Introduction

A large amount of organic waste including diverse kinds of agricultural residues, municipal solid wastes, and animal manures are generated in China every year, possibly more than 5 billion tonnes (wet weight) (Jia et al. 2018; Wang et al. 2020; Wei et al. 2000). Improvement in organic wastes disposal in the years to come is important for the governments in China (Yu et al. 2019; Zhang et al. 2010). Many alternatives for the better treatment of these organic wastes have been compared so far. Composting is an increased accepted method for sustainable agricultural ecosystems to effectively manage solid organic wastes, in which nutrients are biologically stabilized and unstable organic matter is controlled biologically decomposed (Gao et al. 2019; Martínez-Gallardo et al. 2020). Therefore, the performance of composting directly depends on microbial activity (Moreno et al. 2021; Wei et al. 2018). Understanding and regulating the quantity, activity, and biodiversity of indigenous microbial community during different stages are important for high composting efficiency and good quality composts.

Microbial community in composting is mainly influenced by the composition of raw materials to be composted and process control parameters (Wang et al. 2015). For example, physicochemical factors including pH, moisture, electrical conductivity, particle size, C/N, etc. and related operational parameters including forced aeration rate, turning frequency, outdoors or closed composting reactors, etc., both affect the activities of microbes and further have a great effect on final compost quality (Jurado et al. 2015). In fact, various microbial communities also have the ability to influence physicochemical characteristics of the niches they colonize based on their metabolic versatility, such as secreting enzymes or providing precursors of humic substance in the biotransformation process of organics in composting (Moreno et al. 2021; Zhan et al. 2021). Therefore, diverse microbes have a complex role assignment in the typical composting stages, e.g., mineralization of carbohydrates in mesophilic, decomposition of lignocellulose in thermophilic, and microbe-dominated humification in maturation (López et al. 2021; Palaniveloo et al., 2020). Numerous researchers have employed culture-dependent and culture-independent approaches, such as, biochemical identification, polymerase chain reaction denaturing gradient gel electrophoresis (PCR-DGGE), terminal restriction fragment length polymorphism and single strand-conformation polymorphism, and microarray analysis and next-generation sequencing, to characterize microbial community dynamics and ensure the functional taxa in composting (Palaniveloo et al., 2020). Wang et al. (2015) found Bacillus promoted thermophilic process of municipal solid waste composting but a lot of key uncultured bacteria were revealed by PCR-DGGE method. Lopez-Gonzalez et al. (2015) suggested more than 4000 bacterial strains from pure culturing isolation were thermotolerant with high biodiversity in composting from plant residues but still had 202 unidentified strains. Nowadays, high-throughput sequencing technology combined with bioinformatics tools is increasingly used to detect more microorganisms compared to other ways (Meng et al. 2019; Zhong et al., 2020). Metagenomics and metatranscriptomics approaches further provide detailed functional and metabolic information of microbiota during composting (Antunes et al., 2016; Braga et al., 2021). However, there is still limited comprehensive understanding of bacterial transient or general roles in composting from diverse sources.

Commensal relationships are common among microbial community, while the nature of a role played by a microorganism in a symbiotic relationship may alter due to the prevailing conditions. During composting, dynamics of physiochemical factors especially temperature will determine the interaction patterns of microbial populations, favoring the selection of core bacteria capable of intervening in the recycling of carbon and nutrient (López et al. 2021). Numerous studies supported that there existed obvious differences in microbial networks as well as their response to the changes of environmental conditions by understanding “keystone species” on co-occurrence patterns (Zhang et al. 2020). Nowadays, the random matrix theory-based network is increasingly thought as an effective correlation-based relevance network method for analyzing microbiological interaction patterns and their relationship with composting process (Dai et al. 2020; Zhang et al. 2021). Thus, the network method combined with the use of microbial mining and cogent data platforms may allow understanding the core bacterial taxa in the various stages of diverse composting.

This study explored the bacterial community composition, diversity, and co-occurrence relationship in composting from different sources including chicken manure, duck manure, sheep manure, food waste, and vegetable waste by random matrix theory-based network method and redundancy analysis. The objectives of this study were (1) to excavate the information of the biodiversity and composition of bacterial microbiota in diverse composting; (2) to assess the bacterial role associated to composting key environmental factors; and (3) to identify the core bacterial members and their co-occurrence network in diverse composting processes.

Materials and methods

Experimental design

Five composting treatments were performed in 30 days in the 130 L reactors according to Zhan et al. (2021). Chicken manure (CM), duck manure (DM), sheep manure (SM), food waste (FW), and vegetable waste (VW) were collected from different livestock farms, cafeteria, and vegetable market in Dongshan town (Suzhou, China) as the main raw materials, respectively. Sawdust (0.1–1 cm) from a timber mill in Suzhou was added to adjust the initial C/N of composting and moisture. The characteristics of the above materials are shown in Table 1. Different raw materials and sawdust were mixed at a ratio of 4:1 (wet weight). The forced-draft aeration system was used and the aeration rate was 0.2 L·kg−1·min−1. The piles turnings were set on the 4th, 7th, 10th, 15th, 20th, and 25th days with manual operation. Homogeneous samples were collected on days 0, 4, 10, 15, 20, and 30 by multipoint sampling method from different depths and each sample was about 300 g. Mixed samples were deposited in refrigerator at 4 and − 20 °C for physicochemical and microbial analysis, respectively.

Table 1 Some basic characterization of the resource materials
Full size table

Physicochemical parameters

Electrical conductivity (EC) and pH were determined with a digital pH and EC meter using fresh samples. Moisture content was determined at 105 °C for 24 h by calculating weight loss. Germination index (GI) was analyzed as the appendix of Chinese National Standard (NY 525–2021), and cucumber seeds were used for GI analysis. Total organic carbon (TOC), total nitrogen (TN), and total phosphorus (TP) were determined by organic carbon analyzer (TOC-Vcp, Shimadzu, Japan), Kjeldahl method and ascorbic acid/molybdate reagent blue color method as the description of Wei et al. (2021). Olsen P was extracted using 0.5 M NaHCO3 and HClO4–H2SO4 and determined as the same method of TP analysis.

DNA extraction and sequencing

The Soil DNA Isolation Kit (Omega Biotek Inc, USA) was used for microbial DNA extraction from fresh composts (0.50 g) according to the manufacturer’s protocol after washing by corrosion buffer solution to prevent the influence of humus. Bacterial 16S rRNA gene fragments was amplified by the primers pairs 515F (GTGCCAGCMGCCGCGGTAA) as well as 806R (GGACTACVSGGGTATCTAAT) with a 12 nt unique barcode for each sample. The amplification and sequencing of gene were conducted as previous methods of Zhang et al. (2021). Purified 16S rRNA fragments were sequenced based on Illumina sequencing platform (Hiseq2500) in Novogene. The sequences with length > 200 bp, and an average base quality score > 20, without ambiguous base “N” were further analyzed. The operational taxonomic unit (OTU) was assigned from the quality-filtered sequences at 97% similarity according to the SILVA database. Data from sequencing were performed with QIIME 1.7.0.29 and RDP MultiClassifier using the “trainset 16” (Caporaso et al., 2010).

Statistical analysis

Difference was shown with the probability level of significance (P < 0.05) based on SPSS 21.0. Physicochemical parameters, DNA extraction, and sequencing were conducted with three replications. Statistical analyses were conducted by Statistix 8. Redundancy analysis (RDA) was used to analyze the correlation among microbial community and environmental parameters based on Canoco 5.0 (Sheng et al. 2021). The bacterial composition and diversity were compared using the R3.1.2 with related tools in a galaxy instance (www.freebioinfo.org). All data obtained were plotted using Origin 2021. After screening, bacterial taxa with > 50% presence at OTU level among the top one thousand OTU abundance at each sample were used for microbial co-occurrence network. Network analysis was performed by the Pearson correlation coefficients (P < 0.05, |R|> 0.6) using gephi-0.9.2 (Han et al. 2021). Structural equation modeling (SEM) was built by AMOS 21.0 according to Wu et al. (2020).

Results and discussion

Composting development

As for the physical indicators, composting from CM, DM, SM, FW, and VW had four typical degradation phases including mesophilic, thermophilic, cooling, and curing phase as shown in Fig. 1a (Luo et al. 2014). The temperature reached > 50 ℃ (thermophilic phase) at around 2–3 days after the beginning of composting. The duration of thermophilic phase was nearly a week for CM, DM, SM, and VW. Interestingly, there is a 2nd thermophilic phase in FW after turning, whose temperature was significantly higher than other materials composting during days 12 to 22. It may be related to the higher organic matter content and microbial activity in FW (Wei et al., 2018). The pH values of CM, DM, VW, and SM maintained in the range from 7.12 to 9.37 during composting with a slight increase at the beginning (Fig. 1b). The initial pH of FW was low (4.06), which might be caused by a possible anaerobic fermentation in FW storage or transportation process. The pH of FW increased distinctly in the 2nd thermophilic phase and then remained about 8.5. Composts from the five treatments reached the alkaline standard for pH according to Yang et al. (2019). The EC of SM composting was higher than others and there was a significant increase of EC in FW from days 10 to 20 (Fig. 1c). At the final stage of composting, the EC of five treatments was all less than 4 mS/cm (ranged from 1 to 3 mS/cm), which met the composting maturity standard (NY/T 525–2021).

Fig. 1
figure 1

Change of a temperature, b pH, c electrical conductivity (EC), d total organic carbon (TOC), e total nitrogen (TN), f C/N ratio, g total phosphorus (TP), h Olsen phosphorus (Olsen-P), and i germination index (GI) during composting from different sources. CM, chicken manure compost; DM, duck manure compost; SM, sheep manure compost; FW, food waste compost; VW, vegetable waste

Full size image

As for the chemical indicators, TOC content decreased during composting in diverse groups because of the degradation of organic matter by microbes (Fig. 1d). The initial TOC content of FW and VW was 50% and 45%, significantly higher than that in different kinds of manures (P < 0.05). The TOC content of CM, DM, SM, and VW declined by 4–9% after composting. The highest degradation rate of TOC was about 12% for FW and TOC decreased more during the 2nd thermophilic stage, which might be related to the biodegradation of compounds such as carbohydrates and proteins after physical structure improvement (Awasthi et al. 2018). TN varied widely among composts from 1.5% in VW to 3.5% in FW (Fig. 1e). TN content of samples from FW before the 20th day was all higher than that in others (P < 0.01). There was a significant TN decrease of 26% in FW during composting but TN in SM was increased of 30%, which may be related to the lower ammonia volatilization loss (Shi et al. 2021). TN content fluctuated slightly in CM, DM, and VW, ranging from 1.5 to 1.9%. TP content increased distinctly after composting from different sources (Fig. 1g), which is consistent with the degradation of organic materials (Wu et al. 2020). TP content of DM (~ 10.4 g/kg) and CM (~ 9.8 g/kg) was higher than other composts at day 30. The results showed that composts from animal wastes including CM, DM, and SM contained higher amounts of TP than that from municipal solid waste (mainly FW and VW) as the observations of Wei et al. (2015). Olsen P content at the final of composting was all significantly higher than the beginning of composting from diverse materials (P < 0.05) (Fig. 1h). Olsen P ranged from 0.45 to 2.56 g/kg during composting. The ratio of Olsen P relative to TP at the end of composting was the highest for CM 26% and the lowest for FW (14%).

As for the humification indicators, there was a fluctuated decrease in C/N, ranging from 12 to 25 during composting (Fig. 1f). The C/N of CM, DM, FW, and SM at the end of composting was obviously lower than that in VW (P < 0.05). The composts in CM, DM, SM, and VW except FW were basically mature after the 4th day with the GI > 80% (Fig. 1i). At the initial stage of composting until day 10, GI of FW was relatively low possibly due to the low pH (Zhang et al. 2021). At day 30, the GI in different groups was varied from 86 to 107%, suggesting that composts from diverse sources were free of phytotoxic substances and mature.

Anatomy of composting bacterial community

Bacterial composition and diversity

Microbial succession is crucial for an effective composting process. Analysis of the composting indigenous bacterial communities is great significance to find the key bacteria that might be selected as vital inoculant additives participated in different organic fractions transformation (Jurado et al. 2015). Rarefaction curves indicated that the sequencing depth was sufficient to present nearly all the bacterial communities (> 90% coverage). According to the results of high throughput sequencing for 16S rRNA, there were more than 160 million high-quality reads acquired for 30 samples with about 30,000–70,000 reads per sample. These sequences from composting samples were classified into 187,573 different gene clusters (> 97% identity). The dominant bacteria during composting were Firmicutes, Proteobacteria, Actinobacteria, and Bacteroidetes at phylum level, consistent with the national-level survey of Wang et al. (2020) in China (Fig. 2a). However, the changes of bacterial composition were different during composting in the five groups, indicating that the source of raw materials could directly affect the bacterial community. The sum of the relative abundance of these dominant bacteria in five treatments except DM (69%) was above than 80%. The average relative abundance of Firmicutes in FW during composting (58%) was higher than that in other composts (P < 0.05). It was reported that Firmicutes was considered key contributors for the transformation of complex organic compounds (Qi et al. 2021). The results might be one of the reasons of the obvious decrease in TOC in FW. Compared with FW and VW, the relative abundances of Bacteroidetes, Chloroflexi, and Planctomycetes were increased in composting from livestock manure (CM, DM, SM), which were usually identified from animal fecal DNA with certain hosts. The mean proportion of Bacteroidetes in FW (below 1%) was the lowest than other composts, suggesting that the aerobic condition in composting from FW was better than other materials.

Fig. 2
figure 2

Changes of bacterial community during composting from different sources. a Relative abundance of diverse community composition at the phylum level, b at the genus level, c heatmap analysis of dominant genera during composting, and d bacterial community diversity, richness, and evenness

Full size image

The main bacterial genera were Lactobacillus and Thermobifida (Fig. 2b). Pseudomonas and Sphaerobacter was the most abundant native microbes in the raw material of CM, DM, and SM, whereas Thermobifida, Sporolactobacillus, and Thermovum were preponderant in composting, especially in SM. These bacterial taxa were often found in the thermophilic and cooling stages with heat-loving characteristics (López et al. 2021), which were correlated with the higher temperature in CM composting. From the beginning to the final of composting, the relative abundance of Thermobifida enhanced in all composts. It was reported that members of Thermobifida could directly target cellulose for oxidative cleavage of the glucose chains, and even harbor highly thermostable cellulolytic activity, which might support its adaptation for thermophilic composting (Braga et al., 2021; Zainudin et al., 2019). The relative abundance of Lactobacillus in FW and VW was higher than other composts, especially in the beginning of composting, which may be the reason of initial acidic pH (Yang et al. 2019). Bacillus that can form dormant spores under adverse environmental conditions was significantly increased and became dominant in the second thermophilic phase of FW. The results were similar to the observation that Bacillus was one of key bacteria with better collaborative symbiosis with other indigenous microbes in kitchen waste composting (Zhang et al. 2021).

Figure 2c showed that samples from different phases of composting from diverse sources were mainly clustered into 3 groups, suggesting that bacterial community structural patterns basically differed significantly in composting from diverse materials. Obvious distinctions in the composition of bacterial community of VW was observed compared to other composts, which might be related to the higher moisture content of raw material in VW. It is reported that moisture content for composting affect the transportation of nutrients and oxygen flow to maintain the microbial aerobic condition (Rich et al., 2018). The diversity (Shannon) and evenness (Pielous) of bacterial community in composting from diverse raw materials exhibited similar trends except FW (Fig. 2d). The initial Shannon and Pielous indices of FW were lower than other composts, which might be related to the low pH as the organic acids’ accumulation (Wang et al. 2018). However, the bacterial diversity in FW increased in composting, which was not significantly different from other composts at the final stage. The Chao1 index was fluctuant in SM and increased in CM, suggesting that the indigenous microorganism from CM might be more adaptable for composting and being selected as composting inoculant. There was a significant divided increase of Chao1 in FW and the all-time low was appeared at the second thermophilic stage. Considering the advantage of two-stage inoculation (Zhao et al. 2020), the key functional bacterial community with redundancy functionality in FW might need outside help for higher composting efficiency to face the stress from high temperature and acidity. Regression correlation analysis showed that the α-diversity of bacterial community in composting from diverse sources was positively related to pH, GI, and Olsen P but negatively related to TOC and TN (P < 0.05), suggesting that these environmental factors might be crucial for unveiling the core bacteria during composting from diverse sources.

RDA analysis of core bacterial genera

Engineering bacterial communities is crucial to optimize desirable functional processes of composting for diverse raw materials (Qi et al. 2021), which need to analyze the functional roles of bacterial taxa in micro-ecosystem and their relationship with different environmental conditions. Given that in complex communities, bacterial abundance may determine the functional role of phylotypes (Rivett and Bell 2018), RDA was conducted to evaluate how environmental indices interfered the dominant genera with higher relative abundance during composting of CM, DM, SM, FW, and VW (Fig. 3). There was significant correlation between the dominant bacterial genera and environmental factirs in RDA (P < 0.05). TP (P = 0.016) and GI (P = 0.014) showed obvious effects on the main bacterial taxa in CM according to Monte Carlo permutation tests. Thermovum and Saccharophagus were positively related to TP, and Lactobacillus was positively correlated with GI in CM, suggesting its contribution to the humification and maturity of CM. In DM and FW composting, the dominant bacterial compositions were also significantly influenced by GI (P = 0.032 and 0.01). The above results indicated that humus formation as GI increased during composting might reshape the structure of bacterial community in turn (Wu et al. 2020). The bacteria positively correlated with GI in DM were Chelatococcus and Pusillimonas. Chelatococcus and Thermobifida significantly positively affected GI in FW. It has been reported that Chelatococcus was moderately thermophilic and could affect denitrification, indirectly regulating aromatic humic substance formation (Shi et al. 2020). TOC (P = 0.042) and C/N (P = 0.042) showed obvious contribution to the dominant bacterial taxa in VW. RDA showed that Lactobacillus and Pseudomonas were positively correlated with TOC and C/N, suggesting that Lactobacillus and Pseudomonas were conductive to the transformation of organic matter in VW composting. Monte Carlo permutation tests indicated that TP (P = 0.006) and temperature (P = 0.048) had highly obvious effects on the bacterial community structure in SM, which mainly affected Thermobifida, Bacillus, Sporolactobacillus, and Mycobacterium. Moreno et al (2021) reported a specific way to select the thermotolerant microbiota will take charge of the quality of the composts under different facilities or operative strategies. The above results suggested that the primary environmental parameters governing the composition of main microbial species and selecting core functional microbiota were different, which possibly depended on the source of raw materials, composting temperature, and humification process.

Fig. 3
figure 3

Redundancy analysis of dominant bacterial genera and environmental indices during composting from different sources, a chicken manure (CM); b duck manure (DM); c sheep manure (SM); d food waste (FW); e vegetable waste (VW)

Full size image

Core bacterial co-occurrence network and interaction patterns

To deeply understand the core bacteria during composting from different sources, the co-occurrence network was used for the relation of bacterial communities and their interaction patterns. The topological structure of bacteria from OTU level of CM, DM, SM, FW, and VW was compared as shown in Fig. 4, which had an obvious difference among five co-occurrence networks. The number of nodes in the core bacterial network in different composts increased in the order: CM (190), FW (196) < DM (222), SM (224) < VW (236), which were mainly composed of 18 genera in Table 2. However, the number of links in the microbial co-occurrence networks of SM (3741) was the highest, followed by DM (2576), FW (1330), and CM (999). The connection between core bacteria in VW was vulnerable. The above results suggested a more complex and clustering network of SM and DM than other composts. On the other side, the composition of core bacterial network was clearly different. The keystone members in the network at the genus level were Truepera (5.26%), Sphaerobacter (3.68%), and Thermobifida (3.16%) in CM, which were Sphaerobacter (3.15%), Luteimonas (3.15%), and Pseudomonas (3.15%) in DM. These keystones were distinct from the main bacterial composition in composting of CM and DM (Fig. 2). However, the keystones in SM, FW, and VW were similar to their dominant genera with species overlap above 60%, that is, Thermobifida, Pseudomonas, and Lactobacillus in SM, Lactobacillus, Bacillus, and Thermobifida in FW, and Lactobacillus, Enterococcus, and Thermobifida in VW. Overall, the interaction modes of core bacterial community were completely different during composting from diverse sources. Thermobifida was the ubiquitous core bacteria in composting microbial network, while Sphaerobacter and Lactobacillus preferred the starting mesophilic and thermophilic phases of composting from manure (CM, DM, SM) and municipal solid waste (FW, VW) for carbohydrates and lignocellulose degradation.

Fig. 4
figure 4

Co-occurrence network analysis of core bacterial community (OTU level) during composting of chicken manure (CM), duck manure (DM), sheep manure (SM), food waste (FW), and vegetable waste (VW). Only strong “|r|> 0.6” and strongly significant “P < 0.05” interactions were demonstrated in the networks. The nodes with same color belong to same genus according to the figure legend

Full size image
Table 2 The members and proportion of core bacterial genera in the network (only the bacterial taxa with the proportion of node amount > 2% were showed)
Full size table

SEM was structured to reveal the casual relationships between bacterial community composition, bacterial diversity, core bacteria, and composting maturity (C/N and GI) and identify the key biotic factors that driving the maturity during composting from diverse sources. SEM explained 55% of the total variance in C/N and GI (Fig. 5). The total bacterial community composition structure had a significantly positive effect on bacterial diversity (P < 0.001), while the contributions of bacterial diversity to composting maturity were slight and even negligible. Both bacterial community composition and core bacteria showed the positive and direct effects on composting maturity (P < 0.001). Interestingly, bacterial composition structure in composting negatively affected the core bacteria and their interactions, suggesting that the functionally redundant bacterial community might limit the unique components’ function of core bacterial network (Rivett and Bell 2018). In this model system, core bacteria selected from microbial network were the most important parameter influencing composting maturity according to the results of standardized total effects (up to 0.814). Therefore, no matter where the raw materials came from, core bacteria in the co-occurrence network may the key to regulate the biological metabolic strength and energy efficiency of composting, which affect nutrient transformation efficiency on the humification process and product maturity to a certain extent (**e et al. 2021).

Fig. 5
figure 5

Structural equation modeling showing the direct and indirect effects of composting bacterial communities on the compost maturity (C/N ratio and GI value). The r2 values indicate the proportion of the variance explained for the endogenous variable. Red and blue arrows indicate positive and negative relationships, respectively (P < 0.05). Gray arrows indicate no significant relationship. Numbers adjacent to arrows are path coefficients. Significance levels are indicated: *P < 0.05, **P < 0.01, ***P < 0.001

Full size image

In this study, the advanced bioinformatics method based on sequencing analysis, co-occurrence network and SEM are helpful to identify the core bacterial community and their roles for composting from different sources, which presents theoretical basis for improve the full-scale composting efficiency for more and more organic wastes, e.g., garden waste, food and kitchen waste, water plant residues, etc. It was reported that inoculation of functional bacterial agents could accelerate the degradation of organic matter and reconstruct microbial community (Zhan et al. 2021). Wu et al. (2020) suggested endogenous environmental factors could be affected by additives and process control and further change the links between bacteria and the activities of enzymes. According to the above methods to regulate composting microbes, further study would focus on enhance the function of core bacteria to minimize the cost and duration of composting.

Conclusions

This study revealed there were obviously diverse bacterial composition, diversity, and core bacterial genera in composting from CM, DM, SM, FW, and VW, leading to completely different interaction modes of core bacterial community. Thermobifida was the ubiquitous core bacteria in composting bacterial network, while Sphaerobacter and Lactobacillus as core genus preferred the manure (CM, DM, SM) and municipal solid waste (FW, VW), respectively. The core bacteria selected from microbial network had the biggest (above 80%) positive contribution to influencing composting maturity. Sequencing analysis combined with co-occurrence network and SEM could identify the core bacterial community and their roles.