2.1 Study area

Our study was conducted on Hainan Island (18°10′ N–20°10′ N, 108°37′ E–111°03′ E), which located in the northwest of the South China Sea. Hainan Island has a tropical monsoon climate with an average annual temperature of 24.2°C (Meng et al., 2022). The tide types vary from place to place, with a mean tide height of about 2 m. Annual precipitation ranges from 1000 to 2600 mm, with an average annual precipitation of 1639 mm. The rainy season is from May to October each year, with a total precipitation of about 1500 mm, accounting for 70%–90% of the total annual precipitation. The rainy season is from November to April of the next year. There are numerous bays, estuaries and long coastlines in Hainan Island, which provide superior conditions for the growth and reproduction of mangroves. The area of mangroves is 4710 ha (Fu, Tang, et al., 2021), which are mainly distributed in Dongzhaigang, Qinglangang, Sanyagang and Xinyinggang. Dongzhaigang National Nature Reserve is the first mangrove reserve area in China. Hainan Island has the highest biodiversity of all the mangroves in China. There are 37 species of mangroves in China, among which Hainan Island has the largest number of mangroves, with 26 true mangroves and 11 semi-mangroves (Bai et al., 2021). Not only that, the island has the highest carbon storage capacity of any mangrove in China (Liu et al., 2014). Therefore, Hainan Island is considered a hot spot for mangrove research and conservation (Chen, Gu, et al., 2021).

2.2 Sampling and biotic and abiotic factors

According to the distribution characteristics of mangroves in Hainan Island, a total of 56 plots were selected from 8 representative sites and 28 transects (Figure 1 and Table 1). The mangrove biodiversity and its environmental factors were investigated in the rainy and dry seasons of the east and west coasts from November 2021 to October 2022. Additional information on each study site is presented in Table 1.

For molluscs sampling, we set up 1–6 transects of 200–1500 m in length at each site, with an interval of at least 100 m between transects. Two sampling plots (including mangrove forest and mudflat) were set up in each transect. Five 25 cm × 25 cm quadrants were randomly selected from each sampling plot, with a depth of 30 cm to collect sediment samples. Each mud sample was panned with a 1-mm mesh sieve (Ma et al., 2020). Then, the mollusc samples were selected and fixed in 70% alcohol and brought back to the laboratory for classification and identification (Liu et al., 2016).

During the sampling period, the following environmental variables were measured in situ after biological samples were collected in each quadrat. Interstitial water salinity (IS) was measured using a portable salinometer (PAL-SALT Mohr, ATAGO, Japan) in each quadrat for three replicates. Latitude and longitude (coordinates) were determined by a hand-held GPS. Mangrove wetland area was obtained from National Earth System Science Data Center, National Science & Technology Infrastructure of China (http://www.geodata.cn).

Air-dried sediment samples were removed of impurities and pass through a 20-mesh sieve for subsequent processing. Sediment pH (SpH) was prepared according to the soil–water ratio of 2.5:1, and the pH of the extract was measured by a pH meter (Thunder magnetic pHS-3C, China). The median diameter (MD) of the sediment was measured using a laser particle size analyser (Malvern 2000, UK). The remaining sediment samples were passed through a 100-mesh sieve. Sediment organic carbon (TOC) and organic matter (TOM) were determined by potassium dichromate volumetric method. Sediment total nitrogen (TN) was determined by Kjeldahl method. The C/N ratio (CN) was then obtained by dividing sediment organic carbon by sediment total nitrogen.

Biotic factors are defined as the number of mangrove species (MS), mean tree height (H), mean diameter at breast height (DBH) and plant density (PD). The number of mangrove species was based on field survey records. From July to August 2023, three 5 m × 5 m plant quadrats were set in each plot (except mudflats). The DBH and tree height of all trees were measured. The DBH was generally 130 cm for trees and 30 cm for shrubs (Bai et al., 2021). Plant density was calculated as the number of individuals per quadrat divided by the area of the quadrat.

2.3 Statistical analyses

Spearman correlation analysis was used to explore the strength of the association between any two environment variables. Highly correlated environmental factors were eliminated, and variables with the lowest correlation (Spearman's rho < 0.7) were selected to be retained. To explain the importance of each set of variables, a variation partitioning analysis (VPA) was performed. VPA uses R_adj² to estimate unbiased interpretation rates of each variable combination (pure sediment factor, pure plant factor), which is a common method in biogeography research. The p-value of each combination is tested by a series of partial RDA models. In order to investigate the impact of remaining environmental factors on these biodiversity indicators, we calculated the Spearman correlation coefficient and its significance. All analyses in this part were conducted using ‘vegan’ and ‘stats’ in R v.4.1.2.

In order to explore the curve fitting model of the ratio of mudflat area to total area [x] and mollusc biodiversity [f(x)], the following mathematical fitting model was established:

The evaluation of each function involved considering the R² coefficient, standard error (SSE) and regression significance. Subsequently, the model demonstrating the best fit and the strongest correlation was chosen. This part was conducted using SPSS25.0.

3.1 Differences in mollusc biodiversity between unvegetated mudflat and mangrove forest

As shown in Figure 2, the abundance of molluscs in unvegetated mudflats was significantly higher than this in mangroves. However, species richness and functional diversity (FRic, FVu) of molluscs in unvegetated mudflats were significantly lower than those in mangrove forest. The comparison of CWM values of single trait between habitats showed that there was no significant difference in functional traits between mangroves and unvegetated mudflats (Table S2). In addition to differences in length, both unvegetated mudflats and mangrove forests were dominated by non-climbing, free-living, detrital and infaunal molluscs (Table S2).

3.2 Relative importance of biotic and abiotic drivers of mangrove mollusc biodiversity

Six environmental variables (IS, SpH, MD, TN, CN and PD) are selected to assess the relative importance of biotic and abiotic factors for mangrove molluscs through Spearman rank correlation test (Figure S2 and Figure 3b).

VPA (Figure 3a) showed that sediment characteristics have a greater influence on abundance (16.2%) and FVu (25.5%). Abundance was significantly positively correlated with IS and SpH and negatively correlated with CN, while FVu was significantly negatively correlated with IS and SpH and positively correlated with TN (Figure 3b). Species richness and FRic of molluscs were mainly affected by vegetation structure, with an independent contribution of 40.8% and 43.6% to the explained variance, respectively (Figure 3a). According to correlation analysis, PD was significantly positively correlated with both species richness and FRic (Figure 3b).

3.3 The optimal curve for the proportion of unvegetated mudflat area to total area and mangrove mollusc biodiversity

According to Table 2, mollusc abundance, species richness and function richness (FRic) were all the S function showing the best fit, while functional vulnerability (FVu) was the reciprocal function showing the best fit. With the increase of the proportion of unvegetated mudflat area, abundance, richness and FRic increased and FVu decreased (Figure 4). When the proportion of unvegetated mudflat area reached 20%, all indexes showed a slow increase or decrease trend and gradually became stable (Figure 4).

4.1 Differences in habitat use by molluscs in mangrove wetlands

Mangrove wetland in Hainan Island is an important habitat for molluscs. In this study, 88 species of molluscs were recorded, which were more species than Li et al. (2022). Our study confirmed marked differences in mollusc biodiversity between mangrove forests and unvegetated mudflats. Vegetated habitats support relatively higher taxonomic (species richness) and functional diversity (FRic) than unvegetated habitats (Figure 2). The studies by Nozarpour et al. (2024) and Leung and Cheung (2017) had similar results. And richness and FRic were more influenced by vegetation structure (Figure 3a). The complexity of vegetation structure was considered to be one of the factors affecting its diversity (Leung, 2015). For example, mangrove roots provide complex microhabitats that facilitated sediment deposition and enhance food availability, thereby reducing the top-down influence of predators (Corte et al., 2021; Hajializadeh et al., 2020). In addition, the density of mangrove plants was highly significantly positively correlated with the density of Ovassiminea brevicula in the mangrove forest of Avicennia alba in Thailand, possibly because the canopy provided microhabitats conducive to the growth of molluscs (Suzuki et al., 2002), which was consistent with the results of our study (Figure 3b). It had been found that high vegetation density leads to low underground animal community diversity, while medium and low vegetation density promoted underforest animal diversity (Zhang et al., 2003). The ground within mangrove forests was covered with a large amount of leaf litter, providing a substantial surface area for the attachment of macrobenthos (Nozarpour et al., 2024). However, research had shown that there was no significant difference in functional diversity between mangrove vegetation habitat and unvegetated habitat, which indicated that functional diversity may exhibit little or no dependence on vegetation structure (Nozarpour et al., 2023). Macrobenthic organisms possess the capacity to perceive and utilize habitats commensurate with their body size (Robson et al., 2005). Ecologically speaking, body size mirrors disturbance, movement of organic matter and biological interactions within a community (De Roos et al., 2003; Veríssimo et al., 2012). Our study showed that molluscs in unvegetated mudflats were larger in body size than those in mangrove forests (Table S2). This may be due to the fact that dense root structures could provide shelter from predators for small-sized species and limit the movement and feeding of large-sized species (Leung, 2015). In both unvegetated mudflats and vegetated habitats, the proportion of detritivores was higher (Table S2). Detritivores played an important role in nutrient cycling and energy transfer within mangrove ecosystems (Lee & Silliman, 2006). Many mangrove systems supported large communities of leaf litter that were consumed by species such as sesarmid and ocypodid crabs, which, in turn, controlled the fate and distribution of mangrove detritus, affecting detritivore communities (Lopez & Levinton, 2011).

Previous studies of mangrove macrobenthos had reported less abundance within mangrove forests than in adjacent habitats such as seagrass meadows and open sand/mudflats (Alfaro, 2006; Checon et al., 2017; Corte et al., 2021; Leung, 2015). Our results confirmed these findings, showing that the abundance within the mangrove forest was significantly less than that of the adjacent mudflats (Figure 2). Molluscs, especially gastropods, can reach high abundance and biomass in mangrove ecosystems (including mangrove forests and unvegetated mudflats) and occupy different levels in the food web, thus exerting a substantial influence on energy flow and material export to other ecosystems (Ebadzadeh et al., 2024). Sediment properties (mainly particle size, salinity, total organic carbon, total nitrogen and pH) were the main drivers of macroinvertebrate community assembly (Abdullah Al et al., 2022; Alsaffar et al., 2020; Hajializadeh et al., 2020; Wang et al., 2019), which supported our findings (Figure 3). Total organic carbon and total nitrogen serve as surrogate indicators of nutritional resources, fulfilling the energetic requirements of benthic organisms (Kröncke et al., 2003). The leaf litter within mangrove habitats had a high C/N ratio (Bouillon et al., 2002; Nerot et al., 2009; Schwamborn et al., 2002). Our study found a significant negative correlation between the C/N ratio and the abundance of molluscs (Figure 3b). Sediments with a lower C/N ratio generally indicated higher nitrogen content, which allowed animal communities to accumulate sufficient energy in a shorter time, thus enhancing reproductive rates (Lee, 2008). Fluctuations in salinity and pH affect the availability, accumulation and transport of heavy metals in some mangrove plants (Al-Asif et al., 2023), which in turn affect benthos. Ma et al. (2018) found that on the western coast of Hainan Island, the lower salinity due to freshwater inflow in Maniao Bay and Huachang Bay led to a decrease in both the number of mollusc species and their diversity. Changes in plant communities can also indirectly affect benthic ecological processes by altering environmental factors (Chen, Wang, Zhang, et al., 2020). The degradation of mangrove plant communities has become a global issue (Duke et al., 2007; Wang et al., 2020). During changes in mangrove plant communities, environmental factors such as hydrological conditions were altered, affecting associated marine organisms (Chen, Wang, et al., 2021). On one hand, mangroves often lead to habitat sediment acidification and salinization, and changes in plant communities often alter sediment acidification and salinization (Xiao et al., 2020). On the other hand, the mangrove wetlands on the western coast of Hainan Island have numerous aquaculture ponds and residential areas (Fu, Zhang, et al., 2021). The discharge of aquaculture wastewater and domestic sewage lowers water salinity and acidity, affecting the dispersion and colonization of benthic animals (Chen, Wang, Liu, et al., 2020).

Quantifying biodiversity vulnerability (i.e. the extent to which biodiversity and related functions are likely to change in the face of multiple threats) is essential to rationalize ecosystem management and conservation actions (Auber et al., 2022). The functional vulnerability of molluscs in mangrove habitats was significantly higher than that in unvegetated mudflats (Figure 2), suggesting that mangrove habitats may be less resistant to specific disturbances in the future, such as typhoons and extreme low temperatures and so on. Notably, even within biodiverse systems like temperate and tropical regions, the most abundant marine mammal communities exhibited heightened vulnerability (Parravicini et al., 2014), underscoring the notion that species richness and diversity ought not always to be construed as a protective shield but rather as a potential buffer against vulnerability.

In general, although the taxonomic diversity (species richness) and functional diversity (FRic) of molluscs in mangrove forest were higher than those in unvegetated mudflat regions, the unvegetated mudflat regions supported more abundance and had lower functional vulnerability. Therefore, the relative importance of mangrove forest and mudflat regions should be balanced in the process of mangrove wetland protection and restoration.

4.2 Implications of the proportion of mangrove forest and unvegetated mudflat area for mangrove wetland conservation and restoration

In recent years, the area of mangroves in the world has undergone substantial reduction, primarily attributed to land reclamation and aquaculture practices (Friess et al., 2019). Concurrently, urbanization, pollution and extreme climatic events have contributed to extensive mangrove degradation or mortality (Polidoro et al., 2010). Ambitious large-scale mangrove restoration targets have been set around the world (Friess et al., 2022). For example, Indonesia planned to restore 600,000 ha of mangroves by 2024 (Sasmito et al., 2023). In 2020, China introduced the Special Action Plan for Mangrove Protection and Restoration (2020–2025), which aimed to create and restore 18,800 ha of mangroves by 2025 (Choi et al., 2022; Jia et al., 2018). In 2022, Mozambique announced a plan to plant 50–100 million mangrove seedlings across 185,000 ha of land (Friess et al., 2022). The dramatic decline of mangrove area around the world was initially curtailed, with the annual decline rate of global mangrove area decreasing from 1%–2% to 0.16%–0.39% (Goldberg et al., 2020; Hamilton & Casey, 2016; Richards & Friess, 2016). Afforestation of mudflats represents the important approach to mangrove restoration (Agoramoorthy, 2012). Due to a lack of suitable coastal areas for restoration, efforts had been shifted to marginal spaces with less land use rights. The main reason the Philippines was keen on mudflats afforestation was that mudflats were open public land, avoiding issues related to land ownership (Primavera et al., 2014). This was also the case in China. From 2000 to 2019, China's mangrove area increased by about 8000 ha, with over 90% attributed to mudflats afforestation (Wang et al., 2021). However, although these areas had available space, the natural environment was biophysically unsuitable for the establishment and growth of mangroves (Gatt et al., 2022). Reports indicated that despite decades of efforts and funding, the success rate of mangrove restoration remained generally poor (Dale et al., 2014; Hashim et al., 2010; Kairo et al., 2001; Lewis, 2005; Primavera & Esteban, 2008). For example, from 1989 to 1995, West Bengal in India afforested 9050 hm², but only 138 hm² successfully became mangrove forests (Wang et al., 2021). The survival rate of artificially planted mangroves in the Philippines was only 10%–20% (Primavera & Esteban, 2008). An analysis of recent mangrove restoration projects in Sri Lanka found that out of 23 planting sites, 9 had a survival rate of 0%, and only 3 sites had a survival rate exceeding 50% (Kodikara et al., 2017). In Guangxi, China, the survival rate of mangrove afforestation from 2008 to 2015 was only 26.6% (Fan & Mo, 2018), and by 2019, the mangrove survival rate in Zhejiang Province, China, was approximately 22.4% (Chen et al., 2019).

Many factors behind low survival rates stem from the fact that incentive-based restoration (especially area-based or seedling-number-based targets) often provided poor incentives for restoration practitioners (Friess et al., 2022). However, despite the efficacy of these objectives and regulations in promoting mangrove expansion, certain problems and deficiencies persist. For example, the optimal allocation of mangrove and unvegetated mudflat area was not explored based on field investigation to balance the relative benefits of mangrove expansion and mudflat protection. Studies have shown a rapid decline in mudflats around the world over the past few decades (Murray et al., 2019). In addition to direct losses caused by coastal development, mangrove expansion may also contribute to the decline (Murray et al., 2019). Unvegetated mudflats are important habitats for benthic organisms and birds and serve as natural corridors for mangroves to cope with rising sea levels. Planting mangroves on mudflats was considered habitat conversion rather than habitat restoration (Ragavan et al., 2021). The transition from mudflat systems to mangrove forests is likely to entail shifts in hydrological conditions and biomes, including reduced flow rates (Lee & Shih, 2004), substitution of shorebirds and waterfowl by tree-dwelling egrets, reduced production of benthic relative to mangrove litter and alterations in benthic dominance from polychaetes and amphipods to crabs (Huang et al., 2012). Unvegetated mudflats support unique biomes, especially molluscs, which constitute a primary food source for shorebirds. Thus, some studies advocate utilizing historical baselines to guide mangrove reforestation efforts, preferably preserving both mangroves and unvegetated mudflats within the same area (Liu & Ma, 2024). Local managers and stakeholders often resort to the active removal of mangrove seedlings or the felling of mangrove stumps to impede mangrove expansion due to conflicting needs between mangroves and shorebirds in intertidal afforestation, yet such actions are not consistently supported by local policies (Choi et al., 2022). Consequently, during the initial phases of mangrove afforestation, a certain proportion of mudflat area should be conserved to maintain the connectivity and integrity of each geomorphic unit within the mangrove wetland, alongside augmenting mangrove forest area. Yang et al. (2021) had concluded that the appropriate ratio of mudflats area to mangrove area in mangrove ecosystem was 3.5662 or 3.9785 by exploring the richness and abundance of shorebirds. This means that the optimal mudflat area accounts for 78.1% or 79.9% of the total area. Notably, our study, focusing on benthic perspectives, advocates reserving at least 20% of the mudflat area for fauna in mangrove wetland restoration initiatives (Figure 4). This may be due to the fact that shorebirds have a larger home range than molluscs, so they require a larger area. Global data on long-term trends in mudflat ecosystems showed that the rate or scale of maintaining mudflat extent was insufficient to counteract the ongoing loss of these ecosystems (Murray et al., 2019). Given the exacerbation of climate change and anthropogenic disturbances, extensive and substantial conservation actions, restoration, supplementation and remediation are needed worldwide to address the multiple factors causing ecosystem loss. Additionally, it is imperative to integrate diversity considerations into policy formulation and management, particularly in decisions encompassing broader temporal and spatial scales. There is an urgent need for science-based mangrove and mudflat ecosystem protection and restoration efforts to halt and potentially reverse severe wetland degradation. Our findings furnish a scientific underpinning for such endeavours.