Introduction

Enhanced biological phosphorus removal (EBPR) process is carried out by polyphosphate accumulating organisms (PAOs) under alternating anaerobic-aerobic conditions. Under anaerobic conditions, PAOs take up carbon sources by using energy generated from hydrolysis of intracellular polyphosphate (poly-P) and glycogen and accumulate carbon sources as poly-β-hydroxyalkanoates (PHA). Aerobically, PAOs are capable of accumulating excessive amount of phosphate into cells by oxidizing PHA to gain energy (Oehmen et al. 2005c; Zhou et al. 2010; Zhou et al. 2012). PAOs classified under Betaproteobacteria species are tentatively named “Candidatus Accumulibacter phosphatis” and generally referred as Accumulibacter. PAOs are also found in the species of Actinobacteria, Alphaproteobacteria, and Bacteriodetes/Chlorobi group. Depending on the type of carbon sources in the feed, PAO population varies largely (Chua et al. 2006; Fukushima et al. 2007; Gu et al. 2008; He et al. 2008). In general, Accumulibacter is more common in volatile fatty acids (VFAs) fed systems and Actinobacteria seems to be dominant in amino acids fed systems (Zengin et al. 2011).

Glycogen-accumulating organisms (GAOs) are recognized as the major competitor of PAOs that causes EBPR failure. GAOs are able to perform carbon conversions under alternating anaerobic and aerobic conditions in a similar way without contributing to phosphorus accumulation. GAOs classified under Gammaproteobacteria are named “Candidatus Competibacter phosphatis” and generally referred as Competibacter. Competibacter is typically found in glucose or acetate-fed biosystems. Additionally, two GAOs lineages that are classified under Alphaproteobacteria have been described. One is related to Sphingomonas and the other to Defluvicoccus vanus (Burow et al. 2007; Oehmen et al. 2005a; Oehmen et al. 2005b). Based on the previous research, some carbon sources favor GAOs than PAOs. For example, a wastewater stream that contains high concentration of propionate may enrich Defluvicoccus vanus, while glucose was reported to increase Competibacter (Oehmen et al. 2005b; Wang et al. 2010; Zengin et al. 2010).

Considerable studies have been devoted to the understanding of PAO and GAO competition mechanisms and their metabolisms with various carbon sources. Both PAOs and GAOs generally prefer volatile fatty acids (VFAs) as carbon sources. Apart from VFAs, other organic substrates, including various carboxylic acids, sugars, alcohol, and amino acids, can also be utilized anaerobically by EBPR sludge. These carbon sources could affect EBPR performance through altering PAO-GAO population abundance. Selection of PAOs over GAOs is crucial to ensure a more efficient and reliable EBPR process (Lopez-Vazquez et al. 2009; Wu et al. 2010). Meantime, the choice of the carbon source also depends on the operating cost of the process as well as the phosphorus removal efficiency.

To date, there is limited literature that summarizes and compares the pros and cons of studied carbon sources in a systematic manner. Further, complex carbon sources generated from waste material are gaining their popularity to support EBPR process. The performance and feasibility using complex carbon need to be compared with conventional carbon sources-VFAs. Given the fact that commercial carbon source to enhance and stabilize biological nutrient removal (BNR) is costly, a more sustainable carbon source should be explored. This review aims to summarize the carbon sources that are commonly applied in EBPR systems and highlight the roles of carbon sources in microbial community selection. Meanwhile, other environmental friendly and economic carbon sources are proposed and compared with traditional carbon source (e.g., VFAs) from different angles.

Single carbon source

Acetate

Acetate is one of the most popular carbon sources studied in EBPR research. Microbial community of acetate-fed EBPR systems has been investigated in many studies (Gonzalez-Gil and Holliger 2011; He et al. 2010; Lv et al. 2014; Tu and Schuler 2013). Accumulibacter-like bacteria are normally dominant in such system. It is partially due to the much higher acetate uptake rate by phosphate accumulating metabolism (PAM) dominated culture compared to the glycogen accumulating metabolism (GAM) dominated culture (Schuler and Jenkins 2003). Burow et al. (2008) also demonstrated that Defluviicoccus enrichment had a lower acetate uptake rate compared with the Accumulibacter enrichment. However, Gonzalez-Gil and Holliger (2011) concluded that the presence of Accumulibacter cannot ensure a consistently good EBPR performance, while the type of Accumulibacter is one of key factors to determine the robustness of the phosphate removal process. They reported that there was a transient shift between Accumulibacter type I and type II before Accumulibacter type IIA being dominant in EBPR mature stage. The result was different in He et al. (2010), where it concluded that there was no consistent correlation between the reactor performance and the clade distribution. Two acetate-fed reactors were operated in that study. Accumulibacter clade IA remained relatively stable in R1 during the entire 6-months operation, while Accumulibacter clade composition was more dynamic in R2. Interestingly, in another acetate-fed system, PAOs community covered Alphaproteobacteria (20.4 % of PAOs populations), Betaproteobacteria (i.e., Accumulibacter, 22.2 %), Bacteroidetes/Chlorobi group (17.6 %), and Actinobacteria (4.6 %) (Zhang et al. 2005).

Generally, EBPR performance is relatively stable using acetate as carbon source. However, PAM may shift to GAM under certain conditions, e.g., high acetate loadings, which results in a higher fraction of GAOs eventually (Ahn et al. 2007; Ahn et al. 2009; Zhou et al. 2008). pH and temperature are two other important factors affecting PAO and GAO competition (Lopez-Vazquez et al. 2009). A high operational pH was shown to provide PAOs an advantage over both the Competibacter GAOs and Alphaproteobacteria GAOs. In a long-term experimental study comparing the acetate uptake rates of PAOs and GAOs, it was found that GAOs took up acetate faster at a pH below 7.25, while PAOs took up acetate faster at a pH above 7.5 (Filipe et al. 2001). Temperature also appears to be a factor that has an impact on the PAO-GAO competition. It is known that deterioration of EBPR occurs when temperature is higher (25 ∼ 30 °C) (Lopez-Vazquez et al. 2009). By manipulating acetate loading and concentration, it is possible to mitigate the negative impact from these two factors (pH and temperature) on PAOs. Tu and Schuler (2013) reported that EBPR system could be dominated by GAOs at relatively low pH (6.4–7.0) with acetate concentration of 200 mg/L as pulse feed. By controlling the acetate feeding rate and maintaining low concentration of acetate in the reactor, they were able to recover PAOs community and EBPR performance. Ong et al. (2013) demonstrated an effective EBPR system at 28–32 °C with acetate as the sole carbon source. In that study, a lower COD/P ratio (C/P = 3) led to relatively higher P-removal rates as compared to C/P ratio of 10. Therefore, low acetate concentration in bulk and/or low COD loading may favor PAOs than GAOs. These observations could be explained by the results from Burow et al. (2008), which proposed active acetate transport contributed more to overall acetate uptake in Accumulibacter than in Defluviicoccus. Acetate permease activity was thought to play an important role in scavenging low amounts of acetate and increased activity of permease-mediated acetate uptake in Accumulibacter may provide a competitive advantage over Defluviicoccus under the low acetate concentration conditions.

Propionate

Propionate is another popular carbon substrate for EBPR studies. Acetate accounts for 49–71 % of total VFA in septic municipal wastewater, while propionate is the second most dominant VFA (24–33 %) (Lv et al. 2014). Propionate can be taken up by PAOs and converted to propionyl-CoA. Poly-P is hydrolyzed to orthophosphate and released to bulk liquid. Glycogen is hydrolyzed to acetyl-CoA and CO2. Acetyl-CoA and propionyl-CoA are reduced and condensed to form PHA, with the reducing power provided by glycogen hydrolysis (Oehmen et al. 2005c). Propionate is actually considered as a more favorable substrate than acetate. In fact, PAOs require less energy and less P-release for propionate uptake as compared to acetate (Carvalheira et al. 2014; Pijuan et al. 2004).

Propionate can be immediately taken up by Accumulibacter PAOs while its consumption by Competibacter GAOs is insignificant. It is known that only one particular type of GAOs is able to compete for propionate (i.e., Alphaproteobacteria GAOs) (Oehmen et al. 2005b). This competition can be manipulated through pH adjustment. It is reported that Accumulibacter PAOs can be enriched in a propionate-fed system in pH 7–8, where PAOs appeared to outcompete a group of Alphaproteobacteria GAOs (Oehmen et al. 2005a). Indeed, a better P-removal performance is often observed in propionate-fed EBPR systems. The abundance of Accumulibacter PAOs is typically higher in propionate-fed EBPR system than that of acetate-fed system, correspondingly GAOs generally present in a lower number (Oehmen et al. 2006). Lv et al. (2014) showed that Accumulibacter-like species increased from 4.54 to 9.53 % in an acetate-fed reactor during acclimation period, and that changed from 4.38 to 41.5 % in a propionate-fed reactor. GAO-like species varied from 1.54 to 6.92 % and 1.18 to 2.22 % in the two reactors, respectively. The result from Gonzalez-Gil and Holliger (2011) indicates propionate readily favors the dominance of Accumulibacter type IIA, and more importantly, this type of Accumulibacter likely determines the robustness of the phosphate removal.

Alternating acetate and propionate

It was found that propionate uptake rate by Competibacter GAOs is far less efficient than that of acetate, while the Alphaproteobacteria GAOs seem to prefer propionate than acetate. In contrast, Accumulibacter PAOs are able to take up both acetate and propionate at comparable rates (Oehmen et al. 2005b). Thus, a strategy for obtaining highly enriched PAO cultures was developed and demonstrated by Lu et al. (2006), whereby the carbon source was switched between acetate and propionate in every 1–2 solids retention time (SRT). This control strategy was repeatedly shown to be effective in achieving very high enrichment of Accumulibacter PAOs (i.e., greater than 90 %) by eliminating both Competibacter and Alphaproteobacteria GAOs. Other researchers also suggested that the alternating carbon sources (e.g., acetate and propionate, acetate and glucose) can be an effective approach to provide PAOs a competitive advantage over GAOs (Oehmen et al. 2004; Wang et al. 2010).

Glucose

Glucose and/or its polymers (e.g., starch and sucrose) are commonly found in domestic wastewater, where they play major roles in biochemical mechanisms associated with substrate utilization. Their transformation mechanisms in EBPR have also attracted wide interest. Glucose can directly induce glycogen metabolism of GAOs that may lead to deterioration of the EBPR process. Nevertheless, a relatively good EBPR performance was observed in Zengin et al. (2010) at the beginning of the experiment, where lactic acid bacteria were reported to be responsible for anaerobic glucose assimilation and consequent lactic acid production for PAOs and GAOs’ utilization. However, the appearance of 17 % Gammaproteobacteria, which was closely related to the Competibacter, deteriorated EBPR performance from day 29 onwards. Similar deterioration pattern was also reported by other researchers (Wang et al. 2010). Surprisingly, high phosphorus removal efficiency (91.8–94.4 % of 35 mg/L) over a 100-day operation was achieved in an aerobic/extended idle (AEI) system fed with glucose (Wang et al. 2009), although major carbon transformation was through glycogen (Table 1). It is possible that glycogen could replace PHA to provide energy for phosphate uptake, while the microbial community, detailed metabolism, and long-term performance should be evaluated.

Table 1 Summary of stoichiometric ratios of carbon transformation during the anaerobic and aerobic phases with different carbon sources
Full size table

Alcohols

In order to improve EBPR performance, it is common to add external carbon source in wastewater treatment plants. The choice of the external carbon source is critical in terms of operating cost and EBPR performance (Puig et al. 2008). VFAs are preferred carbon sources while it may considerably increase the overall treatment costs. Compared with VFAs, alcohols (i.e., methanol and ethanol) are more economical and sustainable (Wang et al. 2013).

Methanol is a simple and economical alcohol in the market. However, it has been shown that methanol is inappropriate for biological phosphorus removal. Randall et al. (1997) showed that methanol could not be used as carbon source for a glucose-fermentation-products-fed culture. Louzeiro et al. (2002) studied the effect of methanol on a simultaneous N and P removing culture. They concluded that despite methanol was probably not utilized by PAOs, methanol addition may deplete the available nitrate via denitrification and thus allowed EBPR to take place. The feasibility of using methanol for EBPR was also studied by Puig et al. (2008). Similarly, anaerobic P-release was poor with methanol compared with acetate and propionate (7.2 vs 17.0 and 18.2 mg P-PO4 3−/L). PHA storage was also limited with major composition as poly-β-hydroxyvalerate (PHV) (91.7 %). Although P-uptake can be observed in aerobic phase, P-removal efficiency with methanol was far lower than that with VFAs. Recently, a novel process was designed and implemented successfully with methanol as the only carbon source (Taya et al. 2013). In that process, methanol-degrading acetogens were selected and subjected to conventional EBPR conditions. The results showed that EBPR performance was able to sustain in a period of 30 days with a P/C ratio between 0.38 and 0.54 P-mol/C-mol.

Puig et al. (2008) proved that ethanol could be successfully taken up by PAOs with PHV (81.9 %) as major PHA during anaerobic phase and achieved P-removal efficiency of 87.5 %. The P-release and P-uptake rates by using ethanol, acetate, and propionate were comparable. They concluded that ethanol can be considered as long-term additive for EBPR. In another study, P-removal efficiency was 98.5 ± 0.5 % in an AEI system and 85.8 ± 1.6 % in an A/O system by using ethanol as carbon source (Wang et al. 2013). Therefore, ethanol as external substrate to enhance EBPR seems to be more feasible.

In fact, ethanol can select specific PAO and GAO population. For example, Accumulibacter clade IIF genomes contain acetaldehyde dehydrogenase that can process ethanol (Skennerton et al. 2015). Ethanol can also be metabolized by newly discovered PAO ‘Candidatus Halomonas phosphatis’ under both anaerobic and aerobic conditions (Hien et al. 2012). On the other hand, both Cluster 1 and Cluster 2 Defluviicoccus (GAOs) were not able to take up ethanol as evidenced by fluorescence in situ hybridization (FISH) combined with microautoradiography (MAR) (Burow et al. 2007). As such, ethanol can be applied in EBPR to selectively enrich specific PAO groups.

Glycerol

Glycerol is a byproduct from biodiesel fuel production industries. It has been used as an external carbon source for denitrification previously (Akunna et al. 1993). The feasibility of glycerol to support EBPR was evaluated by Yuan et al. (2010) in an acetate-fed EBPR system. They concluded that glycerol was unable to induce EBPR metabolisms unless it was pre-fermented to VFAs. However, Guerrero et al. (2012) reported that it is not necessary to have a separate fermentation step. A single-sludge SBR system with a longer anaerobic phase (4 h anaerobic and 3.5 h aerobic phases) demonstrated efficient net P-removal. The prolonged anaerobic phase would allow degradation of glycerol to VFAs (mainly propionate) that can support PAOs’ metabolism. In this case, the direct carbon source for PAOs was in fact propionate. A low anaerobic P/C ratio (P mol/C mol glycerol uptake) was observed (Table 1). The average distribution of the PHA at the end of the anaerobic phase was 25 % poly-β-hydroxybutyrate (PHB), 45 % PHV, and 30 % poly-β-hydroxy-2-methylvalerate (PH2MV). Enrichment of PAOs was 46 % and a significant amount of GAOs, mostly Defluvicoccus vanus, was present (37 %) of the total bacteria.

Amino acid

Members of the Actinobacterial genus Tetrasphaera are highly abundant in many full-scale EBPR plants where they can be accounted for up to 30 % of the total bacterial population (Kristiansen et al. 2013). Tetrasphaera-related PAOs can take up a mixture of amino acids in wastewater. However, they are not able to use short-chain fatty acids (e.g., acetate), glucose, or ethanol. Their physiology is markedly different from Accumulibacter (Kong et al. 2005). Another putative PAO that is commonly dominant in wastewater treatment plants is Rhodocyclus-related PAOs. Chua et al. (2006) reported Rhodocyclus-related PAO constituted 17, 9, 8, and 7 % of the sludge community in four investigated plants. The Rhodocyclus-related PAOs are capable of assimilating acetate, aspartate, and glutamate under anaerobic condition. Surprisingly, Chua et al. (2006) showed that Rhodocyclus-related PAOs were also able to accumulate poly-P aerobically with aspartate and glutamate present. However, the metabolism of assimilating aspartate and glutamate by Rhodocyclus-related PAOs under aerobic condition is not clear.

Zengin et al. (2011) conducted a detailed study on the response of EBPR sludge to aspartate and glutamate as carbon sources. Rhodocyclus-related PAOs seemed to be more active in the aspartate-fed batch tests, whereas, coccus shaped Actinobacterial PAOs were only observed in glutamate-fed reactor. A higher content of PHA was observed in the aspartate-fed tests compared to glutamate one (16.84–71.90 mg C/L vs. 6.47–36.27 mg C/L). It is interesting to note that Actinobacterial PAOs, which were more dominant in glutamate-fed reactor, seemed incapable to store PHA as evidenced by in situ staining (Kristiansen et al. 2013), and it was hypothesized that PHA accumulation was mainly carried out by non-Actinobacterial PAOs. With amino acids as carbon sources, PAOs can employ a wider variety of metabolism. For instance, PAOs can assimilate aspartate or glutamate simultaneously with poly-P accumulation under aerobic condition (by Rhodocyclus-related PAOs). Further, PHA synthesis was not necessary for Actinobacterial PAOs under anaerobic phase.

Carbon sources involved in tricarboxylic acid (TCA) and glyoxylate cycle

Fumarate, malate, and oxaloacetate are intermediate products involved in the tricarboxylic acid (TCA) and/or glyoxylate cycle. The potential to apply these intermediates for EBPR has not been widely investigated. To date, only one report by Zafiriadis et al. (2013) studied the EBPR performance fed with fumarate, malate, or oxaloacetate. The results demonstrate that there was no typical EBPR metabolism using these carbon sources. It appeared that PAOs remained present during the experimental period but they were inactive regarding their P-removal ability. The P-removal efficiency was only 16, 5, and 10 % for fumarate, malate, and oxaloacetate as substrate. Fumarate and malate favored PHV formation while xaloacetate favored formation of PHB. The poly-P hydrolyses activities were also affected by the kind of carbon sources. Acid phosphatase and endopolyphosphatase displayed higher activities during propionate utilization followed by those for oxaloacetate, fumarate, and acetate, while low activities were detected when malate was utilized as the sole carbon source. Therefore, TCA cycle intermediates seemed not to be suitable as carbon source for PAO’s anaerobic metabolism.

Phosphate and carbon transformation using different single carbon source

Phosphate and carbon transformation ratios are important indicators for EBPR performance. When acetate is used as carbon source, specific P-release rate is around 0.45–0.73 P-mol/C-mol HAc (Carvalho et al. 2007; Hou et al. 2008; Yagci et al. 2007; Zhu et al. 2011) under anaerobic conditions. PHB is the major PHA component, which accounts for 63–100 % of total PHA. The transformation ratios of PHA/HAc and glycogen/HAc (Gly/HAc) are summarized in Table 1 (Carvalho et al. 2007; Yagci et al. 2007; Zeng et al. 2013). Many studies used P-release/HAc and Gly/HAc ratios as indicative parameters for enrichment of PAOs. Based on stoichiometric values in PAOs metabolic model (Smolders et al. 1994), when Gly/HAc ratio is higher than 0.50, it may suggest the existence of GAM. Aerobic phosphate uptake to anaerobic phosphate release ratio is in the range of 1.13 to 1.39 (Abughararah and Randall 1991; Yagci et al. 2007; Zhu et al. 2011).

Anaerobic P-release to HPr uptake (P/C) ratios and PHA/HPr ratio are summarized in Table 1. Unlike the acetate anaerobic metabolism, which mainly induces PHB production, the major PHA fractions produced with propionate are PHV (44.1–84.3 %) and PH2MV (43.2–52.3 %) (Carvalheira et al. 2014; Hsu et al. 2013; Oehmen et al. 2005a; Oehmen et al. 2005c; Vargas et al. 2011; Zeng et al. 2013). The Gly/HPr ratio ranges from 0.05 to 0.49 C-mol/C-mol HPr (Carvalheira et al. 2014; Oehmen et al. 2005a; Oehmen et al. 2005c). Carvalho et al. (2007) reported that the Gly/C ratio was lower in propionate reactor than that in acetate reactor (0.32 vs. 0.69 C-mol/C-mol). This means the involvement of GAM is lower with propionate than that with acetate, which then further supports that propionate favors PAOs to GAOs.

Although glucose was common in domestic wastewater and many researchers employed it in EBPR studies, glucose was proved not to be a suitable carbon source for efficient EBPR. Lower P-release/glucose-uptake ratios were obtained in studies using glucose as carbon source, e.g., 0.0059–0.121 P-mmol/C-mmol (Pijuan et al. 2009; Wang et al. 2010). PHA formation was also in a lower range compared with VFAs as carbon source (Table 1). Synthesized PHA was a mixture of PHB and PHV with a higher content of PHV (accounted for 77.0 %).

Zengin et al. (2011) revealed carbon and P transformation using aspartate and glutamate. The anaerobic P-release to carbon uptake ratios was calculated as 0.73 P-mol/C-mol glutamate and 0.55 P-mol/C-mol aspartate. PHA/C ratios were measured as 0.95 C-mol/C-mol for aspartate and 0.61 C-mol/C-mol for glutamate, respectively. For both carbon sources, P-uptake/P-release ratios varied in the range of 1.14–1.17 which indicated the good performance of P-removal (Table 1).

According to the PAOs models developed by Smolders et al. (1994) and Oehmen et al. (2005c), the stoichiometric ratios summarized above with acetate and propionate as the carbon sources mostly agree with the models (Table 1). When glucose is used as the sole carbon source, all the stoichiometric ratios under anaerobic phase are much lower compared with both models with P/C ratio being particularly low. On the other hand, these ratios with ethanol as the sole carbon source are close to the acetate model which supports ethanol can be used as an alternative source to enhance EBPR. Glycerol despite could not be used by PAOs directly; it can be anaerobically converted to propionate that can in turn be taken up by PAOs. The conversion ratio of glycerol to propionate is about 0.5. This causes low P/C, PHA/C, and glycogen/C ratios observed under anaerobic conditions. In other words, the glycerol utilization efficiency by PAOs is not high. For glutamate and aspartate, higher P/C ratios are obtained under anaerobic condition. Considering the metabolisms of the active PAOs (i.e., Rhodocyclus-related PAOs and Actinobacterial PAOs) are markedly different to Accumulibacter, the implication of PHA/C and glycogen/C ratios on EBPR performance may require more investigation.

Complex carbon source

Yeast extract and peptone

Yeast extract and peptone are commonly used as substrates for culture growth. The feasibility to support EBPR using yeast extract and peptone was studied by Fukushima et al. (2007). The carbon sources in feed solution contained peptone (40 % in terms of COD), yeast extract (50 %), and acetate (10 %). The results showed that the abundance of both Accumulibacter and bacteria affiliated with Actinobacteria were around 10 %. Whereas the batch reactors fed with acetate, aspartate, or glucose showed high phosphorus content which was more than 9 % of mixed liquor volatile suspended solids (MLVSS) and Accumulibacter accounted for over 20 % of all bacteria. However, the results of P-removal performance were not detailed in that paper.

Mixed carbon source

Researchers have also investigated the synthetic carbon mixtures for EBPR application since the carbon sources in wastewater streams generally involve much more diverse compounds. VFAs, glucose, and amino acids are commonly prepared in different mixing ratios. Propionate is particularly of interest in many studies. As discussed in single carbon source section, propionate can be a better carbon source to maintain a stable EBPR process. A wastewater with higher propionate content may also result in a more stable EBPR performance and higher PAO abundance. Chen et al. (2005) reported that a better P-removal performance was obtained in the reactor acclimatized with a higher HPr/HAc wastewater as compared to the reactor receiving low HPr/HAc ratio wastewater (95.76 vs. 65.75 %).

However, results from Ahmed et al. (2008) showed otherwise conclusion. VFA, methanol, and glucose were applied in different mixing ratio in a sequencing anoxic/anaerobic-aerobic membrane bioreactor (SAM) system (Table 2). Propionate-dominant carbon source SAM (SAM-Pr) did not display a better performance in terms of P-removal efficiency as compared to that of acetate-dominant carbon source SAM (SAM-Ac). The lowest P-removal efficiency was observed in glucose-dominant carbon source SAM (SAM-Gl, 24.6 %). This implies that glucose indeed is not suitable to maintain stable EBPR performance. Interestingly, P-removal efficiency in methanol-dominant experiment (SAM-Me) was also higher than that in SAM-Pr (79.8 %). The microbial community and P-removal efficiency are summarized in Table 2. The results from this study indicate that P-removal efficiency was not directly correlated with the abundance of Accumulibacter or total PAO population.

Table 2 Summary of P-removal efficiency and PAOs community with different mixed carbon sources. Adopted from Ahmed et al. (2008)
Full size table

Starch has also been evaluated as potential carbon source for EBPR. Wei et al. (2014) investigated three EBPR reactors fed with mixture of acetate and starch at different ratios. The results revealed that phosphorus removal efficiency decreased with the increase of starch content. It was found that hydrolysis and acidification of starch were necessary to led to VFA generation in the early period of the anaerobic stage. Three mixing ratios were tested. The higher starch content might result in low PHA accumulation and consumption, thereby limiting the amount of phosphate uptake. Previously, Randall et al. (1997) also reported that very minor EBPR phenotype was observed using starch as the carbon source.

Wastes to carbon sources

Crude glycerol (CG)

CG, a byproduct from biodiesel manufacturing, is an organic rich waste. The primary components of CG are glycerol and residual ethanol or methanol. As discussed in section Alcohols and Glycerol , all these individual compounds can be used as carbon source for EBPR processes. Coats et al. (2015) suggested that CG could be used as an organic carbon supplement to accomplish and/or help maintain EBPR stability. Despite there was no clear phenotype of EBPR, the raw CG-fed system could achieve excellent phosphorus removal and showed a PAO population fraction of 0.11 to 4.3 % and GAOs fraction of 65.3 %. CG can also be applied to EBPR system after pre-fermentation, where VFA can be generated (CG-VFA-fed system). However, it was found that the fermented products could not maintain a stable EBPR performance, even though the GAOs population was only 4.9 % in the CG-VFA-fed system. Thus, the authors concluded that successful EBPR can be achieved even with a small fraction of PAOs present in raw CG-fed system. The actual reasons are not clear. In addition, Guerrero et al. (2015) investigated the role of CG in an EBPR system for nitrogen and phosphorus removal. CG was proved to be useful to prevent the EBPR failure that was caused by NOX under anaerobic conditions. The addition of CG favored EBPR because of the extra carbon supply for PAOs and denitrifiers. Additional VFA was also generated via fermentation under anaerobic conditions. All mentioned studies demonstrated that addition of CG could improve EBPR performance.

Sludge and wastes

The performance of both denitrification and biological phosphorus removal highly depends on the availability of readily biodegradable carbon source. From the above discussion, VFAs are certainly preferred carbon sources to improve EBPR performance. However, it is uneconomical and unsustainable to dose commercial VFAs in full scale wastewater treatment plants. Other than glycerol fermentation, VFAs can also be generated from other organic wastes. Wasted sludge fermentation is one way of producing such substrates. Soluble products from sludge fermentation are mainly short-chain fatty acids with two to five carbon atoms (Moser-Engeler et al. 1998), which can be directly used as carbon source for many bioprocesses. Sludge disposal, on the other hand, is of great concern for many wastewater treatment facilities. Therefore, by applying VFAs fermented from the wasted sludge to EBPR system can reduce the sludge volume and solids content as well as provide additional carbon for EBPR and denitrification.

It is noteworthy that P-removal efficiency was in fact higher with fermented liquor as supplementary carbon than that with acetate alone (98.7 vs. 71.1 %) (Tong and Chen 2007). The specific P-release and P-uptake rates were also superior to acetate-fed EBPR. It was found that when fermented liquor was used as carbon source, less PHA was used for glycogen synthesis while more PHA was utilized for aerobic P-uptake. It is possible that PAOs may use the energy derived from PHA degradation for P-uptake rather than glycogen replenishment. This, however, requires further exploration. In another study, fermented liquor was used in a simultaneous nitritation, denitritation, and phosphorus removal (SNDPR) process to reduce the competition between PAOs and denitrifiers (Ji and Chen 2010). Similarly, the P-removal efficiency with sludge fermentation liquid was found higher than that with acetate, and PAOs were also more dominant in sludge fermented liquor-fed SBR. Furthermore, EBPR can also be coupled with enhanced solids treatment process. Park et al. (2011) explored the effectiveness of the ozonolysate as a carbon source for EBPR through a series of batch and continuous experiments. That study presented EBPR could be enhanced when coupled with a sludge ozonation process by providing the supernatant of the disintegrated sludge as a carbon source.

Many types of wastewater, e.g., abattoir wastewater, pulp and paper wastewater, palm oil mil effluent (POME), and agro-food industrial wastewater, are characterized as high strength organic wastewater. Depending on the wastewater origin, some of the wastewater contains large amount of readily used VFAs, while the others may require pre-fermentation to produce simpler organic compounds. These types of wastewater with proper pre-treatment (if necessary) can be explored as external carbon source for wastewater treatment systems. For instance, the addition of agro-food wastewater to domestic wastewater would greatly enhance EBPR performance (Fernández et al. 2011). Thus, it is an economical and environmental friendly concept to use waste to deal with waste.

Remarks

In order to maintain stable EBPR performance, types of carbon sources in wastewater should be carefully evaluated. When choosing an external organic matter for system enhancement, both system performance and operating cost need to be considered. Table 3 analyzes the suitability of reported carbon sources for EBPR in the comparison of system performance, cost, feasibility, and favored microbial community. Among the carbon sources discussed, short chain VFAs, e.g., acetate and propionate, widely exist in wastewater and are the most popular carbon sources for EBPR. Generally, propionate is suggested as preferred carbon sources under normal operating conditions and system configurations. However, certain system conditions may deliver different results. It should be noted that high COD/P ratio and/or high VFA loading may deteriorate EBPR due to proliferation of GAOs. Propionate may favor both Accumulibacter PAO and Defluvicoccus vanus GAO. Alternating VFAs feeding strategy can be very useful to enrich a PAO culture in the laboratory, while there is limitation for its full-scale application. Glucose on the other hand is certainly not an option to improve EBPR process according to reported studies, given the fact that Competibacter GAOs can easily outcompete PAOs with glucose as carbon source. To date, only one study successfully demonstrates using methanol for EBPR enhancement, while ethanol is more popular to assist EBPR and select certain PAOs population (e.g. Accumulibacter clade IIF PAOs). Aspartate and glutamate can be good candidates in amino acid group with Actinobacterial and Rhodocyclus-related PAOs enriched in the systems.

Table 3 Evaluation of different carbon sources from different angles
Full size table

Nevertheless, when cost is a concern, alternate candidates can be CG, fermented sludge or even liquor from sludge pretreatment. For this group of carbon sources, pretreatment or prefermentation is generally required, which can promote production of readily-use carbon source. Actually, side-stream fermentation is currently used in countries such as South Africa and the USA for internal carbon source production for BNR processes (Latimer et al. 2007). However, it is not clear if a long-term stable EBPR can be achieved with the complex carbon sources generated from waste/sludge treatment processes. The mechanisms of fermented carbon sources uptake and their impact on PAO/GAO competition requires further exploration. Further, the potential effect from recalcitrant compounds derived in fermentation step on EBPR should also be investigated.