Sai Wang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Sai Wang has a total of 148 co-authors. Here are the top co-authors by number of shared publications:
- Tuan-Tuan Wang
School of Ecology, Hainan University, Haikou, China
12 shared publications Tuan-Tuan Wang
Center for Eco-Environment Restoration Engineering of Hainan Province, School of Ecology, Hainan University, Haikou 570228, PR China
8 shared publications- Yang Yang
Engineering Research Center of Tropical and Subtropical Aquatic Ecological Engineering, Ministry of Education, Institute of Hydrobiology, Jinan University, Guangzhou 510632, China
6 shared publications - Zhongbing Chen
Department of Applied Ecology, Faculty of Environmental Sciences, Czech University of Life Sciences Prague, Kamýcká 129, Praha-Suchdol 16500, Czech Republic
6 shared publications Ling Mo
Hainan Research Academy of Environmental Sciences, Haikou 571126, China
5 shared publications- Dong-Hai Wu
State Key Laboratory of Marine Resource Utilization in South China Sea Hainan University Haikou China
4 shared publications
Latest Publications
- Open Access
- 17 November 2026
- Published by Wiley in Ecology and Evolution
- Vol. 16 (11)
The Conservation of Seagrass Meadows in the West Sand of Qilianyu Islands Based on the Perspective of Food‐Web Trophic Network and Energy Flow
Seagrass meadows provide critical habitats for marine biodiversity but face accelerating global declines. Using Ecopath with Ecosim (EwE) modeling, we evaluated food web structure, energy flow, and ecosystem health across four habitat types (coral‐seagrass transition, dense, sparse, and bare seagrass), with four replicate sampling stations per habitat, in the West Sand of Qilianyu Islands, South China Sea. Models comprised 20–32 functional groups, with trophic levels constrained below 4.0 due to the absence of apex predators. Transfer efficiencies followed a clear habitat gradient (dense seagrass > coral‐seagrass transition > sparse seagrass > bare seagrass), reflecting the influence of seagrass structural complexity on energy transfer. Keystone species analysis revealed piscivorous fish exerting top‐down control in healthy meadows, shifting to zooplankton and cephalopods as keystone groups in degraded areas. Ecosystem health indices (ascendency/capacity, connectance, omnivory) confirmed a gradient from coral‐seagrass transition (most mature) to bare seagrass (most degraded). Thirty‐year Ecosim simulations predicted seagrass degradation would decrease higher trophic level biomass while increasing primary producers, whereas restoration reversed this cascade. Ecospace spatial simulations showed seagrass loss leads to local functional collapse, while restoration increases carrying capacity and restores habitat connectivity. Our findings demonstrate that seagrass meadows function as ecological engineers whose presence fundamentally determines food web structure and stability. The coral‐seagrass transition zone exhibited superior ecosystem health, highlighting the synergistic benefits of habitat connectivity. These results provide quantitative evidence supporting seagrass conservation prioritization, spatial planning that maintains habitat connectivity, and adaptive management strategies to preserve these critical coastal ecosystems.
- 17 November 2026
- Published by Wiley in Ecology and Evolution
- Vol. 16 (11)
Open AccessThe Conservation of Seagrass Meadows in the West Sand of Qilianyu Islands Based on the Perspective of Food‐Web Trophic Network and Energy Flow
Seagrass meadows provide critical habitats for marine biodiversity but face accelerating global declines. Using Ecopath with Ecosim (EwE) modeling, we evaluated food web structure, energy flow, and ecosystem health across four habitat types (coral‐seagrass transition, dense, sparse, and bare seagrass), with four replicate sampling stations per habitat, in the West Sand of Qilianyu Islands, South China Sea. Models comprised 20–32 functional groups, with trophic levels constrained below 4.0 due to the absence of apex predators. Transfer efficiencies followed a clear habitat gradient (dense seagrass > coral‐seagrass transition > sparse seagrass > bare seagrass), reflecting the influence of seagrass structural complexity on energy transfer. Keystone species analysis revealed piscivorous fish exerting top‐down control in healthy meadows, shifting to zooplankton and cephalopods as keystone groups in degraded areas. Ecosystem health indices (ascendency/capacity, connectance, omnivory) confirmed a gradient from coral‐seagrass transition (most mature) to bare seagrass (most degraded). Thirty‐year Ecosim simulations predicted seagrass degradation would decrease higher trophic level biomass while increasing primary producers, whereas restoration reversed this cascade. Ecospace spatial simulations showed seagrass loss leads to local functional collapse, while restoration increases carrying capacity and restores habitat connectivity. Our findings demonstrate that seagrass meadows function as ecological engineers whose presence fundamentally determines food web structure and stability. The coral‐seagrass transition zone exhibited superior ecosystem health, highlighting the synergistic benefits of habitat connectivity. These results provide quantitative evidence supporting seagrass conservation prioritization, spatial planning that maintains habitat connectivity, and adaptive management strategies to preserve these critical coastal ecosystems.
- Open Access
- 9 November 2026
- Published by Frontiers Media SA in Frontiers in Ecology and Evolution
- Vol. 14, 1960019
eDNA metabarcoding and random forest identify super-connector fish species across mangrove–seagrass–coral reef continuums in the South China Sea
Background: Tropical coastal seascapes rely on ecological connectivity between mangrove, seagrass, and coral reef habitats to sustain fish biodiversity and fisheries productivity. However, identifying key fish species that facilitate cross-habitat energy flows remains challenging due to limitations of traditional survey methods in turbid, structurally complex environments. Methods: We integrated environmental DNA (eDNA) metabarcoding with Random Forest (RF) machine learning to map fish assemblages across a mangrove-seagrass-coral reef continuum in Hainan Island, South China Sea. Results: We detected 234 fish species across 78 families - a 4.03-fold increase over conventional gillnet surveys (58 species, paired t-test, t = 11.34, p < 0.001). Using RF classification, we identified 23 "connector species" with high cross-habitat fidelity, of which three super-connectors - Lutjanus russellii , Siganus fuscescens , and Gerres oyena - accounted for 47% of cross-habitat network flow. Simulated removal of these three species reduced overall network connectivity by 68%. The RF model achieved 90.7% accuracy (Kappa = 0.86) in predicting habitat occurrence from eDNA data. Variable importance analysis revealed a salinity threshold at 28 PSU and a distance-to-reef threshold at 2.5 km as primary drivers of community turnover. Our bipartite network analysis showed significant nestedness (NODF = 47.3, p < 0.001), indicating that generalist species disproportionately maintain seascape connectivity. Discussion: These findings demonstrate that integrating eDNA with machine learning provides a scalable, high-resolution toolkit for defining functional connectivity and prioritizing species for ecosystem-based fisheries management in the Anthropocene.
- 9 November 2026
- Published by Frontiers Media SA in Frontiers in Ecology and Evolution
- Vol. 14, 1960019
Open AccesseDNA metabarcoding and random forest identify super-connector fish species across mangrove–seagrass–coral reef continuums in the South China Sea
Background: Tropical coastal seascapes rely on ecological connectivity between mangrove, seagrass, and coral reef habitats to sustain fish biodiversity and fisheries productivity. However, identifying key fish species that facilitate cross-habitat energy flows remains challenging due to limitations of traditional survey methods in turbid, structurally complex environments. Methods: We integrated environmental DNA (eDNA) metabarcoding with Random Forest (RF) machine learning to map fish assemblages across a mangrove-seagrass-coral reef continuum in Hainan Island, South China Sea. Results: We detected 234 fish species across 78 families - a 4.03-fold increase over conventional gillnet surveys (58 species, paired t-test, t = 11.34, p < 0.001). Using RF classification, we identified 23 "connector species" with high cross-habitat fidelity, of which three super-connectors - Lutjanus russellii , Siganus fuscescens , and Gerres oyena - accounted for 47% of cross-habitat network flow. Simulated removal of these three species reduced overall network connectivity by 68%. The RF model achieved 90.7% accuracy (Kappa = 0.86) in predicting habitat occurrence from eDNA data. Variable importance analysis revealed a salinity threshold at 28 PSU and a distance-to-reef threshold at 2.5 km as primary drivers of community turnover. Our bipartite network analysis showed significant nestedness (NODF = 47.3, p < 0.001), indicating that generalist species disproportionately maintain seascape connectivity. Discussion: These findings demonstrate that integrating eDNA with machine learning provides a scalable, high-resolution toolkit for defining functional connectivity and prioritizing species for ecosystem-based fisheries management in the Anthropocene.
- 10 October 2026
- Published by Springer Nature in Aquatic Sciences
- Vol. 88 (4) , 140
Spatiotemporal patterns of fish community composition and structure in a subtropical reservoir: comparison of eDNA metabarcoding and gillnetting
- 10 October 2026
- Published by Springer Nature in Aquatic Sciences
- Vol. 88 (4) , 140
Spatiotemporal patterns of fish community composition and structure in a subtropical reservoir: comparison of eDNA metabarcoding and gillnetting
- Open Access
- 9 July 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
A coupled LSTM model for predicting blue carbon and fishery dynamics in tropical coastal wetlands under climate change
Coastal wetlands, including mangroves, seagrass beds, and coral reefs, provide critical blue carbon sequestration services and nursery habitats that support fishery resources. However, the bidirectional coupling between blue carbon stocks and fishery abundance remain poorly quantified, particularly under accelerating environmental change in tropical marginal seas. This study presents the first coupled LSTM framework for bidirectional blue carbon-fishery prediction in tropical coastal wetlands. We developed a coupled Long Short-Term Memory (LSTM) neural network model to predict the bidirectional relationship between blue carbon stocks and fishery resource abundance in the coastal ecosystems of Guangdong Province and Hainan Island, China. Field surveys were conducted at 15 representative sites from 2018 to 2025, generating 96 monthly observations of environmental variables (sea surface temperature, salinity, dissolved oxygen, turbidity, nutrients), biological indicators (mangrove above-ground biomass, soil organic carbon, seagrass coverage, coral cover), and fishery metrics (catch per unit effort, species richness, juvenile abundance). The LSTM model achieved superior predictive performance compared to baseline methods, with root mean square error of 0.142 Mg C ha⁻¹ for blue carbon prediction (R² = 0.93) and 0.108 kg h⁻¹ for CPUE prediction (R² = 0.89). Feature importance analysis revealed that mangrove soil organic carbon was the strongest predictor of fishery CPUE (Shapley value = 0.187), while sea surface temperature exerted the greatest influence on blue carbon stock variability. Scenario simulations for 2025–2026 indicate that under a moderate warming scenario (SST increase of 0.8 °C), blue carbon stocks are projected to decline by 7.2% (95% CI: 5.8–8.6%), with associated fishery CPUE reductions of 11.4% (95% CI: 9.2–13.6%). These findings provide a novel predictive framework for ecosystem-based management of tropical coastal wetlands and highlight the vulnerability of coupled social-ecological systems to climate change.
- 9 July 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Open AccessA coupled LSTM model for predicting blue carbon and fishery dynamics in tropical coastal wetlands under climate change
Coastal wetlands, including mangroves, seagrass beds, and coral reefs, provide critical blue carbon sequestration services and nursery habitats that support fishery resources. However, the bidirectional coupling between blue carbon stocks and fishery abundance remain poorly quantified, particularly under accelerating environmental change in tropical marginal seas. This study presents the first coupled LSTM framework for bidirectional blue carbon-fishery prediction in tropical coastal wetlands. We developed a coupled Long Short-Term Memory (LSTM) neural network model to predict the bidirectional relationship between blue carbon stocks and fishery resource abundance in the coastal ecosystems of Guangdong Province and Hainan Island, China. Field surveys were conducted at 15 representative sites from 2018 to 2025, generating 96 monthly observations of environmental variables (sea surface temperature, salinity, dissolved oxygen, turbidity, nutrients), biological indicators (mangrove above-ground biomass, soil organic carbon, seagrass coverage, coral cover), and fishery metrics (catch per unit effort, species richness, juvenile abundance). The LSTM model achieved superior predictive performance compared to baseline methods, with root mean square error of 0.142 Mg C ha⁻¹ for blue carbon prediction (R² = 0.93) and 0.108 kg h⁻¹ for CPUE prediction (R² = 0.89). Feature importance analysis revealed that mangrove soil organic carbon was the strongest predictor of fishery CPUE (Shapley value = 0.187), while sea surface temperature exerted the greatest influence on blue carbon stock variability. Scenario simulations for 2025–2026 indicate that under a moderate warming scenario (SST increase of 0.8 °C), blue carbon stocks are projected to decline by 7.2% (95% CI: 5.8–8.6%), with associated fishery CPUE reductions of 11.4% (95% CI: 9.2–13.6%). These findings provide a novel predictive framework for ecosystem-based management of tropical coastal wetlands and highlight the vulnerability of coupled social-ecological systems to climate change.
- 6 February 2026
- Published by Elsevier in Marine Environmental Research
- Vol. 216, 107907
Presence of per- and polyfluoroalkyl substances (PFASs) in the Nandu River Estuary, Hainan Island
- 6 February 2026
- Published by Elsevier in Marine Environmental Research
- Vol. 216, 107907
Presence of per- and polyfluoroalkyl substances (PFASs) in the Nandu River Estuary, Hainan Island
- 26 January 2026
- Published by Elsevier in Journal of Hazardous Materials
- Vol. 504, 141263
Nonnegligible migration of rubber-derived chemicals from road surface to adjacent greenbelt soil
- 26 January 2026
- Published by Elsevier in Journal of Hazardous Materials
- Vol. 504, 141263
Nonnegligible migration of rubber-derived chemicals from road surface to adjacent greenbelt soil
- Open Access
- 30 September 2025
- Published by Wiley in Ecology and Evolution
- Vol. 15 (10)
The Distribution, Diversity, and Indicator Species of Coral Communities Under the Influence of Environmental Changes in the Subtropical Peninsula of Southern China
Coral reefs are among the most biodiverse and economically important marine ecosystems worldwide. Recently, these ecosystems have faced threats and degradation from multiple factors. In particular, coral species living at the edges of subtropical zones are important “barometers” for indicating the global boundaries of coral communities. From 2023 to 2024, we investigated the coral species composition and coverage surrounding the Dapeng Peninsula, which is located at the northern edge of the subtropical zone, and explored the influence of environmental factors on the coral community structure. Fifty‐one species belonging to 20 genera and 13 families were identified in five habitats (conservation, wind‐wave, tourist, fishery, and pristine areas) across three seasons. Owing to rapid urbanization and global warming, the dominant coral species shifted from branching types, which are essential reef‐builders for creating structural substrates, to clumping types, which are more adaptable to environmental pressures. Notably, compared with the first report four decades ago, our current results demonstrated that coral coverage around the Dapeng Peninsula declined at an approximate rate of 0.54% per year from 1983 to 2023. Redundancy correspondence analysis revealed that substrate type, especially rock and dead coral skeletons, nitrogen and phosphorus levels, suspended particulate matter, chemical oxygen demand, and water transparency were key environmental factors that significantly ( p < 0.05) influenced the distribution of coral species. The lowest diversity of coral communities was recorded in fishery and wind‐wave areas, where intense human disturbance and harsh natural environments limit the survival of sensitive coral species. In contrast, owing to stable hydrological conditions, shallow sunlit water, and diverse substrate types, coral biodiversity was the highest in the pristine habitat. Accordingly, the coral indicator species varied from Pavona decussata in the conservation area to Platygyra carnosus in the wind‐wave area, to Montipora peltiformis and Leptastrea purpurea in the tourist area, to Turbinaria peltata , Porites deformi , Goniopora columna , and Goniopora lobata in the pristine area, while no indicator species were found in the fishery area. The distribution and coverage of these indicator species can effectively reflect the extent of human disturbance to local habitats. In the future, it will be important to conduct real‐time monitoring of coral community characteristics around the Dapeng Peninsula to determine environmental changes in subtropical marine zones, providing insights into the response of global coral communities to anthropogenic disturbances and climate change.
- 30 September 2025
- Published by Wiley in Ecology and Evolution
- Vol. 15 (10)
Open AccessThe Distribution, Diversity, and Indicator Species of Coral Communities Under the Influence of Environmental Changes in the Subtropical Peninsula of Southern China
Coral reefs are among the most biodiverse and economically important marine ecosystems worldwide. Recently, these ecosystems have faced threats and degradation from multiple factors. In particular, coral species living at the edges of subtropical zones are important “barometers” for indicating the global boundaries of coral communities. From 2023 to 2024, we investigated the coral species composition and coverage surrounding the Dapeng Peninsula, which is located at the northern edge of the subtropical zone, and explored the influence of environmental factors on the coral community structure. Fifty‐one species belonging to 20 genera and 13 families were identified in five habitats (conservation, wind‐wave, tourist, fishery, and pristine areas) across three seasons. Owing to rapid urbanization and global warming, the dominant coral species shifted from branching types, which are essential reef‐builders for creating structural substrates, to clumping types, which are more adaptable to environmental pressures. Notably, compared with the first report four decades ago, our current results demonstrated that coral coverage around the Dapeng Peninsula declined at an approximate rate of 0.54% per year from 1983 to 2023. Redundancy correspondence analysis revealed that substrate type, especially rock and dead coral skeletons, nitrogen and phosphorus levels, suspended particulate matter, chemical oxygen demand, and water transparency were key environmental factors that significantly ( p < 0.05) influenced the distribution of coral species. The lowest diversity of coral communities was recorded in fishery and wind‐wave areas, where intense human disturbance and harsh natural environments limit the survival of sensitive coral species. In contrast, owing to stable hydrological conditions, shallow sunlit water, and diverse substrate types, coral biodiversity was the highest in the pristine habitat. Accordingly, the coral indicator species varied from Pavona decussata in the conservation area to Platygyra carnosus in the wind‐wave area, to Montipora peltiformis and Leptastrea purpurea in the tourist area, to Turbinaria peltata , Porites deformi , Goniopora columna , and Goniopora lobata in the pristine area, while no indicator species were found in the fishery area. The distribution and coverage of these indicator species can effectively reflect the extent of human disturbance to local habitats. In the future, it will be important to conduct real‐time monitoring of coral community characteristics around the Dapeng Peninsula to determine environmental changes in subtropical marine zones, providing insights into the response of global coral communities to anthropogenic disturbances and climate change.
- 5 August 2025
- Published by Elsevier in Marine Pollution Bulletin
- Vol. 221, 118521
Presence and sources of antibiotics in Dongzhai Harbor, Hainan Island
- 5 August 2025
- Published by Elsevier in Marine Pollution Bulletin
- Vol. 221, 118521
Presence and sources of antibiotics in Dongzhai Harbor, Hainan Island
- Open Access
- 24 February 2025
- Published by Wiley in Ecology and Evolution
- Vol. 15 (2)
Spatial Variation in Food Web Structures, Energy Flows, and System Attributes Along the Pearl River and Their Indications of Protection and Restoration
River ecosystems are facing significant degradation from human activities, which impact both biotic (e.g., fish and invertebrates) and abiotic components (e.g., water and habitat). A comprehensive comparison of energy flow patterns and system attributes among river food webs under different levels of human interference is highly important for developing management strategies to protect river ecosystems. Along the subtropical Pearl River, six spatial zones, including agricultural, industrial, island, urban, factory, and estuarine areas, were chosen to construct the Ecopath models. The output results revealed that the highest trophic level of the Pearl River was 3.8–4.2, which was occupied by softshell turtles and piscivorous/carnivorous fish. The most diverse functional groups were found in island and estuarine zones due to their heterogeneous habitats (e.g., high submerged vegetation coverage and the transition area between fresh and brackish water). In contrast, the food web structure in the industrial zone was destroyed due to water pollution (e.g., sewage discharge) and habitat degradation. The increase in exotic species and the decrease in native top predators were two factors that result in the low efficiency of energy transmission. A series of trophic (e.g., Lindeman transfer efficiency and mixed predator–prey impacts), structural (e.g., keystoneness, omnivory, and Finn's path length), and theoretical (e.g., connectance and ascendency) indices revealed that the health and maturity of the Pearl River sections can be ranked as island > estuarine > agricultural > urban > factory > industrial zones. The food chains led by softshell turtles ( Pelodiscus sinensis ), piscivores (e.g., Elopichthys bambusa ), molluscivores (e.g., Mylopharyngodon piceus ), and herbivores (e.g., Ctenopharyngodon idelus ) could be used to indicate the health and functioning of river ecosystems. Our results suggest that the ecological management of river ecosystems should focus more efforts on protecting original habitats (e.g., the island zone with fish feeding/spawning grounds), monitoring bioindicators with keystone trophic impacts in the food web, and evaluating the food chains that play important roles in upward energy transmission.
- 24 February 2025
- Published by Wiley in Ecology and Evolution
- Vol. 15 (2)
Open AccessSpatial Variation in Food Web Structures, Energy Flows, and System Attributes Along the Pearl River and Their Indications of Protection and Restoration
River ecosystems are facing significant degradation from human activities, which impact both biotic (e.g., fish and invertebrates) and abiotic components (e.g., water and habitat). A comprehensive comparison of energy flow patterns and system attributes among river food webs under different levels of human interference is highly important for developing management strategies to protect river ecosystems. Along the subtropical Pearl River, six spatial zones, including agricultural, industrial, island, urban, factory, and estuarine areas, were chosen to construct the Ecopath models. The output results revealed that the highest trophic level of the Pearl River was 3.8–4.2, which was occupied by softshell turtles and piscivorous/carnivorous fish. The most diverse functional groups were found in island and estuarine zones due to their heterogeneous habitats (e.g., high submerged vegetation coverage and the transition area between fresh and brackish water). In contrast, the food web structure in the industrial zone was destroyed due to water pollution (e.g., sewage discharge) and habitat degradation. The increase in exotic species and the decrease in native top predators were two factors that result in the low efficiency of energy transmission. A series of trophic (e.g., Lindeman transfer efficiency and mixed predator–prey impacts), structural (e.g., keystoneness, omnivory, and Finn's path length), and theoretical (e.g., connectance and ascendency) indices revealed that the health and maturity of the Pearl River sections can be ranked as island > estuarine > agricultural > urban > factory > industrial zones. The food chains led by softshell turtles ( Pelodiscus sinensis ), piscivores (e.g., Elopichthys bambusa ), molluscivores (e.g., Mylopharyngodon piceus ), and herbivores (e.g., Ctenopharyngodon idelus ) could be used to indicate the health and functioning of river ecosystems. Our results suggest that the ecological management of river ecosystems should focus more efforts on protecting original habitats (e.g., the island zone with fish feeding/spawning grounds), monitoring bioindicators with keystone trophic impacts in the food web, and evaluating the food chains that play important roles in upward energy transmission.