Effects of substrate on shrimp growth, water quality and bacterial community in 1 the biofloc system nursing

This study aimed to investigate the effects of substrate on water quality, shrimp growth 16 and bacterial community in the biofloc system with a salinity of 5‰. Two treatments, 17 biofloc system with (sB) or without (nB) addition of elastic solid packing filler (nylon) 18 as substrate, were set up. Penaeus vannamei postlarvae (PL, ~ stage 15) were stocked at 19 a density of 4000 PL m -3 in each treatment with triplicates for a 28-days culture 20 experiment, taking glucose as carbon source (C:N 15:1). Results showed that the 21 survival rate (96.3±3.6%), FCR (0.76±0.06) and productivity (1.54±0.12 kg m -3 ) in sB 22 treatment were significantly better than those in nB treatment (81.0±7.1%, 0.98±0.08 23 and 1.14±0.09 kg m -3 , P <0.05). All water parameters were in the recommended ranges. 24 Substrate showed significant effect on TAN, TSS, turbidity, biofloc volume, pH and 25 carbonate alkalinity ( P < 0.05). Actinobacteria (4.0-22.7%), Bacteroidetes 26 (10.4-33.5%), Firmicutes (0.2-11.2%), Planctomycetes (4.0-14.9%) and Proteobacteria 27 (29.4-59.0%) were the most dominant phyla for both treatments. However, the bacterial 28 community in sB treatment showed to be significantly different from that in nB 29 treatment (Jaccard distance 0.94±0.01, P =0.001). Substrate showed significant effects 30 on Shannon, Heip, Pielou and Simpson index, as well as relative abundances of 31 Actinobacteria, Bacteroidetes and Planctomycetes ( P < 0.05). The results suggested 32 that addition of substrate affected the shrimp growth, water quality and bacterial 33 community in the biofloc system nursing P. vannamei PL with a 5‰ salinity.

control the problems usually appeared in the traditional prenursery system, such as 57 biosecurity and toxic ammonia and nitrite (Samocha, 2010), due to the advantages of 58 this technology on nitrogen assimilation in situ and pathogen control under minimal or 59 Thirty PL were selected randomly and individually weighed to the nearest 0.1 mg with 145 an electric balance (AUX220, Shimadzu, Japan) each week. Survival rate, weekly 146 increment of body weight (wiW), specific growth rate (SGR), feed conversion rate 147 (FCR) and productivity were calculated according the following formulates: 148  Ltd. The raw data produced from high-throughput sequencing has been deposited at 172 NCBI with accession number of PRJNA646765. 173 Data processing for the high-throughput sequencing data was carried out under the 174 QIIME 2 (Quantitative Insights Into Microbial Ecology, Version 2019.10) framework 175 (Bolyen et al., 2019). In brief, ambiguous nucleotides, adapter sequences and primers 176 contained in reads, and short reads with length less than 30 bp were removed with the 177 cutadapt plugin (Martin, 2011). After that, bases in the two ends of reads with quality 178 score lower than 25 were trimmed. Thereafter, chimeras were filtered, and pair-ended 179 reads were joined, dereplicated, to obtain high-quality reads which were clustered to 180 operational taxonomic units (OTUs) with an identity of 0.97 by using the Vsearch tool 181 (Rognes et al., 2016). Thereafter, the counts of OTUs were normalized by 16S rRNA gene copy number based on rrnDB database (version 5.6) with the QIIME 2 plugin of Data was statistically analyzed with the SPSS platform for windows (

Water quality 213
The levels of dissolved oxygen and temperature in the present study were above 5.0 mg 214 L -1 and 26.0 o C, respectively (Table 1). The effects of time on both parameters were 215 significant (P < 0.05, Table 1). Whereas, substrate showed no significant effect on the 216 levels of dissolved oxygen and temperature (P > 0.05, Table 1). 217 The carbonate alkalinity in sB treatment was maintained at a high level of 218 306.5±51.7 mg L -1 CaCO 3 , although it was lower than that in nB treatment (373.7±12.4 219 mg L -1 CaCO 3 , Table 1). The pH values in nB treatment and sB treatment were 7.20±0.14 and 7.11±0.03, respectively (Table 1). Both parameters were significantly 221 affected by substrate and time, as well as their interactions (P < 0.05, Table 1). 222 The mean concentrations of the three inorganic nitrogen compounds were at low 223 levels in the present study (< 1.0 mg L -1 , The biofloc volume (BFV) levels in nB treatment and sB treatment were similar, 233 with a value of 8.8±2.3 and 8.3±2.3 mL L -1 , respectively (Table 1). However, the total 234 suspended solids (TSS) and turbidity in sB treatment (491.3±150.2 mg L -1 , and 235 299.5±92.0 nephelometric turbidity units, NTU) were higher than those in nB treatment 236 (148.5±31.3 mg L -1 and 111.9±56.3 NTU, Table 1). Substrate showed significant main 237 effects on those three parameters (P < 0.05, Table 1). The effects of time on BFV and 238 turbidity were also significantly, as well as the interaction of substrate and time on BFV 239 (P > 0.05, Table 1). 240

Growth performance 241
During the 28-days culture experiment, the average body weights of shrimp in sB 242 treatment were higher than those of nB treatment ( Fig. 1). At the end, although the final 243 body weight in sB treatment (0.40±0.03 g) was not significantly different from that of B 244 treatment (0.36±0.04 g, P = 0.596, Table 2), the survival rate (96.3±3.6%) and 245 productivity (1.54±0.12 kg m -3 ) of the former treatment were significantly higher than 246 those of the latter (81.0±7.1% and 1.14±0.09 kg m -3 , P < 0.05, Table 2). The feed 247 conversion rate (FCR) in sB treatment (0.76±0.06) was significantly lower than that in 248 nB treatment (0.98±0.08, P = 0.044, Table 2). There no significant difference was 249 observed between both treatments for the weekly increment of body weight (wiW) and 250 specific growth rate (SGR) of shrimp during the culture experiment (P > 0.05, Table 2). 251

Bacterial diversity 252
The Shannon index for bacterial community in sB treatment (6.77±0.18) was 253 higher than that in nB treatment (6.14±0.07, Table 3). Substrate, time and their 254 interaction showed significant effects on this index (P < 0.05, Table 3). The Shannon 255 index in sB treatment obtained the highest value at 7 d and then slightly decreased, but 256 that in nB treatment peaked at 21 d (Fig. 2 a). 257 Higher OTU counts (6238.3±353.8) and Margalef index (387.7±28.0) was 258 observed in sB treatment, when compared to that in nB treatment (5713.5±368.2 and 259 343.9±22.0, Table 3). Both indexes significantly affected by time (P < 0.05), but not by 260 substrate (P > 0.05, Table 3). Those two indexes showed similar changing trends with 261 the Shannon index in each treatment (Fig. 2 b and c). 262 The evenness indexes, Heip index and Pielou index, in sB treatment (0.018±0.001 263 and 0.538±0.003) and nB treatment (0.013±0.001 and 0.493±0.01, Table 3) peaked at 264 14 d and 21 d, respectively (Fig. 2 d and e). The main effects and interactions of 265 substrate and time on both indexes were significant (P < 0.05), with exception of the 266 interaction on Heip index (P = 0.544, Table 3). 267 The Simpson index in sB treatment (0.955±0.003) was higher than that in nB 268 treatment (0.927±0.010, Table 3). This index in sB treatment displayed a peak value at 269 14 d, but that in nB treatment showed a decreasing trend during experiment (Fig. 2 f). 270 Substrate and time significantly affected this index (P < 0.05, Table 3). Whereas, the 271 Berger-Parker index in sB treatment (0.152±0.014) was lower than that in nB treatment 272 (0.193±0.020, Table 3). Substrate was not significantly affected this index (P > 0.05, 273 Table 3). The change trend of Berger-Parker index in the present study was contrary to 274 that of the Simpson index in each treatment (Fig. 2 g). 275 The bacterial community distances within nB treatment and sB treatment were 276 0.76±0.36 and 0.77±0.37, respectively (Table 4). PCA analysis also showed that water 277 samples collected from the same treatment were closer than those from the other 278 treatment, basing on the OTUs composition (Fig. 3). In addition, the bacterial beta 279 diversities between both treatments were significantly different with a Jaccard distance 280 of 0.94±0.01 (P = 0.001, PERMANOVA, Table 4). 281

b). The main effects and 290
interactions of substrate and time on proportions of those seven phyla were significant 291 (P < 0.05), except the substrate effects on Firmicutes and Proteobacteria, and the 292 interaction of substrate and time on Planctomycetes (P > 0.05, Table 5). 293 The bacterial composition profiles on class, order, family and genus levels were 294 showed in Fig. 1S-Fig. 4S, respectively. 295 The LEfSe analysis showed that an unassigned genus and the phylum 296 Verrucomicrobia were the most significant biomarkers for nB and sB, respectively ( In the present study, it was found that the turbidity levels in both treatments were 319 increased continuously. This might be contributed to the removal operation for biofloc 320 during 14-28 d of the experiment in the present study. The turbidity in the present study 321 was determined on water sample without big-size biofloc which settled by using an 322 Imhoff cone for 15 min, indicating that this parameter was correlated to the content of 323 small-size biofloc. Previous studies have showed that the bacterial groups are different 324 between small-and big-size bioflocs (Chen et al., 2019; Huang et al., 2020). Therefore, the removal operation in this study might make settleable solids (big-size biofloc) as 326 well as bacterial groups attached be removed from the water column, which would 327 improve the growth of bacteria adhering on small-size biofloc retained in the water 328 body, by reducing the competition from bacteria relating with big-size biofloc, and in 329 turn, prompting formation of small-size suspended biofloc and increasing the turbidity. 330 Substrate showed significant effect on turbidity in the present study (P = 0.002). It 331 was thought that the addition of artificial substrates may influence the water circulation 332 in the tanks, leading to smaller turbulence in the water, and in turn to facilitating 333 sedimentation or particle aggregation and formation of large flocs, which reduces the 334 suspended solids and thus the turbidity in the water column (Ferreira et al., 2016; 335 Fleckenstein et al., 2020). However, in the present study, the turbidity level in sB 336 treatment was found to be higher than that in nB treatment. It was speculated that 337 substrate might play a positive role in the process of turbidity increasing caused by the 338 removal operation for big-size biofloc discussed above. 339 Substrate also showed significant effect on carbonate alkalinity in the current study 340 (P < 0.001). And the carbonate alkalinity in sB treatment was lower than that in nB 341 treatment in the current study, indicating that more alkalinity was consumed in biofloc