Seasonal variability of groundwater level effects on the growth of Carex cinerascens in lake wetlands

Abstract Groundwater level is crucial for wetland plant growth and reproduction, but the extent of its effect on plant growth can vary along with changed precipitation and temperature at different seasons. In this context, we investigated the effect of two groundwater levels (10 cm vs. 20 cm depth) on growth and reproductive parameters of Carex cinerascens, a dominant plant species in the Poyang Lake wetland, during three seasons (spring, summer, and autumn) and during two consecutive years (2015 and 2016). Carex cinerascens showed low stem number, height, and individual and population biomass in summer compared to spring and autumn. 10 cm groundwater level was overall more suitable for plant growth resulting in higher stem height and biomass. However, the interactive effect between groundwater level and season clearly demonstrated that the effect of groundwater level on plant growth occurred mainly in autumn. After the withering of the plant population in summer, we observed that C. cinerascens growth recovered in autumn to similar values observed in spring only with 10 cm groundwater level. Consequently, we could deduce that lowering groundwater level in the studied Poyang Lake wetland will negatively impact C. cinerascens regeneration and growth particularly during the second growth cycle occurring in autumn. Additionally, our results showed that, independently of the season and groundwater level, population biomass of C. cinerascens was lower during drier year. Altogether, our findings suggest that water limitation due to both reduction in precipitation and decreased groundwater level during the year can strongly impact plant communities.

Seasonal and annual climatic variations, along with hydrological fluctuations, can induce distinct plant responses to groundwater level (Dai et al., 2019;Runhaar et al., 1997) with cascading effect on vegetation distribution (Yuan, Wang, et al., 2017a). Thus, plant survival and growth under the same water level condition are not necessarily consistent at different seasons (Steed & DeWald, 2003).
Surprisingly, previous studies testing the effects of groundwater level on plant growth were mostly conducted during few months and during one growing season only (e.g., Hanke, Ludewig, & Jensen, 2015;Kotowski et al., 2001;Li et al., 2014). Climatic parameters like temperature and precipitation can strongly vary across the seasons  and some previous studies reported that the main hydrologic factors for plant growth and vegetation distribution differ between spring and autumn (Dai et al., 2019;Runhaar et al., 1997). However, to our knowledge, the extent to which variations in groundwater level could affect plant parameters during distinct plant growing periods (i.e., spring vs. autumn) in relation to varying meteorological conditions remains poorly studied.
To fill this knowledge gap, we investigated the effect of two groundwater levels (10 cm vs. 20 cm depth) over three seasons (spring, summer, and autumn) on the growth and reproductive parameters of Carex cinerascens Kük., a dominant plant species in the Poyang Lake wetland (China), during two consecutive years (2015 and 2016). We hypothesized that: (a) C. cinerascens growth parameters are greater in spring and autumn (i.e., during the growing seasons) compared to summer; (b) 10 cm groundwater level favors both C. cinerascens growth and reproductive parameters as compared to 20 cm groundwater level; (c) the effect of groundwater level is also stronger in autumn than in spring and summer because lower autumn precipitation interacts with groundwater level to limit plant growth in the Poyang Lake wetland.

| Study site
Poyang Lake (28.37 to 29.75°N, 115.78 to 116.75°E) is included in Yangtze River drainage basin and is currently the largest freshwater lake in China ( Figure 1a). The climate is subtropical monsoon with a mean annual temperature of 17.6°C and mean annual precipitation of 1,528 mm. Precipitation increases from January reaching the maximum in June and decreases sharply from July to September with dry conditions lasting until December (Guo, Hu, & Jiang, 2008). Groundwater level of the lake is strongly controlled by the precipitation in the basin but also impacted by Three Gorges Dam and sand mining activities that regularly induce a decrease in water level (Lai et al., 2014;Zhang et al., 2012). In the Poyang Lake wetland, there are five main plant communities dominated by Phragmites australis (Cav.) Trin. ex Steud., Triarrhena lutarioriparia L. Liu, Artemisia selengensis Turcz. ex Bess., Carex cinerascens Kük, and Phalaris arundinacea L. in a typical zonation from upland to the shoreline (Wang, Han, Xu, Wan, & Chen, 2014;You et al., 2015).

| Plant material sampling
The experiment was conducted with Carex cinerascens Kük.
( Figure 1b,c), a perennial rhizomatous species, which is a dominant K E Y W O R D S groundwater, hydrological condition, plant growth, plant reproduction, plant species composition, Poyang Lake, wetlands and widely distributed hygrophyte plant species in the wetland zone, mostly subjected to groundwater level fluctuation in the Poyang lake wetland (Yuan, Yang, Liu, & Wang, 2017b;Zhang et al., 2019). C. cinerascens has a phenology characterized by two growing periods in spring and autumn, respectively. In spring, plants emerge with increasing temperature in February and they flower and fruit in April with a peak of biomass production at the end of April. Then, aboveground part of plants dies during the summer flooding. In autumn, after flooding recession, new stems are produced with a peak of biomass around October. Then the plants generally wither with decreasing temperature in winter and new stems sprout again in the following spring.
In February 2014, six turfs (38 cm diameter × 38 cm depth) were collected in a wetland area near the Poyang Lake Wetland Research Station (29.45°N, 116.06°E). The turfs contained both intact C. cinerascens individuals and corresponding undisturbed soil. Each turf was 30 cm depth because the roots of C. cinerascens never exceeded 30 cm depth, as observed during regular field investigations. Soil collected in this area had a pH of 6.3, a bulk density of 1.21 g/cm 3 , a total organic carbon (TOC) concentration of 27.19 g/kg, a total nitrogen (TN) concentration of 1.43 g/kg, and a total phosphorus (TP) concentration of 0.86 g/kg. At the same time, the subsoil below the sampled turfs was also collected and mixed to fill the devices used in the experiment (see below for explanation).

| Experimental design
The experiment was established with two groundwater levels corresponding to a depth of 10 and 20 cm (3 replicates per treatment) and representing the distance from the soil surface to the free groundwater table. Groundwater levels were selected as representative of realistic water fluctuations in Poyang Lake wetland and linked to the rooting depth of C. cinerascens. Previous measurements of C. cinerascens root system showed that 85% of the root biomass is found in the first 10 cm (unpublished results), thus 10 cm groundwater level would slightly limit water access whereas 20 cm groundwater would correspond to a high water access limitation. We used acrylic glass containers (50 cm height × 38 cm diameter) that included three layers: 10 cm of coarse sand at the bottom to ensure the connectivity between the container and a water tank, 10 cm of mixed subsoil collected from the field covered by the 30 cm deep turf of C. cinerascens on the top.  After the week of plant acclimatization, the groundwater level was adjusted daily to the targeted depth (10 cm or 20 cm) with tap water. A valve was installed 5 cm from the bottom of each container. Through this valve, each container was connected to its water tank (25 cm height × 25 cm diameter) by a plastic pipe to adjust the groundwater level. Rulers were adhered to the surface of the containers to measure soil depth and water level in each container. The outside surface of each container was covered with aluminum foil to mimic underground darkness and prevent the temperature in the containers from rising excessively, and at the same time, to allow an easy checking of the inner groundwater level. When it was not raining, the valve was kept on and the groundwater level was adjusted manually every day to reach the desired groundwater levels. The valves were switched off before raining to reduce nutrient leakage, and the redundant water was drained after raining event.

| Meteorological data
The temperature and precipitation data were obtained from the Poyang Lake Wetland Research Station. Since the plant growing duration in spring and autumn is around 3 months long, the average temperature and total precipitation of 3 months before plant traits measurement was used to indicate the meteorological conditions during plant growth (Table 1) corresponding to February 1st to April 30th for spring, May 1st to July 31st for summer, and August 1st to October 31st for autumn.

| Plant growth parameters
Plant growth parameters of C. cinerascens were measured in 2015 and 2016 at the spring peak of biomass production (April), during summer before the autumn sprouting (July), and at the autumn peak of biomass production (October).
The stem number per container and the height (cm) of 5 random stems per container were measured. Three stems among the 5 randomly selected were cut at the soil surface, and oven-dried weight (65°C for 72 hr) of the aboveground biomass (stem and leaves) was measured and used to determine mean aboveground biomass per individual. Aboveground biomass of the entire population of C. cinerascens per pot was estimated based on the mean individual biomass × stem number per pot. In addition to growth parameters, the number of flowering stems per pot was measured in April 2015 and 2016.
Other plant species, in addition to the dominant C. cinerascens, grew in containers over the 3-year long experiment. Stem number of each auxiliary species was recorded in each container, and Shannon diversity index (H) was calculated as:

| Data analysis
Statistical analyses were performed with the R software (version 3.3.1). Significance was evaluated in all cases at p < .05. Data were log-or square root-transformed when necessary to meet the assumption of normality and homoscedasticity.
We used linear mixed-effect models (nlme package), followed by

| Meteorological conditions
The mean temperature in 2016 was slightly higher than in 2015 in spring, summer, and autumn (Table 1). The precipitation in spring and autumn 2016 was lower than in 2015, while the precipitation in summer was higher in 2016 compared to 2015 (Table 1).

| Response of growth parameters
Stem number was 24% lower in 2016 compared to 2015 (Tables 2 and   3). The differences in stem number across seasons were dependent on the groundwater level considered (significant groundwater level × season interaction, Table 2 Table 2). At 10 cm groundwater level, stem height decreased in summer and increased in autumn, but it reached a lower value compared to spring, while it decreased in summer and remained constant in autumn at 20 cm groundwater level ( Figure 2b). By consequence, the difference in stem height between spring and autumn was more marked at 20 cm groundwater level (34%) compared to 10 cm groundwater level (19%). We also observed that in autumn plant height was 23.6% higher at 10 cm compared to 20 cm groundwater level (Figure 2b).
Individual aboveground biomass was higher in autumn compared to the two other seasons (Tables 2 and 3), but this difference was greater at 10 cm groundwater level (64%) compared to 20 cm groundwater level (38%), leading to difference between groundwater levels only in autumn (significant groundwater level × season interaction, Table 2; Figure 2c).
The aboveground biomass of C. cinerascens population was 27% lower in 2016 compared to 2015 (Tables 2 and 3). The aboveground biomass of C. cinerascens population at 10 cm groundwater level was two times higher compared to 20 cm groundwater level, and this difference was due to lower biomass in autumn at 20 cm compared to 10 cm groundwater level (significant groundwater level × season interaction, Table 2; Figure 2d). At 10 cm groundwater level, population biomass decreased from spring to summer and increased in autumn, reaching a value twice higher than in spring, while at 20 cm groundwater level it decreased in summer and increased in autumn to reach similar value as in spring (Figure 2d). No significant effects of groundwater level, sampling year, or their interaction were observed in spring for inflorescence number (Table 2), despite a decreasing trend with decreasing groundwater level (Table 3).

| Plant species diversity response
In contrast to growth parameters, plant species diversity was not affected by groundwater level (Tables 2 and 3). Plant species diversity TA B L E 2 Effects of groundwater level (WL, 10 or 20 cm), season (spring, summer, and autumn), year (2015 and 2016), and their interactions on C. cinerascens growth parameters and plant community diversity. Inflorescence number was only recorded in spring so only WL, year and their interaction were tested for this parameter. F-values and associated p-values (with the respective symbols * for p < .05, ** for p < .01, and *** for p < .001) are indicated  (Figure 3b). The interaction between season and year also showed that there was no difference in plant species diversity across seasons in 2015, while the value in summer was higher than those in spring and autumn in 2016 (Figure 3b). Stem height (cm) Shannon diversity index

| D ISCUSS I ON
Our study aimed at testing the effects of groundwater level and interannual climatic variability on the growth parameters of C. cinerascens. In accordance with our first hypothesis, all measured C. cinerascens growth parameters showed lowest values in summer compared to spring and autumn. After reaching a peak of biomass in April, C. cinerascens started to wither until new plants emerged in August, corresponding to the second growing period (see Section 2.2. for further details). The summer quiescence of the species suggests that precipitation and temperature conditions in summer were not favorable for C. cinerascens growth (Table 1 and Appendix). C. cinerascens had higher stem number and height in spring compared to autumn, but similar population biomass during these two growing periods. Such results were consistent with those of Wang, Xu, Wan, and Chen (2016), but contradictory to those of Hu, Wu, Yao, and Xu (2015), which showed higher biomass of Carex-dominated community in spring in Poyang Lake wetland. Both our study site and the study site of Wang et al. (2016) are located in the same area (northwest side of the lake) and have similar meteorological conditions. By contrast, the study site of Hu et al. (2015) is located at the south of the Lake, and during their experiment there was much more precipitation in spring compared to autumn, which can explain the higher biomass production in spring at this site. During our experiment, C. cinerascens also exhibited lower stem number and population biomass in 2016, which coincided with lower precipitations for both growing periods (spring and autumn) in comparison to 2015 (Table 1). Altogether, such findings further highlight the strong impact of precipitation in driving C. cinerascens productivity in Poyang Lake wetland.
Even though C. cinerascens withered in summer with fewer stem number compared to spring and autumn, some auxiliary species like Artemisia selengensis still grew and other auxiliary species like Conyza canadensis thrived due to the luxurious seed bank available in the Poyang Lake wetland (Wall & Stevens, 2015;Yu & Yu, 2011). In accordance with our second hypothesis, 10 cm groundwater level was more beneficial for C. cinerascens growth compared to 20 cm groundwater level. Indeed, C. cinerascens exhibited higher stem height, individual, and population biomasses with groundwater closer to soil surface. This result is consistent with the findings of Li et al. (2014), which showed that Carex brevicuspis grows better in 20 cm than 40 cm groundwater level in Doting Lake wetlands, China. While increasing water level can have negative effect on plant growth due to oxygen shortage (Kotowski et al., 2001;Li et al., 2017), positive effects of shallow groundwater levels on C. cinerascens growth suggest that water availability is the main limiting factor for plant growth over seasons and years. Despite not significant, we observed a trend of increasing inflorescence number with higher groundwater level, a finding in agreement with the study of Chen et al. (2015) that reported a positive correlation between the sexual reproduction of C. brevicuspis and soil moisture content in Doting Lake wetlands. As Poyang Lake water level is declining due to the effects of Three Gorges Dam, sand mining, and climate change (Lai et al., 2014;Liu, Wu, & Zhao, 2013;Piao et al., 2010;Zhang et al., 2012), our results suggest that deeper groundwater levels would simultaneously reduce vegetative and sexual reproduction with strong impacts on C. cinerascens populations in the future. Wetland ecosystem is known as an important carbon sink (Bernal & Mitsch, 2012;Kayranli, Scholz, Mustafa, & Hedmark, 2010) and as C. cinerascens population is an important component of Poyang Lake wetland, any alteration in the productivity of C. cinerascens may induce cascading effects on wetland functioning.
In line with our third hypothesis, significant interactive effects of groundwater level and season on the different growth parameters of C. cinerascens clearly demonstrated that the effect of groundwater level on plant growth was dependent on the season considered and, as expected, the effects occurred mainly in autumn. In spring, temperature was low, resulting in low soil evaporation and plant transpiration, so that precipitation supplied plants with the required water amount for optimal growth (Appendix).
This situation could probably explain why deeper groundwater level only marginally reduced the inflorescence number of C. cinerascens. Compared to spring, precipitation amount decreased considerably in autumn while the temperature remained high and as a result the groundwater level became the main water source and a limiting factor for plant growth. Our finding is supported by two previous studies which reported that higher groundwater level had no effect on plant biomass in spring, but increased plant biomass in autumn in Dongting Lake wetland, China (Li et al., 2014(Li et al., , 2017.
However, after the withering of the plant population in summer, we observed that C. cinerascens growth recovered in autumn only under 10 cm groundwater level. From this result, we could deduce that decreasing groundwater level in the studied Poyang Lake wetland will negatively affect C. cinerascens regeneration and growth particularly during in autumn.

| CON CLUS ION
Our study provides clear evidence that the groundwater level can have distinct effects on plant growth according to the season, with generally stronger effects occurring in autumn compared to spring and summer. Indeed, under low precipitation, as typically observed in autumn, groundwater level controls water availability and becomes an important driver of plant growth, which significantly influences annual biomass production of C. cinerascens. Overall, lower groundwater level of Poyang Lake impedes C. cinerascens development and constrains population growth but does not affect sexual reproduction in spring, as water availability is ensured by sufficient amount of precipitations. Additionally, our results also showed that, independently of the season and groundwater level, population biomass of C. cinerascens was lower during the drier years (i.e., 2016), further highlighting the role of water limitation on plant growth in Poyang Lake. Altogether, these findings suggest that any change in water availability might strongly impact plant communities.
Interactive effects of groundwater level and precipitation patterns should therefore be taken into account to predict plant community dynamics in Poyang Lake wetlands.

ACK N OWLED G M ENTS
We thank Qiuting Deng, Qi Xie, He Liu, Jinying Xu, and Junxiang Cheng for their contribution to the data collection.

CO N FLI C T O F I NTE R E S T
None declared.

AUTH O R CO NTR I B UTI O N S
WF and LX designed and performed the experiment. WF, PM, and MS analyzed the data and led the writing of the manuscript. All authors contributed critically to the drafts and gave final approval for publication.