Forest type effects on the retention of radiocesium in organic layers of forest ecosystems affected by the Fukushima nuclear accident

The Fukushima Daiichi nuclear power plant disaster caused serious radiocesium (137Cs) contamination of forest ecosystems over a wide area. Forest-floor organic layers play a key role in controlling the overall bioavailability of 137Cs in forest ecosystems; however, there is still an insufficient understanding of how forest types influence the retention capability of 137Cs in organic layers in Japanese forest ecosystems. Here we conducted plot-scale investigations on the retention of 137Cs in organic layers at two contrasting forest sites in Fukushima. In a deciduous broad-leaved forest, approximately 80% of the deposited 137Cs migrated to mineral soil located below the organic layers within two years after the accident, with an ecological half-life of approximately one year. Conversely, in an evergreen coniferous forest, more than half of the deposited 137Cs remained in the organic layers, with an ecological half-life of 2.1 years. The observed retention behavior can be well explained by the tree phenology and accumulation of 137Cs associated with litter materials with different degrees of degradation in the organic layers. Spatial and temporal patterns of gamma-ray dose rates depended on the retention capability. Our results demonstrate that enhanced radiation risks last longer in evergreen coniferous forests than in deciduous broad-leaved forests.

Scientific RepoRts | 6:38591 | DOI: 10.1038/srep38591 depending on the forest type. The tree canopies of evergreen coniferous forests acted as efficient filters of the atmospheric plume of 137 Cs from the Fukushima NPP accident; therefore, a large proportion of 137 Cs was initially intercepted and retained by the tree canopies and subsequently transferred to the forest floor via processes such as throughfall, stemflow, and litterfall 15,16 . By contrast, the canopy-interception effect was less important in deciduous broad-leaved forests compared to evergreen coniferous forests because the deciduous forests were leafless at the time of the accident; therefore, a large proportion of 137 Cs delivered by the Fukushima NPP accident was directly deposited on the forest floor [16][17][18] . Therefore, it is easily assumed that the differences in 137 Cs deposition behavior between the forests will result in different capacities of 137 Cs retention in their organic layers, at least over the first several years following forest contamination.
The rate of decomposition of litter materials in organic layers and the subsequent release of bioavailable 137 Cs are also important factors in controlling the rate of 137 Cs recycling within a forest ecosystem, and these factors are largely dependent on the forest type. Deciduous leaf fall is an annual occurrence and the decomposition of broad leaves in such forests has been shown to be rapid 19,20 , whereas the mean leaf longevity of evergreen coniferous trees is longer than a year 21 and the decomposition of needle-like leaves appears to be much slower compared to broad leaves 22,23 . The rate of decomposition of litter materials in organic layers is also influenced by climatological factors such as temperature and precipitation. Therefore, it is suggested that the behavior of 137 Cs in organic layers differs between European forests (affected by the Chernobyl NPP accident) and Japanese forests (affected by the Fukushima NPP accident) [24][25][26] ; studies under specific climatological and ecological conditions in Japan are urgently required to assess the environmental consequences of the Fukushima NPP accident. In addition, forests have complex stand structures and microtopography, and therefore, the quantity and quality (degrees of degradation) of litter materials accumulated in organic layers are highly spatially variable, even within a forest ecosystem 27 . Spatial heterogeneity can also be a complicating factor in quantitatively assessing the retention capability of organic layers for 137 Cs and its dependence on the forest type 20,28 .
Therefore, even though forest-floor organic layers are of key importance in 137 Cs cycling in forest ecosystems, there is still an insufficient understanding of how much and for how long organic layers can retain 137 Cs in Japanese forest ecosystems affected by the Fukushima NPP accident. To explore the role of organic layers in 137 Cs retention, we conducted plot-scale investigations at two contrasting forest sites in Fukushima, Japan (Fig. 1). We collected samples from the organic and underlying mineral soil (0-5 cm) layers at 25 locations within each of the 20 m × 20 m plot areas established in the deciduous broad-leaved forest (DBF) and evergreen coniferous (Japanese cedar-dominated) forest (CF). Litter samples in organic layers were collected separately from the upper L layer (litter layer, consisting of intact and relatively undecomposed leaves) and the lower F layer (fermentation layer, consisting of partially and well-degraded plant residues). The sample collections were conducted in December 2012 (21 months after the Fukushima NPP accident) at the DBF site and in August 2013 (29 months after the accident) at the CF site, and the collected samples were analyzed for radiocesium isotopes ( 137 Cs and 134 Cs) and for organic carbon (C) and total nitrogen (N). Based on the results, we evaluated the forest type effects on, and plot-scale spatial variability in, the retention behavior of 137 Cs in the organic layers of Japanese forest ecosystems to obtain insights into the radioecological consequences of the Fukushima NPP accident.

Results
Radiocesium concentrations. At both forest sites, 137 Cs concentrations were higher in the organic layer (L and F layers) than in the topsoil (0-5 cm) layer (Table 1; also see Tables S1 and S2 in Supplementary information). Within the organic layer, the concentrations showed a different depth-wise pattern between the sites; 137 Cs was specifically concentrated in the lower F layer at the DBF site, whereas 137 Cs concentration was fairly similar between the L and F layers at the CF site.
The measured 137 Cs and 134 Cs concentrations showed a similar pattern of distribution for all samples. The 134 Cs/ 137 Cs activity ratios were 0.56 ± 0.03 and 0.46 ± 0.02 (mean ± standard deviation) for the litter and soil samples collected at the DBF and CF sites, respectively ( Table 1). The ratios were close to those (0.57 and 0.46, respectively) theoretically predicted for Fukushima-derived radiocesium at the time of sample collection, the initial ratio being unity in March 2011 and decreasing according to different rates of radioactive decay (the physical half-lives of 137 Cs and 134 Cs are 30.1 and 2.1 years, respectively). The ratios indicate that the radiocesium isotopes observed in the present study originate from the Fukushima NPP accident. For simplicity, we will discuss only 137 Cs results in the following parts of this paper.
Total inventory of 137 Cs in the forest surface soils (organic and topsoil layers). The total inventories of 137 Cs in the forest surface soils (organic and topsoil layers) were 49.8 ± 11.0 kBq m −2 (mean ± standard deviation, n = 25) at the DBF site and 43.8 ± 12.1 kBq m −2 (n = 25) at the CF site (Table 1), and showed spatially heterogeneous distributions within the 20 m × 20 m plot areas (Figs 2b and 3b). The difference in the total 137 Cs inventory between the sites was not considered to be statistically significant (P = 0.07 via the unpaired t-test). Coefficients of variation (CV) values of the total 137 Cs inventory, calculated as the ratio of the standard deviation to the mean (i.e., relative standard deviation), were 22.1% and 27.5% for the DBF and CF sites, respectively. In both plots, the spatial distribution of the total 137 Cs inventory was not found to be related to that of the standing trees (Figs 2 and 3). There was no statistically significant (at a 5% significance level) correlation between the total 137 Cs inventory and the stand basal area in the 4 m × 4 m subplot (calculated as the sum of the cross-sectional areas at breast height for all trees in the target subplot) ( Table 2).

Inventory of 137 Cs in organic (L and F) layers.
Although the total 137 Cs inventories in the surface soils at the two sites were similar, the inventories in the organic (L + F) layers were quite different. The inventories of 137 Cs in the organic layers were 10.0 ± 3.2 kBq m −2 (CV: 31.9%) and 23.0 ± 5.5 kBq m −2 (CV: 24.1%) at the DBF and CF sites, respectively ( was found in the topsoil layer below the organic layer. By contrast, at the CF site, more than half (54.0 ± 12.2%) of the total inventory still remained in the organic layer, even though the sample collection at this site was conducted approximately 8 months after the sample collection at the DBF site.
The 137 Cs inventory in organic layers has been investigated several times at these sites since the accident in March 2011 29,30 . This allows us to assess the temporal changes in the 137 Cs inventory in the layers during the first three-year period (Fig. 4). The 137 Cs inventory in the organic layers decreased with time at both sites and this decreasing pattern can be characterized by an exponential decay model: where I t is the 137 Cs inventory (kBq m −2 ) in the organic layer at time t, I 0 is the 137 Cs inventory (kBq m −2 ) in the layer at time t = 0, λ is the decay constant (y −1 ), and t is the elapsed time (y) since the accident. The effective half-life (T eff in years) of 137 Cs in the organic layer can, therefore, be calculated as T eff = ln(2)/λ. Because the 137 Cs inventory data presented in Fig. 4 are decay-corrected to the sampling date, the effective half-life is a measure of the combined effect of physical (radioactive) decay and ecological elimination processes. The ecological half-life (T e in years) of 137 Cs in the organic layer can, therefore, be evaluated as Cs inventory at the CF site (Fig. 3), and there was a significant positive (r = 0.49, p < 0.05) correlation between the two inventories (Table 2). Contrastingly, the total and organic-layer 137 Cs inventories showed different spatial distribution patterns at the DBF site ( Fig. 2), and no correlation was found between the two. Similar to the total 137 Cs inventory, the 137 Cs inventory in the organic layer showed no correlation with the stand basal area in the target subplots at both sites ( Table 2). The inventory of 137 Cs in the organic layer was positively (r = 0.88, p < 0.0001) correlated with that of litter materials in the layer across the two sites ( Fig. 5a).
Within the organic layer, the lower F layer accumulated more 137 Cs than the upper L layer at the DBF site, whereas the L layer accumulated more 137 Cs at the CF site (Table 1). For each layer, there was a significant (p < 0.01) positive correlation between the 137 Cs inventory and the litter-material inventory in the layer ( Fig. 5b; r = 0.74 and r = 0.84 for the L and F layers of the DBF, respectively, and r = 0.65 and r = 0.61 for the L and F layers of the CF, respectively). At the CF site, the 137 Cs inventory appeared to increase linearly with increasing litter-material inventory regardless of the layer (L or F). At the DBF site, however, the 137 Cs inventories for the L and F layers were not observed to be approximated by a single linear relationship. These observations indicate that the larger accumulation of 137 Cs in the organic layer at the CF site can be explained simply by the larger accumulation of litter materials on the forest floor, while the larger accumulation of 137 Cs in the organic layer at the DBF site depends more on the larger accumulation of highly 137 Cs-contaminated materials in the F layer.
Carbon and nitrogen contents. At both sites, the L layer, consisting of intact and relatively undecomposed leaves, showed a high C content (> 440 gC kg −1 dw, Table 1), indicating that the litter materials in the L layers have originated primarily from recent litterfall events. The C content in the topsoil was much lower than that in the organic layer, which was likely due to the mixing of mineral soil particles with organic materials in the topsoil layer. The carbon-to-nitrogen (C/N) ratio generally decreased with depth, indicating that the litter materials in the F layers were older and had experienced higher degrees of microbial decomposition compared to those in the L layers 31,32 . This was consistent with the visual aspect of the litter materials collected from the F layers as a mixture of finely fragmented plant residues and macroscopically unrecognizable organic materials 20 .
Gamma-ray dose rates at the ground surface. Gamma-ray dose rates were measured at the ground surface at both sites using a plastic scintillation fiber (PSF) detection system, and the spatial distributions of the dose rate in the 20 m × 20 m plot areas were recorded (Figs 2 and 3). At the DBF site, the gamma-ray dose rates tended to be higher in areas with a larger accumulation of 137 Cs in the surface soil (organic and topsoil layers). At the CF site, however, the dose rates seemed to be more related to 137 Cs accumulation in the organic layer, rather than in the surface soil. The gamma-ray dose rates were significantly (p < 0.01 via the unpaired t-test) higher at the CF site (0.34 ± 0.04 μ Sv h −1 ) than at the DBF site (0.26 ± 0.09 μ Sv h −1 ).

Discussion
Organic layers on the forest floor are a dynamic component of forest ecosystems and play a key role in ecosystem functioning via the retention and cycling of nutrients 33 . Our plot-scale investigations conducted at two contrasting forests in Fukushima show that organic layers also play a significant role in the retention, and therefore, the behavior of 137 Cs in forest ecosystems. In the DBF, approximately 80% of the 137 Cs that had been deposited onto the forest floor migrated to the underlying mineral soil within two years of the accident (Table 1). Contrastingly, in the CF, more than half of the deposited 137 Cs was retained in the organic layer even two and a half years after the accident. The ecological half-life of 137 Cs in the organic layers was estimated to be 2.1 years at the CF site, which was approximately twice as long as that at the DBF site (Fig. 4). Overall, the correlations found between 137 Cs and the litter-material inventories in the organic layers across the two sites (Fig. 5a) indicate that, in general, the denser (or thicker) organic layers of the CF more efficiently retain 137 Cs on the forest floor than the less-dense (or thinner) organic layers of the DBF 4,28,29,34 . The separate sampling and analysis of the L and F layers provided more detailed information concerning the accumulation of litter-associated 137 Cs within the organic layers and revealed that the retention processes of 137 Cs in the organic layers differed markedly between the two forest sites. The retention processes of 137 Cs in organic layers appeared tightly linked to the ecosystem processes in the two contrasting forests. In the organic layers of the DBF, 137 Cs was primarily retained in association with highly degraded (evidenced by the C/N data) litter materials in the lower F layers (Table 1 and Fig. 5b). In temperate DBFs in Japan, the bulk of the litterfall occurs in autumn (October and November) and new leaves begin to expand in spring (April and May). Trees at the DBF site had no leaves in March 2011 when the Fukushima NPP accident occurred, and therefore, it is assumed that most of the Fukushima-derived 137 Cs was directly deposited onto forest-floor litter materials 17,18 . Because broadleaf litter is known to be relatively rapidly decomposed via microbial activity in the organic layers (a mean residence time of a few years) 19,20,31 , it is likely that the litter materials contaminated by the direct deposition of 137 Cs have been degraded since March 2011 and that most of the 137 Cs was leached from the organic layers and then immobilized in the topsoil, with a minor fraction (14% of the total deposition) being still retained in the F layers in December 2012 17,18,35 .
At the time of sample collection in the DBF, the annual litterfall in 2012 was complete and the litter materials (intact and relatively undecomposed leaves) in the L layer were primarily from the most recent litterfall events. The 137 Cs concentration of the litter materials in the L layers was significantly lower than that in the F layers (Table 1), but three orders of magnitude higher than the pre-accident level, indicating that the newly emerged leaves in 2012 were contaminated with 137 Cs via mechanisms such as translocation from tree stems and uptake by roots from the soil 36,37 . However, the short ecological half-life (0.95 years) of 137 Cs in the organic layers at this site suggests a rapid decrease in the inventory of potentially mobile 137 Cs in the organic layers, and therefore, a rapid reduction in 137 Cs recycling in the soil-plant system 4,14 . This scenario for the DBF site is consistent with the rapid decrease in 137 Cs concentration in fresh broad leaves, which has been observed at various locations in Japan following the Fukushima NPP accident 20,37,38 . It has also been observed that in a DBF in Japan, the amount of 137 Cs that migrated downward through the litter-mineral soil boundary decreased with time at a rate of approximately  50% per year during the period of April 2012-May 2015 25 . This reduction rate for 137 Cs migration is in good agreement with the estimated ecological half-life of 137 Cs in the organic layers at our DBF site.
In contrast to the DBF, the organic layers of the CF showed a high retention capability for 137 Cs, which was primarily dependent on the accumulation of less-degraded, highly contaminated litter materials in the L layers (Table 1 and Fig. 5b). This high retention capability is related to the tree phenology at this site and can be reasonably explained by considering that leaves on the cedar trees, as well as the litter materials on the forest floor, were directly contaminated with 137 Cs as a result of fallout from the Fukushima NPP accident in March 2011. Since then, a portion of the contaminated leaves has gradually been transferred to the forest floor via litterfall over time, and most of the 137 Cs of litterfall origin remained in the L layers until August 2013. This migration pattern is considered reasonable given the observation that more than 60% of the Fukushima-derived 137 Cs was initially intercepted by the tree canopy in evergreen coniferous forests 15 . This pattern is also consistent with the leaf longevity of Japanese cedars (4-8 years) 21,39 . The lagged (indirect) input of the 137 Cs-contaminated leaves to the forest floor 16 , in combination with a possible slower microbial decomposition of needle-like cedar leaves in the organic layers compared to broad leaves 22,23 , is the most likely explanation for the observed higher retention (or longer ecological half-life) of 137 Cs in the organic layers at the CF site compared to that at the DBF site. The results also suggest that additional inputs of leaf-associated 137 Cs to the forest floor over the next several years are possible.
The effective retention of 137 Cs (as potentially mobile 137 Cs) in the thick organic layers of the CF may facilitate the recycling of 137 Cs via 137 Cs uptake by trees from the organic layers and its subsequent re-deposition onto the forest floor via litterfall 5,40 . The scenario for the CF site highly contrasts that for the DBF site (see above), demonstrating that the forest type significantly influences both the short-and long-term behaviors of 137 Cs within a  forest ecosystem 7,39 , and clearly, the retention of 137 Cs in organic layers plays a key role in controlling the overall 137 Cs behavior 5,26 . The observations conducted in European forests after the Chernobyl NPP accident have consistently shown that a large proportion of 137 Cs persisted in forest organic layers for over a decade, resulting in long-lasting 137 Cs bioavailability in the soil-plant system [3][4][5][6][7]28 . This may be partly due to climatological factors such as temperature and precipitation, both of which can affect the microbial decomposition of litter materials in organic layers 23,24,41 . For example, a mean residence time of 28-42 years for litter materials has been reportedly observed for organic layers in a spruce-dominated forest in Scandinavia 42 , which is noticeably longer than that of several years observed for the organic layers of DBFs in Japan 19 .
The 137 Cs inventory in the organic layers was spatially variable at the plot scale (20 m × 20 m) in both forests, with CV values of 32% and 24% for the DBF and CF sites, respectively. In the present study, the spatial distribution of the 137 Cs inventory in the organic layers did not show any clear relationship to that of the standing trees and stand basal area (Figs 2 and 3 and Table 2), suggesting that the forest stand structure was not a factor causing the spatial heterogeneity of the 137 Cs inventory in the organic layers. Instead, the correlations observed between the 137 Cs and litter-material inventories ( Table 2 and Fig. 5) suggest that the accumulation and reduction processes of litter materials on the forest floor were more significant in creating the spatial variability in the 137 Cs inventory in the organic layers 20 . The accumulation and reduction processes include the lateral transport of leaves via wind action while falling and the redistribution of fallen leaf litter on the forest floor 27,43,44 . The decomposition rate of litter materials on the forest floor, which can be modified by microclimate-related microbial activities and soil invertebrate activities, is also a possible factor influencing the increase and decrease in the amount of litter materials in the organic layers 45,46 .
The retention properties of the organic layers for 137 Cs also exerted a significant influence on the spatial and temporal patterns of the gamma-ray dose rate at the ground surface. In the CF where a large proportion of the deposited 137 Cs had been retained in the organic layers, the spatial pattern of the gamma-ray dose rate was found to be relevant to that of the 137 Cs inventory in the organic layers (Fig. 3). In the DBF where approximately half of the deposited 137 Cs had already migrated to the mineral soil, the spatial pattern of the gamma-ray dose rate was found to be more relevant to that of the total 137 Cs inventory, rather than the organic-layer 137 Cs inventory (Fig. 2). The gamma-ray dose rate was significantly higher at the CF site than at the DBF site, even though the total inventory of 137 Cs in the surface soils (organic and topsoil layers) was similar for both sites (Table 1). This is likely a result of the shielding effect because gamma radiation from 137 Cs in deeper soil layers is more efficiently attenuated by the overlying soil 47 .
In conclusion, our study demonstrates that the forest type strongly affects the retention capability of organic layers for 137 Cs in conjunction with the litter dynamics, and subsequently, affects the mobility and bioavailability of 137 Cs within a forest ecosystem. The forest-type-dependent retention capability indicates that enhanced radiation risks delivered from both internal (via consumption of forest products) and external exposures to the local population last longer in CFs than in DBFs. The spatial and temporal patterns of the gamma-ray dose rate and their control factor also depend on the forest type. Continuous observations of the 137 Cs inventory in the organic layers in these two contrasting sites will help improve our understanding of the retention behavior of 137 Cs in organic layers of forest ecosystems in Japan, which is the key to accurately assess both short-and long-term radioecological impacts of the Fukushima NPP accident.

Methods
Study sites. This study was conducted at two forest sites ( Fig. 1) located in the southwestern part of the city of Fukushima (37.71°N, 140.36°E), approximately 70 km northwest of the Fukushima Daiichi NPP. The first site was a DBF dominated by Japanese oak and hornbeam, located along a riverbank. The second site was a CF dominated by Japanese cedar, located along a trail over a small mountain. The distance between the two sites is approximately 1.5 km. The soils have been classified as Fluvisols and Andosols for the DBF and CF sites, respectively, using the classification of the Food and Agriculture Organization 48 . The mean annual temperature and precipitation for the last 10 years at the Fukushima meteorological station are 13.3 °C and 1234 mm, respectively (Japan Meteorological Agency).
The vertical distribution and retention processes of the Fukushima-derived radiocesium in the organic and mineral soil layers of these sites have been studied since the accident in March 2011 29,30,49 . The DBF and CF sites in this study correspond to the sites represented as FR-2 and FR-5, respectively, in the previous studies. For more detail on the site characteristics, including the physicochemical properties of the soils, see Koarashi et al. 29 .
Sample collection and treatment. Field samples were collected in December 2012 (21 months after the accident) at the DBF site and in August 2013 (29 months after the accident) at the CF site. At the time of sample collection in the DBF, the trees had no leaves due to autumn leaf fall (litterfall) in 2012 and the forest floor was slightly covered with snow (see Fig. 1(b)). A 400 m 2 (20 m × 20 m) plot was established at each site and was divided into 25 subplots each having an area of 4 m × 4 m. Litter samples in the organic layer (the upper L and the lower F layers, separately) were collected manually from an area of 900 cm 2 randomly selected within each of the 25 subplots (but generally located near the center of each of the 25 subplots; see Figs 2a and 3a). Three replicate topsoil (0-5 cm) samples were then collected using a cylindrical soil sampler (5 cm in diameter and 5 cm in depth) from the soil surface where the organic layer was removed. The depth range (0-5 cm) of the soil sampling was chosen on the basis of our previous studies, which showed that most (> 94%) of the total 137 Cs inventory was located within the organic and the upper 5-cm soil layers at the sites 29,30 . At each site, the sampling locations and forest stand descriptions (locations and breast height diameters of the standing trees) were recorded. The DBF site was stony, and therefore, the locations and sizes of stones (larger than 10 cm in diameter) were also recorded.
The collected samples (litter and soil) were immediately cooled with dry ice in containers, transported to our laboratory, and then dried to a constant weight at room temperature. The litter samples were finely chopped using Scientific RepoRts | 6:38591 | DOI: 10.1038/srep38591 a mixer to obtain homogenized samples after removing coarse woody debris (fallen branches and twigs). The soil samples were sieved through a 2-mm mesh. Radiocesium analysis. The activity concentrations of 137 Cs and 134 Cs in the litter and soil (< 2 mm) samples were determined using gamma ray spectrometry, and their values were expressed in activity per unit dry weight (Bq kg −1 dw). Samples (or subsamples) were sealed in plastic tubes (5-cm diameter, 7-cm height) and analyzed for 137 Cs and 134 Cs using a high-purity coaxial germanium detector (model GEM25P4-70, ORTEC, USA) at the Nuclear Science and Engineering Center of the Japan Atomic Energy Agency (JAEA). The detector was calibrated with standard gamma sources (each with a relative uncertainty of ~5% for 137 Cs) with different sample heights. The measurement times were 2,000-20,000 s for litter samples and 2,000-100,000 s for soil samples, which allowed us to obtain both 137 Cs and 134 Cs concentration values with relative errors of < 10% (with some exceptions for samples having low 134 Cs concentrations). The activity concentrations were corrected for radioactive decay to the sampling date.
Radiocesium inventories in the L and F layers, I (Bq m −2 ), were estimated as where I is the radiocesium inventory (Bq m −2 ) in the layer, A is the radiocesium activity concentration (Bq kg −1 dw) of the samples collected from the layer, and M is the amount of litter materials per unit area (kg m −2 ) in the layer. The subscripts L and F indicate the L and F layers, respectively. The radiocesium inventory in the topsoil (0-5 cm) layer was estimated to be

S S
where I S is the radiocesium inventory (Bq m −2 ) in the topsoil layer, A S is the radiocesium activity concentration (Bq kg −1 dw) of the soil (< 2 mm) samples, B is the bulk density of the soil (kg m −3 ), g is the gravel (> 2 mm) content of the bulk soil (kg kg −1 ), and d is the thickness (m) of the layer (i.e., 0.05 m). Soil samples were collected and analyzed in triplicate (see above), and therefore, the radiocesium inventory in the topsoil layer at each sampling location was evaluated as the average of the triplicate samples (n = 3).
Carbon and nitrogen analysis. A portion of each sample (both litter and soil) was analyzed for their total C and N contents using an elemental analyzer (vario PYRO cube, Elementar, Germany) 20 .
In situ measurement of spatial distribution of the gamma-ray dose rate. The spatial distribution of the gamma-ray dose rate in the 20 m × 20 m square plot was measured using a PSF detection system. This system consists of a 10-m-long PSF that responds to gamma radiation, light detection units connected to both sides of the PSF, a signal processing unit, and other components 50 . The system can provide numerical values of the gamma-ray dose rate in air along the 10-m-long PSF at an interval of 0.5 m. The 20 m × 20 m square plot established at each site was divided into two 10 m × 20 m areas. For both areas, the distributions of the gamma-ray dose rate at the ground surface in the short axis (10 m) direction were measured by successively setting the 10-m-long PSF on the forest floor at an interval of 1 m along the long axis (20 m) direction (i.e., measurements were conducted 20 × 2 times for each site). We also measured the gamma-ray dose rates at several points within the plot using a NaI scintillation survey meter. Comparing the data obtained by the PSF detection system to those obtained by the NaI survey meter allowed us to perform an in situ calibration of the PSF detection system.
Visualizing the spatial distributions. To visualize the spatial distributions of the 137 Cs inventories and the gamma-ray dose rates in the 20 m × 20 m plot areas, contour maps were created using the observation results and ordinary Kriging with the Surfer ® 12 software (Golden Software, Inc., Golden, CO, USA) 28,51,52 .