Adding an anaerobic step can rapidly inhibit sludge bulking in SBR reactor

Activated sludge from wastewater treatment plants was seeded into a sequencing batch reactor (SBR) in which synthetic wastewater was used as the influent. The sludge was bulked by decreasing the concentration of dissolved oxygen (DO). By adding a 30 min step of anaerobic stirring after the water inflow, the sludge bulking was rapidly inhibited after 10 running cycles, and the sludge volume index (SVI) decreased from 222 to 74 mL·g−1. The results of high-throughput sequencing showed that the relative abundance of bacteria Thiothrix, bacteria norank_o_Sphingobacteriales and fungi Trichosporon was increased by 6.3, 4.3 and 81.2%, after initial SBR stages, but these bacteria were inhibited by the addition of an anaerobic step, as their relative abundances decreased by 0.7, 0.8 and 14.7%, respectively. The proliferation of Thiothrix, norank_o_Sphingobacteriales and Trichosporon was the primary reason for the observed sludge bulking in the reactor. After the anaerobic step was added, the sludge extracellular polymeric substances (EPS) concentration was increased from 84.4 to 104.0 mg·(gMLSS)−1 (grams of mixed liquor suspended solids). Thus, the addition of an anaerobic step can inhibit the growth of filamentous bacteria, increasing the sludge EPS concentration and promoting the precipitation of activated sludge.

Variation of sludge settleability. Sludge settleability was evaluated by determining the SVI of activated sludge. When the SVI is greater than 150 mL·g −1 , the sludge is considered to be bulked. From day 1 to day 118, the SVI of the sludge was 26~123 mL·g −1 , and the sludge had excellent settleability. From day 119 to day 153, the SVI of sludge was 166~212 mL·g −1 , and the sludge was in a bulking state. The anaerobic step was added on day 154, and after only 10 operation cycles on day 158 the sludge settleability was restored, with the SVI of the sludge gradually decreasing to 133 mL·g −1 . Thus, the addition of anaerobic step rapidly inhibited sludge bulking. The average rate at which the SVI of the activated sludge decreased was 8.9 mL·(g·cycle) −1 . The sludge settleability was restored to normal from day 158 to day 208 with an SVI of 56~133 mL·g −1 .
Microscopic examination of sludge. The sludge samples were examined by optical microscopy. As shown in Fig. 1, in sludge samples A1 and A2 a large number of evenly distributed zoogloea were observed with a very small amount of filamentous bacteria. In the A3 bulked sludge sample, very few zoogloea were observed, but an abundance of filamentous bacteria with long and straight filaments were present. The number of filamentous bacteria gradually decreased in the A4 and A5 sludge samples. For the A6 and A7 samples, in which bulking was restored, almost no filamentous bacteria were observed, and the number of zoogloea gradually increased. The microscopic examinations showed that filamentous bulking was the sludge bulking process occurring in the experiment.
Analysis of bacterial community of the activated sludge before and after adding an anaerobic step. Table 1 shows that 30,119~43,862 filtered bacterial sequences were obtained that were grouped into 604~673 Operational Taxonomic Units (OTUs), while 30,711~44,718 filtered fungal sequences were obtained that were grouped into 17~44 OTUs. The bacterial coverage index was greater than 0.996, and the fungal coverage index all reached 1.000.
The bacterial diversity of the sludge from A1 to A7 was principally composed of OTUs associated to the phylum Proteobacteria and Bacteroidetes. The phylum Proteobacteria exhibited the greatest relative abundance among all sludge bacterial phyla, with relative abundances in A1 and A2 samples of 38.4 and 37.5%, respectively. The relative abundance of Proteobacteria increased to 49.3% in the A3 bulked sludge sample, and during the restoration of settleability by adding an anaerobic step, the relative abundance of this phylum gradually increased to 69.6% in A7 sample, which exhibited fully restored settleability. The relative abundance of Bacteroidetes, Chloroflexi, Ignavibacteriae and Acidobacteria were not obvious. The relative abundance of Actinobacteria was 0.05 and 0.02% in A1 and A2 samples before bulking, respectively, increasing to 0.4% in A3 the bulked sludge sample and to 1.4% in the A4 sludge sample at the beginning of the settleability restoration. However, for the A7 sample, which had fully restored settleability, the relative abundance decreased to 0.2%. The relative abundances of the phyla Nitrospirae, Elusimicrobia and Spirochaetae were 5.1, 4.4 and 3.0% in the A1 sludge sample, decreasing to 2.1, 0.5 and 0.1% in the A3 bulked sludge sample, respectively. When the anaerobic step was added, the relative abundances of these bacterial phyla slightly increased.
In total, 342 bacterial genera were observed from sequencing sludge samples. In at least one sludge sample, 14 bacterial genera were present with relative abundances of greater than 1.0%. The relative abundance of bacterial genera is shown in Fig. 3. The primary bacterial genera observed included Candidatus_Competibacter (9.2~31.4%), norank_f_Saprospiraceae (3.0~10.5%), Zoogloea (1.8~6.4%), norank_f_env.OPS_17 (0.2~8.4%), norank_f_PHOS-HE36 (3.1~6.9%), Defluviicoccus (0.9~7.7%), norank_f_Anaerolineaceae (1.9~4.6%) and others. Candidatus_Competibacter had the highest relative abundance among all samples and was the dominant bacterial genus. The relative abundance of Candidatus_Competibacter in the A1 and A2 sludge samples was 9.2 and 10.1%, and that of Defluviicoccus was 0.9 and 1.0%, respectively. The relative abundances of these two genera increased to 13.2 and 4.9% in sample A3 to 22.6 and 7.7% in sample A6 and to 31.4 and 4.1% in sample A7, respectively. Candidatus_Accumulibacter was present at a relative abundance of 1.4% in the A2 sludge samples, decreased to 6.3% in the A3 bulked sludge sample, then gradually increased after adding the anaerobic step, reaching 1.5% in the A7 sludge sample with fully restored settleability. Thiothrix and norank_o_ Sphingobacteriales exhibited notable changes in relative abundance throughout the experiment. Thiothrix was  www.nature.com/scientificreports www.nature.com/scientificreports/ present at a relative abundance of 0.2 and 0.3% in the A1 and A2 sludge samples, respectively, which exhibited excellent settleability at the beginning of the experiment. The relative abundance of Thiothrix increased to 6.3% in the A3 bulked sludge sample, then gradually decreased after adding the anaerobic step, reaching 3.0% in the A6 and reaching 0.7% in the A7 sludge sample with fully restored settleability. The relative abundance of norank_o_Sphingobacteriales was 0.2% in the A1 sludge sample, increasing to 6.6% in the A2 sludge sample and was 4.3% in the A3 bulked sludge sample. The relative abundance of norank_o_Sphingobacteriales gradually decreased after adding the anaerobic step to 0.6 and 0.8% in the A6 and A7 sludge samples, respectively. The relative abundance of norank_f_Saprospiraceae was 8.6 and 10.5% in the A1 and A2 sludge samples, respectively, which had excellent settleability at the beginning of the experiment. Because of the impact of sludge bulking, the relative abundance of norank_f_Saprospiraceae decreased to 5.0% in the A3 bulked sludge sample and to 3.0% in the A7 sludge sample with fully restored settleability. Due to sludge bulking, the relative abundance of Nitrospira decreased from 5.1% in the A1 sludge sample to 2.1% in the A3 bulked sludge sample. This value further decreased to 1.0% in the A7 sludge sample with fully restored settleability. Therefore, the relative abundance was greatly affected by the filamentous sludge bulking.

Analysis of fungal community of the activated sludge before and after adding an anaerobic step.
Sixteen fungal phyla were observed from in the sludge samples. In at least one sludge sample, 6 fungal phyla were present with relative abundances of greater than 1.0%. The relative abundances of fungal phyla are shown in Fig. 4. The primary fungal phyla included Basidiomycota (14.7~81.2%), norank_k -Fungi (17.1~69.3%), Ascomycota (0.3~26.9%) and others. Their relative abundances of these phyla in the A1 sludge sample were 39.1, 29.5 and 26.9%, respectively, thus exhibiting a relatively even distribution. However, the relative abundances of these phyla exhibited significant changes after sludge bulking. In the A3 bulked sludge sample, the relative abundance of Basidiomycota increased to 81.2% but was only 14.7% in the A7 sludge sample with excellent settleability. The relative abundance of Ascomycota was 26.3% in the A1 sludge sample, which exhibited excellent settleability at the beginning of the experiment, decreasing to below 1.0% after sludge bulking, indicating that the  www.nature.com/scientificreports www.nature.com/scientificreports/ relative abundance was greatly affected by sludge bulking. The relative abundance of norank_k -Fungi was 29.5% in the A1 sludge sample, decreasing to 17.1% in the bulked sludge and then gradually increased after adding the anaerobic step. For example, the relative abundance of this phylum was 28.0 and 69.3% in the A6 and A7 sludge samples, respectively.
Thirty-five fungal genera were observed in the sludge samples. In at least one sludge sample, 10 fungal genera were present with relative abundances of greater than 1.0%. The relative abundances of fungal genera are shown in Fig. 5. The relative abundance of Trichosporon was 38.9% in the A1 sludge sample with excellent settleability at the beginning of the experiment, increasing to 81.2% in the A3 bulked sludge sample and decreasing to 14.7% in the A7 sample after adding the anaerobic step. The relative abundance of norank_k_Fungi decreased from 29.5% in the A1 sludge sample to 17.1% in the A3 sample, gradually increasing to 69.3% in the sample A7. However, the relative abundances of norank_o_Salpingoecidae, norank_p_Blastocladiomycota and Spumella were 1.5, 0.9 and 0.1% in the A1 sludge sample, respectively. These three groups were lower than 1.0% in the A3 bulked sludge and the A6 sludge samples, increasing to 7.6, 3.9, and 2.8% in the A7 sludge sample with fully restored settleability, respectively. Therefore, norank_o_Salpingoecidae, norank_p_Blastocladiomycota and Spumella are favourable for the improvement of sludge settleability.
Relationship between the eps concentration and settleability of the activated sludge before and after adding an anaerobic step. The sludge EPS concentration in the reactor was measured when the operation from day 109 to 171 of the experiment. Because the measured humic acid and DNA concentrations were low during the experiment, the total EPS was represented by the sum of polysaccharides and protein concentrations. The protein concentration was 26.8~74.8 mg·(gMLSS) −1 , accounting for 59.2~92.6% of the total EPS, while the polysaccharide concentration was 3.1~32.9 mg·(gMLSS) −1 , accounting for 7.4~40.8% of the total EPS. Thus, proteins were the primary component of EPS in this study.

Discussion
Filamentous sludge bulking occurred to the activated sludge. By adding a 30 min anaerobic step before the aerobic step, the amount of sludge zoogloea was increased while that of filamentous bacteria decreased. The average rate of the SVI decrease of activated sludge was 8.9 mL·(g·cycle) −1 . Only 10 running cycles were required to rapidly and effectively inhibit sludge bulking.
Microbial community analysis provided crucial information to understand adding an anaerobic step can rapidly inhibit sludge bulking. High-throughput sequencing was used to analyse the bacterial community composition of the activated sludge in the SBR reactor. The coverage index was greater than 0.996, and the fungal coverage index all reached 1.000, indicating that the depth of the sequence was sufficient. The higher the Chao index value, the richer the species richness, while the higher the Shannon index, the higher the diversity of the communities 18 . There was little change in the bacterial Chao and Shannon indices for each sample. During the sludge bulking process, the richness of the bacterial community gradually increased, but the diversity gradually decreased. After adding the anaerobic step, the bacterial community richness first decreased and then gradually increased during the sludge settleability restoration process, while the bacterial diversity initially increased and then gradually decreased. The fungal richness and diversity in sample A1 was higher than those of the A3, A6 and A7 samples. Sludge bulking influenced the fungal community in that the fungal richness and diversity decreased due to sludge bulking. After adding the anaerobic step, the fungal richness decreased, while the diversity gradually increased.
The phylum Proteobacteria exhibited the greatest relative abundance among all sludge bacterial phyla, which is consistent with the results of the bacterial communities in soil [19][20][21] and activated sludge 22,23 . Proteobacteria have wide distribution and high diversity 24 , and their primary function during wastewater treatment is to remove organic pollutants, nitrogen and phosphorus 25 . Because of the eurytopic nature of the phylum Proteobacteria, its relative abundance was increased after adding the anaerobic step. Bacteroidetes, Chloroflexi, and Acidobacteria were also often observed in activated sludge 24,26 , Bacteroidetes have strong metabolic capacity for complex organics, proteins and lipids, and can decompose complex macromolecules into simple compounds, which play an important role in the ecosystem 27 . Chloroflexi is mostly filamentous bacteria, and exists in the form of flocs www.nature.com/scientificreports www.nature.com/scientificreports/ skeleton in flocculent sludge clump inside body, play a role of sludge flocculation, at the same time with macromolecular organic matter degradation ability and has good biological phosphorus removal effect 28 . Firmicutes may be related to the degradation of COD and refractory macromolecules 22 . Research has shown that the excessive growth of Actinobacteria can induce filamentous sludge bulking 29 . The relative abundances of the phyla Nitrospirae, Elusimicrobia and Spirochaetae decreased in the bulked sludge sample, indicating that the excessive proliferation of filamentous bacteria inhibited the growth of these microorganisms.
Thiothrix and norank_o_Sphingobacteriales exhibited notable changes in relative abundance throughout the experiment, both Thiothrix and norank_o_Sphingobacteriales showed a noticeable increase. The excessive proliferation of filamentous Thiothrix can cause sludge bulking 30 .These observations indicated that the excessive proliferation of these two bacteria is the primary reason for the observed filamentous sludge bulking in the experiments.
The excessive proliferation of Thiothrix and norank_o_Sphingobacteriales in the bacterial genera resulted in sludge bulking. After adding the anaerobic step as an anaerobic selector, the growth of Thiothrix and norank_o_ Sphingobacteriales were inhibited. The sludge bulking was rapidly and effectively inhibited. Similar results were found in studies by Donkin and Nielsen. The presence of an anaerobic tank ahead of a completely mixed aerobic reactor as an anaerobic selector was to limit/suppress Thiothrix-caused bulking in dairy wastewater treatment plants 31 . Nielsen's 32 research shows that Thiothrix remained physiologically active under prolonged anaerobic conditions, but it seemed that the growth of Thiothrix was limited in such environments with an anaerobic selector ahead of a completely mixed aerobic conditions. In addition, other research shows that 12 adding the anaerobic step as an anaerobic selector generally showed good performances in controlling filamentous bacteria when treating both municipal and industrial wastewaters, namely when bulking occurred following Type 0041 overgrowth. Constantinos et al. 33 proved that under the conditions prevailing in anaerobic selector tank, the growth of M. parvicella was limited.
In activated sludge, both glycogen accumulating organi (GAO) and phosphorusaccumulatingorganism (PAO) were reported to help structuring the biomass into dense and stable aggregates 34,35 and help increasing the settleability of sludge. Most common GAO reported were the Gammaproteobacterium Candidatus_Competibacter and the Alphaproteobacterium Defluviicoccus 36,37 , most common PAO were the Betaproteobacterium Candidatus_Accumulibacter. In the anaerobic step, these two glycogen-accumulating bacteria Candidatus_Competibacter, Defluviicoccus and Candidatus_Accumulibacter adsorb large amounts of organic substrates, such as volatile fatty acids, converting them to intracellular polymers such as polyhydroxyalkanoates for storage and inhibiting the growth of filamentous bacteria. This is also beneficial for the formation of high-density sludge particles, further increasing the settleability of sludge 38 . The filamentous bulking was solved after adding the anaerobic step and a significant population of Candidatus_Competibacter, Defluviicoccus and Candidatus_Accumulibacter was established.
Although sludge bulking caused a decrease in the relative abundance of Nitrospira, members of which played an important role in the biodenitrification process, the effluent ammonia nitrogen concentration did not increase. This result occurred because the DO concentration in the mixed solution was greater than 2.0 mg·L −1 , which improved the activity of Nitrospira 39,40 and its ammonia nitrogen oxidation capability.
High-throughput sequencing was used to analyse the fungi community composition of the activated sludge in the SBR reactor also. The primary fungal phyla included Basidiomycota, norank_k -Fungi and Ascomycota, which are typically present in the activated sludge of urban wastewater treatment plants 41 . In the bulked sludge sample, the relative abundance of Basidiomycota increased and decreased in sludge sample with excellent settleability after adding an anaerobic step before the aerobic step. These results indicate that the excessive proliferation of Basidiomycota is unfavourable to the settlement of sludge, while the addition of the anaerobic step can effectively inhibit the proliferation of this fungal phylum.
The relative abundance of norank_k -Fungi was increased after adding an anaerobic step before the aerobic step, indicating that the improvement in the relative abundance of norank_k -Fungi in sludge is favourable to the settleability of sludge, and the addition of an anaerobic step is favourable for the growth of norank_k -Fungi and can promote the growth and proliferation of norank_k -Fungi, thus improving the settleability of sludge.
The relative abundance of Trichosporon was 38.9% in the sludge sample with excellent settleability at the beginning of the experiment, increasing to 81.2% in the bulked sludge sample and decreasing to 14.7% in the sample after adding the anaerobic step. Trichosporon can induce sludge bulking 6 , the excessive proliferation of Trichosporon resulted in sludge bulking in the SBR reactor in this study. After adding the anaerobic step, the growth of these bacteria and fungi were inhibited.
Akkache et al. 42 showed that the presence of EPS improved the settleability of activated sludge. However, other studies 43,44 concluded that the EPS concentration is positively correlated to the sludge SVI values and is thus unfavourable for the settleability of sludge. The variation in the EPS concentration and SVI values in the sludge in the experiment are shown in Fig. 6. From day 117 to day 131, the polysaccharide concentration in the EPS of the sludge rapidly increased from 3.61 to 25.86 mg·(gMLSS) −1 . The rapid increase in the polysaccharide concentration causes an increase of electrophoretic mobility and an increase of the repulsive force between the sludge flocs. As a result, the settleability of sludge was worsened 45 , and the SVI value increased from 123 to 201 mL·g −1 . The change in the bacterial community composition in the activated sludge from day 154 to day 161 after adding the anaerobic step resulted in a change in the EPS concentration 46

Materials and Methods
Experimental equipment and operation mode. SBR reactor with an effective volume of 11 L were used as the experimental equipment, running for 2 cycles every day with a 10 min water influent step, a 360 min aerobic step, a 60 min precipitation step, a 10 min drainage step with a drainage ratio of 38.5%, and a 290 min standing step. After adding the anaerobic step, the equipment was run with a water 10 min water influent step, a 30 min anaerobic step, a 360 min aerobic step, a 60 min precipitation step, a 10 min drainage step, and a 260 min standing step.
Seed sludge and experimental water. The seed sludge was taken at the end of the aerobic step of the oxidation ditch process in a wastewater treatment plant in the Changji city of Xinjiang.The MLSS of the seed sludge was 4,512 mg·L −1 and the SVI was 202 mL·g −1 . The experimental water used was synthetic water that simulated domestic wastewater. The primary components of the synthetic water included CH 3 COONa·3H 2 O, NH 4  Experimental process and sampling. Bulked sludge was used as the seed sludge used in this study. After excellent settleability was maintained, the experiment was initiated and lasted for 208 days. The experimental process was as follows. From day 1 to day 108, the operation status remained unchanged, with a (DO) concentration of 1.09~4.98 mg·L −1 . From day 109 to day 130, while other parameters remained constant, the DO concentration decreased to 0.20~0.65 mg·L −1 , resulting in the gradual bulking of sludge. From day 131 to day 153, the aeration rate was restored such that the DO concentration was 2.19~3.50 mg·L −1 , although the settleability of sludge was not restored, even with sufficient DO. From day 154 to day 208, after water was added to the reactor, a 30 min anaerobic step was added prior to the aerobic step.
The sampling and numbering scheme is shown in Table 2. Sludge samples A1 and A2 had excellent settleability at the beginning of the experiment; sludge sample A3 was taken when the sludge bulking occurred; sludge samples A4, A5 and A6 were taken from the stage when the settleability of sludge was restored after adding the anaerobic step; and sludge sample A7 had fully restored settleability after the addition of an anaerobic step.  www.nature.com/scientificreports www.nature.com/scientificreports/ Analytical methods for water quality and sludge monitoring. Ammonium and COD were analyzed according to standard procedures. DO was measured with a portable DO analyser; pH was measured with a pH test pen; and the water temperature was measured with a mercury thermometer. Sludge volume index (SVI) were determined by reading the volume of the settled bed in a column after 30 min settling and calculated from the dry weight in (MLSS). Microscopic observations were performed with a photonic microscope. The morphology of filaments and flocs was evaluated on a day-to-day basis.

EPS extraction methods and component analysis methods. EPS was extracted through an ethylene
diamine tetraacetic acid (EDTA) method, the activated sludge was harvested by centrifugation at 5600 rpm for 10 min at 4°C, the collected activated sludge was re-suspended in fresh sterile and harvested by centrifugation at 5500 rpm for 20 min at 4°C, and the collected activated sludge was re-suspended in fresh sterile and 2%EDTA (1:1) at 4 °C for 3 h, then harvested by centrifugation at 12300 rpm for 20 min at 4°C.Extractant residues in the solution were removed by the dialysis membrane filtration in the subsequent treatment. The supernatants were collected as EPS solution and stored at 4 °C or freeze-dried for further analysis.The carbohydrate content in EPS was measured by the anthrone method 48 using glucose as the standard. The contents of protein in EPS were measured by the modified Lowry method 49 using bovine serum albumin as the respective standards. Data analysis. Data analysis was conducted using the i-sanger platform (http://www.i-sanger.com/) provided by Majorbio Bio-Pharm Technology Co. Ltd. (Shanghai, China). The microbial phylotype richness levels were calculated using the Chao estimator, the Shannon diversity index. The Chao estimator, the Shannon diversity index and the coverage percentage were also calculated by the Mothur program version v.1.30.1 (http:// www.mothur.org/wiki/Schloss_SOP#Alpha_diversity). These analyses were performed using the R Programming Language software.