Global wetlands store a large proportion of the global land carbon and can play a critical role in the global carbon cycle (Mitra et al., 2005; Lal, 2008). In particular, wetlands account for 20%–30% of the global soil organic carbon (SOC) pool (Bridgham et al., 2006), and have been a net carbon sink in recent decades with an average net carbon retention rate of 118 g C m^-2 yr^-1 (Mitsch et al., 2013). However, changes in climate and wetland ecological features could turn them from carbon sinks into carbon sources (Oechel et al., 1993; Hadden and Grelle, 2016; Zhou et al., 2018). Freshwater bodies, such as streams, rivers and lakes, have produced a large amount of CO₂ (Schrier-Uijl et al., 2011; Wang et al., 2011). It is reported that global lakes alone emit 1943 Tg CO₂ to the atmosphere every year (Williamson et al., 2009). Nevertheless, most studies on wetland carbon cycles mainly focus on boreal peatlands (Fan et al., 2013; Wang et al., 2015; Karlsson et al., 2021), while research on the carbon cycle of tropical and subtropical wetlands, especially of the subtropical shallow lakes, is somewhat limited (Hu et al., 2015; Amaral et al., 2020; Calamita et al., 2021; Paranaíba et al., 2021).

Many subtropical wetlands experience seasonal cycles of water table (Carmignani and Roy, 2017). With the fluctuation of water table, some parts of the wetland soils periodically shift between an aerobic and anaerobic environment. This periodical water table fluctuation can strongly impact soil properties and the processes of subtropical wetland carbon cycle (Marcé et al., 2019; Keller et al., 2020; Machado dos Santos Pinto et al., 2020). A high water table may sustain the carbon stock of wetland by preventing aerobic respiration, whilst exposure to the air under low water table can greatly influence the CO₂ emissions from the terrestrial ecosystem to the atmosphere (Couwenberg et al., 2010; Swails et al., 2022). However, the extent to which the water table fluctuation may impact soil humidity, bulk density, soil carbon and the temporal dynamics of ecosystem carbon exchanges, still remain obscure, largely due to the lack of monitoring data. Moreover, with climate change and enhanced anthropogenic activities, many subtropical wetlands have experienced substantial land transformations and land cover changes, and can permanently alter the water table level (Zhao et al., 2002; Maertens et al., 2022; Richardson et al., 2022). Such land cover changes and land transformation-induced water table level alternation, together with periodical water table cycle, have profound implications for wetland carbon balances (Hu et al., 2015; Johnson et al., 2022). In particular, ecosystem respiration, a key component of the carbon cycle, can change dramatically in response to changes in land cover and water table level (Zhou et al., 2014; Ma et al., 2022; Li et al., 2023a). Quantifying the ecosystem respiration responses to water table variations in lacustrine mudflats, therefore, is important for understanding and predicting subtropical wetland carbon cycles.

The Dongting Lake in Hunan Province, China, is a typical subtropical ephemeral lake located in the middle of the Yangtze River Basin. As one of the last two large Yangtze-connected lakes, the average monthly water level of the Dongting Lake fluctuates between 20.19 and 30.24 m (Shi et al., 2012). Consequently, the littoral areas are completely inundated in summer flood season, but are exposed to the air in autumn–winter with an annual drawdown period lasting about 4–6 months. In recent decades, due to the water volume control of the Three Gorges Dam, the hydro-condition of downstream wetlands has greatly changed (Feng et al., 2013), resulting in even longer winter drawdown periods in Dongting Lake (Lai et al., 2013). During the drawdown periods, littoral plants, such as Carex, extensively emerge and grow continuously almost throughout the drawdown period, turning mudflats into meadows (Lei et al., 2014). This mudflat-to-meadow shift during the drawdown period suggests that autotrophic respiration may also substantially contribute to the total ecosystem respiration, which has been scarcely investigated to date. The main objective of this study, therefore, is to measure and understand ecosystem respiration from both mudflats and meadows in the Dongting Lake wetlands during the winter drawdown period. The dynamics of ecosystem respiration (including both plant autotrophic respiration (R_p) and soil respiration (R_s)) can be influenced by varied biotic and abiotic factors, such as vegetation activity (Urbanova et al., 2012; Ives et al., 2013; Wang et al., 2021), the quantity and quality of soil properties (Fan et al., 2013) and soil moisture (Zhou et al., 2014). To understand how these factors may influence ecosystem respiration, we also measured soil and climate factors, including soil water content (SWC), SOC, dissolved organic carbon (DOC), total nitrogen (TN), temperature at 5 cm soil depth (T₅) and air temperature in the meantime.

The sampling zones in this study were located close to the open water body during the drawdown period. With the water level fluctuating in Dongting Lake all the time, sampling zones exposed to the air were usually saturated during the drawdown period, which induced SWC at the surface soil that was almost same as that at the deep soil depth (Figure 2). Stable SWC among three layers of soil led to somewhat stable SOC and DOC between the surface and deep soil. It was reported that the decomposition of SOC is greatly reduced in anaerobic environments, such as marsh and swampland (Luo and Zhou, 2006). A higher value of SWC creates a more anaerobic environment, restraining soil respiration and advancing carbon sequestration in soil (Hudson, 1994). However, SWC showed the trend of a lower value at the surface soil, which is not related to the slightly higher SOC value at 0–10 cm depth soil in this study (Figure 2). Consequently, it is concluded that sedimentation during the flooded period brings external organic matter to the surface soil (Dai et al., 2005). Using the system of Surface elevation table, the height of the lake basin in the Dongting Lake area was monitored. The average annual sedimentation in the past 5 years is approximately 7–8cm (personal communication), confirming the influence of exogenous organic matter on the surface soil. TN also showed the trend of a higher value at the surface layer than the deep layer due to the deposition of exogenous nitrogen during the drawdown period.

The content of SOC and TN in the soil was positively correlated with the time of mudflat exposure in the short term due to the supplementation of exogenous organic matter. With the intensification of CO₂ producing a reaction, nitrification and denitrification, the contents of SOC and TN (Gu et al., 2004) began to decline in the long term (Figure 3).

The value of C/N is one of the most important factors which determine the mineralization rate in soil. In this study, the C/N value was around nine, not significantly influenced by soil depth and CDE, which reflected that nitrogen is not the limit factor for soil mineralization in Dongting Lake. With the emergence of mudflat, soil microbial metabolic activities gradually increased, resulting in a significant increase in soil DOC content with the increase in exposure time (Figure 3).

It is believed that water content, temperature and other environmental parameters interact to determine the CO₂ release from the soil (Juszczak et al., 2013; Keller et al., 2020). In this study, only DOC was found to be significantly correlated with soil respiration at the mudflat, mostly due to organic carbon dissolved in the surface soil supplying a substrate for CO₂ production. When soil was saturated at the mudflats in Dongting Lake during the drawdown period, the mechanism of the CO₂ cycle was limited by the amount of active substrate that could be estimated by the DOC value.

During the drawdown period in Dongting Lake, the surface soil is almost fully exposed to the CO₂ reaction and consumption of DOC, making DOC at the surface soil slightly lower than in the deep soil (Figure 2). With the increase in the exposure time of the mudflat, the consumption of DOC could significantly decrease the respiratory flux from the mudflat. However, no significant correlation was found between CDE and the respiratory flux from the mudflat, which is derived from the increasing DOC induced by the organic matter deposition during the flooded period and the enhancement of microbial metabolism in the soil after mudflat exposure (Vonk et al., 2017). At the deeper soil, the DOC content increased with the increase in CDE (Figure 7), continuously supplying DOC for ecosystem respiration. Consequently, the substrate for ecosystem respiration is sufficient, intensifying ecosystem respiration shortly after exposure, then resulting in a gradual decrease in DOC at the surface soil depth after a short-term increase (Figure 7). Therefore, due to the accumulation of exogenous organic matter and the increase in soil microbial metabolism, the respiration substrate of the soil increased with the increase in CDE in the short term, resulting in a gradual increase in the respiration intensity of the sediment, which reached a high value at around 60 days. Followed by the continuous consumption of DOC, the respiratory intensity decreased gradually (Figure 5). How DOC determines the respiration process was also reflected by the significant correlation between respiration flux and DOC value (Figure 6).

When vegetation colonized, the root system interspersed, squeezed, divided the soil and absorbed soil moisture, creating a more aerobic environment at the surface soil layer. In Dongting Lake, no significantly difference was found between the soil properties from the meadow and the mudflat at the deeper soil layer, also suggesting that vegetation colonizing only impacts soil properties at the surface layer during the drawdown period by changing the aeration and generating more external organic matter (Jobbágy and Jackson, 2000; Tuo et al., 2018). The values of SOC, DOC and C/N at the surface layer were significantly higher in the meadow than the mudflat, reflecting the fact that the dominant Carex produces more organic carbon by photosynthesis and maintains aerobic environment advancing respiration. Thus, organic carbon was not the limit factor for respiration, which was merely determined by the soil temperature instead in the meadow (Figure 6). Furthermore, due to the lower value of C/N in Dongting Lake, SOC could be consumed as a substrate for soil respiration and root respiration in the short term. Consequently, it was found that the ecosystem respiration flux from the meadow was three times greater than that from mudflat.

Due to the exposure of the mudflat and the improvement of soil ventilation, the mean value of the respiratory flux from the mudflat of Dongting Lake in this study was 0.716 ± 0.114 μmol m^-2s^-1. This was far higher than the CO₂ flux of 0.2 μmol m^-2s^-1 in maximum from the terrestrial–aquatic interface in Chinese subtropical shallow lakes (Liu et al., 2013b; Xiao, 2014; Xiao et al., 2020; Li et al., 2023b). It is concluded that more SOC in the lake basin is involved in the respiration process after exposure to the air, significantly emitting more CO₂ into the atmosphere.

A previous study stated that fluxes from poplars, reeds and farmlands in Dongting Lake during the drawdown period were 0.70, 0.56 and 0.45 μmol m^-2s^-1, respectively (Tang, 2011), much lower than that from the Carex meadow and slightly lower than that from the exposed mudflat, and this is the opposite in the flooding season. During the flooding season, fluxes from poplars, reeds and farmlands in Dongting Lake were 4.80, 8.21 and 1.27 μmol m^-2s^-1, respectively, significantly higher than the flux from the terrestrial–aquatic interface and the flux from water surface. Consequently, the exposed mudflat, instead of poplar, reed or farmland, became the most active area of CO₂ emissions during the drawdown period. The intensity of respiration flux was strengthened after Carex colonizing. Reducing the area and duration of mudflat without littoral plants exposure to the air by manual control may be conducive to reducing CO₂ emissions to the atmosphere.

With CDE increasing during the drawdown period in Dongting Lake, exposure to the air and Carex colonizing together increased the soil DOC value in exposed mudflat, strengthening the intensity of ecosystem respiration. DOC was the limit factor of ecosystem respiration that reached a peak period around 60 days after exposure of the mudflat, while only soil temperature determined respiration in the newly colonized meadow with a sufficient supply of organic carbon. The ecosystem respiration value reached 0.716 ± 0.114 μmol m^-2s^-1 and 2.240 ± 0.375 μmol m^-2s^-1 in the exposed mudflat and newly colonized meadow, respectively, both of which became the most active areas of CO₂ emissions during the drawdown period.