Impact of nitrogen fertilization on soil–Atmosphere greenhouse gas exchanges in eucalypt plantations with different soil characteristics in southern China

Nitrogen (N) fertilization is necessary to sustain productivity in eucalypt plantations, but it can increase the risk of greenhouse gas emissions. However, the response of soil greenhouse gas emissions to N fertilization might be influenced by soil characteristics, which is of great significance for accurately assessing greenhouse gas budgets and scientific fertilization in plantations. We conducted a two-year N fertilization experiment (control [CK], low N [LN], middle N [MN] and high N [HN] fertilization) in two eucalypt plantations with different soil characteristics (higher and lower soil organic carbon sites [HSOC and LSOC]) in Guangxi, China, and assessed soil–atmosphere greenhouse gas exchanges. The annual mean fluxes of soil CO2, CH4, and N2O were separately 153–266 mg m-2 h-1, -55 –-40 μg m-2 h-1, and 11–95 μg m-2 h-1, with CO2 and N2O emissions showing significant seasonal variations. N fertilization significantly increased soil CO2 and N2O emissions and decreased CH4 uptake at both sites. There were significant interactions of N fertilization and SOC level on soil CO2 and N2O emissions. At the LSOC site, the annual mean flux of soil CO2 emission was only significantly higher than the CK treatment in the HN treatment, but, at the HSOC site, the annual mean flux of soil CO2 emission was significantly higher for both the LN (or MN) and HN treatments in comparison to the CK treatment. Under the CK and LN treatments, the annual mean flux of N2O emission was not significantly different between HSOC and LSOC sites, but under the HN treatment, it was significantly higher in the HSOC site than in the LSOC site. Correlation analysis showed that changes in soil CO2 and N2O emissions were significantly related to soil dissolved organic carbon, ammonia, nitrate and pH. Our results suggested significant interactions of N fertilization and soil characteristics existed in soil–atmosphere greenhouse gas exchanges, which should be considered in assessing greenhouse gas budgets and scientific fertilization strategies in eucalypt plantations.


Introduction
Forest plantations with fast growing tree species have quickly expanded in recent years to meet the increasing demand for timber and timber products [1]. Eucalypt, an important afforestation tree species, has been introduced to many tropical or subtropical regions [2]. In southern China, eucalypt plantations covered about 3,680,000 hm 2 in 2010 [3]. Nitrogen (N) fertilization is necessary to sustain plantation productivity [4], but it also increases the risk of greenhouse gas emissions [5]. The greenhouse gas emissions have been extensively studied in crop lands and forests [6][7][8][9]; however, greenhouse gas fluxes in introduced eucalypt plantations and their responses to N fertilization have been seldom reported in southern China [10]. This limits our understanding of the effects of eucalypt planting on greenhouse gas emissions in this area.
The effects of N fertilization on soil-atmosphere exchanges of greenhouse gas have been extensively studied, but the results have not been consistent, varying from positive to negative in different studies [11][12][13][14][15][16][17][18] or even in the same study [19,20]. These variations impeded accurate assessment of global greenhouse gas fluxes [21]. It has been reported that the response of soil greenhouse gas fluxes to N addition depends on the N status of ecosystems [15], or is influenced by soil properties, such as dissolved organic matter or ratios of N to other nutrients [17,22,23]. Soil organic carbon (SOC) is a very important soil property and might influence responses of soil greenhouse gas fluxes to N addition through coupling between carbon and N cycles [24,25]. However, the interactions of N fertilization and SOC on soil greenhouse gas fluxes have rarely been studied.
In order to assess soil greenhouse gas fluxes in eucalypt plantations and investigate the interactive effects of N fertilization and SOC level on soil greenhouse gas fluxes, two eucalypt plantations in southern China with different SOC contents were selected as study sites (higher and lower soil organic carbon sites [HSOC and LSOC]), and four N fertilization treatments [no fertilization control (CK), low (LN), middle (MN) and high (HN) level N fertilization] were applied in each plantation. The soil-atmosphere exchanges of greenhouse gas were measured using static chambers and chromatography during May 2013 to April 2015. The early results of the first growing season had been published as communications in Chinese [26,27]. We hypothesized that (1) N fertilization would increase soil CO 2 and N 2 O emissions and decrease CH 4 uptake, and (2) the responses of soil-atmosphere exchanges of greenhouse gas to N fertilization would be larger in HSOC site than in LSOC site.

Site description
This study was conducted in the Dongmen Forest Farm (with the permission of Guangxi Dongmen Forest Farm) (22˚16 0 -22˚30 0 N, 107˚13 0 -107˚59 0 E), which is located in Dongmen county, Chongzuo city, Guangxi province, southern China, with an altitude varying from 140 to 250 masl and an area of 22,000 hm 2 on hilly terrain. This region is characterized by a typical subtropical monsoon climate, with mean annual temperatures of 21.2-22.3˚C and annual rainfall of 1100-1300 mm. Soils in the region are mainly lateritic red soils derived from arenaceous shale with a profile depth of about 80 cm. Soils' pH range from 4 to 5 [3].
Eucalypt (Eucalyptus urophylla × grandis) is the dominant tree species planted in the Dongmen Forest Farm and Guangxi province. In the eucalypt plantations, tree density was about 1400 trees hm -2 and the rotation was 5 years. Prior to eucalypt planting, clear-cutting, fire clearance and reclamation were completed. Then, a base fertilizer (500 g/seedling, N:P: K = 10:15:5) was placed into a 10-cm-deep soil hole and covered with soil. Topdressing was performed once a year during the first 3 years after eucalypt planting (1 st year, 250 g/seedling; 2 nd year, 500 g/seedling; 3 rd year, 500 g/seedling; N:P:K = 10:15:5). The application of herbicide (glyphosate) was performed once a year during the first 3 years after planting and consequently the coverage of understory plants was less than 50%. The understory plants were dominated by Eupatorium odoratum, Rhodomyrtus tomentos and Miscanthus floridulus. The leaf, branch and bark litter were kept in the plantation during the plant growth period, but at harvest time most branch litter was removed and burned.

Experimental design
The aim of this paper was to study the impacts of N fertilization on soil-atmosphere greenhouse gas exchanges in eucalypt plantations with different SOC contents. In order to select two eucalypt plantations with significantly different SOC contents within the Dongmen Forest Farm, soils from 20 eucalypt plantations (1-2 years in age) with similar above ground conditions were collected for physical and chemical analysis. The eucalypt plantation with the highest SOC was selected as the HSOC site and that with the lowest SOC was selected as the LSOC site. The land use histories of the two eucalypt plantations were similar (each represented the second rotation of eucalypt planting after conversion from Pinus to Eucalyptus). The eucalypt in both study sites were planted during May to October 2012 and the tree density was about 1400 tree hm -2 . The tree height (TH), diameter at breast height (DBH) and their increments during May, 2013 to May, 2014 were described in Table 1. Understory coverage was less than 50% due to herbicide application. The initial soil properties were described in Table 2. At each site, 12 plots (10 m × 10 m) were separated by a 5 m wide buffer strip. High, middle, and low level N fertilization treatments (HN: 334 kg NÁhm -2 , MN: 167 kg NÁhm -2 and LN: 84 kg NÁhm -2 ) and a no fertilization control (CK: 0 kgÁhm -2 ), with three replications, were randomly arranged in the 12 plots. Urea formaldehyde was used in this experiment as N fertilizer to simulate the controlled release of fertilizer used in the Dongmen Forest Farm. On May 19-20, 2013 and May 25-30, 2014, urea formaldehyde was fertilized in soil holes (about 10 cm depth) that were about 30 cm away from each tree and covered with soil.
In each plot, three trees aligned in a diagonal pattern were selected. For each tree, gas sampling was performed at four points, a fertilized point and three non-fertilized points (Fig 1). The tree-level soil greenhouse gas fluxes were calculated by averaging the fluxes of the four points and the plot-level fluxes were calculated by averaging the three tree-level fluxes at each plot.

Greenhouse gas flux measurements
Soil-atmosphere exchange of CO 2 , CH 4 , and N 2 O was measured using a static chamber technique and gas chromatography. The static chamber consisted of two parts, a stainless-steel base box (without top and bottom, but with a 4 cm deep × 4 cm wide trough on top of the side wall, length × width × height = 0.4 m × 0.4 m × 0.1 m) and a removable transparent acrylic cover box (without bottom, length × width × height = 0.4 m × 0.4 m × 0.4 m). At the gas sampling location, the base box was permanently inserted directly into the soil about 5 cm, and the cover box was placed on top of the base box and sealed by filling water into the trough during sampling and removed afterwards. On the side wall of each chamber, a fan (10 cm in diameter) was installed to mix the air during sampling and a tube was installed to balance air pressure between the inside and outside of the chamber. The sampling tube was installed on the top wall of each chamber for gas sampling. Gas samples (300 ml) were collected in 500 mL gas sampling bags (Dalian Hede Technologies LTD., Dalian, China) using a gas sampler QC-1 Gas samples were collected between 09:00 and 11:00 once per month for laboratory analysis and calculation of the greenhouse gas fluxes. Samples were analyzed for CO 2 , CH 4 , and N 2 O concentrations using an Agilent 7890A gas chromatograph (Agilent, Wilmington, DE, USA). The gas fluxes were calculated from the changes in gas concentration in relation to time after chamber closure. The calculation was conducted using the following equation [26]: Where F is gas flux (mg m -2 h -1 for CO 2 and μg m -2 h -1 for CH 4 and N 2 O), ρ is gas density under normal conditions (mg m -3 ), V is the volume of the static chamber (m 3 ), A is the area that the static chamber covered, Δc/Δt is changes in gas concentration (Δc) during a certain time (Δt), and T is air temperature (˚C).

Measurement of environmental and soil factors
Air temperature and precipitation in the Dongmen Forest Farm (Fig 2) were monitored using a rainfall recorder L99-YLWS (Shanghai Fotel LTD., Shanghai, China). In

Statistical analysis
An independent-samples t test was used to compare initial soil physical and chemical properties between the HSOC and LSOC sites. Two-way analysis of variance (ANOVA) was used to analyze the impact of N fertilization, SOC, and their interaction on soil properties (August 2014 and August 2015) and greenhouse gas exchanges (annual mean fluxes during May 2013 to April 2014 and during May 2014 to April 2015). When significant F tests were obtained, Tukey's test was used to identify the significance of differences between N treatments in the HSOC and LSOC sites. The relationship between soil greenhouse gas fluxes and soil properties were analyzed using Pearson correlation analysis. All statistical analyses were performed using SPSS 16.0 for windows (SPSS Inc., Chicago, IL, USA).

Annual mean fluxes and seasonal variations of greenhouse gas exchanges in eucalypt plantations
During our experiment, the annual mean fluxes of soil CO 2 , CH 4 , and N 2 O were separately 153-266 mg m -2 h -1 , -55 --40 μg m -2 h -1 , and 11-95 μg m -2 h -1 under different N fertilization treatments at both sites. In the plantations with no N fertilization, soil CO 2 and N 2 O emissions showed significant seasonal variations, which were higher in warm and wet months (June-September) and lower in cold and dry months (December-March) (Figs 3 and 4), but soil CH 4 uptake did not show significant seasonal variation (Fig 5).

Differences in nitrogen-induced greenhouse gas exchanges between the two study sites
The responses of soil CO 2 and N 2 O emissions to N fertilization were significantly different between HSOC and LSOC sites. Two-way ANOVA showed significant interactions of N fertilization and SOC level on soil CO 2 and N 2 O emissions. At the LSOC site, the annual mean soil CO 2 flux was only significantly higher under the HN treatment in comparison to the CK treatment. However, at the HSOC site, soil CO 2 fluxes were significantly higher under LN andHN treatments (2013-2014) or under MN and HN treatments (2014-2015) than under the CK treatment (Fig 6a and 6b) HN treatment (2014-2015), the annual mean flux of N 2 O was significantly higher in the HSOC site than in the LSOC site (Fig 6e and 6f).

Soil properties and their relationship with greenhouse gas exchanges
In August 2013, after 3 months of N fertilization, soil DOC, NH 4 + -N and NO 3 --N were measured (Table 3). Soil DOC concentrations showed an increasing trend with the amount of fertilized N, but the differences were not significant. Soil   (Table 3). Soil pH showed a decreasing trend with the amount of fertilized N, but the differences were not significant. Soil  (Table 4).

Discussion
The annual mean soil-atmosphere greenhouse gas fluxes in our study fall in the range of greenhouse gas fluxes reported by previous studies [19,20,[28][29][30]. Our result suggested that N fertilization stimulated soil CO 2 emissions (Fig 6a and 6b). N fertilization might stimulate soil CO 2 emissions through two pathways. The first way is by stimulating light organic carbon decomposition and increasing soil DOC concentration (Table 3), resulting in higher soil heterotrophic respiration and higher soil CO 2 emission fluxes. The second way is by stimulating plant root growth and activity, resulting in higher autotrophic respiration [4,31]. This was not tested in this study due to lack of root data.
A higher stimulative effect of N fertilization on soil CO 2 emissions was found at the HSOC site. On the one hand, the higher SOC generally related to higher soil available carbon, microbial biomass, and microbial activities [32], which would stimulate soil respiration [33,34] and increase soil CO 2 emissions in response to N addition. In this study, soil DOC concentrations were 407-477 mg kg -1 in the LSOC site and 465-509 mg kg -1 in the HSOC site, which might explain the higher priming effect of N fertilization on soil CO 2 emission in HSOC site. On the other hand, N fertilization could inhibit soil respiration by decreasing soil pH [35]. Our results showed that soil pH significantly decreased with increases in fertilized N  (Table 3), which likely reduced soil respiration and counteracted the stimulation effect of N fertilization. Changes in soil pH induced by N fertilization could be influenced by soil cation exchange capacity, which usually correlated with SOC content with positive relationship [36,37]. With increases in fertilized N, soil pH decreased from 4.51 to 4.20 at the LSOC site and from 4.44 to 4.27 at the HSOC site, suggesting that soils with higher organic carbon could buffer N-induced decreases of soil pH. Thus, the significantly higher soil CO 2 emission fluxes under the HN treatment at the HSOC site might be the result of higher soil DOC concentrations and smaller decreases in soil pH.
N fertilization significantly decreased CH 4 uptake, which was consistent with previous studies [12,23]. The inhibition of N fertilization on soil CH 4 uptake might be caused by competitive inhibition of CH 4 monooxygenase [38], toxic inhibition by hydroxylamine and nitrite produced during nitrification process [39], high osmotic pressure caused by high N ammonia and nitrate concentration [40], and toxicity caused by decreased soil pH [41]. We found that soil NO 3 --N concentration increased and soil pH decreased with increases in fertilized N (Table 3), which explained the decreased soil CH 4 uptake in the MN and HN treatments.
N fertilization significantly increased soil N 2 O emission fluxes, which might be explained by higher N availability caused by N fertilization (Table 3). That result was consistent with Zhang et al. [19], who studied the responses of N 2 O emissions to simulated N deposition in three tropical forests in southern China and found that N fertilization increased soil N 2 O emissions through increasing N availability and stimulation of nitrification and denitrification processes. The increase of soil N 2 O emission fluxes caused by N fertilization was significantly higher at the HSOC site (from 12 μg m -2 h -1 to 95 μg m -2 h -1 ) than at the LSOC site (from 18 μg m -2 h -1 to 85 μg m -2 h -1 ) (Fig 6e and 6f). That might be explained by differences in soil carbon availability between the two sites. Denitrification, an important soil process producing N 2 O, is generally restricted by substrate availability, energy sources, and presence of anaerobic environments. The higher DOC concentrations at the HSOC site could offer more energy and electron acceptors for denitrification [42], enlarging the soil N 2 O emission responses to N fertilization. Additionally, higher respiration at the HSOC site might exhaust soil O 2 and create an anaerobic environment, which would be suitable for denitrification and N 2 O production [43].
Our results showed that the responses of soil-atmosphere CO 2 and N 2 O exchanges to N fertilization were different at the LSOC and HSOC sites. So, when assessing N-induced soil greenhouse gas emissions, soil characteristics, such as SOC, should be considered in attempt to reduce the variation of emission factors [21]. Higher SOC is generally related to higher soil fertility and plant productivity [44], but it also stimulates the responses of soil CO 2 and N 2 O emissions to N fertilization. Thus, in order to reduce the risk of N-induced greenhouse emissions, N fertilizer should be applied according to soil properties.

Conclusion
Our study found that the annual mean fluxes of soil CO 2 , CH 4 , and N 2 O in eucalypt plantations in Guangxi, China were separately 153-266 mg m -1 h -1 , -55 --40 μg m -1 h -1 and 11-95 μg m -2 h -1 . N fertilization significantly increased soil CO 2 and N 2 O emissions and decreased soil CH 4 uptake. The stimulative effects of N fertilization on soil CO 2 and N 2 O emissions were significantly higher at the HSOC site than at the LSOC site. Thus, the interaction of N fertilization and soil characteristics should be considered in assessing greenhouse gas budgets and scientific fertilization in eucalypt plantations.