Introduction

Methane contributes up to 30% to total net anthropogenic radiative forcing of the atmosphere1. Atmospheric CH4 has increased 151% since 1750 as a result of an imbalance between increasing sources and decreasing sinks2. The largest biological sink of CH4 is consumption by soil aerobic bacteria3. Upland soils (e.g., forest, grassland and desert) have been shown to be significant sinks for CH44, contributing approximately 6% of the total consumption5. Among upland soils, forest is probably the most efficient CH4 sink4. Urbanization has increased rapidly worldwide and 70% of the world population will be urban by 20506. The urban population of China rose from 18% in 1978 to 45% in 2005 and is projected to be 60% by 20307. Urbanization is generally accompanied with environmental changes in chemistry and climate, which consequently impact the capacity of CH4 consumption in urban forests8,9.

Previous studies showed that CH4 uptake was commonly lower in urban than in suburban and rural forests8,10,11. However, there was disagreement regarding the mechanisms underlying the reductions of CH4 uptake. Goldman et al. ascribed the suppressed CH4 uptake to lower rates of organic matter degradation and nutrient cycling caused by air pollution (especially as O3)8. Groffman and Pouyat reported that increases in N deposition and CO2 levels might be the possible mechanisms10. Costa and Groffman found that the differences in N cycling associated with urbanization led to a reduction in the microbial populations responsible for CH4 uptake12. So far, the effects of urbanization on CH4 uptake in tropical and subtropical forest soils remain unclear.

The objective of this study was to investigate how CH4 fluxes from subtropical forests varied along an urban-to-rural gradient in Guangzhou City metropolitan area and to determine what factors underlie urbanization effects on CH4 uptake. We hypothesized that CH4 consumption would be lowest in urban forests, followed by the suburban and rural forest sites, due to the exposure to urbanization-induced environmental changes of the former.

Results

Soil CH4 uptake

Heishiding (HSD, rural site) forests had the highest rates of CH4 uptake, followed by Dinghushan (DHS, suburban site) and then Maofengshan (MFS, urban site) forests (Fig. 1 a, b, all P < 0.05). For pine and broadleaf mixed forests (MF), the mean rate of CH4 uptake in HSD was higher by 38% than in MFS site (Fig. 1 a). In broadleaf forests (BF), the average rate of CH4 uptake was higher by 37% in DHS and 44% in HSD than that of MFS site (Fig. 1 b, all P < 0.05). In DHS forest site, average CH4 uptake for BF was 37% higher than that of MF (P = 0.04). However, there were no differences between MF and BF stands at other forest sites. Monthly mean CH4 uptake in all forests showed a similar seasonal pattern, with the highest consumption occurring in the fall (August to October) (Fig. 1 c, d).

Figure 1
figure 1

Comparisons of the mean rates and seasonal patterns of CH4 uptake.

(a) and (b) depict the mean rates of CH4 uptake and (c) and (d) show the seasonal patterns of soil CH4 uptake (Mean value ± 1 standard error, n = 5); Different letter “a” and “b” denote significant differences (P < 0.05) for the same forest type. MF, mixed forest; BF, broadleaf forest; MFS, Maofengshan; DHS, Dinghushan; HSD, Heishiding.

Full size image

Biotic and abiotic variables

The amount of N deposition in rainfall was higher in MFS and DHS than in HSD site (P < 0.05), with no significant difference between MFS and DHS (Table 1). Soil temperature and WFPS exhibited clear seasonal patterns (Fig. 2 a, b), similar to those of air temperature and rainfall at each site (Fig. S1). Higher soil WFPS and temperatures were found in MFS forests than that of HSD site (Fig. 2 c, d, all P < 0.05).

Table 1 Site characteristics along an urban-to-rural gradient in southern China
Full size table
Figure 2
figure 2

Soil temperature and WFPS at forests along the urban-to-rural gradient.

(a) Seasonal patterns of soil WFPS at 0–10 cm depth; (b) seasonal patterns of soil temperature at 5 cm depth; (c) annual mean soil WFPS; and (d) annual mean soil temperature (Mean value ± 1 standard error, n = 5). In panel (a) and (b), letters M and B following the site abbreviations denote mixed forest and evergreen broadleaf forest, respectively. Different letter “a” and “b” denotes significant difference between forests for the same forest type (P < 0.05). MFS, Maofengshan; DHS, Dinghushan; HSD, Heishiding.

Full size image

For both forest types, soil NO3− contents were higher in HSD than that of MFS and DHS sites, whereas soil NH4+ and total N (TN) showed an opposite pattern (Table 2). Soil bulk density and pH values were significantly lower in MFS and DHS than in HSD forests, conversely, soil Al3+ contents were higher in MFS and DHS sites (Table 2). Soil CH4 uptake showed negatively related with soil NO3− and Al3+ contents (Fig. 3 a, c) and positively related to pH values (Fig. 3 b). Soil microbial biomass C (MBC) contents were lowest in MFS, followed by DHS and HSD for the same forest stands, whereas microbial biomass N (MBN) had no significant differences across the gradient sites (Table 2).

Table 2 Soil properties (0–10 cm depth) of forests along the urban-to-rural gradient
Full size table
Figure 3
figure 3

Linear regressions between annual CH4 uptake rates and soil properties.

(a) soil NO3− contents and CH4 fluxes, (b) soil pH values and CH4 fluxes and (c) soil Al3+ contents and CH4 fluxes. CH4 fluxes used in this figure were measured in field at the same day of soil sampling (July 2009). MF, mixed forest; BF, evergreen broadleaf forest.

Full size image

Discussion

Methane fluxes at all sites were predominantly negative during the study period, indicating CH4 uptake from the atmosphere. While many studies have quantified differences in CH4 cycling under the changes of land uses11, there is considerable uncertainty about the unmanaged or intact forests within urban areas acting as CH4 sinks that have been observed8,10,12. The rates of CH4 uptake in the suburban and rural forests (2.4 to 3.3 kg CH4-C ha−1 yr−1) were comparable with previous studies in tropical and subtropical regions of southern China (2.5 to 4.3 kg CH4-C ha−1 yr−1)13,14,15. In the suburban (DHS) site, the soil of BF oxidized more CH4 than that of MF stand, a trend observed in previous studies13,16. The BF is an old-growth forest (more than 400 years), which represents a forest type in an advanced successional stage within the study region17. Forest composition, species assemblages and soil density (Table S1) are more favorable than MF, leading to increased CH4 consumption in this BF stand4,13,16.

Consistent with our previous hypothesis, the urban forest soils had the lowest capacity of CH4 consumption compared to rural and suburban sites. Our results were comparable with previous reports that CH4 uptake would be markedly lower in urban than in rural forest soils. For example, Goldman et al. found that CH4 uptake rates in intact forests in urban center of New York City (NY, USA) were 30% lower than that of rural forest soils8. Similar results were found in urban forests in the Baltimore metropolitan area (MD, USA)10,11. These results have suggested that there is an urban atmospheric effect on CH4 consumption in urban forests12. In the present study, the reduction of CH4 uptake in urban forests might be influenced by several factors as follows.

Firstly, the reduction of CH4 uptake was primarily influenced by the lower diffusion of CH4 into urban forest soils. Multiple regression analyses indicated that soil WFPS accounted for 52% of the variance in CH4 uptake for both MF and BF stands (Table 3). Higher soil WFPS and bulk density in the urban forests could increase resistance to atmospheric CH4 and O2 transport into the soils, reducing the activity of methanotrophic bacteria4,18. The seasonal changes in CH4 uptake were also best explained by gas diffusion and were most related with lower soil WFPS during the fall. There was no significant relationship between CH4 uptake and soil temperature, which was consistent with previous studies in this region13,15,19. These results suggest that soil temperature is not the key factor for controlling CH4 consumption in subtropical forests of southern China.

Table 3 Responses of CH4 uptake rates to soil variables: multiple regression analysis and partitioning the variance in R2
Full size table

Secondly, higher N cycling rates and soil NO3− contents, which are caused by high rates of atmospheric N deposition, significantly decreased soil CH4 uptake in the urban forests. Soil NO3− contents accounted for 30% of the variance in CH4 uptake (Table 3). Previous studies have shown that NO3− and/or NO2− produced from the NO3− reduction were possibly toxic to CH4-oxidizing microbes20,21, which might be a potential role for the reduction of CH4 consumption in our urban forests. Although we did not measure net N mineralization and nitrification rates, the fact that these results were consistent with patterns of N deposition and soil N cycling already published for this urban-to-rural gradient22, strongly suggest that higher rates of N cycling underlie the reductions of CH4 consumption in the urban forests10,12. We found that soil NH4+ and TN contents tended to be higher in the rural than that in suburban and urban forest sites, which was in conflict with the generally accepted idea that soil N contents tend to be higher in urban than rural forests23. However, our result was comparable with previous studies within the same region22,24. The cause may be a higher rate of N losses in the urban site22.

Thirdly, CH4 uptake might also be reduced by the changes in soil pH value and Al3+ content of the urban forests. A decrease of soil pH in urban forests might lead to decreased CH4 uptake25, which was noted in the Broadbalk Experiment in southeastern England26. Soil acidification might cause a release of heavy metals, such as Al3+, inhibiting soil CH4 uptake26. Al3+ toxicity could inhibit CH4 uptake27, which potentially explains the results from our study.

Finally, the suppressive effect of urbanization-induced environmental changes on CH4-consuming bacteria might be another mechanism for the CH4 reduction. Urban environments might decrease CH4 uptake indirectly through changes in the habitat of the methanotrophic bacteria28. Long-term N deposition could cause a decrease in methanotrophic populations via niche competition with nitrifiers29, leading to a reduction in CH4 consumption30. In the present study, a significant decrease in soil MBC at the urban site compared to the rural and suburban forest sites, might negatively affect the size, composition and activity of the CH4 oxidizing community10,31. The urbanization-induced environmental changes might have profound implications for CH4 consuming communities and further research should be conducted.

In summary, the largest consumption of CH4 was found in the rural, followed by the suburban and urban forest sites. The data presented here strongly indicated that the reduction of CH4 uptake in urban forests was due to higher soil WFPS and adverse effects of urbanization-induced environmental changes, such as atmospheric N deposition, higher soil NO3− content and Al3+ toxicity. This is the first in situ study to attempt to clarify the effects of urbanization-induced environmental changes on soil CH4 uptake in tropical and subtropical forests. Our results suggest that the projected environmental changes associated with urbanization would decrease CH4 consumption in urban forest soils of subtropical regions and this reduction of CH4 consumption needs to be considered in global CH4 budgets.

Methods

Study area

The study sites were located along a 150 km gradient that extends from MFS (an urban site near Guangzhou City), through DHS (a suburban site), to HSD (a rural site) within the Pearl River Delta (PRD) region of Guangdong Province, South China (Fig. S2). Some characteristics of the study sites were presented in Table 1. The PRD region is one of the regions experiencing rapid urbanization with its population increasing nearly two fold from 1982 (54 million) to 2010 (104 million)32. The study region has a subtropical monsoon climate. There was a gradient for environmental variables from Guangzhou City to its surroundings (Table 1).

At each site, a MF and a BF were selected for experimental plots. Several criteria were used in study site selection to ensure comparability among the forests: (1) no disturbance after planting (such as fire, insect infestations, logging and fertilization); (2) forest age between 50 and 70 year (excluding the BF in DHS); (3) soil of lateritic red earth (Table S1). The dominant species and plant characteristics are described in Table S1. Within each forest, five 10 m × 10 m plots were randomly established at least 100 m from the edge to avoid “edge effects”.

Measurement of CH4 fluxes

Soil CH4 fluxes were measured from April 2009 to March 2010 using the static chamber method. Gas samples were collected biweekly during the growing season (April to September) and monthly at other times. The chamber design and the measurement procedure were adopted from Zhang et al13. Gas samples were collected with 60 ml plastic syringes at 0, 15 and 30 min after the cover chamber closure and immediately injected into special gas bags (Yile CO., LTD, Shanghai, China). The samples were transferred to a lab and analyzed within 2–3 days by a gas chromatography with flame ionization detection (FID). Atmospheric pressure, air temperature (inside chamber), soil temperature (5 cm depth) and moisture (0–10 cm depth) were measured during each sampling event. Soil moisture was converted to WFPS (%). Details of the measurement were referenced from Zhang et al.13.

Soil sampling and analysis

Soil samples (0–10 cm depth) were collected on July 2009. Three soil cores (3.5 cm internal diameter, ID) were collected randomly from each plot and combined to one composite sample. The samples were passed through a 2 mm sieve and divided to two parts. One subsample was used for the analysis of NH4+, NO3−, MBC and MBN. The other was air dried at 25°C for the estimation of other chemical parameters. Gravimetric water content was determined through oven drying at 105°C for 48 h.

Soil NH4+ and NO3− contents were determined by extraction with 2 M KCl solution followed by colorimetric analysis on a flow-injection autoanalyzer (Lachat Instruments, Milwaukee, USA). TN content was determined by micro-Kjeldahl digestion33, followed by detection of ammonium with a UV-8000 Spectrophotometer (Metash Instruments Corp., Shanghai, China). Soil MBC and MBN contents were estimated by chloroform fumigation-extraction method34. For each sample, two fresh soil subsamples (10 g dry soil equivalent) were prepared. One subsample was fumigated with chloroform for 24 h at 25°C. The other was treated as control. Soil MBC and MBN contents were calculated as the difference in extractable C, N between fumigated and non-fumigated subsamples, using the conversion factors of 0.33 and 0.45 for MBC and MBN, respectively35,36.

Atmospheric N deposition

We used ion-exchange resin (IER) columns to monitor N deposition in precipitation at the three sites during the study period. The IER columns and the measurement procedure were adopted from Fang et al.22. The resin columns were collected with interval for three months. The concentrations of NH4+ and NO3− were determined by 2 M KCl extraction as described above.

Statistical analysis

Repeated Measures Analysis of Variance (ANOVA) was used to examine the differences of soil CH4 fluxes among the three sites. One-way ANOVA was performed to compare the differences in soil properties, MBC and MBN within the same forest type. Linear regression analysis was performed to quantify the relationships between CH4 fluxes and individual soil variables. Multivariate linear regression analysis was performed using CH4 uptake rate as dependent variable and soil WFPS, NO3−, Al3+ contents and pH values as independent variables. R2 was partitioned to quantify the importance of each independent variable. All statistical analyses were conducted using SPSS 16.0 for windows (SPSS Inc., Chicago, IL, USA). Statistically significant difference was set at P ≤ 0.05 unless otherwise stated. Mean values ± 1 standard error were reported.