The microbial community in filamentous bulking sludge with the ultra-low sludge loading and long sludge retention time in oxidation ditch

Sludge bulking is a major problem that restricts the development of the activated sludge process. The microbial community responsible for sludge bulking varies depending on water quality and operational conditions. This study analysed the microbial community of bulking sludge in oxidation ditch with ultra-low sludge loading and long sludge retention time using high-throughput sequencing. The study found that the relative abundance of bacterial genus Saprospiraceae_norank was the highest in bulking sludge, reaching 13.39–28.83%, followed by Comamonadaceae_unclassified, Ardenticatenia_norank and Tetrasphaera, with the relative abundance of 4.59–11.08%, 0.52–16.60% and 0.17–8.92% respectively. In contrast, the relative abundance of bacteria that easily caused sludge bulking including Microthrix (0.54–2.47%), Trichococcus (0.32–1.71%), Gordonia (0.14–1.28%), and Thiothrix (0.01–0.06%) were relatively low. Saprospiraceae_norank was predominant and induced sludge bulking in oxidation ditch. The relative abundance of fungal genus Trichosporon was the highest in bulking sludge, reaching 16.95–24.98%, while other fungal genera were Saccharomycetales_unclassified (5.59–14.55%), Ascomycota_norank (1.45–13.51%), Galactomyces (5.23–11.23%), and Debaryomyces (7.69–9.42%), whereas Trichosporon was the dominant fungal genus in bulking sludge. This study reported that excessive Saprospiraceae_norank can induce sludge bulking for the first time, which provides important knowledge to control sludge bulking.

Factors such as water temperature 22 , dissolved oxygen (DO) 3 , sludge retention time (SRT) 23 , pH 24 , influent quality 25 , nutrient ratio 26 and sludge loading 27 are responsible for filamentous sludge bulking. The microbial community responsible for sludge bulking varies depending on the water quality and operational conditions. For example, for bacterial communities, Microthrix proliferated at low sludge loading and low temperature 6,28 , whereas Eikelboom type 021N induced sludge bulking at high sludge loading and high temperature 10 . Flavobacterium proliferated and caused sludge bulking at low influent carbon/nitrogen(C/N) ratio and long hydraulic retention time (HRT) 11 . The mass propagation of Haliscomenobacter hydrossis caused sludge bulking and resulted in high sludge loading and long SRT 29 , whereas Nocardia induced sludge bulking when the sludge loading was less than 0.5 kg BOD 5 /(kg MLSS·d) 13 . Thiothrix proliferated and caused sludge bulking at high chemical oxygen demand(COD) concentration, low DO and low nutrient 30 . Excessive proliferation of Tetrasphaera and Trichococcus Nostocoida limicola I caused sludge bulking at low temperature 31 . Further, Beggiatoa proliferated and resulted in sludge bulking when the sludge loading was less than 0.51 kg BOD 5 /(kg MLSS·d) and the DO lower than 1.5 mg/L 17 . For fungal communities, excessive propagation of Trichosporon caused sludge bulking at low DO 18 , while Geotrichum caused sludge bulking at low pH and high temperature 19 .
Although filamentous bacteria causing sludge bulking under different operational conditions have been widely investigated, sludge bulking is still a major problem hindering the operation of the activated sludge process. High-throughput sequencing is a revolutionary reform to traditional sequencing since the former does not require a pure culture and can sequence hundreds of thousands to millions of deoxyribonucleic acid (DNA) molecules rapidly and accurately 32 . In this study, the microbial community in the sludge collected from an oxidation ditch that has been experiencing sludge bulking constantly in recent years was analysed using high-throughput sequencing technology with the ultra-low sludge loading and long SRT. The outcomes of this study are expected to provide valuable knowledge required to control the sludge bulking.

Results
Treatment efficiency of the WWTP. The removal efficiencies of the WWTP from January 2016 to January 2018 are presented in Table 1. The average influent biochemical oxygen demand (BOD) to COD ratio was 0.46, indicating a good biochemical property of sewage. The COD, BOD 5 , suspended solids (SS), total phosphorus (TP), and total nitrogen (TN) in the design influent to the WWTP were 450 mg/L, 200 mg/L, 250 mg/L, 5 mg/L, and 40 mg/L. As evident in Table 1, the actual influent COD, BOD 5 , SS, TP, and TN concentrations during the sampling period were 2-3 times higher than the design influent concentrations in the WWTP. The transformation of influent TN may generate substantial levels of free ammonia and free nitrous acid, which can have adverse impacts on microbial community 33,34 . The effluent from the WWTP met the second-level discharge standard 35 . Despite the sludge bulking, the sewage treatment efficiency was good. Activated sludge settling property. As shown in Fig. 1, SVI of the activated sludge samples were 162-250 mL/g indicating poor settling property and the activated sludge in the oxidation ditch was in the state of constant sludge bulking. Furthermore, filamentous sludge bulking occurred in the WWTP according to the microscopic examination.
Bacterial community analysis based on 16S rRNA sequencing. The total effective readings of the seven bulking sludge samples were between 30457 and 55170. The coverage indexes of all samples were more than 0.988, indicating the detection of most bacterial communities in this sequencing with high data reliability. The operational taxonomic units (OTUs), Chao, Shannon values are presented in Table 2. The Chao and Shannon indexes represent the richness and diversity of the microbial community, respectively. Higher Chao index indicates higher species richness and higher Shannon index suggests higher diversity of the communities 36 . In January 2016 (CJ1), the SVI value was the largest, while the Chao and Shannon values were the lowest, suggesting the lowest richness and diversity of the bacterial community. In January 2018 (CJ7), SVI value was the lowest, whereas the Chao and Shannon values were the highest, indicating the highest richness and diversity of the bacterial community. The richness and diversity of the bacterial community are lower when significant sludge bulking occurred.
A total of 35 bacterial phyla were detected in seven sludge samples. In at least one sample, there were 12 bacterial phyla with relative abundances of over 1%, accounting for 97.05-99.25% of the total bacterial effective sequences (Fig. 2) www.nature.com/scientificreports www.nature.com/scientificreports/ 531 bacterial genera were present in seven sludge samples. In at least one sample, there were 73 bacterial genera with relative abundances of over 0.5%, accounting for 78.32-83.21% of the total bacterial effective sequences (Fig. 3). The dominant bacterial genera observed included Saprospiraceae_norank (11.87-28.83%), Comamonadaceae_unclassified (4.59-11.08%), Ardenticatenia_norank (0.52-16.60%) and Tetrasphaera (0.17-8.92%). The relative abundance of filamentous bacteria related to sludge bulking such as Microthrix (0.54-2.47%), Trichococcus (0.32-1.71%), Gordonia (0.14-1.28%) and Thiothrix (0.01-0.06%) was relatively low, among which Saprospiraceae_norank was predominant in all bacterial genera.    Table 3. In December 2016 (CJ3), SVI value was the largest, while the Chao and Shannon values were the lowest, suggesting the lowest richness and diversity of the fungal community. In January 2018 (CJ7), the SVI value was the smallest, whereas the Chao and Shannon values were the highest, indicating that the highest richness and diversity of the fungal community. The richness and diversity of the fungal community are generally lower for significant sludge bulking to occur.

Discussion
The dominant bacterial phyla obtained were Bacteroidetes (25.86-47.56%), Proteobacteria (21.98-37.77%), Chloroflexi (4.28-24.96%), and Actinobacteria (3.29-14.12%) and are not significantly different from the dominant bacterial phyla in the oxidation ditch process of WWTP in China [37][38][39] . However, the relative abundance of these phyla is different. Bacteroidetes, which plays an important role in wastewater treatment by degrading macromolecular organic pollutants 40 , were present at 25.86-47.56% in bulking sludge samples, and were the dominant bacterial phylum. Kragelund et al. found that the relatively high abundance of Bacteroidetes can cause sludge bulking problems 41 . Proteobacteria, which is a conventional bacterial phylum in WWTPs with the ability to degrade organic pollutants and remove nutrients such as biological nitrogen and phosphorus 42 , were present at 21.98-37.77% in all bulking sludge samples. Xu et al. found that Proteobacteria (33.90-50.90%) was the dominant bacterial phylum in the oxidation ditch without sludge bulking 38 . Chloroflexi is chiefly filamentous bacteria, which exists in flocculent sludge clump inside the body in the form of flocs skeleton. It plays a role in sludge flocculation, but rarely induces sludge bulking 43 . The relative abundance of Chloroflexi was between 4.28% and 24.96% in all bulking sludge samples. Furthermore, the mass proliferation of Actinobacteria can cause filamentous sludge bulking 6 . Wang et al. found that Actinobacteria was dominant with a relative abundance of 50% in   www.nature.com/scientificreports www.nature.com/scientificreports/ the WWTP, where excessive sludge bulking occurred 31 . In this study, the relative abundance of Actinobacteria   www.nature.com/scientificreports www.nature.com/scientificreports/ (≤12-15 °C) and is the dominant bacterial genus responsible for sludge bulking in cold areas 28 . Xinjiang is a dry and cold region with the winter lasting for five months. The influent temperature remains at 7-15 °C in winter and 16-24 °C in summer in the WWTP. However, in this study, Saprospiraceae_norank was the predominant bacterial genus in oxidation ditch bulking sludge and its relative abundance varied between 11.87% and 28.83%, whereas Yang et al. found that the relative abundance of Saprospiraceae_norank was between 2.10% and 3.53% in non-bulking activated sludge in WWTP 21 . Muszynski et al. reported that the abundance of Saprospiraceae_ norank was dependent on the season 45 . Additionally, Saprospiraceae_norank, which existed in sludge flocs and is capable to produce extracellular enzymes to degrade protein and is crucial for partial nitrification, denitrification and sludge fermentation 46 . Saprospiraceae_norank belongs to phylum Bacteroides, class Sphingoleifera, order Sphingoleiferae, and family Saprospiraceae. Shchegolkova et al. found that Saprospiraceae_norank was the inductor of activated sludge bulking and foaming 47 . Yao et al. found that sludge bulking was inhibited due to the addition of an anaerobic step 48 . In this study, Saprospiraceae_norank was the predominant bacterial genus that induced sludge bulking in oxidation ditch, while the relative abundance of Microthrix was only between 0.54% and 2.47% in bulking sludge, and was far less than the relative abundance reported when the proliferation of Microthrix caused sludge bulking in WWTPs. Microthrix was not the dominant bacterial genus that caused sludge bulking in oxidation ditch in WWTP.
Tetrasphaera belongs to Actinobacteria and plays a role in the biological phosphorus removal in WWTPs 49 . According to Wang et al., Tetrasphaera (6.75%) was generally found in activated sludge systems, where filamentous sludge bulking occurred at low temperature and contributed to sludge bulking in WWTPs 31 . The relative abundance of Tetrasphaera was between 0.17% and 8.92% in all bulking sludge samples. Trichococcus (3.91%) was the dominant filamentous bacterial genus that caused sludge bulking in WWTP at low temperature and low DO 16,31 . The relative abundance of Trichococcus (0.32-1.71%) was low and was not the dominant filamentous bacterial genus in oxidation ditch bulking sludge. Similarly, Gordonia (5.1%) was the dominant bacterial genus in WWTPs with low DO, long SRT and low temperature 16 . A long-term study that was conducted to identify the dominant filamentous bacteria in a full-scale WWTP found that Thiothrix (51.9%) was the dominant filamentous bacterial genus with the high COD concentration, low DO and nutrient deficits 30 . However, in this study, the relative abundance of Gordonia (0.14-1.28%) and Thiothrix (0.01-0.06%) were low and were not the dominant filamentous bacterial genera in oxidation ditch bulking sludge. Speirs et al. studied the bacterial community structure in oxidation ditch of a WWTP with severe sludge bulking in South Australia and found that the dominant bacterial genus was Type 0914 (35%) with the long SRT 50 Table 4. Though Past studies did not report that excessive proliferation of Saprospiraceae_norank can induce sludge bulking, this study found that excessive Saprospiraceae_norank can induce sludge bulking for the first time.
Ascomycota (58.83-69.69%) and Basidiomycota (23.71-25.68%) were the dominant fungal phyla obtained by fungal sequencing in WWTP, which is consistent with the results obtained in other WWTPs in China. Ascomycota and Basidiomycota were conventional fungal phyla in activated sludge of WWTPs and Ascomycota (51.82%) was dominant in all fungal phyla 51 . Basidiomycota mainly affects the formation of sludge flocs by reducing sludge settling property 52 and thereby inducing sludge bulking. In this study, Basidiomycota was the dominant fungal phylum that caused sludge bulking in the oxidation ditch.
Trichosporon (16.95-24.98%) was the dominant fungal genus obtained by fungal sequencing and was responsible for inducing sludge bulking 53 , whereas Trichosporon (25%) was the dominant fungal genus when sludge bulking occurred in WWTP 18 . This is consistent with the dominant fungal genus obtained in this study. Further, Trichosporon induced sludge bulking in a sequencing batch reactor (SBR) in a previous laboratory study 14 . The study found that Trichosporon was the filamentous fungal genus causing sludge bulking under different operational conditions. Saccharomycetales_unclassified (5.59-14.55%) and Debaryomyces (7.69-9.42%) are common yeasts found in the oxidation ditch of WWTPs in China 54 . Saccharomycetales_unclassified and Debaryomyces can produce hydrolytic enzymes and collectively degrade some pollutants in sewage 55 . In this study, Saccharomycetales_unclassified and Debaryomyces existed in all bulking sludge in the oxidation ditch of the WWTP. However, they are not filamentous fungi and do not generally induce sludge bulking. Galactomyces is a common filamentous fungal genus in activated sludge of WWTPs and can induce sludge bulking 20 . Comparatively high relative abundance of Galactomyces (5.23-11.23%) in oxidation ditch bulking sludge could contribute to the sludge bulking in WWTP, while the relative abundance of Geotrichum 19 related to sludge bulking was low,  www.nature.com/scientificreports www.nature.com/scientificreports/ only 2.17-3.02%. In other words, Trichosporon was the dominant filamentous fungal genus when sludge bulking occurred in WWTP, which was consistent with past study results. In this study, sludge bulking in WWTP occurred due to high influent pollutant concentrations, ultra-low sludge loading and long SRT. They are discussed in detail below.

Materials and Methods
Description of the WWTP and sample collection. The WWTP located in the Changji city of Xinjiang, northern China, with the treatment design capacity of 10 × 10 4 m 3 /d uses the carrousel oxidation ditch process and has been operating since 2000. The influent of Changji WWTP is mainly domestic wastewater. Activated sludge samples, CJ1, CJ2, CJ3, CJ4, CJ5, CJ6, and CJ7, were collected from the end of the aerobic stage of the oxidation ditch. Sampling date, sludge index volume and operating parameters of WWTP are presented in Table 5.
The SVI values of the samples were more than 150 mL/g, confirming that all samples were bulking sludge.
Analysis methods. COD, BOD 5 , TN, TP, SS and mixed liquor suspended solids (MLSS) were assayed according to the standard method 58 . The temperature was measured using a thermometer. SVI values were determined by reading the percentage of sludge volume in the mixture of water and sludge after 30 min settling in a 100 mL measuring cylinder and counted from the dry weight in MLSS. Microscopic examination was conducted using a photonic microscope. The morphology of activated sludge filaments and flocs was characterized daily.  China) for conducting data analysis. The Chao estimator and the Shannon diversity index were used to calculate the microbial phylotype richness levels. The Mothur program version v.1.30.1 (http://www.mothur.org/wiki/ Schloss_SOP#Alpha_diversity) was used to calculate the Chao estimator, the Shannon diversity index, and the coverage percentage. These analyses were performed using the R Programming Language software.  Table 5. Sampling date, sludge settling property, sludge concentration and water temperature of the operation.