Multivariate Statistics, Radioactivity and Radiological Hazard Evaluation in Marine Sediments of Selected Areas from Sicily, Southern Italy

Abstract

This work reports the findings of an investigation aimed at assessing, for the first time, the natural and anthropogenic radioactivity content of marine sediments collected from selected areas of Sicily, Southern Italy. In particular, it focused on evaluating the average activity concentration of detected radionuclides and the radiological hazard for humans, above all considering the use of this material for nourishing actual eroded beaches. To this aim, the quantification of the average specific activity of ²²⁶Ra, ²³²Th, and ⁴⁰K natural and ¹³⁷Cs anthropogenic radioisotopes was addressed through the employment of High-Purity Germanium (HPGe) gamma-ray spectrometry. Furthermore, the absorbed gamma dose rate (D), the annual effective dose equivalent outdoor (AEDE_out), the external hazard index (H_ex), and the excess lifetime cancer risk (ELCR) were also calculated to evaluate the radiological hazard for humans related to external exposure to ionizing radiations. Furthermore, the average specific activity of ¹³⁷Cs was found to be less than the lowest detectable activity in all cases, excluding anthropogenic radioactive contamination of the investigated samples. Finally, Pearson correlation, principal component analysis (PCA) and hierarchical cluster analysis (HCA), i.e., multivariate statistics, were carried out by analyzing detected radioactivity and radiological characteristics to evaluate their relationship with the sampling locations.

1. Introduction

Natural radioactivity in the environment is primarily caused by cosmogenic, i.e., ¹⁴C, ⁷Be, ³H, and primordial radionuclides in the Earth’s crust, accounting for the majority of the population’s exposure to radiation [1]. In addition, anthropogenic radioisotopes, such as ⁶⁰Co, ¹³⁴Cs, and ¹³⁷Cs (corrosion products of nuclear power plants, deposited in the soil as fallout), as well as ⁹⁹Tc, due to the nuclear medicine application, can contribute to radiation exposure [2]. Of them, radiocaesium is the most significant one, due to its longer radioactive half-life [3].

Among environmental matrices, an interesting deposit of material for artificial nourishment of beaches currently undergoing erosion is represented by paleo-beaches, for this reason defined as Deposits of Marine Relict Sediments (the Italian acronym is DSMR) [4,5,6].

Nevertheless, it has to be noticed that sediments are known to accumulate natural radionuclides, i.e., uranium (²³⁸U and ²³⁵U) and thorium (²³²Th) decay chain products and ⁴⁰K, principally via weathering, erosion, and depositional processes of various geological materials, with concentrations which generally increase upon grain size decreases [7]. In this context, it is important to highlight that erosion and accretion processes, i.e., sediment dynamics, can be identified using radioactivity maps in the sediments themselves [8]. In fact, among scientific methods utilized to investigate the evolution of sediment sources and deposition rates, the measurement of radionuclide concentrations, the latter acting as tracers, is largely employed to determine the geochronology of sediments [9,10,11].

In particular, in the case of marine sediments, in view of their usage for beach nourishment, even if the material has been considered compatible from the point because of the grain size, the color, and the composition, the evaluation of the natural and anthropogenic radioactivity appears crucial for an estimation of the potential public health hazard of radio contamination in coastal areas [12,13,14,15], considering the great attraction these beaches exert on tourists, other than providing important information regarding the origin and fate of radionuclides in aquatic ecosystems [7].

To this aim, it is worth noting that radioactivity concentration in the marine environment has been identified as a significant pollution hazard [16]. Radionuclides adhere to particles in the marine ecosystem and eventually accumulate in sediments. As a result, the sediment is a significant source of contamination in aquatic settings [17], and thus there is the need of mapping radioactivity on seabed using lab-based and in situ methods, especially in many areas with high interest (e.g., industrial areas with enhanced NORMs, coastal areas near to nuclear power plants) [18,19,20]. In addition, radionuclides can accumulate in marine organisms’ tissues through a variety of processes. As a result, radionuclides enter the food chain through marine foods [21,22].

Some researchers reported natural radioactivity in the sediments of the Mediterranean Sea [8,17,23].

In terms of environmental laws and regulations, the European Union has made substantial efforts to achieve a satisfactory ecological condition in aquatic habitats for all Member States. Although such approaches are made clear in the Water Framework Directive (Directive 2000/60/EC), the Marine Strategy Framework Directive (Directive 2008/56/EC), or the Environmental Quality Standards in the context of water policy (Directive 2008/105/EC), radionuclides are not specifically addressed in these documents [24]. Conversely, it is plausible to hypothesize that the introduction of further regulatory frameworks would demand precise field data on NORM exposure, in order to identify polluted areas and design risk-based management methods [25].

In this study, marine sediments from the Gulf of Termini Imerese, Sicily, Southern Italy, were analyzed to identify and quantify natural and artificial gamma-emitting radionuclides. The analysis was carried out by means of High-Purity Germanium (HPGe) gamma spectrometry, in order to investigate the activity concentrations of ²²⁶Ra (in secular equilibrium with ²³⁸U), ²³²Th, and ⁴⁰K natural radioisotopes, with the aim to effectively record the radioactivity background levels. Furthermore, the specific activity of ¹³⁷Cs was examined to ascertain the presence of any potential artificial radionuclides. Moreover, in order to estimate any possible radiological hazard for human beings, related to external exposure to ionizing radiations, the absorbed gamma dose rate (D), the annual effective dose equivalent outdoor (AEDE_out), the external hazard index (H_ex), and the excess lifetime cancer risk (ELCR) were evaluated.

Lastly, multivariate statistical analyses, such as Pearson correlation, principal component analysis (PCA), and hierarchical cluster analysis (HCA), were performed on the gathered data to examine the radioactivity and radiological parameters in the studied samples and find any potential relationships between the variables under investigation [26,27].

2. Materials and Methods

2.1. Sampling

Fifty samples of marine sediments, around 500 g each, were collected by Mediterranea Sabbie Srl society in ten selected sites (ID#, # = 1, …, 10), five samples for each site, of the Gulf of Termini Imerese, Sicily, Southern Italy (see

Table 1. Site IDs, GPS coordinates, and sampling depth.

Site ID	GPS Position (Reference System WGS84/UTM Zone 33 N EPSG: 32633)		Sampling Depth (m)
	Latitude	Longitude
1	376901	4217704	121
2	377100	4217430	121
3	377642	4217097	128
4	378101	4216713	126
5	378217	4216657	128
6	378453	4216513	128
7	379008	4215999	124
8	380841	4215441	129
9	383289	4214482	131
10	384164	4214475	137

and Figure 1), for subsequent laboratory measurements.

The collection of samples was performed out in the following manner: the device used for sampling (carotaker) was armed and lowered at an even rate to reach the bottom in the proper place. In connection with the background, the technician recorded the GPS data and recovered the sampling gear. As soon when the sampler reached the surface, it was swiftly retrieved to prevent stress that influenced its contents, rinsed externally to prevent contamination, and its contents filled into a tub before being stored in acidified polyethylene holders that were tightly sealed and labeled to prevent radionuclide precipitation and absorption on the exterior of the container.

2.2. HPGE Gamma Spectrometry Analysis

Investigated sediments were dried until all moisture was gone and a consistent mass was reached in order to do HPGe gamma spectrometry analysis. After that, they were put in 250 mL hermetically sealed Marinelli containers so that they could be distributed evenly throughout the detector. The samples were ready for gamma spectrometry counting after 40 days, when the secular radioactive equilibrium between ²²⁶Ra and its daughter products was achieved [28]. Samples were counted for 70,000 s, and the spectra were examined to determine the specific activity of ²²⁶Ra, ²³²Th, ⁴⁰K, and ¹³⁷Cs in order to lower statistical uncertainty. In particular, the ²²⁶Ra activity concentration was calculated using the 295.21 keV and 351.92 keV ²¹⁴Pb and 1120.29 keV ²¹⁴Bi gamma-ray lines, and the ²³²Th specific activity was determined using the 911.21 keV and 968.97 keV ²²⁸Ac γ-ray lines. Then, for ⁴⁰K, the evaluation was performed from its γ line at 1460.8 keV and, finally, to investigate the artificial radioactivity content, the ¹³⁷Cs specific activity was quantified from its γ line at 661.66 keV.

The experimental set-up consisted of a negatively biased Ortec HPGe detector (GMX) (Ametek Ortec, Oak Ridge, TN, USA), the operating parameters of which are given in

Table 2. GMX settings.

GMX Detector
Parameter	Value
FWHM	1.94 keV
Peak/Compton	65:1
εr	37.5% (at the 1.33 MeV 60Co γ–line)
ΔV	−4800 V
ΔE	5 keV–2 MeV

[29].

The detector was placed in lead wells to safeguard it from the background radiation coming from the environment, and the energy and efficiency calibrations were performed using a 250 mL multipeak Marinelli geometry gamma source (BC-4464) (Eckert & Ziegler, Braunschweig, Germany), with an energy range from 60 keV to 1836 keV.

The activity concentration (Bq kg⁻¹ dry weight, d.w.) of the investigated radionuclides was estimated as follows [30,31]:

$$\mathrm{C}={\displaystyle \frac{{\mathrm{N}}_{\mathrm{E}}}{{\epsilon }_{\mathrm{E}}\mathrm{t}{\gamma }_{\mathrm{d}}\mathrm{M}}}$$

(1)

where N_E indicates the net area of a peak at energy E, ε_E and γ_d are the efficiency and yield of the photopeak at energy E, respectively, M is the mass of the sample (kg), and t is the live time (s).

Finally, the Italian Accreditation Body (ACCREDIA) has formally acknowledged the elevated quality of HPGe analysis outcomes [32]. This implies the continuous verification (with an annual periodicity) of the maintenance of the performance characteristics of the gamma spectrometry method. In particular, the accuracy of the results was verified by analysis of certified reference materials by means of the u-test method, according to the following acceptability criterion [33]:

$${\mathrm{u}}_{\mathrm{t}\mathrm{e}\mathrm{s}\mathrm{t}}={\displaystyle \frac{\mathrm{m}\mathrm{e}\mathrm{a}\mathrm{s}\mathrm{u}\mathrm{r}\mathrm{e}\mathrm{d}-\mathrm{r}\mathrm{e}\mathrm{f}\mathrm{e}\mathrm{r}\mathrm{e}\mathrm{n}\mathrm{c}\mathrm{e}}{\sqrt{{\mathrm{u}}_{\mathrm{m}\mathrm{e}\mathrm{a}\mathrm{s}}^2-{\mathrm{u}}_{\mathrm{r}\mathrm{e}\mathrm{f}}^2}}}$$

(2)

where “measured” and “reference” indicate the experimentally measured and the reference values, respectively, while u_meas and u_ref are their uncertainties.

2.3. Evaluation of Radiological Hazard

To evaluate any possible radiological harm to humans, a number of radiological parameters were computed, including the absorbed gamma dose rate, the annual effective dose equivalent outdoor, the external hazard index, and the excess lifetime cancer risk.

2.3.1. Absorbed Gamma Dose Rate

The calculation of the absorbed gamma dose rate constitutes the initial significant stage in the evaluation of health hazard [34]:

D (nGy h⁻¹) = 0.462C_Ra + 0.604C_Th + 0.0417C_K

(3)

where C_Ra, C_Th, and C_K are the mean specific activities of ²²⁶Ra, ²³²Th, and ⁴⁰K in the investigated samples, respectively.

2.3.2. Annual Effective Dose Equivalent Outdoor

The annual effective dose equivalent received by an individual outdoors has been quantified using the following equation, assuming an outdoor occupation of 20% [35]:

AEDE_out (mSv year⁻¹) = D (nGy h⁻¹) × 8760 h × 0.7 Sv Gy⁻¹ × 0.2 × 10⁻⁶

(4)

This value must be lower than 1 mSv year⁻¹ to ensure a negligible radiological hazard for humans [36].

2.3.3. External Hazard Index

In order to limit the radiation dose to 1 mSv year⁻¹, the external radiation hazard index was defined [37]:

H_ex = (C_Ra/370 + C_Th/259 + C_K/4810)

(5)

It must not exceed the limit of unity for the radiation hazard to be negligible.

2.3.4. Excess Lifetime Cancer Risk

The excess lifetime cancer risk index expresses the probability of developing cancer during a lifetime at a given level of exposure [38]. It represents the number of additional cancers expected in a given population due to exposure to a carcinogen at a given dose:

ELCR = AEDE_out × D_L × R_F

(6)

where AEDE_out is the annual effective dose equivalent outdoor, D_L the average duration of life, i.e., 70 years, and R_F the risk factor (Sv⁻¹), i.e., fatal cancer risk per Sievert. For stochastic effects, International Commission on Radiological Protection (ICRP) recommends a value of 0.05, for the public, for this last parameter [39].

2.4. Statistical Treatments

Multivariate statistical analyses were performed using Origin Lab Pro 2021 version 9.4 (OriginLab, Northampton, MA, USA) [40].

Specifically, principal component analysis (PCA) was employed as an exploratory tool to reduce data dimensionality [41], assess correlations among the original variables via Pearson’s correlation coefficients and identify the principal components (PCs) that capture the largest proportion of variance [42]. Prior to PCA, the Kaiser–Meyer–Olkin (KMO) measure of sampling adequacy and Bartlett’s test of sphericity were conducted to evaluate the data’s suitability for factor analysis [43]. The resulting KMO value of 0.64 and a significant Bartlett’s test (χ² = 74.89, p < 0.0001) indicated that the correlation matrix was suitable for PCA. Furthermore, hierarchical cluster analysis (HCA) with Ward’s linkage and Euclidean distance was used to group samples based on their similarity in radiological parameters and radionuclide activities [44], effectively reducing the number of individual observations.

3. Results and Discussion

3.1. The Specific Activities of Detected Radioisotopes

Table 3. Mean specific activities C_Ra, C_Th, C_K, and C_Cs of ²²⁶Ra, ²³²Th, ⁴⁰K and ¹³⁷Cs, respectively, in investigated samples.

Site ID	CRa (Bq kg−1 d.w.)	CTh (Bq kg−1 d.w.)	CK (Bq kg−1 d.w.)	CCs (Bq kg−1 d.w.)
1	22.4 ± 3.2	41.1 ± 6.3	500 ± 71	<0.10
2	20.7 ± 4.9	40.1 ± 9.7	522 ± 122	<0.12
3	21.4 ± 3.4	38.6 ± 8.2	426 ± 101	<0.09
4	19.4 ± 4.6	39.3 ± 9.5	510 ± 120	<0.13
5	20.4 ± 4.8	37.4 ± 9.1	487 ± 114	<0.11
6	19.8 ± 2.3	38.6 ± 4.9	452 ± 51	<0.09
7	19.9 ± 4.7	38.5 ± 9.4	521 ± 119	<0.08
8	16.5 ± 3.9	31.4 ± 4.5	431 ± 75	<0.10
9	20.5 ± 2.9	37.8 ± 5.8	449 ± 64	<0.10
10	16.6 ± 4.4	36.9 ± 5.1	420 ± 72	<0.09

reports the mean specific activities C_Ra, C_Th, C_K, and C_Cs, of detected radionuclides, i.e., ²²⁶Ra, ²³²Th, ⁴⁰K, and ¹³⁷Cs, in the investigated marine sediments.

In particular, for the natural radioactivity content, it is important to point out that the mean activity concentration of ²²⁶Ra is lower than the world average value of 35 Bq kg⁻¹ for all site IDs [1]. In addition, the mean specific activities of ²³²Th and ⁴⁰K exceed the global average of 30 Bq kg⁻¹ and 400 Bq kg⁻¹, respectively, in all cases. Moreover, obtained results are in good agreement with similar data from other regions worldwide [7,17,23,45,46].

Furthermore, the average specific activity of ¹³⁷Cs was found to be less than the lowest detectable activity in all cases, excluding significant anthropogenic radioactive contamination of the samples analyzed.

It is worth underlining that, as frequently reported in the literature, values of C_Ra, C_Th, and C_K turned out to be contingent on the chemical composition and mineralogical phases of the samples under investigation [47,48,49]. In this sense, future investigations through multi-scale analytical techniques, such as X-ray fluorescence spectroscopy, micro-Raman scattering, and X-ray diffraction will explore the chemical and mineralogical composition of these samples in more detail.

Additionally,

Table 4. Statistical data pertaining to CRa, CTh, and CK.

Statistically Function	CRa	CTh	CK
Min	16.5	31.4	420
Max	22.4	41.1	521
Mean	19.76	37.96	471.7
Geometric mean	19.76	38.50	475.58
Median	20.4	38.51	493.5
Standard deviation	1.89	2.61	40.39
Skewness	−0.67	0.51	−0.24
Kurtosis	−0.86	−1.00	−1.96

provides key statistical parameters of the dataset. These fundamental statistical parameters are essential for characterizing the dataset’s distribution. It should be noted that the data exclusively pertain to the specific activities of natural radionuclides.

Statistical analysis revealed distinct distribution characteristics for each radionuclide. Skewness values indicated asymmetrical distributions across all samples. Specifically, C_Ra and C_K exhibited negative skewness, suggesting a higher probability of observing activity concentrations above the mean, with a tail extending towards lower values. Conversely, C_Th displayed positive skewness, indicating a distribution skewed towards lower activity concentrations, with a tail towards higher values.

Kurtosis values provided further insights into the distributional shape. The negative kurtosis observed for all three radionuclides suggests platykurtic distributions, characterized by flatter peaks and thinner tails compared to a normal distribution. This implies a lower likelihood of encountering extreme activity concentrations within the dataset [50].

The evaluation of natural radionuclide activity concentrations was also conducted to develop a frequency distribution model, as illustrated through specialized graphical representations (Figure 2).

From a statistical perspective, the skewness values obtained indicate an asymmetric distribution, while kurtosis coefficients confirm that the empirical distribution of C_Ra, C_Th, and C_K, in agreement with the null kurtosis value, can be considered as normal.

It is worth noting that values of C_Ra and C_Th were higher than the mean ones in 60% of the samples, whereas in the 50% of the marine sediments the ⁴⁰K specific activity exceeded the corresponding average value.

Regarding ¹³⁷Cs, the mean specific activity was determined to be lower than the minimum detectable activity in all cases, thereby ruling out the possibility of anthropogenic radioactive contamination of the investigated samples.

3.2. Radiological Hazard Assessment

Table 5. Radiological hazard indices in sampling sites investigated.

Site ID	D (nGy h−1)	AEDEout (µSv year−1)	Hex	ELCR (×10−3)
1	56.0	68.7	0.32	0.24
2	55.4	68.0	0.32	0.24
3	51.1	62.5	0.30	0.22
4	54.0	66.2	0.31	0.23
5	52.3	64.2	0.30	0.22
6	51.3	62.9	0.30	0.22
7	54.2	66.4	0.31	0.23
8	44.6	54.6	0.26	0.19
9	51.0	62.6	0.29	0.22
10	47.5	58.2	0.27	0.20
Average	51.7	63.4	0.30	0.22

reports the calculated values of D, AEDE_out, H_ex, and ELCR.

In particular:

(i)

The variability of the absorbed doses was found to be correlated with the different lithological components of the sampling sites under consideration [51,52];

(ii)

The annual effective dose equivalent outdoor was lower than the threshold limit of 1 mSv year⁻¹ [36] in all cases;

(iii)

The external hazard index was lower than unity in all cases.

(iv)

A linear relationship between ELCR and AEDE_out was found, as clearly visible in Figure 3, according to the literature [53].

3.3. Statistical Features

As reported in [54,55], in parametric statistical elaboration, determining a specific data distribution is critical. This is because if the assumption of a normal distribution is violated, the results may be erroneous or untrustworthy. As a result, before starting with any relevant statistical processes, it is critical to test the sufficiency of the assumption of a normal distribution of data. For this objective, the Shapiro–Wilk, Anderson–Darling, and Lilliefors tests were done [56].

The results of these tests are presented in

Table 6. p-value results of various statistical tests utilised to verify normal distribution of data.

Variables	p-Value
	Shapiro–Wilk	Anderson–Darling	Lilliefors
CRa	0.547	0.583	0.529
CTh	0.088	0.289	0.186
CK	0.224	0.455	0.386

, which reports the p-values provided by the various algorithms used to verify the distribution of the data.

The C_K data were log₁₀₀-transformed prior to testing to improve normality. The Shapiro–Wilk test indicated that C_Ra and C_Th data were normally distributed (p > 0.05), while C_K data were not normally distributed (p < 0.05). After log-transformation, the C_K data also exhibited a normal distribution (p > 0.05). The Anderson–Darling and Lilliefors tests yielded consistent results, supporting the normality of C_Ra and C_Th data. These tests also confirmed that the log-transformed C_K data followed a normal distribution. Therefore, parametric statistical methods are appropriate for further analysis.

Pearson’s correlation analysis was then employed to assess the interdependency between the specific activities of the detected radionuclides and the radiological parameters, as well as to ascertain any existing correlations between radiological hazard indices and radioisotope activity concentrations. The findings of this investigation are provided in

Table 7. Pearson’s correlation matrix.

Variables	CRa	CTh	CK	D	AEDEout	Hex	ELCR
CRa	1	0.789	0.475	0.821	0.820	0.847	0.847
CTh	0.789	1	0.571	0.907	0.906	0.894	0.907
CK	0.475	0.571	1	0.839	0.841	0.806	0.807
D	0.821	0.907	0.839	1	0.999	0.986	0.991
AEDEout	0.820	0.906	0.841	0.999	1	0.984	0.991
Hex	0.847	0.894	0.806	0.986	0.984	1	0.987
ELCR	0.847	0.907	0.807	0.991	0.991	0.987	1

.

The analysis reveals predominantly strong positive intercorrelations among the measured variables. Notably, C_Ra demonstrates a high degree of correlation with C_Th (r = 0.789), D (r = 0.821), AEDE_out (r = 0.820), H_ex (r = 0.847), and ELCR (r = 0.847). C_Th also exhibits strong positive correlations with D (r = 0.907), AEDE_out (r = 0.906), H_ex (r = 0.894), and ELCR (r = 0.907). However, it is important to note that while C_K shows positive correlations with D (r = 0.839), AEDE_out (r = 0.841), H_ex (r = 0.806), and ELCR (r = 0.807), these are notably weaker compared to those observed for C_Ra and C_Th. Furthermore, C_K exhibits a moderately positive correlation with C_Ra (r = 0.475) and C_Th shows a similar moderately positive correlation with C_K (r = 0.571). The lower correlation of C_K with C_Ra and C_Th may suggest that C_K is influenced by different geological or geochemical processes compared to these other radionuclides. As expected, a very strong positive correlation is observed between D and AEDE_out (r = 0.999), as well as between AEDE_out and both H_ex (r = 0.984) and ELCR (r = 0.991). H_ex and ELCR are also highly correlated (r = 0.987). These high correlations among D, AEDE_out, H_ex, and ELCR are expected due to their direct or indirect dependence on the radionuclide activity concentrations.

Principal component analysis (PCA) was subsequently performed on the seven variables (C_Ra, C_Th, C_K, D, H_ex, AEDE_out, and ELCR) to identify the principal components explaining the variance in the dataset. The extracted factors are presented in

Table 8. Eigenvalues, variability, and percentage of total variance explained by the first five principal components (PC1–PC5).

	PC1	PC2	PC3	PC4	PC5
Eigenvalues	6.18	0.58	0.20	0.02	0.009
Variability	88.38	8.29	2.91	0.29	0.14
% Total Variance Explained	88.38	96.67	99.57	99.86	100

.

To assess the suitability of the dataset for principal component analysis (PCA), the Kaiser–Meyer–Olkin (KMO) test and Bartlett’s test of sphericity were conducted. The KMO value of 0.64 indicated acceptable sampling adequacy for PCA. Bartlett’s test of sphericity yielded a highly significant result (p < 0.001), confirming the presence of significant correlations within the dataset, thus justifying the application of PCA.

Prior to analysis, a log₁₀₀-transformation was applied to address non-normality and heteroscedasticity, a standard approach in environmental studies involving compositional data or variables with wide ranges [48,49].

The obtained PCA biplot is reported in Figure 4. The variable “Site IDs” was also inserted, in an attempt to evaluate the possibility of regrouping for those samples that demonstrate homogeneous behavior in terms of activity concentrations and radiological characteristics.

The biplot clearly separates sampling sites into two groups along PC1, which ex-plains 88.38% of the total variance. This separation suggests that PC1 captures most of the variability in the data. The distinction is mainly driven by the positive loadings of C_Ra, C_Th and ELCR on PC1. C_Ra and C_Th are indicators of sediment sources such as terrigenous input [55] or weathering processes [57]. The positive correlation between C_Ra and C_Th suggests a shared origin or similar mobilization mechanisms. ELCR also correlates positively with C_Ra and C_Th, reinforcing the connection between these isotopes and potential health risks.

PC2, which explains 8.29% of the variance, further differentiates samples within the groups identified by PC1. This component is positively loaded by C_K, AEDE_out, and H_ex.

A notable pattern in the PCA biplot is the distribution of samples relative to sampling depth. In detail, samples from site IDs 3, 6, 8, 9, and 10, collected from greater depths, tend to cluster on the left side of the plot. This suggests that depth may be a key factor influencing sediment composition, likely reflecting different environmental conditions or geochemical processes at varying depths.

Several depth-related factors could contribute to this differentiation. It is well-established in sedimentology that finer sediments (silts and clays) accumulate in deeper waters [58], which can affect the adsorption of metals and contaminants, as well as the distribution of radioactive isotopes. The quantity and composition of organic matter also change with depth [59], influencing the bioavailability and distribution of contaminants. Additionally, diagenetic processes [60] can alter the composition of deeper sediments, while changes in hydrodynamic conditions [61] can influence sediment deposition and transport. Deeper waters, with lower energy, tend to favor the accumulation of fine sediments and associated contaminants. Depth may also correlate with distinct sources of anthropogenic pollution, as industrial or urban discharges may vary depending on depth or location.

The distribution within the biplot indicates environmental heterogeneity in the Gulf of Termini Imerese. Future research, incorporating additional environmental data (e.g., bathymetry, grain size, total organic carbon content), will provide further insights into the complex factors that govern the distribution of contaminants in marine sediments.

The dendrogram generated from the HCA reveals distinct groupings, thus offering valuable insights into the compositional heterogeneity of investigated samples [26] (Figure 5).

The tight clustering of samples from site IDs 1 and 2, both collected at a depth of 121 m, strongly suggests a marked similarity in their composition. This cohesion may be attributed to a shared origin, such as a common source of sediment input, or comparable environmental conditions prevailing at their respective sampling locations. However, the dispersion of samples collected at identical depths (samples from site IDs 3, 5, and 6, at 128 m), across various clusters, underscores the influence of factors beyond mere bathymetry in shaping sediment composition. This observation potentially reflects variations in proximity to specific pollutant sources, diverse hydrodynamic regimes, or localized geological features (e.g., variable lithology and morphology of the seabed).

Furthermore, the observed clustering patterns may be influenced by factors such as grain size distribution, mineralogy, and organic matter content within the sediment matrix. HCA results provide a robust foundation for subsequent investigations into the intricate interplay of factors governing sediment composition in the Gulf of Termini Imerese. Future research endeavors exploring potential correlations between the HCA-derived clusters and environmental parameters, such as heavy metal concentrations or organic carbon content, could offer further insights into the ecological implications of sediment composition in this coastal marine environment.

It is noteworthy that the integration of HCA results with PCA offers a more comprehensive perspective on the variability of investigated marine sediments. The PCA, shown in the biplot, reveals a clear separation along PC1. This separation partially reflects the groupings observed in the HCA. The concordance between HCA and PCA results highlights the robustness of the identified groupings and allows us to establish that both methods capture relevant aspects of sediment composition. The PCA further elucidates the drivers of these groupings, indicating that variables such as C_K, AEDE_out, H_ex, and ELCR, strongly correlated with PC1, contribute to the separation of samples.

Interestingly, samples such as those from site IDs 3, 5, and 6 appear more dispersed along PC2 in the PCA biplot, suggesting greater variability in their chemical composition. This observation is consistent with the HCA results, where these samples do not form a tightly grouped cluster, reinforcing their distinct characteristics.

The concordance between HCA and PCA suggests that each method captures different dimensions of variability within investigated samples. Specifically, HCA groups samples based on shared traits, whereas PCA identifies the key variables driving these similarities and differences. By integrating HCA and PCA, a more comprehensive understanding of the factors influencing sediment composition emerges, allowing for more robust and nuanced conclusions about their complex interactions.

4. Summary and Future Perspectives

The natural and anthropogenic radioactivity of marine sediments from ten different selected sites of the Sicilian region, Southern Italy, i.e., in the Gulf of Termini Imerese, was investigated for the first time through High-Purity Germanium (HPGe) gamma-ray spectrometry, thus filling the knowledge gap on the subject. Furthermore, D, AEDE_out, H_ex, and ELCR, were also calculated to evaluate the radiological hazard for humans related to external exposure to ionizing radiations. Moreover, the average specific activity of ¹³⁷Cs was determined to be less than the lowest detectable activity in all cases, ruling out a significant anthropogenic radioactive contamination of the analyzed samples.

In addition, Pearson correlation, principal component, and hierarchical cluster multivariate statistical analyses were carried out to establish the correlation of the observed radioactivity and radiological parameters with the sampling locations. Overall, this integrated statistical approach provided valuable insights into the spatial distribution and interrelationships of radiological parameters in the studied marine environment, emphasizing the need for future research incorporating additional environmental data to fully elucidate the complex factors governing sediment composition and potential ecological implications.

Finally, as future perspectives, it is noteworthy that: (i) the evaluation of radiation hazard of alpha and beta emitters in case of incorporation, by using beta and alpha spectroscopy, has been scheduled for development; (ii) the application of cost-effective and rapid response in situ methods and sensors in this marine environment may also support national authorities to enrich the existing database for radioactivity data into the sediments; and (iii) the sampling, treatment, and analysis of the investigated marine sediments are in full accordance to what is reported in the manual of the Italian Environmental Radioactivity Surveillance Network, both for routine and emergency cases. It is important to point out that one of the main objectives of this network is to follow the spatio-temporal trend of the activity concentrations of natural and anthropogenic radionuclides in various environmental matrices across the entire national territory, in order to assess the state of the environment, as well as to promptly identify anomalies arising from events, also occurring outside the national territory, that involve contamination, acting as a means of alerting and as a tool for timely and post-incident assessment in support of decision-making. Therefore, it is fully reasonable to believe that the proposed method is capable of sustaining national authorities in an emergency situation, in order to provide support on the measures to be taken in the case of a contaminated area.