase the accuracy of the results, displaying the characteristics of indicator families in all samples. At the genus level, four genera of nine dominant genera, summarized in Figure 5A, showed statistically significant differences (P < 0.05) in all samples, namely, Aeromonas (Figure 5B), Flavobacterium (Figure 5C), Fluviicola (Figure 5D), Unidentified_Synechoccales (Figure 5F), and Unidentified_Oxyphotobacteria (Figure 5G). There were significant differences (P < 0.001) in TRS, DLS, and DLG of the Flavobacterium and Polynucleobacter compared with TRG, but there were no statistically significant differences between the three groups of DLS:TRS, DLG:TRS, and DLG:DLS. In Fluviicola, the change was similar to the former. However, the significant increase was lower (P < 0.05) in the TRS:TRG, TRS:DLG, and TRS:DLG groups. In addition, three groups (TRS:TRG, DLS:TRG, and DLS:DLG) in the Unidentified_Synechoccales showed significant differences (P < 0.001), and TRS:DLG showed a substantial increase (P < 0.01) in the Unidentified_Oxyphotobacteria, additional information about the relative abundances at genus level among the four groups can be acquired in Appendix S2.

Three of the eight dominant phyla showed statistically significant differences (P < 0.05) in all samples (Figure 6A), namely, Bacteroidetes (Figure 6B), Firmicutes (Figure 6C), and Verrucomicrobia (Figure 6F). There were significant differences (P < 0.001) in TRS, DLS, and DLG of the Proteobacteria (Figure 6E) compared with TRG, and slight trends of TRS (P < 0.05), DLS (P < 0.05), and DLG (P < 0.1) in Oxyphotobacteria (Figure 6D) compared with TRG. However, there were no statistically significant differences between the three groups (DLS:TRS, DLG:TRS, and DLG:DLS). In Planctomycetes (Figure 6G), the change was similar to that of the former. However, a significant increase was observed (P < 0.001) in the three groups DLG:DLS, TRS:DLG, and TRS:DLG. Four groups (DLS:DLG, DLS:TRG, TRS:DLG, and TRS:TRG) in the Bacteroidetes showed significant differences (P < 0.05), and two groups (TRG:TRS and TRG:DLS) increased at a similar level in the Firmicutes, and groups of TRG:TRS and TRG:DLS showed a significant increase (P < 0.05). In Verrucomicrobia, there were significant increases in DLG:TRG (P < 0.001), DLS:TRS (P < 0.05), and DLG:DLS (P < 0.05), additional information about the relative abundances at phylum level among the four groups can be acquired in Appendix S3.

PICRUSt illustrated significantly changed functions of intestinal microbiota in shrimp intestines and living waters. Compared with the taxonomic profiles, the main functions checked from the samples were broken into 11 parts that showed more similarity in all four groups. The projected capabilities can be prioritized to define the function of gut microbes based on the percentage summarized in Figure 7: unknown function (33.12–33.99%) > membrane transport (9.81–12.89%) > amino acid metabolism (9.87–11.15%) > carbohydrate metabolism (9.71–10.28%) > replication and repair (6.98–7.61%) > energy metabolism (5.35–7.55%) > poorly characterized (4.92–5.20%) > metabolism of cofactors and vitamins (4.10–5.04%) > translation (4.19–4.88%) > lipid metabolism (3.45–3.86%) > cellular processes and signaling (3.21–4.02%). These results showed that the gastrointestinal microbiome of shrimp exhibits biological roles similar to those found in alkaline water or freshwater. Many metabolic activities are aided by microbiome-like membrane transport, amino acid metabolism, glucose metabolism, energy metabolism, replication, and repair.