Materials

Vegetable residues, mainly potato peels, were collected from restaurants around the university campus. Cow manure was obtained from a farm and used as a nitrogen source. The residues and manure were separately ground using a blender and subsampled using the quartering method (^{Jolivet et al., 2023}). Approximately 15 grams of vegetable residues were mixed with 1 gram of cow manure and used in the anaerobic digestion.

Experimental setup and cultivation conditions

All experiments were performed under anaerobic conditions. Seed culture was prepared in one-liter screw-capped bottles (Schott, Duran) containing 900 ml of 1 %(w/v) organic residues (i.e., the mixture of vegetable residues and cow manure) dissolved in sterile distilled water and incubated at 26 (C for 20 days in an incubator (Innova 42, New Brunswick Scientific). The caps in the bottles included two outlets, one for allowing gas dissipation and the other for collecting samples or adding an 8 %(w/v) sodium hydroxide and 2 %(w/v) sodium bicarbonate solution to raise the pH of the medium above 5 when a decrease in this value was detected. Ten milliliters from the liquid medium of the reactor were taken and discarded before the sample was obtained to eliminate residuals in the tubing. The pH of the samples was determined by using a FiveEasy pH meter (Mettler Toledo). Subsequently, 100 ml of the seed culture were added to bottles containing 800 ml of 1 %(w/v) of the same substrate used in the seed culture. Duplicate experiments were performed at controlled temperatures, first at 35 (C for 17 hours and continued at 55 (C until 26 hours of cultivation. An experiment involving cultivation at ambient temperature (10 - 26 (C) was used as the reference.

Chemical characterization of compounds produced during the anaerobic digestion

The concentrations of acetic and propionic acids in the anaerobic digestion medium were analyzed using high-performance liquid chromatography (HPLC) (Thermo Scientific, UltiMate 3000). Before quantifying the acids, 10 ml samples were centrifuged at 10000 g for 7 minutes using a Sorvall ST 40R centrifuge (Thermo Scientific). The supernatant was filtered through carbon-activated impregnated cotton placed in a syringe. One milliliter of each sample was acidified with 1 μl concentrated (85 %) H₃PO₄. The compounds were separated on a C18 Ultra AQ RESTEK column, which is a silica reversed-phase column. The temperature of the column was maintained at 35 °C in the oven. A solution containing 10 ml/L acetonitrile and 340 μL/L H₃PO₄ dissolved in deionized water was used as the mobile phase with a 0.8 ml/min flow rate. A UV diode array detector was used to quantify the acids' concentrations at 210 nm and 35 °C. All samples were analyzed in triplicate.

For the quantification of metals, samples of the medium were digested following the APHA standard method 3030 G: Nitric and sulfuric acid digestion (^{Lipps et al., 2018}). A combined flame and furnace atomic absorption spectrometer (AA500FG model, PG instruments) equipped with deuterium lamp background correction, hollow cathode lamps, and an air-acetylene burner was used to quantify iron, zinc, cadmium, nickel, cobalt, selenium, copper, magnesium, potassium, lead, and manganese. The instrumental parameters used were those recommended by the manufacturer. Methane, carbon dioxide, and hydrogen sulfide were quantified at the outlet of the flasks containing the medium using a portable multi-gas detector (SAW4, Shi'An).

DNA sequencing and metagenome construction

Duplicate liquid samples from the anaerobic digestion were vacuum filtered through a 0.45 µm sterile filter (S-Pak, Merk) until the filter was clogged for the experiments performed at 35 and 55 (C. Similarly, a single sample was obtained from the seed culture, and the anaerobic digestion performed at ambient temperature. These filters were used to isolate DNA using the DNeasy Power Water kit (Qiagen), following the instructions indicated by the supplier.

DNA concentrations obtained from the samples were measured using a Qubit 2.0 fluorometer (ThermoFisher Scientific, USA). The DNA samples were sent to Omega Bioservices (GA, USA) for sequencing. The V3-V4 region of the bacterial 16S rRNA gene sequences was amplified using the primer pair containing the gene‐specific sequences and Illumina adapter overhang nucleotide sequences. The complete length primer sequences were 16S rRNA amplicon PCR forward primer (5'-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGCCTACGGGNGGCWGCAG-3´) and reverse primer (5'-GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGACTACHVGGGTATCTAATCC-3´). The libraries were normalized with Mag-Bind® EquiPure Library Normalization Kit (Omega Bio-tek, Norcross, GA) and then pooled. The pooled library was checked using an Agilent 2200 TapeStation and sequenced (2 x 300 bp paired-end read setting) on the MiSeq (Illumina, San Diego, CA).

All sequencing reads are available at the Sequence Read Archive (SRA) in the National Center for Biotechnology Information (NCBI) under the project name PRJNA1092678.

Metagenomic data analysis

Individual per-sample fastq files were quality-trimmed using the DADA2 pipeline v1.26.0 (^{Callahan et al., 2016}). The software was further used for quality control and error rate learning. Inference of amplicon sequence variants (ASVs) was performed separately for the forward and reverse sequencing runs to account for run-specific error profiles. The reads were merged into a single-sequence table. Chimeras were removed from the merged sequence table, and taxonomy was assigned using the DADA2 implementation of the naive Bayesian classifier and the Silva reference database v138.1. Phylum- and genera-level assignments were performed using the DADA2 pipeline, allowing multiple matches per ASV. Relative abundances were obtained by normalizing the ASV count data that, for this purpose, were divided by the total number of counts in the sample. Bar plots were constructed using the graphics v4.2.2 package (Murrell, 2006).

Statistical analysis

All statistical analyses were performed using R v4.2.2 (^{RCoreTeam, 2021}). The Shannon index for alpha diversity of the microbial phyla was estimated in Vegan v2.6-4 using the Shannon-Weaver method based on the counts for the ASVs (^{Oksanen et al., 2022}). To check if the Shannon index was significantly different (p < 0.05) between the sample groups, we performed the Wilcoxon rank-sum test using Stats v4.2.2. Relative abundances of identified microbial phyla with statistical difference p < 0.05 were obtained with edgeR v.3.41.9 (^{Chen et al., 2016}; ^{McCarthy et al., 2012}) after comparing the abundances found in the samples. The beta diversity was evaluated using a hierarchical clustering based on the pairwise distances from the Bray-Curtis distance of the normalized ASV counts using Vegan v2.6-4. Heatmaps were constructed using the package pheatmap v1.0.12 (^{Kolde, 2019}). Furthermore, we used the Bray-Curtis distance of the normalized ASV counts for the metaMDS function of the Vegan package to run nonmetric multidimensional scaling (NMDS) analysis and ANOSIM to test whether there was a significant difference between microbial communities, both identified and non-identified, in the samples and regarding their relationship with cultivation time and temperature. One-way ANOVA was used to determine the significant differences between the concentrations of the chemicals in the medium. However, when Levene’s test showed that the standard deviation of the data was not equal, Welch's method was employed using the t1way function in the WRS2 v.1.1-4 package.

Identification of bacterial groups during the anaerobic digestion of organic residues

Organic residues, including restaurant vegetable disposals and cow manure, were used as the substrates for anaerobic digestion. The microorganisms in the substrates were first adapted to anaerobic conditions in the seed culture. After 10 and 20 hours of cultivation, the phyla identified were Firmicutes, Bacteroidota, and Proteobacteria. Among the bacterial genera within these phyla, Veillonella and Megasphaera had high relative abundances at 10 hours of cultivation, although Dialister and Dysgonomonas had higher abundances after 20 hours. These genera include anaerobic and microaerophilic species (^{Joyce et al., 2018}; ^{Lim et al., 2020}; ^{Sayara & Sánchez, 2019}; ^{Stoyancheva et al., 2023}; ^{Zhou et al., 2021}). The phyla Bacteroidota and Firmicutes are frequently found in human and animal intestines and related biomes (^{Casals-Pascual et al., 2018}; ^{Santiago-Rodriguez et al., 2020}). Moreover, Firmicutes and Bacteroidota are the phyla containing most hydrolytic bacteria found in bioreactors where AD proceeds (Lim et al., 2020). These phyla, along with Chloroflexi, Proteobacteria, and Atribacteria, contain many acidogenic species reported in anaerobic reactors (Lim et al., 2020). Indeed, Veillonella, Megasphaera, Dialister, and Dysgonomonas form part of the bacterial consortium found during the acidogenesis of different organic residues (Joyce et al., 2018; Sayara & Sánchez, 2019; Stoyancheva et al., 2023; Zhou et al., 2021).

The experiments were continued by adding 100 ml of the seed culture to bottles containing 800 ml of the substrate used in the seed culture. An experiment performed at ambient temperature was used as the reference. The anaerobic digestion started at 35 (C, and Bacteroidota, Firmicutes, and Proteobacteria were the phyla with the highest relative abundance (Figure 1). The anaerobic fermentation of food waste from the hospitality sector and cow manure showed that Firmicutes was the most representative phylum in the microbiomes, followed by Bacteroidota (^{Rahman et al., 2021}; ^{Rasi et al., 2022}). The predominance of Bacteroidota in our experiments was most likely related to the organic residues utilized. After 9 hours of cultivation, the abundance of the phylum Bacteroidota decreased (p < 0.05), and the abundance of Actinobacteriota significantly (p < 0.05) increased. Additionally, the abundance of the phylum Bacteroidota at 9 hours was much lower (p < 0.05) than that at 17 hours of cultivation (Figure 1). However, at 55 (C after 17 hours, the abundances of all phyla were similar (p > 0.05) to those found at the beginning of the digestion.

Figure 1 Identified bacterial diversity during the anaerobic digestion of organic residues at different temperatures and cultivation times. The figure shows the relative abundance of the identified bacterial phyla. Results are based on data analysis from duplicate metagenome sequences developed at 35 and 55 (C. The experiment performed at an ambient temperature was used as the reference. DADA2 v1.26 was used for bioinformatic analysis. 

Previously, anaerobic digestion of cow manure showed a decreased relative abundance of Bacteroidota and Actinobacteria (^{Rahman et al., 2021}). Moreover, the growth of Firmicutes was favored in the reference experiment in samples obtained at 9 and 17 hours. Several genera comprising anaerobic and microaerophilic bacteria were identified in these cultures. The abundance of these genera changed during the experiments, although the presence of Erysipelatoclostridium was notable at ambient temperature and 9 and 17 hours. Species of Erysipelatoclostridium have been found in human and chicken guts before (^{Gilroy et al., 2021}; ^{Milosavljevic et al., 2021}) and may be related to the cow dung used in our cultivations.

Evaluation of the whole bacterial community modification in the course of anaerobic digestion

Although many amplicon sequences were detected, not all of them were identified. However, the abundance of all sequences and their relationships with cultivation parameters can be analyzed. Bacterial community richness was evaluated using the Shannon diversity (Figure 2A), which revealed no significant differences (p > 0.05) between the communities detected at 35 and 55 (C. Additionally, the microbial community dynamics (i.e., beta diversity) showed that the microorganisms did not form separate clusters for each of these temperatures at different cultivation times, albeit the communities shared some similarities among the samples obtained between 9 and 26 hours at ambient temperatures (Supplementary Figure S3). A report on the anaerobic digestion of cow manure described significant differences in alpha- and beta-diversity in a bioreactor with non-temperature control and at various cultivation times (^{Rahman et al., 2021}), whereas community richness and dynamics reduced significantly over long operation times (after 54 days) of the anaerobic co-digestion of microalgae with potato processing waste and glycerol (^{Zhang et al., 2020}). To further investigate the differences in microbial communities at different temperatures and cultivation times, we used nonmetric multidimensional scaling (NMDS) analysis, resulting in a statistical stress of 0.118. Figure 2B also depicts no significant distances for the ordination of the combined relationship of the microbial distribution at the temperatures tested and cultivation times (ANOSIM statistic R = 0.3522, significance = 0.1122). Consequently, the cultivation time and type of substrate used in anaerobic digestion are shown to determine the variations observed in the microbial communities, which remained approximately the same in our experiments.

Figure 2 Analysis of the anaerobic digestion microbiome performed at different temperatures and cultivation times. The figures show A) the Shannon index for alpha-diversity and B) the NMDS analysis of the bacterial population related to the temperatures and cultivation times during the anaerobic digestion. DADA2 v1.26 and Vegan v2.6-4 were used for bioinformatic and statistical analysis, respectively. 

Organic acids and metals concentrations amid the anaerobic digestion

Acetic and propionic acids were quantified during the seed culture and anaerobic digestion of the organic residues (Figure 3). Acetic acid was the most prevalent compound observed in our cultivations. It increased from 3.5 g/L to 8.2 g/L in the seed culture. Acetic acid concentration notoriously increased (p < 0.05) from those determined at 35 (C to those at 55 (C and reached a high concentration of 11.8 g/L at 55 (C and 17 hours in the anaerobic digestion. Furthermore, the propionic acid concentration gradually increased in the seed culture and reached a maximum of 3.2 g/L at 55 (C and 9 hours of cultivation. At ambient temperature, the maximum concentrations of acetic and propionic acids were 6.7 g/L and 1.8 g/L, respectively (Supplementary Figure S4). On the other hand, the pH of the culture medium was maintained over 6 in the seed culture and was allowed to decrease to 5.2 during the digestion to favor the production of organic acids and restrict the formation of methane (^{Zhou et al., 2021}). In fact, the methane concentration was lower than 20 % (v/v) in our cultivations, whereas the concentrations of CO₂ and H₂S steadily increased (Supplementary Figure S5). The anaerobic digestion of organic residues yields more significant amounts of acetic acid than propionic acid at pH 5 and 6 (Zhou et al., 2021). Commonly, these acids are anaerobically produced with others, such as butyric and hexanoic acids, which are appropriate for producing polyhydroxyalkanoates (^{Lee et al., 2014}).

Figure 3 Evolution of pH and acetic and propionic acids concentration during the development of the seed culture and the anaerobic digestion of organic residues. Error bars show the standard deviations of the average values of two independent experiments. 

Metals concentrations in the medium were also determined during the anaerobic digestion (Table 1). A significant increase in the concentration of iron was observed between 35 and 55 (C (p < 0.05), although the metal concentration also increased in the experiment performed at ambient temperature. These results suggest that the modification in the metal concentration was due to microbial digestion of the organic matter rather than the cultivation conditions. All other metals had approximately the same concentrations during the cultivation. Analysis of previous reports revealed that low concentrations of Cu²⁺ (0-100 mg/L), Fe²⁺ (50-4000 mg/L), Ni²⁺ (0.8-50 mg/L), Cd²⁺ (0.1-0.3 mg/L), and Zn²⁺ (0-5 mg/kg) promote biogas production (^{Guo et al., 2019}). Therefore, the concentrations of the metals in our experiments were not inhibitory to the anaerobic process. In addition, releasing iron into the medium can benefit anaerobic digestion because it detoxifies sulfide inhibition in the microbial community (Guo et al., 2019).

Table 1 Metals concentrations during the anaerobic digestion of organic residues. Two independent experiments were performed at 35 and 55 (C; the average values of the concentrations are followed by their standard deviations. An experiment was performed at ambient temperature as a reference.