The relationship between benthic nutrient fluxes and bacterial community in Aquaculture Tail-water Treatment Systems

Constructed-wetlands, Biofilms, and sedimentation are potential aquaculture tail-water treatments however their roles on the distribution of benthic microbial community and the way they affect the interaction between microbial community and inorganic nutrient fluxes have not been fully explored. This study applied 16S rRNA high-throughput sequencing technology to investigate the microbial community distribution and their link with nutrient fluxes in an aquaculture tail-water bioremediation system. Results showed that bacterial community compositions were significantly different in constructed-wetland and biofilm treatments (p<0.05) relative to sedimentation. The composition of the 16S rRNA genes among all the treatments was enriched with Proteobacteria, Bacteroidetes, Firmicutes, and Flavobacteria. NMDS analysis showed that the bacterial composition in constructed-wetland and biofilm samples clustered separately compared to those in sedimentation. The Functional-Annotation-of-Prokaryotic-Taxa analysis indicated that the proportions of sediment-microbial-functional groups (aerobic-chemoheterophy, chemoheterotrophy, and nitrate-ammonification combined) in the constructed-wetland treatment were 47%, 32% in biofilm and 13% in sedimentation system. Benthic-nutrient fluxes for phosphate, ammonium, nitrite, nitrate and sediment oxygen consumption differed markedly among the treatments (p<0.05). Canonical correspondence analysis indicated constructed-wetland had the strongest association between biogeochemical contents and the bacterial community relative to other treatments. This study suggests that the microbial community distributions and their interactions nutrient fluxes were most improved in the constructed-wetland followed by the area under biofilm and sedimentation treatment.


40
Intensive aquaculture farming practices have contributed overwhelming allochthonous 41 organic matter (OM), excreta, food-wastes, and dissolved nutrients (e.g., nitrogen and occupying 400 sq. m of the total system area. The planting densities of S. anglica were 50% of 95 the total wetland cover. These plants grew rapidly to colonize the wetland and they were not 96 harvested during this study. The biofilm system was deployed with suitable aeration facilities 97 and suspended carriers in the form of fiber threads (adhesive matrix of extracellular polymeric 98 substances) for enhancing the surface area for microorganism attachment. The physical 99 sedimentation consisted of bare sediment surface, overlying aquaculture effluent water, and 100 aeration. This system was involved in filtering and settling large particles of the incoming 101 effluent water from the production center. 102 Sampling started one year after the project launched to let the ecological succession develop. 103 To ensure a representative sampling strategy, data was collected three times consecutively, 104 between April to July. Four different sampling points from each system were identified and 105 sampled. 0.5L of the overlying water was collected from each system for water quality analysis. 106 Using a handheld sediment corer four undisturbed sediment cores (8 cm height) from each 107 system were gently collected in cylindrical plastic tubes (i.d. 6.4 cm, height 19.4 cm). The 108 sediment cores and water samples were immediately brought back to the laboratory for 109 physicochemical analysis and incubation. Water samples were stored at 4°C, whereas the 110 sediment cores were kept ready for the incubation experiment.

111
The incubation experiment was done as previously described [29]. Water samples for the 112 determination of benthic flux rates for the total ammonia nitrogen (TAN), nitrate (NO 3 --N) and 113 nitrite (NO 2 --N), and soluble reactive phosphate (SRP) were collected, filtered in 0.45 GF/F and 114 stored under -20°C until analysis. After the incubation experiment, using a clean stainless steel 115 microspatula, the sediment cores were sliced into three sub-sampling points (surface 0-2 cm, 116 middle 2-4 cm, and bottom 4-8 cm). These subsamples were thoroughly homogenized and 6 6 117 divided into two portions. One potion was freeze-dried for physicochemical contents analysis 118 and the other potion was stored in clean polypropylene tubes at -20°C for the 16S rRNA 119 extraction.

120
Analysis of physicochemical contents and nutrient flux rates 121 The physicochemical water parameters (dissolved oxygen, temperature, and salinity) were 122 measured in situ during sampling using a handheld automated YSI 6000 multi-parameter probe  Where: Flux is the nutrients or sediment oxygen fluxes (mmol m -2 h -1 ); ΔC (mgL -1 ) is the change in 131 concentration of oxygen/nutrients (prior and after incubation); V (m 3 ) the volume of overlying water; 132 A (m 2 ), is the cross-sectional area of the incubation chamber; t (h) is the duration of incubation. The sediment grain size distribution was determined using sieves with different mesh sizes 134 [31]. Briefly, grain-size parameters were conducted mechanically from oven-dried subsamples 135 using standard sieving methods for the sand content (500-63 µm) and sedigraph techniques for  The sediment samples were freeze-dried, pulverized, and pre-weighed before being placed in a 139 muffle furnace at 475°C for 4 h. Then the samples were reweighed with the difference equals to 7 7 140 the %OM content. TOC (total organic carbon), TON (total organic nitrogen), and C/N 141 (carbon/nitrogen) ratio were analyzed commercially by using Carbon Elemental Analyzer.

142
Briefly, during pretreatment, 5g of the post-freeze-dried wet-sediment were ground using a 143 mortar into powder to pass through a 1-mm mesh sieve. Before analysis, further pretreatment 144 procedures necessary especially for TOC were done to remove the carbon dioxide by adding 1:1 145 HCl and oven-dried at 80°C, overnight to a constant weight.  Bioinformatics analysis 164 The sequencing process of the paired reads was initially joined with FLASH using default 165 settings [33], then, the Raw FASTQ files were processed using Quantitative-Insights-Into- 166 Microbial-Ecology (QIIME version 1.8.0, [34]. The operational taxonomic units (OTUs) 167 assigned at a 97% similarity cut-off point in all samples were clustered using USEARCH 168 (version 7.1, http://drive5.com/uparse/). The sequences were quality filtered based on sequence 169 length, quality score, chimera, and primer mismatch thresholds. In a nutshell, homopolymer runs 170 exceeding 6 bp were screened-out by PyroNoise. Sequences with the same barcodes were 171 assigned to the same sample. The phylotypes were performed using the UCLUST algorithm [35].

172
The most abundant sequences of each phytotype were selected as the clean sequence and were 173 taxonomically assigned (Greengenes database, release 13.8) using PyNAST [36]. Diversity 174 indices (Shannon index, Simpson, Chao1, and observed OTUs) were performed using the 175 phylogenetic tree (QIIME pipeline).

177
The variations of the different physicochemical variables were analyzed by a one-way or 178 two-way repeated ANOVA. Post Hoc tests were performed to determine the significant groups.

179
The normal distribution and homogeneity of variances among treatments were verified before the 180 ANOVA test. All the data were Hellinger transformed post statistical analyses, and then 181 normalized by using the function decostand/p-p plot in the "vegan" package to improve    The sediment physicochemical contents are described in Table 1. The results indicate 208 distinct differences in sediment organic and inorganic contents among the systems. The surface 10 10 209 sediments of the constructed-wetland consisted of 79% medium sand, 17% very fine sand, and 210 4% silt/clay whereases the compositional contents in the biofilm were 68% medium sand, 25% 211 very fine sand, and 7% silt. The sedimentation system was dominated by 84% (medium sand), 212 11% (very fine sand), and 5% (silt). Cores from the sedimentation system had significantly    The bacterial community compositions varied among depths and between the treatment 242 systems (Fig 2). A total of 519, 692, and 837 OTUs were identified for sedimentation, 20-50-fold higher (biofilm) relative to constructed-wetland and sedimentation (jointly 6-12-fold).

261
The distribution of the most dominant bacterial community (at the genera level) among the 262 treatments is represented by heatmap (Fig 3). The heatmap includes the top thirty genera, which     (Fig 4). Some studies suggest that soil microbial distribution can be 353 regulated by the different vegetation types [43,44]. In this study, microbial diversity was highly In the current study, constructed-wetland showed higher SRP, NO 2 --N, NO 3 --N, and TAN 377 flux rates relative to other treatments (Fig 1). This release pattern is ascribed to promoted 378 physicochemical-microbial mediated activities such as mineralization, nitrification-379 denitrification, and redox reaction [8,19]. Literatures show that the root system of the The content of TON, TOC, TP, and TOM varied significantly among the treatments (Table   418 1). The observed distribution trend was likely due to improved bacterial community activities