Introduction

The accident at the Fukushima Daiichi nuclear power plant (NPP), triggered by a catastrophic earthquake (M9.0) and resulting tsunami on March 11, 2011, caused serious radioactive contamination over a wide area of eastern Japan1. Of the radionuclides found in the atmospheric fallout from the accident, 137Cs with a physical half-life of 30.1 years is the largest source of concern because of its potential impact on humans and ecosystems over the coming decades.

Among the terrestrial ecosystems, forest ecosystems have received the most attention because a majority (~70%) of the land area contaminated by the accident is covered by forests2. Forest ecosystems consist of tree biomass (aboveground: boles, branches and leaves; belowground: roots), forest-floor litter (fallen dead leaves and branches and their decomposed materials) and underlying mineral soil2. Fukushima-derived 137Cs was first deposited on forest-floor litter directly3,4, or intercepted by forest canopies and subsequently deposited on the forest floor through processes such as throughfall (precipitation wash-off) and stemflow5,6. As a result, 137Cs deposited on the forest floor was observed mostly in litter layers at forest sites in 2011 (ref. 3, 7), although some of the 137Cs was observed to be retained by abiotic components (minerals and organic materials) of the underlying soil at shallow depths8,9. Studies conducted in European forests after the accident at the Chernobyl NPP showed that a large part of Chernobyl-derived 137Cs persisted in forest-floor litter layers over a decade and was a prolonged source for 137Cs recycling in plants10,11,12. Therefore, knowledge about the behavior of Fukushima-derived 137Cs in Japanese forest ecosystems—particularly in forest surface environments—is of great importance in the assessment of associated radiological risks from both external and internal (via consumption of forest products) radiation exposure.

Forest-floor litter is a dynamic component of forest ecosystems in Japan, with a mean residence time of a few years13,14. The temporal pattern of litter accumulation is a function of litter input and decomposition15. In temperate deciduous forests, a bulk of deciduous leaf fall (litterfall) occurs in autumn (October and November), which is the largest source of litter input on a forest floor. Litter decomposition on a forest floor involves a complex set of processes including physical, chemical and biological breakdown of leaf-litter materials16. These decomposition processes operate simultaneously at varying rates depending on the litter quality and environmental conditions by which the mass (and size) and chemical composition (e.g., carbon and nitrogen content and organic carbon structure) of leaf litter change during decomposition14,17. Because of variations in input and decomposition behavior, the forest-floor litter comprises a variety of organic materials of different degrees of degradation, from undecomposed and partially decomposed leaves to finely fragmented and macroscopically unrecognizable materials.

Local topography, notably hillslopes, can further influence litter accumulation on a forest floor through its effect on microclimate and differential lateral transport of litter materials15,16. Litter decomposition is generally faster at hillslope bottoms than at upper-slope positions because of the moderated environmental conditions at the bottom; however, the downslope movement of litter materials more than offsets the enhanced decomposition and increases the accumulation of litter materials at the bottom15,18. The topographically controlled accumulation of litter materials at hillslope bottoms should be a rapid ecological process; it is thus hypothesized that this process causes a rapid topographic heterogeneity in the distribution of Fukushima-derived 137Cs on the forest floor. Once 137Cs reaches the soil, it can be rapidly and strongly adsorbed by fine soil particles19 and subsequently redistributed within the landscape primarily through physical processes20,21. Therefore, it is well documented that tracing the soil-associated 137Cs provides a very effective tool for estimating long-term (20–50 years) rates of 137Cs (soil) redistribution after deposition21,22. However, this technique is difficult to apply to litter layers as 137Cs cannot be strongly fixed by forest-floor litter materials, probably because cesium forms weak bonds with natural organic matter11. There is still insufficient understanding of the short-term dynamic processes that influence the distribution, migration and accumulation of 137Cs in litter layers on a forested hillslope.

Here we investigated the distribution of 137Cs in litter layers on a steep hillslope in a Japanese deciduous forest, the Ogawa Forest Reserve (Fig. 1), affected by the Fukushima NPP accident. We collected litter samples at three slope positions (bottom and 8 and 12 m above the bottom) on the hillslope in August 2013 (before the bulk of litterfall occurred in 2013), fractionated the samples into four litter fractions (F1 to F4) that were characterized by different sizes and physical status and then determined the concentrations and inventories of Fukushima-derived 137Cs in the fractions. Furthermore, mean ages since litterfall for the fractions were estimated based on their chemical (carbon and nitrogen) composition. We also observed the yearly change in 137Cs concentrations of fresh (newly emerged) leaves at the site. Based on these observations, we show evidence for a significant, short-term topographic heterogeneity in the accumulation of Fukushima-derived 137Cs on the forest floor, which is driven by biologically mediated processes.

Figure 1
figure 1

Location of the Ogawa Forest Reserve (a) and a photograph of the forested hillslope (b) investigated in this study.

The 137Cs inventory map (a) was generated using the website “Extension Site of Distribution Map of Radiation Dose, etc.” prepared by MEXT, Japan40. Photograph by E. Takeuchi.

Full size image

Results

Litter accumulation

The inventory of leaf-litter materials (excluding coarse woody debris such as fallen branches and twigs) on the forest floor was significantly greater at the bottom of the hillslope (2.6 ± 0.4 kg m−2; mean ± standard deviation of triplicate samples) than at 12 m (0.26 ± 0.06 kg m−2) and 8 m (0.56 ± 0.16 kg m−2) above the bottom (see Table S1 in Supplementary information). The inventory of coarse woody debris was 0.14 ± 0.07 kg m−2 (range: 0.07–0.31 kg m−2, corresponding to 4.8%–31.2% of the total amount of litter materials) and showed no significant difference between the slope positions.

The leaf-litter materials were fractionated into four fractions (Fig. 2): leaves showing no visible signs of degradation (F1); chipped or degraded leaves > 3 cm × 3 cm (F2) and < 3 cm × 3 cm (F3) in size; and fine leaf fragments < 1 cm × 1 cm in size, including macroscopically unrecognizable materials (F4). The fractionation method recovered 95.7% ± 2.6% (by weight) of leaf-litter materials (Table S1). The results revealed that the bottom of the hillslope accumulated more leaf-litter materials in all fractions than the upper parts of the hillslope (Fig. 3a). However, the distribution pattern of leaf-litter materials among the fractions was quite different between the slope positions. The F4 fraction represented the largest fraction (nearly half of the total mass) at the upper parts of the slope, whereas leaf-litter materials were approximately equally distributed among the four fractions at the bottom of the slope.

Figure 2
figure 2

Examples of litter materials in four litter fractions.

(a) Leaves showing no visible signs of degradation (F1); (b) chipped or degraded leaves >3 cm × 3 cm in size (F2) and (c) <3 cm × 3 cm in size (F3); and (d) fine leaf fragments <1 cm × 1 cm in size, including macroscopically unrecognizable materials (F4).

Full size image
Figure 3
figure 3

Inventories of leaf-litter materials (a) and 137Cs (b) in litter fractions (F1 to F4) at three slope positions.

Full size image

C and N content of litter fractions

Overall, the C content of the litter fractions decreased with decreasing size of leaf-litter materials in the fraction (F1 to F4), but the N content remained nearly constant (Table 1). As a result, the C/N ratio of the litter fractions decreased with decreasing size of leaf-litter materials. Considering the general trend that the C/N ratio progressively decreases with litter mass loss during an early stage of decomposition17,23, the C/N ratios obtained in this study indicate that the degree of degradation of leaf-litter materials increases in the order of F1 < F2 < F3 < F4. There were no consistent changes in the C and N content and the C/N ratio along the three slope positions.

Table 1 Concentrations of 137Cs, carbon (C) and nitrogen (N) and estimated mean ages of litter fractions at three slope positions
Full size table

Characterization of litter fractions based on their C/N ratios

A three-year litterbag experiment14 previously conducted in the Ogawa Forest Reserve showed that the C/N ratio of decomposing leaf litter decreased exponentially with time for two dominant species (beech and oak). A relationship between the C/N ratio and the incubation period (or the age of leaf-litter materials) was derived using the litterbag experiment data14 (see Fig. S1 in Supplementary information) and was employed to estimate the mean ages of litter materials in the four fractions. The mean ages increased with decreasing size of litter materials (F1 to F4), ranging from 0.5 to >3.0 years (Table 1). The mean ages of litter materials in the F1 fraction were generally less than one year, demonstrating that the leaves showing no visible evidence of degradation in this fraction were for the most part newly shed leaves originating from most recent litterfall events (i.e., October–November 2012). The mean ages for the F2 and F3 fractions suggest that the fractions mainly contained litter materials added to the forest floor in October–November 2011 (after the Fukushima NPP accident) at the bottom and at 12 m above the bottom of the hillslope. The finely fragmented F4 fraction showed mean ages of 1.2–2.4 years (except for the fraction obtained at 8 m above the bottom), indicating that the fraction was dominated by litter materials added via litterfall in 2010 (before the accident), as well as those added via litterfall in 2011 (after the accident).

The leaf-litter materials collected at 8 m above the bottom gave consistently lower C/N ratios (and thus, older mean ages) in all fractions than those collected at the other slope positions. Furthermore, a lower C content was observed in the fractions at this slope position, suggesting adhesion of soil mineral particles interacting with highly microbially transformed organic materials (humus) to leaf-litter materials24.

Radiocesium concentrations of litter fractions

The litter fractions largely varied in 137Cs concentration from 1,610 to 6,090 Bq kg−1 dry weight (dw) (Table 1). The concentrations were generally higher in fractions comprising smaller-size litter materials and at higher slope positions. Of particular interest was the observation of high 137Cs contamination (1,610–3,200 Bq kg−1 dw), even in the F1 fraction at all slope positions.

The measured 137Cs and 134Cs concentrations showed a similar pattern of distribution for all the litter fractions (see Table S1). The 134Cs/137Cs activity ratios of the litter fractions were 0.48 ± 0.02 (mean ± standard deviation), independent of the litter fraction category. The ratios corresponded well to the ratio (0.47) theoretically predicted for Fukushima-derived radiocesium at the time of sample collection (August 2013), the initial ratio being unity in March 2011 (ref. 25) and decreasing according to different rates of radioactive decay (the physical half-lives of 137Cs and 134Cs are 30.1 and 2.1 years, respectively). Therefore, the observed 137Cs in the present study was considered to originate from the Fukushima NPP accident.

Radiocesium inventory in litter layers

The total inventory of 137Cs in litter layers was significantly greater at the bottom (6.8 ± 0.6 kBq m−2) than at the upper-slope positions (1.3 ± 0.6 and 1.8 ± 0.4 kBq m−2) (Fig. 3b). At the upper-slope positions, approximately two-thirds of the total 137Cs inventory was apportioned to the F4 fraction; the other three fractions (F1, F2 and F3) retained only a small amount (~0.5–0.6 kBq m−2 in total) of 137Cs. In contrast, a large proportion (~65%) of Fukushima-derived 137C was observed in the F1, F2 and F3 fractions at the bottom of the hillslope, which amounted to ~4.4 kBq m−2, far larger than the total inventory of 137Cs at the upper-slope positions.

Radiocesium concentrations of fresh leaves

Concentrations of 137Cs in fresh leaf samples were 286–3,310 Bq kg−1 dw (Table 2), high in May 2011 (two months after the Fukushima NPP accident) and low in October 2013 and May 2014. The 134Cs/137Cs activity ratios were close to the theoretically predicted ratios (0.95, 0.45 and 0.38 for samples collected in 2011, 2013 and 2014, respectively), indicating that the radiocesium isotopes observed here came from the Fukushima NPP accident. There was no difference in 137Cs concentrations between the washed and non-washed leaf samples in 2014.

Table 2 Radiocesium concentrations of an archived litter sample and fresh beech leaf samples
Full size table

No fresh leaf samples were collected from the site before the accident. To determine the pre-accident level of 137Cs in beech leaves at the site, an archived litter sample collected from the forest floor in 2007 was analyzed for radiocesium. The 137Cs concentration of the archived litter sample was 5.0 Bq kg−1 dw, which was three orders of magnitude lower than that of fresh leaves collected in May 2011. As expected, 134Cs was not detected in the sample.

Discussion

The results of the present study showed that hillslope topography has a great effect on the accumulation of litter materials and consequently of Fukushima-derived 137Cs on the forest floor. A similar topographic pattern of surface litter accumulation along a hillslope has been observed throughout a growing season (April to October) in an American beech and maple forest15. The total inventory of Fukushima-derived 137Cs in the litter layer varied by a factor of five, from 1.3 kBq m−2 at the highest position to 6.8 kBq m−2 at the bottom position (Fig. 3b). More importantly, a significant fraction (65%) of the 137Cs inventory was associated with the younger litter materials (F1 to F3 fractions: mean ages of 0.5–1.5 years, dominated by litter materials added after the Fukushima NPP accident) at the bottom of the hillslope (Fig. 4).

Figure 4
figure 4

Relationship between mean age and 137Cs inventory for litter fractions (F1 to F4) at three slope positions.

Full size image

The picture that emerges is that biological recycling of 137Cs (i.e., uptake of 137Cs by trees and subsequent re-deposition on the forest floor via litterfall) plays an important role in causing topographic heterogeneity in the accumulation of Fukushima-derived 137Cs on the forest floor. The fresh leaves collected in May 2011 were highly contaminated (1,790–3,310 Bq kg−1 dw) with Fukushima-derived 137Cs (Table 2). The leaves were newly emerged ones; such a contamination therefore cannot be explained without invoking mechanisms such as uptake of 137Cs by roots and translocation of 137Cs from tree stems6,26,27. The contamination of leaf surfaces by adhering resuspended soil particles may be possible; however, the 137Cs concentration of fresh leaf samples was not reduced by washing (Table 2), suggesting that this process is of minor importance compared with root uptake28. Given the finding in the Ogawa Forest Reserve that leaves disperse along hillslopes within 20 m of source trees by lateral transport driven by wind action while falling29, it is possible that newly emerged leaves contaminated with “biologically recycled” 137Cs were carried to the bottom of the hillslope via litterfall. This is consistent with the observation that 137Cs concentrations (2,180–3,230 Bq kg−1 dw) of the litter materials in the F2 and F3 fractions (considered mainly from the 2011 litterfall events) at the bottom of the hillslope were similar to those of the fresh leaves collected in May 2011.

With an annual litterfall input of 0.43 kg m−2 y−1 at this site30, we estimate the annual input of biologically recycled 137Cs on the forest floor to be 0.77–1.42 kBq m−2 y−1 in 2011. This input corresponds to ~21%–39% of the 137Cs inventory in the F2 and F3 fractions at the bottom of the hillslope in August 2013, which is sufficiently large to possibly explain the observed accumulation of Fukushima-derived 137Cs on the forest floor at the bottom of the hillslope.

Furthermore, our observations suggest that the biologically recycled 137Cs has been supplied, but at a reduced rate, to the bottom of the hillslope via litterfall until the present. At the bottom of the hillslope, a substantial amount (0.77 kBq m−2) of Fukushima-derived 137Cs was found in the youngest F1 fraction, the nearly intact leaf-litter materials originating mainly from the 2012 litterfall events (Fig. 4). The fresh leaves collected in 2013 and 2014 still had 137Cs concentrations ranging from 286 to 875 Bq kg−1 dw; the concentrations were much lower than that of fresh leaves collected in May 2011 (two months after the accident), but were far higher than the pre-accident level (5.0 Bq kg−1 dw) (see Table 2). The contaminated leaves could still mediate ~0.13–0.38 kBq m−2 of biologically recycled 137Cs via annual litterfall.

On hillslopes, leaf-litter materials that have already been deposited on the forest floor may be redistributed by wind and gravity during snow-cover-free periods15,31. The trees at our site had no leaves in March 2011 when the Fukushima Daiichi NPP accident occurred; therefore, a majority of Fukushima-derived 137Cs was very likely to have been directly deposited onto the forest-floor litter materials4. The larger accumulation of Fukushima-derived 137Cs in the oldest F4 fraction at the bottom may indicate the redistribution (downslope movement) of litter materials contaminated with “directly deposited” 137Cs to the bottom. However, the younger mean ages for the F4 fraction at the bottom compared with the upper positions suggests a supply of younger litter materials to the fraction at the bottom; this indicates that the 137Cs accumulation observed in the F4 fraction at the bottom may be partly due to the supply of biologically recycled 137Cs. At the upper-slope positions, Fukushima-derived 137Cs was observed mainly in the oldest F4 fraction (Fig. 4). At the bottom of the hillslope, a large amount (4.4 kBq m−2) of litter-associated 137Cs was accumulated in the three younger (F1, F2 and F3) fractions. These results suggest that the short-term (within 2–3 years) topographic heterogeneity in the distribution of Fukushima-derived 137Cs on the forest floor has been established primarily by the redistribution of biologically recycled 137Cs, rather than by the redistribution of directly deposited 137Cs.

These findings have major implications for the assessment of future impacts of radioactive contamination of forest ecosystems from the Fukushima NPP accident. The biologically mediated redistribution of 137Cs on forested hillslopes significantly alters the distribution of litter-associated 137Cs on the forest floor and thus alters the radiation situation of not only external but also internal exposure to the population. The local (secondary) accumulation of litter-associated 137Cs at hillslope bottoms is likely to be a main source for 137Cs recycling in plants in the long term11, which may result in unexpected 137Cs contamination levels in some forest products. The redistribution may further influence the discharge of Fukushima-derived 137Cs from forest ecosystems. Studies conducted after the Chernobyl NPP accident have shown that forest ecosystems act as effective long-term reservoirs of the deposited 137Cs (ref. 32). However, the litter-associated 137Cs allocated to hillslope bottoms probably has more opportunities to be carried away by stream flows (particularly those caused by heavy precipitation events) and thus to be transferred from forest ecosystems to downstream areas via aquatic pathways33,34. This biologically mediated redistribution seems particularly important in Japan where forests are concentrated in mountainous and hilly regions with steep terrain. Clearly, this is worth further investigation (including field observations and modeling) to improve our understanding of the dynamics of 137Cs in Japanese forest ecosystems. Finally, the results of this study suggest that even more than three years after the Fukushima NPP accident, the removal of forest-floor litter materials preferentially accumulated at hillslope bottoms is still an effective countermeasure option to reduce forest contamination35.

Methods

Study site

The study was conducted on a steep, flat-bottomed hillslope (slope length: 28.5 m, slope angle: 35°–40°) in the Ogawa Forest Reserve (36°56′N, 140°35′E) in Ibaraki prefecture, Japan (Fig. 1). The Ogawa Forest Reserve is a temperate broad-leaved deciduous forest dominated by Japanese beech (Fagus crenata) and Japanese oak (Quercus crispula), located on an undulating plateau at the southern edge of the Abukuma mountain region. The area of the forest catchment is 58.4 ha and the elevation ranges from 588 to 724 m (ref. 36). The bulk of litterfall in the forest occurs during October and November and the annual litterfall input on the forest floor is 0.43 kg m−2 y−1 (ref. 37). The trees had no leaves in March 2011 when the Fukushima Daiichi NPP accident occurred. The mean annual temperature and precipitation are 10.7°C and 1,910 mm, respectively38. The soils of this area are heterogeneous, exhibiting a mosaic-style pattern of distribution of Cambisols and Andosols39. The parent materials are metamorphic rock and Late Quaternary volcanic ash39. The site was located approximately 70 km southwest of the Fukushima Daiichi NPP and affected by radioactive fallout from the Fukushima NPP accident at a level of 10–60 kBq m−2 of 137Cs deposition according to an airborne monitoring survey40.

Litter sample collection and fractionation

In August 2013 (before newly emerged leaves began to fall), litter samples were collected from litter layers on the surface of the soil at three slope positions: bottom and 8 and 12 m above the bottom of the hillslope. Litter samples were collected from three replicate plots (each having a 30 cm × 30 cm square) at each slope position. The litter samples were immediately transported to our laboratory with special care to avoid artificial fragmentation and then gently spread on wide trays to dry to a constant weight at room temperature.

To investigate the distribution of Fukushima-derived 137Cs among litter materials of different degrees of degradation, the litter samples were physically separated by hand into the following four fractions (Fig. 2). The first fraction (F1) consisted of leaves showing no visible signs of degradation. The second (F2) and third (F3) fractions both consisted of leaves that are more or less chipped or degraded and also included “skeleton leaves” in which the leaf parenchyma has been largely decomposed, but the midrib and veins (leaf tissues more resistant to microbial decomposition) persist. These two fractions were differentiated by leaf size: >3 cm × 3 cm in size for the F2 fraction and <3 cm × 3 cm for the F3 fraction. The last fraction (F4) consisted mainly of fine leaf fragments (<1 cm × 1 cm in size), including petioles detached from leaves and macroscopically unrecognizable materials. Here we assumed that the degree of degradation of litter materials is closely related to their sizes and therefore increases in the order of F1 < F2 < F3 < F4 in the fractionation method. Note that coarse woody debris (fallen branches and twigs) in the collected samples were removed before fractionation.

The amount of litter materials per unit area (or litter inventory; in kg m−2) in a given fraction was estimated as the mass of litter fraction (kg dw) divided by the area (m2) where the litter sample was collected (i.e., 900 cm2).

Fresh leaf sample collection

Fresh leaves were collected from a single beech tree growing around the hillslope in October 2013. In May 2014, newly emerged leaves were collected from a single beech tree at each slope position on the hillslope. The leaf samples collected in May 2014 were divided into two in our laboratory: one was dried without pretreatment; and the other was washed with water to remove adhering soil particles before drying. We did not collect leaf samples at the hillslope site in 2011, but had a sample (newly emerged beech leaves) collected ~180 m away from the hillslope within the Ogawa Forest Reserve in May 2011 (two months after the Fukushima NPP accident). This sample was used to estimate the radiocesium contamination level in fresh leaves in 2011.

There was also no sample collection at the site before the Fukushima NPP accident. Fortunately, we had an archived litter sample that was collected in January 2007 at the point where the 2011 leaf sample was collected. This archived litter sample was used to estimate the pre-accident level of 137Cs concentration in beech leaves at the site. All the leaf samples were dried to a constant weight at room temperature before the following analyses.

Radiocesium analysis

The activity concentrations of 137Cs and 134Cs in the litter fractions (F1 to F4) and fresh leaf samples were determined using gamma ray spectrometry and their values were expressed in activity per unit dry weight (Bq kg−1 dw). Samples (dried) were finely chopped using a mixer, sealed in plastic tubes (5 cm diameter, 7 cm height) and analyzed for 137Cs and 134Cs 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. The detector was calibrated with standard gamma sources (each with a relative uncertainty of ~5% for 137Cs) with different sample heights. The measurement times were normally 2,000–4,000 s for litter samples and 5,000–50,000 s for fresh leaf samples, which allowed us to obtain both 137Cs and 134Cs concentration values with relative errors <10% (with some exceptions for fresh leaf samples having low 134Cs concentration). The activity concentrations were corrected for radioactive decay to the sampling date.

Radiocesium inventory (Bq m−2) in a given litter fraction was estimated by multiplying the radiocesium activity concentration (Bq kg−1 dw) of the litter fraction by the inventory of the litter fraction (kg m−2).

C and N analysis and mean age estimation

The litter fractions were further analyzed for their total C and N content using an elemental analyzer (vario PYRO cube, Elementar). A three-year litterbag experiment14 previously conducted in the Ogawa Forest Reserve showed that the C/N ratio of decomposing leaf litter decreased exponentially with time (and thus with decreasing litter mass) for two dominant species (beech and oak). The data were used to derive a relationship between the C/N ratio (R) and the field incubation period (or the mean age of litter materials since the deposition: Y in years) (see Fig. S1 in Supplementary information):

and then the mean ages for the litter fractions obtained in the present study were estimated from their C/N ratios through this relationship.

Data analysis

The inventories of litter materials and 137Cs at different slope positions were statistically analyzed using analysis of variance. Means that exhibited significant differences were subjected to Tukey's honestly significant difference test to find which means were significantly different from one another at a 5% significance level.