Microbial community composition of food waste before anaerobic digestion

Anaerobic digestion is widely used to process and recover value from food waste. Commercial food waste anaerobic digestion facilities seek improvements in process efficiency to enable higher throughput. There is limited information on the composition of microbial communities in food waste prior to digestion, limiting rational exploitation of the catalytic potential of microorganisms in pretreatment processes. To address this knowledge gap, bacterial and fungal communities in food waste samples from a commercial anaerobic digestion facility were characterised over 3 months. The abundance of 16S rRNA bacterial genes was approximately five orders of magnitude higher than the abundance of the fungal intergenic spacer (ITS) sequence, suggesting the numerical dominance of bacteria over fungi in food waste before anaerobic digestion. Evidence for the mass proliferation of bacteria in food waste during storage prior to anaerobic digestion is presented. The composition of the bacterial community shows variation over time, but lineages within the Lactobacillaceae family are consistently dominant. Nitrogen content and pH are correlated to community variation. These findings form a foundation for understanding the microbial ecology of food waste and provide opportunities to further improve the throughput of anaerobic digestion.

DNA extraction and PCR.Genomic DNA was extracted from 0.2 g of food waste samples (wet weight) using the QIAamp PowerFecal Pro DNA extraction kit (Qiagen).Manufacturer protocols were followed, except DNA-free PCR grade water (Sigma) was used instead of the C6 solution for final preservation.Genomic DNA concentration was measured using a Qubit 2.0 fluorometer.Polymerase chain reactions (PCR) were performed on the extracted genomic DNA.The primers used were 1048F (5′-GTGSTGC AYG GYT GTC GTCA-3′) and 1294R (5′-GCC TAC GAT CTG AAC TGA GC 3′) for bacteria, and ITS1 (5′-TCC GTA GGT GAA CCT GCG G-3′) & ITS4 (5′-TCC TCC GC TTA TTG ATA TGC-3′) for fungi.The thermal cycle setup of bacteria PCR is initial denaturation at 95 °C for 5 min, then followed by 30 cycles of 30 s of denaturation at 94 °C, 30 s of annealing at 58 °C, and 40 s of extension at 72 °C and the final extension period is 10 min at 72 °C.The fungus PCR setting begins with an initial denaturation at 94 °C for 5 min, then followed by 30 cycles of 45 s of denaturation at 94 °C, 30 s of annealing at 60 °C, and 45 s of extension at 72 °C and the final extension period is 10 min at 72 °C.Gel electrophoresis was used to identify the integrity of the PCR product with 1% agarose gel run at 90 V for 30 min.
Quantitative PCR (qPCR) was used to quantify bacteria and fungi in food waste samples with a Bio-Rad CFX qPCR machine.The primers used for qPCR are 1048F (5′-GTGSTGCA YGG YTG TCG TCA -3′) and 1194R (5′-ACG TCA TCC CCA CCT TCC -3′) for the total copy count of the 16S rRNA gene representing the bacteria community14 and ITS1 (5′-TCC GTA GGT GAA CCT GCG G-3′) and 4 (5′-TCC TCC GCT TAT TGA TAT GC-3′) for the total copy count representing the fungi community.The working solution mix contained 5 µL of SsoFast EvaGreen Supermix (BIO-RAD), 0.1 µL of each primer, 0.1 µL of BSA (20 mg/mL) and 2.7 µL of molecular water.When plating, 96 well plates with 8 µL of working solution and 2 µL of sample were used in each well.The total bacterial qPCR protocol was 3 min of denaturation at 98 °C, forty cycles of DNA segment replication at 90 °C for 20 s, and 62 °C for 50 s.The temperature decreased to 60 °C and was maintained for 10 s for fluorescent reads, following a temperature increase of 60 to 90 °C in 0.5 °C intervals for melt curve analysis.Fungi quantification qPCR protocol was 2 min denaturation at 95 °C, forty cycles of DNA segment replication at 90 °C for 20 s, 55 °C for 30 s, and 72 °C for 60 s.The temperature stays at 72 °C for 10 min 15,16 .The average efficiency of the qPCR of the 16S rRNA gene and ITS gene copies was 95% and 96%, with an average slope of (− 3.42) and (− 3.46) and an average R2 equal to 0.993 and 0.995, respectively.DNA sequencing.Illumina sequencing was used to identify species of bacteria and fungi.The sequencing platform was MiSeq v2 2 × 250 bp.The Ramaciotti Center at UNSW provided the sequencing service, including library preparation and sequencing runs.The bacteria assays were prepared with the 16S rRNA gene amplicon library and the fungi assays with the ITS2 amplicon library (UNSW, Sydney, Australia).Paired-end sequencing results were analysed using QIIME 2 2019.7 pipeline 17 .The sampling depth was set to 1600 for quality control.The Dada2 plugin 18 was used to denoise the data with trimming length determined using FastQC (https:// www.bioin forma tics.babra ham.Ac. uk/ proje cts/ fastqc) for quality assurance.Taxonomy was assigned using the q2-feature classifier 19 classify-sklearn naïve Bayes taxonomy classifier on Silva 138 release 20 for bacteria and UNITE 8.0 21 for fungi taxonomy classification.The reference sequence was modified to increase accuracy to exclude primer overhang.The reference sequence was not cut with the advice of the QIIME2 team 22 .The genomic sequence entries were then BLASTed (https:// blast.ncbi.nlm.nih.gov) to acquire species information.Sequence reads extracted from the Illumina results were aligned using MUSCLE.A maximum likelihood phylogenetic tree was generated using MEGA-7 23 .Bacterial growth modelling.Bacterial growth in the digestor feed tank was modelled using the Monod equation and key reaction kinetic parameters (Figure S1).The equations reflect the specific growth rate of bacteria ( µmax), which depends on the temperature, pH, and nutrient level of the medium.The chosen ( µmax) used in the model was conservative since the aim of the model was to determine if there was enough residence time (steady state) for bacterial growth to occur between the two sampling points.Other kinetic parameters used include the biomass yield (Yx/s) and the saturation constant (Ks).Growth kinetics were based on Lactobacillus, which represents more than 70% of the bacterial community.
Ethical approval.This article does not contain any studies with human participants performed by any of the authors.

Results
The abundance of bacteria and fungi in food waste feedstock.To investigate the variation in the microbial communities of food waste, samples were taken from the hydropulper shredder and digestor feeding tank (Fig. 1a) approximately weekly for 3 months from winter to spring.DNA was extracted and the total abundance of 16S rRNA (bacteria) and ITS gene sequences (fungi) was determined using quantitative PCR (Fig. 1b,c).Archaeal relative abundance was below 1%.The 16S rRNA gene copy abundance was 4-6 orders of magnitude higher than the abundance of ITS sequences, indicating that bacteria are numerically dominant.The abundance of fungi was similar in the hydropulper and digestor feed samples (average 4.3 ± 2.0 × 10 4 and 4.4 ± 2.0 × 10 4 copies/g, respectively).Bacterial abundance in the digestor feed (average 1.6 ± 2.4 × 10 10 copies/g) was 26-fold higher than in the hydropulper (average 6.2 ± 6.4 × 10 8 copies/g) indicating proliferation.
To estimate the growth rate and threshold of bacteria in food waste, a bacterial growth model was developed based on growth rates of Lactobacillus (> 70% relative abundance of bacteria community, Fig. 3a,b) under comparable conditions and estimates of available growth substrates (Supplement information 1).The model was used to determine whether the residence time between the hydropulper and the digestor feed tank (~ 16 h) was sufficient to account for the observed 26-fold increase in bacterial abundance.From a starting point of 6.2 × 10 8 copies/g, the model reached a 16S rRNA gene copy density of 1 × 10 10 copies/g after 16 h and plateaued at 1.3 × 10 10 copies/g after 18 h.Growth model data and qPCR data are consistent with the proliferation of bacteria in food waste between the hydropulper and the anaerobic digestor feeding tank.
Composition of the fungal community in food waste.Figure 2 shows the structure variation of the fungal community in the hydropulper and digestor feed samples.Sequencing was done on the Illumina ITS platform, and QIIME 2 pipelines were used to process reads.The most abundant fungal lineages belong to the genera Saccharomyces and Kazachstania.The digestor feeding tank samples held more Saccharomyces and the < 1% abundance fungi strains decreased in relative abundance.The composition of the fungal community in the feedstock of food waste was relatively consistent over the 3 month sampling period.The consistent community structure with the lack of proliferation of fungi between the hydropulper and digestor feed samples indicated limited fungal activity in the food waste.
Composition of the bacterial community in food waste .Figure 3a,b shows the variation of the bacterial communities in the hydropulper and digestor feed samples.Illumina 16S rRNA sequencing was used to generate raw reads, and QIIME 2 pipelines were used to process the reads.Lactobacillaceae lineages (including Lactobacillus and Lactiplantibacillus) dominated throughout the sampling period, with an average total relative abundance in the hydropulper of 64% and 78% in the digestor feed.The relative abundance of Lactobacillaceae was more variable in the hydropulper than in the digestor feed, ranging from 32 to 93%.Sequences within the Lactobacillaceae family and closest relatives from the NCBI database were used to generate a phylogenetic tree to allow species assignation (Fig. 3c).Leuconostoc and Klebsiella strains were added as outgroups.Three of the most abundant Lactobacillaceae lineages belonged to Lactobacillus amylovorus (Lin1), Lactobacillus sanfranciscensis (Lin2), and Lactiplantibacillus plantarum (Lin3).Lactobacillus amylovorus (Lin1) was the most abundant species in the hydropulper, while Lactobacillus sanfranciscensis (Lin2) and Lactiplantibacillus plantarum (Lin3) were the most abundant in digestor feed samples.Other bacterial lineages above 1% relative abundance include Klebsiella, Leuconostoc, Acetobacter, Pseudomonas, Weissella, Brachymonas, Cloacimonadaceae W5 and Unclassified Enterobacteriaceae.They were more abundant in hydropulper samples than digestor feed samples.Bacterial lineages with less than 1% relative abundance in the hydropulper had further decreased in relative abundance in the digestor feed.

Correlations between environmental parameters and community variation.
To understand the relationship between food waste physico-chemical characteristics and the composition variations of the microbial community, three characteristics of food waste were measured (Supplementary information).Figure S2a shows the water content and pH variation of the samples over 3 months.Figure S2b shows the proportions of carbon, nitrogen and sulfur in the samples.The digestor feed and hydropulper samples have similar carbon (average of 43.38% and 42.47%, respectively), nitrogen (average of 2.40% and 2.46%, respectively) and sulphur content (average of 0.18% and 0.32%, respectively).Although the sample characteristics between the two sampling points are similar in average, on a daily basis, the sample can be vastly different.Although average elemental contents were similar, the carbon content variation pattern in hydropulper were different from the digestor feed sample from the same day.Similarly, nitrogen content had disagreed abundance variation between same day samples from the two sampling locations.
Figure 4 shows the bacterial community diversity among samples.The fungal community (not shown) did not return any prominent grouping for all metadata categories.For the bacterial community, two metadata categories, pH and nitrogen element content, were identified with nestings associated with specific ranges.The bacterial community experienced structural shifts from the hydropulper to digestor feeding tank samples.The lower the pH or higher the nitrogen content, the more closely related the bacterial communities were.The pH range of 3.5-4 and the nitrogen element content of 3-3.5% showed a strong influence on the bacteria community.
The diversity within each sample regarding the three metadata criteria, type of sample, pH, and N% of feedstock, are summarised in Table 1.Hydropulper samples had higher bacterial diversity (richness) than digestor feed samples (P = 4.72E−07).When pH was higher than 5, the richness of the bacterial community increased (P = 0.039), and a higher percentage of nitrogen was associated with an increase in bacterial community evenness (P = 0.040).
Figure 5 illustrates the correlations between environmental parameters, physico-chemical characteristics, abundance species variation, and community variance for both bacterial and fungal communities.Factor scales from − 1 to 1 indicate a total negative or positive nonparametric Spearman correlation.Underlined correlation scores had significant differences (P < 0.05).
Between the hydropulper and digestor feed tank, nitrogen content was positively correlated with abundant bacterial genera (Lactobacilus, Leunocostoc, and Klebessila) and negatively correlated to pH.Furthermore, the BLAST search of Illumina entries allowed for analysing the abundance variations of Lactobacillaceae species under various conditions.The highly abundant L. plantarum (Lin3), L. sanfranciscensis (Lin2), and L. amylovorus (Lin1) were correlated with pH, nitrogen and carbon content with varied dependency.In the fungal community (Fig. 5c,d), the sulphur and nitrogen element content impacted the hydropulper community, where fungal strains are negatively connected.The pH was not clearly correlated with the relative abundance of fungal strains or the number of ITS gene copies.Table 1.Alpha (within-sample) diversity of bacterial and fungal communities, including Margalef richness (a count of different species in the sample, higher means more species) and Simpson evenness (a measure of relative abundance of different species in the sample, higher means less even spread of population among species) indices.

Discussion
Methane yield during anaerobic digestion correlates with the relative abundance of specific bacterial and archaeal lineages in the incumbent microbial community 28 .The composition of microbial communities is determined by the growth and death rates of community members and immigration from outside the system harbouring the community 29 .In the context of anaerobic digestion of food waste, immigration is logically driven by the indigenous community of the foodwaste feedstock.The microbiota associated with feedstock is additionally relevant with respect to chemical transformations it catalyses prior to feedstock entering a digester (biological pretreatment).Despite its relevance to anaerobic digestion of food waste, relatively little is known about the composition of microbial communities in food waste.One study described the microbial community in food from a canteen from the moment it was discarded and over the ensuing 72 h 13 .Another described changes in community composition of food waste in response to aeration 30 .In addition, a study has explained the variation of food waste composition in response to storage environment 31 .None have described how the community composition of food waste feedstock received by a digestion facility varies over time or within different units prior to entering digesters.The food waste that undergoes anaerobic digestion is mainly composed of bacteria, especially Lactobacillaceae.However, the bacterial community is not stable and varies every day along with the pH and nitrogen levels of the waste.Ammonia, as a form of nitrogen, can inhibit the growth of various species of bacteria and pose selective pressure and shift the microbial community in food waste 32 .This knowledge underpins future attempts to rationally engineer the foodwaste community for improved downstream resource recovery.
In the current study of an industrial anaerobic digestion facility handling food waste, bacteria and fungi were quantified in food waste samples from the hydropulper unit and from the downstream storage unit from which food waste is transferred into the digester.pH ranged from 3 to 5.5 trending downwards throughout the sampling campaign.Water content sat between 80 and 90%.The pH shift of the food waste could result either from changes in the food waste composition as lower pH is associated with carbohydrate-rich food waste and higher pH associated with protein-rich food waste or the storage temperature that affects the production of acidic fermentation products 31 .
Quantification of fungal and bacterial sequences revealed that bacteria are numerically dominant in food waste by several orders of magnitude.This suggests that fungi play a limited role in foodwaste decay or depolymerisation prior to anaerobic digestion.The closely related yeasts Saccharomyces and Khazachstania were the dominant fungal lineages.Khazachstania has been observed previously in food waste and is common in food fermentation communities 33 Saccharomyces are commonly found in food fermentation waste such as the washdown of brewery and bakery factories 34,35 .The low abundance of moulds may be due to the homogenisation of food waste in handling, transport and ultimately pulping, and the limited availability of oxygen 36 , but the low abundance of fungi generally was suprising.Fungi can be tolerant of low pH and oxygen limitation, so it is possible that the fungi in foodwaste are suppressed through production of antifungal agents by bacteria.The bacteria Lactobacillus sanfranciscensis (Lin2) and Lactiplantibacillus plantarum (Lin3), observed to dominate the bacterial community in this study, are known to produce antifungal substances 37 .Regardless of the driver of low fungal relative abundance, it appears that the biodegradative abilities of fungi are not being exploited in foodwaste.
The bacterial community was dominated by lineages within the Lactobacillaceae Family (principally Lactobacillus species).Lactobacillus species are well known for their ability to ferment sugars under oxygen limitation and for the associated release of volatile fatty acids that results in a drop of pH 38 .This has long been exploited in the preservation of foods because lowering the pH below 5 creates conditions unfavourable to most microbes 39,40 .In essence, Lactobacillus species exclude competitors in food waste through pH manipulation.This limits the biodegradation ability of the microbial community in foodwaste to degradative enzymatic activity generated by Lactobacillus.Regarding extracellular enzyme activity, abundant L. amylovorus, L. sanfranciscensis, and L. plantarum lineages, are known to produce amylase [41][42][43][44] .L. amylovorus and L. plantarum are also known to produce lipase and protease [45][46][47] though proteolytic activity is moderate 48,49 .L. plantarum also produces extracellular feruloyl esterase 50 , while L. amylovorus has only been reported to produce intracellular esterase 51 .Therefore, the three most abundant bacteria observed can produce amylase, lipase, protease, and feruloyl esterase, but they are not known to produce cellulase.They probably play a significant role in starch degradation, but have a limited ability to hydrolyse lignocellulosic biomass.There may be potential in bioaumenting cellulase producing bacteria to increase downstream digestion efficiency.
Proteobacteria were notable in their absence in foodwaste 13,40,52 .Proteobacteria are capable of degrading various complex substances and are prevalent in anaerobic digestion systems 53,54 .This supports the potential for bioaugmenting foodwaste with Proteobacteria or altering conditions for the growth of Proteobacteria to exploit their degradation capabilities including cellulase activity 55 .
Aerobic pretreatment of organic substrates can improve anaerobic digestion [56][57][58] .This is believed to be a consequence of reducing the concentration of easily digestible substrates that can result in rapid decreases in pH due to VFA production under anaerobic conditions, accelerated oxidation of VFAs and the depolymerisation of relatively recalcitrant biopolymers resulting in more extensive digestion and reduced biosolids 59,60 .Aerobic pre-treatment can increase digestion stability and efficiency [61][62][63] .These benefits derived from aeration logically hinge on the composition of microbial communities indigenous to food waste and the catalytic abilities encoded therein 30 .
This study showed the variability of microbial community structure in food waste.The samples showed dominance and growth of the bacterial community but not the fungal community.Lactobacillaceae were dominant, and their activity was mainly influenced by pH and nitrogen content.The results inform the potential for adjusting the community via bioaugmentation or supplying air to achieve higher downstream digestion efficiency.

Figure 1 .
Figure 1.(a) Basic process flow chart of the food waste digestion facility.The sampling points are shaded in grey at the hydropulper and digestor feed tank.Liquid food waste includes mainly grease traps and spadable liquid waste.The abundance of (b) bacteria and (c) fungi per gram of sample (wet weight) taken from the hydropulper (triangles) and digestor feed (squares) was determined by quantitative PCR targeting 16S rRNA gene copies for bacteria and ITS copies for fungi.Gene copy counts represent cell concentration measurements, thus are relative to the wet weight of the sample, and are the average number of technical triplicates. https://doi.org/10.1038/s41598-023-39991-wwww.nature.com/scientificreports/

Figure 4 .
Figure 4. Weighted PCoA graphs show the relationship between environmental parameters and bacterial community composition.Hydropulper (squares) and digestor feed (stars) communities are distinct (a).pH (b) and nitrogen (c) content showed specific grouping at pH 3.5-4 and nitrogen content 3-3.5% (indicated by black color).

Figure 5 .
Figure 5.The bacterial heat map indicates the relationships among the environmental parameters (pH, nitrogen content, the variance in relative abundance, and the number of bacterial 16S rRNA gene copies in the sample separated into hydropulper (a) and digestor feed (b) samples.The negative (dark green) heat map cell indicates a negative nonparametric Spearman correlation and vice versa.The fungal heat map indicates the relationships among environmental parameters, the variance in strain relative abundance, and the number of fungal ITS gene copies in the sample separated into hydropulper (c) and digestor feed (d) samples.The negative (dark green) heat map cell indicates a negative nonparametric Spearman correlation and vice versa.The underlined entries have a P value less than 0.05 from sigificance test.
27cteria and fungi samples were presented in weighted Unifrac PcoA graphs27ategorized by sample type, pH, and nitrogen element content.The bacteriological taxonomy data were filtered to exclude chloroplast sequences.The relative abundance of OTUs is considered the reference abundance of microorganisms in the samples.Due to countless substrains that are low in abundance (< 1%) or only present in a few samples, the top four bacteria species (making up > 40% of community) were kept for further correlation calculations.Similarly, the analysis of the community of fungi included the four most abundant species.GraphPad Prism 9 was used to analyse the correlation between environmental parameters and variances in community structure.Correlations are examined using nonparametric Spearman correlations to cover more than linear relations and to generate heat maps.The P values were calculated with the student t-test in Prism 9. Vol:.(1234567890)