Introduction

High-rate algal oxidation ponds (HRAOP) are a fundamental component of integrated algal pond systems (IAPS), an advanced configuration of algal integrated wastewater pond systems (AIWPS) pioneered by Oswald and others (Ludwig et al. 1951; Oswald et al. 1953a, b; Oswald 1991, 1995). The demonstration of IAPS in South Africa has shown that this nature-based technology solution has merit for use in municipal sewage treatment and the treatment of wastewater from various industrial and agricultural sources, with the potential for resource recovery (Mambo et al. 2014; Jimoh and Cowan 2017; Laubscher and Cowan 2020). The incorporation of HRAOP in the design of IAPS was to specifically promote the uptake and assimilation of organic matter and nutrients into biomass coupled with photosynthetic oxygenation of the effluent for the secondary treatment of wastewater (Oswald 1991; Green et al. 1995). Beyond nutrient removal, the design and performance of the HRAOP also favour the aggregation of microbiota to form microalgal-bacterial flocs (MaB-flocs). Aggregated MaB-flocs are assemblages of microalgae and bacteria loosely joined to form unstructured aggregates (Chen et al. 2022). The in-situ formation of MaB-flocs is critical to the performance of IAPS. It is directly responsible for enhancing the quality of the final treated water to meet regulatory standards and facilitate cost-effective valorization of the harvested biomass (Van Den Hende et al. 2014; Jimoh and Cowan 2017; Ramanan et al. 2016; Dube and Cowan 2023).

By design, the microbial community structure and HRAOP performance tend to be affected by operational and environmental factors. At high HRT and low F/M ratio, microalgal cells reach a steady state of growth that allows for aggregation and incorporation of individual cells into MaB-flocs, reducing the concentration of juvenile and planktonic cells (Gutzeit et al. 2005; Medina and Neis 2007). Park et al. (2013) also reported that recycling of gravity-harvested biomass to HRAOP promoted the formation of flocs >500 µm, which enhanced productivity, harvest efficiency and energy yield. In addition, research in our laboratory using a 500-person equivalent (PE) demonstration-scale IAPS supplied continuously with municipal sewage has demonstrated extracellular polymeric substance (EPS) production by the consortium, a role for these macro-molecular structures in the aggregation process, flocculation, floc stability, settling, and improved water quality (Jimoh and Cowan 2017; Jimoh et al. 2019).

Macro-molecular EPS comprises various metabolites that contribute to the growth dynamics and microbial interaction for successful attachment, aggregation, and adhesion for floc formation within an aquatic microenvironment (Jimoh et al. 2019; Perera et al. 2021; Ashraf et al. 2023). Despite the role of EPS in MaB-floc formation, the core microbial participants responsible for aggregation to ensure settleable MaB-flocs remain largely unidentified. Furthermore, while changes in season, temperature, irradiance, pH, etc., are understood to affect microbial population dynamics in HRAOP (Park et al. 2011; Sutherland et al. 2014), whether these factors influence MaB-floc formation and stability has yet to be demonstrated.

This study was carried out to firstly establish potential correlations between the community structure of the microbiota in the HRAOP of an IAPS and the environmental/operational parameters that impact the composition, formation, and stability of MaB-flocs in the system during domestic sewage treatment. A second purpose of this study was to determine whether a species associated with disaggregation of MaB-flocs could be identified by monitoring biotic/abiotic-induced changes in microbial community structure. This was achieved using metagenomics to monitor the seasonal succession of microalgal and bacterial abundance and diversity within the HRAOP over one year. A principal components analysis (PCA) was used to reduce the dimensionality of the biotic (i.e., MaB-floc core microbial components) and abiotic (i.e., environmental/operational) variables and to determine which variables most strongly correlate to influence MaB-floc stability in HRAOP.

Materials and methods

Location and operation of the integrated algal pond system

The IAPS used in this study was located at the Belmont Valley Wastewater Treatment Works (WWTW), Makhanda, South Africa (33° 19׳ 07״ South, 26° 33׳ 25״ East). The 500 PE-designed demonstration system was operated at an average flow rate of 75 m3 day-1. The design, process flow, operation and performance have been described elsewhere (Mambo et al. 2014; Jimoh and Cowan 2017; Laubscher and Cowan 2020; Titilawo et al. 2021; Dube and Cowan 2023). Briefly, the system comprised an advanced facultative pond (AFP), two HRAOP each with its own algal settling pond (ASP) connected in series. The 840 m2 AFP with total volume of 1500 m3 and HRT of 20 days, incorporated a 225 m3 in-pond digester (IPD) located at least 3 m below surface water level, where anaerobic biodegradation of suspended and dissolved solids could take place. Each HRAOP, connected to a 12.5 m2 ASP, had a surface area of 500 m2, a total volume of 150 m3 and an operating depth of 0.3 m. Mixing and turbulent flow essential for biomass productivity and MaB-floc formation were achieved using paddlewheels powered by electric motors (0.25 kW). Effluent from the AFP decanted under gravity to the first HRAOP with HRT of 2 days, and then to the first ASP for half a day. Partially treated effluent entered the second HRAOP where it was detained for 4 days and gravity-fed to a second ASP with HRT of 0.25 d. MaB-flocs were separated from the treated water by gravity and the final treated effluent returned to the municipal WWTW.

Sampling procedures

Sampling was carried out fortnightly between February and December 2019. Mixed liquor containing suspended MaB-flocs (i.e., MLSS) was collected from the second of two HRAOPs directly in front of the paddlewheel between 16:00 h and 17:00 h (Jimoh and Cowan 2017). Grab samples were immediately transferred to the laboratory or stored at 4°C for analysis the following day. For next-generation sequencing (NGS) analysis of microbial community structure, six samples were collected during three seasonal windows within the experimental period. Two samples were collected in late summer (February - March 2019, S1 and S2), another two in the winter (June - July 2019, S3 and S4), and the last two in the spring (November - December 2019, S5 and S6).

Physicochemical analysis

Physicochemical parameters, including water temperature, pH, and electrical conductivity (EC), were measured in situ during the study period. Temperature and EC were measured using an EC Testr11 Dual range 68X 546 501 meter (Eutech Instruments, Singapore), while pH was measured using a microcomputer pH meter (Eutech Instruments, Singapore). Meteorological data, including atmospheric temperature and direct normal irradiance (DNI) for the study period, were obtained from the Southern African Universities Radiometric Network (SAURAN, Brooks et al. 2015) and the University of Fort Hare (32° 47׳ 13.4״ south, 26° 50׳ 56.7״ east, elevation: 540 m a.s.l.).

Pond productivity, MaB-floc stability and settling

Biomass from the HRAOP (measured as the concentration of MaB-flocs present in MLSS samples) and aerial productivity (g m-2 day-1) were estimated as previously described (Jimoh and Cowan 2017). The settleability of MaB-flocs was determined according to standard methods (APHA 1998; 2540 F) using Imhoff cones and a settling time of 1 h.

Metagenomics

Aliquots of the collected MLSS were filtered successively using 0.45 and 0.22 µm membrane filters to first capture MaB-flocs and other larger colonial microalgae and microbes and then to capture the smaller microbes, including bacteria. Retained biomass from both filters was pooled, and the total genomic DNA extracted using a DNA isolation kit (Zymo Research), according to the manufacturer’s instructions. The 16S rRNA gene in the extracts was targeted for amplification using the universal primer pair 515f (5′-GTGYCAGCMGCCGCGGTAA-3′) and 926r (5′-CCGYCAATTYMTTTRAGTTT-3′) at a region of 450 bp. The 18S rRNA target site was amplified using universal primer pair 1391f (5′-GTACACACCGCCCGTC-3′) and EukBr (5′-TGATCCTTCTGCAGGTTCACCTAC-3′) at a region of approximately 260 bp.

The mixture prepared for amplification contained 5 µL of extracted genomic DNA, 12.5 µL master mix, 0.5 µL of each of the forward and reverse primers, and 6.5 µL molecular grade water to make a final reaction volume of 25 µL. The 16S rRNA gene was subjected to initial denaturation at 94°C for 3 min, followed by 35 cycles at 94°C for 45 s, 50°C for 60 s, and 72°C for 90 s of denaturation, annealing, and extension, respectively, and a final extension at 72°C for 10 min using an Applied Biosystems SimpliAmp Thermal Cycler (Thermo Fisher Scientific, USA). The 18S rRNA target was amplified by initial denaturation at 90°C for 15 min, followed by 94°C for 60 s, annealing at 60°C for 60 s, and 72°C for 60 s for extension, and a final extension at 72°C for 7 min. The quality of the amplified products was checked by agarose gel electrophoresis, and the purity of the DNA bands tested using a FavorPrep gel purification mini kit (Favorgen, Austria) according to the manufacturer’s instructions. The purified products were sequenced using GS Junior Titanium Sequencing Platform (454 Life Sciences, Roche).

Data curation

Datasets were curated using Mothur 1.41.3 to remove low-quality reads and ambiguous nucleotides from the set of sequences (Schloss et al. 2009). After taxonomic identification, chimeric sequences were also identified and removed using the UChime algorithm. Classification and alignment of sequences were done using SILVA (version 132) as a reference database. Following these, a distant matrix of 0.03 was created on the curated datasets to cluster sequences into operational taxonomic units (OTU) at a similarity level of 97%. Following the removal of singletons (OTUs assigned only a single read), the OTUs were classified to genus level using the SILVA database, which assigns taxonomic identity at a confidence threshold of 80%. The identity of the dominant OTU sequences was confirmed by comparing them with standard sequences contained in the NCBI-BLAST database. The sequence dataset used in this study has been submitted to the NCBI Sequence Read Archive under BioProject PRJNA955810.

Bioinformatics and statistical analysis

The phylogenetic relatedness of the dominant prokaryotic and eukaryotic populations in HRAOP was carried out using the Neighbour-Joining method in Mega 6.0 (Tamura et al. 2013). The microbial phylotype richness and evenness were estimated using an abundance-based coverage estimator (ACE), Chao, Shannon, and Inverse Simpson (Invsimpson) index incorporated in Mothur software. Rarefaction calculations were carried out within Mothur, and curves were generated for a distance value of 0.03 to visualise that all samples reached the plateau. The community structure and diversity are presented in bar plots from data computed using the statistical function in SigmaPlot 11 (Systat Software Inc., USA). Heat maps showing the taxonomic profile of the most abundant microbes on a seasonal basis were constructed using an online platform (Babicki et al. 2016). Jvenn, an online interactive Venn diagram viewer, was employed to identify the unique and similar OTUs among seasonal samples (Bardou et al. 2014).

To determine the key indicator (i.e., feature or variable) in the HRAOP that underpins MaB-floc formation, pond productivity, floc stability, settling and biomass recovery, and hence final effluent quality, a principal components analysis (PCA) was performed using R (Version 4.3.0). The most abundant prokaryotes and eukaryotes were selected as the biotic variables while the abiotic variables used were the environmental/operational parameters, including direct normal irradiance (DNI), water temperature, air temperature, pH, MLSS as an indicator of biomass concentration, and MaB-floc settling. Before analysis, data were standardised to ensure the normality and linearity of statistical assumptions. Results of the analysis were visualised using the “devtools”, “factoextra” and “ggbiplot” packages in R.

Results

Environmental and operational variables affecting HRAOP performance

Parameters considered important in mediating the dynamics of MaB-flocs were monitored to determine their effect on pond performance and productivity. Results presented in Fig. 1a show the variation in daily normal irradiance (DNI), as expected for this Eastern Cape Province isophote, varied between 146-442 W m-2 with a mean DNI of 287 W m-2 (Power and Mills 2005). Air temperature varied from a low of 11.1°C in winter to a maximum of 28.9°C in summer (December), while the mean annual daytime air temperature during the experiment was 20.4°C. Pond water temperature followed a similar trend and varied throughout the sampling period from a low of 12.5°C in winter to a high of 29.5°C in summer (Fig. 1b). Pond pH varied over the sampling period from a low of 8.6 in the cooler winter months to a high of 11.3 in the summer (Fig. 1b), presumably due to H+ absorption from the wastewater by photosynthetically active microalgal cells (Zerveas et al. 2021).

Fig. 1
figure 1

Time course of changes in environmental and HRAOP operational variables monitored from February to December 2019. Air temperature and daily normal irradiance (a), water temperature and pH (b), and MLSS and settled biomass (c). Red diamonds along the X-axis indicate sampling of mixed liquor for metagenomic analysis. Error bars represent the S.E. of replicated samples

Biomass concentration in HRAOP (measured as MLSS) also varied and ranged from 70 mg L-1 to 209 mg L-1 during the experimental period (Fig. 1c). A steady decline in biomass concentration from a near maximum of 190 mg L-1 in February (late summer) to a near minimum of 80 mg L-1 in May (late autumn) was followed by a gradual increase, commencing in late winter, and rising to >200 mg L-1 in December (Fig. 1c).

Biomass settling was not apparently linked to MLSS concentration but rather to the forming of well-aggregated and stable flocs. While the settling efficiency of biomass after 1 h in Imhoff cones ranged between 36 and 95% (Fig. 1c), from late summer (February) through autumn (May), the formation of considerably larger flocs resulted in an average settleability of 78% (Fig. 1c). Likewise, floc settleability in spring (November - December) was ~95%. On the other hand, settling was substantially lower in the winter sampling period (June - July), with a mean settled biomass of 41%.

Structure of MaB-flocs and effect on HRAOP productivity

Light microscope examination showed that samples of mixed liquor collected monthly from the HRAOPs contained an abundance of MaB-flocs of various compositions and structures. Flocs were discernible as suspended aggregates which settled easily and rapidly and comprised an agglomeration of extracellular materials and microorganisms, including bacteria, microalgae, and occasionally rotifers and ciliates (Fig. 2).

Fig. 2
figure 2

Light micrographs showing MaB-flocs from HRAOP mixed liquor detected during sampling. Typical compact flocs in summer between February and May were dominated by Pediastrum and Desmodesmus/Scenedesmus (a-b), flocs dominated by filamentous bacteria in winter from June to July (c-d), loose and disaggregated flocs observed early in October (e), spring flocs dominated by filamentous cyanobacteria (f), Micractinum (g) and diatoms (h) from November to December. Scale bar = 54 µm

During the study period (February - December 2019), the abundance of the dominant microalgal species varied coincident with changes in the environment, MaB-floc size, composition, structure and settling. Dominant microalgal species within the aggregated biomass between February and May were the chlorophytes Pediastrum sp., Desmodesmus sp. Micractinium sp., and Scenedesmus sp. (Fig. 2a and b). During the winter period (June - July), MaB-flocs were characterised by the presence of chlorophytes such as Desmodesmus/Scenedesmus sp., Micractinium sp., Chlorella sp. and Actinastrum sp., but dominated by filamentous bacteria (Fig. 2c and d). Later and before the onset of spring, MaB-flocs appeared either loose, sparse or completely disaggregated, resulting in biomass mostly in a planktonic state (Fig. 2e). Between October and December, the mixed liquor typically contained MaB-flocs dominated by filamentous cyanobacteria, Micractinium sp. and several diatoms (Fig. 2f-h). Other occasionally present species during the experimental period included Dictyosphaerium sp., Coelastrum sp., Pyrobotrys sp. and Closterium sp.

Overall, the period of good flocculation and settling was associated with dominance by colonial microalgae, filamentous cyanobacteria and various diatoms contained as part of the aggregated structure. Thus, when floc formation was greatest and aggregates were stable productivity of the HRAOP was highest (Table 1). Meanwhile, disaggregation of flocs in the mixed liquor coincided with dominance by a filamentous bacterium, suggesting that this species may somehow be responsible for the poor biomass settling. Thus, the observed changes in species composition and diversity within the mixed liquor and suspended microbial aggregates may play a significant role in the formation and stability of MaB-flocs.

Table 1 Biomass productivity in periods of good and poor MaB-floc formation in HRAOP of an IAPS treating domestic sewage. Values are mean ± S.E.

Diversity and community structure of HRAOP

After removing ambiguous or chimeric sequences, 131,599 and 243,883 sequence reads were generated for the prokaryote (including cyanobacteria) and eukaryote populations in HRAOP mixed liquor, respectively (Table 2). The number of prokaryotic reads observed in spring was relatively low, indicating that the PCR library used for the sequencing reaction may have been insufficient. Thus, sequences of rare species as part of the population may have been overlooked. Nevertheless, all samples showed good coverage, especially the eukaryotes, indicating that the depth of sequences used for analysis adequately represented the microbial community structure in HRAOP mixed liquor during sampling. Diversity indices comprising ACE, Chao, Shannon, and Invsimpson were determined to assess microbial richness and diversity. Results revealed, and perhaps as expected, that microbial community richness (measured as ACE, Chao and the number of OTU generated) was generally higher in the summer months. At the same time, the lowest values were recorded during winter (Table 2). Even though the number of bacterial sequences was relatively low in spring, the microbial population appeared well-distributed and more diverse than in winter. Thus, the diversity metrics obtained from the Shannon and Invsimpson indices were higher in the summer months, while the lowest values were obtained in winter.

Table 2 Summary of the sequence reads and alpha diversity indices of the microbial population in HRAOP of an IAPS treating domestic sewage

Phylogenetic classification of sequence reads using the SILVA database revealed that the bacterial population in HRAOP mixed liquor comprised 29 phyla, and the dominant phyla are shown in Fig. 3. Proteobacteria was the most abundant phylum, accounting for 27-79% of the classified sequences (Fig. 3A). Within the Proteobacteria, Gammaproteobacteria was the most abundant (51-98%), especially in winter (S3 and S4) and accounted for almost all derived sequences (>98%). The abundance of other members of the Proteobacteria (i.e., Alphaproteobacteria and Deltaproteobacteria) was relatively low during the winter sampling intervals S3 and S4 (≤2%) but significantly higher in summer and spring. Cyanobacteria was the most abundant phylum in the spring, S5 (49%), and was also present in significant numbers in winter (S3 and S4) and late spring and accounted for 18-21% of the sequences. Also abundant were members of the phyla Bacteroidetes and Planctomycetes, which accounted for 12-34% and 3-18% of sequences collected in spring and summer (Fig. 3a). Other representatives but in relatively low abundance (≤4%) were from the Gemmatimonadetes, Verrucomicrobia, Firmicutes and Actinobacteria, and some unclassified bacteria for which the taxonomy remains unavailable.

Fig. 3
figure 3

Microbial diversity and community structure in HRAOP. The dominant phyla of bacteria (a) and eukaryotes (b) are expressed as the relative percentage of the total sequences in each sample

Within the eukaryotic domain, a total of 24 phyla were detected in the microbial population. Phylum Chlorophyta was the most abundant (71-99%) and comprised largely of the family Chlorophyceae, accounting for 62-86% of the total sequences (Fig. 3b). Sample S2 was dominated by the phylum Rotifera (44%), with fewer representatives of the family Chlorophyceae (35%). Other dominant chlorophytes included those of the Tribouxiophyceae family (>23%) and some unclassified chlorophytes (>15%), most of which were sequences derived from samples collected in the winter (S3 and S4). Furthermore, a significant number of fungi (16%) belonging to the phylum Blastocladiomycota were observed in S6.

Additionally, some unclassified eukaryotes (>20%) were present in samples S1 and S2, which may indicate entirely novel phylotypes. It is equally possible that these sequences were absent from the version of the SILVA database used for analysis in the present study. This notwithstanding, it was evident that summer samples showed relatively higher diversity and variation in the distribution of prokaryotic and eukaryotic phylotypes.

Sequences were further grouped into OTUs at the 97% confidence level, and a total of 166 OTUs were successfully clustered for bacterial sequences, while 174 OTUs clustered for eukaryotes. Identification of the major OTUs and determination of the dominant species in each sample was achieved using SILVA. The results were confirmed by reference to the NCBI database (Table S1), and the microbial composition of HRAOP mixed liquor at each sampling interval is presented as % relative abundance and shown in Figure 4.

Fig. 4
figure 4

Relative abundance of the dominant bacterial (a) and eukaryotic (b) species present in MLSS of HRAOP. Sequences were clustered into OTUs, and phylogenetic classification was carried out using SILVA as the reference database. S1 and S2 = summer (February-March 2019); S3 and S4 = winter (June-July 2019); and S5 and S6 = spring (November-December 2019)

Thiothrix spp. was the most abundant bacterial genus (17%) in the mixed liquor from HRAOPs (Fig. 4A), clustered into three OTUs (B_OTU001, B_OTU083, and B_OTU086), was dominant in winter and accounted for almost all the sequence reads present (>80%). A more diverse bacterial population was observed during summer when species of Porphyrobacter (B_OTU005 and B_OTU006), Silanimonas (B_OTU002), Mariniradius (B_OTU003), SM1A02 (B_OTU004 and B_OTU009), Algoriphagus (B_OTU007 and B_OTU136), Flavobacterium (B_OTU042 and B_OTU046), Cyanobacterium (B_OTU014, B_OTU017, B_OTU034, B_OTU103, and B_OTU120), Dokdonella (B_OTU011), among others, accounted for a significant proportion of the population. Also present, but in lesser proportions, were species such as Zoogloea, Thauera, Malikia, Candidatus Halomonas phosphatis, Truepera, Roseomonas, Ideonella, and others (Fig. 4a). Only three OTUs (i.e., Mariniradius, Hydrogenophaga, and an uncultured Rhizobiales) were shared by all samples suggesting that, of the bacterial species present in HRAOP mixed liquor, 9% occurred irrespective of the season (Fig. S1).

The OTUs generated for eukaryotes were dominated by chlorophytes, including Desmodesmus, Pseudopediastrum, Scenedesmus, Chlorella, Micractinium, and Hariotina (Fig. 4b). Pseudopediastrum (OTU003 and OTU031) was the most abundant in sample, S1 and accounted for 58% of the reads but declined to 27% by the second summer sampling interval S2, possibly due to the emergence of the rotifer, Brachionus (OTU004, OTU033, and OTU052), which accounted for 49% of the population. On the other hand, Desmodesmus (OTU001, OTU002, OTU009, OTU014, OTU017, OTU020, OTU028, and OTU039) was present in all samples but more abundant in the springtime, when it accounted for 85% of the sequences. Other abundant chlorophytes included species of Chlorella, Micractinium, and Scenedesmus. Nitzschia (OTU024) was the only diatom species detected and was present during sampling interval S6, which coincided with the transition from spring to summer. In addition, evidence for the presence of various fungi (Sanchytrium and Paraphysoderma), ciliates (Amphileptus and Telotrochidium), rotifers (Cephalodella) and arthropods (Moina and Stenocypris) was also obtained (Fig. 4b). Compared to the number of bacterial sequences, the number of eukaryotic OTUs shared across all samples was higher (a total of 12). These OTUs accounted for 69% of the eukaryotic population and were predominantly of microalgal origin and included Pseudopediastrum, Desmodesmus, Scenedesmus and Chlorella (Fig. S1).

Heatmaps show the relationship between parameters using a change in cell colour intensity across each axis to display patterns reflecting the change in value for one or both parameters. Thus, the dominant OTUs from each season were pooled, and the relationship between dominant microbial genera and season was determined as illustrated in Figure 5. Thirty-three bacterial genera were classed within the Proteobacteria, Bacteroidetes, Planctomycetes, or Cyanobacteria and these constituted the dominant prokaryotic groups and represented 86% of the total bacterial sequence reads. Changes in colour intensity revealed that variations in bacterial genera appeared to be decidedly seasonal. Except for a few, most of the OTUs (28 of 44) were more abundant in summer than in either winter or spring, emphasising the higher diversity of bacterial genera during this season (Fig. 5a). OTUs identified as Thiothrix, Cyanobacterium, Polynucleobacter and Algoriphagus were peculiar to winter, whereas bacteria such as Flavobacterium, Sediminibacterium, and Cyanobacterium were dominant in spring (Fig. 5a).

Fig. 5
figure 5

Heatmap analysis showing the dominant OTU of the bacterial (a) and eukaryotic (b) communities in mixed liquor from HRAOP. The taxonomic classification of each OTU is listed in Table 3

Heatmap analysis of the eukaryotic population showed 21 genera dominant in the HRAOPs during the sampling period and that microbial diversity was greater in the spring and summer than in the winter (Fig. 5b). Although chlorophytes were dominant in all seasons, as might be expected for HRAOP mixed liquor, surprisingly, this was particularly so in winter and spring. Non-microalgal microbes such as rotifers and fungi were also present at other times. So, seasonality appears to exert a major impact on population dynamics in HRAOP mixed liquor, which it is posited, must play a fundamental role in microbial aggregation, accumulation, and stability of MaB-flocs, settling, biomass recovery and, ultimately, water treatment efficiency.

Influence of biotic/abiotic parameters on seasonal emergence of MaB-flocs

To further explore microbial community dynamics in HRAOP, prevailing biotic factors were assessed regarding species co-occurrence, and PCA was used to scrutinise both biotic and abiotic factors further and to delineate the best possible characteristics or combination thereof that underpin the emergence of MaB-flocs.

Results in Figure 6 illustrate the co-occurrence between the five most abundant microalgae species and species of the five most abundant fungal pathogens and micro-zooplanktonic grazers. The rotifer, Cephalodella sp., was dominant in the summer sampling periods and, whereas the population of this grazer increased between S1 and S2, that of the ciliate Telotrochidium sp. declined (Fig. 6a). A similar trend was observed for the chytrid, Sanchytrium sp. (Fig. 6b), and the microalgae Pseudopediastrum sp. and Desmodesmus sp. (Fig. 6c). Thereafter, the populations of rotifers, ciliates and crustaceans, and fungi remained relatively low until spring sampling (Fig. 6a and b). In early spring, S5, populations of the crustacean Stenocypris spp. and an uncultured choanoflagellate (protozoan) peaked briefly, and in S6, the crustacean Moina sp. was at a maximum (Fig. 6A). Chytrids also increased during in the spring (S5 and S6), with the population of Paraphysoderma sp. rising to a maximum by S6 (Fig. 6b). These changes were not associated with the relative abundance of Desmodesmus sp., which remained elevated throughout the winter (S3 and S4) and spring (S5 and S6).

Fig. 6
figure 6

Co-occurrence between micro-zooplankton (a), fungal pathogens (b) and microalgae (c) in HRAOP mixed liquor suspended solids

Principal component analysis revealed the clustering patterns for the selected variables, genus (biotic) and HRAOP parameter (abiotic) as a function of the season for the prokaryote and eukaryote populations (Fig. 7).

Fig. 7
figure 7

Principal component analysis biplot showing the dominant prokaryote (a) and eukaryote species (b) response to abiotic (grey ovals) and biotic (blue ovals) parameters in HRAOP of an IAPS used for municipal sewage remediation. Data points within each cluster include SM1A02 sp. (1), Porphyrobacter sp. (2), Mariniradius sp. (3), Thiothrix sp. (4), pH (5), settleability (6), MLSS (7), irradiance (8), Sanchytrium sp. (9), Desmodesmus_009 sp. (10), Pseudopediastrum sp. (11), Brachionus sp. (12), Chlorella sp. (13), Micractinium_010 sp. (14), Paraphysoderma sp. (15), Desmodesmus_002 sp. (16), Micractinium_006 sp. (17), and Desmodesmus_001 sp. (18)

Results for the most abundant prokaryotic species (i.e., Thiothrix sp., Gemmatimonas sp., Ideonella sp., Algoriphagus sp., Mariniradius sp., Silanimonas sp., Porphyrobacter sp. and SM1A02) within the HRAOP microbiome show that dominant microbes in the same season are closely associated with one another, suggesting that samples from a similar abiotic dynamic likely comprise a similar biotic component. In Fig. 7a, 93.4% of the total variation was explained by PCA1 and PCA2. Variation along the first axis (PCA1) showed a strong correlation to all dominant species in summer except for Thiothrix, which was dominant in winter. Variation along the second axis (PCA2) correlated with all abiotic variables. Furthermore, the occurrence of Thiothrix was negatively correlated with temperature, pH, MLSS and settleability, which may indicate that this biotic component was a major contributor to floc disaggregation in winter. Thus, abiotic variables appear to exert greater effect in spring, less so in summer and not at all in winter.

For the most abundant species within the eukaryotic population of the HRAOP microbiome (i.e., Desmodesmus sp., Pseudopediastrum sp., Chlorella sp., Micractinium sp., Brachionus sp., Paraphysoderma sp. and Sanchytrium sp.) and as shown in Fig. 7b, the first two principal components accounted for 81.8% of the total variation. The first axis (PCA1) strongly correlated with all abiotic variables. Most chlorophytes, the ciliate Brachionus, a trophic upgrader and the parasitic fungi Sanchytrium and Paraphysoderma tended more towards the second component, PCA2 (Fig. 7b). Nevertheless, other dominant chlorophytes, including Chlorella sp. and Micractinium_10 sp., appeared sensitive to both biotic and abiotic variables, including settleability. Colonial microalgal species responded better to these variables. Interestingly, periods in which formation and stability of MaB-flocs were best, viz. spring and summer, were characterised by the presence of chytrids in association with the host microalgae. Thus, the abundance of Paraphysoderma sp. increased in the spring, coincident with a proliferation of Desmodesmus sp. Similarly, Sanchytrium sp. dominated in the summer along with the chlorophytes Pseudopediastrum sp., Desmodesmus sp. and the ciliate Brachionus sp., suggesting that these trophic upgraders may significantly contribute to MaB-floc formation in HRAOP.

Discussion

For many algal-based WWT technology solutions such as IAPS (Laubscher and Cowan 2020), AIWPS (Oswald 1995), and enhanced pond systems (EPS;Craggs et al. 2003, 2015; Sutherland and Ralph 2020), efficient treatment relies on the ability of the bioprocess to form good, settleable sludge (Garcia et al. 2000, 2006; De Schryver et al. 2008; Jimoh et al. 2019) precipitated by the synergy between microalgae and the HRAOP microbiome (Ashraf et al. 2023). Indeed, the natural phenomenon of bio-flocculation, which in algal-based systems, results in the formation of MaB-flocs with good settling characteristics, has been established for HRAOP of IAPS (Jimoh and Cowan 2017; Wang et al. 2022; Biliani and Manariotis 2023) and is reconfirmed in this study. Formation and passive settling of flocs ensures that the discharged water from IAPS meets the standards set by regulatory authorities (Jimoh et al. 2019; Dube and Cowan 2023). Even so, there are periods when floc formation in HRAOP is either compromised or once formed, flocs disaggregate, resulting in low or no settled biomass, poor recovery, and markedly reduced water quality. Disaggregation of MaB-flocs is likely an active process and may be due to biotic impact. Thus, a major purpose was to identify culprit species responsible for disaggregation of MaB-flocs by monitoring biotic/abiotic-induced changes in microbial community structure.

This study used NGS to identify the key biocatalysts, describe the microbial community, and elucidate seasonal changes in structure and diversity in mixed liquor from the HRAOPs of an IAPS used to treat municipal sewage. Analysis revealed that the living component of the HRAOP mixed liquor was composed of bacteria (including cyanobacteria) and a suite of eukaryotes, including microalgae, fungi, ciliates, rotifers, and arthropods either directly or indirectly associated with the suspended aggregates or MaB-flocs. While species composition and structure of the microbial community are purported not to affect nutrient removal efficiency and biomass production by HRAOP (Sutherland and Ralph 2020), results from the current study show that changes in species diversity and abundance can be associated with changes in MaB-floc formation and settling. And the generation of easily settleable MaB-flocs and removal of biomass from algal settler ponds produces a final treated effluent with COD and soluble solids content permissible for irrigation or discharge (Dube and Cowan 2023).

Seemingly ignored and under-studied is the prokaryotic component in mixed liquor from HRAOPs and, particularly, the identity, community structure and dynamics and role of the bacteria in MaB-floc formation and stability. Metagenomics revealed that Silanimonas sp., first isolated from laboratory cultures of the bloom-forming cyanobacterium, Microcystis (Chun et al. 2017); Mariniradius sp., a member of the family Cyclobacteriaceae isolated from marine aquaculture pond water (Bhumika et al. 2013; Shahinpei et al. 2020); bacteria belonging to the SM1A02 group which are present in many activated sludges and like Nitrospira and Nitrosomonas, are a genus related to ammonia transformation (Chu et al. 2015; Cui et al. 2016; Tian et al. 2017; Vico et al. 2021) and recently, the first microalgae-associated SM1A02 planctomycete to be recognised (Rambo et al. 2020); and, Porphyrobacter sp., an aerobic bacteriochlorophyll-synthesizing budding bacterium isolated from freshwater (Fuerst et al. 1993) were the dominant bacteria in HRAOP during the summer sampling intervals.

Other abundant genera present in the summer included Algoriphagus, a genus belonging to the phylum Bacteroidetes with type species Algoriphagus ratkowskyi IC025T (Bowman et al. 2003); Ideonella, a genus containing the species Ideonella sakaiensis 201-F6 which was recently reported capable of using poly(ethylene terephthalate) as a carbon and energy source for growth (Austin et al. 2018; Son et al. 2019) and the less abundant soil-associated bacteria Gemmatimonas, a member of the phylum Gemmatimonadetes which is said to be one of the nine most abundant phyla in soils (Janssen 2006; Youssef and Elshahed 2009) with an incidence of 2.2% of the total soil bacterial population (DeBruyn et al. 2011) and potentially chlorophototrophic (Zeng et al. 2014, 2021); and, Dokdonella sp., a species of bacterium of the family Xanthomonadaceae originally isolated from agricultural soil (Yoon et al. 2006).

In the winter, the bacterial population was dominated by Thiothrix, a genus of filamentous sulphur-oxidizing bacteria related to the genera Beggiatoa and Thioploca (Williams and Unz 1985). Also present was an uncultured cyanobacterium and an uncultured Rhodocyclaceae. Thiothrix is typically Gram-negative, rod-shaped, and appears as long sulphur globule-carrying filaments known to co-occur with archaea where they protect and facilitate attachment of organic and inorganic materials, respectively, particularly in the spring (Moissl-Eichinger and Huber 2011). This latter character is indicative of EPS production and biofilm formation. Indeed, a function of aggregation of Thiothrix into a biofilm together with other facultative anaerobic, chemolithoautotrophic, sulphur-oxidizing bacteria, on the surface of water bodies is purported to be mandatory for survival in oxygen-rich environs. Similarly, Rhodocyclaceae, like the Xanthomonadaceae and Sphingomonadaceae, contain bacteria capable of producing EPS in anaerobic granular sludge (AGS) systems (Szabó et al. 2017). Cyanobacteria are well known to produce EPS and other high molecular mass hetero-biopolymers used to impart resilience to biofilm and aggregate formation while serving a fundamental role in carbon and nitrogen assimilation and sequestration (Pereira et al. 2009; Jimoh et al. 2019; Mota et al. 2021; Laroche 2022). In the spring, bacterial diversity increased and was initially dominated by an uncultured cyanobacterium and the bacteria Dokdonella, Mariniradius, an uncultured Rhizobiales, species of the planctomycete Pirellula, Flavobacterium, Thauera and others typically associated with water and soil environments. In late spring, the population became dominated by an uncultured cyanobacterium and species of the bacteria Mariniradius, Dokdonella and Flavobacterium.

Due to its larger size and dominance, Pseudopediastrum sp. (Jena et al. 2014; Lenarczyk and Saługa 2018) apparently enhances gravity-based harvesting efficiency by forming larger MaB-flocs than other colonial microalgae (Park et al. 2013). In addition, selection for this as the dominant species in an IAPS-like process also increases biomass yield and resilience to grazing (Sutherland and Ralph 2020). Desmodesmus sp. were secondary in summer but rose to dominance in winter (June - July) and persisted into the spring. While grazers may also contribute to the disaggregation of flocs and loss of settleability due to the well-known and devastating effects these organisms can have on microalgae populations (Day et al. 2017), the two most likely candidates based on the present results are, the filamentous bacterium Thiothrix that dominated throughout periods of poor floc settleability and the presence of chytrids. When viewed from a co-occurrence perspective (Vass et al. 2022), the latter sheds new light on our understanding of the HRAOP fungal metacommunity assembly and reveals the potential of host-parasite interactions in MaB-floc formation, wastewater treatment and settleability of harvestable biomass.

Dominance by the filamentous bacterium Thiothrix coincided with poor flocculation and very low settling (<40%). Generally, an overgrowth of filamentous microorganisms (including cyanobacteria) causes bulking during wastewater treatment by lowering floc density significantly, causing poor settling and loss of important microbial species necessary for COD and nutrient removal (Henriet et al. 2017). Thiothrix has been especially associated with biomass washout and loss of performance in wastewater treatment efficiency (Henriet et al. 2017; Stauch‑White et al. 2017). The emergence of this species during IAPS wastewater treatment may explain the onset of poor flocculation and biomass settleability when this bacterium was present. The dominance of Thiothrix is associated with sludge bulking, a common problem in which sludge fails to separate in the sedimentation tanks and negatively impacts settling by interfering with oxygen transfer (Wu et al. 2019). In an aerobic system like HRAOPs, where the mutualistic transfer of CO2 and O2 takes place between microalgae and bacteria, dominance by Thiothrix, which hinders O2 transfer, might have tampered with system operation and performance, hence poor MaB-floc formation. Indeed, the disappearance of Thiothrix from the microbial population coincided with the restoration of natural floc formation in the system, irrespective of the dominating microalgae.

Chytrids, or the Chytridiomycota, are predominantly aquatic fungi, and some are known to parasitise algae, particularly microalgae (Kagami et al. 2007; Hoffman et al. 2008; Karpov et al. 2018; Van den Wyngaert et al. 2018, 2023; Kobayashi et al. 2023) and are considered by some authors to be ‘trophic upgraders’ in which organic and inorganic nutrients are transferred from lower to higher trophic organisms (Rasconi et al. 2020). For example, experiments show that protists, including heterotrophic nanoflagellates (Bec et al. 2003, 2006), dinoflagellates (Veloza et al. 2006), and ciliates (Martin-Creuzburg et al. 2005) can enhance (i.e., ‘upgrade’) the diets of the higher trophic level organisms by supplying essential nutrients such as lipids (Rasconi et al. 2020; Laundon et al. 2022). Thus, trophic upgraders may also change host metabolism, leading to greater quantities of mucilage or EPS for protection from parasitism. The latter likely also contributes material for floc formation.

A PCA confirmed the linear combinations in response to abiotic and biotic variables as a function of season and their potential association with MaB-floc formation and stability. In particular, the associations in summer and spring between the colonial microalgae Pseudopediastrum sp., Desmodesmus sp. and Micractinium sp. and their respective chytrids, Sanchytrium sp. and Paraphysoderma sp. Less significant were the abiotic parameters, which clustered together. While several studies indicate that temperature is a dominant environmental variable affecting microbial community structure (Wang et al. 2014; Meerbergen et al. 2016; Xu et al. 2018), lower irradiance appears to promote the formation of MaB-flocs that settle rapidly and easily (Arcila and Buitrón 2017). This notwithstanding, aggregation/disaggregation of MaB-flocs in high-rate oxidation ponds is clearly a response to biotic/abiotic-induced changes in microbial community structure. And MaB-floc formation is essential for settling to occur and is needed to ensure that the final treated effluent is within regulation for either re-use or discharge (Dube and Cowan 2023).

Conclusion

Use of NGS in this study has improved our understanding of the community dynamics in HRAOP of an IAPS treating domestic sewage. In addition, this is the first report of a detailed analysis of the bacterial, fungal, microalgal, and arthropod community structure and diversity associated with MaB-floc formation in HRAOP. Thiothrix, notorious for causing sludge bulking and settleability problems during wastewater treatment was the dominant bacterial species in samples collected in the winter and appeared to be highly sensitive to temperature, and occurred coincident with reduced biomass settleability. It is proposed that the appearance of this culprit bacterium resulted in the disaggregation and disappearance of MaB-flocs in the HRAOP. And that parasitic chytrids, known for upgrading nutrient availability, may also have contributed to seasonal changes in the microalgal population dynamic to enhance floc formation. These findings indicate that seasonal change, colonial microalgae and associated parasitic fungi, and diverse bacterial species participate in the forming and disaggregation of MaB-flocs, which impacts biomass recovery from the HRAOP and final effluent quality.