Vertical migration of radiocesium and clay mineral composition in five forest soils contaminated by the Fukushima nuclear accident

Abstract

In forest soils contaminated by radiocesium (¹³⁴Cs and ¹³⁷Cs), deposition from the Fukushima nuclear accident, clay minerals might play important roles in long-term cesium (Cs) dynamics through sorption. To determine whether radiocesium can be retained within the organic layer and the upper mineral soil layers in the Fukushima region, we investigated the vertical distribution of ¹³⁴Cs and ¹³⁷Cs and the clay mineral composition in five soil profiles of varying radiocesium deposition levels and vegetation types. X-ray diffraction analyses and oxalate extraction suggested that hydroxy-interlayered vermiculites and short-range-ordered aluminum (Al) and iron (Fe) compounds (i.e, allophane and ferrihydrite) were major clay mineral species of the upper soil layers. The vertical soil distribution of ¹³⁴Cs and ¹³⁷Cs suggested that most of them were retained in the organic layer and upper mineral soil layer under different levels of deposition. Within 1.5 years after the accident, both ¹³⁴Cs and ¹³⁷Cs were leached from the organic layer, and most of these (59–73%) were accumulated in the upper soil layer (0–5 cm). The proportion of ¹³⁷Cs (or ¹³⁴Cs) leaching from the organic layer was greater at sites receiving greater amounts of precipitation. The substantial accumulation of ¹³⁷Cs in the upper soil layer, irrespective of the ¹³⁷Cs deposition level or clay mineral composition, suggests that sorption capacities of clays and organic matter are sufficiently high to retain ¹³⁷Cs in the surface soil during at least the initial stage of contamination.

Key words:

• Cesium
• mica
• radiocaesium
• sorption
• volcanic soil

INTRODUCTION

The accident at the Fukushima Dai-ichi Nuclear Power Plant (FNPP), which resulted from the large tsunami after the March 11, 2011 Tohoku-Oki earthquake, led to the environmental contamination of radiocesium species (¹³⁴Cs and ¹³⁷Cs) (Yasunari et al. 2011; 6–36 × 10¹⁵ Bq from Adachi et al. 2013). Because ¹³⁴Cs and ¹³⁷Cs have relatively long half-lives (2.06 and 30.17 yr, respectively), their contamination has the potential to persist for years to decades (Schimmack et al. 1997; Hashimoto et al. 2013). Therefore, monitoring the initial deposition of ¹³⁴Cs and ¹³⁷Cs and their migration is crucial to forestry and agriculture for the prediction of short- and long-term dynamics within a forest ecosystem and transfer to downstream ecosystems.

Most of the radiocesium deposited onto the canopy and litter layers is predicted to translocate into the mineral soil within the forest ecosystem (Hashimoto et al. 2013). Monitoring of ¹³⁷Cs dynamics in Chernobyl forests has also shown that soils have been the dominant sink of ¹³⁷Cs during the decades following the accident (Konopleva et al. 2009). However, the cesium (Cs) retention capacities of soils do vary depending on climate, vegetation and soil types (Rosén et al. 1999). The larger amounts of precipitation can increase Cs leaching, especially in organic soils (Sigurgeirsson et al. 2005). The smaller ¹³⁷Cs retention capacities were also reported for the Fukushima forest soils rich in organic matter (Koarashi et al. 2012). The dynamics of radiocesium need to be studied for Fukushima soils, which are influenced by large amounts of precipitation and volcanic ash deposition (Shoji et al. 1985).

The retention of radiocesium in soil can occur mainly by sorption onto clays (Cremers et al. 1988). Due to the larger surface areas of finer soil particles, a dominant proportion of ¹³⁷Cs was reported to be accumulated within the clay fraction [e.g., 41% of ¹³⁷Cs in 12% clay from Livens and Baxter (1988); 69–93% of ¹³⁷Cs in 24–27% clay from Gerzabek et al. (1992)]. Because ¹³⁷Cs retention in the surface soil layers can increase with increasing clay content (Koarashi et al. 2012), clay content has been considered as one of important factors regulating Cs retention. Further, the laboratory studies revealed that clay mineral composition, especially abundance of the micaceous clay mineral (or illite), can also have influence on long-term ¹³⁷Cs stabilization (Konopleva et al. 2009; Giannakopoulou et al. 2012), although its effects on initial Cs dynamics have not been observed from the field survey (Matsunaga et al. 2013).

Of a variety of clay minerals, mica and its weathered products may play important roles in Cs retention in soil profiles due to a high layer charge of 2:1 phyllosilicate (Hird et al. 1996; Konopleva et al. 2009). Illite exhibits the highest selectivity of Cs sorption (Delvaux et al. 2001). Cs is selectively sorbed and fixed in wedge-shaped expanded zones of illite clay interlayers, called frayed edge sites (FES) (Cremers et al. 1988; Hird et al. 1996). FES account for a small proportion (ca. 1%) of the cation exchange capacity (CEC) but dominate the sorption of trace amounts of Cs (Cremers et al. 1988). Mica weathering products [especially vermiculite and hydroxy-interlayered vermiculite (HIV)] exhibit lower Cs selectivity than illite, however, they remain illitic in their core structure, and Cs fixation can occur on vermiculitic sites neighboring illitic wedge zones (Delvaux et al. 2001). The volcanic ash soils are typically poor in FES from illite clays, and the abundant clay mineral species, allophane, generally exhibit smaller Cs sorption capacities (Vandebroek et al. 2012). These factors, along with low pH, high organic matter content and high extent of aluminum (Al) interlayering (formation of HIV minerals), can be hypothesized to limit Cs retention in the surface layers of volcanic soils (Maes et al. 1999; Staunton et al. 2002; Nakao et al. 2008, 2009). The mobility of ¹³⁴Cs and ¹³⁷Cs in Fukushima forest soils might differ from that in Chernobyl grassland or forest soils derived from loess and glacial deposits.

In addition to being affected by chemical equilibria between Cs and clays, the depth distribution of radiocesium in soil profiles is regulated by various physical and biological processes, such as the accumulation and decomposition of organic matter, migration of particulate matter and recycling via litterfall (Nimis 1996; Kruyts and Delvaux 2002). Also, the migration of radiocesium from organic layers into the mineral soil can vary with vegetation type, depending on the thickness and decomposition rate of the organic layers (Andolina and Guillitte 1990; Rafferty et al. 2000). Larger proportions of ¹³⁷Cs were reported to be held within thick (moder-type) organic layers even 2 years after the Chernobyl accident (Andolina and Guillitte 1990). Because organic layers are usually thin in Japanese forests (Takahashi et al. 2010), the migration rates into the mineral soil layers of Cs might be different from those in Chernobyl grassland or forest soils.

In the present study, we investigated clay mineral composition and the initial vertical distribution of ¹³⁴Cs and ¹³⁷Cs in five forest soil profiles, which were selected based on the gradient of radiocesium deposition level and vegetation type. We sought to address three questions: (1) whether ¹³⁷Cs (or ¹³⁴Cs) could be retained within the soil profiles under different levels of radiocesium deposition, (2) whether ¹³⁷Cs deposited onto organic layers could migrate into mineral soil layers within the initial period following the accident and (3) whether the potential adsorbents of radiocesium, mica and its weathering products, are present or not in the Fukushima forest soils affected by volcanic materials.

MATERIALS AND METHODS

Experimental sites

We selected three study sites, Kawauchi, Ohtama and Tadami, along the deposition-level gradient (Fig. 1). Deposition of ¹³⁴Cs and ¹³⁷Cs at the forest sites amounted to 1020 kBq m⁻², 73 kBq m⁻², and 16 kBq m⁻² at Kawauchi, Ohtama and Tadami, respectively [Forestry and Forest Products Research Institute (FFPRI) 2011a, 2011b]. The Tadami site was located on a steep slope (29°), whereas the other sites were located on moderate slopes having an average inclination of 8° to 16°. The Kawauchi site [37°17’N, 140°47’E, 660 m above sea level (a.s.l.)] consisted solely of Japanese cedar plantation (KC plot; 43-year-old Cryptomeria japonica D. Don). The KC soil, derived from volcanic ash and granite, has been classified as Andisols (Soil Survey Staff 2006) or Brown forest soils (Forest Soil Division 1975; Kaneko et al. 2013). The Ohtama site (37°34’N, 140°18’E, 730–780 m a.s.l.) included three plots: one of pine forest (OP), one of mixed deciduous forest (OM), and one of cedar forest (OC). The OP plot was dominated by 43-year-old Pinus densiflora S. et Z., the OM plot was dominated by broad-leaved species such as Quercus serrata M. and Castanea crenata S. et Z. mixed with Pinus densiflora, and the OC plot was a 42-year-old Cryptomeria japonica plantation. The soils of the OP, OM and OC plots were derived from volcanic ash and andesite. The OP and OM soils have been classified as Andisols or Brown forest soils. The OC soil, in which a buried A horizon was observed in the subsoil (49–82 cm), was classified as an Andisol or Black soil. The Tadami site (37°19′N, 139°31′E; 680 m a.s.l.) consisted solely of cedar plantation (TC plot; 41-year-old Cryptomeria japonica). The TC soil, which was derived from volcanic ash and tuffaceous sandstone, has been classified as Inceptisols or Brown forest soils. The mean annual air temperature in the most recent 10 years at the study sites ranged from 10.3 to 12.2°C. Annual precipitation was 1546 mm yr⁻¹ at the KC site and 1261 mm yr⁻¹ at the Ohtama site, but it was considerably higher (2573 mm yr⁻¹) at the TC site due to large amounts of snowfall.

Figure 1 Locations of the experimental sites. FNPP: Fukushima Dai-ichi Nuclear Power Plant; KC: Kawauchi cedar site; TC: Tadami cedar site; OP, OM and OC: pine, mixed deciduous and cedar forests, respectively, of the Ohtama site.

Soil sampling

The inventory study of radiocesium (kBq m⁻²; aboveground biomass and soil) was carried out by FFPRI (2011a, 2011b). For measurement of soil radiocesium activity, both the organic layers and mineral soil layers were sampled from August to September of 2011 and 2012, i.e., 0.5 and 1.5 years after the accident, at all sites. Briefly, organic layers were sampled from 25-cm quadrates in 12 replicates. Surface soil layers were collected using a core sampler (diameter of 110 cm and height of 5 cm) in 12 replicates, whereas deeper soil layers (5–10, 10–15 and 15–20 cm) were collected in three replicates. For analyses of physicochemical and mineralogical properties, soil samples were collected from pedogenetic horizons. The data of physicochemical properties from pedogenetic horizons were converted to depth increments (0–5, 5–10, 10–15 and 15–20 cm). All soil samples were air-dried and crushed to pass through a 2-mm sieve and were analyzed for radiocesium activity and physicochemical and mineralogical properties.

Radiocesium activity in organic and mineral soil materials

Each soil sample (ca. 50 g) was compressed into a plastic container (100 mL), and the total concentration of ¹³⁴Cs and ¹³⁷Cs in the bulk soil of the core samples was determined using a high-purity Ge gamma-ray detector (GC2020-7500SL-2002CSL, CANBERRA, Meriden, USA) connected to a multichannel analyzer system. The stocks of ¹³⁴Cs and ¹³⁷Cs in the soil layers (0–5 cm, 5–10 cm, 10–15 cm and 15–20 cm) were calculated using the respective concentrations and bulk density of the soil. The downward fluxes of ¹³⁷Cs (or ¹³⁴Cs) into the organic layer and the mineral soil layer (0–5 cm) were calculated from the net increase of ¹³⁷Cs (or ¹³⁴Cs) in the underlying soil layers. To estimate the radioactive decay of the stocks, the remaining proportions of ¹³⁴Cs and ¹³⁷Cs at time t (yr) were calculated as follows:

(1)

where N(t) and N(0) represent the activities of radiocesium at time t (yr) and time 0, respectively, and λ represents the decay constant (λ of ¹³⁴Cs = 0.346 and λ of ¹³⁷Cs = 0.023).

Physicochemical properties of soils

Soil pH was measured using a soil to solution [H₂O or 1 M potassium chloride (KCl)] ratio of 1:5 (w/v) after shaking for 1 h. Total carbon (C) and nitrogen (N) contents in soils were determined using a CN analyzer (Thermo Fisher Scientific, Flash EA 1112). Particle size distribution was determined by a pipette method. The exchangeable basic cations and CEC were determined using the ammonium acetate (1 M and pH 7.0) method. The short-range-ordered Al, iron (Fe), and silicon (Si) compounds (Al_o, Fe_o, and Si_o) were determined by extraction in the dark with acid (pH 3) 0.2 M ammonium oxalate (McKeague and Day 1966). The concentrations of Al, Fe and Si in extracts were measured using inductively coupled plasma atomic emission spectrometry (ICP-AES, Thermo Fisher Scientific, iCAP6500Duo). The contents of allophane in the soil samples were estimated by multiplying Si_o with a factor (7.14), while those of ferrihydrite were estimated by multiplying Fe_o with a factor (1.7) (Parfitt and Henmi 1982; Parfitt and Childs 1988). The contents of crystalline clay minerals were estimated by subtracting the contents of allophane and ferrihydrite from the clay content, which was re-calculated on the basis of bulk soil.

Mineralogical properties of soils

The clay mineral composition of selected soil samples was examined using the X-ray diffraction (XRD) method on parallel-oriented clay (< 2 μm) specimens subjected to magnesium (Mg) saturation, potassium (K) saturation, glycerol solvation and heat treatment (300°C, 550°C) for K-saturated specimens. XRD patterns were obtained using an X-ray diffractometer (Rigaku, RINT 2000) with CuKα radiation. The relative contents of illite, vermiculite (Vt) and quartz (Qz) were estimated based on the following formulae (Matsue and Wada 1988, 1989):

(2)

(3)

(4)

where I_1.0 and I_1.4 are the XRD intensities (peak heights) of K- and Mg-saturated and air-dried clay samples at 1.0 and 1.4 nm, respectively. The other 1.4-nm clay minerals, HIV, have been estimated from the 1.4-nm peak height of K-saturated clay samples when chlorite (Ch) was negligible (Matsue and Wada 1989; Ohta et al. 1993). Because small amounts of chlorite were detected in our study, the relative contents of HIV and Ch were estimated from the 1.4-nm peak height of K-saturated clay samples. The degrees of Al interlayering were estimated using the intensity ratio of HIV and Ch to 1.4-nm clay minerals (Matsue and Wada 1988):

(5)

(6)

The sum of relative contents of Illite, Vt, HIV and Ch equals 1.

Calculations and statistics

All results were expressed on an oven-dried (24 h, 105°C) weight basis and as means ± standard error (SE) of 12 determinations for organic layers, nine determinations for surface soil layers (0–5 cm) and three determinations for deeper soil layers (5–10, 10–15 and 15–20 cm). The significance of differences in mean values between groups (horizons, sites and 2011 vs. 2012) were tested using an analysis of variance (ANOVA) at the p < 0.05 significance level for Cs concentrations and stocks. Spearman’s linear correlations were used to assess the relationships between Cs loading and soil Cs stock, and between Cs loading and soil Cs fluxes. The statistical analyses were performed using SYSTAT ver. 11.0 (SPSS Inc., 2008).

RESULTS

Soil physicochemical properties

The soils were moderately acidic (pH of 4.4 to 5.2), consistent with low base saturation (Table 1). Exceptionally, base saturation [especially exchangeable calcium (Ca)] was high in the surface layers of the OC soil due to incorporation of Ca-rich cedar litter. Soil C contents were high from the top to the 15-cm depth at all sites except for TC soil, which was derived from sandstone. The fine fraction (clay + silt) was high at all sites (45 to 75%), but it was lower at the TC site than at the other sites, probably due to the influence of the bedrock (tuffaceous sandstone) (Table 2). Soil C contents were positively correlated with fine (clay + silt) fractions (r = 0.68, p < 0.01, n = 15). Variable but high contents of Si_o indicated that most of the soil horizons have been influenced by volcanic materials (Table 1). The contents of oxalate-extractable Al and Fe compounds (Al_o + 1/2 Fe_o) satisfied the criterion of andic soil properties (> 20 g kg⁻¹ soil) throughout the profiles of all sites, except for the A horizon of the KC soil and the TC soil profile (Table 1). The contents of Al_o + 1/2Fe_o in the TC soil were low due partly to surface erosion on the steep slope.

Table 1 Physicochemical properties of the five forest soils

										Exchangeable basic cation					Oxalate extractable^‡
		Depth	pH		Total C	Total N		Bulk density	CEC	Na	K	Mg	Ca	BS^†	Al	Fe	Si	Alo + 1/2Feo^†
Site	Horizon	(cm)	H2O	KCl	(g kg^−1)		C:N ratio	(g cm^−3)	(cmolc kg^−1)					(%)	(g kg^−1)			(g kg^−1)
KC	O	+2–0
	A1	0–5	4.7	4.0	173.5	9.9	17.6	0.4	40.8	0.1	0.4	1.1	5.0	15.5	12.9	10.8	1.0	18.3
	A2	5–10	4.9	4.2	127.0	6.8	18.6	0.5	28.3	0.1	0.3	0.6	1.1	7.3	17.7	12.3	1.7	23.9
	A2	10–15	4.9	4.3	114.6	6.1	18.6	0.6	25.7	0.1	0.3	0.5	0.9	6.4	18.6	12.5	2.2	24.8
	A3	15–20	5.0	4.6	65.0	3.4	18.8	0.6	15.1	0.1	0.1	0.1	0.2	3.0	22.0	13.2	4.2	28.6
OP	O	+2–0
	A1	0–5	5.1	4.4	141.0	7.7	18.4	0.4	31.9	0.1	0.4	0.6	3.0	12.8	17.0	9.9	2.3	22.0
	A2	5–10	5.1	4.4	141.0	7.7	18.4	0.6	31.9	0.1	0.4	0.6	3.0	12.8	17.0	9.9	2.3	22.0
	A3	10–15	5.1	4.6	110.7	6.1	17.9	0.5	26.1	0.1	0.3	0.5	2.2	10.3	20.2	10.7	4.1	25.6
	B	15–20	5.2	5.1	40.0	2.4	16.7	0.5	12.6	0.1	0.1	0.1	0.3	4.4	27.7	12.5	8.2	34.0
OM	O	+2–0
	A1	0–5	5.1	4.1	181.0	9.8	18.4	0.4	45.7	0.1	0.3	0.6	2.3	7.2	18.7	10.1	2.9	23.8
	A2	5–10	5.0	4.4	118.6	6.8	17.5	0.5	30.7	0.1	0.2	0.4	0.9	5.0	24.3	11.7	5.1	26.8
	A3	10–15	5.0	4.7	75.5	4.6	16.2	0.6	20.3	0.1	0.2	0.2	0.5	4.7	28.6	13.1	7.6	32.3
	AB	15–20	4.9	4.9	48.0	3.1	15.1	0.6	13.7	0.1	0.1	0.1	0.4	4.9	31.4	14.1	9.5	38.5
OC	O	+2–0
	A1	0–5	5.0	4.8	249.0	14.5	17.2	0.3	74.4	0.2	0.6	3.4	47.4	69.4	16.8	8.2	2.1	20.9
	A2	5–10	5.0	4.8	249.0	14.5	17.2	0.4	74.4	0.2	0.6	3.4	47.4	69.4	16.8	8.2	2.1	20.9
	A3	10–15	4.8	4.4	111.3	6.3	17.6	0.5	48.7	0.1	0.2	1.0	11.5	23.6	27.1	11.5	4.2	32.8
	A4	15–20	4.9	4.6	149.0	8.7	17.2	0.6	23.8	0.1	0.1	0.1	1.0	5.2	27.9	12.9	5.6	34.4
TC	O	+2–0
	A1	0–5	4.4	4.1	90.0	4.7	19.0	0.3	19.0	0.1	0.2	0.2	0.5	5.1	9.2	6.1	0.9	12.3
	A2	5–10	4.4	4.1	90.0	4.7	19.0	0.5	19.0	0.1	0.2	0.2	0.5	5.1	9.2	6.1	0.9	12.3
	B1	10–15	5.0	4.4	17.0	1.1	14.9	0.6	8.6	0.1	0.2	0.1	0.2	6.5	12.8	4.5	3.2	15.1
	B2	15–20	5.0	4.4	16.1	1.1	14.3	0.7	8.6	0.1	0.2	0.1	0.2	6.8	7.7	4.0	1.4	9.7

Oven-dry basis. A proportion of data were cited from Kaneko et al. (2013). ^†BS represents base saturation calculated using exchangeable base and CEC. ^‡Poorly crystalline Al and Fe oxides and organic Al and Fe compounds.

Clay mineralogical properties

Allophane and ferrihydrite accounted for a significant proportion of the total clay fractions for all of the soil samples (Table 2). Especially in the deeper soil layers of Ohtama three plots, allophane and ferrihydrite accounted for high proportions (23 to 55%) of clay fractions. The clay mineral species are dominated by crystalline clay minerals in the upper mineral soil layers at all plots (Table 2).

Table 2 Physical and mineralogical properties of the five forest soils

			Particle size distribution^†					Crystalline
		Depth	Clay	Silt	Sand	Allophane^‡	Ferrihydrite^‡	minerals^‡	Relative content of mineral^§
Site	Horizon	(cm)	(%)			(%)			Qz	Illite	HIV + Ch	Vt	Extent of interlayering
Kawauchi cedar (KC)	A1	0–4	20	41	39	0.6	1.8	11.3	0.02	0.03	0.95	0.03	0.97
	A2	4–18	19	36	46	1.2	2.1	11.0	0.05	0.04	0.96	0.01	0.99
	A3	18–40	19	27	55	3.0	2.2	11.2	0.06	0.02	0.95	0.02	0.98
Ohtama pine (OP)	A	0–17	24	40	37	1.6	1.7	14.3	0.16	0.04	0.94	0.01	0.98
	B	17–30	24	37	40	5.9	2.1	13.9	0.13	0.05	0.95	0.00	1.00
	2A	30–48	26	35	40	3.3	2.0	19.1	0.09	0.09	0.87	0.01	0.99
Ohtama mixed deciduous (OM)	A1	0–6	22	45	34	2.1	1.7	10.8	0.14	0.05	0.87	0.11	0.89
	A2	6–15	22	40	39	4.0	2.1	11.6	0.13	0.09	0.94	0.00	1.00
	AB	15–32	18	36	46	6.8	2.4	7.5	0.13	0.06	0.91	0.02	0.98
Ohtama cedar (OC)	O/A	0–11	20	55	25	1.5	1.4	8.1	0.13	0.06	0.90	0.02	0.98
	A	11–30	20	47	33	3.1	2.0	11.2	0.15	0.06	0.93	0.01	0.99
	AB	30–49	19	40	41	4.0	2.2	7.6	0.11	0.07	0.92	0.00	1.00
Tadami cedar (TC)	A	0–10	17	18	65	0.6	1.0	12.7	0.05	0.05	0.90	0.06	0.94
	B1	10–29	17	17	66	2.3	0.8	13.8	0.05	0.05	0.82	0.14	0.85
	B2	29–48	17	12	71	0.9	0.7	15.0	0.08	0.04	0.86	0.10	0.89

Oven-dry basis. Note that the data of mineralogical properties were obtained for different pedogenetic horizons. ^†Clay (< 0.002 mm); silt (0.002–0.05 mm); sand (0.050–2 mm). ^‡Allophane content (%) = Si_o content (%) × 7.14; ferrihydrite content (%) = Fe_o content (%) × 1.7 (bulk soil basis); crystalline mineral content (%) = clay content (%) × (100 – total soil C (%) × 1.8)/100 – allophane content (%) – ferrihydrite content (%). ^§Qz = quartz, HIV = hydroxy-interlayered vermiculite, Vt = vermiculite, Ch = chlorite. The sum of illite, HIV, Vt and Ch (other than Qz) equals to 1.

According to XRD patterns of clay fractions, the illite peak was identified from the reflection at 1.0 nm (Fig. 2), but the relative contents of illite in our study were small for all of the clay samples (Table 2). The absence of a peak at 1.8 nm after glycerol solvation suggests that smectite was negligible for all of the samples (not shown). The chlorite peak was identified from the 1.4-nm reflection when the K-saturated clay samples were heated at 550°C (Fig. 2). At the same time, the intensities of the 1.2-nm reflection, which corresponds to HIV, increased after heating (550°C) of the K-saturated clay samples at the expense of the 1.4-nm reflection at 20 and 300°C (not shown). The chlorite peak was consistently small and the relative contents of HIV and chlorite corresponded predominantly to HIV for all the samples studied (Table 2). Quartz was identified from reflections of 0.43 nm (Fig. 2) and 0.33 nm (not shown). The decrease of the 1.4-nm reflection after K saturation was small, and the degree of Al interlayering was consistently high for the clay samples studied. The degree of Al interlayering was positively correlated with the contents of Al_o (r = 0.61, p < 0.01, n = 15) and Al_o + 1/2Fe_o (r = 0.66, p < 0.01, n = 15) (Tables 1 and 2). Accordingly, the relative contents of vermiculite were negatively correlated with these parameters (Tables 1 and 2).

Figure 2 X-ray diffraction patterns of clay fractions from the surface soil layers of three cedar sites of Kawauchi, Ohtama, and Tadami (KC, OC and TC). “Mg-” and “K-” represent magnesium saturation and potassium saturation, respectively, and the following numbers, “20” or “550”, are the temperatures (°C) at which the samples were dried or heated.

Initial depth distribution and migration of radiocesium within the soil profiles

In 2011, the concentrations of radiocesium (¹³⁴Cs and ¹³⁷Cs) in the soils increased toward FNPP in the following order: KC > OP, OM, OC > TC. At all sites, the concentrations of both ¹³⁴Cs and ¹³⁷Cs were largest in the organic layer and decreased with depth (Table 3). The concentrations of ¹³⁴Cs and ¹³⁷Cs decreased significantly (p < 0.05) in the organic layers from 2011 to 2012, while they increased significantly (p < 0.05) for the upper mineral soil layer (0–5cm). Significant differences in ¹³⁷Cs (or ¹³⁴Cs) concentrations between 2011 and 2012 could not be observed for the deeper soil layers (below 5 cm); Cs concentrations were highly variable in these layers.

Table 3 Depth distribution of ¹³⁴Cs and ¹³⁷Cs in the soil profiles

			^137Cs		^134Cs
Site	Horizon	Depth	2011^†	2012^†	2011^†	2012^†
		(cm)	(Bq kg^−1 soil)		(Bq kg^−1 soil)
Kawauchi cedar (KC)	O	+2–0	153770 ± 11757a	59850 ± 9324b	175430 ± 13497a	94551 ± 14878b
	A1	0–5	10151 ± 3611a	19753 ± 3599b	11945 ± 3509a	31448 ± 5732b
	A2	5–10	299 ± 121a	1245 ± 426a	405 ± 134a	1965 ± 658a
	A2	10–15	97 ± 20a	280 ± 106a	140 ± 19a	465 ± 177a
	A3	15–20	54 ± 20a	137 ± 90a	68 ± 21a	231 ± 153a
Ohtama pine (OP)	O	+2–0	21930 ± 1134a	6857 ± 384b	25922 ± 1346a	10870 ± 585b
	A1	0–5	564 ± 129a	1263 ± 243b	741 ± 146a	2103 ± 395b
	A2	5–10	104 ± 70a	104 ± 12a	158 ± 90a	191 ± 19a
	A3	10–15	54 ± 24a	35 ± 3a	65 ± 22a	81 ± 2a
	B	15–20	44 ± 17a	19 ± 5a	42 ± 16a	36 ± 11a
Ohtama mixed deciduous (OM)	O	+2–0	24800 ± 1533a	6605 ± 480b	28973 ± 1730a	10634 ± 808b
	A1	0–5	646 ± 164a	1646 ± 388b	809 ± 182a	2692 ± 615b
	A2	5–10	42 ± 11a	30 ± 6a	64 ± 13a	70 ± 4a
	A3	10–15	30 ± 19a	83 ± 42a	35 ± 19a	160 ± 81a
	AB	15–20	9 ± 3a	5 ± 2a	11 ± 4a	9 ± 5a
Ohtama cedar (OC)	O	+2–0	12348 ± 893a	5282 ± 408b	14307 ± 1020a	8361 ± 617b
	A1	0–5	643 ± 106a	1847 ± 418b	826 ± 131a	3001 ± 657b
	A2	5–10	75 ± 40a	38 ± 11a	112 ± 40a	107 ± 22a
	A3	10–15	11 ± 3a	14 ± 7a	17 ± 3a	41 ± 23b
	A4	15–20	5 ± 1a	13 ± 4b	6 ± 0a	32 ± 13b
Tadami cedar (TC)	O	+2–0	2670 ± 507a	1523 ± 279b	3187 ± 600a	2431 ± 482a
	A1	0–5	152 ± 36a	308 ± 78b	234 ± 44a	585 ± 133b
	A2	5–10	19 ± 9a	9 ± 4a	28 ± 9a	31 ± 10a
	B1	10–15	6 ± 0a	8 ± 4a	6 ± 1a	15 ± 8a
	B2	15–20	9 ± 2a	6 ± 2a	10 ± 3a	8 ± 4a

Oven-dry basis.^†All results are the mean ± standard error [n = 12 for organic layer and soil (0–5 cm), n = 3 for the deeper soil layers]. The different letters (a, b) indicate that mean values of each horizon are significantly different (p < 0.05) between sites.

Because the vertical distribution patterns for ¹³⁴Cs and ¹³⁷Cs were similar, the ¹³⁷Cs stocks in the soil are mainly discussed hereafter. The total ¹³⁷Cs stock in the soil (August 2011, organic layer to 20 cm depth) varied widely from 12 kBq m⁻² (TC) to 558 kBq m⁻² (KC) (Table 4). Regarding the ¹³⁷Cs concentration, the total stock of ¹³⁷Cs in the soil increased toward FNPP. Within the soil profiles, ¹³⁷Cs stocks in 2011 were largest in the organic layers at all sites. When the total soil ¹³⁷Cs stock was regarded as the “¹³⁷Cs loading” by neglecting the pre-accident ¹³⁷Cs stock, the ¹³⁷Cs stocks in the organic layer corresponded to 44 to 64% of the ¹³⁷Cs loadings at the point of August 2011 (after the initial 0.5 years from the accident).

Table 4 Percentages of ¹³⁷Cs retention in the soil compartments relative to ¹³⁷Cs loading

Site	Date (Time after the accident)	Total stock of ^137Cs in soil^†	Organic layer^‡	Mineral soil 0–5 cm^‡	Mineral soil 5–20 cm^‡
		(kBq m^−2)	(%)
Kawauchi dedar (KC)	2011 (0.5 year)	384	55 ± 8A	41 ± 12A	4 ± 1B
	2012 (1.5 year)	589	18 ± 4b	72 ± 14a	10 ± 3a
		(+205)
Ohtama pine (OP)	2011 (0.5 year)	44	56 ± 7A	29 ± 6B	15 ± 5A
	2012 (1.5 year)	61	27 ± 4a	60 ± 11a	13 ± 2a
		(+17)
Ohtama mixed deciduous (OM)	2011 (0.5 year)	41	60 ± 10A	33 ± 8AB	6 ± 1B
	2012 (1.5 year)	62	21 ± 4ab	69 ± 15a	10 ± 3a
		(+21)
Ohtama cedar (OC)	2011 (0.5 year)	38	62 ± 7A	30 ± 5B	7 ± 2B
	2012 (1.5 year)	50	26 ± 5ab	68 ± 14a	7 ± 1a
		(+12)
Tadami cedar (TC)	2011 (0.5 year)	8	50 ± 8A	37 ± 6A	13 ± 3A
	2012 (1.5 year)	14	18 ± 4b	69 ± 17a	13 ± 4a
		(+6)

^†The sum of the ¹³⁴Cs and ¹³⁷Cs stocks from the organic layer to 20 cm depth. ^‡All results are expressed as the mean ± standard error. The mean values of each layer were compared between sites for 2011 and 2012, respectively. The different letters (A, B for 2011; a, b for 2012) indicate the significant difference (p < 0.05) between sites.

After the subsequent 1 year (August 2011 to August 2012), the total ¹³⁷Cs stock in the soil (organic layer to 20 cm depth) increased significantly (p < 0.05) due to radiocesium input from throughfall and litterfall (Table 4). However, the ¹³⁷Cs stock in the organic layer decreased significantly (p < 0.05) from 2011 to 2012 at all sites (Fig. 3). The proportions of ¹³⁷Cs stocks relative to their loadings decreased from 44–64% in 2011 to 14–27% in 2012 (Table 4). When ¹³⁷Cs retention in the organic layer was compared among the five sites, the remaining proportion of ¹³⁷Cs in the organic layer was lower at the KC and TC sites, which received higher amounts of precipitation (1546 and 2573 mm yr⁻¹) than the three Ohtama sites. For the three Ohtama sites, the remaining proportion of ¹³⁷Cs in the organic layer did not vary significantly between vegetation types (Table 4).

Figure 3 Vertical distribution of radiocesium in the five soil profiles from 2011 to 2012. Bars represent standard errors [n = 12 for organic layer and soil (0–5 cm); n = 3 for the deeper soil layers].

When the radioactive decay of the radiocesium stocks after 1 year (August 2011 to August 2012) was calculated using Eq. 1, the stocks of ¹³⁴Cs and ¹³⁷Cs would decrease by 29% and 2%, respectively. In the case of the upper 0–5-cm soil layer of KC, the decrease of ¹³⁷Cs stock by radioactive decay amounted to 4 kBq m⁻². Despite this, the ¹³⁷Cs stock in the upper soil layer increased significantly (p < 0.05) from 159 kBq m⁻² (2011) to 422 kBq m⁻² (2012). Similarly, the ¹³⁴Cs stock also increased from 135 to 265 kBq m⁻², despite greater radioactive decay (40 kBq m⁻²). In 2012, the ¹³⁴Cs and ¹³⁷Cs stocks of the upper soil layer exceeded those of the organic layer. This trend was observed at all sites, although the magnitude tended to decrease with increasing distance from FNNP (Fig. 3). No significant increase in ¹³⁷Cs (or ¹³⁴Cs) stocks was observed for the deeper soil layers (Fig. 3). The ¹³⁷Cs retention in the upper soil layer (0–5 cm) increased from 29–41% (2011) to 60–72% (2012) of the ¹³⁷Cs loading (Table 4). The percentages of ¹³⁷Cs (or ¹³⁴Cs) retention in the upper soil layers had no correlation with the relative contents of clay minerals, the contents of allophane and clay percentages, respectively (Tables 2 and 4).

DISCUSSION

Characteristics of clay mineral composition in Fukushima soils

The clay mineralogy of forests soils in the Fukushima region is strongly influenced by volcanic ash deposition. The results of XRD and oxalate extraction showed that the major clay mineral species in our study were HIV and short-range-ordered Al and Fe compounds (i.e., allophane and ferrihydrite), irrespective of the variation of bedrock (Fig. 2; Table 2). The relative contents of illite in our study were lower than those in soils from sedimentary rocks (0.19 to 0.85; Matsue and Wada 1989; Ohta et al. 1993) due partly to the lower contents of illite in volcanic materials. However, it should be noted that abundance of illite could not be evaluated solely from XRD patterns (Thompson and Ukrainczyk 2002). The greater degree of Al interlayering in the soils from our study (Table 2), compared to other soil types (0.12–0.38 from Pai et al. 2004), is consistent with the finding that HIV is the most abundant clay mineral species of nonallophanic Andisols in the Fukushima region (Shoji et al. 1985). This can be supported by (1) less acidic soil conditions (pH of 4.6 to 5.8 from Rich 1968; Bain et al. 1990), (2) low concentrations of chelating organic acids (Malcolm et al. 1969; Fujii et al. 2008, 2010), and (3) a supply of Al from short-range-ordered Al compounds. Regarding the third point above, the high amount of hydroxyl-interlayered vermiculite requires sources of micaceous minerals and interlayer Al compounds. Micas and chlorite can be supplied from bedrock and aeolian dust (Bautista-Tulin and Inoue 1997), whereas interlayer Al compounds can be supplied from short-range-ordered Al compounds (Ndayiragije and Delvaux 2003; Yagasaki et al. 2006). The high amounts of Al released from short-range-ordered Al compounds are favorable for Al interlayering of vermiculite, as suggested by the positive correlation between the degree of Al interlayering and Al_o contents (Table 2).

Effects of deposition levels on the distribution of radiocesium in Fukushima soil profiles

The temporal fluctuation of radiocesium in soils after the accident needs to be discussed along with the pre-accident stocks and the time-dependent decrease caused by radioactive decay (Almgren and Isaksson 2006). The pre-accident concentration of ¹³⁷Cs, which was derived largely from atmospheric nuclear tests and the Chernobyl accident, has been reported to be ~48 Bq kg⁻¹ in Japanese forest soils (Takenaka et al. 1998; Yamaguchi et al. 2012). Assuming a similar pre-accident fallout level at our sites, most of the measured radiocesium would have originated from FNPP (Table 3). Accordingly, the total stock of ¹³⁷Cs in the soil profile (organic layer to 20-cm depth), which can be regarded as the ¹³⁷Cs loading to the soil, increased with decreasing distance from FNNP (Table 4).

About 50–62% of the ¹³⁷Cs loading were retained in the organic layer after the initial 0.5 years (March 2011 to August 2011). The stock of ¹³⁷Cs (or ¹³⁴Cs) in the organic layer in 2011 increased with increasing ¹³⁷Cs loading (deposition level) (Table 4). This suggests deposition of ¹³⁷Cs from FNPP and its accumulation in the organic layer. After the following 1 year (August 2011 to August 2012), ¹³⁷Cs stock in the organic layer decreased to 14–27% of the ¹³⁷Cs loading at all sites. The significant loss of ¹³⁷Cs in the organic layer could be caused by both radioactive decay and leaching (Nakanishi et al. 2014). Judging from the facts that (1) the 1-year radioactive decay can account for only a 29% loss of ¹³⁴Cs and a 2% loss of ¹³⁷Cs and (2) the dominant sink of ¹³⁴Cs and ¹³⁷Cs changed from the organic layer to the upper soil layer (Fig. 3), the loss of radiocesium from the organic layers probably resulted largely from leaching.

The proportions of ¹³⁴Cs and ¹³⁷Cs stock in the mineral soil layer relative to their loading increased from 29–43% in 2011 to 59–73% in 2012 (Table 4). Radioactive decay could cause a decrease in the ¹³⁴Cs and ¹³⁷Cs stocks by 29% and 2%, respectively, from 2011 to 2012; however, these increased in the upper 0–5-cm soil layers (Fig. 3). The significant gain of radiocesium in this layer within the 1 year could have resulted from an overwhelming magnitude of leaching from the organic layer. The changes in the ¹³⁴Cs and ¹³⁷Cs stocks between 2011 and 2012 indicated input of radiocesium from aboveground biomass (canopy) and organic layer and their accumulation in the mineral soil (Fig. 3). Although the downward flux of ¹³⁷Cs (or ¹³⁴Cs) from the organic layer increased with ¹³⁷Cs loading, most of the flux was retained in the upper 0–5-cm soil layer irrespective of deposition level (Fig. 3; Table 4). The markedly low mobility of Cs relative to the water flux suggests that Cs mobility is strongly limited by chemical processes (i.e., sorption), rather than physical or biological processes. Even with a high ¹³⁷Cs loading, the upper layer of the KC soil exhibited high efficiency of retention without approaching saturation (Table 4). Although Andisols typically exhibit low sorption capacity of ¹³⁷Cs (Vandebroek et al. 2012), a variety of retention sites and processes in the volcanic soils can contribute to ¹³⁷Cs retention (Joussein et al. 2004; Sigurgeirsson et al. 2005), as is discussed later. The ¹³⁷Cs accumulation within the upper 5-cm soil layer in our study (Fig. 3) is consistent with reports for grasslands and forests in Germany and agricultural lands in Sweden, where most of the ¹³⁷Cs was retained even 7 to 20 years after the Chernobyl accident (Schimmack et al. 1997; Almgren and Isaksson 2006; Konopleva et al. 2009). Although the greater precipitation and volcanic soils at our sites are different from the conditions of their studies, low mobility of ¹³⁷Cs might be common to soils rich in sorption and fixation sites, which contrasts with the rapid migration of ¹³⁷Cs observed in Histosols or Spodosols (Rosén et al. 1999; Sigurgeirsson et al. 2005).

Role of organic layer thickness in ¹³⁷Cs retention

Both ¹³⁴Cs and ¹³⁷Cs exhibited similar patterns of initial migration and retention in our study; however, an increasingly large gap will occur due to their differences in half-life. ¹³⁷Cs retention within the soil needs to be focused to predict long-term dynamics. The factors regulating ¹³⁷Cs retention and migration in the organic and mineral soil layers are discussed in the present and following sections to answer the questions (1) and (2) raised in the introduction.

Because ¹³⁷Cs retention capacities in organic layers are typically small compared to those in mineral soil layers (Shand et al. 1994), the migration of ¹³⁷Cs into mineral soil layers is hypothesized to be rapid in the case of a thin organic layer. In accordance with this, ¹³⁷Cs retention in the organic layer was consistently small 1.5 years after the accident (Table 4). The initial vertical migration of ¹³⁷Cs from the organic layer is driven by (a) initial washout and (b) decomposition and leaching. Regarding (b), the mass of the organic layer is mostly small in Fukushima forest soils (Kaneko et al. 2013), suggesting a rapid turnover of litter. The rapid loss of ¹³⁷Cs (or ¹³⁴Cs) from the organic layer is considered to be caused by rapid wash-out of deposited ¹³⁷Cs from the thin organic layer and rapid decomposition of litter, which can mobilize ¹³⁷Cs associated with organic matter (Nakanishi et al. 2014). Small retention capacities of ¹³⁷Cs were also reported from a thin (mull-type) organic layer, but this contrasts with the greater retention (60 to 75%) in a thick organic layer (moder-type; 6 cm thick) 2 to 5 years after the Chernobyl accident (Andolina and Guillitte 1990; Rafferty et al. 2000). Thick organic layers can retain higher amounts of ¹³⁷Cs due to sorption onto clay minerals admixed into humified internal layers (Shand et al. 1994). In our study, larger proportions of ¹³⁷Cs were leached from the organic layer at the TC and KC sites, which received greater precipitation than the OC site (Table 4), highlighting the importance of precipitation as a key factor regulating ¹³⁷Cs leaching flux from the organic layer. Small Cs retention and rapid leaching appear to be common in “mull-type” organic layers in this region, irrespective of vegetation species.

Potential importance of clay minerals in soil ¹³⁷Cs dynamics

The clay content and clay mineral composition can influence ¹³⁷Cs retention and stabilization process (Rosén et al. 1999; Konopleva et al. 2009). In our study, irrespective of clay properties or ¹³⁷Cs deposition levels, ¹³⁷Cs retention in the surface soil layer (59 to 73% of ¹³⁷Cs deposition) was consistently high at all sites (Table 4). Although volcanic soils typically exhibit small ¹³⁷Cs fixation capacities (Vandebroek et al. 2012), the upper mineral soil layers appeared to have retention capacities of ¹³⁷Cs (Table 4). The possible processes of ¹³⁷Cs retention in soil are fixation on the vermiculitic sites associated with illitic wedge zones and sorption on exchangeable sites derived from organic matter and clay minerals (Chorover et al. 1999; Matsunaga et al. 2013). The importance of mica and its weathering products (illite, vermiculite and HIV) in ¹³⁷Cs stabilization is supported by the positive correlations between FES and abundance of illite and mixed-layer minerals (Cremers et al. 1988), vermiculite (Delvaux et al. 2001) and illite (Nakao et al. 2014). The abundant 2:1 phyllosilicate mineral species in our study, HIV, generally exhibits considerably smaller fixation capacities (Maes et al. 1999; Nakao et al. 2009), but it can retain ¹³⁷Cs through fixation in FES situating at the fringe of non-interlayered structures. The fixation of ¹³⁷Cs within illite and vermiculite could not be supported in our study due to their small XRD peak intensities (Fig. 2; Table 2); however, this might be due to the limitations of XRD detection (Thompson and Ukrainczyk 2002). In addition, illite in minor abundance, which is not detectable from XRD patterns, may possess sufficient fixation capacities relative to trace amounts of ¹³⁷Cs (Hird et al. 1995). Allophane, which is another major constituent of clay minerals in our study (Table 2), can also possess low but significant amounts of FES (De Koning et al. 2007). Thus, these clay minerals have the potential to be ¹³⁷Cs reservoirs in the volcanic soils studied. The high CEC from organic matter and clays can also contribute to ¹³⁷Cs retention in the surface soil layers (Table 1). Based on these findings, the high ¹³⁷Cs retention in our study can be explained in part by the presence of selective sorption sites from a variety of clays (e.g., HIV, illite, allophane) and exchangeable sites from clays and organic matter in all surface soil samples (Fig. 3; Table 2), although quantitative importance of the individual processes remains to be evaluated.

When the amount of ¹³⁷Cs retention and the number of Cs-specific sorption sites are compared at the KC site, the ¹³⁷Cs concentration (31,448 Bq kg⁻¹) in the 0–5-cm layer is calculated to be 7.2 × 10⁻⁸ mmol kg⁻¹ (1 Bq ¹³⁷Cs ≡ 2.3 × 10⁻¹⁵ mol). The number of ¹³⁷Cs-specific sorption sites is reported to range from 0 to 1 × 10⁻⁵ mmol kg⁻¹ in the organic layer (Hird et al. 1996; De Koning et al. 2007) and from 3.0 × 10⁻² to 2.8 mmol kg⁻¹ soil in the mineral soil layer (Cremers et al. 1988; Hird et al. 1996). These data suggest that the sorption capacity of clays could be sufficiently high to fix ¹³⁷Cs in even highly contaminated soil. However, our data showed only that most of the radiocesium can be retained in the surface soil layer during at least the initial stage of contamination. Histosol and sandy Spodosol soils devoid of FES were reported to exhibit a deeper distribution of ¹³⁷Cs eight years after the Chernobyl accident (Rosén et al. 1999).

To predict the long-term ¹³⁷Cs dynamics in Fukushima forests, long-term monitoring and evaluation of FES are required for the volcanic soils. In addition, there are several uncertainties regarding the contribution of fixation to whole ¹³⁷Cs retention, fixation/desorption of mobile ¹³⁷Cs with aging, and mobility of ¹³⁷Cs (e.g., physical migration of ¹³⁷Cs-fixed particles by preferential water flow, erosion and mixing by soil animals) in volcanic soils. With regard to ¹³⁷Cs cycling between plant and soil, as the organic layer acts as a source of mobile ¹³⁷Cs for plants (Kruyts and Delvaux 2002), the migration of ¹³⁷Cs into the mineral soil layer is expected to decrease the risks of ¹³⁷Cs transfer to plants in our study. An understanding of the mobility and bioavailability of ¹³⁷Cs in relation to retention forms for volcanic soils is also needed.

CONCLUSIONS

All of the soil horizons contained hydroxy-interlayered vermiculites and short-range-ordered Al and Fe compounds including allophane and ferrihydrite. The deposited ¹³⁷Cs was leached from the organic layer, and most of the ¹³⁷Cs accumulated in the upper soil layer within 1.5 years after the accident. A larger proportion of ¹³⁷Cs was leached from the organic layers at sites receiving greater amounts of precipitation. Accumulation of ¹³⁷Cs in the upper 0–5 cm soil layer, irrespective of ¹³⁷Cs loading and clay mineral composition, suggests that the sorption capacities of clays and organic matter are sufficiently high to retain ¹³⁷Cs even in the upper mineral layers. However, it should be noted that the results obtained during the initial stage of contamination could not be extrapolated for long-term ¹³⁷Cs dynamics; therefore, long-term monitoring is warranted.

ACKNOWLEDGMENTS

This study was conducted as an FFPRI project study (F1P07) and was also funded by the Forestry Agency. We would like to thank the Kanto Regional Forest Office and the Rural Development Division of Kawauchi Village for allowing us to conduct studies at the experimental sites. We are also grateful to the members of FFPRI for the field sampling and measurements.