3 Results and discussion

3.1 Reactor performance analysis

3.1.1 Decolorization performance

The decolorization performance of the three reactors was evaluated over a period of nearly 100 days, as shown in Figure 2. All three reactors showed high decolorization efficiency for RB5, but the MBBR reactor exhibited better decolorization stability. One-way ANOVA results indicated significant differences (p <0.05) in RB5 decolorization rates among the three reactors, which was probably due to the fluctuation observed in the biochar and PPC reactors. At around day 21, samples of the carrier material were taken and preserved, and almond shell biochar was added to the biochar reactor to supplement the insufficient amount of almond shell biochar. Despite a slight decline due to the lack of biochar, the biochar reactor’s discoloration rate for RB5 remained stable above 80%. Around day 70, partial detachment of the biofilm in the PPC reactor caused fluctuations in its treatment performance for RB5. Overall, all the reactors’ treatment performance was relatively stable, but the partial detachment of the biofilm caused fluctuations in the biochar and PPC reactors. Obvious fluctuation in the decolorization rate was observed in the biochar and PPC reactors.

3.1.2 Analysis of RB5 degradation pathways

(1) UV-Vis analysis
The high concentration of dye in the influent exceeded the upper limit of the UV spectrophotometer; therefore, a four-fold dilution was necessary for both the influent and effluent samples. Figure 3 presents the full spectrum scan results of the diluted samples. The characteristic peaks of RB5 at 310 nm and 595 nm, which were caused by the auxochrome and chromophore of RB5, were significantly reduced (Droguett et al., 2020). The biological decolorization of the dye could have resulted from adsorption onto biochar or biodegradation. Biochar adsorption caused all peaks in the full spectrum to decrease proportionally, while biodegradation was accompanied by the disappearance of absorption peaks and the formation of new peaks (Chen et al., 2003). The UV-Vis absorbance of the biochar reactor effluent was slightly lower than that of the MBBR and PPC reactors in the 260-700 nm range, which could be attributed to the adsorption effect of biochar. The characteristic peaks of RB5 in the effluent at 310 nm and 595 nm gradually decreased and tended to be flat after passing through the biochar, MBBR, and PPC reactors, indicating that the auxochrome and chromophore of RB5 were destroyed, and the reactors had a biodegradation effect on RB5. A new characteristic peak was formed at around 250 nm and 380 nm in the effluent of the biochar, MBBR, and PPC reactors, which might have been due to the absorption peak of bacterial metabolic products or colorless intermediate products (Daneshvar et al., 2007). According to the location of the new absorption peak (200-400 nm), it could be preliminarily determined that they were likely to be monocyclic compounds, and it was speculated that substances such as amine might have been produced (Tan et al., 2016). It is speculated that there might be some similarity in the effluent products between the biochar reactor and the MBBR/PPC reactors.
(2) FTIR analysis
The FTIR analysis (Figure 4) revealed changes in the positions and intensities of certain functional groups in the effluent of biochar, MBBR, and PPC reactors. In the high wavenumber region (4000-2500 cm-1), stretching vibrations of O-H, N-H, C-H, and S-H bonds were typically observed. The peak at 3386.7 cm-1, which might have corresponded to the stretching vibration of -NH2 bonds, appeared as a broad peak. The variation in peak area indicated differences in the content of functional groups. The range of stretching vibration of C-H bonds was 2850-2710 cm-1, with the peak at 2829.67 cm-1 corresponding to the stretching vibration of -CH2 bonds. The peak at 2829.67 cm-1in the effluent of biochar, MBBR, and PPC reactors was smaller than that in the influent, indicating replacement of some -CH2bonds.
In the wavenumber region below 1900 cm-1, stretching vibrations of -C=C-, -C=N-, -C=O bonds, and skeletal vibrations of aromatic rings were observed, indicating the presence of benzene rings. The peak at 1589.77 cm-1 might have corresponded to the stretching vibration of -N=N- bonds, which became weaker after decolorization, indicating that -N=N- bonds might have been destroyed during the degradation process. The stretching vibration of the R-SO2-R group was observed in the range of 1370-1290 cm-1 (Martorell et al., 2017). Additionally, a weaker peak at 1117.43 cm-1 might have corresponded to the bending vibration of -OH bonds. The characteristic peak at 835.20 cm-1 in the effluent of biochar, MBBR, and PPC reactors suggested the presence of polycyclic aromatic hydrocarbons, and the appearance of a small peak at 696.62 cm-1indicated the occurrence of out-of-plane bending vibrations of the N-H bonds of amine compounds. The effluent of biochar, MBBR, and PPC reactors might have generated similar compounds.
(3) LC-MS analysis
Figure 5 displays the liquid chromatography profiles of the influent, biochar, MBBR, and PPC reactors. The chromatograms illustrate significant variations between the effluents of the biochar reactor and the MBBR and PPC reactors. In the MBBR and PPC reactor effluents, a prominent peak was observed around 9.86 min, which was absent in the influent chromatogram. This indicates that the production of organic compounds with higher concentrations occurred in the MBBR and PPC effluents that were not present in the biochar reactor effluent. It should be noted that the soluble nature of RB5 in water prevented its extraction in the organic phase, and therefore it was not detected in the influent liquid chromatography.
Based on the findings in Figure 6, it can be concluded that there was a significant increase in the ion peak at m/z 399 in both the MBBR and PPC reactor effluent at 9.86 min. The molecular ion peak [M-H+] + at m/z 399 suggested a relative molecular mass of 400 for this substance, indicating that it was only present in the effluent of the MBBR and PPC reactors. It is tentatively suggested that this substance might be a benzene-containing structure formed by further ring-opening of the naphthalene ring structure in the active black 5 molecule after the double azo bond was opened. The maximum absorption peak at around 250 nm in the characteristic absorption of the benzene ring supports this conclusion. However, the results also indicate that the MBBR and PPC reactors were unable to completely mineralize the dye, and additional reaction systems may be required to achieve thorough degradation.
The UV-Vis and FTIR analysis results indicated that the effluent from the biochar reactor shared similarities with that of the MBBR and PPC reactors. This suggests that these reactors may have produced similar substances during the RB5 degradation process. However, the LC-MS analysis showed significant differences between the biochar reactor and the MBBR and PPC reactors. While parallel sample repeatability was good, the degradation pathways of RB5 in the biochar and polymer carrier reactors differed to some extent. High-efficiency degradation bacterial isolation results showed that the microbial species responsible for the RB5 degradation in the polymer carrier MBBR and PPC reactors were similar. Therefore, it is possible that the differences and overlaps in the degradation mechanism can be attributed to the presence of high-efficiency degradation bacteria. The carrier material could have an impact on the high-efficiency degradation bacterial community in the reactor, which may affect the RB5 degradation mechanism.

3.2 Analysis of microbial community structure in reactors

3.2.1 Analysis of microbial community structure on the carrier

(1) Statistical analysis of diversity indices
Based on Table 1, the ACE and Chao1 indices indicated that P1 had the highest richness, while Simpson and Shannon indices showed that P1 had good evenness. M2 and P1 had similar richness, as shown by the ACE and Chao1 results. A1 had good evenness according to the Simpson index, while P1 had good evenness according to the Shannon index. Overall, the richness of microorganisms on the biochar carrier was lower than that on MBBR and PPC, possibly due to the smaller pores of the almond shell biochar carrier, which could be easily covered by a dense biofilm, limiting dissolved oxygen diffusion into the inner layer of the carrier, and leading to decreased overall richness. Simpson and Shannon observations showed that the evenness of microorganisms on MBBR carriers was slightly lower than that on biochar and PPC carriers. The PD_whole_tree results indicated that the abundance of microorganisms on MBBR and PPC was higher than that on biochar, suggesting differences in microbial diversity among different carriers.
(2) Species distribution histogram
Based on the results presented in Figure 7a, the dominant phyla on the three different carriers were Proteobacteria , Firmicutes , and Bacteroidetes . Specifically, the biochar carrier had a higher proportion of Firmicutes (53.3±2.0%),while the MBBR and PPC carriers had relatively abundant Proteobacteria andBacteroidetes . Previous studies had shown thatProteobacteria (Wang et al., 2008a) and Firmicutes(Ramalho et al., 2007a; You and Teng, 2009) were effective in degrading RB5, and a mixed community of Bacteroidetes could also effectively degrade RB5 (Forss et al., 2013). Synergistetes were more abundant on the biochar carrier, while Chloroflexi bacteria were more commonly found on the MBBR and PPC carriers.
At the family level (Figure 7b), the dominant families on the biochar carrier were f-Family_XIV , f-Peptococcaceae , andf-s74e-6049 , while on the MBBR carrier, the dominant genera wereParacoccus , Flavobacterium , and Taibaiella , and on the PPC carrier, the dominant genera were Paracoccus ,f-Peptococcaceae , Desulfuribacillus , andDesulfitibacter . Paracoccus had been shown to effectively degrade dyes (Bheemaraddi et al., 2014). The results suggested that the carrier played an important role in the growth and enrichment of microorganisms. Different carriers had different microbial functional structures, depending on the characteristics of the carrier material, porosity, surface area, and other factors (Wang et al., 2014). In addition, there were very low levels of Shewanella on all three reactor carriers, with only 0.01%±0.004%, 0.01%±0.01%, and 0.004%±0.006%, respectively, which might have been due to competition between different genera (Hibbing et al., 2010; Zhu et al., 2019).
The microbial community structure on the carriers in this study was complex, and the efficient degrading bacteria for RB5 was likely to be a microbial community. Previous studies had shown that microbial communities could effectively degrade RB5. For example, the natural microbial community DDMZ1, composed of Burkholderia ,Achromobacter , Stenotrophomonas , Pseudomonas , andEnterobacter , showed stable decolorization efficiency for RB5 (Zhang et al., 2019). Another study isolated a microbial community consisting of A. hydrophila, R. mucilaginosa , G. pseudocandidum , and E. coli from dye wastewater and found that this community achieved a decolorization rate of over 95% for RB5 within 24 hours, with a more stable decolorization effect than a single strain (Flórez Restrepo et al., 2018). Compared to single strains, mixed microbial communities had stronger advantages in completely mineralizing azo dyes due to the synergistic effect between different degrading microbial communities, which could better degrade azo dyes (Xie et al., 2020; Saratale et al., 2010).

3.2.2 Isolation of high-efficiency degradation bacterial community

The effluent from the reactors was acclimatized with increasing concentrations of RB5 (Figure 8), starting from 100mg/L. After 8 hours, no significant discoloration effect was observed, and the OD600 did not show a significant increase. However, gradual discoloration began to appear in the effluent of all three reactors after 8 hours. At a concentration of 200 mg/L, discoloration was observed in the effluent of all three reactors at 4 and 8 hours. At a concentration of 300 mg/L, discoloration first appeared in the MBBR reactor group at 4 hours, with MBBR reactor group having a slightly higher OD600 than biochar reactor group and PPC reactor group. This early discoloration may be attributed to the effect of acclimatization and the accumulation of a large number of degrading bacteria. At 24 hours, all three reactors had completed discoloration, with MBBR reactor group having the highest OD600. At 48 hours, the discoloration rate of all three reactors was nearly 97%, with MBBR reactor group showing the highest OD600, indicating the completion of acclimatization. Subsequently, the enriched liquid after acclimatization was diluted and spread plated five times, resulting in the isolation of highly efficient RB5-degrading bacterial communities from the effluent of the three reactors.

3.2.3 Analysis of the RB5 efficient degradation bacterial community structure

(1) Statistical analysis of diversity indices
Table 2 shows that the ACE and Chao1 indices were highest in M1 and M2 in the MBBR reactor, indicating a higher abundance of the microbial community with high RB5 degradation efficiency in the MBBR reactor. Additionally, the Simpson and Shannon indices were also highest in M1 and M2, suggesting greater diversity in the MBBR reactor. In contrast, A1 and A2 had the lowest abundance based on the ACE and Chao1 indices. Overall, the abundance of the microbial community with high RB5 degradation efficiency was smaller in the biochar reactor than in the MBBR and PPC reactors based on the ACE and Chao1 indices. However, based on the Simpson and Shannon indices, the functional bacterial composition was more complex in the MBBR reactor than in the biochar and PPC reactors, indicating greater diversity in the MBBR reactor. The PD_whole_tree analysis showed that the abundance of the three reactors was similar.
(2) Species distribution histogram
At the phylum level, Figure 9a shows that in the efficient RB5 degrading bacterial communities of the biochar reactor, Clostridiales andPseudomonadales had the highest proportions, accounting for nearly 100% of the community. In the MBBR efficient RB5 degrading bacterial community, Pseudomonadales accounted for approximately 27-30%, Enterobacterales accounted for approximately 23%,Burkholderiales accounted for approximately 25-28%, andBacillales accounted for approximately 20%. In the PPC reactor’s efficient RB5 degrading bacterial community, Pseudomonadalesaccounted for approximately 65-67%, and Enterobacteralesaccounted for approximately 31-33%. The biochar reactor had a higher abundance of Clostridiales than the other reactors, butEnterobacterales were not detected in the biochar reactor.Burkholderiales and Bacillales were present in higher numbers in the MBBR reactor than in the biochar and PPC reactors.Burkholderia , a member of the Burkholderiales order, can synergistically degrade RB5 with other bacterial species, which may be why the MBBR reactor showed the earliest decolorization effect among the efficient RB5 degrading bacterial communities.
According to Figure 9b, at the genus level, the highly efficient degrading bacterial communities in the biochar reactor were dominated by the Pseudomonas genus (approximately 62-63%), followed by theClostridium_sensu_stricto_1 genus (approximately 35-37%). In the MBBR reactor, the dominant genera were Pseudomonas(approximately 27-30%), Shewanella (approximately 23-24%),Delftia (approximately 25-28%), and Bacillus(approximately 20%). In the PPC reactor, the dominant genera werePseudomonas (approximately 65-67%) and Shewanella(approximately 31-33%). The Shewanella genus was not present in the biochar reactor but accounted for a large proportion in the MBBR and PPC reactors, possibly due to the electron transfer efficiency and affinity between Shewanella and the carrier (Shi et al., 2007).Clostridium sensu stricto 1 was found only in the biochar reactor and was the dominant species. In the MBBR reactor, a large number ofDelftia and Bacillus species were detected, but they were almost absent in the biochar and PPC reactors. Pseudomonas , which was present in all three reactors, had been extensively documented to have good decolorization effects on RB5 (Wang et al., 2008b; Hussain et al., 2013). Bacillus and Delftia were also capable of effectively degrading RB5 and solubilizing phosphate in the substrate, respectively, indicating their potential synergistic action (Dafale et al., 2008; Liao et al., 2013). In bottle experiments, the efficient degradation microbial community in the MBBR reactor showed earlier decolorization of RB5 than those in the biochar and PPC reactors, possibly due to the diverse functional bacteria present. The efficient degradation microbial communities in the MBBR and PPC reactors had high similarities and had consistent degradation mechanisms.Shewanella , which was artificially inoculated as an RB5-degrading bacterium during reactor setup, did not become the dominant species on the reactor carrier, but screening showed that it was a dominant functional bacterium for RB5 degradation in both MBBR and PPC reactors, working in conjunction with other functional bacteria to effectively degrade RB5.

3.3 The impact of carrier physicochemical properties on reactor performance

Microbial attachment and fixation on carrier surfaces is an important aspect of reactor performance, influenced by both the physicochemical properties of the carriers and the surface characteristics of the microorganisms. Previous studies have shown that carrier roughness, pH, and pore size affect the growth and stability of biofilms in packed bed reactors (Al-Amshawee et al., 2021; Jaroszynski et al., 2011).
Almond shell biochar showed slightly alkaline pH, with a pH value of 8.28±0.20. It is worth noting that most biochar samples reported in the literature had an alkaline pH value (Greenough et al., 2021). The pH of the MBBR carrier was 7.41±0.22, which was also slightly alkaline. Alkaline conditions were conducive to microbial growth and the development of microbial communities on the surface of the MBBR carrier, as well as the formation of biofilms and diverse functional bacteria (Dai et al., 2021). The pH of the PPC carrier was 6.80±0.04.
Surface roughness was an important factor in carrier selection and reactor design, as it influenced mass transfer efficiency, effective contact area, and protection of microorganisms (Messing and Oppermann, 1979). As shown in Figure 10, biochar carrier had a highly porous surface with small pore sizes (0.1 - 0.5 μm), which might limit the initial growth of microorganisms (Messing and Oppermann, 1979). In the reactor that effectively degraded bacteria with a length of about 0.6 to 1.5 μm, these pores could not serve as their living place, and microorganisms attached to the top or nearby areas of the pores. After the reactor had been in operation, these pores were filled with microbial metabolites, and the pore structure was no longer obvious, resulting in the appearance of a large number of spherical substances. Most of the surface of the biochar carrier was covered with a layer of attached biofilm growth, and there were rod-shaped bacteria present on the surface and grooves of the biochar carrier.
In contrast, MBBR carriers had many protrusions, grooves, and micro-cracks on their surface, and their overall structure had a toothed shape. The high surface roughness of the carrier resulted in slow growth of biofilm on it, which could form a biofilm structure that was more resistant to flushing (Al-Amshawee et al., 2020). There were mainly two reasons for this. Firstly, compared with smooth surfaces, the rough surface of the carrier increased the effective contact area between bacteria and the carrier. Secondly, the rough parts of the carrier’s surface, such as holes and cracks, protected the attached bacteria (Al-Amshawee et al., 2021). After the reactor operated, microorganisms accumulated in large numbers on the carrier surface, and the biofilm structure became compact, with spherical and rod-shaped bacteria interweaving and growing together. As the biofilm matured, it became more stable and less likely to detach, which was beneficial for the sustained and stable operation of the reactor.
The PPC carriers had an average diameter of 1cm and a relatively smooth surface, but with many large pores ranging from 0.5 to 1.5 μm in diameter. These pores could serve as attachment and survival sites for microorganisms, providing protection against hydraulic shear stress. However, some studies had shown that the porous structure of biofilm carriers might have led to low mass transfer efficiency (Cao et al., 2017). After the reactor operates, these large pores were filled with a large number of rod-shaped bacteria, which grew abundantly and covered the pores and their surroundings. The microbial abundance on the surface of PPC carriers was relatively high, and this carrier could effectively enrich microorganisms. However, there was a risk of the biofilm being washed away from the polyurethane carriers (Al-Amshawee et al., 2020).
Initially, all carriers had clean surfaces without any attached biofilms. After reactor operation, dense biofilms formed on all carriers. The pore size of the almond shell biochar was too small to serve as a habitat for microorganisms. As a result, the abundance of microorganisms on the almond shell biochar carrier was lower than that on the MBBR and PPC carriers. The surface of the MBBR carrier was rougher than that of the PPC carrier, and the rough surface reduced the possibility of biofilm detachment (Huang et al., 2018). The MBBR carrier could form a robust and impact-resistant biofilm structure. At the same time, the surface heterogeneity of the MBBR carrier provided a better living space for microorganisms and increased the utilization rate of adsorbed organic matter by attached microorganisms (Dutta et al., 2014). Therefore, microorganisms could more easily utilize adsorbed compounds and convert them into final products.