Concentration of fallout Cesium-137 in soils of the Rio de Janeiro State , Brazil .

The distribution of fallout Cs in soil samples of the Rio de Janeiro State was investigated. Two-hundred sixty-two samples of soil were analyzed for Cs by gamma spectrometry. A reference data set was built containing thirty-six detected values (14%) and two hundred twenty-six (86%) multiple detection limits of non-detected observations. Distribution pattern of the radionuclide, as well their controlling factors were evaluated based on statistical methods for censored data set (with non-detected observations). The mean of concentration values of thirty-six detected observations was 1.25 Bq.kg, while for the censored data sets the mean value was 0.51 Bq.kg. The references values for the Rio de Janeiro State were estimated as 1.44 Bq.kg (75) and 1.78 Bq.kg1 (90) (considering the year of 2015). Cs concentrations tend to be higher in soils with high levels of organic matter and clay and locations with high levels of these parameters would provide higher Cs fixation. No influence of the rainfall index and soil pH on the Cs content in soil was observed. The data and information gathered in this study provide valuable information for decision making purposes in case of soil contamination.


INTRODUCTION
The testing of nuclear weapons introduced in the environment man-made radionuclides, which were widely scattered in the atmosphere, and deposited on the entire surface of Earth.
Although in a regional scale, the nuclear accidents, as Chernobyl and Fukushima, also spread man-made radionuclides in atmosphere and on the ground.Nuclear power plants also routinely release man-made radionuclides to the environment, nonetheless in a controlled way.Among the man-made released radionuclides, 137 Cs is one of the most important [1].
Due to its chemical characteristics, 137 Cs is strongly adsorbed on soil particles and, in consequence, its mobility in soil depends largely on physical processes [2].Nevertheless, 137 Cs mobility can be affected by some characteristics of the soil.The element is nearly immobile in soils with high clay or organic matter content, thus making it difficult to move to deeper layers.[3,4,5].Because of its low mobility, some studies have showed that the distribution in undisturbed soil decreases exponentially with depth [6,7,8].However, in agricultural soils 137 Cs tend to exhibit a uniform concentration in the arable layer due to ploughing [9,10].As a consequence of its limited migration, 137 Cs is used on studies to quantify the erosion of soil [3,11].
The distribution of 137 Cs spread on the Earth crust is not homogeneous.As few atomic tests were carried out in the southern hemisphere, the highest concentrations of 137 Cs in soil are found mainly in the middle and high latitudes of the northern hemisphere and in the central area of Australia [3,12,13].According to UNSCEAR [14], around 25 per cent of the total 137 Cs released by the testing of nuclear weapons and other atmospheric emissions were deposited in the southern hemisphere, while the prediction of the distribution of global inventory for 137 Cs for the equatorial region is to reach the minimal values.
Although many studies have been conducted to establish the 137 Cs baseline in the northern hemisphere, the southern hemisphere studies are still scarce [15].In Chile, the radionuclide deposition density was higher than the predicted to the hemisphere and similar to that found in Europe, ranging from 450 Bq.m -2 to 5410 Bq.m -2 [16,17,18].No dependence between concentration values and geographic location was found, but of the average annual rainfall.Labrecque et al. [19] found concentrations of 137 Cs in Venezuelan soils varying between 16.3 and 30.8Bq.kg -1 , which are considered anomalous (high) for the region that is located near the equator.These high concentration values were attributed to the high annual rainfall and to the organic soils of the area.
Levels of background and baseline on distribution of radioactivity in soils are performed worldwide with the aim of providing useful information to assess human exposure to radiation, epidemiological studies and to building a "reference" database that can be useful to evaluate changes of environmental radioactivity caused by nuclear, industrial or other human activities [20][21][22][23][24][25].In this context, the knowledge about the relationship between levels of 137 Cs in soils and environmental variables may provide valuable information about the region vulnerability to contamination, which are good sources of information for decision making purposes in case of soil contamination.
As results of these studies, undisturbed soils showed concentration values of 137 Cs ranging between 0.3 and 5.2 Bq.kg -1 .A comparison among soils from different latitudes showed that the 137 Cs deposition increases with the distance from the equator: as farther from the equator, as higher is the 137 Cs deposition, which endorsed the information provided by SCHULLER et al.
The knowledge of the "current concentration" of 137 Cs in soil is important, because it's concentration in the environment depends only on human activities as atomic weapon tests, power plant releases, nuclear accidents and so on.The reference values can be established at the same methodology as the one preconized by the Brazilian National Council of Environment (CONAMA), in which the samples are collected in the undisturbed soil in a depth of 0-20 cm [31].This way, a baseline can be established to serve as a reference to assess risks of potential contamination and the influence of anthropic activities in the environment.In the absence of a baseline, and if international values or the ones defined in other regions are used, inadequate evaluations can be generated, since there are marked differences in geological, water and soil characteristics among regions [32].
The Brazilian state of Rio de Janeiro is located in the Southeastern region and comprises different types of soils, geology features and climatic conditions.The State has the only Brazilian nuclear power plants, two in operation and one under construction.Surveys for assessing the baseline data of 137 Cs in a systematic way in soil were not reported for Rio de Janeiro State.Therefore, this study contributes to increase the knowledge of the state environment and to establish a "baseline" of the content of 137 Cs in the soil of the State.With the aim of investigating trends of 137 Cs accumulation, the relationships between 137 Cs and the main characteristics of environment (rainfall) and soils (pH, organic matter and granulometry) that are reported as influencing 137 Cs accumulation on soil, were studied.

Study Area
Rio de Janeiro is a state located in the Southeastern region of Brazil, with about 43,000 km 2 , and is the second Brazilian state regarding population and industry [33,34].Its geology consists of sequences of igneous and metamorphic rocks as granite, gneiss and migmatites, forming mountains chain and sedimentary areas mainly in the coastal plains of fluvial or marine origin.In small areas spread by the state it occurs alkaline rocks with uranium and thorium associated to the minerals.The state has a variety of soils, influenced by the parent material and diversification of climate types.The rainfall is seriously influenced by the geographical localization and by the orographic factors.At the mountain chains of Serra dos Órgãos and Serra do Mar, the precipitation is higher than in the Metropolitan, North and Medium Valley of Paraíba do Sul River regions [35].The climate types are classified as Aw, Am, Af, BSh, Cfa, Cfb, Cwb e Cwa, according to Köppen System [36].

Sample collection and laboratory analyses
Following the methodology established by CONAMA [31], this study analyzed a total of 262 samples of top soil (0-20 cm) that were collected by Lima [37] in undisturbed areas to stablish quality reference values of toxic metals.The sampling points were selected on the basis of a joint analysis of maps of soils, geology, land cover and use of the State of Rio de Janeiro, all on scale 1:500,000, covering a wide state area.The superimposition of maps was carried out using the program ArcGIS Desktop 10, produced by ESRI, and to obtain the sampling points it was used the cLHS program -Conditioned Latin Hypercube System.It was also used the road map of the State of Rio de Janeiro, to orient the offset into the sampling areas.The sampling areas were defined to be at least 200 m from the roads to avoid contamination.Still in the field the samples were identified and georeferenced.The geographical location of the soils sampling points can be seen in the figure 1.
The samples were air dried and sieved using 2-mm nylon sieve before further analysis by the staff of Laboratory of Soil Pollution at Universidade Federal Rural do Rio de Janeiro -UFRRJ who classified the soils samples according to Brazilian Classification System [38] and FAO/WRB Soil Classification System [39] and chemical and physical-chemical characteristics.
The granulometry quantification (sand, silte and clay), pH in H2O and organic matter were determined in accordance with the methodology described by DONAGEMMA et al. [40].

Statistical Analyses
Environmental data sets commonly contain analytical values that are lower than limits deemed reliable enough to report as numerical values [41].The Minimum Detectable Concentration (MDC), also denoted by detection limit (DL), is the lowest concentration of a chemical that a particular protocol or methodology is capable of consistently quantify.Due to the fact that levels below the MDC (LD) are usually found in environmental samples, the environmental data sets unsurprisingly contain non-detect (ND) values and may result in leftcensored data.A ND value means that the analyte concentration in the sample is between zero and the MDC value.
Due to diversity of the density of the analyzed material, the chemical recovery of the sample processing, the efficiency of the utilized equipment, among other, the environmental data set typically contain more than one value of MDC.Data sets containing various values of MDCs or multiple detection limits are called multiply censored data sets.

Descriptive statistical analysis and distribution and comparison tests have been specially
developed for multiply censored data set, and have been employed for environmental statistics [42,43].Basically, three data-analysis procedures are available for multiply censored data [42,44]: i) Simple-Substitution Methods: recommendations are to replace the ND with half of the DL (DL/2), DL/10 or by the DL value itself, but this replacement is known to cause a frequency spike at that value [45].Since any single value between zero and the DL is arguably as good as another, simple substitution method, particularly when used on multiply censored data, may introduce a signal that is not present in the data, or obscure a signal that is present.In consequence, this method is considered to be the least precise method [47,48].ii) Parametric Methods that require sufficient data to validate the use of a specific distributional modela requirement that is difficult to meet with small multiply censored datasets and iii) Nonparametric Methods.Maximum likelihood estimation (MLE), regression on order statistics (ROS) and Kaplan-Meier are the methods most often cited in the literature for handling non-detect values.
The regression on order statistics (ROS) is a probability plotting and regression procedure that models censored distributions using a linear regression model of detected concentrations versus their normal quantiles.The regression line is used to fill in values imputed from the straight line for the data below the detection limit.
Nonparametric distributional modeling, such as methods based on Kaplan-Meier statistics, do not require the assumption of a specific distribution to estimate summary statistics for multiply censored datasets.This least procedure is the one adopted by the ProUCL software package [49], a computational code made available to free download by the US Environmental Protection Agency's (USEPA).It has graphical, estimation and hypotheses testing methods for uncensored-full data sets and for left-censored data sets consisting of NDs observations with multiple detection limits (DLs).For data sets with NDs, ProUCL has several estimation methods including the Kaplan-Meier (KM) method and regression on order statistics (ROS) methods.
Graphical methods in ProUCL include histograms, multiple quantile-quantile (Q-Q) plots, and side-by-side box plots.The Oneway Analysis of Variance (ANOVA) module in ProUCL has both classical and nonparametric Kruskal-Wallis (K-W) tests.
From a theoretical point of view, ProUCL can compute various statistics based upon data sets consisting of at least 3 detected observations [49].
The maps of sampling points and of 137 Cs distribution were performed with the use of the free software Qgis v.2.12.2 Lyon [50], which is free to download.

RESULTS AND DISCUSSION 137 Cs concentrations in Rio de Janeiro soils
The data set comprises two hundred and sixty-two samples, in which the values of concentrations of 137 Cs varied between <0.2 and 4.23 Bq.kg -1 .From the data set, thirty-six samples (14%) were detected values and two hundred twenty-six (86%) were multiple detection limits of non-detected observations.Tough the same counting geometry (300 ml) and time were used to analyze all the soil samples, to carry out the analysis of such big number of samples in a shorten time, six gamma detectors were used.Considering that each sample has its specific density and each detector has its efficiency of counting for 137 C, 662 keV, many DL values were observed.The range of the DL values ranged from <0.2 to <2.3 Bq.kg -1 .
According to the recommendations of the technical guide of the ProUCL, prior the statistical analysis, the outliners, observations coming from population(s) other than the main dominant population, should be identified.The Dixon test point out the value of 4.23 Bq.kg -1 as an outliner and in consequence, this value was not considered in the subsequent statistic tests.
Consistent with the Saphiro Wilk GOF test, the thirty six samples with detected values do not follow normal, lognormal or gamma distribution, with 95% of significance.The concentration values for the detected observations varied between 0.32 and 3.42 Bq.kg -1 , with mean value of 1.18 Bq.kg -1 , standard deviation (SD) value of 0.47.The range of detected values found for Rio de Janeiro State is within of the range found for São Paulo State [30]. Figure 2 shows the distribution of the detected values of 137 Cs concentration in soil of Rio de Janeiro.
The mean of the concentrations of the censored datasets (with detected and non-detected values) was estimated by Kaplan Meier Method as 0.51 Bq.kg -1 , the SD of 0.47 Bq.kg -1 and the 75 th percentile, which means that values expected from observation coming from the population (or comparable to the population) will be ≤ X0.75 with probability 0.75, was 1.44 Bq.kg -1 and the 90 th percentile was 1.78 Bq.kg -1 .

State (color figure available online).
Influence of rainfall index, organic matter, clay and pH in 137 Cs soil concentration.
As described before, some environmental parameters have been pointed out by others authors as having influence on 137 Cs concentrations in soils, as the rainfall index, pH, organic matter and clay contents.These parameters were counterpoise to 137 Cs levels in soil.For the comparison among the data sets, only those populations with more than 3 detected data were statistically evaluated.
Rainfall index-The deposition of 137 Cs fallout has been described by other studies as directly related with rainfall.This way, regions with high rainfall index would present higher 137 Cs in their soil [3,[7][8][9][10] The state regions have different rainfall index, varying between 1805 -2202 mm of the Centre South region and 766 -946 mm of the North region, table 1 [35].Figure 3 shows a visual comparison among the data sets of the State´s region.

Figure 3. Box plot of the data set of different regions of Rio de Janeiro State (colored figure available online).
No influence of rainfall index in the mean/median concentrations of 137 Cs in soils was observed, since the analysis of variance ANOVA by Kaplan-Meir test suggested that mean/median characteristics of the various groups are comparable at 0.05 level of significance.Also, the visual comparison with multiple Q-Q plots did not pointed out any significant difference among the data sets (Figure 4).Organic Matter -The content of organic matter (OM) in the Rio de Janeiro soils varied from 5.3 to 69 g.kg -1 .The goodness-of-fit test showed the distribution of data fit a lognormal distribution, with 95% of significance.The median, mean and standard deviation values were 23.2, 24.8 and 9.2 g.kg -1 respectively.For studying the OM influence on the 137 Cs concentration in the soil, the content of OM was grouped in: high OM, values higher than 45 g kg -1 (4.5%); medium values, ranging from, 15 g to 45 g kg -1 (1.5% a 4.5%); and low values, values lower than 15 g kg -1 (1.5%) of OM [48].According with this classification most of the soils had medium content of OM (87%), 10% of the soils had low content of OM and 3% had high levels of OM.The Nonparametric Oneway ANOVA (Kruskal-Wallis Test) suggested that mean/median characteristics of the various groups of OM are comparable.However, the visual comparison with multiple Q-Q plots showed that the upper halves of the "High" and "Low" data sets are quite different (figure 5)."High" group has generally higher values than the "Low" group.On the other hand, the "mean or medium" group showed many values compatible with the "Low" data set, but high values compatible with the "High" data set were also observed.pH -Quantitatively, the degree of acidity of the soil is identified as neutral, with pH around 7; slightly acidic, with a pH between 6.0 and 7.0; moderately acid pH between 5.5 and 6.0; strongly acidic pH between 5.0 and 5.5; very acid, pH 4.5 to 5.0; and extremely acidic pH less than 4.5 [51].According with this degree of acidity, most of the soil samples, 29%, were soil moderately acidic, 26% were strongly and 26% very acidic soils, 11% were slightly acidic, 4% extremely acidic and 3% neutral.The statistical test showed that the means/medians are compatible.Also, no influence of the degree of acidity over 137 Cs level in the soils was observed by visual comparison with Q-Q multiple plots.
Clay -As the content of clay was pointed out as an important factor for immobilizing 137 Cs on soil [3][4][5], the samples were grouped into intervals of clay content according to the following criteria (http://www.pedologiafacil.com.br/textura.php):0-15%, sandy (37 samples); 15-35%, mean (149); 35-60%, clayed (73 samples) 60-100%, very clayed (3 samples).According with the statistical test for data sets with Non-Detects, no statistical differences were found among the content of clay groups.However, the visual comparison with multiple Q-Q plots showed that the upper halves of the "Clayed" and "Mean or Medium" data sets are compatible, but both are moderately different from the sandy group."Clayed and Medium" groups had higher values than the "Sandy" group (Figure 6).

CONCLUSION
137 Cs was analyzed in two-hundred sixty-two samples of soil collected over all state of Rio de Janeiro.The samples were analyzed by gamma spectrometry for 137 Cs determination.Due to the different efficiencies of detection of the six used gamma spectrometry systems and to the different densities of the samples, multiple detection limits were obtained.A multiply censored data set was built containing thirty-six detected values (14%) and one two hundred twenty-six (86%) multiple detection limits of non-detected observations.The specific statistical methods for censored data with multiple detection limits available in the ProUCL software were used to estimate mean/median, standard deviations, make comparison among mean/median of the data sets and build visual comparison plots.
The mean value of concentration values of thirty-six detected observations was 1.25 Bq.kg -1 , with standard deviation (SD) value of 0.78 Bq.kg -1 .The mean of the concentrations of the censored datasets (with detected and non-detected data) was estimated as 0.51 Bq.kg -1 , the SD of 0.47 Bq.kg -1 and the 75 th and 90 Th percentiles as 1.44 Bq.kg -1 and 1.78 Bq.kg -1 , respectively.This way the reference value of 137 Cs in the Rio de Janeiro State is estimate as 1.44 or 1.88 Bq.kg -1 (considering the year of 2015), depending of the percentile considered.
Since the higher concentrations of 137 Cs was observed in the soils with high content of organic matter and medium and high content of clay, a trend of influencing of the two parameters on the fixation of 137 Cs on soils was pointed out.And in consequence, the fixation of 137 Cs would not be favorable for soils with low content of organic matter and clay.Therefore, we suggest that specific studies focused on Cs accumulation on Rio de Janeiro soils are performed to verify this trend and provide useful information for decision-making in case of accidents.No influence was observed of the rainfall index and pH on the 137 Cs content in the Rio de Janeiro soils.
Finally, it is important to note that, despite the high number of undetected values, this study provides valuable information about the occurrence and behavior of 137 Cs in the soil of Rio de Janeiro State and in the Southern hemisphere.

Figure 1 .
Figure 1.Location of the sampling points.

Figure 4 .
Figure 4. Visual comparison among the State´s regions with multiple Q-Q plots.

Figure 5 .
Figure 5. Visual comparison among the 137 Cs and Organic Matter content in soils with multiple Q-Q plots.

Figure 6 .
Figure 6.Visual comparison among the 137 Cs and the intervals of Clay content.with multiple Q-Q plots.