Phased ecological responses and effectiveness assessment of subtidal oyster reef restoration: a case study from Dashentang, Bohai Bay

Subtidal oyster reef restoration presents greater challenges for implementation and monitoring than intertidal restoration, and evaluations of relatively large-scale ecological restoration projects remain limited. This study evaluated the restoration project of Dashengtang using monitoring data from five field surveys collected from 2022 to 2026, and evaluated stage-specific ecological changes in oyster populations, aquatic environments, and macrobenthic communities. Oyster abundance generally increased following restoration. By March 2026, mean oyster density on natural and artificial reefs reached 1,830.4 and 286.0 ind. m -2, respectively. The recruitment density on artificial reefs increased to 204.2 ind. m -2, although the population remained dominated by small- and medium-sized individuals. Mean macrobenthic density increased to 255.6 ind. m -2 during post-restoration monitoring, and community composition differed significantly among monitoring periods. Water-quality variables showed different stage-specific patterns. Mean suspended particulate matter (SPM) concentration decreased from 19.3 mg L -1 before restoration to 7.28 mg L -1 in March 2026. Dissolved oxygen (DO) returned to levels comparable to pre-restoration conditions. Reactive phosphate remained low during the late post-restoration period, whereas chemical oxygen demand (COD), dissolved inorganic nitrogen (DIN), and chlorophyll-a fluctuated among monitoring periods. The Dashentang case provides insights into the implementation of large-scale subtidal oyster reef restoration under the specific environmental conditions of Bohai Bay. Future work should strengthen long-term continuous monitoring and observation of key ecological processes to further examine the relationship between environmental variability and oyster reef ecological development, evaluate the long-term stability and ecosystem services of artificial reefs, and progressively establish evaluation thresholds for different restoration stages to support adaptive management.

Oyster reefs are key coastal habitats composed of live oysters, dead oyster shells, and organisms inhabiting the reefs. They provide several ecosystem services, such as water filtration, provision of habitat, carbon storage, reduction of waves and shoreline stabilisation. Healthy oyster reefs provide habitat to fish, crustaceans, polychaetes and other benthic invertebrates. Using filtration by oysters, biodeposition and the structure of the reefs, they play an important role in modifying suspended particulate concentrations, nutrient cycling, and near-bottom hydrodynamic processes (Newell, 2004; Coen et al., 2007; Grabowski and Peterson, 2007; Fitzsimons et al., 2019). However, overfishing, coastal development, sediment accumulation, and water quality degradation have resulted in the loss of approximately 85% of oyster reefs worldwide, making them one of the most severely degraded coastal habitats (Beck et al., 2011). Oyster reefs were once widely distributed along the Chinese coastline. Among them, the Dashentang oyster reef in Tianjin is a representative northern subtidal oyster reef. Its area has declined from approximately 35 km 2 in the 1970s to 1.69 km 2 in recent years. Urgent conservation and restoration measures are therefore required to maintain the ecological functions of the remaining oyster populations (Zhao et al., 2025a).

Numerous field-based oyster reef restoration projects have been implemented worldwide in recent decades. Key indicators used to evaluate restoration performance include oyster density, biomass, recruit and adult composition, shell budget, and reef spatial extent (Oyster Metrics Workgroup, 2011; Baggett et al., 2015). Unlike intertidal reefs, subtidal oyster reefs remain continuously submerged, and their recovery is more strongly influenced by hydrodynamic conditions, sediment stability, sediment burial, larval dispersal, and regional environmental suitability, including salinity and DO. As a result of these constraints, the effectiveness of artificial reef restoration is highly site-specific and uncertain (Puckett et al., 2018; Graham et al., 2017; Lipcius et al., 2021). Current subtidal oyster reef restoration efforts have primarily involved engineering demonstrations and small-scale, site-specific trials, with particular attention to reef materials, habitat suitability, oyster colonisation, and biological communities associated with artificial reefs (Wang et al., 2022; Xu et al., 2023; Quan et al., 2025; Zhao et al., 2025a, 2025b). However, continuous monitoring and integrated assessments of oyster populations, the aquatic environment and macrobenthic communities following large-scale restoration remain limited.

In this study, the subtidal oyster reef restoration area at Dashentang, Tianjin, was selected as the study site. Using field-monitoring data collected from 2022 to 2026, the phased ecological responses to restoration were assessed in terms of oyster populations, the aquatic environment, and macrobenthic communities. The specific objectives were to: (1) compare temporal changes in oyster density, recruitment density, and population structure between natural and artificial reefs, and examine differences in population establishment; (2) assess temporal changes in macrobenthic density, wet biomass, diversity, and community composition; (3) analyze phased changes in major water-quality variables and their correlations with biological indicators; and (4) integrate responses across these ecological components to support restoration-effectiveness assessment and adaptive management of the Dashentang subtidal oyster reef.

Study area and restoration project

The study area is located in the nearshore environment to the south of Dashentang Village, in the Binhai New Area of Tianjin. The seabed is generally flat, with water depths of 2–5 m, and consists mainly of silty sand, clayey silt, and sandy silt. Historical fishing activities, shoreline development, and changes in seabed conditions have resulted in marked degradation of the natural oyster reef at Dashentang. To improve the oyster-reef habitat and facilitate population recovery, a subtidal oyster-reef restoration project was implemented in the study area from August 2022 to December 2023. This project is the largest oyster reef restoration project undertaken in northern China. A total of 4,500 reinforced-concrete artificial reef structures and 41,250 frame-type oyster-shell reef structures were deployed, creating an oyster-reef complex with a total volume of 345,000 m 3. The frame-type oyster-shell reefs were deployed using a large-scale, evenly dispersed configuration to increase the substrate available for oyster attachment and the habitat space available to benthic organisms. The reinforced-concrete artificial reefs were arranged primarily in double-layered strips around the periphery of the restoration area, with a denser configuration on the western side to support both oyster population recovery and ecological hazard-mitigation needs. The study area is shown in Figure 1. Details of structural design for restored reef units are shown in Supplementary Material S1.

Monitoring design and indicators

Based on five monitoring surveys conducted between 2022 and 2026, this study evaluated restoration effectiveness across the pre-restoration, construction, and post-restoration phases. Monitoring was conducted before restoration in May 2022, during construction in November 2023, during the early post-restoration period in March and June 2024, and during the late post-restoration period in March 2026. The monitoring stations are shown in Figure 1. Before restoration, six stations were established in areas where natural oyster reefs were concentrated. During construction and after restoration, ten replicate stations were established, comprising five stations each on natural and artificial oyster reefs. The natural reef served as an ecological reference within the restoration area. In accordance with the technical requirements specified in the “Technical Guideline for Investigation and Assessment of Coastal Ecosystems— Part 7: Oyster Reef” (China Association of Oceanic Engineering [CAOE], 2020), oyster-reef monitoring indicators included oyster density, recruitment, shell-height distribution, adult proportion, and empty-shell rate. Recruitment was quantified as the number of live oysters with a shell height of <20 mm, while the adult proportion was defined as the percentage of individuals with a shell height of ≥20 mm. The empty-shell rate was calculated as the ratio of oyster-shell weight to the combined weight of live oysters and oyster shells. Oyster reef area was mapped using a NORBIT iWBMS multibeam echosounder, with an unmanned surface vessel used as survey platform. Surface seawater was sampled and analysed in accordance with the “The Specification for marine monitoring—Part 4: Seawater analysis” (AQSIQ and SAC, 2007b) and “ Specifications for oceanographic Survey—Part 4: Survey of Chemical Parameters in sea water” (AQSIQ and SAC, 2007a). Monitoring parameters included temperature, salinity, pH, DO, SPM, COD, DIN, reactive phosphate, and chlorophyll-a. Macrobenthos were collected using a 0.1-m 2 grab sampler. Following sieving, fixation, taxonomic identification, enumeration, and weighing, density, wet biomass, and species richness were calculated.

Data processing and analysis

The analytical framework is shown in Figure 2. The indicators were classified into three domains, oyster populations, the aquatic environment, and macrobenthic communities, to assess population establishment, environmental responses, and the recovery of the macrobenthic community, respectively. Given the small sample size and the non-normal distributions of some indicators, Kruskal–Wallis tests were initially used to assess differences among restoration-stage × reef-type groups (Kruskal and Wallis, 1952). When an overall difference was significant, Dunn’s post hoc tests were performed (Dunn, 1964), and P values for multiple comparisons were adjusted using the Benjamini–Hochberg method (Benjamini and Hochberg, 1995).

Macrobenthic community composition was analysed using a site-species density matrix. Species-density data were square-root transformed before the Bray–Curtis dissimilarity matrix was calculated. Principal coordinates analysis (PCoA) was then used to visualise patterns of separation in community composition among samples from different restoration stages (Legendre and Gallagher, 2001). Permutational multivariate analysis of variance (PERMANOVA) was used separately for natural and artificial reefs to assess differences in community composition (Anderson, 2001), while permutational analysis of multivariate dispersions (PERMDISP) was used to evaluate differences in within-group dispersion (Anderson, 2006). For fixed monitoring stations, permutations were restricted by station. When the overall PERMDISP test was significant, pairwise comparisons between groups were further conducted and P values were adjusted using the Benjamini–Hochberg method. Similarity percentage (SIMPER) analysis was subsequently performed separately for natural and artificial reefs in March 2024 and March 2026 to identify the species contributing most strongly to differences in community composition between the two periods (Clarke, 1993). To minimise the influence of short-term construction disturbance on the comparisons of post-restoration community changes, samples collected during construction were excluded from the community–composition analysis.

Water-quality variables were used primarily to characterise temporal changes in the overall aquatic conditions of the restoration area. For statistical analysis, data from natural and artificial reef sites within the same survey period were pooled to compare overall aquatic conditions among survey periods. Associations between water-quality variables and biological indicators were examined using Spearman’s rank correlation, and P values for multiple comparisons were adjusted using the Benjamini–Hochberg method.

Oyster population characteristics

Oyster population parameters are presented in Figures 3 and 4. Across the monitoring periods, oyster density, recruitment density, and population structure of oysters showed significant temporal dynamic changes. The Kruskal–Wallis test results showed significant differences in both oyster density and recruitment density among the monitoring-period × reef-type groups (P < 0.001).

Before restoration, the average density of live oysters in natural reefs was only 80.33 ind m -2, while the empty-shell rate was as high as 88.59%. The population was dominated by adults, with 44.7% of individuals having shell heights exceeding 60 mm. During the construction period, mean oyster density on natural reefs increased to 1782.8 ind. m -2, and the average recruitment density reached 1398.6 ind. m -2. Individuals with shell heights less than 20 mm accounted for 77.69%. During the same period, the density of artificial reef oysters was 65.0 ind. m -2, and the recruitment density was 54.8 ind. m -2. Individuals with shell heights less than 20 mm accounted for 89.47%, and the empty shell rate was zero.

During the early post-restoration period, oyster density on the natural reef remained at the high level observed during construction. In March 2024 and June 2024, oyster density on natural reefs reached 1769.6 ind. m -2 and 1698.4 ind. m -2, respectively. The recruitment density decreased compared to the construction period, averaging 265.6 ind·m -2 by June 2024. The population structure shifted from mainly juvenile oysters with shell heights less than 20 mm in March to mainly adult oysters in June. During the same period, oyster density on artificial reefs remained relatively low. The densities in March and June 2024 were 58.8 ind·m -2 in March 2024 and 86.6 ind·m -2 in June 2024 respectively, showing no significant change compared to the construction period. The recruitment density decreased, averaging 6.8 ind·m -2 by June 2024. The population was dominated by adults with shell heights of 20–40 mm.

During the late post-restoration period in March 2026, the density of natural reef oysters remained at a relatively high level at 1,830.4 ind. m -2. Recruitment density increased compared to June 2024, averaging 379.4 ind. m -2. The adult proportion was 79.19%. Oyster density and recruitment density on artificial reefs increased compared to the early post-restoration period, reaching 286.0 and 204.2 ind. m -2, respectively. Individuals with shell heights of 20–40 mm accounted for 65.71%, whereas individuals >60 mm accounted for only 5.71%. The population therefore remained dominated by small and medium-sized individuals.

Dunn’s post hoc tests showed that oyster density on the natural reef was significantly higher than that of the artificial reef in November 2023, March 2024, and June 2024 (P_ adj = 0.0074, 0.0074, and 0.0180, respectively). However, the difference was no longer significant by March 2026 (P_ adj = 0.2017). Recruitment density was significantly higher on the natural reef during the construction period and in March 2024 (P_ adj = 0.0244 and 0.0013, respectively), but the differences between reef types were not significant in June 2024 and March 2026 (P_ adj = 0.0667 and 0.5190, respectively). The above results indicate that both oyster density and recruitment density of artificial reef oysters increased during the late post-restoration period. The difference in oyster density between reef types was no longer significant, although the artificial-reef population remained dominated by small- and medium-sized individuals.

According to the multibeam survey data (Figure 5), the coverage area of artificial oyster reefs in the deployment area was 64.7 hectares, and the natural reef area had also increased by 0.7 hectares compared to before restoration.

Macrobenthic community characteristics

The increase in macrobenthic density lagged behind the increase in oyster density (Figure 6). Before restoration, the macrobenthic density was 140.5 ind·m -2 and declined to 30.4 ind·m -2 during construction. Density remained low during the early post-restoration period in March and June 2024. By March 2026, however, mean density had increased to 255.6 ind·m -2, exceeding the pre-restoration level. From March 2024 to March 2026, mean macrobenthic density increased from 19.0 to 106.0 ind. m -2 on the natural reef and from 29.4 to 405.2 ind. m -2 on the artificial reef. Wet biomass and the Shannon–Wiener diversity index followed a similar pattern, initially declining and subsequently increasing.

The PCoA ordination (Figure 7a) showed partial separation of macrobenthic samples from different restoration stages. Pre-restoration samples were distributed mainly in the lower-left quadrant of the ordination plot, whereas samples from the early post-restoration period in 2024 were concentrated predominantly on the right. By March 2026, the samples had shifted towards the upper-left quadrant. PERMANOVA showed a significant difference in community composition on the natural reef between pre-restoration and the late post-restoration period (R ² = 0.340, P = 0.0028), consistent with the clear separation of the two periods in the PCoA. The PERMDISP test was not statistically significant (P = 0.886), indicating that the observed differences primarily reflected changes in species composition rather than differences in within-group dispersion. Among the three post-restoration monitoring periods, community composition on the natural reef differed significantly (R ² = 0.282, P = 0.0042), whereas PERMDISP was not significant (P = 0.0626). Community composition on the artificial reef also differed significantly among the three post-restoration monitoring periods (PERMANOVA, R ² = 0.305, P = 0.0026). PERMDISP indicated significant differences in within-group dispersion among monitoring periods (P = 0.0002). Mean distances to the group centroid were 0.586 in March 2024 and 0.398 in March 2026, and the pairwise difference between these periods was significant (P_ adj = 0.0469). Community-composition analysis (Figure 7b) showed that molluscs overwhelmingly dominated the macrobenthic community before restoration, accounting for 98.1% of relative abundance. During the early post-restoration period, the relative abundance of molluscs declined to 26.2% in March 2024, whereas those of echinoderms, arthropods, and annelids increased substantially. By June 2024, annelids had become the dominant taxon, accounting for 49.5% of the community. In March 2026, the relative abundance of molluscs had recovered to 66.3%, making them the dominant group once again, although annelids and arthropods remained relatively abundant. In terms of major species densities, mean density of Raeta pulchella on the natural reef decreased from 111.7 ind. m -2 before restoration to 6.0 ind. m -2 in March 2026. The species Theora lata and Ruditapes philippinarum, which were not recorded before restoration, reached 18.6 and 26.0 ind. m -2, respectively, in March 2026. Major species densities also changed substantially on the artificial reef. From March 2024 to March 2026, mean density of Ruditapes philippinarum increased from 4.8 to 115.2 ind. m -2. The average density of Arcuatula senhousia increased from 0 to 76.0 ind. m -2. The average density of Theora lata increased from 0 to 31.6 ind. m -2. SIMPER analysis (Figure 7c) showed that the major taxa contributing to changes in community composition from March 2024 to March 2026 differed between reef types. Molluscs and annelids were the main contributing phyla on both reef types, whereas the cumulative contribution of arthropods was more pronounced on the artificial reef and the contributions of nemerteans and echinoderms were relatively higher on the natural reef. Six of the ten leading contributing species were shared between reef types, with Theora lata and Ruditapes philippinarum both ranking in the top three.

Aquatic environmental characteristics

Figure 8 shows stage-specific changes in six major aquatic environmental variables across monitoring stages within the reef area. Kruskal–Wallis tests indicated overall differences among monitoring periods, with significant temporal differences in SPM, DO, COD, DIN, and reactive phosphate (P < 0.05). The mean concentration of SPM in the reef habitats was 19.3 mg L -1 before restoration and decreased to 7.28 mg L -1 by March 2026. DO concentrations declined slightly during construction and the early post-restoration period before recovering to 7.87 mgL -1 by March 2026, a value comparable to the pre-restoration level. COD and DIN fluctuated among monitoring periods. Reactive phosphate concentrations remained consistently low during the post-restoration stage. Chlorophyll-a concentrations increased during the early post-restoration period but subsequently returned to levels comparable to the pre-restoration baseline, with no significant overall difference among monitoring periods.

Correlation analysis (Figure 9) showed significant relationships between DO, SPM, DIN, reactive phosphate, and several biological indicators. Among them, DO was significantly positively correlated with oyster shell height, macrobenthic density and wet biomass (r = 0.52, r = 0.72, r = 0.62, respectively). There was a significant negative correlation between SPM and the density and biomass of macrobenthos (r = -0.48 and -0.45, respectively). DIN was significantly positively correlated with oyster shell height and density of macrobenthos (r = 0.54 and 0.50, respectively). A significant negative correlation (r = -0.56) between reactive phosphate and the density of macrobenthos was indicated. In contrast, COD and chlorophyll-a showed weaker correlations with the biological indicators.

Compared with intertidal oyster reefs, the outcomes of subtidal oyster-reef restoration are less readily observed and therefore require an integrated assessment of progress towards restoration objectives, ecological process responses, and the recovery of ecosystem functions. This study systematically evaluated stage-specific ecological dynamics during subtidal oyster-reef restoration at Dashentang across three dimensions: oyster population dynamics, aquatic environmental variability, and changes in macrobenthic communities. The principal findings and their ecological implications are discussed below.

Mechanisms of oyster population recovery and artificial-reef habitat formation

Our findings indicate substantial increases in oyster density following restoration at Dashentang. Previous surveys reported that the density of live oysters on natural reefs at Dashentang ranged from 237 to 2,680 ind. m -2 between 2021 and 2023 (Liu et al., 2026). In this study, mean oyster density on natural reefs in March 2026 was relatively high at 1,830.4 ind. m -2, whereas that on artificial reefs was 286 ind. m -2, close to the lower limit of the previously reported range. Although the higher adult populations on natural reefs may serve as potential larval sources for surrounding waters, transport and settlement of larvae onto artificial reefs also depend on the hydrodynamic connectivity between the reefs and larval transport pathways (Puckett et al., 2018; Lipcius et al., 2021). Therefore, recruitment on artificial reefs not only depends on the size of the surrounding adult populations, but may also be limited by the larval input and transport efficiency.

At the end of the construction period, oyster recruitment was observed on artificial reefs, with a density of 54.8 ind. m -2. However, recruitment density declined to 6.8 ind. m -2 by June 2024 during the early post-restoration period. Several mechanisms may have contributed to this pattern. First, sediment deposition and resuspension may be one of the important factors affecting the survival of oyster larvae in this restoration area. The study area has a shallow water depth and is dominated by fine unconsolidated substrate. Under the influence of near-bottom water flow, sediment deposition and resuspension are prone to occur. Sediment burial may reduce the suitable attachment surfaces and reduce the survival rate of newly colonized oysters through burial and interference with filter feeding, thereby increasing post-settlement mortality (Lenihan, 1999; Esquivel-Muelbert et al., 2022; Caretti et al., 2024). The height and three-dimensional structure of the reef can also affect the accumulation of sediment on the reef surface and the subsequent survival of larvae by altering the near-bottom flow field and sediment transport.

In addition, the artificial reef substrate may also affect the settlement and subsequent survival of oyster larvae. Oyster larvae preferentially settle on shell substrates coated with biofilms. Biofilms, epibionts, and shell accumulations on the newly deployed artificial reef surface require a certain period of time to develop (Tamburri et al., 2008; Walles et al., 2016). By March 2026, while recruitment density of artificial reefs had risen to 204.2 ind. m -2, the population remained dominated by small and medium-sized oysters. One possible explanation is the gradual development of substrate conditions and microhabitats on artificial reefs. Additionally, biological processes such as predation and spatial competition may also affect the survival of oysters after settlement (Nestlerode et al., 2007; Esquivel-Muelbert et al., 2022).

Recovery of macrobenthic communities and restoration of habitat function

Our results show that the density, wet biomass, and diversity of macrobenthos were low both during the construction phase and in the early stage of restoration. A marked recovery was observed by March 2026. Mean macrobenthic density in the restoration area reached 255.6 ind. m -2 in March 2026, which was relatively high compared with historical survey results of the area (Liu et al., 2026). Previous oyster reef restoration studies have shown that the density and diversity of reef-associated macroinvertebrates can increase accordingly as reef structure and microhabitats develop. However, the response rates among taxa are not consistent (Rodney and Paynter, 2006; Grabowski and Peterson, 2007; Searles et al., 2022).

Community-composition and ordination analyses have further indicated considerable changes within the macrobenthic community. Before the restoration process, the macrobenthic community was composed mainly of molluscs, with the deposit-feeding bivalve Raeta pulchella as the main dominant species. By March 2026, the density of this species had decreased markedly, while bivalves such as Ruditapes philippinarum and Theora lata had increased. Raeta pulchella mainly utilizes organic matter in muddy sediments (Morton, 2010). Ruditapes philippinarum mainly filters suspended particles in the water (Koo and Seo, 2020). Theora lata can contribute to material exchange across the sediment–water interface through deposit feeding and bioturbation (Yamada and Kayama, 1987). In addition to bivalves, annelids, arthropods, and echinoderms were more strongly represented after restoration, indicating broader taxonomic and trophic representation within the community.

Post-restoration patterns of community change also differed between natural and artificial reefs. SIMPER analysis showed that molluscs and annelids were the major contributors to community dissimilarity between March 2024 and March 2026 in both reef types. However, the cumulative contribution of arthropods was higher on the artificial reef and the contributions of nemerteans and echinoderms were relatively higher on the natural reef. The community changes in the later stage of the artificial reef also involved various feeding modes and life forms, including the suspension-feeding Arcuatula senhousia, the tube-dwelling amphipod Monocorophium insidiosum, and the surface deposit-feeding annelid Aphelochaeta multifilis (Crooks, 1998; Nair and Anger, 1979; Rengaiyan et al., 2017). In addition to changes in taxonomic composition and life forms, PERMDISP analysis indicated significant differences in multivariate dispersion among post-restoration periods on the artificial reef, with a lower mean distance to the group centroid in March 2026 than in March 2024. During the same period, major contributing species identified by SIMPER, such as Ruditapes philippinarum, Theora lata, and Arcuatula senhousia, were recorded in more artificial-reef stations, which is consistent with the PERMDISP results.

These changes in community composition and life forms may be related to the structure and developmental stage of the oyster reef habitat. The three-dimensional structure created by live oysters, empty shells, and epibionts increases bottom complexity, provides hard substrate for colonisation, and creates interstitial spaces that offer attachment surfaces, refuge, feeding, and reproduction grounds for molluscs, crustaceans, polychaetes, and other small invertebrates. Furthermore, the filtering activity of oysters and their biodeposition may affect the movement of organic material between water and sediment, thus providing food and energy for benthic organisms belonging to different feeding types (Grabowski and Peterson, 2007; Coen et al., 2007). Previous studies on natural oyster reefs have shown that the reef habitat can support relatively high abundances of benthic groups such as echinoderms, arthropods, and molluscs (Crawford et al., 2020). Restoration studies in Chesapeake Bay have also shown higher numbers of benthic macrofauna, sessile organisms, crabs, and demersal fishes in restored reefs than in adjacent non-restored areas (Rodney and Paynter, 2006). Similarly, the contribution of multiple feeding modes and life forms to late-stage community change in this study is consistent with a progressive increase in micro-habitats as the three-dimensional reef structure develops.

Changes in the aquatic environment and their implications for the ecological functions of oyster reefs

Water-quality variables showed different stage-specific changes before and after restoration, among which the concentration of suspended solids decreased significantly. Oysters can reduce particulate concentration in the water column through filtration of phytoplankton, organic debris and inorganic particulates. Through biodeposition, oysters transfer filtered particles to the seabed as faeces and pseudofaeces, thereby influencing material cycling across the sediment–water interface (Newell, 2004). Studies of natural and artificial oyster reefs show that reef restoration can result in a reduction of suspended particulates and chlorophyll-a, although their removal rate is dependent on the number of oysters, reef size and hydrodynamic conditions (Grizzle et al., 2008). In this study, mean SPM decreased from approximately 19.3 mg L -1 before restoration to 7.28 mg L -1 in March 2026 and was significantly negatively correlated with macrobenthic density and wet biomass. Lower SPM concentrations can reduce fine-particle deposition on shell substrates and reduce interference with filtration and respiration. This indicates that particle concentration may be an important environmental factor affecting benthic habitat conditions in the restoration area (Wilber and Clarke, 2010; Whitman and Reidenbach, 2012). These observations are consistent with the expected effects of oyster filtration and sediment stabilization, although these processes were not directly quantified in this study.

DO concentrations dropped during construction and early restoration but increased to pre-restoration levels during the late restoration phase. DO was positively correlated with the oyster shell heights, macrobenthic density and wet biomass. Low DO can constrain oyster feeding, growth, and energy acquisition and may alter macrobenthic behaviour and survival (Breitburg, 2002; Diaz and Rosenberg, 2008; Vaquer-Sunyer and Duarte, 2008; Porter and Breitburg, 2016). Reef structure and elevation can enhance near-bottom flow and reduce exposure to hypoxic conditions (Lenihan, 1999).

Temporal patterns in nutrients, COD, and chlorophyll-a were more complex. Reactive phosphate and DIN were significantly correlated with some biological indicators, while COD and chlorophyll-a showed weaker correlations. Nutrients and chlorophyll-a in coastal waters are jointly influenced by terrestrial input, tidal exchange, vertical mixing of the water column, light levels, and resuspension of sediments (Cloern, 2001; Monbet, 1992; Nixon, 1995; Rios-Yunes et al., 2023). COD is more influenced by terrestrial organic-matter inputs and hydrodynamic transport. Studies in the Bohai Sea have shown that coastal river inputs, as well as tidal currents and residual circulation, can influence COD fluxes and spatial distributions (Li et al., 2022; Chi et al., 2020). Historical surveys at Dashentang also revealed marked interannual variation in nutrients, COD, and chlorophyll-a, with long-term influences from terrestrial inputs and nearshore hydrodynamic processes (Xu et al., 2022). Oyster reefs can themselves contribute to nearshore nutrient cycling through filtration, biodeposition, nutrient retention, and sediment denitrification (Kellogg et al., 2013). The processes of oyster filtration, biodeposition, and exchange across the sediment–water interface may further influence particle deposition, oxygen conditions, and nutrient cycling.

Implications of subtidal oyster reef restoration and limitations of this study

Our findings suggest that in cases where bay environments have suffered extensive anthropogenic impacts like Bohai Bay, the conservation of natural oyster reefs should be incorporated into the overall design of subtidal oyster reef restoration programme. Given the differences in oyster population establishment between natural and artificial reefs, subtidal oyster-reef restoration planning should consider not only the deployment area, number, and configuration of artificial reefs, but also the distribution of natural reefs, recruitment, and connectivity among reef units. From the perspective of adaptive management, the formulation of restoration goals should closely follow the core ecological processes and be dynamically adjusted according to the restoration stages. Early recovery should focus on oyster settlement and population establishment. In the middle stage, attention should be paid to the reef structure, sediment environment and habitat changes. At later stages, evaluations should focus on the ecosystem functions and services such as population maintenance and water purification. If insufficient larval recruitment, sediment burial of reefs or limited population establishment is found, targeted intervention can be carried out through measures such as supplementing suitable substrate, adjusting reef height and layout, and protecting adjacent natural reef populations.

This study has several limitations. First, the current monitoring period was relatively short compared with the time required for artificial reef structures to develop into self-sustaining biogenic reefs, limiting a comprehensive assessment of their long-term stability. Furthermore, this study did not include external control sites outside the restoration area, and therefore a strict BACI (Before–After–Control–Impact) framework could not be applied to evaluate restoration effects. The survey stations of natural reefs before and after restoration were not completely consistent, and thus the corresponding comparisons mainly reflected overall changes at the reef scale. Due to differences in survey timing and temporal coverage, available historical regional data were only used to provide long-term ecological background for the Dashentang and could not serve as contemporaneous controls. Besides, surveys conducted at different stages covered different seasons and years, and thus some biological and environmental variations might have been simultaneously influenced by seasonal fluctuations and interannual variability. Although surveys conducted in the same month (March 2024 and March 2026) improved comparability between post-restoration years, the effects of interannual variation and restoration could not be completely separated. Spearman correlation analysis primarily reflects statistical associations between environmental factors and biological responses, and some correlations may have been influenced by concurrent temporal changes among multiple environmental variables.

Second, several key ecological processes and ecosystem services were not directly quantified in this study. Oyster filtration capacity, nutrient fluxes across the sediment–water interface and denitrification rates were not measured directly. Quantitative assessments of ecosystem services, including water purification and nutrient removal, therefore remain limited. Meanwhile, quantitative field measurements of artificial-reef subsidence and sediment burial were also lacking, limiting the understanding of physical reef development and its potential influence on oyster population establishment. Future work could combine synchronous monitoring of planktonic oyster larvae with standardized settlement plates on natural and artificial reefs, together with measurements of post-settlement survival and reef-surface sedimentation. Persistent differences in larval abundance and initial settlement between artificial and natural reefs would indicate potential limitation by larval input, whereas similar initial settlement but reduced survival associated with increased sediment accumulation would provide stronger evidence for sediment burial effects.

Third, this study has not yet established quantitative thresholds for cross-stage ecological restoration assessment. Long-term monitoring, combined with long-term data on natural-reef condition and the recovery trajectory of artificial reefs, could be used to progressively establish stage-specific evaluation thresholds for population-establishment indicators (e.g., oyster density, recruitment density, and proportion of mature oysters), structural-stability indicators (e.g., reef area and elevation), and indicators of community condition and ecological function. Such thresholds would support restoration-effectiveness assessment and adaptive management.

Monitoring results from 2022 to 2026 indicated clear stage-specific changes in oyster populations, macrobenthic communities, and the water environment following subtidal oyster reef restoration at Dashentang. By March 2026, oyster density on the natural reef had increased substantially relative to the pre-restoration period. Oyster density and recruitment density on the artificial reef had increased relative to the early post-restoration period, but the population remained dominated by small and medium-sized individuals. Changes in the macrobenthic community were comparatively delayed. The density remained relatively low during the early post-restoration period but increased to 255.6 ind. m -2 by March 2026. Macrobenthic community composition differed significantly among monitoring periods, mainly through changes in the relative abundance of dominant taxa and the composition of major species. Water-quality variables also showed stage-specific patterns. SPM showed significant temporal variation and was lower in March 2026 than before restoration. DO returned to near the pre-restoration level during the late post-restoration period. Other variables fluctuated to varying degrees. Overall, the results indicate positive ecological responses following oyster-reef restoration, while these responses should be interpreted within the context of seasonal and interannual environmental variability. Future management should continuously integrate natural reef conservation with artificial reef deployment, and assess restoration effectiveness and ecological functions through long-term monitoring and phased evaluations.