Temporal changes in vertical distribution of ¹³⁷Cs in litter and soils in mixed deciduous forests in Fukushima, Japan

ABSTRACT

Downward migration of ¹³⁷Cs in soils was studied in three mixed deciduous forests c.a. 40 km northwest of the Fukushima Daiichi Nuclear power plant (FDNPP), Japan. We selected three different types of forest regarding to environmental condition such as slope inclinations and snow coverage conditions at the time of deposition. We examined temporal changes in the vertical distribution of ¹³⁷Cs from litter layers to 10 cm soil depth for two years (2.3 to 4.3 years after the FDNPP accident in 2011). At all three study sites, the ¹³⁷Cs in the litter layer had largely migrated to surface soil by 2013. After 2014, about 80% of the ¹³⁷Cs in forest soils (litter layer to 10 cm soil depth) remained within 0–5 cm soil layer. The vertical distribution had not changed substantially since 2014, suggesting that changes to the downward migration rates of ¹³⁷Cs in soils drastically decreased with time. In addition, small amounts of migrating ¹³⁷Cs could not be detected by the present method because there was a large spatial variation in the distribution of soil ¹³⁷Cs. The results showed similar patterns of soil ¹³⁷Cs distribution among the three study sites although there were differences in the environmental conditions.

KEYWORDS:

• Downward migration
• forest soils
• cesium 137
• spatial variation
• accident
• adsorption

1. Introduction

Large amounts of radionuclides were released into the environment during the Fukushima Daiichi Nuclear Power Plant (FDNPP) accident in 2011 [1,2], and were widely dispersed in the surrounding forest areas. Forest areas account for more than 70% of land area in Fukushima Prefecture [3]. Radioactive Cs, especially ¹³⁷Cs, will remain in these areas for a considerable period of time because ¹³⁷Cs has a long half-life of 30.17 years. Therefore, the radioactive contamination of forests and forest products will be a matter of serious concern for some time to come [3].

Most of the ¹³⁷Cs that was deposited in mixed deciduous forests in the Fukushima region after the FDNPP accident was transferred to the forest floors by precipitation (stemflow and throughfall) and litterfall [4,5]. The high adsorption ability of radioactive Cs on soil particles [6], and the events after the Chernobyl accident [7] led to the expectation that ¹³⁷Cs in the forests in Fukushima would migrate from the litter layer to the surface soil and deeper soil over time, and that the ¹³⁷Cs will remain in the surface layer of soils for a long time [8]. Therefore, the downward migration of ¹³⁷Cs in the soils will be the most important transport of ¹³⁷Cs in a forest ecosystem to be considered on a temporal basis. However, a few studies into the migration of ¹³⁷Cs in forest soils in Fukushima regions have been published since the FDNPP accident [9,10], and these studies have focused on the early stage of the ¹³⁷Cs migration in soils, which show relatively rapid migration rates. In addition, because the topography in the northwest of FDNPP (the present study site) largely consists of steep slopes, there is a possibility that large amounts of ¹³⁷Cs deposited on forest floors are transported from ridge to valley with litter and surface runoff water, and consequently discharges to outside ecosystems. The ¹³⁷Cs movement in a forest ecosystem may also be dependent on site characteristics; for example, the amount of snow fall that remained on the forest floors at the time of the accident was variable by site. ¹³⁷Cs deposited on snow as initial deposition migrates through the snow and/or with melting of the snow. The subsequent downward migration process of ¹³⁷Cs in litter layer and soil may differ depending on initial deposition place; snow or litter layer and soil. Such spatial heterogeneity effects must be considered in the study of downward migration of ¹³⁷Cs in forest soils.

In the present study, we examined temporal changes in the vertical distributions of ¹³⁷Cs in soils in mixed deciduous forests in Fukushima Prefecture to improve our understanding of the downward migration of ¹³⁷Cs in forest soils. We selected different forest areas as study sites and aimed to clarify the effects on the migration based on steep slope angles of the forests (horizontal movement along steep slopes) and the difference in snow coverage at the time of the deposition. In this study, we focused on downward migration in the transition phase from rapid migration phase to slow migration phase. The field surveys were conducted for 2 years (from 2.3 to 4.3 years after the FDNPP accident).

2. Material and methods

2.1. Study sites

Three study sites were selected (Sites A, B, and C). These study sites were located in secondary mixed deciduous forests in Fukushima Prefecture (Figure 1). The field conditions of each study site are shown in Table 1. Distances from FDNPP to three study sites were 39–46 km. The total atmospheric deposition of ¹³⁷Cs after the accident in the study areas ranged from 100 to 600 kBq m^–2, based on the third airborne monitoring survey by the Japanese government [11]. Site A had no snow coverage at the time of the deposition of the radioactive materials, while Site C had a substantial snow layer on the forest floor, judging from a satellite image on 14 March 2014. Snow cover at Site B was at an intermediate level between Sites A and C. The potential for ¹³⁷Cs to move horizontally because of the presence of a steep slope was investigated by comparing sites with steep slopes (Sites A and C, which had slopes of 30° and 20°, respectively) and a flat site (Site B, which had a slope of 5°). Mean annual air temperature and precipitation recorded at the nearest meteorological station to all study sites were 10.0 °C and 1361.6 mm, respectively (Iitate Meteorological Station, 9–12 km from the study sites, extracted from the Japan Meteorological Agency database available at http://www.jma.go.jp/jma/index.html). The study sites are typical naturally regenerating forests (secondary forests) in the northern part of Japan and are dominated by deciduous broadleaf trees (e.g. Oak; Quercus mongolica subsp. crispula, Quercus serrata) and sporadic evergreen coniferous trees (e.g. Japanese fir, Abies firma). The soil types at the study sites are brown forest soil and Andisol [12]. The thickness of the organic layer (L, F, and H layers) at each site was approximately 1.5 cm on average. The densities of the soil at 0–5 and 5–10 cm depths as of November 2014 were 0.48 and 0.63 g cm^–3 on a dry weight basis, respectively (Table 1).

Figure 1. Location of the study sites and the Fukushima Daiichi Nuclear Power Plant (FDNPP).

Table 1. Environmental conditions for each study site.

Study site	A	B	C
Longitude	37°43′N	37°45′N	37°44′N
Latitude	140°29′E	140°44′E	140°43′E
Altitude (m)	285	510	445
Distance from FDNPP (km)	39	46	43
Total atmospheric deposition (kBq m^−2)*	100–300	300–600	300–600
Snow coverage on forest floor at the time of the accident**	Not at all	A little	Quite a lot
Average degree of slope (°)	30	5	20
Direction of slope	North	Northwest	Southeast
Soil type***	Brown forest soils	Brown forest soils	Andisol
Soil density (g cm^−3)****
0–5 cm	0.44	0.38	0.60
5–10 cm	0.59	0.46	0.74

* MEXT (2011) results of the third airborne monitoring survey.

^** Judging from a satellite image on 14 March 2011.

^*** Kanno et al. (2008) Pedologist 52, 129–133.

^**** Sampling date, November 2014 on dry weight basis.

2.2. Sample collection and measurements

Field surveys were continued for 25 months from August 2013 to August 2015. Sample collection was conducted every 3–5 months (August 2013, December 2013, March 2014, August 2014, November 2014, March 2015, and August 2015). We collected samples of the litter layer (organic layer), and 0–5 and 5–10 cm soil layers because most ¹³⁷Cs from the FDNPP accident remains within the upper 10 cm of soil [10]. To attempt to confirm this, we took soil samples from the 10–15 cm soil depth in March 2015 only. On each sampling date, litter from the organic layer was collected from an area of 10 × 10 cm, and soil samples were collected with a 100 ml soil core sampler (20 cm² and 5 cm depth). The locations for the sample collection at each study site were randomly chosen at each field survey, and eight samples from the litter layers, 0–5 and 5–10 cm soil layers were collected from each site. Four samples were collected from 10–15 cm soil layer in March 2015. Samples were collected from the middle sections of the slopes at Sites A and C. In the present study, the soil ¹³⁷Cs inventories were examined in 5 cm depth sections because ¹³⁷Cs inventories within the 0–5 cm soil layer have a large spatial heterogeneity (a difference of up to 30 times between the lowest and highest values) [13]. We, therefore, decided to discuss downward migration at a scale of 5 cm soil depth in the present study.

The collected litter and soil samples were dried at 80 °C for 48 h and lightly ground to pass through a 2-mm sieve. The samples were packed into 100-ml plastic polystyrene containers (U-8), and analyzed for ¹³⁷Cs and ¹³⁴Cs using a low-background Ge spectrometer (GEM-110225, Seiko EG&G). The measurement times were between 600 and 30,000 s, depending on radioactivity of the samples. The associated errors were composed of 5% from the detection efficiency and 1%–10% from peak counting error. The activity of ¹³⁷Cs and ¹³⁴Cs in all samples showed a similar pattern, and the ratio of ¹³⁷Cs/¹³⁴Cs activity was almost constant (1.0–1.1 in March 2011). In the present study, we only employed ¹³⁷Cs for the analysis. The ¹³⁴Cs data were not used because the relatively short half-life (2.06 years) is unsuitable for this analysis more than four years after the FDNPP accident. Activity of ¹³⁷Cs was corrected for radioactive decay to the first sampling day of this survey, 1 August 2013. The ¹³⁷Cs activities of the entire soil sample collected with the 100-mL soil sampler were measured, and the inventory data were calculated by dividing each ¹³⁷Cs activity by the area of the sampler (20 cm²).

2.3. Statistical analysis

Application of the Shapiro–Wilk normality test showed that some data-sets in the present study were logarithmically normal distributions (P > 0.05), although the remaining data-sets were marginally not normal (P < 0.05). Many studies after the Chernobyl accident showed that approximation of soil ¹³⁷Cs inventory with data showing a logarithmically normal distribution is suitable for statistical analysis [14,15]. Therefore, all the data-sets in the present study were log-transformed for statistical analysis.

Geometric means of the ¹³⁷Cs inventory (kBq m^–2) and the coefficient of variation (ratio of the standard deviation of log-transformed ¹³⁷Cs inventory to the geometric mean, hereafter CV) were calculated. The one-way analysis of variance (ANOVA) test with multiple comparisons (Tukey's test) was used to compare ¹³⁷Cs inventories among different layers and sampling periods.

3. Results

Temporal changes in the ¹³⁷Cs inventories in the litter layer, and at 0–5 and 5–10 cm soil layer for two years are shown in Figure 2 and Table S1 (Supplemental Online Material). The highest ¹³⁷Cs inventory value, 295 kBq m^–2, was found in the 0–5 cm soil layer at Site C in November 2014. The lowest ¹³⁷Cs inventory value, 7 kBq m^–2, was found in 5–10 cm soil layer at Site A in August 2013. Generally, the ¹³⁷Cs inventory in the litter layer decreased over time, the inventory at 0–5 cm soil layer increased over time, and the inventory at 5–10 cm soil layer slightly increased over time (one-way ANOVA, P < 0.001 at Sites A, B, and C). The largest and statistically significant change in each layer occurred from August to December 2013. Large decreases and increases were observed in the ¹³⁷Cs inventories in the litter and the 0–5 cm soil layers, respectively. In litter layers, no significant changes were observed in the litter inventories in 2014 after the decrease in 2013. Subsequently, the litter inventories in Site A and B significantly increased from November 2014 to March 2015 that in Site C also increased although it was not statistically significant. The ¹³⁷Cs inventories in the soil 0–5 cm soil layers did not significantly increase since 2014. In all the three study sites, the ¹³⁷Cs inventories in the soil 5–10 cm soil layers did not shoed statistically significant increases in and after December 2013.

Figure 2. Temporal changes in the geometric means of the ¹³⁷Cs inventories for the litter layers, 0–5 cm soil layer, and 5–10 cm soil layer and the total inventories (for the litter and soil to a depth of 10 cm) from August 2013 to August 2015 at Sites A, B, and C. Inventories for 10–15 cm soil layer for March 2015 are also shown. Each vertical bar indicates the standard deviation.

The ¹³⁷Cs inventories in the litter layer exceeded the sum of the ¹³⁷Cs inventories in the 0–5 cm layer and the 5–10 cm layer at the three sites observed in August 2013. The litter layer inventory in August 2013 was higher than the 0–5 cm soil layer inventory, but in November 2013, the 0–5 cm soil layer inventory was higher than the litter layer inventory. The litter layer ¹³⁷Cs inventory accounted for 60%–70% of the total ¹³⁷Cs inventory (for the litter and the soil to a depth of 10 cm) in August 2013 but accounted for only around 10% one year later. The litter layer ¹³⁷Cs inventory then remained stable for several months, but increased by a factor of two to three in 2015. The contributions of the 0–5 cm soil layer inventory to the total inventory increased from 20%–30% to 70%–80% for one year between August 2013 and August 2014. The ¹³⁷Cs inventories at increased by a moderate amount overall during the study period, but then substantially increased by a factor of two to three between August and December 2013. No dramatic changes in the vertical distributions (such as those that occurred in 2013) occurred in any layer after 2014. Although there were moderate fluctuations in the total ¹³⁷Cs inventories, statistically significant changes in the total ¹³⁷Cs inventories (for the litter and soil to a depth of 10 cm) were not found during the study period at Sites A and B (one-way ANOVA, P > 0.05), which is because the standard deviations were large. These large standard deviations were caused by the high degree of spatial heterogeneity at 0–5 cm soil layer. However, at Site C, the total inventory significantly but marginally differed during the study period (P = 0.011). The ¹³⁷Cs inventory for 10–15 cm soil layer was determined only in March 2015, and, at that time, the ¹³⁷Cs inventory at 10–15 cm soil layer was 43%–70% of the ¹³⁷Cs inventory for 5–10 cm soil layer. This ¹³⁷Cs at the 10–15 cm soil layer was derived from not only the FDNPP accident in 2011 but also the global fallout 1950s–1960s [16]

Temporal changes in the CVs of the inventories for the different layers and the total inventories during the study period are shown in Figure 3 and Table S2 (Supplemental Online Material). No clear differences were found between the CVs for the study sites with steep slopes and the flat study site, and the CVs were not related to snow coverage conditions that had occurred. The CVs for the litter layers did not show clear temporal changes. The CVs for the soils decreased most markedly in 2013. The deeper soils had larger CVs. Except at Site A, the CVs for the litter layer were higher than the CVs for the 5–10 cm soil layer. The CV of the total inventories did not fluctuate substantially during the study.

Figure 3. Temporal changes in the coefficients of variation of the ¹³⁷Cs inventories (the ratio between the standard deviation of the log-transformed ¹³⁷Cs inventory and the geometric mean) for the litter layer, 0–5 cm soil layer, and 5–10 cm soil layer, and of the total inventories (from the litter to 10 cm soil depth) from August 2013 to August 2015 at Sites A, B, and C. Results for 10–15 cm soil layer in March 2015 are also shown.

4. Discussion

The ¹³⁷Cs inventories for the litter layers decreased and the ¹³⁷Cs inventories for the soils increased. These results indicated that the ¹³⁷Cs in the litter layer migrated into the soils during the study period, as shown in many previous studies [9,17]. This pattern was observed at all three of our study sites; at Sites A and C with steep slopes, significant temporal changes of the total inventories (litter layer to 10 cm soil depth) were not observed, more than two years after the FDNPP accident. In addition, we did not observe any clear effects of initial deposition relating to snow coverage on the forest floors at the time of deposition (Figure 2). The total inventories in all the study sites increased even though those in Site A and B were not statistically significant. These increases were reflected by the increases in ¹³⁷Cs in soil 0–5 cm layers, but they are not totally supported by migration from the litter layers. One of the causes of specific accumulation of ¹³⁷Cs in soil 0–5 cm layer is due to lateral migration, and another may be direct input from post-accident deposition.

The large decreases in the ¹³⁷Cs inventories in the litter layers in 2013 are likely to have been partly caused by contaminated litter decomposition, as suggested in many previous studies [17–19]. Although the present study started 2.3 years after the FDNPP accident, previous study earlier started in Fukushima showed decrease in ¹³⁷Cs in litter layers in 2011 and 2012 [19]. In present study sites, ¹³⁷Cs in litter layers is also assumed to have migrated into soil in 2011 and 2012. According to an experimental study, litter is largely decomposed within three years in deciduous forests in northern Japan [20], therefore the most contaminated litter in 2011 is thought to have almost finished by 2013, explaining that difference in decrease rate of ¹³⁷Cs in the litter layer between in the first year of our study (2013) and in the following years (2014 and 2015). However, the ¹³⁷Cs activities in the litter layers decreased much more quickly than litter decomposes, on a weight basis, in deciduous forests in northern Japan in 2013 [20]. The decreases in the ¹³⁷Cs inventories in the litter layers in our study may have been caused by a combination of the litter decomposition and other environmental factors, such ¹³⁷Cs being washed off the litter surfaces by precipitation and the ¹³⁷Cs being leached from the litter [18,21]. The ¹³⁷Cs inventories in the litter layers increased slightly in 2015 at all study sites. This was assumed to be caused by translocation of the ¹³⁷Cs from the soil to the litter layer by fungi [17,19], and the additional contamination by fresh litterfall containing ¹³⁷Cs by biological recycle in the forest ecosystems [22]. The translocation by fungi [17,19] could have occurred in 2013 and 2014, but it would be difficult to determine whether it had occurred because the ¹³⁷Cs inventories in the litter layers decreased rapidly.

The ¹³⁷Cs activities increased much more at 0–5 cm soil layer than at 5–10 cm soil layer, suggesting that the ¹³⁷Cs that migrated from the litter layers to the soils remained in the surface layers of the soils. This result agrees with the results of many previous studies in forests [8,18,22,23]. The ¹³⁷Cs remained in the surface layers of the soils because the ¹³⁷Cs migrated into soils rapidly and subsequently was strongly adsorbed onto soil particles [6]. At Sites A and C, ¹³⁷Cs inventories significantly increased at 5–10 cm soil layer in 2013, but thereafter, there were no clear signs of ¹³⁷Cs migrating from 0–5 to 5–10 cm soil layer. This may be explained the hypothesis that just after the FDNPP accident, ¹³⁷Cs quickly migrated into deeper soil layer via pore water without adsorption onto soil particles, which contained ¹³⁷Cs leaching from the litter layer on the forest floors, throughfall and stemflow. As a result, the ¹³⁷Cs migration to 5–10 cm soil layer decreased with decrease of dissolved ¹³⁷Cs in soils [18]. In addition, if the vertical ¹³⁷Cs migration was smaller than the spatial variation of ¹³⁷Cs in soils, the vertical ¹³⁷Cs migration from 0–5 to 5–10 cm soil layer may be behind the spatial variation. Therefore, this study period of only two years may be too short to detect slow vertical migration in soil columns since 2014. A study with a longer duration (at least dozens of years) should be conducted to provide clear evidence to support previous studies [7,24]. Similarly, a larger sample should have been taken to detect the slow ¹³⁷Cs migration in soils after 2014 (e.g. more than 100 soil samples were required to detect statistically significant differences based on power analysis). However, such a considerable large sampling size was impractical.

The amounts of ¹³⁷Cs in each layer was not observed to increase or decrease markedly after 2014 and showed relatively stable, indicating that the present study sites entered a new phase which ¹³⁷Cs in forest soils migrates very slowly. This may be termed the ‘quasi-equilibrium phase,’ ‘relatively stable phase,’ or ‘steady state,’ as was observed after the Chernobyl accident [18,22,23]. Thus, the dramatic downward migration of large amounts of ¹³⁷Cs is unlikely to continue to occur in forests around the FDNPP in the future.

The CVs of the ¹³⁷Cs inventories in the litter layers (i.e. spatial variations in the ¹³⁷Cs inventories in the litter layers) did not follow any noticeable temporal changes related to the decreases in the ¹³⁷Cs inventories of the litter layers during the study period. The CVs of the ¹³⁷Cs inventories for the soils decreased, especially in 2013, and the decrease occurred at the same time as the ¹³⁷Cs inventories of the litter layers and soils decreased and increased, respectively. This indicated that the ¹³⁷Csthat was deposited onto the forest soils shortly after the FDNPP accident as initial deposition has a heterogeneous spatial distribution and then became more homogeneous with increasing migration. The temporal changes in the spatial variability of ¹³⁷Cs in the soils occurred during the final phase of ¹³⁷Cs inputs to the forest floors and soils via stemflow and throughfall, as observed in a mixed deciduous forest in Fukushima [5]. The high degree of spatial variability in the forest soils was thought to be because of heterogeneity of initial deposition, and then be strongly affected by the sporadic transport of ¹³⁷Cs in stemflow and throughfall [13]. The spatial variability then decreased because large amounts of ¹³⁷Cs migrated from the litter layers to the soils.

5. Conclusions

Our results suggest that large amounts of ¹³⁷Cs had migrated from the litter layers to the surface soils by 2013 at the study sites. In addition, the ¹³⁷Cs migrated from the litter layers much more quickly than could be explained by the rate at which litter decomposes in the study region. The decrease in spatial variability in the ¹³⁷Cs inventories in the soils in 2013 suggested that the ¹³⁷Cs in the soils had been predominantly supplied by stemflow and throughfall by 2013. The dominant source then gradually changed to the decomposition and leaching of the contaminated litter on the forest floors. At more than two years after the FDNPP accident at the present study sites, the differences in initial atmospheric deposition, snow coverage on the forest floors at the time of the deposition, and horizontal movement along slopes did not cause any clear differences in the downward migration patterns of ¹³⁷Cs in forest soils.

The results indicated that the study sites had reached the phase in which the ¹³⁷Cs in forest soils migrates very slowly (known as the quasi-equilibrium/relatively stable/steady-state phase [18,22,23]). The small amounts of ¹³⁷Cs that would have migrated during this phase could not be determined by the present study method because of the high degree of spatial variability in the ¹³⁷Cs inventories of the soils. A different method is, therefore, required to evaluate the future downward migration of ¹³⁷Cs in soils. One way of achieving this would be to use a lysimeter method that allows small amounts of ¹³⁷Cs migrating through the soil to be measured.

Acknowledgements

We sincerely thank the Iwaki District Forest Office for supporting our field survey. We also thank Prof. K. Shizuma at the Graduate School of Engineering, Hiroshima University for the many useful suggestions and support for sample measurements. This study was supported by JSPS KAKENHI [grant number 15J03548], and Phoenix Leader Education Program (Hiroshima Initiative) for Renaissance from Radiation Disaster, Organization of the Leading Graduate Education Program. The radioactivity measurements were performed at the radiation research facility of Faculty of Engineering, Hiroshima University.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This study was supported by JSPS KAKENHI [grant number 15J03548].