Introduction

Studies of the distribution of radioactivity in soils have been performed worldwide to obtain radiological maps (of radioactivity levels), which provide useful information for mineral exploration and geological and geophysical investigations and can be used to assess the exposure of humans to radiation, to perform epidemiological studies and to build a “reference” database that can be useful for evaluating changes in environmental radioactivity due to nuclear, industrial or other human activities [1–14].

Natural environmental radioactivity levels are related to the presence of the primordial radioactive isotopes 238U and 232Th, their decay products, and 40K. These radionuclides are present at different concentrations in all components of the environments. However, their levels in the soil depend on the geological and geographical conditions and vary in different parts of the world [15]. Although soil radionuclide contents are directly related to the radionuclide contents in the parent rock, the processes of soil formation may promote changes in radionuclide contents due to their addition, loss, transformation and translocation, which depend on additional factors such as topography, climate, organisms and time. Natural radioactive isotopes are responsible for approximately 86 % of human exposure to radiation [15].

Besides natural radioactive isotopes, which have been present since the beginning of the earth, atomic weapon tests and the Chernobyl and Fukushima accidents introduced man-made radionuclides to the atmosphere; these radionuclides were spread throughout the environment and have been incorporated in all components of the environment [16, 17]. Because few atomic tests were carried out in the southern hemisphere, the highest concentrations of 137Cs in the soil were mainly found at middle and high latitudes in the northern hemisphere and in central Australia [17–20]. According to the estimated global distribution of 137Cs in 1996 that originated from nuclear testing, the fallout in the equatorial region was minimal [21]. Therefore, many studies have been conducted to establish the 137Cs baseline in soils in Europe, America, China, and Australia [16–20]. However, limited amounts of data from systematic surveys of this radionuclide in the equatorial region are available in the literature [22].

Some surveys have described the abiotic characteristics of the Fernando de Noronha archipelago (FNA), raising important data for the preservation of the ecosystem [23–28]. However, no studies concerning the occurrence of natural or man-made radioactive isotopes on these islands have been performed. Therefore, this study contributes to the knowledge of the island environments and establishes a “baseline” of the natural radioactive isotope and 137Cs contents in the soils of this low-latitude volcanic archipelago in the southern hemisphere. Overall studies of these small oceanic islands can be used to gather relevant information for improving our understanding of the nature of the oceanic crust and the processes related to its spreading.

Materials and methods

Study site

The Fernando de Noronha archipelago is located in the Southern Equatorial Atlantic around 400 km from the northeast coast of Brazil (at 3º51′S and 32º25′W). The archipelago consists of one main island encircled by 20 islands and islets [24]. The total area of the archipelago is approximately 20 km2. The main island, which the archipelago is named after, comprises 16.4 km2 and is the only inhabited island.

The climate is tropical (Awi type according to the Köppen classification system), marked by the oceanic domain, with annual average temperature of approximately 25 °C, and the annual rainfall is approximately 1400 mm. There are two distinct seasons, a wet season from March to July and a dry season with a water deficit from August to January [25].

The archipelago has approximately 2700 inhabitants, and 100 % of the population resides in the urban area. Tourism is the main economic activity (in 2014 were about 90,000 visitors), followed by the service sector [26]. The inhabitants practice subsistence farming and artisanal fisheries.

The archipelago is the top of an extinct volcano, with a wide variety of rocks of igneous origin, ranging from ultrabasic to intermediate. Almeida [23] reports that an area of only 1.5 km2 of the archipelago contains 14 different types of volcanic rocks. Geologically, the 21 islands and islets that form the Archipelago represent the top of a mountain chain that developed along an east–west fracture zone of the ocean floor and was built up by volcanic and subvolcanic alkaline and subsaturated rocks. These rocks are products of two distinct volcanic episodes. The first episode, which is represented by the Remédios Formation of the Late Miocene, is composed of pyroclastic deposits at the base and cut by intrusions in the form of necks, plugs, domes and dikes of subsaturated alkaline rocks. The composition ranges from intrusive basic-ultrabasic (lamprophyres, tephrites, basanites and alkali basalts) to intermediate (trachytes and phonolites) rocks. The second episode, represented by the Quixaba Formation of the Late Pliocene to the Early Pleistocene, is composed by a stack of melanocratic ankaratrite lava flows, subordinate pyroclastic deposits and some nefelinite dikes. A small amount of basanite flow with a controversial age is represented by the São José Formation [26].

The relief of the main island is made up of beaches, dunes, plains, slopes, low plateaus, hills, shore cliffs and a high plateau, which is where the small town of Vila dos Remédios lies and where most of the inhabitants of the island live.

The volcanism represented by the exposed rocks in the archipelago is characterized by its strongly sodium-alkaline nature, subsaturation of silica (there is no quartz in the rocks, although it is the most abundant mineral of the Earth’s crust) and a high degree of differentiation. The silica contents of the rocks vary between 34.4 (melilite ankaratrite) and 60.8 % (alkali-trachyte). Among the oceanic volcanic islands of the world, the magmatic province of Fernando de Noronha is one of the most alkaline [24]. The lighter fractions of the magmatic differentiates were produced earlier, originating from the Remédios Formation, which corresponds to the oldest rocks of the archipelago that are dated between 8 and 12 Ma. After a period of time that may have lasted 5 to 6 Ma, ultrabasic volcanism of the Quixaba Formation was initiated, with an age of between 1.7 and 3 Ma [27].

The strongly melanocratic rocks are the youngest rocks of the Remédios Formation and contain little or no feldspar. Special emphasis is given to the xenoliths and ejectolites of the Remédios Formation because they may represent very deep source rocks of the volcanic edifice.

The Quixaba Formation is a sequence of black flows of alternating ankaratrite lava and pyroclasts containing components of the same lava [28]. This formation rests on two step-like plateaus of the main island. The eastern one may reach an exposed thickness of 180 m above sea level. This formation also occurs on some islands, particularly on the Rata Island and the small São José, Cuscuz and de Fora Islands.

Another formation called the Calcarenite of Caracas consists of light cream coloured sediment, also called Modern Sediment, and was built by fluctuations of sea level that covered the island platforms with deposits of sand, gravel, reef coralline algae and marine sands. During the Pleistocene, when the sea level was approximately six metres below the current sea level, extensive sandy beaches were built to the south and southeast of the archipelago. These sands were moved by wind and formed dunes that likely reached a thickness of approximately 20 m. With the increase in sea level that followed, some of these sands were submerged and currently form the Calcarenite of Caracas Formation. These sediments are light cream in colour and are almost entirely composed of calcium carbonate grains of biogenic origin (quartz is virtually nonexistent in the islands). On Rato Island, phosphate deposits have been partially replaced by calcium phosphate that originated from seabird guano. In addition, the soils on Rata Island are rich in phosphates due to the interactions between the weathering of volcanic rocks and guano [28].

Sample collection, preparation and measurement

This survey is part of a cooperative agreement between the Geological Survey of Brazil (CPRM) and the Instituto de Radioproteção e Dosimetria (IRD/CNEN). The CPRM collects soil samples throughout all Brazil territories as part of its Low Density Geochemistry Project, and some of the collected samples have been sent to the IRD for radionuclide analysis.

Soil samples were collected at a depth of approximately 20 cm from residual soils that had not been displaced or used for agriculture by using a manual stainless steel auger. The sampling grid was approximately 500 × 500 m. Seventy samples were collected to achieve a sample density of 3.7 samples.km−2. A Garmin Global Positioning System Receiver (GPS 12 XL) was used to record the latitude and longitude of each sample location. Figure 1 shows a map of the area and the locations of the sampling points.

Fig. 1
figure 1

Geologic map (adapted from Almeida [33]) and locations of the sampling points

Full size image

Afterwards, 1–3 kg of the samples were stored in polythene bags and sent to the IRD. Later, approximately 1 kg subsamples of the soil samples were oven dried at 40 °C in the laboratory in a stove for 48 h or until a constant weight was achieved. Next, the samples were ground and passed through a nylon 2 mm sieve before further analysis. For gamma spectrometry analysis, sediment subsamples of approximately 300 g were transferred to polyethylene tubes, sealed and stored for at least 30 days to reach radioactive equilibrium among 226Ra and its decay products [29].

Then, the samples were analysed for 60,000 s using two high purity germanium detectors (HPGe) and associated electronic devices (amplifier, HV source and MCA Analyser) from Canberra, Inc. (USA) with a relative efficiency of 20 %. Energy and efficiency calibrations were performed using standard sources supplied by the National Laboratory of Metrology of Ionizing Radiation (LNMRI) of the IRD. The activity of 228Ra (232Th) was derived from the 911.6 keV peak of 228Ac. The peaks at 351.9 keV (214Pb) and 609.3 keV (214Bi) were used to determine the activity of 226Ra. 40K activities were calculated from its photo peak of 1460.8 keV, and 137Cs activities were calculated from its photo peak of 661 keV [30]. The activities concentrations and respective uncertainties were determined according to the statistical uncertainties of the peak areas provided by the Genie2000 software. The value of minimum detectable activity (MDA) was 12.1 Bq kg−1 for 40K, 2.3 Bq kg−1 for 226Ra, 2.8 Bq kg−1 for 232Th (228Ra) and 0.3 Bq kg−1 for 137Cs when a counting time of 60,000 s was used. The certificated reference material MAPEP-09-MaS21 and EML Analysis QAP59 were used for quality assurance.

All statistical data processing in this study was carried out using the commercial statistics software package SPSS 19.0 for Windows (IBM, USA) and ProUCL 5.0 [31, 32]. Maps were designed using the QGIS 2.8.2 software (Open Source Geospatial Foundation Project), which is free to download [33].

Results and discussion

Radionuclide concentrations

As usually found in environmental surveys, the dispersion of the radionuclide concentrations resulted in a skewed data distribution in this study (Fig. 2). The goodness-of-fit (GOF) tests for normal, lognormal and gamma distributions showed that the concentrations of 232Th (228Ra) and 226Ra do not fit the three distributions, and the 40K data distribution could be fit as a lognormal distribution (0.05 of significance level) [31]. Thus the statistical analysis should be performed using nonparametric methods [32]. For comparison purposes, the central tendencies of the data, including the average (arithmetic mean) and median, are reported in Table 1 that also shows the observed range of values of the radionuclide concentrations and the coefficients of variation obtained from the datasets. The uncertainties of the measurement values were around 4 % for 226Ra, 3 % to 228Ra and 9 % for 40K.

Fig. 2
figure 2

Distribution of natural radionuclides in the FNA soil samples

Full size image
Table 1 Descriptive statistical analysis of the natural radionuclide concentrations (in Bq kg−1) in the soils from Fernando de Noronha
Full size table

The most abundant of these three nuclides is potassium, which is found in alkali feldspars and micas. A significant correlation was found between 226Ra and 228Ra, with a correlation coefficient of 0.542, and between 226Ra and 40K, with a correlation coefficient of 0.433 (0.05 significant). No correlation was found between 228Ra and 40K. Figure 3 shows the spatial concentration of 40K obtained from the interpolation of the obtained results.

Fig. 3
figure 3

Distribution of 40K in the soils of the Fernando de Noronha Archipelago

Full size image

For most of the samples (90 %), the 226Ra concentrations are lower than the 228Ra concentrations. The ratio of radium isotopes 228Ra/226Ra varied between 0.16 and 4.37, with average and median values of 1.85 and 1.75, respectively. The high value of the radium isotope ratio resulted from the high stability of thorium during weathering processes. Although thorium remains stable in all minerals, uranium is only stable in zircon and monazite and is very soluble in other minerals containing carbonates, sulfates and phosphates [34]. Thus, during the weathering process, uranium tends to weather out (as the soluble form of uranyl) as metamorphism begins while thorium remains in the rock, resulting in a higher thorium concentration in the igneous rocks than in the sedimentary rocks. However, this trend only changed for samples collected from the sandstones and phosphates of Rata Island, where the concentration of 226Ra activity was higher than that of 228Ra. This outcome can be attributed to the greater affinity of 238U than 232Th to the biological phosphate (guano) present on this island [35].

The highest concentrations of 226Ra and 228Ra were found in some small areas, as shown in Figs. 4 and 5.

Fig. 4
figure 4

Spatial distribution of 226Ra in the soils of the FNA

Full size image
Fig. 5
figure 5

Spatial distribution of 228Ra in the FNA

Full size image

For comparison, the reported average radionuclide concentrations in rocks from different countries are shown in Table 2.

Table 2 Worldwide averages of soil radionuclide concentrations
Full size table

The average concentrations of 232Th (228Ra) and 226Ra in this survey are higher than the values reported world wide, while the average concentration of 40K is lower than average. Regarding natural radioactivity, igneous rocks usually exhibit higher contents of Th and U than sedimentary rocks, except for some shale and phosphate rocks [16]. Thus the relatively high concentrations of 226Ra and Th232 (Ra228) in the soil of the FNA are related to the igneous origins of the archipelago. On the other hand, the low 40K concentration of the soil in the FNA is related to the low feldspar content of most of the rocks on the island.

Considering the fractionation processes that occur as a result of radioactive decay and the distinct chemical characteristics of the decay products, radioactive equilibrium among the radionuclides of uranium and thorium decay chains cannot be expected to exist in environmental samples. Due to its greater environmental mobility in comparison with 238U and 230Th, 226Ra may be present at slightly different concentrations than 238U in soil samples [2, 16]. Although similar considerations can be made for 232Th and 228Ra concentrations, the 228Ra half-life (5.75 years) may allow a quite similar concentration values for these radionuclides in soils. Despite of the above considerations, an approach of 238U and 232Th concentrations can be estimated considering the equilibrium in the chains. The concentrations of 238U (226Ra) and 232Th (228Ra) can be converted to the elemental concentrations (µg g−1) of the radionuclides by using 238U and considering that 1 µg g−1 of 238U is equivalent to 12.35 Bq of 238U kg−1. For 232Th, 1 µg g−1 of 232Th was considered to correspond to 4.06 Bq of 232Th kg−1. The percentage of K (%) was obtained by dividing the 40K concentration by 2.76, and the concentration of K was obtained by dividing the 40K concentration by 2.76 10−2.

The elemental concentrations ranged from 0.38 to 44.5 µg g−1, with an average concentration of 3.00 and a median concentration of 5.02 µg g−1 for 238U, from 2.59 to 73.3 µg g−1, with an average concentration of 20.2 µg g−1 and a median concentration of 16.5 µg g−1 for 232Th, and from 0.05 to 4.64 %, with an average concentration of 0.65 % and a median concentration of 0.32 % for K. The worldwide median values reported by UNSCEAR (2000) are 35, 30 and 400 Bq kg−1 for 238U, 232Th and 40K respectively, which correspond to 2.8 mg kg−1, 7.4 mg kg−1 and 1.4 %. Thus, the U and Th contents in the FNA soil are higher, while the K content is smaller than the median value for these radionuclides worldwide. The mean value of the Th/U ratio (5.6 ± 2.6) is higher than the theoretically expected Th/U ratio of 2.8 for the Earth’s crust. The average K/U ratio found for the FNA (1.7 103, with a median value of 9.7 102) is lower than the estimated terrestrial and ocean basalt values of approximately 1.0 104 and lower than the lower value of the range for the Earth (from 0.5 104 to 1.4 104) [36]. The low K/U ratio observed for FNA is consistent with the depletion of K observed in the archipelago. This value is similar to the determined value of 2.8 103 for the superficial soil of Cyprus [37] The K/Th values range from 1.9 101 to 3.2 103, with an average of 3.9 102 and a median of 8.5 102.

From the seventy investigated samples, 137Cs was detected in only sixteen samples (23 % of the analysed samples). The statistical code ProUCL 5.0 can be used to perform statistical analysis of a data population with non-detected values. Because the number of non-detected data points is not lower than ten [31], 137Cs descriptive analysis was performed. The 137Cs was calculated using the Kaplan–Meier method, which resulted in a mean of 0.62 Bq kg−1 and a range of <0.3–2.4 Bq kg−1. The 50 % percentile and 99 % percentile are 1.1 and 2.4 Bq kg−1, respectively. Considering only the sixteen detected values, the values range from 0.4 to 2.4 Bq kg−1, with a mean value of 1.04 and a median value of 0.97 Bq kg−1. These data are comparable with the data obtained by LaBrecque and Cordoves [22] for topsoil in Venezuela, where only 28 % of the 90 analysed samples had 137Cs values higher than 2 Bq kg−1. The values of the detected samples varied between less than 2 and 33.4 Bq kg−1. The low 137Cs content in the FNA soil confirms the predictions reported by UNSCEAR [21] regarding the low level of fallout radionuclides that reached the equatorial region.

Relationships with the main geological formations

The maximum and minimum values and the average and median concentrations of radionuclides obtained in this study for the main formations of FNA, Quixaba (Q), Remédios (R) and Modern Sediments (MS) are summarized in Table 3.

Table 3 Average, median and maximum and minimum 226Ra, 228Ra and 40K concentrations in Bq kg−1 according to geological formation: Quixaba (Q), Remédios (R) and Modern Sediments (MS)
Full size table

Most of samples were collected from the Quixaba Formation (52 samples, corresponding to 74 % of all the samples), 14 samples were collected from the Remédio Formation, only 4 samples were collected from the Modern Sediments Formation and no samples were collected from the San José Formation.

ANOVA using Fisher’s test showed that the mean of the data from the Quixaba Formation (Q) for 226Ra is lower than the mean of the data from the Remédios and Modern Sediments Formations, which had similar means (with 95 % of significant). The 228Ra and 40K concentrations are both higher for the Remédios Formation than for the Quixaba and Modern Sediments Formations, which are similar. The box-and-whisker graphs in Fig. 6 show the distributions of the concentration data obtained from the different formations.

Fig. 6
figure 6

Distributions of 226Ra, 228Ra and 40K data in the main geological formations of the FNA (R Remédios formation, Q Quixaba formation and MS Modern sediments formation)

Full size image

Thus, except for the similar data of Remédios and Modern Sediments Formations for 226Ra, the data set from the Remédios Formation, which is the oldest formation in the archipelago, had the highest concentrations of natural radionuclides.

The greater affinity of 238U than 232Th for biological phosphate (guano) in the Modern Sediment Formation was underlined by the nonparametric Wilcoxon–Mann–Whitney (WMW) test, which showed a significant difference (95 % significance) among the data of the 228Ra/226Ra ratios of the Modern Sediment Formation (data ranging from 0.24 to 2.38, average of 0.95 and median of 0.58) and the 228Ra/226Ra ratios from Quixaba (data ranging from 0.16 to 4.36, average of 1.99 and median of 1.79) and Remédios (data varying between 0.26 and 2.57, average and median values of 1.60 and 1.71, respectively). No significant difference was observed between the radium isotope ratios of the Quixaba and Remédios Formations. Considering the elemental concentrations, the WMW test indicated a significant difference between the U/Th ratios of the sediment from the Modern Sediment Formation and the sediments from the Quixaba and Remédios Formations, which are similar. However, the WMW test showed no significant differences among the K/U ratios and among the K/Th ratios for the three geological formations.

Absorbed gamma dose rate and corresponding annual effective dose

The gamma dose rate can be estimated from the concentrations of soil 226Ra, 232Th (228Ra), and 40K (Eq. 1) [21]. Using this approach, the infinite source geometry, homogeneous distribution of radionuclides and radioactive equilibrium in the 238U and 232Th decay series should be considered. For this estimation, the concentrations of natural radionuclides are multiplied by the corresponding values of the kerma rate conversion factors (K) in nGy h−1 per Bq kg−1 (Eq. 1).

$${\text{D}}_{\text{nat}} \left( {\eta {\text{G}} . {\text{h}}^{ - 1} } \right) = 0.0417 \times {\text{C}}_{\text{K40}} + 0.463 \times {\text{C}}_{\text{Ra226}} + 0.604 \times {\text{C}}_{\text{Ra228}}$$
(1)

where D is the rate of the dose in the air (nGy h−1), the Ks values were obtained by Saito and Jacob [39] and correspond to 0.463 for the 238U series, 0.604 for the 232Th series and 0.0417 for 40K, respectively, and C is the radionuclide concentration in Bq kg−1.

The estimated gamma rate dose varied between 27 and 369 nGy h−1, with mean and median values of 86 and 72 nGy h−1, respectively. The range of values estimated for FNA surpass the range of variability of the measured absorbed dose rates in the air cited by UNSCEAR [16], which range from 10 to 200 nGy h−1, and the median values are slightly higher than the worldwide median of 51 nGy h−1.

The annual effective dose due to outdoor external exposure Em (mSv a−1) from the soil can be calculated using the following equation:

$${\text{E}}_{m} = {\text{D}}_{\text{nat}} \times {\text{O}}_{f} \times 0.7\,{\text{Sv}}\,{\text{Gy}}^{ - 1} \times 10^{ - 6}$$
(2)

where O f is the fraction of the year for which a hypothetical member of the public is exposed to the outdoors. The suggested O f value is 20 % (0.2), which means that each individual spends an average of 1752 h outdoors each year. The factor 0.7 Sv Gy−1 is the conversion factor from the absorbed dose in the air to the effective dose received by adults at a height of 1 m above the ground surface [15].

The calculated median value of the annual effective external dose outdoors was 0.08 mSv a−1, with an average value of 0.10 mSv a−1 and a range of 0.03–0.45 mSv a−1.

If no flooring cover would be considered, and considering that individuals spend 80 % of their time indoors, the external dose due to soil exposure indoors varies between 0.17 and 2.26 mSv a−1. Thus, the average and the median total contributions of natural radionuclides in the soil to the total external effective dose are 0.52 and 0.31 mSv a−1, respectively, which are similar to the worldwide values of 0.48 and 0.45 mSv a−1 established be UNSCEAR [15]. However, one location with Aphyric phonolites in the south and middle area of the main island would present a total annual external dose of between 1.1 and 2.3 mSv, which surpasses the worldwide average external dose of 0.3–0.6 mSv (Fig. 7).

Fig. 7
figure 7

Dose resulting from the natural radionuclides in the soils from the FNA

Full size image

Conclusions

This research generated a database on naturally occurring radionuclides and 137Cs levels in the soils of the Fernando de Noronha archipelago, whose soils were analysed using gamma spectrometry. The median concentrations of 226Ra, 232Th and 40K in the soil samples from FNA were 43, 70 and 108 Bq kg−1, respectively. The contents of 226Ra and 232Th in the soil samples were higher than the world median concentrations of 30 for 226Ra and 35 for 232Th Bq kg−1. However, due its rock characteristics, the levels of 40K in the soils from the FNA are four times lower than the worldwide 40K estimate of 400 Bq kg−1. The distribution of radionuclides in the soils of the archipelago is very heterogeneous, which could be attributed to geological characteristic of the archipelago. The soils from the Remédios Formation, which is the oldest formation of the archipelago, have the highest concentrations of natural radionuclides. No significant differences were observed between the U/Th ratios of the Quixaba and Remédios Formations. However, this trend changed for samples collected from the soils overlying sandstones and phosphates on Rata Island (Modern Sediment), where the concentrations of 226Ra were higher than the concentrations of 228Ra due to the greater affinity of 238U than 232Th for sorption to the biological phosphate (guano) present on this island. In this study, the average concentrations of 3.00 µg g−1 for 238U, 20.2 µg g−1 for 232Th and 0.65 % for K were derived. The measured elemental ratios exhibit a wide range of values. Specifically, the Th/U ratios varied between 0.5 and 13, with an average value of 5.6, which was higher than the expected value when considering the average value for the Earth’s crust of approximately 2.8. This result potentially indicates that the fractionation of radioactive isotopes during weathering occurred. The K/Th ratio ranged from 1.9 101 to 3.2 103, with an average value of 3.9 102, and K/U ratio varied between 6.9 101 and 1.2 104, with an average value of 1.7 103, which is similar to the ratio observed in the soils of Cyprus Island. The K/U and K/Th ratios were similar for the three geological formations. Although 137Cs was detected in only sixteen samples, the range of values from less than 0.3 to 2.4 Bq kg−1 can be used as a reference for assessing the contamination of FNA soils. The low soil 137Cs content confirms the prediction of low radionuclide fallout in equatorial regions. A gamma dose rate of 72 nGy h−1 was estimated, which slightly exceeded the world median rate of 51 nGy h−1. The main contributor for the dose rate is the Th decay series, which accounts for 58 %, followed by the U-series and 40K, which account for 33 and 9 %, respectively. The level of external exposure to naturally occurring radionuclides in soils indicates that the population in most areas of the FNA are exposed to normal radiation levels. However, a small area containing Aphyric phonolites has higher levels of external exposure to radiation than the worldwide range.