Investigating effect of environmental controls on dynamics of CO2 budget in a subtropical estuarial marsh wetland ecosystem

In this study, we quantified the ecosystem-scale CO2 exchange of two different but typical low-latitude vegetation types, para grass and reed, in a subtropical wetland ecosystem by integrating flux observation with the parameterization of environmental variables. In addition, we explored how seasonal dynamics of environmental factors affected variations in CO2 budget. The results suggest that gross primary production (GPP, in the order of 1700 gC m−2 yr−1) of CO2 was higher in this site than in previous studies of northern peatlands and estuarial wetlands because of the direct effect of environmental factors. Temperature and radiation had a larger effect than water status (soil moisture content and vapor pressure deficit) on GPP for the two low-latitude ecosystems, which differ from the results for high-latitude regions. Environmental variables had a strong but different impact on the CO2 budget for para grass and reed areas. This diversity led to different potential shifts and trends of biomass accumulation and distribution of these two typical low-latitude vegetation types under different scenarios of environmental change. The findings from this study can sufficiently provide quantitative understanding of CO2 budgets in low-latitude wetlands.


Introduction
Concerns related to regional and global environmental change have stimulated much interest in investigating the control mechanisms on potential shifts of carbon (C) exchange over different ecosystems. Wetlands cover only 6%-7% of the Earth's surface (Lehner and Döll 2004), but the C storage is estimated to be up to 450 Gt C or approximately 20% of the total C storage in the terrestrial biosphere (Gorham 1991, Maltby andImmirzi 1993). A recent report by Intergovernmental Panel on Climate Change stated that coastal wetlands are highly sensitive to climate change, and one of the greatest challenges in understanding the functioning of estuarine ecosystems is to accurately estimate their C storage and budget (Nicholls et al 2007).
On the global scale, wetlands are found over different latitudes under varying climate conditions and have a unique role in regulating global biogeochemical cycles (Reddy and DeLaune 2008). The characteristics of worldwide wetland ecosystems vary widely because of differences in environmental factors and human disturbance. In recent years, despite some studies on C storage and budget in wetland ecosystems, most were of northern peatlands (Leverenz 1987, Battin et al 2009. Northern peatlands mainly function as long-term sinks for atmospheric carbon dioxide (CO 2 ) and sources of atmospheric methane (CH 4 ), and their net primary productivity is mainly controlled by the environmental factors and the availability of nutrients (Alm et al 1997, Friborg et al 2003, Sottocornola and Kiely 2005, Dunn et al 2007. In contrast, estuarine wetland ecosystems are important for C storage for terrestrial ecosystems and are very productive because of the regular input of nutrients (Sverdrup et al 2002, Thornton andMcManus 1994). Several recent studies concluded that although the global estuarine area is relatively small, its CO 2 degassing flux is as large as the CO 2 uptake by the continental shelf, ranging from 0.4 to 0.6 PgC yr −1 and that both flux terms are substantial in terms of global CO 2 flux (Borges 2005, Cai et al 2006, Chen and Borges 2009). The range of CO 2 flux reflects the heterogeneity and complexity of these highly active biogeochemical environments at the interface between land and ocean. To date, regions of estuarine CO 2 flux studies are mostly in Europe and eastern North America (Laruelle et al 2010); little attention has been paid to estimating CO 2 flux over low-latitude tropical and subtropical estuarine wetland ecosystems.
In addition to determining C budget, quantifying the ecosystem response to seasonal variations of environmental factors is an important issue. Several studies have addressed how the C cycle in peatlands react to seasonality in middle to high-latitude ecosystems (Howes et al 1985, Roulet et al 1992, Moore et al 1998, Ågren et al 2007. A recent study of an estuarial reed ecosystem in northern China found strong seasonal variation in the diurnal course of net ecosystem exchange (NEE) (Zhou et al 2009). A substantial proportion of this variation could be explained by environmental variables; indeed, soil water content (SWC) and temperature had greater effect on gross primary production (GPP) than photosynthetically active radiation (PAR) and vapor pressure deficit (VPD) in this study site (Zhou et al 2009). However, most studies related to environmental controls were conducted in high-latitude ecosystems, and few have investigated estuarial wetlands in tropical or subtropical areas.
Low-latitude wetlands have a broad array of ecosystem services. The ecosystems often show high productivity and feature a unique assembly of aquatic and terrestrial biodiversity. Because of the great influence of tropical and subtropical wetlands on global C cycling, considerable scientific efforts have been invested in quantifying wetland C storage, turnover, hydrologic exports and C interchange between wetland soils and the atmosphere (Mitsch et al 1988, Zhuang et al 2004, Cui et al 2005. In this study, to quantify the CO 2 exchange and characterize how environmental factors affect the seasonal dynamics of this exchange in low-latitude estuarial wetland ecosystems, we used continuous eddy-covariance (EC) measurement to conduct an investigation in a subtropical estuarine wetland ecosystem in northern Taiwan. In this site, two flux towers were built over two different but representative tropical-to-subtropical ecosystems, para grass and reed. We used EC techniques to analyze temporal variation in CO 2 exchange and determined the effect of environmental factors on the dynamics of CO 2 budget in these two different ecosystems. We aimed to (1) quantify the CO 2 budget, including GPP, ecosystem respiration (R eco ), and NEE by examining EC data in terms of environmental variables; (2) analyze the effect of environmental factors on the CO 2 budget; and (3) interpret possible shifts and trends in CO 2 dynamics in this representative subtropical estuarine wetland ecosystem under environmental change.

Study site
Data was collected from an EC flux site (25°7'N, 121°2 8'E, and 4 m a.s.l.) in the western part of Guandu Nature Park in northwest Taipei City (figure 1). This site is a grass Marsh ecosystem along the riverbank of the Tamsui River, the major river system of the Taipei Basin and is a natural succession-processing area without direct artificial management on the landscape and ecosystem within the study area. The predominant species are para grass (Brachiaria mutica) and reed (Phragmites australis) (Chang 2011), with mean canopy heights of about 1.2 and 3.0 m, respectively. The major soil type is alluvium clay and is uniformly distributed through the entire rooting zone. The climate is humid subtropical and is strongly influenced by typhoons during the summer and the East Asian monsoon over different seasons. The annual mean temperature is 23.0°C, and the annual precipitation is 2405.1 mm, with a very mild dry-wet seasonal variation. As for the surface water status, the soil moisture content in reed is mostly close to saturation (from 41.0% to 45.5%) while the variation of soil moisture content for para grass is relatively greater (from 24.5% to 45.5%). Because the precipitation in this study site is more abundant during the summertime, the SWC is high during the wet season and the reed area occasionally experiences surface inundation up to 10 cm in depth for couple hours.
Because the location of the site is only 10 km away from the estuary of Tamsui River, the impact caused by interactions of the estuarial tides and stream flow should play important roles in the hydrological properties, energy budget, and biogeochemical cycles in the area. Several studies provided sufficient information and data on the patterns of nutrients and soil characteristics (Lee 2011) and basic dataset for the local vegetation distributions (Shao 1999) in this subtropical wetland ecosystem.

Instruments
The EC flux and micrometeorological data were collected from two observation towers in the center of the site. The first tower (hereafter T1) is a 3.5 m high walk-up tower located in the center of the site and covered by para grass. Data have been collected at T1 since November 2010. The sensors for EC flux were mounted at 3 m and included an open-path infrared gas analyzer (Li-7500, LI-COR, Lincoln, NE, USA) for measuring molar densities of CO 2 and water vapor (H 2 O), and a 3D ultrasonic anemometer (Young-81000, R M Yang Corp., Traverse City, MI, USA) for simultaneously measuring wind speed components, wind direction, and virtual temperature.
The second tower (hereafter T2) is a 5 m high walk-up tower located 125 m to the east of T1 and covered by reeds. Data collection at T2 began in January 2013. The EC-system sensors were mounted at 5.0 m and include a Li7500 gas analyzer and a 3D ultrasonic anemometer (CSAT-3, Campbell Scientific, Logan, UT, USA). All scalars and fluxes on both towers were sampled at 10 Hz and averaged every 30 min. Data from these two towers were collected by use of a data logger (CR3000, Campbell Scientific, Logan, UT, USA) mounted on T1. To estimate the source area of the flux measurement of these two towers, the footprint model developed by Kormann and Meixner (2001) was applied, and the results show that 90% of the footprint is 56 m for T1 and 82 m for T2. According to the results, the footprint extents for both ecosystem areas are spatially homogeneous and are representative of both towers.
In addition to the EC data, meteorological variables were collected at both sites. Data for PAR (collected by LI-190 quantum sensor, LI-COR, Lincoln, NE, USA) and radiation components (CNR1 fourcomponent net radiometer, Kipp and Zonen, Delft, The Netherlands) were collected at the top of T1 and T2. Meanwhile, PAR (BF5 sunshine sensor, Delta-T Devices, Cambridge, UK) and radiation components were sampled at the top of T2. Air temperature (T a ) and relative humidity were measured by use of HMP45C probes (Campbell Scientific) at both towers. SWC was measured at 0.02 and 0.20 m by using timedomain reflectometry probes (CS616, Campbell Scientific). Precipitation were measured with use of a tipping-bucket rain gauge (RS-102-D, Ogasawara Keiki, Japan) mounted on T1. All meteorological data were recorded by use of the CR3000 data logger on T1 and averaged every 30 min.
In this study, the data from January 2011 to May 2014 for para grass and from January 2013 to May 2014 for reed was used for analysis and comparison. Energy, momentum, and scalar fluxes were calculated for each 30 min period. Data quality control in this study applied the following standard procedure widely used in FluxNet, AsiaFlux, and CarboEurope (Webb et al 1980, Foken and Wichura 1996, Vickers and Mahrt 1997, Wilczak et al 2001, Foken et al 2004. The high-frequency raw data quality check, spike detection, and data quality classification (Foken et al 2004) were performed by using the EdiRe software developed by University of Edinburg (Clement 2004). Afterwards, the half-hourly flux data was calculated by applying planner-fit correction (Wilczak et al 2001), stationarity test (Vickers andMahrt 1997, Foken et al 2004), and Webb-Pearman-Leuning correction (Webb et al 1980).
According to the statistics of the measurement data, the gaps (missing data) of the entire study period were 17.8% for para grass and 21.1% for reed, and the gaps are mainly due to the interference caused by precipitation on the LI-7500 open-path gas analyzer, as well as the thermal stratification during the nighttime. To fill the gap, the gap less than 3 h were recovered by using a simple interpolation and the gaps of several hours were filled by applying the mean diurnal variation method (Falge et al 2001).

Quantifying CO 2 budget
On the ecosystem scale, the NEE of CO 2 could be partitioned into two components, GPP and R eco , by the following equation: 10 where Q 10 is the temperature coefficient), an indicator of sensitivity to temperature, and defined as the change rate of R eco on increasing the temperature by 10°C (Xu and Qi 2001).
In addition to R eco , NEE could be parameterized by appropriate environmental factors. During the daytime, maximum ecosystem photosynthesis rate A max (mg CO 2 m −2 s −1 ) and apparent quantum yield α (μmol CO 2 μmol −1 photon) were used to indicate plant photosynthetic capacity and characterize the response pattern to radiation (Ruimy et al 1995, Frolking et al 1998, Zhang et al 2006. These two parameters were derived from the relationship between daytime NEE (mg CO 2 m −2 s −1 ) and PPFD (μmol photon). For this purpose, mean daytime values (binned-averages by time of day) for NEE and PPFD were calculated for parameterization. From the daytime 30 min averaged data, the values for A max and α on the ecosystem scale were estimated by the Michaelis-Menten model (Ruimy et al 1995): From the measured NEE, the GPP was calculated by equation (1). Daily and monthly values for GPP and R eco were summed from the 30 min data.

Path analysis
To evaluate how environmental variables affect GPP of CO 2 , we performed path analysis (Li 1981, Liu et al 2011 with flux data for both ecosystem areas at Guandu to characterize the detailed relationship between environmental factors and components of CO 2 budgets for the two ecosystem areas at seasonal to annual temporal scales. Path analysis, an extension of multivariable regression analysis, provides estimates of the magnitude and significance of hypothesized causal connections of different environmental variables. The method has been widely used in many ecological studies (Mitsch et al 1988, Xu and Qi 2001, Zhuang et al 2004, Whittaker et al 2009, Mitsch et al 2010. In this study, we assumed that T a , PAR, SWC and VPD were the primary factors in driving or constricting photosynthetic CO 2 fixation, and their interrelated paths provide the means to evaluate direct and indirect responses (primarily of PPFD). To quantify the path coefficient for each component, we conducted four multiple linear regressions on data collected during the entire study period that did not violate conditions of statistical stationarity in the measured turbulent fluxes. We used standardized partial-regression coefficients to describe the path strengths generated from the following analyses: (1) GPP regressed on T a , PAR, SWC and VPD; (2) environmental factors regressed on each other. In addition, the test of statistical significance was performed for each regression to characterize how significant the effect was based on a given significance level (0.05).
In order to process the regression analysis, an appropriate time step was determined first to maintain the statistical stationarity (Liu et al 2011). Because flux components of grass ecosystems could be affected by short-term weather events (typhoons, fronts, etc) for the time scale of a couple of days (typically three to five days in subtropical East Asia) (Chen and Chen 2003), we used the time step of four days as an average duration to smoothen the effects of meteorological fluctuations.

Comparison of temporal variations of NEE
Climate data suggests there is no onset or end of the growing season in terms of temperature and water status at this site, and the seasonality was relatively mild compared to other wetlands at higher latitudes. To characterize the difference in seasonal variations, we simply divided the data into two groups: the 'fast growing season' for the summer months from May to September, with monthly mean temperature ⩾ 27°C, and the 'slow growing season', from October to April. Here the terms 'fast' and 'slow' are just for naming of groups.
At T1, the hourly dynamics of NEE during fast and slow growing seasons showed similar trends but different magnitudes (figure 2). Mean NEE was −8.0 and −0.4 μmol m −2 s −1 during the fast and slow growing seasons, respectively, and annual mean NEE was about −5.6 μmol m −2 s −1 . At T2, the difference between the fast and slow growing seasons was smaller than that at T1. The mean NEE was −10.8 and −8.7 μmol m −2 s −1 during the fast and slow growing seasons, respectively, and the annual mean NEE was −9.6 μmol m −2 s −1 . As for the components of CO 2 budget in the annual scale, from the EC data and the corresponding environmental factors, NEE, R eco and GPP were estimated to be −53.48, 1732.13 and −1785.64 gC m −2 yr −1 , respectively, for the para grass area and −376.12, 1351.44 and −1727.62 gC m −2 yr −1 , respectively, for the reed area (table 1).
The comparison of NEE dynamics for these two ecosystems showed that seasonal variation in CO 2 flux between the fast and slow growing seasons for the para grass was relatively larger than reed. The results imply that variations in environmental factors over different seasons were more important for CO 2 uptake in para grass than in reed in this area.

Responses of NEE to environmental factors
To parameterize R eco , we analyzed the relationship between nocturnal NEE and T a by equation (2) for both areas (figure 3(a)). At an annual time scale, the temperature dependence of R eco was lower for reeds (T2) than para grass (T1). The temperature coefficient Q 10 was estimated at 1.97 for T1 and 1.49 for T2. Therefore, the results suggest that reeds had higher adaptive resilience to fluctuation in environmental temperature than did para grass in this study site. In addition, the difference in surface water status (the flooding is more frequent in T2 than in T1) is expected to be another cause to suppress CO 2 release of reed.
To parameterize NEE in terms of radiation, we used the Michaelis-Menten curve (equation (3)) to fit the whole observed NEE data at both towers. However, the regression curve underestimated NEE at low radiation and could not properly capture the patterns throughout the entire PPFD range. Previous studies suggested that the light response curve may have distinct patterns under different radiation conditions (Leverenz 1987, Kull andKruijt 1998). Therefore, we divided the data into two groups with different radiation regimes: the data with PPFD < 600 μmol m −2 s −1 was calculated by linear regression (Kull and Kruijt 1998, Leverenz 1987, Lewis et al 2000, and the other group with PPFD ⩾ 600 μmol m −2 s −1 was estimated with the Michaelis-Menten model. The fitted curves suggest that the magnitude of daytime NEE at both towers could be appropriately parameterized by the Michaelis-Menten model (R 2 = 0.695 for T1 and R 2 = 0.680 for T2) ( figure 3(b)). The difference in NEE patterns in the comparison implies that the two   figure 3(b) shows that CO 2 assimilation was 20% higher for reeds at lower radiation condition, but para grass was more capable of assimilate CO 2 at higher radiation condition of PPFD ⩾ 1200 μmol m −2 s −1 .

Path analysis
We used path analysis to examine the effects of major environmental variables on controlling GPP at 4-day time step as mentioned earlier in the method section for both para grass and reed areas. The environmental variables included T a , PAR, and VPD for both ecosystems. In addition, we considered SWC for T1 only because the swamp-covered land surface around T2 caused saturated condition all the time. At T1 (figure 4(a)), all four environmental variables had significant and positive effects on the magnitude of GPP. However, T a and PAR had greater path coefficients than did VPD and SWC ( figure 4(a)), for stronger control on GPP. At T2 ( figure 4(b)), path analysis revealed a similar pattern, with both T a and PAR playing more important roles than VPD on GPP. However, the path coefficients for the environmental variables were smaller than at T1. The ecosystem response in terms of GPP was relatively less sensitive to fluctuations of environmental factors for reeds than para grass at this study site.
To further investigate how the environmental factors affect the dynamics of GPP throughout the entire 'fast growing season', we quantified the relationship between environmental factors and EC-based GPP during the 4-day time step over three different periods of the summer season (May-September, the 'fast growing season' in this study) as well as the transition periods just before (April) and after (October) (Liu et al 2011). These three periods include the early stage (from April to May), the middle stage (from June to August), and the late stage (from September to October), and the relationship was regressed by using scatter plots (figure 5). At T1 ( figure 5(a)), at all three stages of the growing season, PAR and T a had the greatest effect on GPP. However, at T2 ( figure 5(b)), these effects were relatively mild. The result was consistent with the path analysis findings.

Comparison of CO 2 budget for para grass and reed
The two ecosystems in this study are adjacent to each other and experience almost identical environmental conditions. It makes these two ecosystems good examples to characterize their sensitivity to environmental changes. The results of CO 2 flux show stronger seasonality for para grass than reeds (figure 2). Thus, according to results shown in table 1, the quantities of GPP for these two areas are almost identical, but NEE and R eco are showing different magnitudes in annual scale. The annual ecosystem respiration was 28% higher at T1 than at T2. One of the reasons is probably because of relatively more plant debris in para grass area, which provided more sources for CO 2 release. Another reason to cause such difference is mainly attributed to the difference in surface water status between this two sites. As stated earlier, soil moisture content in T2 is close to saturation, which could significantly suppress the release of CO 2 around T2 (Pezeshki 2001).
The analysis revealed that CO 2 assimilation was 20% higher for reeds than para grass at PPFD ⩽ 800 μmol m −2 s −1 , but para grass was more capable of uptaking CO 2 than reeds at higher radiation condition of PPFD ⩾ 200 μmol m −2 s −1 (figure 3(b)). Thus, para grass was more sensitive to the light environment than reeds and less productive at lower radiation. According the observation data, about 63.6% of the PAR records were ⩽ 800 μmol m −2 s −1 , and only 20.3% were ⩾ 1200 μmol m −2 s −1 . Therefore, daytime NEE was relatively greater for reeds than para grass in this area. This finding agreed with path analysis results, which suggests that PAR had a greater effect on para grass than reeds. The results of path analysis and scatter plots also indicate that the variations in both T a and atmospheric water status (in terms of VPD) had stronger effects on para grass than reeds.
Para grass is native to tropical areas of western and northern Africa (Thornton andMcManus 1994, Chanson et al 2004) and spread to Southeast Asia, Australia and Latin America. In contrast, reeds are found on every continent over different latitudes, except Antarctica (Marks et al 1994, Chambers et al 1999. Over the last 150 years, the distribution of reeds has greatly increased and the species are now considered aggressive invaders in many regions (Chambers et al 1999, Rice et al 2000. A previous study found that when the invasion of reeds progresses from pre-invasion conditions to 90% cover, C sequestration in wetlands significantly increases (Battin et al 2009). Similar conditions were also found in Guandu site. According to the long-term field observation conducted by Guandu Nature Park (Shao 1999, Lee 2011, these two species are native in the area. However, in the early 1990s, the area was dominated by para grass and the coverage of reeds was small, yet over time, reed dominated ecosystem expanded into areas previously dominated by para grass. Because the two ecosystems in this study showed different sensitivity to environmental factors and both are widely distributed in tropical and subtropical regions, variations in environmental factors under global or regional environmental changes may lead to changes in C budgets in low-latitude wetland ecosystems. Our results suggest that the para grass is relatively more sensitive than reed to the influences of environmental factors. Therefore, under the variations in environmental controls, the change rate of CO 2 exchange will be greater for para grass than other species such as reed. The highly sensitive para grass has less resilience than other species, which may lead to the less-sensitive species such as reeds becoming more abundant in low-latitude areas. Such changes in coverage can alter ecosystem processes and change biogeochemical cycling (Roehm and Roulet 2003) via an aggressive growth strategy over different temporal scales.

CO 2 flux compared with other wetland ecosystems
We used EC measurement with environmental factors to estimate the CO 2 budget for two ecosystem areas in the subtropical wetland site and compared our data with those from other studies with different methods conducted in other wetland ecosystems (table 1). We found the C sequestration at Guandu is greater than most studies.
The diversity of C budgets among these study sites is due first to differences in biomass among the study sites (the leaf area index LAI is 1.7 for the bog in The Netherlands, 1.5 for the peatlands in Canada, 4.0 for reed wetland in Panjin and 8.0 for para grass in this study). Ecosystems with lager biomass have more capacity for C fixation, which is consistent with other studies (Frolking et al 1998, Moreno-Sotomayor et al 2002. Furthermore, environmental factors play key roles in influencing C exchange over different sites. Our study site had no dry season, which is a typical climatic condition in tropical to subtropical regions in East Asia and Southeast Asia. However, in other studies, the monthly precipitation at higher latitudes showed large seasonal variation, so the dry condition in these sites may suppress the photosynthetic capacity. As well, low temperature restricts ecosystem photosynthesis (Lafleur et al 2001, Zhang 2006). Therefore, low temperatures at high-latitude sites results in short growing seasons, low photosynthetic capacity, and lower annual GPP.

Factors affecting ecosystem CO 2 exchange
Several previous studies tried to characterize how environmental controls alter C exchange in grassland ecosystems. However, most of these studies focused on patterns in northern ecosystem. An analysis of a high-latitude northern European peatlands area showed that PAR was the dominant factor controlling gross and net uptake of CO 2 (Lindroth et al 2007). In addition, soil water status played an important role as well (Forrest andSmith 1975, Moore 1989). Other studies of high-latitude peatlands ecosystems found that the fixation and transformation of N could alter dynamics of C cycling by a complicated mechanism (Hemond 1983, Urban and Eisenreich 1988, Rochefort et al 1990, Li and Vitt 1997, Williams and Silcock 1997, Williams et al 1999. Therefore, understanding the key ecological properties such as CO 2 dynamics and the influences of environmental fluctuations on estimating CO 2 storage is important for characterizing low-latitude wetlands. In this study, we provide solid quantitative evidence to conclude that the seasonal dynamics of CO 2 exchange in a typical tropical-to-subtropical wetland ecosystem is strongly associated with the seasonality of several environmental factors. CO 2 storage and flux of different dominant species in low-latitude Marsh wetlands have different sensitivities to seasonal variations in meteorological conditions. For low-latitude Marsh wetland ecosystems such as at Guandu and all other ecosystems in East to Southeast Asia, temperature and PAR have stronger effects than do soil water status and VPD on CO 2 exchange. These findings differ from those for high-latitude estuarial wetland ecosystems in Northeast China (Zhou et al 2009), which found that SWC and temperature had greater effect on GPP than PAR and VPD for high-latitude wetlands.

Conclusions
In this study, we used EC measurement to quantify the ecosystem-scale budget of CO 2 exchange at two different but representative wetland ecosystem in northern Taiwan. These two vegetation types showed different NEE response during different seasons. Temperature and radiation (PAR) had a greater effect than water status (SWC and VPD) on GPP for these two low-latitude ecosystems, which differs from findings for high-latitude regions. The results from path analysis and regression revealed that environmental variables have strong but different impacts on CO 2 budget for para grass and reed in this area. It implies that in this low-latitude wetlands area, the sensitivities of these species to the variations in environmental controls are statistically different, and these two ecosystem types would possibly respond at different magnitudes under different environmental conditions. Furthermore, such discrepancy indicates different capabilities of CO 2 exchange, so this diversity leads to different potential shifts and trends of biomass accumulation and distribution of these two typical low-latitude vegetation types under environmental changes. The finding from this study can sufficiently provide quantitative understanding on managing wetland ecosystem at different temporal and spatial scales.
Understanding the dynamics of C stocks, release and sequestration in subtropical wetlands would improve current understanding of C cycling at regional and global scales. Adequate approaches to quantifying these fluxes and stocks remain challenging. Standardized methods of monitoring, reporting and verification need to be developed. Finally, CO 2 temporal variability ranges from daily to interannual. C-rich tropical and subtropical wetland ecosystems should be considered high priorities in climate-change adaptation and mitigation strategies throughout the world. For a more robust evaluation of C flux in these areas, an intensive, integrated, international and interdisciplinary program of observational efforts is required.