3.4.1 Microbial Relative Abundance
The inoculum source showed a relative abundance for Archaea and Bateria Domain as 40% and 60%, respectively (Fig. 6). In the sample from R1 (Fermentative Reactor), the relative abundance was just 4% for the Archaea Domain, while it was 96% for Bateria. Thus, the effectiveness of inoculum pre-treatment and operational conditions on microorganisms selection were confirmed. As discussed previously, those can be claimed for having selected fermentative bacteria responsible for hydrogen and organic acids production.
A different tendency was observed in the relative abundance for reactors R2 and R3 compared to the inoculum (Fig. 6). The relative abundance for Archaea and Bateria Domain were 27% and 73% for R2. Since R2 was fed with the reactor's effluent without any dilution, a high volatile organic acids concentration can be caused by the inhibition of the Archaea Domain. On the contrary, for the R3, an increase in the relative abundance for Archaea Domain was observed (55%), and this result corroborates with methane production observed in this reactor (85.7% of CH4 in the composition of the biogas).
From the analysis of the relative microbial abundance in the reactors R1, R2 and R3 operated sequentially, it is possible to attest to its efficiency in creating different environments for selecting specific microbial populations to optimize the whole process of residues digestion and biofuels production.
A considerable change in the phyla composition was observed in all reactors compared to the inoculum in natura (Fig. 7). The most abundant phyla observed in the inoculum were Firmicutes (28%), Cloacimonetes (19%), Bacteroidetes (16%), and Proteobacteria (14%). The Phylum Firmicutes is known to group bacteria capable of tolerating adverse conditions by endospores forming and resisting the sludge's pre-treatment; this justifies its dominance in R1 (74%) in comparison to the reactors R2 and R3 [47]. The second most abundant phylum in the inoculum was the Cloacimonetes, which was recently classified, and it clusters syntrophic bacteria involved in the degradation of propionate [48]. Taken into account the inoculum in natura came from a methanogenic UASB reactor, in which the production of organic acids is known, the presence of anaerobic propionate-degrading bacteria was expected. The presence of phyla Bacteroidetes and Proteobacteria were verified in samples from the inoculum, R1, R2, and R3, which groups facultative fermentative and anaerobic bacteria that have certainly contributed to the hydrogen/organic acids production verified during our experiments [49]. In reactors R2 and R3, there was no H2 detection, but instead, methane production, the most abundant phylum was also the Firmicutes (55% and 40%, respectively). It is most likely that this abundancy is due to the diversity of strict and facultative anaerobic bacteria belonging to this phylum that can perform a wide diversity of metabolic pathways such as alcohols and volatile fatty acids [5].
Lactobacillales, belonging to the Bacteria Domain, was the most abundant order in the inoculum (13.1%) (Fig. 7). The presence of microorganisms in this order increased in R2 (21%), while a meager percentage was observed in R1 and R3 (3.4 and 2.3%). Regarding the genus Lactobacillus, there was a relative abundance of 91% in reactor R2. The presence of Lactobacillus is often described in fermentative bioreactors in which acidic pHs are observed [17, 50]. However, the presence of those microorganisms is known to be prejudicial for biohydrogen production due to the inhibition of Clostridia cells by the biocines secreted [42].
A different tendency was observed for the orders Bacteroidales and Clostridiales that showed relatively abundancy increased from 2.5 and 1.5% in the inoculum to 11.8% and 10.6% in R3, respectively. Thus, it is possible that the operational conditions applied to the reactors, including the carbon source used, favored their establishment in the reactors.
Members of this order, such as Clostridium sp. have already been described with the ability to consume different carbon sources through hydrolysis, culminating with organic acids and hydrogen generation, among other metabolic pathways [51]. The high relative abundance of the order Clostridiales in reactor R3 (10.6%) suggests there were still compounds to be consumed or an environmental condition that could favor/select these endospore-forming cells.
In reactor R1, the order with the highest relative abundance was Selenomonadales (61.3%) and genus Megasphaera (29%), which groups lactic acid consuming microorganisms that can produce H2 [52]. However, there was a reduced production of H2 in R1 when compared to the theoretical production.
The maintenance of different hydrogen-producing microorganisms can be related to the synergism between species that have already been reported in several bioreactors. Strains of Enterobacteriaceae were observed contributing to the production of H2 and Clostridium in a CSTR fed with glucose [53]. The synergism between the orders Enterobacteriales and Bacteroidales was reported in packaged bed reactors in which Clostridium spp. developed an association with Klebsiella and Prevotella, which can agglutinate with other microorganisms [49, 54]. These groups of microorganisms in the R1 reactor may explain the low relative abundance of individuals of the order Clostridiales in a fermentative reactor with H2 production. The microbial consortium can benefit from this synergism between species and stand out in the production of H2 when compared to communities dominated by Clostridium.
Luo et al. [55] used a continuous flow bioreactor inoculated with thermally treated agricultural soil and fed with glucose. They observed a diversified selection of microbial communities (Selenomonas, Enterobacter, and Clostridium spp). This fact resulted in the highest hydrogen production yield compared to the community dominated by Clostridium observed in higher OLRs.
For the Archaea domain present in the acidogenic reactor (R1), the relative abundance of 4% of Methanobacteriales was higher when compared to the inoculum (0.8%). In addition, these known H2-consuming microorganisms are also capable of assimilating methanol [56], which may be present in crude glycerol, favoring its permanence in R1 and justifying the reduced H2 production.
The order Methanosarcinales had the highest relative abundance in the methanogenic reactors, presenting 17.6% and 35.7% in the R2 and R3 reactors, respectively. Thus, these microorganisms were likely responsible for the expressive CH4 generation that occurred during the experiments.
The genetic sequencing of the inoculum in natura and the composing samples retrieved from the reactor at the end of phase 1 have clearly shown the microbiota changes due to substrate modification operational conditions imposed on the 3-stage bioreactors.
3.4.2. Microbial Community Structure - DGGE
3.4.2.1 Bacteria Domain
Figure 8 shows a dendrogram and the Jaccard similarity index calculated for the Bacteria Domain. The samples were clustered in 2 main groups: (1) all reactor R1 and (2) samples from reactors R2, R3, and the inoculum.
A high similarity (93%) was verified between the central and end portion of reactor R1 (R1P2b and R1P3b). Probably, the highest OLR and potentially toxic compounds could have impacted the bacterial community of the first section of the reactor R1 (R1P1b) than the intermediate and final sections. However, it is worth emphasizing that the similarity of R1P2b and R1P3b with the R1P1b is 60%.
Another essential comparison is with the inoculum (INOCb) with the bacterial community from reactor R1, which was 44%. In other words, the pre-treatment of the inoculum inactivated populations, causing many of them to disappear along the first reactor. The higher the OLR, the lower the community similarity; organic loading rate shocks act by selecting more resistant individuals [36].
A similar pattern was observed by Silva et al. (2019), who studied a fluidized bed reactor to treat dairy wastewater under different OLRs. 90% similarity was observed at 28.7, 53.2 Kg COD (m3.d)−1 conditions, and 70% at 95.76 Kg COD (m3.d)−1. The piston flow of the AFBR acted selecting specific bacteria on different sections inside the reactor, as the initial portion receives major OLRs, it allows and pushes for a longitudinally oriented selection of species.
In Reactor R2, 74% of similarity was observed between the initial and central portions of the reactor (R2P1b e R2P2b). This cluster was 52% similar to the end portion of this reactor (R2P3b) (Fig. 8). The changes in the bacterial community were expected due to the piston flow and the design of the system set-up operated in series. As per reactors and system configuration, the effluent of one reactor will be used as the influent for the next one contributing to the microbial selection.
To summarize, similarly to what was observed in R2, a methanogenic reactor, the intermediate and end sections of R3 were more similar between themselves. These sections were possibly colonized by microbial populations capable of generating methane from similar substrates. Different results were reported when batch reactors were used for H2 production; higher similarities between their populations were observed 87% by Zhao et al. (2010) and 98% by Maintinguer et al. [35]. Pre-treatments and operational conditions such as higher pH and temperature range play an essential role. This difference may have occurred due to factors such as the reactor configuration and the more extended experimental period (260 days) in our HARFB system, which contributed to species selection, increasing the diversity in the reactors.
According to the Shannon-Wiener index (H index) for the Bacteria Domain, higher diversity was observed in the R3 reactor: 3.17, 3.27, and 3.19 for sampling points P1, P2, and P3, respectively (Table 4). These results are comparable with the inoculum diversity index (3.28). Furthermore, it is noteworthy that the inoculum came from a methanogenic reactor and the operational conditions applied for the R2 and R3 reactors were set to favor methane generation, corroborating the high diversity in the R3 reactor, as observed in the inoculum.
Table 4
Shannon-Wiener index (H index) to Bacteria domain
Inoculum
|
R1
|
R2
|
R3
|
P1
|
P2
|
P3
|
P1
|
P2
|
P3
|
P1
|
P2
|
P3
|
3.28
|
2.60
|
2.91
|
2.80
|
3.25
|
3.15
|
3.09
|
3.17
|
3.27
|
3.19
|
The lowest diversity index (2.60) was found for the first sampling point in R1; this can be explained due to the pre-treatment applied to the inoculum for fermentative microorganisms' selection and acidic pH imposed (5.5). The same was observed by Abreu et al. (2011) in their study for hydrogen production from a mixture of glucose (13 mM) and L-arabinose (16 mM), at 37ºC, and pH 5.5. Additionally, the highest OLR these microorganisms were exposed to also contributed to its lowest diversity. Contrary, reactors R2 and R3, being feed with reactor's R1 and reactor's R2 effluent, respectively, received lower glycerol load in comparison to R1, but certainly higher in volatile organic acids [60–62].
Although showing the lowest diversity index if compared to all reactors, R1's diversity was still high. Pachiega et al. [17] studied batch reactors fed with sucrose and inoculated with granular sludge from a UASB reactor treating brewery wastewater; the authors observed an H index of 0.78 for the sample of tropical sludge pretreated (pH 5.5 + heat treatment). Reactor's type, system configuration, inoculum source (initial diversity), pre-treatment method, carbon source, and operational conditions (OLR, temperature, pH) play a role in microorganisms' selection, and it is impossible to determine which of all the listed factors weighed more. The main drawback of the high microbial diversity is having hydrogen-consuming microorganisms co-existing with hydrogen-producers [63, 64]. Silva et al. [57] observed a correlation between OLR and microbial diversity. For these authors, the higher the OLR, the higher the diversity (2.205 to 2.811); however, the opposite trend was seen for H2 production. In the present study, the unstable and lowest hydrogen production obtained during the start-up phase could be due to the impact of OLR on microbial diversity. The same trend was observed in a study aiming to evaluate the microbial profile of water reservoir sediments against organic matter concentration and nutrients such as nitrogen and phosphorus [65].
In reactors R2 (>3.09) and R3 (>3.17), the microbial diversity was higher in comparison to the first reactor (<2.91). As the system comprises 3-series reactors, the volatile fatty acids and alcohols rich-effluent with reduced glycerol load seem to have played an important role in microorganisms' diversity in R2 and R3. Furthermore, NaHCO3 supplementation in R2's influent to mimic an optimal pH environment for establishing methanogens could have also affected positively for its diversity.
3.5.2 Domain Archaea
DGGE profiles for the Domain Archaea revealed significant structural differences between the inoculum and samples from the reactors after 260 days; the similarity index was less than 15%. This result can be attributed to the pre-treatment applied to the inoculum and the operational conditions that the microorganisms were exposed to during the experimental period. It was also observed that a reduced similarity between the samples of each reactor. In reactor R1, the lowest similarity was between sampling points R1P2a and R1P3a (50%) and in reactor R2 between R2P1a and R2P3a (40%). Based on this analysis, it was possible to observe the impact of inoculum pre-treatment, substrate composition, and operational conditions on the microbial community. The spatial selection of microorganisms throughout the reactor could be attributed to the reactor's configuration – plug flow - as per its definition, the carbon source/substrate the microorganisms were exposed to within the same bioreactors varies (Fig. 9).
Figure 9 clearly shows that all the samples taken from the reactors were clustered together; however, in a separate branch from the inoculum (~10%). These results confirm the selection of different microbial communities either due to the acidic pre-treatment to the inoculum was subjected to the environmental conditions the microorganisms were exposed to during the experimental period (260 days).
The sample from the first point of the system (R1P1a) was not grouped with other samples (less than 30% similar) nor with the inoculum, suggesting the pre-treatment impact in the microbial populations. The central and end regions of R1 (R1P2a and R1P3a) showed 50% similarity, indicating a more similar environmental condition in these two sections of the reactor. Samples from the reactor R2 were either grouped with samples from R1 and R3. The central portion of the reactor R2P2a was grouped with the central and end region of reactor R1. In a different cluster, it was grouped R2P1a and R2P3a (>40% similarity).
In reactor R3, the sample R3P1a, collected from the reactor's initial part, was grouped with samples from the reactor R2: R2P1a and R2P3a (35% similarity) (Fig. 9). As the R3 was fed with the effluent from R2, probably the substrate composition favored the selection of a similar microbial community. However, the last sampling point of the system (R3P3a) was not grouped with any other sample analyzed, showing less than 20% similarity with them. As being the last part of a system composed of 3-reactors operated in series and fed with the effluent of the precedent reactor, the lack of carbon source (99.9% of overall glycerol removal was achieved) could be the reason for such difference. It was impossible to analyze the sample R3P2a (sampling point R3 central position) due to a DNA amplification limitation.
Archaea microorganisms have high sensitivity to environmental fluctuations. It includes substrate-fed shocks occurring during the reactor's start-up phase and justifies the low similarities of 15–50% among all present study samples. Silva et al. [66] obtained similar results by analyzing the microbiota from the inoculum and the first chamber of a pilot-scale compartmented reactor used for the treatment of sanitary sewage and in reduced OLR (0.10 kg COD (m3 d)−1); it was observed a 51% of similarity.
The highest diversity Shannon-Wiener index for the Archaea Domain was observed for sampling points R2P2a for R2 (2.25) and R3P3a for R3 (2.27). Both values were higher than that observed for the inoculum (1.99). The lowest diversity index, as expected, occurred in the first sampling point of the fermentative reactor R1 (1.58) since the sludge was subjected to the acidic pre-treatment, which aimed to select endospore-forming bacteria, including the methanogens (Table 5).
Table 5
Shannon-Wiener index (H index) to Archaea domain
Inoculum
|
R1
|
R2
|
R3
|
P1
|
P2
|
P3
|
P1
|
P2
|
P3
|
P1
|
P2
|
P3
|
1.99
|
1.58
|
1.71
|
1.99
|
2.09
|
2.25
|
2.01
|
2.22
|
-
|
2.27
|