Dongting Lake (28°30′–30°20′ N, 111°40′–113°10′ E) is the second-largest freshwater lake in China, with an area of 2625 km². It is located on the southern bank of the middle reaches of the Yangtze River. The surrounding wetlands have a northern subtropical monsoonal climate, with a mean annual temperature of 16.8°C and mean annual precipitation of 1382 mm. The floodplain wetlands experience large water level fluctuations (up to 15 m) and are completely inundated from June to October and exposed from November to May.

In the Dongting Lake wetlands, C. brevicuspis and M. sacchariflorus are the most abundant species in the two dominant plant communities, the Carex community and Miscanthus community, respectively (Xie et al., 2015). Carex brevicuspis is distributed at low elevations, ranging from 22–27 m. The pseudoculms of C. brevicuspis plants have overlapping leaf sheaths and are usually 20–55 cm in height (Chen et al., 2014). The shoots of C. brevicuspis decompose during the flooding season, and new shoots emerge immediately after the flood has receded. In contrast, M. sacchariflorus is mostly distributed at elevations greater than 27 m (Xie et al., 2015; Huang et al., 2022). Miscanthus sacchariflorus plants are 100–500 cm tall with robust and erect culms (Chen et al., 2015b). The shoots of M. sacchariflorus can go through the flooding season and subsequently flower and fruit.

Three lake shores where the two communities are typically distributed, Xiaoxihu (112.993684 N, 29.372535 E), Tuanzhou (112.858114 N, 29.336779 E), and Junshan (112.796314 N, 29.458161 E), as study sites ( Figure 1 ). A 1 km transect parallel to the lakeshore at each site in the middle of each community zone was established. In November 2017, five permanent plots (5 × 5 m) 100 m apart on each transect at each site was established for sampling aboveground vegetation, soil seed banks, and belowground bud banks. Thirty permanent plots (5 plots × 3 sites × 2 community types) was established. We collected samples after floodwaters receded (in early November 2017), during the coldest month (mid-January 2018), after spring sprouting (early March), and prior to flooding (early May 2018).

The bud/seed bank density was calculated as the number of buds/seeds per m². We analyzed the differences in seed and bud densities among different sampling seasons and between the two community types using linear mixed models, with community type and sampling season included as the main factors and sampling sites as a random factor. We performed multiple comparisons of means using Tukey’s honestly significant difference (HSD) test at a significance level of 0.05. If necessary, the data was transformed by using square root or log10 to reduce variances’ heterogeneity and tested homogeneity using Levene’s test.

The relationship between seed/bud densities and environmental variables was analyzed using the partial least squares (PLS) regression model, which avoids autocorrelation between independent variables (Huang et al., 2021). The variable importance value (VIP) and model coefficient (MC) were calculated for each independent variable. Variables with VIP values greater than or equal to 0.8 are considered important in the PLS model (Guo et al., 2015). A multiple regression model was used on data that did not conform to the PLS model to analyze the relationship between the dependent and independent variables. All statistical analyses was performed by using the program R (version 3.6.1; R Core Team, 2019).

Seed density varied significantly among sampling seasons, with significant interactions between the sampling season and community type ( Table 1 ). The seed densities of the Carex and Miscanthus communities were relatively low from November (after floodwaters receded) to March (441 ± 165 and 627 ± 98 seeds/m², respectively) and increased dramatically in May (prior to flooding) (15307 ± 4674 and 17330 ± 2486 seeds/m², respectively) ( Figure 2A ). The seed density of the Carex community did not differ significantly from that of the Miscanthus community, except in March ( Figure 2 ).

Bud density varied significantly among sampling seasons and between community types, with significant interactions ( Table 1 ). The bud densities of the Carex and Miscanthus communities peaked in January (980 ± 99 and 261 ± 62 buds/m², respectively), decreased markedly in March, and were the lowest in May (prior to flooding) (41 ± 9 and 52 ± 11 buds/m², respectively) ( Figure 2B ). The bud density of the Carex community was significantly higher than that of the Miscanthus community in all sampling seasons except May ( Figure 2 ).

In total, there were 23 and 19 species in the seed banks of the Carex and Miscanthus communities, respectively ( Table 2 ). However, the species composition of soil seed banks varied across seasons. In the Carex community, the number of species ranged from 6 to 13 and was at its lowest in January. In the Miscanthus community, the number of species ranged from 7 to 14 and was the lowest in November.

The soil seed bank of the Carex community was composed of 17 annual and 6 perennial species, and that of the Miscanthus community was composed of 15 annual and 4 perennial species. The seed density of perennial species was very low, usually less than 50 seeds/m² ( Table 2 ). The species present in the bud banks did not occur in the seed banks. Only one species, C. brevicuspis, was found in the bud bank of the Carex community. There were three species in the bud bank of the Miscanthus community: C. brevicuspis, M. sacchariflorus, and Phragmites australis.

There were nine annual and seven perennial species in the aboveground Carex community, with perennial species accounting for 62.5% of the total shoots ( Table 3 ). There were 16 annual and 11 perennial species in the aboveground Miscanthus community, with perennial species accounting for 44.8% of the total shoots ( Table 3 ).

VIP results showed that the seed density of the Miscanthus community was positively associated with aboveground shoot density and soil bulk density (VIP > 0.8; Table 4 ; Figure 3 ). The seed density of the Carex community was positively associated with plant biomass, total phosphorus, and moisture content (VIP > 0.8; Table 4 ; Figure 3 ). The bud density of the Miscanthus community was negatively correlated with soil total phosphorus and soil bulk density (P < 0.05; Table 4 ; Figure 3 ). The bud density of the Carex community was positively associated with shoot density and soil total phosphorus, and negatively associated with plant biomass, soil moisture content, and soil bulk density ( Table 4 ; Figure 3 ).

Seasonal changes in propagule bank densities and species compositions may indicate when a community will be most resilient or susceptible to disturbance (Ott et al., 2019). The seed densities of both plant communities were relatively constant from November to March and showed a marked increase in May. Variation in the density of soil seed banks in flooded wetlands depends mainly on seed rain, dispersal, germination, and mortality (Nielsen et al., 2018). During the long flooding season, many seeds are lost from the seed bank by flushing, predation, decomposition, and aging (Baskin et al., 2019). Therefore, the seed bank density declined after the flooding in November. Although many seedlings were observed after flooding, such as Cardamine hirsute, germination during this period did not cause a significant decline in the seed bank. The seed bank density remained low until May, when abundant seeds, such as Alopecurus aequalis and C. hirsute, were deposited onto the seed bank.

The bud densities of the two macrophyte communities peaked in winter and declined dramatically after spring sprouting. During the growing season before flooding (March–May), the buds continually sprouted into aboveground shoots, and we observed new ramets of M. sacchariflorus, even under the closed canopy. The bud formation rate of parent plants was lower than the rate of bud sprouting; therefore, bud density further decreased before the flooding season. A small bud bank can reduce maintenance costs during a long flooding season (Vesk & Westoby, 2004; Chen et al., 2015b), and it can ensure the regeneration of the aboveground shoots of M. sacchariflorus after flooding (Chen et al., 2015d).

Our first hypothesis, which stated that seed and bud bank densities peaked in the winter and were the lowest after the flood recession, was rejected. In the Dongting Lake wetlands, seasonal patterns of seed and bud densities showed nearly contrasting patterns; that is, seed bank density peaked before flooding and was lower in other seasons, whereas bud bank density peaked in the winter and was the lowest before flooding. The contrasting seasonal patterns of seed and bud densities may represent a life-history strategy of wetland plants for adapting to seasonal changes in wetland habitats, such as flooding and cold temperatures (Bao et al., 2018).

Our results indicate that seed bank density did not differ between the low-elevation Carex community and the high-elevation Miscanthus community. Seed banks in wetlands are linked to vegetation type, hydrological regime, or other types of disturbances (Abernethy & Willby, 1999; Hou et al., 2012; Greulich et al., 2019). Carex brevicuspis flowers, produces seeds and completes the first growing season before the flooding, whereas M. sacchariflorus flowers and fruits after the flooding in the fall (Xie & Chen, 2008). Some seeds were directly scattered in litters on the ground, while the others were buried in the soil due to the influence of the external environment (Vesk & Westoby, 2004). In the Dongting Lake wetlands, flood water may flush seeds, sediment, and litter to low-elevation sites when the flood rises (Hou et al., 2012). However, seeds may float for thousands of meters before sinking or being deposited, and deposited seeds may be resuspended and deposited downstream or elsewhere in frequently flood-disturbed floodplains (Gurnell et al., 2008; Fryirs & Carthey, 2022). Therefore, seed bank density may not differ among macrophyte communities along an elevational gradient in the Dongting Lake wetlands.

Nonetheless, we found that the low-elevation distributed Carex community had a larger bud bank than that of the high-elevation distributed Miscanthus community, which was consistent with a previous study (Chen et al., 2015b). A large bud bank may facilitate rapid regeneration of Carex after more frequent flooding disturbance at low-elevation sites, whereas Miscanthus may benefit from a small bud bank because of its lower production and maintenance costs and the low probability of flooding disturbance at high-elevation sites (Vesk & Westoby, 2004; Chen et al., 2015b). Overall, our second hypothesis, which stated that the low-elevation distributed Carex community has a larger seed and bud bank than the high-elevation distributed Miscanthus community, was only partially supported.

The seed banks of the two macrophyte communities were mainly composed of annual species. Some perennial species, such as Phalaris arundinacea and Paederia foetida, also appeared in the seed bank, but their densities were very low. The bud banks of the two macrophyte communities were composed of only three perennial species: C. brevicuspis, M. sacchariflorus, and P. australis. Unexpectedly, the three perennial species in the bud bank, which were also the most dominant species in the community, were not present in the seed bank. Consistent with a previous study, no seedling recruitment was found in mature populations of C. brevicuspis in the Dongting Lake wetlands (Deng et al., 2015). The absence or rarity of dominant perennials in the seed bank, which may be caused by low seed numbers and/or low seed viability and environmental factors such as flood regime (Fraaije et al., 2015; Cui et al., 2016), has also been observed in other wetlands and grasslands (Galatowitsch & van der Valk, 1996; Benson & Hartnett, 2006).

Therefore, our results suggest that seed and bud banks are complementary in species composition for the recruitment of macrophyte communities; that is, the seed bank ensures the persistence of annual species, and the bud bank enables the regeneration of dominant perennials. Vegetative reproduction via bud banks and sexual reproduction via seeds may constitute two different strategies for wetland plants to colonize flooding-disturbed wetland habitats (Klimešová and Klimeš, 2007; Herben et al., 2016). Bud banks mediate the demography of dominant perennials and maintain local population processes (Dalgleish & Hartnett, 2009), whereas seed banks contribute to the regeneration of annuals after flooding and are involved in regional processes through seed dispersal (Herben et al., 2016).

On a large scale, bud density is correlated with precipitation along a climatic gradient in the temperate steppe of northern China and the North American grasslands (Dalgleish & Hartnett, 2006; Qian et al., 2017). At the microsite scale, the soil water status is crucial for bud production, with overly wet conditions reducing the capacity of bud production in Leymus chinensis (Zhang et al., 2009). Our results indicate that seed and bud densities were influenced by different soil variables and differed between the two communities. However, the bud densities of both the Carex and Miscanthus communities were negatively correlated with soil bulk density. The seed bank of the Carex community was associated with soil moisture content and soil total phosphorus content, whereas that of the Miscanthus community was associated with soil bulk density. These results suggest that soil compactness is an important factor influencing propagule banks in flooded wetlands. Low soil bulk density often implies high soil porosity, aerobic rhizosphere conditions, and high soil organic matter content, which may facilitate rhizome growth and bud production (Deng et al., 2013). The complicated relationships between propagule banks and microhabitat factors need further investigation to understand vegetation succession in response to habitat changes in the context of hydrological and climatic changes.

To the best of our knowledge, the present study is the first to simultaneously investigate the abundance, species composition, and seasonal patterns of seed and bud banks. We obtained some unexpected results. Seasonal seed and bud densities showed nearly contrasting patterns for the Carex and Miscanthus communities in the Dongting Lake wetlands. Furthermore, the species composition of the seed and bud banks were complementary. Therefore, both seed and bud banks play important roles in the community recruitment of wetland macrophytes (i.e., bud banks regulate the demography of dominant perennials, and seed banks contribute to the recruitment and dispersal of annual species). The regeneration of macrophyte communities after disturbances depends primarily on the abundance and species composition of propagule banks. In recent years, seed banks have been widely used for wetland restoration (Nishihiro et al., 2006; Hanson et al., 2017). Given the high abundance of annuals and the near absence of most dominant perennials in the soil seed bank, bud banks of dominant perennial species should be introduced during the restoration of wetland vegetation.