Introduction

Island ecosystems, which account for merely 5% of Earth’s landmass, harbor 20% of documented species (Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services-IPBES, 2019), highlighting their disproportionate ecological significance. The classical theory of island biogeography demonstrates that geographic barriers and niche differentiation jointly drive rapid evolutionary processes in these ecosystems, establishing them as ideal natural laboratories for studying speciation mechanisms1. However, the very attributes that contribute to their ecological uniqueness, such as limited land area, constrained environmental carrying capacity, and finite resource availability, also render them acutely vulnerable2,3. This vulnerability is manifested through heightened susceptibility to perturbation-induced damage and reduced recovery capacity via endogenous regulatory mechanisms4,5.

The Pressure-State-Response (PSR) model, originally formalized by the OECD (1993), provides a systematic framework for evaluating these dynamics. Within this paradigm, (1) Pressures (P) encompass human activities such as tourism infrastructure expansion, industrial development, and coastal urbanization, which disrupt natural habitats and alter land-use patterns6,7,8,9; (2) State (S) reflects the resultant ecological degradation, including diminished ecosystem functions and service values10; and (3) Responses (R) involve adaptive management interventions aimed at restoring ecological balance. Empirical validations of this framework include Jeju Island’s “Blue Economy” initiative in South Korea, which increased fishery incomes by 35% while restoring 22% of depleted marine stocks11and the Philippines’ community-based mangrove banking system, which reduced restoration costs by 40% 12. These cases underscore the PSR model’s effectiveness in reconciling ecological preservation with sustainable development goals.

Islands in China have an archipelagic distribution pattern, forming groups of islands with distinct geographic characteristics within a relatively homogeneous geological base and natural environment. Empirical studies have shown that islands within a single group, although sharing a homogeneous geological background, exhibit significant spatial heterogeneity in terms of basic geographic attributes (area index, shoreline morphology, etc.)13. Spatial differences are also evident within individual islands in relation to topography and soil factors. These variations, combined with the uneven development of human activities both between and within islands, contribute to distinct spatial heterogeneity of ecosystem-related risks on islands. Many researchers have developed risk assessment methods and case studies from various perspectives and for different ecosystems14,15,16and risk assessment methods specific to island ecosystems have also been proposed17,18. However, current methods often struggle to accurately capture the combined terrestrial and marine characteristics of island ecosystems and their spatial heterogeneity.

The Miaodao Archipelago’s North Changshan and Miao Islands, located 16.5 km and 4.3 km from the mainland coast respectively, function as critical ecological corridors in the Bohai Sea. Due to their proximity to the mainland, these two islands have experienced distinct development trajectories under varying anthropogenic pressures. North Changshan Island, which is connected to South Changshan Island via a sea-crossing bridge, is primarily developed for tourism. This has led to accelerated urbanization and increased habitat fragmentation. In contrast, although Miao Island supports biodiverse habitats for migratory birds and endangered marine species, it faces intensive aquaculture use, with 40% of its coastal zone occupied by kelp farming.

These human-induced pressures highlight the need to apply the PSR framework to assess growing ecological risks, particularly as the archipelago’s resource and environmental carrying capacity approaches critical thresholds. The specific objectives of this study are to: (i) assess ecological risk using the PSR model and examine temporal changes in that risk, and (ii) identify key variables influencing system development in composite environments subject to intensive human activity. Central to this analysis is understanding how socioeconomic drivers, including tourism expansion, fisheries intensification, and coastal urbanization, reconfigure ecological networks, disrupt native biogeochemical cycles, and reinforce livelihood dependencies. The findings are expected to inform spatially targeted conservation strategies that balance ecological integrity with sustainable development in these ecologically significant island systems.

Materials and methods

Study area

The South Island group of Miaodao Archipelago (120°35–120°45’E, 37°53–38°00’N), in the northern part of Yantai City, Shandong Province, China (Fig. 1), comprises South Changshan, North Changshan, Daheishan, Xiaoheishan, and Miao islands. Situated within the Bohai Strait ecological corridor that links the Shandong Peninsula and Liaodong Peninsula19,20these islands maintain ecological connectivity with mainland ecosystems through frequent bird migrations and marine species dispersal. We select North Changshan and Miao islands for comparative analysis due to their contrasting development trajectories, despite their geographical proximity (< 3 km apart) and shared climatic conditions under the northern temperate monsoon regime, which is moderated by marine influences.

Fig. 1
figure 1

Location (a) and island composition (b) of surveyed islands.

Full size image

North Changshan Island (120°42.5′E, 37°58.5′N), the largest island in the group (7.98 km2), has experienced rapid tourism-driven urbanization since its designation as a national 4 A-level scenic area in 2015. Over 68% of its land area is now occupied by tourism infrastructure, including resorts, golf courses, and a deep-water cruise terminal, contributing to a high exploitation rate (47.3%). This development has led to the fragmentation of the original brown soil habitats on low hills (approximately 50% of the island area), although remnant Pinus thunbergii forests remain on protected slopes. Recent studies have reported habitat loss for migratory raptors such as Accipiter soloensis, with nesting sites declining by 32% since 2010 21. In contrast, Miao Island (1.59 km2) supports a mixed subsistence economy, integrating small-scale aquaculture (40% shoreline utilization) with traditional dryland farming. Its exploitation rate is lower (approximately 35%). Although its vegetation structure resembles the coniferous cover found in North Changshan, Miao Island maintains more intact understory diversity, including the vulnerable fern species Cyrtomium devexiscapulae, which is endemic to Shandong’s offshore islands.

Data sources

The Geospatial Data Cloud platform administered by the Computer Network Information Center (CNIC) of the Chinese Academy of Sciences (CAS) provided the Landsat 8 imagery with 30 m spatial resolution and minimal cloud cover (< 10%) used for landscape analysis in the study area in July 2016. A macroscopic survey of the islands was conducted using unmanned aerial vehicles (UAVs, DJI Phantom 2 Vision + modified with hyperspectral camera integration) between August and October 2016. Land cover classification accuracy within the study area is significantly improved through the use of high-resolution orthophotos acquired via UAV remote sensing, while the operational efficiency of vegetation mapping is enhanced through automated spectral feature extraction. A total of 35 and 15 ground control points were established on North Changshan Island and Miao Island, respectively, using a differential GPS (DGPS) measurement system that achieved sub-meter accuracy (in Fig. 2). An 8-category landscape classification system comprising 17 subclasses was systematically applied to generate standardized land cover datasets across the study area. The vegetation classification system was implemented following the Technical Procedures for Satellite Remote Sensing Survey of Coastal Areas of Marine Islands (SOA, 2005), a nationally mandated framework developed through systematic validation across more than 11,000 islands in China. This protocol integrates spectral reflectance characteristics with phenological patterns derived from multi-temporal Landsat imagery, ensuring compatibility with the IUCN Global Ecosystem Typology22 for characterizing coastal ecotones. For non-vegetated areas, the modified Anderson Level II classification was adopted, incorporating high-resolution (≤ 1 m) texture features from UAV orthophotos to differentiate built-up land subcategories. The classification achieved an overall accuracy of 89.2% (Kappa = 0.84) based on stratified random validation. Vegetation types were classified into woodland, shrubland, or agricultural land. Based on surface-use characteristics, non-vegetated areas were categorized as urban and rural buildings, tourist facilities, traffic infrastructure, or bare land.

Vegetation and soil classification surveys were conducted on North Changshan and Miao islands during July and August 2017. A total of 18 French-Swiss school23 sample plots were established on North Changshan Island to quantify structural vegetation metrics, including species abundance, projective cover, and sociability indices. In addition, 4 Anglo-American school24,25 sample plots were set up to systematically track successional dynamics through dominant species turnover rates and community stability thresholds. On Miao Island, 14 French-Swiss school sample plots and 5 Anglo-American school sample plots were similarly established. Vegetation data were used to calculate relative species importance values for each stratum and to classify community types reported by Ellenberg26. Five levels of vegetation cover in sample plots were converted to relative species importance values for each stratum. The combination of the list method and the TWINSPAN quantitative classification improved the accuracy of the vegetation classification27. Diversity indices (Simpson’s (D), Shannon–Wiener (\(\:{H}_{e}^{{\prime\:}}\)) and Pielou’s (\(\:{J}_{e}\))) were calculated as follows28:

$$\:D=1-\sum\:_{i=1}^{S}\frac{{n}_{i}({n}_{i}-1)}{N(N-1)}$$
(1)
$$\:{H}_{e}^{{\prime\:}}=-\sum\:_{i=1}^{S}\frac{{n}_{i}}{N}\text{ln}\frac{{n}_{i}}{N}$$
(2)
$$\:{J}_{e}=\frac{{H}_{e}^{{\prime\:}}}{\text{ln}S}$$
(3)

where S is the number of all species at the standard site, N is the total number of individuals of all species, and ni is the number of individuals of the ith species.

The Natural Composition Index (NCI) is employed to quantify successional transitions within plant communities, building upon the Composition Index (CI) framework originally established by Curtis and McIntosh (1951)29. Plant communities are systematically classified according to their successional stages using the following computational protocol. The NCI is calculated as:

$$\:NCI={V}_{i}\times\:{CAV}_{i}$$
(4)

.

where Vi represents the relative importance value proportion of each species group, and CAVi denotes the corresponding Climax Adaptation Value. Successional progression is numerically encoded using a five-stage classification system: pioneer (CAV = 1), sub-pioneer (3), transitional (5), sub-climax (7), and climax species (9), based on their ecological succession characteristics.

Environmental monitoring data, including water quality assessments, offshore sediment characterization, and marine biota analyses, were obtained from the August 2016 survey conducted by the Changdao County Marine Environmental Monitoring Center (CCMEMC), with intertidal organism data sourced from the center’s August 2015 monitoring campaign. All measurements were carried out in accordance with the Chinese National Standard GB 17,378 − 2007 (2007)30 for marine environmental quality. Quantitative sampling in the intertidal zone was performed using a 0.25 m2 (50 × 50 cm) quadrat to define the sampling area. Owing to the narrow width of the intertidal zone (averaging approximately 25 m) in the study area, standardized tidal zone stratification (high/mid/low) was not applied during sampling. Instead, representative stations were established exclusively in the lower intertidal zone to facilitate systematic specimen collection during low tide periods. At each station, three discrete sampling points were selected and composited into a single mixed sample representing the station. Surface seawater samples were analyzed within 24 h of collection (petroleum hydrocarbons within 10 h) for dissolved oxygen (DO), chemical oxygen demand (COD), dissolved inorganic nitrogen (DIN), dissolved inorganic phosphorus (DIP), petroleum hydrocarbons, and chlorophyll a. Primary productivity in the nearshore waters surrounding the island was calculated based on chlorophyll a and transparency data. DO, COD, DIN, DIP, and petroleum hydrocarbons were selected as assessment factors to compute a comprehensive seawater quality index. At each sampling station, a single integrated phytoplankton sample was collected via vertical hauls from the bottom to the surface at a towing speed of 0.5 m/s using a Type III shallow-water plankton net (mesh size: 0.077 mm). Samples were fixed in 5% formaldehyde solution and stored in 0.5 L polyethylene (PE) bottles at 4 °C in the dark. Phytoplankton species were identified and enumerated using the Utermöhl method31. Cell abundance (cells/m3) was calculated based on the settled sample volume and the volume of seawater filtered during sampling. Phytoplankton biodiversity indices were subsequently derived from the abundance data. For zooplankton, one comprehensive sample per station was obtained via vertical hauls (bottom to surface) at 0.5 m/s using a Type I plankton net (mesh size: 0.505 mm) and preserved in 5% formaldehyde. In the laboratory, zooplankton were taxonomically identified using a stereomicroscope and enumerated. Abundance was expressed as individuals per cubic meter (ind./m3). For biomass determination, gelatinous organisms were excluded; wet weight biomass was determined and reported as milligrams per cubic meter (mg/m3). Zooplankton biodiversity indices were then calculated based on abundance data.

Probability values for risk assessment were derived by tallying the frequency of disaster occurrences per unit time. For stochastic events such as droughts, strong winds, and tropical cyclones, risk statistics were calculated using regional historical data collected over a 50-year period (1951–2000). In the case of cold waves, probabilistic statistics were formulated based on a 40-year dataset (1951–1990). The severity level of each disaster was categorized based on historical records and documentation, with data mainly sourced from Typhoon Online, the Statistical Bureau of Shandong (1986–2018), and Wang32. Data were processed using ArcGIS 10.2, Geomagic, Surfer 21.1.158, and SPSS 19.0. Spatial distribution maps were generated using ordinary kriging implemented in Surfer. An exponential variogram model with 15 lags and a 90% tolerance was applied for all interpolations. Model performance was validated using leave-one-out cross-validation (LOOCV), which yielded the following root mean square errors (RMSE).

Fig. 2
figure 2

Sampling plots on surveyed islands.

Full size image

Research methodology

Theoretical framework

Islands have characteristics of terrestrial ecosystems while simultaneously being influenced by oceanic climate and hydrological factors. There is a general consensus among studies that four distinct microhabitats exist on islands (land, island foundation, coast, and the shallow sea surrounding the island), each with its own biological communities1,33,34. We focus on island management and risk mitigation by identifying relationships of support and constraint among subsystems. Our ecological risk assessment framework, based on island landmass, intertidal zones, and adjacent aquatic environments, underscores the connectedness of terrestrial and marine systems. Within this framework, terrestrial characteristics are encapsulated in its land area, and those of intertidal zones and surrounding waters, its marine features. These resources and habitats are important for island development.

The PSR model, originally developed by Friend and Rapport (1979)35 to systematically analyze environmental feedback mechanisms between anthropogenic pressures, ecological states, and institutional responses, has been established as the predominant analytical framework for coastal ecosystem management under joint OECD–UNEP protocols. This model is operationalized through three core metric categories: Pressure Metrics: Characterized by anthropogenic stressors including intensive resource extraction, cumulative pollutant discharge, and landscape fragmentation indices. State Metrics: Quantified through composite ecosystem health parameters encompassing structural integrity (biodiversity indices), functional efficiency (nutrient cycling rates), and adaptive resilience thresholds. Response Metrics: Evaluated via mitigation efficacy ratios, incorporating both socio-institutional adaptation measures (policy implementation indices) and ecological self-regulation capacities (successional recovery rates).

The PSR model, which addresses state and response in the environment, has become a fundamental framework for exploring environmental issues. Based on the concept of the PSR model and combined with the actual conditions of the islands, a model for evaluating island ecological security is constructed, comprising pressure indicators, state indicators, and response indicators. Pressure indicators mainly refer to the loads imposed on natural ecosystems by human activities, such as resource utilization and pollutant emissions. State indicators reflect ecosystem health, including vitality, organizational structure, and resilience. Response indicators represent the degree of change in ecosystem health, encompassing both environmental transformations and shifts in human behavior.

Indicator system

Risk analysis involves analysis and identification of risk sources and receptors within ecosystems, and exposure-hazard analysis between them. Risk sources involve natural and anthropogenic factors, where the frequency, intensity, and degree of hazard associated with a risk source is investigated and evaluated36. Among risk receptors are various ecosystem and landscape types, along with important species and their habitats21,37. Exposure and impact characterization are central to ecological risk assessment, because they clarify the nature and interaction between risk sources and receptors, reveal potential impacts, and provide a scientific basis for risk management38,39.

Risks to island ecosystems are mainly natural or anthropogenic. Natural risk sources are typically a series of frequent disasters such as droughts, tropical cyclones, pest infestations, cold spells, and strong winds40. Anthropogenic risk sources focus on human activities and events that pose threats to or destruction of local ecosystems. Ecological risks involve pollution caused by overdevelopment, wastewater discharge, chemical and oil spills, and harmful algal blooms—all of which may threaten or damage ecosystem integrity41.

In ecological risk assessments that incorporate land use and remote sensing data, different land use categories—including farmland, urban and rural habitats, and other built environments—serve as key indicators to define the main pressures and risk sources related to human activity42. Based on current land-use classification standards, and considering the feasibility of remote-sensing image interpretation, we identify the following land-use types with characteristics of human activity: agricultural, industrial, mining and storage, residential, tourism and recreation, transportation, and other construction land. Given the current development and use-status of offshore and intertidal waters in China, we recognize four major human activities (aquaculture, industry construction, port transportation, and tourism development) for this region. Because of limitations in information acquisition, we cannot consider risk sources related to pollution emissions.

The main natural risk sources faced by islands originate from series of diverse and ecological disasters such as storm surges, cold spells, droughts, tsunamis, earthquakes, landslides, debris flows, potential sea-level rise, coastal erosion, harmful algal blooms, and invasive alien species. These risk sources are mainly high-frequency disasters that significantly affect island ecosystems, with significant and quantifiable consequences. We do not examine the effects of gradual sea-level rise or tsunamis, which have a lower probability of occurrence in China.

The selection of island ecosystems should be based on a series of ecological parameters that reflect the reduction in the number and quantity of locally endemic species, the decline in bird population numbers and quantities, the disappearance of rare and endangered bird species, and intertidal zone degradation. These ecological impacts may adversely affect the structure and function of island ecosystems. However, we focus on the vegetation, intertidal zone populations, offshore wetlands, and rare and endangered species, rather than bird and other animal populations.

Determining indicator weights

The Analytic Hierarchy Process (AHP) offers a systematic and logically structured methodology for multi-criteria decision-making. Standardized datasets provide the basis for constructing hierarchical judgment matrices through pairwise factor comparisons using the 1–9 ratio scaling technique. Eigenvalues and eigenvectors are then calculated to determine the relative weights of each factor. A consistency validation is performed by calculating the consistency ratio (CR), with values below 0.1 indicating acceptable coherence of the matrix and ensuring statistically reliable weight distributions across the hierarchical structure. Hierarchical aggregation of these weights produces global priority rankings, which are further validated through repeated consistency checks (Table 1).

Table 1 The identified factors of ecological risk for islands.
Full size table

Ecological risk index

Using a multi-factor integrated assessment method, the dimensionless values of each indicator were multiplied by their corresponding weights and summed to obtain an ecological risk index, calculated as follows43:

$$\:\text{E}\text{R}\text{I}=\sum\:_{i=1}^{n}{w}_{i}\cdot\:{A}_{ij}^{{\prime\:}}$$
(5)

.

where ERI is the ecological risk index, Wi is the corresponding weight of the ith indicator, 𝐴𝑖𝑗′ is the result of dimensionless processing of the jth indicator in the ith system.

Ecological risk index values were standardized to fall within a range of 0 to 1. Using the equidistant method44risk levels were classified into five categories: extremely high, high, medium, low, and ultra-low. Risks above the medium level were considered to fall in the risk-warning category.

Results

Distribution pattern and characteristics of land use

On North Changshan Island, eight land categories were delineated (trees, shrubs, grassland, agricultural land, bare land, urban areas, transport land, and tourism land). An overview of the spatial distribution of each land category across sub-regions is depicted in Fig. 3. Vegetation was mostly concentrated in the northern, eastern, and central regions of the island. The primary sources of risk were urban livelihoods and agricultural activities, mainly in northwestern and southwestern coastal areas, respectively, while agricultural pursuits were mainly in the central part of the island.

Fig. 3
figure 3

Spatial distributions of each land category across sub-regions on both islands. (a: North Changshan Island; b: Miao Island).

Full size image

There was no bare land on Miao Island; and the remaining land-use types were consistent with those on North Changshan Island. Vegetation on Miao Island was primarily concentrated in the central region. The main sources of ecological risk originated from urban life and agricultural activities. Agricultural practices and urban habitation were identified as the primary risk factors on both islands.

Type and diversity of vegetation communities

In island ecosystems, vegetation serves as the foundational component, and a robust vegetation cover is necessary to ensure island habitability. Healthy vegetation is essential for hydrological regulation, stabilization of coastal geomorphology, and the maintenance of ecological and biological processes. Table 2 presents the percentage of vegetation cover and soil fertility across each zone of both islands. The soils of North Changshan and Miao islands comprised mainly brown and cinnamon soils, and provide ideal conditions for agriculture and vegetation. The vegetation cover and communities of both islands were similar, comprising mainly plantation forests (42%) and evolved secondary forests (44.8%).

Table 2 Percentage of vegetation cover and soil fertility in each zone of both islands.
Full size table

Dominant communities included Pinus thunbergii, Robinia pseudoacacia, Quercus acutissima, and mixed forests (Table 3). The Shannon–Wiener index for different vegetation types ranged from 2.41 to 3.56, with an average of 3.09. Mixed forests exhibited the highest Shannon–Wiener index, while Robinia pseudoacacia forests showed the lowest, likely due to poor rooting and low light penetration. Pielou’s index was proportional to the Shannon–Wiener index. Simpson’s index correlated negatively with Shannon–Wiener and Pielou’s indices, and varied between 0.13 and 0.38, indicating that species dominance was more pronounced in Robinia forests because of higher species concentration, whereas species dominance was not evident in mixed forests. In terms of vegetation, the peak of plant community evolution would be expected in deciduous forests dominated by Quercus acutissima. The Shannon–Wiener index for Quercus acutissima was lower than for the mixed forest, indicating that the most stable communities were not the most diverse.

Table 3 Characteristics of the main plant communities in the study area.
Full size table

Frequency of natural disasters

Historical data reveals the key factors contributing to natural disasters on both islands were droughts, cold waves, storm surges, insect infestations, and earthquakes. Anthropogenic disasters and accidents mainly involved fires, oil spills, and harmful algal blooms, albeit at a lower frequency. Excluding infrequent and less-threatening stochastic risks, frequent hazards such as droughts, cold waves, storm surges, and insect infestations were the main risks. In general, natural disasters exert impacts on a widespread scale, and the distribution of natural hazard intensities will remain consistent at a regional level. Droughts, cold waves, and storm surges could be deemed to be homogeneous.

Probability values for risks were derived by tallying the frequency of disasters per unit of time. For stochastic droughts, strong winds, and tropical cyclones, risk statistics were calculated using regional historical data collected over 50 years (1951–2000). As for cold waves, probabilistic statistics were formulated based on historical data spanning 40 years (1951–1990). Categorization of the severity level of each disaster was determined based on historical accounts for each (data were mainly sourced Typhoon Online, Statistical Bureau of Shandong 1986–2018, and Wang32. In general, natural disasters exert impacts on a widespread scale, and the distribution of natural hazard intensities will remain consistent at a regional level. Droughts, cold waves, and storm surges could be deemed to be homogeneous.

Ecological status of the circum-island sea area

Spatial variations in macrozoobenthos community metrics were examined across summer intertidal habitats of two island ecosystems. The dominant species on North Changshan Island were Ophiuroidea and Nicon sinica, whereas Batillaria zonalis was the dominant species on Miao Island. A 5.7-fold greater abundance was recorded on Miao Island compared to North Changshan Island, while the latter exhibited a 281% higher Shannon-Wiener diversity index and a 196% increase in Pielou’s evenness (Table 4). The Margalef richness index was observed to be 145% higher on Miao Island, indicating an enhanced capacity for species recruitment. Compared to Wang et al.45who documented 68 species (32 mollusks and 19 crustaceans) with a higher mean abundance (1,383 ind./m2) and biomass (372.41 g/m2) in broader archipelago surveys, the present study recorded fewer species and lower biomass, but comparable species richness. This faunal shift was associated with intensified aquaculture activities after 2010, which were found to increase habitat complexity for epifaunal crustaceans while destabilizing sediment structures essential for burrowing mollusks46. Notably, North Changshan Island exhibited 281% higher Shannon diversity compared to Miao Island, likely reflecting its sheltered bays with heterogeneous substrates that support a more balanced species distribution. These findings are consistent with previous studies highlighting substrate heterogeneity as a key driver of benthic community structure in temperate intertidal systems47.

Table 4 Macrozoobenthos community metrics in the intertidal zones of both Islands in summer*.
Full size table

Annual primary productivity in the surrounding coastal waters was characterized by low values, peaking at levels below 100 g C m−2 yr−1 (Fig. 4). In contrast, relatively high biodiversity was observed in both phytoplankton and zooplankton communities. Seawater quality parameters remained within acceptable thresholds, with key indicators meeting Class I marine water quality standards (GB3097-1997).

Fig. 4
figure 4

Environmental characteristics of surrounding waters: (a) comprehensive seawater quality index; (b) primary productivity; (c, d) phytoplankton diversity; (e, f) zooplankton diversity.

Full size image

Risk assessment index for Island ecosystems

Risk compartments are fundamental building blocks in risk evaluation and classification. Each compartment functions independently. Assessment and comparative analysis of ecological risk across various compartments enable the stratification of ecological risk levels and the generation of spatial maps within the assessment area. In general, risk-parcel division can be based on natural or anthropogenic boundaries, such as mountain ranges, rivers, valleys, or protected area boundaries48. For areas that include water bodies, however, risk-cell delineation is often conducted using isobaths. A primary objective of island ecological risk assessment is to define risk-cell classifications in conjunction with natural boundaries.

Fig. 5
figure 5

Spatial distributions of risk on both islands. a: North Changshan Island; b: Miao Island.

Full size image

The risk value of the North Changshan Island onshore assessment unit was medium (0.50, Fig. 5). The onshore ecosystem was found to be relatively stable, yet ecological risks were deemed to be potential occurrences. The risk value of the intertidal zone and offshore waters was high (0.74). Ecosystems within these zones were in an early warning state, suggesting that ecological problems were likely.

An overview of the percentage of area exposed to risk on both islands is depicted in Table 5. According to assessment results for terrestrial ecosystems for North Changshan Island, approximately 55.46% of the land was categorized at low ecological risk, 33.74% as ultra-low ecological risk, and 4.63% as moderate ecological risk. Overall, the ecological risk level for terrestrial ecosystems on the island was low. However, the ecological risk in central and coastal areas was slightly higher. For the intertidal zone, 24.91% of the area was at low risk and 75.81% at high risk, and risk was consistently greater in the southern region. For coastal waters, 73.23% were at medium ecological risk, and 26.77% were at high ecological risk, the latter mostly in eastern and western waters. The ecological risk of terrestrial ecosystems on the island was mostly low and ultra-low, whereas for the intertidal zone and coastal waters, ecosystems were in a state of early warning.

Table 5 The percentage of area exposed to risk on both islands.
Full size table

The risk value assigned to terrestrial ecosystems of Miao Island (0.40) was low, and that for the intertidal zone and coastal waters was medium (0.50). The terrestrial ecosystem remained fundamentally stable with a low probability of ecological risk, whereas that for intertidal and offshore ecosystems was deemed to be potential. Of terrestrial ecosystems, 27.23% were classified as ultra-low risk, 65.26% as low ecological risk, and 7.51% as medium ecological risk. Overall, the ecological risk level on the island was low, with this being slightly higher in central and coastal regions. Ecological risk characteristics of the intertidal and coastal waters of Miao Island were lower than those of North Changshan Island.

Discussion

Exploring assessment models

This study established a risk assessment framework for island ecosystems based on a PSR model that includes both landmass and adjacent aquatic subsystems. The pressure layer consists of three types of stressors: development intensity, natural hazards, and ecosystem risk. It reflects the spatial distribution characteristics of the type, scale, and intensity of human activities and natural hazards on the islands and their impacts on the ecosystem. Among the state layer indicators, vegetation and marine primary productivity reflect ecosystem vitality, determine carbon sources/sinks, and are important factors regulating ecosystem processes49. Diversity plays a fundamental role in regulating material cycles, energy flows, and ecosystem stability50. Soils not only provide sites and nutrients for plant growth but also play an important role in pollutant remediation and the cycling of elements51. The quality of seawater reflects the capacity of marine waters to support biological or human survival and development under the pressure of pollutant discharge. The response layer includes measures to prevent pollution and restore seawater quality, and it assesses the extent to which effective management contributes. The analysis of changes in productivity, biodiversity, and environmental quality of island ecosystems under human development pressures provides a more complete picture of the probability of ecological risk in island systems.

Furthermore, the assessment model fully reflects the dual characteristics of island ecosystems and their spatial heterogeneity. The island ecosystem is divided into two parts with clear boundaries and interrelationships. An assessment index with spatial heterogeneity is selected to reflect the status of the two subsystems. Risk probability calculations are performed using corresponding assessment metrics to achieve spatial uniformity, standardization, and comparability of risks for terrestrial island ecosystems and marine atoll ecosystems. This model is designed to assess the risk of island ecosystems at both the ecosystem and block scales, and its calculation methodology is clear, simple, applicable, and comparable, making it a useful reference for the risk assessment of island ecosystems.

It should be noted that the indicator system used in this study was designed primarily for the characteristics of islands in the Bohai-Yellow Sea region, and its applicability to tropical and reef islands has not been tested. The rate of degradation in the terrestrial subsystem may be underestimated, as soil erosion dynamics are not explicitly parameterized. The timeliness of bioindicator monitoring is currently inadequate, which limits the ability to resolve short-term variability in stress responses. As sessile photoautotrophs, algal communities are particularly effective as bioindicators of coastal eutrophication and trace metal accumulation. Future monitoring protocols should incorporate both morphological and metabarcoding analyses of perennial macroalgae to improve the comprehensiveness of environmental assessments. Future studies are recommended to focus on four key directions to enhance methodological rigor and interdisciplinary integration.

Impact factors

The economic foundation of North Changshan Island is anchored in two dominant sectors: marine aquaculture and tourism. As of 2015, the total fisheries output reached 6.13 billion RMB, reflecting a 2.7% annual growth rate. Aquaculture production attained 4.3095 million metric tons, marking a 1.2% year-on-year increase, with mariculture operations contributing 306,100 metric tons, a 1.1% rise compared to baseline levels. Concurrently, tourism—specifically fishery-based homestay services—has emerged as a critical secondary economic pillar. Since its institutionalization in 2000, this sector employed 22,000 individuals by 2017, accounting for 38% of local household income. The proliferation of coastal tourism infrastructure correlates with a 15% expansion in shoreline development between 2010 and 2017.

The vulnerability of island ecosystems in the study area is characterized by land-sea interaction dynamics, as revealed through systematic analysis. Regarding ecological conditions, island soil quality and the primary productivity of surrounding coastal waters are low. The diversity of island vegetation, as well as the diversity of phytoplankton and zooplankton in coastal waters, is falling within intermediate ranges for the Bohai-Yellow Sea ecoregion, while seawater quality is relatively good. In terms of development intensity, coastal and island development is high, while offshore development intensity is relatively low. These findings demonstrate the predictive capacity of the pressure-state-response (PSR) framework: when anthropogenic drivers exceed ecosystem carrying capacities, the degradation of service functions follows threshold-driven nonlinear decay patterns.

The results of the block-scale assessment clearly illustrate the spatial heterogeneity of risk to island ecosystems. The spatial heterogeneity in the carrying capacity of marine ecosystems surrounding the islands is strongly influenced by development intensity. Different modes of marine use during the process of development and utilization exert varying degrees of impact on marine ecosystems. The intertidal zone experiences most development and use pressure, whereas coastal waters face threats from aquaculture, tourism, and recreation activities. Some of the offshore region of North Changshan Island occurs within the core of the “Miao Islands Spotted Seal Nature Reserve,” and is of significant ecological significance. Aquaculture is more prevalent along the northwestern coastline of North Changshan Island, than elsewhere. The coastline is largely occupied by aquaculture ponds, and the intertidal zone is degraded. These ponds both mar the coastal landscape and significantly impact the local ecology by eroding foraging and roosting habitats for numerous rare waterfowl. This mirrors trends identified on Changxing Island through the Resource Risk Model (RRM) evaluation, which demonstrated that tourism constitutes the predominant ecological stressor (contributing 58% to cumulative risk indices), followed by mariculture (32%) and agricultural land use (10%)52. Managing shoreline risk is imperative. Construction of new aquaculture ponds must cease, indiscriminate private construction must be prohibited, and abandoned and illegal breeding ponds should be dismantled to restore the coast to its original state, particularly in an area adjacent to the seal sanctuary.

Our assessment reveals that North Changshan Island is at a relatively high level of potential ecological risk, consistent with Wang32 and Qin53. In the island-land subsystem, high-risk areas are concentrated in zones with dense vegetation distribution on North Changshan Island. Given that the potential risks on Miao Island are comparably low, and there are no notable stressors or habitat types within the island’s vicinity that require urgent attention, the development of specialized risk-management measures is less urgent. Nevertheless, the island’s terrestrial ecosystem subsystem plays an important role in vegetation preservation, and the risks within this ecosystem are elevated. Therefore, consideration should be given to safeguarding the terrestrial environment and its flora and fauna.

Conservation and restoration of Island vegetation based on risk assessment

Vegetation naturalness is an important metric of environmental health and vitality. As naturalness increases, so too does the overall condition of vegetation. However, these islands support low levels of naturalness, with vegetation being almost artificial forests and remote natural forests54. Two key factors influence island vegetation naturalness. First, it is closely related to the type and developmental stage of vegetation. For instance, climax forests are the most natural, followed by transitional successional forests, pioneer forests, native species plantations, and exotic species plantations. Unfortunately, vegetation on neither North Changshan nor Miao Island has reached a climax stage. Second, increased urbanization and exploitation of island environments have significantly degraded island vegetation. When natural vegetation is damaged, efforts to restore it using single-type plantations often hinder natural succession.

Semi-plantations (semi-artificial forests) on North Changshan Island exhibit sluggish understory renewal, a relatively homogeneous community structure, and low stand density, resulting in limited ecological functions in terms of windbreak, soil consolidation, and fertilizer retention. In response to these characteristics, a mixed-forest transformation strategy is recommended, involving the gradual replanting of native tree species adapted to local ecological conditions to enhance stand structure richness, species diversity, and overall community stability.

For Miao Island, a remote natural forest area, appropriate artificial nurturing measures are suggested to accelerate the natural succession of vegetation. Specific approaches might include thinning out of poorly growing trees, reducing stand closure, and creating forest gaps conducive to seedling growth to promote natural vegetation renewal.

Assessing the semi-naturalness of island vegetation serves as a foundation for restoring degraded forest ecosystems on islands. Based on our evaluation, the development plans for islands can be formulated more scientifically, informed use-intensities can be set, and targeted vegetation restoration and protection measures can be implemented. These actions will ensure the health, stability, and sustainable development of island ecosystems.

Conclusions

We assess the ecological risk to two islands in the Miaodao Archipelago. To assess the exposure, sensitivity, and adaptability of islands, our risk assessment examined relevant natural and human factors, and considered their unique characteristics and developmental context. Based on a literature review, field investigations, and previous research, incorporating a Geographical Ternary Structure System (island landmass, intertidal zones, and adjacent aquatic system), an ecological risk assessment system was formulated. This system provides a robust framework for evaluating ecological risks on islands. The assessment model reflects the dual characteristics of land and sea ecosystems and their spatial heterogeneity, and conducts risk assessment of island ecosystems at multiple scales, providing a useful reference for future evaluations of island ecological risk.

The spatial risk heterogeneity observed in the two islands necessitates tiered governance strategies. (1) Terrestrial subsystems exhibit mild risk, while offshore zones show moderate risk due to cumulative aquaculture impacts. Implement Dual-Zone Management has been recommended under China’s Marine Ecological Red Line Policy: enforce Grade III shoreline controls limiting artificial coastlines to < 5% of total length, prioritizing seawall-to-reef conversions; establish 500 m aquaculture buffer zones with density caps to reduce nutrient loading. (2) 24.7% of North Changshan’s high-vulnerability areas correlate with tourism infrastructure density. Tourism carrying capacity optimization requires the implementation of scientifically validated visitor density thresholds, operationalized through a three-tier zoning system enforced. (3) Construction footprints require strict containment, accompanied by slope stabilization using native pioneer species and biochar-amended soil remediation. (4) To address the limitations of conventional monitoring methods in detecting critical biological thresholds, a synergistic “real-time/periodic” monitoring system is proposed. Specifically, regular monthly and seasonal monitoring will be supplemented with hyperspectral drone imaging for terrestrial ecosystems and eDNA macrobarcoding for aquatic ecosystems, thereby establishing an integrated real-time biological monitoring network.

Because complete data were unavailable for certain indicators in our indicator system, this affects the precision of evaluation results. Accumulating data over time would enable trend assessments to be made, ultimately rendering our evaluation of island ecosystems more instructive and valuable.