Multistability and Reversibility of Aerobic Granular Sludge Microbial Communities Upon Changes From Simple to Complex Synthetic Wastewater and Back

Abstract

Aerobic granular sludge (AGS) is a promising alternative wastewater treatment to the conventional activated sludge system allowing space and energy saving. Basic understanding of AGS has mainly been obtained using simple wastewater containing acetate and propionate as carbon source. Yet, the aspect and performances of AGS grown in such model systems are different from those obtained in reactor treating real wastewater. The impact of fermentable and hydrolyzable compounds on already formed AGS was assessed separately by changing the composition of the influent from simple wastewater containing volatile fatty acids to complex monomeric wastewater containing amino acids and glucose, and then to complex polymeric wastewater containing also starch and peptone. The reversibility of the observed changes was assessed by changing the composition of the wastewater from complex monomeric back to simple. The introduction of fermentable compounds in the influent left the settling properties and nutrient removal performance unchanged, but had a significant impact on the bacterial community. The proportion of Gammaproteobacteria diminished to the benefit of Actinobacteria and the Saccharibateria phylum. On the other hand, the introduction of polymeric compounds altered the settling properties and denitrification efficiency, but induced smaller changes in the bacterial community. The changes induced by the wastewater transition were only partly reversed. Seven distinct stables states of the bacterial community were detected during the 921 days of experiment, four of them observed with the complex monomeric wastewater. The transitions between these states were not only caused by wastewater changes but also by operation failures and other incidences. However, the nutrient removal performance and settling properties of the AGS were globally maintained due to the functional redundancy of its bacterial community.

Keywords: Candidatus Accumulibacter; PAO; Tetrasphaera; bacterial communities; functional redundancy; multistability; polymeric substrates.

Figure 1

Summarizing diagram of the reactor operation and the two experiments. Experiment 1 started in RB with AGS acclimated to a simple wastewater [containing mainly volatile fatty acids (VFA)]. The wastewater feeding the AGS was progressively replaced by a complex monomeric wastewater (containing VFA, glucose, and amino acids). On day 333, half of the sludge of RB was used to start reactor RA. The two reactors were fed with complex monomeric wastewater. On day 600, RB was stopped and emptied. On day 607, half of the sludge from RA was used to re-start RB. In RB, the complex monomeric wastewater was progressively replaced by a complex polymeric wastewater (containing VFA, glucose, amino acids, starch, and peptone). In RA, the complex monomeric wastewater was progressively replace by a simple wastewater.

Figure 2

${\text{PO}}_4^{-}$-, NH₄-, and TN-removal performance of the AGS during experiment 1. The process events susceptible to having had a negative impact on nutrient removal performance are numbered from 1 to 12. (1) No C-medium in the influent wastewater during eight cycles. (2) The leak of O₂ detected since several days is corrected. (3) Few biomass in the reactors due to the duplication of the reactors and the feeding was not adapted. (4) Problem with pumping of NP-medium. (5) The pH of the bulk water was higher than 11 during at least 2 h. (6) The pH of the bulk water was higher than 11 during a period of time estimated to 24 h. (7) Problems with the recirculation pump possibly leading to lower shear stress during 2 days. (8) No trace elements in the influent during 4 days. (9) The influent was composed of twice the NP-medium and 200 mg O₂ ml⁻¹ of glucose during 2 days. (10) No feeding during the night. (11) Loss of biomass due to accidental change from 5 to 3 min of settling. (12) Lower concentrations of C-medium in the influent due to malfunctioning of the pumping.

Figure 3

${\text{PO}}_4^{-}$-, NH₄-, and TN-removal performances during the transition back from complex monomeric to simple monomeric influent wastewater. The vertical bar with the number (13) corresponds to the day with no feeding during several cycles.

Figure 4

The seven bacterial stable states detected during experiment 1 and 2.

Figure 5

Principal component analysis based on the Bray-Curtis distance of the bacterial communities (genus level) at stable states. The two first coordinates PCo1 and PCo2 explain 35.47 and 22.05% of the variance, respectively. The plots of the three first coordinates are provided in Supplementary Figure 4.

Figure 6

Evolution of the bacterial community composition during the transition from simple to complex monomeric wastewater and to complex polymeric wastewater. The genera are colored according to the class they belong to, with the exception of the Betaproteobacteriales, previously the class of Betaproteobacteria, colored in red. They were recently merged with the Gammaproteobacteria (Parks et al., 2018), here colored in light green. Only the abundant bacteria (mean > 1% with at least one wastewater type) are shown.

Figure 7

Evolution of the bacterial communities during the transition back from complex monomeric to simple wastewater. The genus are colored according to the class they belong to, with the exception of the Betaproteobacteriales, colored in red, whereas the other Gammaproteobacteria are colored in light green. Only the abundant bacteria (mean > 1% with at least one wastewater type) are shown.

Figure 8

Granules and flocs heatmap showing the average relative abundance of the abundant taxa in AGS fed with complex monomeric wastewater during stable states 2, 4, and 5. The flocs and granules data of stable state 3 were too sparse for this analysis. Green indicates a higher proportion in granules, purple a higher proportion in flocs. In order to lower the possible effect of noise in the very low abundant taxa, a pseudo-count of 0.5% was added to each abundance for the color code and the significance analysis. A yellow asterisk indicates the differences that are significant with an adjusted p-value of 2.94 × 10⁻⁴ (0.01/34).