Co-Regulations of Spartina alterniflora Invasion and Exogenous Nitrogen Loading on Soil N2O Efflux in Subtropical Mangrove Mesocosms

Both plant invasion and nitrogen (N) enrichment should have significant impact on mangrove ecosystems in coastal regions around the world. However, how N2O efflux in mangrove wetlands responds to these environmental changes has not been well studied. Here, we conducted a mesocosm experiment with native mangrove species Kandelia obovata, invasive salt marsh species Spartina alterniflora, and their mixture in a simulated tide rotation system with or without nitrogen addition. In the treatments without N addition, the N2O effluxes were relatively low and there were no significant variations among the three vegetation types. A pulse loading of exogenous ammonium nitrogen increased N2O effluxes from soils but the stimulatory effect gradually diminished over time, suggesting that frequent measurements are necessary to accurately understand the behavior of N-induced response of N2O emissions. With the N addition, the N2O effluxes from the invasive S. alterniflora were lower than that from native K. obovata mesocosms. This result may be attributed to higher growth of S. alterniflora consuming most of the available nitrogen in soils, and thus inhibiting N2O production. We concluded that N loading significantly increased N2O effluxes, while the invasion of S. alterniflora reduced N2O effluxes response to N loading in this simulated mangrove ecosystem. Thus, both plant invasion and excessive N loading can co-regulate soil N2O emissions from mangrove wetlands, which should be considered when projecting future N2O effluxes from this type of coastal wetland.


Introduction
Although nitrous oxide (N 2 O) contributes only 5% to estimated global warming potentials [1], its global warming strength is 265 times more powerful than CO 2 over a 100-year time frame [2]. N 2 O has also contributed to destruction of ozone in the stratosphere [3]. During the past few decades, the global mean atmospheric concentration of N 2 O increased from 270 ppb in pre-industrial times to 324 ppb in 2011 [2]. The rapid increase in atmospheric N 2 O concentrations has gained much attention in quantifying N 2 O effluxes from various sources.
Coastal wetlands, such as mangroves in tropical and subtropical regions have been recognized as a major marine source of atmospheric N 2 O [4,5]. In general, N 2 O effluxes from wetlands are related to several biological processes, mainly including nitrification (ammonium oxidation and nitrifier denitrification) [6], denitrification and nitrate-ammonification [7]. These processes can be affected by many abiotic and biotic factors, such as soil temperature, oxygen level, and substrate availability (bioavailable carbon, ammonium and nitrate) in the soil [8,9]. In addition, it is widely recognized that vegetation type can significantly affect N 2 O effluxes [10]. Plants living in wetland environments can supply oxygen to the rhizosphere via aerenchymous tissue of plants, which creates oxidized microzones surrounding roots and rhizomes and favors nitrification and nitrification-denitrification reactions, facilitating N 2 O production [11][12][13][14][15]. Meanwhile, root exudates and debris can be used as an organic carbon source for microbes. The high activity of microbes could increase soil N mineralization [16] and promote N 2 O production. Therefore, the change in vegetation type, such as plant invasion, may have a potential effect on N 2 O effluxes.
Spartina alterniflora, a C 4 salt marsh grass native to the east coast of the USA, was introduced to China in 1979. It has grown vigorously in China and spread over coastal wetlands since that time. During the last two decades, it has aggressively invaded habitats of native mangroves on the southern coast of China, especially in mangrove seedlings established areas [17][18][19], influencing the native species composition of mangrove ecosystems. As a result, it would alter soil available inorganic nitrogen (N) content and microbial activity [20,21], which would further change N 2 O production and effluxes. So, it is essential to understand the change of N 2 O effluxes from mangrove wetlands following S. alterniflora invasion.
Globally, many mangrove ecosystems are also under significant threat from N pollution because of human activities, such as sewage discharge and surrounding aquaculture operation [22]. Exogenous N loading has greatly changed functions of many mangrove wetlands in tropical and subtropical regions, from pristinely oligotrophic [23] to eutrophic [24]. Usually, pristine mangrove soils, in N-limited environments, show low or negative N 2 O effluxes [25,26]. Exogenous N loading often stimulate N 2 O effluxes by directly stimulating nitrification and denitrification in the soil or sediment [27,28]. For example, the Futian mangrove swamp in China, an area with intense human activities, had the highest N 2 O effluxes compared to three other mangrove swamps with slightly anthropogenic disturbance [27]. Similarly, the effluxes of N 2 O from the Puerto Rico mangrove swamp in the northeastern Caribbean Sea, where anthropogenic activities increase nitrogen loading to mangrove sediments, was much higher than those reported previously for intertidal estuarine sediments [29].
In the coastal wetlands, N loading from surrounding aquaculture ponds are always periodic. For example, during the common process of complete pond sediment cleaning, large quantities of sediment with high nutrient levels are washed out and discharged into the adjacent mangrove swamps in a very short period of time [30]. N 2 O effluxes may vary greatly over time because of dynamic changes of soil inorganic N concentration after such exogenous N loading. In order to accurately estimate N loading effect on N 2 O effluxes, attention should be paid to dynamic changes of N 2 O effluxes after a pulse of exogenous N loading. However, to our knowledge, there has been limited research focusing on the dynamic change of N 2 O effluxes after an N-pulse loading in a wetland (but see [31]).
Although the previous research has pointed out that N 2 O effluxes in mangroves have positive response to excessive N loading, the response ability would be different between vegetation types [32]. After N loading, the divergent growth rate between various species meant dissimilar uptake capabilities of ammonium nitrogen (NH 4 + -N) and nitrate nitrogen (NO 3 --N), which are the important substrates participating in nitrification and denitrification processes. The difference in the competition between microorganisms and plants for N may lead to a differential in soil N 2 O production [33,34]. Thus, to gain an insight into the relationship between excessive N loading and N 2 O effluxes from mangrove wetlands after S. alterniflora invasion, the varying physiological responses of various plant species should also be considered.
In order to understand individual and possible interactive effects of S. alterniflora invasion and exogenous N loading on N 2 O effluxes from mangrove ecosystems, we conducted a mesocosm experiment with three vegetation types including the monocultures of K. obovata (a common native mangrove species in China) and S. alterniflora and their mixture and two N loading treatments (with or without N loading). The aims of the study were (1) to determine the dynamic change of N 2 O effluxes after a pulse N loading, (2) to test whether and how the invasion of S. alterniflora alters N 2 O effluxes, and (3) to investigate the interactive effect of plant invasion and N loading on mangrove N 2 O effluxes. We hypothesized that N loading will increase N 2 O effluxes from both mangrove and salt marsh mesocosms, but the invasive S. alterniflora mesocosms would have a lower increment of N 2 O effluxes after N loading due to larger N uptake compared with the native mangrove mesocoms.

Experimental design
The tide-system mesocosm experiment was conducted in a roof greenhouse at the Graduate School at Shenzhen, Tsinghua University, China (22°59 0 N, 113°97 0 E). The mesocosm system consisted of 18 cement tanks (2.40 m × 1.10 m × 0.50 m in volume) as experimental mesocosms and 2 cement tanks (7.50 m × 1.00 m × 0.70 m in volume) as seawater reservoirs (Fig 1). One reservoir was connected to all the mesocosms without N addition treatment and the other was connected to all the mesocosms with N addition treatment by pipelines of water inputting (used during tide flooding periods) and water outputting (used in tide falling periods). The artificial seawater at a salinity of 14 g l -1 , representing an average salinity in a typical mangrove environment of southern China, was prepared by dissolving natural sea salts in tap water. Submersible pumps and timers were installed to simulate diurnal tides for all the mesocosms. For the transplanted plants to establish and acclimate, each mesocosm was conditioned under the tidal regime with one tidal cycle a day, 12 h high tide and 12 h low tide for 2 months before starting the experiment on August 31, 2012. After the formal start of the system operation, artificial seawater was pumped from the reservoir to each mesocosm to a depth of 5 cm between 14:00 and 20:00 (local time) to simulate high tide, then seawater was drained back by gravitational force to the reservoir during the low tide period. Tidal flux took about 30 minutes from the initiation of the ebb or flood tide. Tidal water in each reservoir was rotated for about 15 days as a water cycle, then drained and replenished by freshly prepared artificial seawater. In this way, we mimicked field conditions because daily tidal dynamic change might greatly change nitrification and denitrification potential, allowing us to determine whether the effects were robust with respect to natural perturbations.
Each mesocosm was filled with fresh mangrove soils. On June 5, 2012, the mangrove soils were collected from the mudflat connected to the Futian Mangrove Natural Reserve (22°51'N, 113°96'E), a typical mangrove swamp along the southern coast of Shenzhen. We collected the soils after receiving the permission of the Futian Mangrove Nature Reserve in this research. The substrate was sandy, consisting of 73.7% sand, 14.8% silt and 11.6% clay. After stones, benthic animals and plant residues in the soil were carefully removed, the soils was then used to fill each mesocosm to a depth of 30 cm (Fig 1). In the present study, K. obovata was chosen because it was the most dominant species among the eight true mangrove species in South China [35]. The mature propagules of K. obovata and ramets of S. alterniflora were collected from the mangrove forest in Zhangjiangkou Mangrove Nature Reserve in Yunxiao, Fujian (23°56'N, 117°25'E). We collected the plants after receiving the permission of the Zhanjiang Mangrove National Nature Reserve in this research. The propagules of K. obovata were cultivated in the sand for 2 months before transplantation to the mesocosms. Then healthy K. obovata seedlings with approximately equal height of 15 cm (usually with 3-5 green leaves) were selected for this study. Meanwhile, the young ramets of S. alterniflora of approximately equal size with four true leaves were selected for this study. Each K. obovata seedling and ramet of S. alterniflora covered an area of 0.15 m × 0.15 m and 0.25 m × 0.25 m, respectively. This mimics typical seedling densities of these two species according to our observations in the field.
In this experiment, three vegetation types represented three stages of S. alterniflora invasion to K. obovata: (i) monoculture of K. obovata (no invasion), (ii) mixture of K. obovata and S. alterniflora (partial invasion), and (iii) monoculture of S. alterniflora (complete invasion) ( Fig  1). All vegetation types were tested in the absence and presence of nitrogen addition. Ammonium chloride (NH 4 Cl) of 450 mg was added to synthetic seawater of 3000 ml in the reservoir connected to mesocosms with N addition treatment one day before the start of water cycle (i.e. one day before tide rising of the first day) in order to make NH 4 Cl dissolve thoroughly. During a typical 15 day water cycle period, no more NH 4 Cl was added into the reservoir. Because the N of water may be lost by transformation by microbes and volatilization in form of ammonia gas in the reservoir during low tide periods, we measured the actual rates of N loading, which were shown in Fig 2. The N levels we applied are within the range of sewage and wastewater and sludge from aquaculture ponds [36,37].

Measurements of N 2 O effluxes
For the results reported in this paper, we selected a typical 15 day water cycle period. Artificial seawater pumping started at 14:00-20:00 on June 18, 2013 (Day 1) from a new pool of tidal water with desired salinities and N levels specified above. We started the collection of gas samples on the day after a new seawater change (i.e. on June 19, 2013) to investigate the effect of N loading on N 2 O effluxes during daytime of tide falling periods. During this water cycle period, the N 2 O effluxes in each mesocosm were quantified using a static chamber-gas chromatograph method on the day 2, 4, 6, 8, 10, 14 and 16. Each static chamber, which was made of PVC, covered a soil area of 0.025 m 2 and had an internal volume of about 4000 ml. The open end of the chamber was inserted 3-5 cm into the soil with the air-tight valve open. Gas sampling was carried out 30 minutes after the chamber was inserted. Generally, five 8 ml gas samples of chamber air were collected by passing a hypodermic needle attached to a 10 ml glass syringe and was injected into pre-evacuated vials for laboratory analysis. Meanwhile, to minimize any effect of diurnal variation in N 2 O effluxes, samplings were conducted at roughly the same time of the day for the same mesocosm between 9:00 and 11:00. The deployment time was set to 45-90 min with sampling at 15-30 min intervals depending on the N 2 O efflux rates.
The N 2 O concentration of gas samples collected was analyzed within 24 h after sampling using a gas chromatograph (GC, Agilent 6820, Agilent Technologies, USA) equipped with an electron capture detector (mECD) and a HP-Plot/column (J&W GC Columns, Agilent Technologies, USA). The temperatures of the injector, column and detector were 50°C, 50°C and 300°C, respectively, with a carrier gas (N 2 ) flow rate of 15 ml min -1 . The relative standard deviation (RSD) of replicate standard measurements was 0.8%. During gas measurements, standard samples were analyzed with every 10 samples of determination to ensure that each sample run maintained a RSD below 3% in 12 h. N 2 O concentrations were quantified by comparing the peak areas of samples against the standard curves. N 2 O efflux rates were calculated from the linear change of the measured gas concentrations in the chamber headspace with known headspace volume and sampling time. Estimates of cumulative N 2 O emissions for each mesocosm were based on linear interpolation, with the sum of cumulative experiment period emissions representing one water cycle.

Sampling and analyses of water and soil
Everyday input water samples from each reservoir were collected before the start of tide flooding and output water samples from each mesocosm were collected before the start of tide falling during the experiment water cycle period. The water samples were transported to the laboratory and analyzed immediately. The samples were filtered through a 0.45 μm filter and analyzed for NH 4 + -N, NO 3 --N, nitrite nitrogen (NO 2 --N). We used standard procedures and methods for the determination, including the Sodium salicylate-sodium hypochlorite Method for NH 4 + -N, the UV Spectrophotometer Method for NO 3 --N and the α-Naphthylamine Method for NO 2 --N. Total inorganic nitrogen (TIN-N) was calculated by summing NH 4 + -N, NO 3 --N and NO 2 --N [38]. Rates of N loading were calculated by multiplying N concentration of input water by the area and the flooding depth (5cm) of each mesocosm. Three soil cores (0-15 cm) were collected around the chamber using hand-held PVC corers from each mesocosm at day 4 after gas sampling. Soil total carbon and nitrogen contents were measured by the Elementar (Vario ELⅢ, German). The NH 4 + -N and NO 3 --N contents in the KCl (2 M) extracts were also determined by the methods above. Both soil nitrification potential activity (PNA) and denitrification potential activity (PDA) were measured according to the method of Chen et al. (2012) [39].

Measurements of plant biomass
K. obovata biomass, measured on July 5, 2013, was evaluated by a non-destructive allometric technique [40]. S. alterniflora biomass was measured also on July 5, 2013 but was evaluated by a nondestructive height-weight method [41]. Plant samples were analyzed for N concentration with a Elementar elemental analyzer (Vario ELⅢ, German) and the total N accumulation was calculated by multiplying the N concentration by dry weight.

Statistical analyses
The main and interactive effects of sampling time, vegetation type and nitrogen addition treatment on N 2 O effluxes were tested with a parametric three-way analysis of variance (ANOVA). A one-way analysis of variance (ANOVA) performed to examine the significant differences of N 2 O effluxes between vegetation types for each period. One-way ANOVA was also used to analyze the differences in parameters of soil properties and plant traits between the treatments, followed by the least significant difference (Duncan) test at P < 0.05. Linear regression analyses were performed to evaluate the relationships of NH

Rate of N loading
In the treatment without N addition, NO 3 --N was the dominant dissolved N form of the input water and contributed 70.08% of TIN-N load. The rate of NO 3 --N loading declined gradually with time. With N addition, after the refreshment of tidal water, the rate of NH 4 + -N loading in the mesocosms decreased with time from 1.54 to 0.002 g m -2 , However, the rate of NO 3 --N loading in the mesocosms increased to a peak by the day 8 (0.21 mg m -2 ), and then decreased gradually to 0.04 mg m -2 . A similar variation pattern was observed for nitrite (NO 2 --N), with the highest rate (0.06 mg m -2 ) also on day 8. The rate of TIN-N loading generally decreased with time (Fig 2).

Soil properties and plant traits
In treatments without N addition, concentrations of soil total carbon, nitrogen, NH 4 + -N and NO 3 --N were not significantly affected by plant community composition, except that the NH 4 + -N concentration in mixture mesocosms was significantly lower than that in S. alterniflora mesocosms and K. obovata mesocosms. In treatments with N addition, NH 4 + -N concentrations differed among the three plant communities, ranging from 2.25 mg kg -1 in the S. alterniflora mesocosms, and 3.02 mg kg -1 in the mixture mesocosms to 3.27 mg kg -1 in the K. obovata mesocosms. PNA also varied with plant species. N loading increased the soil total carbon, NH 4 + -N and PNA significantly. However, soil total N, NO 3 --N and PDA were not significantly different between the N addition and without N addition mesocosms (Table 1).
Without N addition, no significant differences in aboveground biomass were found among vegetation types, while the belowground biomass in S. alterniflora mesocosms was 1.02 kg m -2 significantly higher than that in K. obovata mesocosms. N addition increased total biomass in the S. alterniflora, K. obovata and their mixture mesocosms by 55.26%, 176.76% and 148.78%, respectively. Under N addition condition, above-and below-ground biomass of S. alterniflora were evidently higher than that of K. obovata mesocosms. However, the ratio of aboveground to belowground biomass was 3.21 in K. obovata mesocosms, significantly higher than 0.99 in S. alterniflora mesocosms. Therefore, S. alterniflora was more productive compared to K. obovata (Table 1).

N 2 O effluxes
Without N addition, the N 2 O effluxes from K. obovata, the mixture and the S. alterniflora mesocosms ranged from 0.27 to 0.93, from 0.52 to 1.11 and from 0.32 to 0.70 μmol m -2 h -1 in the, respectively (Fig 3). N addition significantly increased mean N 2 O effluxes ( Table 2). With N addition, the N 2 O effluxes ranged from 0.56 to 7.17, from 0.23 to 4.14 and from 0.19 to 4.10 μmol m -2 h -1 in the K. obovata, the mixture and the S. alterniflora mesocosms, respectively (Fig 3). N 2 O effluxes for all three vegetation types generally decreased with sampling time. In addition, there was a significant interaction of plant species × N fertilization on N 2 O effluxes ( Table 2). We further analyzed N 2 O effluxes over the 15-day period by dividing the sampling time into three periods according to the N loading rate and main species of N in the input water with N enrichment as shown in Fig 3. NH 4 + -N was the major species of N in the first period during day 1 to 6 with high N loading (TIN-N loading rates ranging from 0.51 to 1.87 g m -2 ); NO 3 --N was the major species of N in the second period during day 7 to day 13 with modest N loading (TIN-N loading rates ranging from 0.12 to 0.39 g m -2 ) and in the last period total N loading rates have decreased to background level during day 14 to day 16 (TIN-N loading rates ranging from 0.047 to 0.057 g m -2 ) (Fig 2).Without N addition, the N 2 O effluxes from K. obovata, the mixture and the S. alterniflora mesocosms did not vary significantly between vegetation types during all three periods, while there were trends that N 2 O effluxes were higher from mixture mesocosms than that from K. obovata and S. alterniflora mesocosms during all three periods. With N addition, compared to that from K. obovata mesocosms, the mean N 2 O effluxes were significantly lower from the S. alterniflora mesocosms in period 1 and 2 (Fig 4). However, in period 3, there were no significant difference among the vegetation types.

Relationship between N 2 O effluxes and N loading
When integrating all data either with or without N addition for analysis in each vegetation type respectively, N 2 O effluxes were found to significantly correlate to the rate of TIN-N, NH 4 + -N, tionship between rate of TIN-N loading and N 2 O effluxes for S. alterniflora and mixture mesocosms were significantly lower than that for K. obovata mesocosms by student's t-test (P < 0.05) (Fig 5). Thus, the increase rate of N 2 O effluxes by N addition was much lower in the S. alterniflora and mixture than native K. obovata mesocosms.

Discussion
Our results from this mesocosm experimental study showed that a pulse loading of exogenous ammonium nitrogen resulted in a stimulatory effect on N 2 O effluxes but the effect gradually diminished over time and that the invasion of S. alterniflora could mitigate N 2 O effluxes following N loading. In addition, there was a significant antagonism effect between plant invasion and exogenous N loading on mangrove N 2 O effluxes. Our findings offer new insights into how plant invasion and N loading modulate N 2 O effluxes from mangrove wetlands.

Pulse effect of N addition on N 2 O effluxes
Compared to the treatment with N addition, we found that the N 2 O effluxes were relatively low (<1.50 μmol m -2 h -1 ) in the treatment without N addition in the K. obovata, S. alterniflora and their mixture mesocosms, which was consistent with the conclusions from the previous studies that natural coastal wetlands were weak sources for atmospheric N 2 O, such as undisturbed mangrove and salt marshes ecosystems [26,27]. However, we did not observe any N 2 O sink in either pure S. alterniflora stand or the mixture of S. alterniflora and native mangrove seedlings, which was different from the results of Yuan, et al. (2014) [26]. A likely explanation for this difference is that in our study soil nitrate levels never reached the very low level (< 1 mg N kg -1 ) [42]. N 2 O effluxes were significantly increased by the N loading in both mangrove and Spartina mesocosms, which was also found in salt marshes [34]. Although N 2 O emissions accounted for a very low proportion of total N loading, ranging from 0.23-0.47% for the mesocosms with N addition treatment and from 2.22 to 3.55% for those without N loading (S1 Table), we could not neglect N loading effect on N 2 O emission, considering 265 times higher potential warming strength than CO 2 . Previous research also found relative lower proportions (ranging from 0.05-0.48% at N loading rate of 1.4 g m -2 in field experiment of salt marsh coastal wetland) [31]. The stimulating effect of N loading on N 2 O effluxes results might be attributed to the fact that enzyme activities and the microbial population sizes including nitrifiers and denitrifiers in surface sediment had a dramatic increase after N addition [43,44] and N loading supply the source of substrate for nitrification and denitrification in the soil or sediment. However, for the exgenous N loading, N 2 O effluxes may be underestimated or overestimated if they were not measured continuously after the N loading, as shown in this study that the stimulatory effect of N loading on N 2 O effluxes decreased gradually over time. A contributing factor to this result is that N 2 O fluxes were significantly related to N loading rates and these N loading rates usually decreased gradually after a pulse of nitrogen loading due partly to assimilation by wetland plants, transformation by microbes, and adsorption to sediments. So, N addition significantly increased the mean N 2 O effluxes in the first (Day 0-5) and second period (Day 6-10) in S. alterniflora mesocosms, K. obovata mesocosms and their mixture, but had no effect in the third period (Day [11][12][13][14][15]. This result suggests that discretely or infrequently measuring is inadequate to accurately predict the behavior of exotic N-induced response of N 2 O effluxes. Meanwhile, with N addition, we found the mean N 2 O effluxes in the first period were higher than in the second period. We infered that, except for the higher N loading rates in the first period, the higher N 2 O effluxes may also be attributed to varying reponses of N 2 O effluxes to different N species. Previous study pointed out that N 2 O effluxes were larger when NH 4 + -N was the dominant substrate and nitrification was the main source of nitrous oxide compared to the condition when nitrate and nitrite predominate [29]. In our study, major nitrogen species of input water were transformed from ammonium in the first period to nitrate in the second period with time, which could partly explain the reason for the different rates of N 2 O effluxes between the two periods. Therefore, in order to get more accurate results especially in coastal wetlands which are frequently affected by tide, we should pay more attention to sampling time, daily N loading rates and N species of input water after N pulse addition. In addition, we added ammonium into seawater as the initial nitrogen species to assess the anthropogenic effects on N 2 O effluxes from coastal wetland, as it constitutes a major form of human inputs such as sewage and wastewater and sludge from aquaculture ponds. Future studies should also investigate the effects of other N species such as NO 3 --N on N 2 O effluxes.

Effects of plant invasion on N 2 O effluxes
Without N addition, there were no significant difference of N 2 O effluxes between the invasive S. alterniflora and native K. obovata mesocosms and more productive S. alterniflora did not result in higher N 2 O effluxes. This finding implied that the invasion of S. alterniflora played a limited role in mediating N 2 O effluxes from mangrove in N-limited condition. Previous research work have shown that fast-growing species such as S. alterniflora, a salt marsh species, increased N mineralization of soil to a greater extent than more conservative species such as K. obovata, a mangrove forest species [16,45]. However, it has also been found that the rate of net N uptake was four-to six-fold higher for faster growing species than slower growing plant species (e.g., Poorter et al. (1991) [46]). Considering the high growth and high root uptake capacity of S. alterniflora [47], the mineralized N in soil could be largely consumed by S. alterniflora, which would result in no buildup of the soil mineral N pool. Therefore, without N addition, no difference in N 2 O effluxes between S. alterniflora and K. obovata mesocosms could be attributed to the similar concentration of soil N pool. Meanwhile, we found significant antagonism effects of plant invasion and nitrogen loading on soil N 2 O effluxes. With N addition, the mean N 2 O effluxes from the invasive S. alterniflora were lower than that from native K. obovata mesocosms. In addition the increase rates of N 2 O effluxes by N addition were also lower in the S. alterniflora than native K. obovata mesocosms, indicating the invasion of S. alterniflora reduced N 2 O effluxes response to N loading in this simulated mangrove ecosystem. This result may be attributed to a crucial effect of vegetation type on the quality of N 2 O effluxes. On one hand, as important substrates participating in the processes of nitrification and denitrification by microbes [48], NH 4 + -N and NO 3 --N would be absorbed by vegetation, which may inhibit N 2 O production due to the competition between the plant and microorganism for soil inorganic N. In our study, compared to mangrove mesocosms, higher biomass and higher uptake of N in S. alterniflora mesocosms were found, which might mean higher photosynthetic capacity, leading to stronger competition ability for N source and lower soil inorganic N content for N 2 O production by microbes. On the other hand, in wetland, plants with aerenchyma can supply oxygen to the rhizosphere [12,13], which can form oxidized zone in the rhizosphere that stimulates nitrification and nitrificationdenitrification processes [15,49], therefore facilitating N 2 O effluxes. Previous researches have shown that stronger oxidation activity in the rhizosphere of mangrove species like K. obovata than salt marsh species such as S. alterniflora [50,51]. Indeed, we found significant higher PNA in K. obovata mesocosms compared to S. alterniflora mesocosms. Therefore, we inferred that the higher N 2 O effluxes in the K. obovata mesocosms than in S. alterniflora mesocosms could be partly attributed to the stronger oxidation in the rhizosphere of mangrove species. The mixture mesocosms of S. alterniflora and K. obovata, reflecting partly invasion of S. alterniflora, had higher plant richness compared to the monoculture K. obovata or S. alterniflora. With N addition, both partly and complete invasion (in monoculture S. alterniflora and mixture mesocosms) reduced N 2 O effluxes from mangrove soil in the first period. However, only complete invasion of S. alterniflora reduced N 2 O effluxes from mangrove soil in the second period. Moreover, both complete and partly invasion did not reduce N 2 O effluxes from mangrove soil with N addition in the third period and without N addition for all three periods. In conclusion, S. alterniflora invasion could reduce N 2 O effluxes regardless of invasion degree under relatively higher nitrogen availability, but no relief effect of S. alterniflora invasion on N 2 O effluxes occurred under N-limiting condition. The results indicated that N addition can improve the effect of partly S. alterniflora invasion on N 2 O effluxes in mangrove, perhaps because N greatly increased S. alterniflora growth and competition with mangrove seedlings [52], leading to exhibition of S. alterniflora characteristic in mixture mesocosms with N addition.
Although substrate condition and results of our controlled mesocosm experiment were similar to field experiment [27], cautions must be exercised when extrapolating results from controlled mesocosm studies to field-scale processes.
Conclusions N 2 O effluxes were significantly increased by the N loading in the K. obovata, the S. alterniflora and their mixture mesocosms, but the stimulatory effect of N loading on the N 2 O effluxes decreased gradually with time and dispeared when the N loading rates returned to the background level. The lower N 2 O effluxes and a weaker response to N addition in the S. alterniflora mesocosms than the K. obovata mesocosms were partly due to significantly higher growth of S. alterniflora, which could have led to lower increment of available N in the sediments for N 2 O production; and also due to higher oxidation capacity in the rhizosphere of K. obovata, which stimulated nitrification and nitrification-denitrification processes. Thus, S. alterniflora invasion into mangrove habitats could reduce N 2 O effluxes in case of amount of exogenous N loading, such as the input of sewage and wastewater and sludge from aquaculture ponds. These findings should be considered when projecting future N 2 O effluxes from global mangrove wetlands under the influences of both biological invasion and increasing exogenous N loading.