Process-Based Source Apportionment and Radiological Baseline of Multi-Radionuclides in Soils of a Tourism-Oriented Island

Abstract

Islands have high ecological and tourism value; however, owing to their remoteness and limited accessibility, environmental radioactivity is often less systematically evaluated than in mainland regions. This study investigates the distribution, source partitioning, and radiological implications of multi-radionuclides (⁷Be, ¹³⁷Cs, ²¹⁰Pb, ²³⁸U, ²²⁶Ra, ²³²Th, and ⁴⁰K) in surface soils of Zhoushan Island, a representative tourism-oriented island in the East China Sea. Activity concentrations of ⁷Be, ¹³⁷Cs, ²¹⁰Pb, ²³⁸U, ²²⁶Ra, ²³²Th, and ⁴⁰K ranged from 3.4 to 585.5, below detection limit −5.7, 45–1490, 33.3–72.4, 32.3–58.9, 37.8–91.7, and 439.6–872.3 Bq/kg, respectively. Using multivariate statistics and geochemical interpretation, we classified radionuclides into three groups: (i) atmospheric deposition-driven nuclides (⁷Be, ²¹⁰Pb_ex), (ii) lithogenic background-controlled nuclides (²³⁸U−²²⁶Ra−²³²Th), and (iii) the alkali-metal-like behavior group (¹³⁷Cs−⁴⁰K). This shows that soil radionuclide patterns result from atmospheric inputs, geological inheritance, and land-use disturbance, rather than simple concentration variability. Spatial analysis revealed that agricultural disturbance enhances ¹³⁷Cs redistribution, low-lying terrains preferentially accumulate atmospheric fallout nuclides, and lithogenic radionuclides are higher in the northern island due to parent material and weathering. No significant ⁴⁰K enrichment was observed in cultivated soils, indicating limited fertilizer influence. Although radiological indices remain within international safety thresholds, several parameters exceed global background levels, indicating elevated natural radiation driven primarily by thorium-rich lithology. Importantly, we show that radiological risk assessments based solely on bulk activity may overestimate environmental significance without considering process controls. This study provides a process-informed radiological assessment for island systems, offering insights for environmental monitoring and risk evaluation in similar coastal and tourism-dominated regions.

1. Introduction

Islands serve as strategic hubs and essential platforms for ocean exploration and utilization, thereby underpinning the development of the marine economy [1]. China possesses more than 6500 islands with areas exceeding 500 m², totaling ~72,800 km², and collectively forming an extensive “Second Marine Economic Zone”. Situated in the coastal region of the East China Sea, the Zhoushan Archipelago comprises more than 1390 islands, accounting for more than one-fifth of all Chinese islands [2]. It is the only prefecture-level city in China established on an archipelago (with Hainan as a province-level administrative unit) [3], thus occupying a strategically pivotal position in the national framework for island conservation and development. However, due to inherent natural constraints such as strong geographical isolation, limited land availability, and scarce natural resources, island ecosystems generally exhibit high ecological fragility [1,4]. Although environmental radioactivity has been investigated on several islands, including the Canary Islands [5], the Marshall Islands [6], Ustica Island [7], the Aeolian Islands [8], the Princes’ Islands [9], and Taiwan Island [10], existing studies have primarily focused on radionuclide inventories, whereas the mechanisms controlling radionuclide partitioning among atmospheric deposition, lithogenic background, and anthropogenic disturbance remain insufficiently understood. In this context, a process-based assessment refers to an interpretive framework that links radionuclide occurrence to specific environmental transport and transformation mechanisms, such as wet and dry deposition, weathering, and soil disturbance, rather than relying solely on bulk activity measurements. In addition, overall awareness of and attention to radiological issues and associated risks in island environments remain limited. This is largely due to the unique environmental settings of islands, poor data comparability among studies, insufficient long-term monitoring, and socioeconomic constraints. Consequently, significant gaps persist in systematic scientific data regarding baseline environmental radioactivity, radionuclide migration processes, and potential public health impacts. In addition, recent studies have increasingly emphasized radionuclide transport mechanisms in coastal and island systems, including geochemical fractionation [11], sedimentary redistribution [12], and watershed-scale migration [13]. These investigations provide critical insights into the processes governing radionuclide behavior and offer an important foundation for process-based interpretations in dynamic coastal environments.

Since the beginning of the 21st century, the rapid expansion of the marine economy and accelerated urbanization in coastal island regions—particularly following the opening of the cross-sea bridge connecting the Chinese mainland to Zhoushan Island after 2010—have transformed the Zhoushan Archipelago into a well-developed tourist destination [1,2]. In addition to its unique natural landscapes, the archipelago is highly valued for its thematic tourism that integrates historical heritage, Buddhist culture, and folk customs [14]. According to the Zhoushan Municipal Bureau of Statistics, tourist arrivals in the archipelago grew rapidly since 1980, reaching a peak of 70.52 million in 2019, declined during the COVID-19 period, and rebounding to 27.80 million in 2024. Against this backdrop, ecological and environmental pressures in the Zhoushan Archipelago have intensified, drawing increasing concern from multiple sectors of society [4]. To date, research on the ecological environment of the Zhoushan Archipelago and surrounding waters has primarily focused on heavy metal contamination in seawater, sediments, and seafood [15,16,17], the evolution of land-use and landscape patterns [2], and assessments of island ecosystem vulnerability [4,18]. However, systematic studies and baseline assessments of radiation environments on islands in this region are still lacking.

Radioactive nuclides are a major class of environmental pollutants characterized by insidiousness, irreversibility, and persistence [19]. They contribute approximately 80% of the annual cumulative radiation dose received by humans [20]. Based on their origin, natural radionuclides can be classified into two main categories: (1) cosmogenic radionuclides (e.g., ⁷Be), which are produced by interactions between cosmic rays and atmospheric constituents and contribute minimally to the global natural radiation dose; and (2) primordial radionuclides, which are the principal source of natural radiation and mainly include decay products of the thorium, uranium, and actinium series (e.g., ²¹⁰Pb, ²¹⁰Pb_ex, ²²⁶Ra, ²²⁸Ra, ²²⁸Th, ²³⁸U) as well as long-lived single-decay nuclides such as ⁴⁰K [19]. Soil acts as a major reservoir and key environmental medium for the migration and transformation of radionuclides [21]. Natural radionuclides (²³⁸U−²³²Th-series and ⁴⁰K) in soils are primarily derived from the weathering of underlying bedrock. Their distribution is influenced by processes such as weathering, sedimentation, leaching/adsorption, precipitation from percolating groundwater, and dilution by materials of different composition, resulting in substantial variability in activity concentrations [22]. Anthropogenic radionuclides such as ⁹⁰Sr, ¹³⁴Cs, ¹³⁷Cs, ²³⁹Pu, ¹²⁹I, and ²⁴¹Am originate mainly from atmospheric nuclear weapon tests in the 1950s–1960s and the Chernobyl nuclear accident (Ukraine) in 1986 [23]. Although ¹³¹I and radiocesium were occasionally detected in food and seafood in China after the Fukushima Daiichi accident (2011), the negligible concentrations indicated no significant radiological impact, and ¹³⁷Cs in Chinese soils remains dominated by historical fallout rather than Fukushima-derived sources [24,25]. Extensive studies have investigated their behavior in soil environments for radiological protection purposes [21,26,27,28,29]. These radionuclides emit radiation during decay and represent a major source of external exposure to the public. Thus, both natural and anthropogenic radionuclides can accumulate in island soils. When such soils are used for agricultural, pastoral, or tourism-related purposes, such as grazing livestock, cultivating crops and vegetables, or constructing tourist facilities, elevated radionuclide activity levels may pose potential radiation risks to food safety and public health. Therefore, characterizing the spatial distribution of radionuclides in island soils and tracing their sources are essential for guiding the sustainable development and conservation of tourism-oriented islands.

In this study, 43 surface soil samples (0–5 cm depth) were collected from evenly distributed sites across Zhoushan Island to analyze radionuclide characteristics. The objectives of this study are to: (1) determine the spatial distribution patterns and activity levels of radionuclides in surface soils under the unique geographical context of an island environment; (2) identify the sources of radionuclides in island soils; (3) locate and analyze areas with anomalous radioactivity levels; and (4) assess radiation risks using hazard indices, with particular attention to sensitive areas such as parks, ports, and schools. This study attempts to link multivariate grouping results with geochemical processes, enabling a process-based assessment of radionuclide behavior. Ultimately, rather than focusing solely on radiological hazard indices, this study aims to evaluate whether such indices adequately reflect environmental processes and actual risk relevance in island settings.

2. Materials and Methods

2.1. Study Area

Zhoushan Island (29°50′–30°12′ N, 121°30′–122°25′ E), the largest island within the Zhoushan Archipelago, is located in the central coastal region of Zhejiang Province, China (Figure 1). The island trends northeast–southwest, measuring approximately 44 km in length and 18 km in width, with a total coastline of 170.16 km and a land area of about 491 km² [2,30]. The terrain is predominantly hilly. Rocky coastlines account for 75.8% of the shoreline, whereas artificial coastlines account for 21.6% [31]. The island has a northern subtropical monsoon climate, with a mean annual temperature of 15.8–16.7 °C. Annual precipitation ranges from 921.6 to 1318.8 mm, whereas annual evaporation ranges from 1225.9 to 1374.2 mm [32]. The tidal regime is characterized as irregular semidiurnal, with a long-term average tidal range of 1.91–3.31 m and a maximum tidal range of up to 4.96 m [32,33]. The island is frequently affected by tropical cyclones during summer and autumn, with an average of 4.3 events per year [4]. Geologically, the island is underlain by extensive Mesozoic andesitic–granitic igneous rocks, overlain by semi-consolidated to unconsolidated Cenozoic terrigenous sediments that may reach several hundred meters in thickness [34]. Soils are generally shallow, nutrient-poor, and relatively saline [4]. Major soil types include Acrisols, Leptosols, Anthrosols (paddy soils), and Solonchaks according to the WRB soil classification system [35].

Zhoushan Island, the principal island of the Zhoushan Archipelago, serves as the political and economic center of Zhoushan City, accommodating 70% of the archipelago’s population and contributing 74% of its economic output [2]. In recent decades, Zhoushan Island has experienced rapid development driven by tourism expansion, aquaculture, port construction, and urbanization, making it representative of the socio–ecological–economic characteristics of tourism-oriented islands. Therefore, it provides a representative case for analyzing vulnerability within the socio–ecological–economic complex of tourism-oriented islands [1]. The demand for land has increased substantially, leading to large-scale land reclamation and marine engineering over the past forty years, which have profoundly altered the coastal landscape [3,4,31]. Between 1984 and 2020, construction land expanded from 18.19 to 115.97 km², while forest land, mudflats, cropland/grassland, and bare land decreased from 205.72 to 194.39 km², 19.24 to 13.73 km², 210.98 to 203.87 km², and 19.05 to 6.09 km², respectively [2]. Furthermore, from 2000 to 2020, the average Remote Sensing Ecological Index (RSEI) in the Zhoushan Islands declined from 0.748 to 0.681, accompanied by a 13.54% reduction in areas rated as excellent and a 3.43% increase in areas rated as poor or fair, indicating clear ecological degradation [18]. In addition, the total shoreline length of the Zhoushan Archipelago decreased by 7.05 km from 1984 to 2018 [33]. Collectively, these findings demonstrate the high ecological vulnerability of Zhoushan Island.

2.2. Sampling and Preparation

A total of forty-two sampling sites was selected in coastal zones, mountainous areas, croplands, agricultural fields, industrial zones, and urban residential areas across Zhoushan Island (Figure 1). In June 2025, 43 samples were collected: one from each site and two from site ZS31, where two samples (ZS31 and ZS31a) were collected in close proximity (one from inside a park and the other from a roadside) to account for local-scale variability. At each sampling site, a square quadrat of approximately 5 m² was delineated. Five subsampling points were randomly selected within each quadrat, and the topsoil collected from these points was combined to form one composite sample (approximately 1 kg). The 0.0–0.2 cm surface layer, consisting of overlying vegetation and the thin veneer of adhering soil, was removed by cutting at the soil interface. Subsequent soil samples within the upper 5.0 cm were collected using a stainless-steel blade and placed in polyethylene Ziplock bags for laboratory analysis.

2.3. Radioactivity Measurement

Soil samples were oven-dried at 80 °C for 24 h, ground, sieved to <150 μm (100 mesh), packed into cylindrical containers of identical geometry, and sealed for 30 days prior to radionuclide analysis using a high-purity germanium (HPGe) γ-spectrometer (Canberra-BE3830, Meriden, CT, USA). After sealing, secular equilibrium between ²²⁶Ra and its progeny radionuclides (²²²Rn, ²¹⁴Pb, ²¹⁴Bi, ²¹⁴Po and ²¹⁰Pb) was achieved. The activity concentrations of ²¹⁰Pb, ²²⁶Ra, ¹³⁷Cs, ⁴⁰K, ²²⁸Th, ²²⁸Ra, and ²³⁸U were determined using an ultra-low background HPGe γ-spectrometer (Canberra-BE3830, Meriden, USA; energy range: 10 keV–2000 keV; relative efficiency >35%; background ~0.9 counts/s in 3 keV–2 GeV). The instrument shielding consisted of 15 cm of low-background lead, lined with tin and high-purity copper.

To ensure analytical accuracy, certified reference materials (GBW08034a, GBW04127, IAEA-312, and IAEA-327), provided by the National Institute of Metrology of China and the International Atomic Energy Agency (IAEA), were used to determine full-energy peak efficiencies for ¹³⁷Cs and naturally occurring radionuclides. Comparisons with certified values of reference materials showed an accuracy of 85–109% for 8 radioisotopes in soil samples analyzed in our laboratory. Decay correction and background subtraction were performed for all samples. Each sample was counted for 24 h. The specific activity of ²¹⁰Pb was determined from the 46.5 keV peak (branching ratio: 4.25%). ²³⁸U activity was quantified directly from the 63.5 keV gamma line, which has a branching ratio of 4.8% of its progeny ²³⁴Th. ²²⁶Ra activity was calculated from its progeny ²¹⁴Bi (609.3 keV, 46.3%) and ²¹⁴Pb (295.2 keV, 19.3%, and 352.0 keV, 37.6%). Excess ²¹⁰Pb (²¹⁰Pb_ex) was obtained by subtracting ²²⁶Ra activity from total ²¹⁰Pb. ²²⁸Ra activity was determined using ²²⁸Ac peaks (338.5 keV, 11.27% and 911.1 keV, 26.6%), whereas ²²⁸Th activity was calculated from the mean activity of ²⁰⁸Tl (583.2 keV, 85.2%) and ²¹²Pb (238.6 keV, 43.3%). The specific activity of ⁷Be was determined from the 477 keV peak (10.44%) with a detection limit of 5 Bq/kg. ⁴⁰K and ¹³⁷Cs were quantified using the 1460.8 keV (10.55%) and 661.5 keV (84.62%) peaks, respectively. The counting error for all radionuclides was <10%.

The activity of a radionuclide in a soil sample can be calculated by the following equation:

$$A_i=\left({\displaystyle \frac{N_i}{T_i}}-{\displaystyle \frac{N_{bg}}{T_{bg}}}\right)/(Y_i\times {\epsilon }_i\times M)$$

The minimum detectable activity (MDA) values in soil samples can be given as:

$$MDA(i)={\displaystyle \frac{2.71\times \sqrt{N_{bg}(i)}}{{\epsilon }_i\times M\times T_{bg}}}$$

where A_i (Bq/kg) denotes the specific activity of interested radionuclide, i; N_i and N_bg are the peak areas of interested radionuclide at sample determining time and background determining time, respectively; T_i and T_bg (s) denote the measuring time at sample measurement and background measurement, respectively; Y_i is the branching ratio of the interested radionuclide; ε_i is the detection efficiency of radionuclide i; M is the mass of soil sample (kg).

The typical MDA values were 0.58 Bq/kg for ²¹⁰Pb, 0.8 Bq/kg for ²³⁸U, 1.30 Bq/kg for ²²⁶Ra, 0.27 Bq/kg for ²²⁸Ra, 0.19 Bq/kg for ²²⁸Th, 3.0 Bq/kg for ⁷Be, 1.91 Bq/kg for ⁴⁰K, and 0.21 Bq/kg for ¹³⁷Cs.

2.4. Calculation of Radiological Indices

In order to evaluate the radiological risks associated with natural radionuclides in soils, several internationally accepted hazard indices were calculated, including radium equivalent activity (Ra_eq), outdoor absorbed gamma dose rate (DR), annual effective dose equivalent (AEDE), annual gonadal dose equivalent (AGDE), excess lifetime cancer risk (ELCR), and external hazard index (Hext). These indices collectively reflect radiation intensity, potential health effects, and long-term carcinogenic risk.

Natural radionuclides (²²⁶Ra, ²³²Th, and ⁴⁰K) are heterogeneously distributed in soils; therefore, standardized indices that combine the activities of multiple nuclides are required for reliable assessment of radiation exposure. The Ra_eq was developed to represent the overall gamma output from different radionuclides. It is based on the assumption that 370 Bq/kg of ²²⁶Ra, 259 Bq/kg of ²³²Th, and 4810 Bq/kg of ⁴⁰K produce the same gamma dose rate [36]. Thus, Ra_eq is calculated using the following equation [36]:

$${\mathit{Ra}}_{\mathit{eq}}={\mathit{C}}_{\mathit{Ra}}+{1.43\mathit{C}}_{\mathit{T}\mathrm{h}}+0.077{\mathit{C}}_{\mathit{K}}$$

(1)

where C_Ra, C_Th, and C_K are the activity concentrations (Bq/kg) of ²²⁶Ra, ²³²Th, and ⁴⁰K, respectively. Ra_eq should not exceed 370 Bq/kg, which corresponds to an annual dose of 1.5 mGy/y [36], and is commonly used as a screening index for gamma radiation hazard.

The DR (nGy/h) reflects the amount of ionizing radiation energy deposited in air at a height of 1 m above the ground. Assuming uniform radionuclide distribution in soil, DR can be calculated using dose conversion factors as [36]:

$$\mathit{DR}=0.462{\mathit{C}}_{\mathit{Ra}}+0.604{\mathit{C}}_{\mathit{T}\mathrm{h}}+0.0417{\mathit{C}}_{\mathit{K}}$$

(2)

This parameter is directly related to potential biological effects of gamma radiation.

AEDE is a key parameter used in radiation protection to evaluate potential health effects from long-term exposure to ionizing radiation. AEDE can be calculated using dose conversion factors as [36]:

$$\mathit{AEDE}=\mathit{DR}\times \mathit{DCF}\times \mathit{OF}\times \mathit{T}$$

(3)

where DR is the absorbed gamma dose rate, DCF is the dose conversion coefficient (0.7 Sv/Gy), OF is the outdoor occupancy coefficient (20%), and T is the time coefficient (8760 h) [36].

The ELCR estimates the probability of developing cancer over a lifetime (70 years) due to exposure to natural radionuclides. ELCR is derived from AEDE using a fatal cancer risk factor of 0.057 Sv⁻¹ [37]:

$$\mathit{ELCR}=\mathit{AEDE}\times \mathit{DL}\times \mathit{RF}$$

(4)

where DL is the life expectancy (70 years), and RF is the fatal risk factor per year, equal to 0.057 Sv⁻¹ [37].

The Hext assesses the external gamma radiation hazard from soil or building materials [38]. For safe use, H_ext must be less than 1, ensuring that the external dose does not exceed 1 mSv/y [38]. It is calculated as:

$$\mathit{Hext}={\displaystyle \frac{{\mathit{c}}_{\mathit{Ra}}}{370}}+{\displaystyle \frac{{\mathit{c}}_{\mathit{T}\mathrm{h}}}{259}}+{\displaystyle \frac{{\mathit{c}}_{\mathit{K}}}{4810}}$$

(5)

AGDE is particularly important for assessing hereditary radiation effects in exposed populations. The AGDE (μSv/y) evaluates the genetic significance of radiation exposure to reproductive organs and is calculated using the following equation [31]:

$$\mathit{AGDE}=3.09{\mathit{C}}_{\mathit{Ra}}+4.18{\mathit{C}}_{\mathit{T}\mathrm{h}}+0.314{\mathit{C}}_{\mathit{K}}$$

(6)

Statistical analyses were performed using SPSS Statistics (IBM) 17 and Origin 2026 (OriginLab), while spatial distribution maps were generated in ArcGIS 10.8.0 (Esri). Relationships between radionuclide ratios and soil properties were assessed via Pearson correlation. Hierarchical Cluster Analysis (HCA) and Principal Component Analysis (PCA) were conducted using Ward’s method with Pearson correlation distance, and PCA components were varimax rotated for clearer interpretation.

3. Results

3.1. Statistical Analysis of the Specific Activities of Radionuclides in Topsoil

The specific activities of naturally occurring radionuclides (²¹⁰Pb, ⁷Be, ²²⁶Ra, ²²⁸Ra, ²²⁸Th, ²³²Th, ²³⁸U, and ⁴⁰K) and the anthropogenic radionuclide (¹³⁷Cs) were measured in 43 topsoil samples. Statistical analysis, including the mean, standard deviation, skewness, kurtosis, and range of activity values for each radionuclide, is summarized in Table S1 and illustrated in Figure 2. The data reveal significant variability, particularly for airborne radionuclides (⁷Be and ²¹⁰Pb). For instance, ⁷Be activities ranged from 3.4 to 585.5 Bq/kg, with a mean of 44.5 ± 91.6 Bq/kg, while ²¹⁰Pb activities ranged from 45 to 1490 Bq/kg, with a mean of 181.5 ± 279.0 Bq/kg. For terrestrial radionuclides, activities were as follows: ²³⁸U ranged from 33.3 to 72.4 Bq/kg (mean 52.0 ± 8.5 Bq/kg), ²²⁶Ra from 32.3 to 58.9 Bq/kg (mean 44.3 ± 5.3 Bq/kg), ²²⁸Ra from 37.3 to 91.2 Bq/kg (mean 56.6 ± 8.6 Bq/kg), ²²⁸Th from 38.2 to 92.1 Bq/kg (mean 59.9 ± 8.9 Bq/kg), and ⁴⁰K from 439.6 to 872.3 Bq/kg (mean 631.0 ± 87.4 Bq/kg). Since ²³²Th is the parent nuclide of ²²⁸Ra and ²²⁸Th, radioactive equilibrium between ²³²Th and its daughters was evaluated. In equilibrium, the average activity of ²²⁸Ra and ²²⁸Th can represent ²³²Th activity. As shown in Figure S1, ²²⁸Ra and ²²⁸Th exhibit a strong linear correlation (R²: 0.93, slope: 1.07, p < 0.001), confirming secular equilibrium. Consequently, ²³²Th activities were calculated, ranging from 37.8 to 91.7 Bq/kg (mean 58.2 ± 8.7 Bq/kg;

Table 1. Factor loadings for the first three components.

	Component
	PC1	PC2	PC3
210Pbex	−0.026	0.962	−0.016
7Be	0.016	0.943	0.125
137Cs	0.258	−0.198	−0.759
238U	0.887	0.012	−0.125
226Ra	0.917	0.003	0.029
232Th	0.801	−0.060	0.508
40K	0.350	−0.064	0.729

Extraction Method: Principal Component Analysis; Rotation Method: Varimax with Kaiser Normalization. Note: Factor loadings indicate statistical associations only. The grouping of ¹³⁷Cs and ⁴⁰K in PC3 reflects shared alkali-metal chemistry but does not imply common sources or transport pathways (see Section 3.2 for detailed interpretation).

). The anthropogenic radionuclide ¹³⁷Cs showed activities ranging from below the detection limit (<0.2 Bq/kg) to 5.7 Bq/kg, with a mean of 1.5 ± 1.4 Bq/kg.

The variability in radionuclide activities is further depicted in Figure 2, which shows their frequency distributions. Positive skewness coefficients indicate right-skewed distributions. While most radionuclides exhibit slight positive skewness, ²²⁶Ra, ²³⁸U, and ⁴⁰K have distributions close to normal (skewness 0.19–0.48). In contrast, ⁷Be, ²¹⁰Pb, ²¹⁰Pb_ex, and ¹³⁷Cs display highly skewed, non-normal distributions (Figure 2, Table S1). Low kurtosis values for most terrestrial radionuclides suggest near-normal distributions, whereas airborne radionuclides show exceptionally high kurtosis: ²¹⁰Pb (16.43), ²¹⁰Pb_ex (16.35), and ⁷Be (28.93), indicating heavy-tailed distributions. Thus, except for ²¹⁰Pb, ²¹⁰Pb_ex, and ⁷Be, most lithogenic radionuclides approximate Gaussian distributions (Figure 2). Notably, the highest measured values of ²²⁶Ra, ²³⁸U, ⁴⁰K, and ¹³⁷Cs fall within 3σ of the mean, while those of ²³²Th, ²²⁸Ra, and ²²⁸Th fall within 4σ, and the maximum values of ²¹⁰Pb and ⁷Be fall within 5σ and 6σ, respectively.

Comparative data from the literature are presented in Table S2. The activities of ²³⁸U, ²²⁶Ra, ²³²Th, and ⁴⁰K in this study align with those reported for mountain, agricultural, and industrial soils (Table S2). However, activities in Zhoushan Island soils exceed global averages reported by UNSCEAR (35 Bq/kg for ²³⁸U and ²²⁶Ra, 30 Bq/kg for ²³²Th, and 400 Bq/kg for ⁴⁰K) [36] and values from Svalbard, an Arctic island [39,40]. Conversely, ²²⁶Ra, ²³²Th, and ⁴⁰K activities in Zhoushan are lower than those in volcanic tuff soils from Cappadocia, Turkey [41], likely due to differences in mineral composition and soil provenance. The ¹³⁷Cs activity in Zhoushan soils (<0.2–5.7 Bq/kg) is comparable to values reported in the Southern Rechna of Pakistan (<1.35–6.5 Bq/kg) [42], the South Shetland Archipelago (1.59–15.6 Bq/kg) [43], and the Alexandria Region of Egypt (0.11–7.86 Bq/kg) [44], but significantly lower than in areas heavily impacted by the Chernobyl accident (Table S2). Notably, ²¹⁰Pb and ⁷Be activities in Zhoushan surface soils are substantially higher than those reported in most other regions and countries (Table S2).

For broader context, our results can be compared with the comprehensive review of soil radioactivity in Serbia by Kuzmanović [45], which reported mean activities of ⁴⁰K (620–720 Bq/kg), ²²⁶Ra (35–55 Bq/kg), and ²³²Th (30–50 Bq/kg) across diverse geological settings. The corresponding values in Zhoushan (⁴⁰K: 631 Bq/kg; ²²⁶Ra: 44.3 Bq/kg; ²³²Th: 58.2 Bq/kg) fall within or slightly above these ranges, consistent with the thorium-enriched lithological background of the study area. A broader international comparison of soil radionuclide concentrations is provided in Table S2, enabling direct benchmarking of our dataset against global references.

3.2. Multivariate Exploratory Analysis

To explore potential groupings among radionuclides and to generate hypotheses regarding their sources and controlling processes, hierarchical cluster analysis (HCA), principal component analysis (PCA), and Pearson correlation analysis were employed as exploratory statistical tools. These methods were used to identify underlying statistical patterns, which were subsequently evaluated against independent spatial distributions and geochemical evidence (Section 3.3). It should be noted that statistical associations alone do not necessarily imply causal relationships; therefore, all process-based interpretations were further validated using geomorphological characteristics and land-use information. HCA was applied to classify radionuclides according to their similarity, producing dendrograms that illustrate the degree of association among variables. In the dendrograms, shorter linkage distances indicate stronger similarities between groups. Pearson correlation analysis was used to quantify linear relationships among radionuclides, whereas PCA was employed to identify dominant association patterns, distinguish radionuclide behaviors, and infer their potential sources and controlling mechanisms.

The HCA dendrogram (Figure 3a) shows that the seven radionuclides in Zhoushan Island soils form three clearly separated clusters. The first cluster contains ²¹⁰Pb_ex and ⁷Be, which are strongly grouped together. The second cluster includes the lithogenic radionuclides ²³⁸U, ²²⁶Ra, and ²³²Th. The third cluster consists of ⁴⁰K and ¹³⁷Cs. PCA was applied to further explore these relationships and identify the underlying factors controlling radionuclide distribution. According to Kaiser’s criterion (eigenvalue > 1), three significant principal components (PCs) were extracted, explaining a total of 81.75% of the variance (Table S3). The factor loadings of each radionuclide on the PCs are shown in Table 1.

PC1 accounts for 37.01% of the total variance and exhibits high positive loadings for ²³⁸U (0.887), ²²⁶Ra (0.917), and ²³²Th (0.801), indicating a common terrestrial origin and association with natural background radioactivity. PC2 explains 27.44% of the variance and shows high positive loadings for ⁷Be (0.943) and ²¹⁰Pb_ex (0.962), indicating a common atmospheric source, likely from wet/dry deposition. PC3 accounts for 17.30% of the variance and is characterized by high loadings for ⁴⁰K (0.729) and 137Cs (−0.759). Although both radionuclides are alkali elements and may exhibit similar ion-exchange behavior in soils, their opposite loading signs indicate inverse spatial distributions. This contrast reflects fundamentally different sources: ⁴⁰K is predominantly lithogenic, whereas ¹³⁷Cs is anthropogenic and likely distinct retention mechanisms in soils, such as the strong fixation of ¹³⁷Cs in illite-rich clay minerals [46]. It is important to emphasize that chemical similarity alone is insufficient to infer shared controlling processes.

The score plot of PC1 versus PC2 (Figure 3b) clearly delineates three distinct radionuclide groups, aligning with the HCA results. The combination of HCA and PCA confirms three distinct behavioral patterns of radionuclides in Zhoushan Island soils: (1) ²³⁸U, ²²⁶Ra, and ²³²Th, associated with terrestrial sources and natural background radioactivity; (2) ²¹⁰Pb_ex and ⁷Be, derived from atmospheric deposition; and (3) ⁴⁰K and ¹³⁷Cs, exhibiting similar geochemical characteristics due to their alkali metal properties.

The Pearson correlation coefficients among the measured radionuclides are shown in Figure 4. Intermediate but statistically significant positive correlations were observed among the terrestrial radionuclides ²³⁸U, ²²⁶Ra, ²³²Th and ⁴⁰K: ²³⁸U−²²⁶Ra (r = 0.71, p < 0.05), ²³⁸U−²³²Th (r = 0.64, p < 0.05), ²²⁶Ra−²³²Th (r = 0.70, p < 0.05), and ⁴⁰K−²³²Th (r = 0.58, p < 0.05). These correlations further support the interpretation that these radionuclides share a common lithogenic origin. A strong positive correlation was also observed between ²¹⁰Pb_ex and ⁷Be (r = 0.83, p < 0.05), consistent with their atmospheric deposition source. In contrast, ¹³⁷Cs showed no significant correlation with any other radionuclide, indicating distinct behavior and sources relative to both terrestrial and atmospheric nuclides.

3.3. Spatial Distribution Characteristics and Influencing Factors

Although the spatial coverage and resolution of the sampling points are limited, predictive distribution maps were generated using inverse distance weighting (IDW) interpolation (power parameter = 2). Given the relatively small sample size (n = 43) over an area of 491 km², substantial uncertainty remains in unsampled regions, particularly in coastal zones characterized by strong environmental gradients. Leave-one-out cross-validation indicates poor predictive performance, with high normalized RMSE values (13–307% across radionuclides) and negative R2 values (Table S4), suggesting that IDW performs worse than a simple mean-based estimator. This limited performance is primarily attributable to low sampling density (~0.09 samples/km²) and pronounced spatial heterogeneity of atmospherically derived radionuclides (⁷Be, ²¹⁰Pb, and ²¹⁰Pb_ex), which are strongly influenced by episodic deposition events. In addition, the presence of extreme outliers (e.g., ZS31a with ⁷Be = 585.5 Bq/kg and ZS42 with ²¹⁰Pb = 1490.3 Bq/kg), far exceeding local background levels, further reduces interpolation stability and reliability. Accordingly, the resulting spatial distribution maps (Figure 5) should be interpreted as qualitative representations of broad spatial trends rather than precise predictions at unsampled locations. The geostatistical approach nevertheless provides a first-order spatial characterization of radionuclide distributions across the island, including unsampled areas. The spatial distributions of ¹³⁷Cs, ⁷Be, ²¹⁰Pb_ex, ²¹⁰Pb, ²²⁶Ra, ²³⁸U, ²³²Th, and ⁴⁰K in Zhoushan Island surface soils are presented in Figure 5a–h.

The surface soils in the northwestern, northeastern, and southeastern regions of Zhoushan Island exhibit distinctly low levels of ¹³⁷Cs. In contrast, scattered patchy high-value zones are observed in the central part of the island from east to west. Further analysis shows that these high-value areas are predominantly located in farmland, vegetable plots, and cultivated land, while forested, residential, and industrial areas generally display very low ¹³⁷Cs activities (Table 1; Figure 5a). This spatial pattern suggests a potential influence of agricultural disturbance on the redistribution of ¹³⁷Cs. In particular, plowing and repeated soil mixing may have transported legacy fallout ¹³⁷Cs from subsurface horizons to the topsoil, thereby elevating surface activities in cultivated areas. However, this interpretation is based solely on spatial correspondence and remains tentative. In the absence of supporting information on soil texture, organic matter content, particle size distribution, or detailed land-use and cultivation history, agricultural disturbance cannot be identified as the dominant controlling factor. Alternative processes, such as differential erosion, sediment reworking, or spatially variable atmospheric deposition, may also contribute to the observed distribution and cannot be ruled out.

The spatial distribution maps of ⁷Be, ²¹⁰Pb_ex, and ²¹⁰Pb are highly similar, with higher concentrations in the central region of the island and lower concentrations in the eastern and southwestern regions (Figure 5b–d). These patterns are controlled by multiple environmental factors, including spatial variability in rainfall, elevation, and terrain. For example, the highest concentrations of ⁷Be and ²¹⁰Pb were recorded at sampling sites ZS31a (curbside) and ZS42 (residential area), both of which are located near sea level. This observation is consistent with the hypothesis that low-lying areas may preferentially accumulate atmospheric fallout radionuclides, potentially due to enhanced retention and/or reduced erosion intensity. However, in the absence of concurrent meteorological information (e.g., precipitation amount and intensity and wind field characteristics) and a detailed topographic analysis based on high-resolution digital elevation data, this interpretation remains speculative and should be treated with caution.

The interpolated maps of ²²⁶Ra, ²³⁸U, and ²³²Th (Figure 5e–g) display very similar spatial patterns. When the island is divided along a diagonal from the northwest to the southeast, higher activities are concentrated in the northern part of the island, while lower values dominate the southern part. This distribution likely reflects the natural radioactive background controlled by soil-forming processes. The weathering of parent rocks, rock type, and mineralogical composition are the primary factors influencing the concentrations of these lithogenic radionuclides in surface soils.

In contrast to ²²⁶Ra, ²³⁸U, and ²³²Th, the spatial distribution of ⁴⁰K (Figure 5h) shows an inverse pattern: the lowest values occur in the central region, surrounded by higher values in the eastern, northwestern, and southwestern parts of the island. Although agricultural land is typically enriched in ⁴⁰K due to potassium fertilizer application, no significant increase in ⁴⁰K content was observed in agricultural topsoils on Zhoushan Island (Table S1; Figure 5h). This indicates that land-use type exerts minimal influence on ⁴⁰K distribution, and geogenic factors may be more dominant.

The spatial distributions of both fallout and terrestrial radionuclides in Zhoushan Island surface soils show no abnormally elevated areas. The spatial variability reflects the combined effects of human agricultural activity (¹³⁷Cs), atmospheric deposition processes (⁷Be and ²¹⁰Pb_ex), and geological background conditions (²²⁶Ra, ²³⁸U, ²³²Th, and ⁴⁰K). Conclusively, the strong coherence within each group suggests that radionuclide behavior is governed by distinct transport pathways rather than random spatial variability.

3.4. Radiological Hazards

While we acknowledge the inherent limitations of conventional radiological indices, particularly their inability to capture radionuclide mobility, bioavailability, and process-specific redistribution pathways, these standard metrics are still reported to facilitate direct comparison with global baseline studies [36] and to meet regulatory screening requirements. Importantly, values below established safety thresholds should not be interpreted as indicating negligible environmental risk, especially where process-based evidence suggests ongoing redistribution (e.g., ¹³⁷Cs mobilization associated with agricultural disturbance). Future assessments should therefore integrate complementary approaches, such as sequential extraction procedures or dose coefficient frameworks adjusted for bioaccessibility, to better constrain radionuclide behavior and improve environmental risk characterization beyond conventional screening indices.

The Ra_eq values (Figure 6a) ranged from 123 to 251 Bq/kg, all lower than the recommended limit of 370 Bq/kg. A Ra_eq value of 370 Bq/kg corresponds to an annual effective dose of 1.05 mSv [36], which is considered the threshold for radiotoxicity [47]. Therefore, there is no significant gamma radiation hazard in the Zhoushan Island area. Among the radionuclides, ²³²Th was the dominant contributor to Ra_eq, accounting for 43.2–52.5%.

The DR ranged from 57.8 to 116 nGy/h (Figure 6b). The lowest DR was observed at site ZS25, located in the central mountainous area of Zhoushan Island, while the highest was observed at ZS22 in the north–central part of the island. Only ZS25 showed a lower DR than the worldwide average of 59 nGy/h [36], whereas all other sites exceeded this value. Nevertheless, all DR values were well below the recommended safety threshold of 1000 nGy/h set by the International Commission on Radiological Protection [48]. The AEDE and ELCR varied from 0.071 to 0.142 mSv/y and from 0.248 × 10⁻³ to 0.497 × 10⁻³, respectively (Figure 6b). AEDE values were higher than the global mean of 0.07 mSv/y [36] but remained lower than the ICRP-recommended limit of 1 mSv/y [48]. Except for ZS08 and ZS25, ELCR values at all sampling sites exceeded the global average of 0.290 × 10⁻³ [36]. Similar to Ra_eq, ²³²Th was the primary contributor to DR, AEDE, and ELCR, accounting for 38.8–48.5%. The Hext ranged from 0.333 to 0.679 (Figure 6c), with a mean of 0.476 ± 0.057, which is below the recommended limit of 1 [48], suggesting insignificant external radiological risk. The AGDE ranged from 408 to 816 μSv/y (Figure 6d), with a mean of 579 ± 68 μSv/y. All samples exceeded the global average of 300 μSv/y [36]; however, AGDE values were still lower than the ICRP threshold of 1 mSv/y [48]. The relative contributions of ²²⁶Ra, ²³²Th, and ⁴⁰K to AGDE were similar to those for DR, with median contributions of 24.6%, 42.7%, and 32.0%, respectively, indicating ²³²Th as the major source.

Overall, the radiological hazard indices exhibited wide spatial variation, likely due to differences in lithological composition across the study area. High values were mainly distributed in the northern and south-central regions, while low values were concentrated in the southwestern and central mountainous areas. The lowest hazard indices were observed at ZS25 (central mountainous region), whereas the highest were recorded at ZS22 (north–central region). Although DR, AEDE (outdoors), AGDE, and ELCR were generally higher than the worldwide mean values reported by UNSCEAR [36], all Ra_eq, DR, AEDE, and AGDE values remained below the threshold levels recommended by ICRP [48].

The radiological parameters were evaluated for three land-use types relevant to tourism management: parks, schools, and ports (Figure 6). Mean Raeq values were 168.34 Bq/kg (parks), 156.49 Bq/kg (schools), and 173.50 Bq/kg (ports), with corresponding DR values of 78.52, 73.32, and 80.21 nGy/h, and AEDE (outdoors) of 0.096, 0.090, and 0.098 mSv/y, respectively. ELCR, Hext, and AGDE followed similar trends, all remaining well below international safety limits [48], indicating negligible health risks. Ports consistently showed slightly higher values than parks and schools, likely due to accumulated materials from cargo and industrial activities. The process-based grouping of radionuclides suggests that anthropogenic disturbance can enhance redistribution. From a tourism management perspective, we recommend routine monitoring in parks and schools to maintain public reassurance, and periodic assessments in ports given their elevated background. While radiological indices indicate low risk, their interpretation is limited as they do not account for radionuclide mobility, bioavailability, or process-driven redistribution, which may be particularly relevant in sensitive areas such as parks, ports, and schools. These findings underscore the importance of integrating regular radiological assessments into sustainable environmental management for recreational and transport areas on tourism-oriented islands like Zhoushan.

4. Conclusions

This study demonstrates that radionuclide distributions in Zhoushan Island soils are governed by three distinct process domains: atmospheric deposition (⁷Be and ²¹⁰Pb_ex), lithogenic inheritance (²³⁸U, ²²⁶Ra, ²³²Th), and disturbance-modulated redistribution (¹³⁷Cs and ⁴⁰K). By integrating multivariate statistical analysis with process-based interpretation, this study extends conventional descriptive approaches by providing a framework that explicitly links radionuclide occurrence to underlying environmental controls and transport mechanisms. The results reveal that spatial variability is not merely a function of concentration differences but reflects the interaction of atmospheric input, geomorphological conditions, and human activities. In particular, agricultural disturbance plays a key role in redistributing ¹³⁷Cs, while low-lying terrains enhance the accumulation of fallout radionuclides.

Although radiological hazard indices remain within international safety limits, their interpretation is constrained by the lack of process considerations. Elevated values relative to global averages are primarily associated with natural lithogenic enrichment rather than anthropogenic contamination, underscoring the importance of distinguishing between background variability and pollution signals. From a broader perspective, this study highlights that radionuclide behavior in coastal and island environments is highly sensitive to both climate variability and land-use change. The findings emphasize the need for process-informed radiological assessment frameworks that incorporate source identification, environmental transport, and redistribution mechanisms.

This study is limited by the spatial density of surface sampling and the absence of depth-resolved profiles, which restricts the quantification of radionuclide migration and temporal dynamics. Future work should integrate vertical profiling, isotope ratio analysis, and sediment flux quantification to better constrain transport processes and improve environmental risk assessment.