Introduction

Landfills are widely used around the world because of their low cost and ease of management (Long et al. 2008). Conventional landfill (anaerobic landfill) always needs long-term maintenance because of the slow rate of decomposition, while decomposition can be enhanced in bioreactor landfill with the help of leachate recirculation (Long et al. 2010a). However, there is a limited capacity for landfills with increased amounts of municipal solid waste (MSW), and creating new landfills once an old landfill is closed is very difficult because of a shortage of land (Sener et al. 2006). Compared with traditional anaerobic landfill, semi-aerobic landfill has become popular because of its faster degradation rate and higher leachate quality without an obvious increase in cost (Hirata et al. 2012). Regardless of the mode of operation, the risk of odor pollution from landfills still exists. Hydrogen sulfide (H2S), with an extremely low odor threshold (around 0.5 ppb) and high toxicity (Firer et al. 2008), is one of the major contributors to odor pollution from landfill and it has been found at concentrations ranging from hundreds of micrograms per cubic meter to thousands of milligrams per cubic meter in air at different landfill sites (Ding et al. 2012; Kim et al. 2005).

H2S is produced by the conversion of organic and/or inorganic compounds containing sulfur (CCS) during decomposition. Landfills are complex artificial habitats and the microbial communities resident in landfill are responsible for the conversion of CCS. For example, sulfate-reducing bacteria (SRB) can play an instrumental role in biogenic production of H2S. Meanwhile, biooxidation of sulfide and intermediary sulfur compounds carried out by sulfide-oxidizing bacteria (SOB) are crucial in the removal of H2S. Sulfur-metabolizing bacteria play an important role in sulfur cycling in landfills. Sulfur-metabolizing bacteria have been the focus of many previous studies. Zhang et al. (2012) found that the bacterial and archaeal communities from a white hydrothermal plume were dominated by sulfur-reducing Nautilia and Thermococcus, whereas a yellow hydrothermal plume and the surface water were dominated by sulfide-oxidizing Thiomicrospira and Euryarchaeota Marine Group II, respectively. Lou et al. (2013) concentrated on enrichment, isolation, and identification of SOB from a sulfide removing bioreactor. Rhodopseudomonas and Halothiobacillus were found to be the main SOB in the sulfide-removing reactor, and were responsible for sulfur oxidation in the treatment system. Unfortunately, studies targeting the composition of sulfur-metabolizing bacteria in landfill refuse are scarce. Differences in H2S emissions from landfills might be attributed to the differences in activity of some sulfur-metabolizing bacteria. Therefore, it is necessary to focus on the relationship between sulfur-metabolizing bacterial composition and landfill scenarios. This could provide a reference for controlling landfill bacterial community composition and transforming landfill technology based on the identification of H2S emission behaviors.

In this study, the distribution of H2S emission at different depths inside landfill was first examined in six lab-scale simulated landfills with different operation modes. The microbial community composition inside different landfills during each designated landfill phase, especially the sulfur-metabolizing bacterial community, was investigated using high-throughput sequencing technology. The study aimed to evaluate the relationship between the H2S emission behavior and the corresponding sulfur-metabolizing bacterial community in different landfill scenarios such as landfill modes and depths.

Materials and methods

Experimental set-up

Six different sets of simulated bioreactor landfills were designed. R1 and R3 were operated as anaerobic landfill; R2 and R4 were semi-aerobic landfill; R5 and R6 were semi-aerobic landfills switched from anaerobic mode after day 175. The leachate of R1, R2, and R6 were single pass leaching, while the leachate of R3, R4, and R5 were directly recirculated by a peristaltic pump into simulated landfill after collection every two days. Complete configurations of the six landfills are shown in Fig. 1. Each simulated landfill had a diameter of 0.5 m and a height of 2.0 m and was operated at room temperature. The simulated landfills with semi-aerobic operation mode had three inlets at the bottom of each landfill. The three inlets were set in the same horizontal position, and the angle between them was 120°. A center vent pipe with a diameter of 5 cm was set in the middle of each semi-aerobic landfill. The semi-aerobic system achieved through a convection process. Before day 175, the inlets and center vent pipes in R5 and R6 were in a closed state to maintain anaerobic conditions. From top to bottom, each landfill had a 100-mm-thick layer of headspace, a 1600-mm-thick layer of landfill site, and a 300-mm-thick layer of leachate collection. A 100-mm-thick layer of gravel was placed at the bottom of each landfill site to simulate a leachate collection system and to prevent clogging of the leachate withdrawal outlets. The MSW was loaded in 1450-mm layers and compacted using a shovel and a sledge hammer. A 50-mm-thick layer of sand was placed on top of the MSW to simulate intermediate cover and an upper drainage layer and to provide even distribution of the recirculated leachate. Headspace on top of each landfill created a leachate distribution system. The MSW layer was divided into three layers: shallow, middle, and deep layers and each layer had three sampling ports. Each simulated landfill was packed with fresh refuse collected from the transport station of Hangzhou (Zhejiang, east China), with a wet density of 880 kg m−3. The characteristics of MSW used in this experimental were (by wet weight, w/w): food and fruit waste, 66.8 %; plastic, 3.2 %; paper, 9.2 %; dirt, 1.8 %; glass, 0.4 %; cellulose textile, 0.5 %; metal, 0.2 %; timber, 1.6 %; residue, 16.5 %. The moisture content of MSW was 68 %.

Fig. 1
figure 1

Schematic diagram of simulated landfill systems (1 Leachate outlet 2 Leachate collection tank 3 Gravel layer 4 Landfill site 5 Sampling port 6 Sandy layer 7. Headspace 8 Gas outlet 9 Leachate recirculation tank 10 Peristaltic pump 11 Leachate distribution system 12 Center vent pipe 13 Inlet)

Full size image

Sample collection

Based on our previous landfill studies (Fang et al. 2009a, b; Long et al. 2009, 2010b), the decomposition of simulated landfills approached stabilization after about 200 days’ operation. In this study, we observed that the cover layer of six simulated landfills significantly settled down after 175 days. This meant that the rapid degradation phase finished by this time. To evaluate the further stabilization behavior of landfill, we turned on the ventilation system of R5 and R6 to switch them from anaerobic to semi-aerobic. Therefore, after 175 days operation, the six landfill reactors were anaerobic landfill (R1 and R3), semi-aerobic landfill (R2 and R4), and semi-aerobic landfills switched from anaerobic (R5 and R6).

At day 200, gas and refuse were sampled from three layers of the sampling ports around the side of each simulated landfill. Gas samples in the positions of 5, 12, and 20 cm from the center vent were collected. Approximately 100 g refuse samples were collected from each refuse sampling port and the refuse samples from all three ports in one layer were mixed as a sample (~300 g), of which 50-g refuse samples were stored at room temperature to dry and 20-g refuse samples were stored at −80 °C for the following bacterial community analysis. Because of the landfill settlement caused by the rapid degradation of refuse, the refuse samples in the shallow layer of R2 and R4 could not be collected.

Chemical analysis

Gas samples were analyzed for H2S and CO2. The H2S in the gas samples was analyzed using a gas chromatograph equipped with a flame photometric detector (GC 7890A; Agilent Technologies, Santa Clara, CA, USA) (Fang et al. 2015). A gas chromatograph equipped with a thermal conductivity detector (GC 7890II; Shanghai Tianmei Scientific Instruments Co., Ltd., Shanghai, China) (Fang et al. 2015) was used to determine the CO2 concentrations in the gas samples.

pH, dissolved organic carbon (DOC), sulfate (SO4 2−), sulfide (H2S, HS−, S2−), ferrous (Fe2+), nitrate (NO3 −), and nitrite (NO2 −) concentrations of the refuse samples were analyzed on the day the samples were collected. Distilled water was used as the extraction solution (solid–liquid ratio of 1:10) for analysis of pH, DOC, and SO4 2−. The pH was determined using a pH meter (SevenEasy, Mettler-Toledo, Switzerland). The DOC was determined using a TOC analyzer (TOC-V CPN, Shimadzu, Japan). The SO4 2− was determined using an ion chromatograph described by Fang et al. (2015). The sulfide was determined by the method described by Qiu et al. (1992). The ferrous concentration was determined by the o-phenanthroline method after extraction with 1 mol L−1 HCl (solid-to-liquid ratio of 1:100) (Li et al. 2010). Then, 2 mol L−1 KCl solution was used as the extraction solution (solid–liquid ratio of 1:10) for analysis of NO3 − and NO2 −. NO3 − and NO2 − were determined by ultraviolet spectrophotometry and N-(1-naphthyl)-ethylenediamine dihydrochloride, respectively (APHA 1999). The dry samples were milled for the subsequent determination of TN by the H2SO4–H2O2 digestion indophenol-blue colorimetric method (Lu 1999).

All the analyses were carried out in triplicate from three samples to ensure the validity of the results, and all the results of chemical analyses were calculated on a dry-weight basis.

Bacterial community analysis

The genomic DNA of the sample was extracted using an extraction kit (DR4011; Bioteke Corporation, Beijing, China) according to the manufacturer’s instructions. The bacterial 16S rRNA gene of the extracted DNA was amplified using the primer pair 338F (5′-ACTCCTACGGGAGGCAGCAG-3′) and 806R (5′-GGACTACATCGACGGGTATTCTAAT-3′) (Masoud et al. 2011). The bacterial community was investigated by Illumina high-throughput sequencing, which was conducted by Majorbio Bio-Pharm Technology Co., Ltd (Shanghai, China) (Zhang et al. 2015). Sequences were clustered into operational taxonomic units by setting a 0.03 distance limit (equivalent to 97 % similarity) using the Usearch program. Sequences were then phylogenetically assigned to taxonomic classifications using an RDP classifier and were allocated to different levels (Zhang et al. 2015). Raw sequence files for the 16 samples analyzed here have been deposited in the NCBI Sequence Read Archive.

Statistical analysis

Statistical analyses were conducted using surfer 10, R Statistics Program, and SPSS 19.0. Relative abundance heat maps were created from phylum-level bacterial annotations. Prior to statistical analysis, relative abundance was log-transformed to achieve a more normal distribution. Linear regressions were performed using SPSS software to relate environmental factors to community composition. Multivariate analysis was conducted using canonical correspondence analysis (CCA) of genus-level bacterial taxa and refuse property data. Calculations were performed using the CCA function of the vegan package in R Statistics Program using refuse property data as environmental parameters. Data were transformed and scaled prior to analysis as follows: taxa abundances were log(x + 1)-transformed to dampen the contribution of abundant taxa and each environmental parameter was converted to its z-score.

Results and discussion

Distribution of H2S emission in landfill

The H2S concentration varied in different landfills, with relatively higher concentrations under anaerobic conditions (R1 and R3) and lower concentrations under semi-aerobic conditions (R2 and R4) (Fig. 2). These results corresponded to the fact that during the degradation of sulfur-containing substrates under anaerobic conditions in landfill, H2S will be produced and released (Mescia et al. 2011). For R1 and R3, the highest H2S concentrations reached 4.46 ± 0.68 and 6.13 ± 0.32 mg m−3, respectively, under anaerobic conditions, while in R2 and R4, the highest H2S concentrations reached 3.70 ± 0.13 and 3.17 ± 0.07 mg m−3, respectively, under semi-aerobic conditions. The difference in H2S emission behavior between anaerobic and aerobic conditions can be further verified by the special landfill mode, namely semi-aerobic switched from anaerobic. After transformation from anaerobic to semi-aerobic conditions (R5 and R6), the H2S concentrations substantially decreased with air exposure. The highest H2S concentrations in R5 and R6 decreased to 2.92 ± 0.13 and 2.73 ± 0.08 mg m−3, respectively, after exposure to air. However, the lowest H2S concentrations in R5 and R6 were 1.4 and 1.5 times higher than in R2 and R4. The H2S emissions were lowest in R2 and R4 of all the landfill configurations tested. The difference in H2S emission behavior was mainly attributed to the different environment for biodegradation. Under anoxic and anaerobic conditions, some SRB can grow with archaea and use SO4 2− as an electron acceptor to oxidize CH4 (Boetius et al. 2000; Michaelis et al. 2002; Caldwell et al. 2008), whereas under oxic conditions SOB oxidize H2S to S0 and SO4 2− (Kelly et al. 1997). The semi-aerobic mode of landfill operation could substantially attenuate the risk of H2S pollution in landfill. One of the main reasons for this might be the difference in biological processes related to sulfur metabolism driven by functional microbes.

Fig. 2
figure 2

Distribution curve of H2S emission in simulated landfill

Full size image

Diversity of sulfur-metabolizing bacteria in landfill

To investigate the differences in H2S emission behavior described above, refuse samples at day 200 were sampled to analyze the microbial diversity. High-throughput sequencing results showed that the microbial community structures in landfill at the phylum level varied with operation modes and sampling depths (Fig. 3). Five major phyla accounted for more than 94 % of the total bacterial abundance in the refuse samples investigated here, including Firmicutes, Proteobacteria, Actinobacteria, Bacteroidetes, and Cyanobacteria. Some members of the Firmicutes and Proteobacteria are known to be involved in the reductive and oxidative pathway of the sulfur cycle (Köchling et al. 2015; Luo et al. 2013). Schauer et al. (2011) found that Bacteroidetes was highly abundant in sulfur-rich sediments. Phylum Firmicutes was predominant in the deep layer of R1 (55 %) and R3 (49 %), and also accounted for relatively high proportions in the middle layer of R2 (50 %) and R4 (52 %). After semi-aerobic transformation, the abundance of phylum Firmicutes increased to 66 and 73 % in the top layer of R5 and the middle layer of R6, respectively. Phylum Proteobacteria was most abundant in the middle layer of R1 (51 %) and R3 (50 %) and accounted for 44 and 26 % in the deep layer of R2 and R4, respectively. With air exposure, a large shift in the proportion of Proteobacteria was observed in R5 and R6, showing the largest relative abundance in the deep layer of R5 (35 %) and in the shallow layer of R6 (41 %). The proportion of Bacteroidetes varied greatly in the refuse samples (0.5–21 %). Actinobacteria and Cyanobacteria occurred in the shallow layer of R6 with a relative abundance of 3.8 and 0.05 %, respectively, but they were detected at higher proportions in other refuse samples. Comparison among the five major phyla abundances showed that Proteobacteria was most consistent with the H2S emission behaviors. Based on the relative abundance of bacterial phyla, cluster analysis showed that the refuse samples from the landfills under the same oxygen conditions grouped together. Samples from the anaerobic landfill, semi-aerobic landfill, and the semi-aerobic transformation landfills were separated from each other, which was consistent with the distribution of H2S emission in landfills. These results indicated that the oxygen condition might affect the composition of sulfur-metabolizing bacterial communities and then cause the final H2S emission behavior difference, of which Proteobacteria abundance might have a greater influence on H2S emission.

Fig. 3
figure 3

Relative abundance heatmap of phylum-level bacterial communities. Dendrogram of sample sites (average clustering technique) based on community similarity along top axis; Left dendrogram depicts clustering of phyla by co-occurrence (average clustering technique)

Full size image

Typical sulfur-metabolizing bacteria being reported were detected in these six landfills (Table 1), including Acinetobacter, Bacillus, Comamonas, Ochrobactrum, Paracoccus, Pseudomonas, and Rhodococcus. Among them, Bacillus affiliated with the phylum Firmicutes, Rhodococcus affiliated with the phylum Actinobacteria, and the remaining five genera affiliated with the phylum Proteobacteria. In these six landfills, the relative abundances of the known sulfur-metabolizing bacteria ranged from 1.1 to 12 %. Acinetobacter was the most abundant known sulfur-metabolizing genus. Acinetobacter can use dimethyl sulfide as the sole sulfur source (Horinouchi et al. 1997) and participates in thiosulfate oxidation (Luo et al. 2013). It was detected with the highest abundance in R3 and with the lowest abundance in R4. Correspondingly, it was observed that the H2S emission was the highest in R3 and the lowest in R4. Similarly, the abundance of Paracoccus was related to the H2S emissions. Paracoccus are facultative aerobes that can grow heterotrophically with various carbon sources or chemoautotrophically with thiosulfate and sulfide as electron donors under aerobic conditions (Frierich et al. 2008). This suggested that Acinetobacter and Paracoccus might play important roles in the metabolism of H2S in landfills. The abundances of Ochrobactrum and Pseudomonas were relatively low in R4. They can not only use sulfide and thiosulfate for growth under aerobic conditions, but can also use them as a source of electrons to reduce nitrite and/or nitrate anaerobically (Euzeby 1997; Sorokin et al. 1999; Zhang et al. 2008; Mahmood et al. 2009). The abundances of Bacillus, Comamonas, and Rhodocccus were low in all landfills. The differences in abundance of Acinetobacter and Paracoccus might be the main reason for the observed differences in H2S emission.

Table 1 The relative abundances (% of total reads) of the known sulfur-metabolizing bacteria in the simulated landfills, based on the abundances of 16S rRNA gene sequences detected in the DNA by use of pyrosequencing
Full size table

Effect of environmental parameters on bacterial community composition

CCA was performed to investigate the possible relationship between microbial community composition and the detailed surrounding environment of different landfill scenarios (Fig. 4). Based on variance inflation factors, 10 significant environmental variables (Fe2+, SO4 2−, sulfide, total nitrogen (TN), NO3 −, NO2 −, pH, DOC, H2S, and CO2) were selected for inclusion in the CCA biplot. The first canonical axis explained 24.65 % of the detected microbial diversity and was positively correlated with SO4 2−, CO2, H2S, TN, NO2 −, pH, and DOC. The second axis represented 13.66 % of variance and was positively correlated with Fe2+, sulfide, and NO3 −. The length of an environmental parameter arrow in the ordination plot indicates the strength of the relationship of that parameter to community composition. As such, pH, NO2 −, and sulfide appeared to be the most important environmental factors. The stress of pH has a significant effect on the overall diversity and composition of microbial communities in a range of terrestrial and aquatic environments (Fierer and Jackson 2006; Hornstrom 2002). In addition to pH, NO2 −, sulfide, and DOC were strongly and significantly linked to bacterial community variance in the CCA. Microbial community structure was also influenced by Fe2+, NO3 −, H2S, and CO2. The relatively small magnitude of SO4 2− and TN vectors indicated that SO4 2− and TN were not as strongly correlated to community composition as other environmental parameters. Sulfide was positively related to Fe2+ and negatively related to NO3 − whilst Fe2+ was also positively correlated to H2S. There was no significant relationship between NO3 − and H2S. In general, R1 and R3 had a positive relationship with H2S, while the opposite relationship appeared in R2, R4, R5, and R6. These results showed that H2S concentration and microbial community composition greatly differed with the operation modes. The landfills with different oxygen distribution could lead to different degradation rates and biological processes. In contrast, the semi-aerobic mode might be more favorable for controlling H2S emission in landfill.

Fig. 4
figure 4

Canonical correspondence analysis of log (x+1)-normalized genus-level bacterial communities, constrained by the independent, normalized refuse properties. Samples are labeled in black. Blue vectors indicate the effect of refuse properties on sample community outcomes. Red taxa points indicate the average CCA location of the genus-level bacterial taxa. The percentage of variation explained by each axis is shown. The CCA model is significant (P = 0.007)

Full size image

Pairwise linear regressions were further performed to assess the potential influence of environmental attributes on the sulfur-metabolizing bacterial community or environmental outcomes (Table 2). Any pairwise relationships not listed had a regression p value greater than 0.1. The higher pH levels observed were correlated with higher NO2 − and sulfide levels, lower SO4 2− level, and decreased relative abundance of Acinetobacter, Ochrobactrum, and Rhodococcus (p values of 0.000, 0.003, 0.017, 0.086, 0.071, and 0.048, respectively). The relative abundance of Paracoccus and Rhodococcus were found to be positively correlated with the levels of DOC (p < 0.05). The relative abundance of Acinetobacter and Rhodococcus showed a negative correlation with the level of NO2 −, while the relative abundance of Bacillus and Comamonas showed a positive correlation with the level of NO2 −. Paracoccus was positively correlated with the levels of H2S and CO2, while Ochrobactrum was positively correlated with the level of SO4 2−. Pseudomonas showed no significant correlation with the determined physicochemical parameters. Paracoccus had the most direct relationship with H2S emission behavior. Relatively higher H2S concentration in landfill was associated with increased relative abundance of Paracoccus, the genus commonly associated with the microbial sulfur cycle.

Table 2 Significant pairwise linear regressions for community and environmental parameters (Community included Acinetobacter, Bacillus, Comamonas, Ochrobactrum, Paracoccus, Pseudomonas, Rhodococcus richness; Environmental parameters included ferrous, sulfate, sulfide, total nitrogen, nitrate, nitrite, pH, DOC, H2S and CO2 content)
Full size table

In summary, high-throughput sequencing was employed to examine the bacterial communities of refuse samples from landfills with different operation modes and sampling depths. In addition to certain unique genera of bacterial populations in each sample, the known sulfur-metabolizing genera were commonly shared by all samples. The CCA analysis and pairwise linear regressions results showed that a myriad of factors might mutually shape the microbial community composition in landfills. The sulfur-metabolizing bacterial community composition influenced by the operation modes had a strong relationship with H2S concentration. The semi-aerobic mode with relatively low abundance of Paracoccus might reduce the risk of H2S pollution in landfills. Further work is necessary to elucidate the factors regulating the landfill bacterial community.

Conclusions

H2S concentration and microbial community composition might vary depending on the landfill operation modes. The semi-aerobic mode can substantially attenuate the risks of H2S pollution in landfill, which might be because of the difference in biological processes related to sulfur metabolism driven by functional microbes. The difference in abundance of the genera Acinetobacter and Paracoccus (phylum Proteobacteria) caused by environmental factors might be the main reason for differences in H2S emission in landfill.