On Establishing Microbial Community-Process Function Relationships in Anaerobic Digestion

Abstract Municipal sludge anaerobic digestion is an engineered biological treatment process wherein a mixture of diverse microorganisms mediates the biochemical transformation of wastewater sludge organics to methane. To anaerobically digest organics more efficiently to methane, the process-enabling microbial communities need to be further investigated and related to process function. The current industry standard for anaerobic digestion process modeling relies on anaerobic digestion model 1 (ADM1). However, ADM1 typically uses default kinetic coefficients to model bioconversion rates considering only an average microbial community. But the average microbial community in municipal anaerobic digesters may be a poor representation of the complex and diverse microbial communities known to dwell among various anaerobic reactors. Consequently, current models can be inaccurate, and this can negatively impact implementation and process efficiency. In this study, we build upon our previous work to relate process function and microbial community information in anaerobic digestion. The central goal of this work is to better model methane production from municipal sludge by leveraging quantitative microbial community data to select the most appropriate Monod kinetic parameters based on the identity and abilities of the specific microbes present in a given digester. Anticipated outcomes will be broadly applicable and of particular interest to the municipal anaerobic digestion and waste management industry, and the methods developed can be potentially applicable to other biological treatment, such as nitrification or anammox processes. To achieve this goal, 20 full-scale operating mesophilic anaerobic digesters were sampled. Source digesters were selected to represent regionally significant sectors and a diversity of microbial community profiles, a vital factor when seeking to relate process function and microbial information. Community activity was measured via specific methanogenic activity (SMA) assays. Prior to SMA setup all bulk digester biomass was normalized to 7 g/L VSS, macro and micronutrients were supplemented, alkalinity was standardized to 6 g/L sodium bicarbonate, titanium citrate reducing agent added and allowed to completely degas at 35 °C with continuous impeller mixing. To determine SMA and kinetic parameters, each bulk digester sample was partitioned into 24 different reactors, each 250 mL working volume and fed 8 model substrates in triplicate representing critical intermediates in the biochemical transformation pathway from complex organics to methane; a custom, automated respirometer system recorded high resolution methane production data via biogas scrubbing for liquid and solid substrates and headspace pressure measurement was used for hydrogen gas substrate (Figure 1). Hydrogen SMAs were evaluated differently due to concerns regarding gas permeability and substrate to microbe ratio, so 160 mL working volume serum bottle reactors were used with 26.5 mL biomass volume. DNA was extracted from digester set samples including 3 samples from endogenous decay control reactors and 3 reference samples from initial digester inocula, and repeated for all 20 digesters surveyed, yielding 600 microbial community samples for analysis. Absolute quantitative microbial community data were generated via a microbial spike-in method, wherein a known quantity of two commercially available microbes (ZymoBIOMICSTM Spike-in Control I D6320) were added to digester samples prior to DNA extraction, and subsequent 16S rRNA gene amplicons were sequenced. Artificial neural network (ANN) modeling will be used to correlate significant microbial consortia from microbial communities fed model substrates to kinetic parameters and evaluated by a leave-one-out cross validation. The resulting model will predict kinetic parameters relevant for ADM1 based upon the microbes present in a digester sample. This study is currently in progress with SMA values, Monod kinetic parameters, and microbial community analysis partially completed. All laboratory evaluations to generate SMA datasets and DNA extraction and sequencing are completed. Preliminary results from SMA evaluations for 6 digester sets showed satisfactory variation among substrates, with median values as low as 1.2 and up to 12.3 mL CH4/g VSS-h for cellulose (CEL) and glucose (GLU) & hydrogen (H2), respectively (Figure 2). Within substrate groups, less variability is observed among SMA values from different digester sources with the exception of hydrogen, varying between 11 and 34 mL CH4/g VSS-h in the first and third quartiles (Figure 2). Among the 6 digester sources presented, two were municipal with both receiving primary solids and WAS and one also co-digesting dairy processing waste from cream cheese. Notably, the municipal digester treating only biosolids had among the lowest SMAs for all substrates, whereas the municipal co-digester was more similar to industrial digesters and was significantly higher for butyrate SMA among all 6 digesters. Additionally, municipal digester SMAs for cellulose were significantly higher than industrial (p-value 0.00002), 3.3 ± 0.7 and 1.9 ± 0.03 mL CH4/g VSS-h, respectively. These comprehensive SMA profiles using 8 model substrates can serve as a holistic method to evaluate and compare activity between digesters. Figure not included due to space constraints. Variability in terms of SMAs within substrates groups or in other terms, between digester sets is vital for being able to detect microbial consortia involved with specific substrate conversion to methane and ultimately to quantitatively correlate function (Monod kinetic coefficients) to microbial community. Initial results of microbial community analysis revealed little change in microbial communities following substrate additions, but interesting clustering has emerged to reveal clear patterns. The detection of microbes found in a control group containing a gut microbiome standard (ZymoBIOMICSTM) allowed for sensitivity analysis. Findings from a subset of data show that our microbial community DNA sequencing and bioinformatics method can detect down to 0.097 ± 0.017 % relative abundance, based on 0.066 % of Methanobrevibacter 16S rRNA sequences present in gut microbiome standard. Additionally, 99.99 % of DNA sequences from the gut microbiome standard were detected, representing 12 of 15 microbial genera with 0.01 % found in the remaining 3 microbes undetected. Comparing the same 6 digester sets represented in Figure 2, a 2-dimensional principal coordinate analysis showing 47 % of total variability based on dissimilarity, shows 6 distinct clusters (Figure 3). Each point in the figure represents a single microbial community from a reactor receiving a model substrate. The relative distance between points indicates the degree of dissimilarity, so points closer together show less dissimilarity and point further apart more dissimilar. Colors denote digester source and native feedstock shown in the legend. All clusters uniquely represent a digester source apart from two municipal digesters in yellow and purple, while the shapes denote substrate fed during SMAs. These municipal digesters were significantly different in terms of their SMA activity profile, the more active of the two co-digests dairy processing waste; however, their tight clustering in PCoA analysis still shows that relative to four industrial digesters their microbial communities as a whole, despite different substrates and activity were nearly identical. Differences between industrial and municipal digester microbial communities will help to inform Work will continue to further investigate these microbial communities and will ultimately help to unravel their functional connections. Holistic SMA profiles using 8 model substrates and quantitative absolute abundance methodologies using commercially available internal standard spike-in controls are effective tools applicable to all anaerobic digestion studies, but may be especially relevant for municipal biosolids digestion and co-digestion where differences in SMA and microbial communities can be more challenging to differentiate. We anticipate that predicted Monod kinetic parameters are grounded in digester microbial community data and can be an input to ADM1 for improved process modeling.