Tracing radioactive cesium in stem wood of three Japanese conifer species 3 years after the Fukushima Dai-ichi Nuclear Power Plant accident

To understand the dynamics of accident-derived radioactive cesium (137Cs) in stem wood that had a substantial amount of heartwood at the time of the Fukushima Dai-ichi Nuclear Power Plant accident, the radial and vertical distributions of 137Cs activity concentration in stem wood of Japanese cedar (Cryptomeria japonica), cypress (Chamaecyparis obtusa), and larch (Larix kaempferi) were investigated. In addition, the natural distribution of stable cesium (133Cs), rubidium (85Rb), and potassium (39K) concentrations was analyzed to determine the characteristics of 137Cs distribution. Wood disks were collected from the tree stems of six cedars, three cypresses, and two larches at multiple heights in 2014, and the concentrations were measured every 2 cm in the radial direction. 137Cs distribution in stem wood differed among tree species, sampling site, and vertical position of the stem within a tree. Statistical analyses suggested that the radial distribution of 137Cs within the heartwood can be explained by the heartwood moisture content and the distance from the treetop, regardless of species, while the distribution between sapwood and heartwood was dependent on the heartwood cross-sectional area and was additionally different between larch and other species. Similarly, the heartwood/sapwood concentration ratios of stable alkali metals differed between larch and the other species. In the larch, the ratio was ca. 0.5 for all elements, but the ratio was over 1.0 and differed among elements in the other species. Consequently, the species-specific difference in the distribution of 137Cs between sapwood and heartwood was considered to be due to different activity levels of radial transport toward the heartwood. The radial variation pattern of the 137Cs/133Cs concentration ratio showed that less 137Cs was transferred to the inner heartwood compared with the 133Cs distribution pattern in many trees; however, there was also a tree in which 137Cs was excessively transferred to the inner heartwood compared with the 133Cs distribution pattern. Such patterns may result from a combination of significant foliar uptake of 137Cs and poor root uptake after the accident, in addition to the high moisture content of the heartwood.


Introduction
Radioactive cesium ( 137 Cs) dispersed into the atmosphere by a nuclear weapon test or power plant accident is transferred into trees mainly by two pathways, namely, foliar uptake and root uptake [1,2]. A major part of the foliar uptake of 137 Cs is assumed to occur shortly after a 137 Cs deposition event, when the fraction of dissolved 137 Cs in throughfall and precipitation is high [3], and can be the major source of initial 137 Cs input to trees that have needles or leaves at the time of the deposition event [4]. Subsequently, as the amount of 137 Cs in throughfall decreases exponentially with time [3,5], root uptake becomes the dominant source of 137 Cs input to trees over the long term [1]. Recently, bark uptake was experimentally demonstrated to be another possible pathway [6,7]; however, its contribution in the actual situation remains unclear, and its main contribution seems to occur shortly after the deposition event for the same reason as foliar uptake.
Regardless of the pathways, a portion of the 137 Cs absorbed into trees can be distributed in stem wood. As with other alkali metals (e.g., K and Rb), Cs absorbed from the surface of foliage can be translocated to tree stems via the phloem (inner bark) [8] and partly transported to the sapwood via the rays [9]. Cs absorbed from the roots moves upward with sap flow in the sapwood. Within the sapwood, Cs can be redistributed widely by transport via the axial and ray parenchyma cells and by diffusion in free water [10][11][12]. Although the transfer processes of alkali metals from sapwood to heartwood are not fully understood, many studies have observed accident-derived 137 Cs in the heartwood [e.g., 13,14], and an experimental study suggested the possibility of active transport via ray parenchyma cells and diffusion in the cell walls of the intermediate wood (the transition zone between sapwood and heartwood) [11,12]. Finally, within the heartwood, because there are no living cells, alkali metals transfer by diffusion in free water, which is generally less available in the heartwood than in the sapwood.
Due to the complex and uncertain transfer processes described above and the heterogeneous distribution of water, especially in the heartwood [15,16], 137 Cs is expected to show non-uniform distribution in stem wood. Actually, the radial and/or vertical distributions of 137 Cs activity concentration in stem wood affected by the Fukushima Dai-ichi Nuclear Power Plant (FDNPP) accident have been found to be heterogeneous and species-dependent [4,14,[17][18][19][20]. In addition, a monitoring survey has suggested the possibility of temporal change in 137 Cs distribution (progression of 137 Cs transfer to the inner parts of the heartwood) [18]. Thus, understanding the current situation and temporal change of 137 Cs distribution in stem wood is necessary for predicting 137 Cs contamination of wood. It is also important for estimating the 137 Cs contamination of an entire stem wood from a small sample (e.g., a sapwood core at breast height).
When the 137 Cs dynamics in a forest ecosystem reach the equilibrium state, the distribution pattern of 137 Cs in a tree is assumed to be similar to that of the naturally distributed stable isotope of cesium ( 133 Cs) [21]. However, it is uncertain whether the distribution patterns of 137 Cs and 133 Cs really match well in the equilibrium state, because their transfer processes are different in mature trees that had a substantial amount of heartwood at the time of the deposition event. Accident-derived 137 Cs is transferred into trees by both foliar and root uptake and into the heartwood after its formation, whereas 133 Cs is transferred into trees by root uptake and is incorporated into the heartwood with its formation. Although the influence of foliar uptake on the distribution of 137 Cs in stem wood is still unknown, similarity between 137 Cs and 133 Cs distribution patterns in stem wood has been found in a Scots pine (Pinus sylvestris) collected 12 years after the Chernobyl Nuclear Power Plant (CNPP) accident [21]. However, because the reported pine was about 20 years old at the time of the accident and probably had a small amount of heartwood, it is unclear whether this similarity can be found in older and larger trees with a substantial amount of heartwood. This is a big concern, especially in the case of the FDNPP accident, because Japanese cedar (Cryptomeria japonica) and cypress (Chamaecyparis obtusa), which are key species for forestry in the contaminated area, have larger heartwood fractions than Scots pine and also because they have the species-specific characteristic that the Cs concentration is higher in the heartwood than in the sapwood [22,23].
Our long-term goal is to clarify temporal changes in the distribution of accident-derived 137 Cs in stem wood from the non-equilibrium state to the equilibrium state, especially in mature trees that had a substantial amount of heartwood at the time of the accident. As the first step, this study aims to understand the characteristics of the radial and vertical distributions of 137 Cs activity concentration in stem wood 3 years after the FDNPP accident, assuming that the distributions were still in the non-equilibrium state. We targeted three major conifer species in the contaminated area: two evergreen conifer species, namely, Japanese cedar (Cryptomeria japonica [L.f.] D.Don) and cypress (Chamaecyparis obtusa [Siebold et Zucc.] Endl.), and one deciduous conifer species, namely, Japanese larch (Larix kaempferi [Lamb.] Carrière). The characteristics of 137 Cs distribution in stem wood were evaluated by statistical analyses and by comparison with stable alkali metals ( 133 Cs, 85 Rb, and 39 K).

Sample collection and preparation
Samples were collected from three sites in August and September 2014. Sites 1 and 2 were located in Kawauchi Village of Fukushima Prefecture, and site 5 was located in Ishioka City of Ibaraki Prefecture ( Fig. 1; site identifications correspond to those in Ohashi et al. [24]). As of December 28, 2012, the deposition densities of 137 Cs after the FDNPP accident were 390, 140, and 11 kBq m −2 at site 1, 2, and 5, respectively [25]. Japanese cedar samples were collected from sites 1 and 2, cypress samples from site 5, and larch samples from site 2. Although collecting wood samples from the same individual over a long period is ideal, we collected wood disks from multiple individuals in this study because it was difficult to obtain an adequate amount of sample for 137 Cs measurements by core sampling and to get samples from the upper parts of the stems without felling the trees. We cut down three individuals from different classes of diameter at breast height (DBH) at each site for the cedar and cypress (sizes were identified as large (L), medium (M), and small (S) according to relative size of the DBH in each site) and two individuals for the larch (sizes were identified L and M). Before collecting the wood disks, the bark was removed with a saw and scraper to prevent highly contaminated bark from mixing into the wood. Then, several wood disks about 5 cm thick were cut from multiple heights of the stems with a chainsaw. The sampling heights were 1, 5, and 10 m above the ground for size L and M trees (also 15 and 20 m for tall trees in site 2) and 1 m for size S trees. In addition, disks were collected from 2.5 and 4.5 m above the ground to evaluate the heartwood moisture content of each tree. Details of the sampled trees are shown in Table 1.
Each wood disk was divided every 2 cm in a radial direction with the sapwood-heartwood boundary as a starting point. The sapwood-heartwood boundary, which was mostly narrow and indistinct, was regarded as the middle of the intermediate wood for the cypress and larch samples. On the other hand, the intermediate wood was treated as an independent sample for the cedar samples, because it was wide (about 1 cm in the radial direction) and distinct in all disks. The edge of a disk sector (i.e., the outermost sapwood or innermost heartwood) was included in the adjacent sample when its radial length was less than 1 cm, and it was obtained as an independent sample when its radial length was 1 cm or longer; however, the innermost heartwood for which the  radial length was less than 1 cm was not analyzed because its amount was not enough for 137 Cs measurement.

Determination of radioactive cesium ( 137 Cs) activity concentration
The samples were chipped using a cutting mill with a 6-mm sieve (UPC-140, HORAI, Higashiosaka, Japan) and packed into polystyrene containers (type U-8 [100 mL] or V-1 [550 mL]) for measurement of radioactivity. Because of the low heat resistance of the container, the sample mass was measured after oven-drying at 75 °C for 48 h and then converted to a dry mass at 105 °C by multiplying by the conversion factor (0.99) [24]. The radioactivity of 137 Cs in the samples was determined using gamma-ray spectrometry with a high-purity Ge detector (GEM20

Statistical analysis of 137 Cs distribution
To understand the characteristics of 137 Cs distribution in stem wood, statistical analyses were performed using linear mixed models (LMMs) with a random effect of individual trees. Because the 137 Cs transfer mechanism from sapwood to heartwood and that within the heartwood are assumed to be different, we used two different models: sapwood to heartwood (SW-to-HW) and within-HW models.
The response variables were the concentration ratio of 137 Cs in the heartwood to sapwood (HW/SW CR of 137 Cs) in the SW-to-HW model and the concentration ratio of 137 Cs in the innermost heartwood to the outermost heartwood (HW inner /HW outer CR of 137 Cs) in the within-HW model. If the sapwood or heartwood of the wood disk was divided into subsamples, we used the mass-weighted mean concentration of 137 Cs in the sapwood or heartwood in the SW-to-HW model.
Based on the following hypotheses or reported observations, we designated eight variables as fixed effects in both the SW-to-HW and the within-HW models ( Table 2). (1) Heartwood cross-sectional area (HWA) of wood disk: in a wood disk with larger HWA, the 137 Cs concentration in the heartwood is more diluted, and it takes a longer time for 137 Cs to transfer to the inner parts of the heartwood. (2) Heartwood circumferential length per HWA (HWCL-per-HWA) of wood disk: when HWCL-per-HWA is longer (or heartwood diameter is smaller), 137 Cs transfers from sapwood to heartwood more efficiently. (3) Whether a wood disk is located in the tree crown or not (or is located at a higher position than the lowest living branch or not) (CROWN): the 137 Cs concentration in the inner parts of a wood disk located within the tree crown is likely to be higher than that in the inner parts of a wood disk located below the tree crown [26]. (4) Distance from the treetop (D-from-T) of wood disk: Ogawa et al. [18] reported that the 137 Cs activity concentration in the heartwood tended to be higher in positions closer to the treetop and suggested that D-from-T is a better explanatory variable than distance from the ground to explain the vertical distribution of 137 Cs in the heartwood. Although D-from-T is somewhat correlated (not significantly) with HWA, HWCLper-HWA, and CROWN, we designated this variable to consider unexplained effects related to vertical position in the stem. (5) Heartwood moisture content (HWMC) of the tree: 137 Cs transfers more easily to inner parts of the heartwood when HWMC is higher. Iizuka et al. [20] reported that the radial distribution of 137 Cs activity concentration in Japanese cedars with high HWMC was relatively uniform throughout the heartwood compared with that in cedars with low HWMC. In this study, HWMC of the tree (mean of HWMC values of wood disks sampled from 2.5 and 4.5 m above the ground) was used because HWMC data for each disk were not available. (6-8) Tree species and sampling site (LARCH, SITE2, CYPRESS-SITE5): to detect other effects that are not explained by the above five variables and arise from differences between tree species and sampling sites, the effects of larch (LARCH) and site 2 (SITE2) were included in the models, setting "cedar" as the reference category of species and "site 1" as the reference category of site. The effects of cypress and site 5 were combined into one variable (CYPRESS-SITE5), because the cypress samples were collected from the single site (site 5) that was different from the sampling sites of the other two species, and we cannot distinguish the effects of cypress and site 5. All statistical analyses were performed using R version 3.6.1 [27]. The response variables were transformed to their common logarithms. Data from size S trees were not used in the analyses, because these data are only from disks sampled at 1 m above the ground. In addition, the wood disk from the size L larch collected at 20 m above the ground was eliminated from the within-HW model analysis because it was not divided into subsamples. The four fixed effects, CROWN, LARCH, SITE2, and CYPRESS-SITE5, are categorical variables, and the other fixed effects are continuous variables that were centered by subtracting their means and scaled by dividing them by their standard deviations. Multicollinearity in the model including all the fixed effects was checked using the variance inflation factor (VIF) [28], and none of the fixed effects were eliminated here, judging by the criterion that VIF should not exceed 10. Tentative models derived from all combinations of the fixed effects were fitted to the data using the lmer function in the lme4 package [29], and Akaike's information criterion (AIC) value was calculated for each model. Finally, we selected the model with the lowest AIC value as the most reasonable model (selected independently for the SW-to-HW and within-HW models).

Radioactive cesium ( 137 Cs)
The radial distribution of 137 Cs activity concentration in stem wood varied not only among species, but also among trees of the same species collected from different sites and among vertical positions of the stem within tree (Fig. 2). Nevertheless, the radial distribution at the same sampling height showed high reproducibility among trees from the same site. The vertical distribution of 137 Cs activity concentration in the sapwood was almost uniform in all trees, whereas that in the heartwood was nonuniform and was higher in the upper parts of the stems in most trees.
Comparison of the 137 Cs activity concentration in the sapwood and the heartwood showed that most wood disks of cedar and cypress had higher concentrations in the heartwood than in the sapwood (Fig. 2a-i). However, disks from site-2 cedars collected at or below 10 m above the ground and disks from site-5 cypresses collected at or below 5 m above the ground had lower concentrations in the heartwood than in the sapwood (Fig. 2d-i). In contrast to these two species, larch had lower concentrations in the heartwood than in the sapwood at all sampling heights (Fig. 2j, k). The activity concentration in the intermediate wood of the cedar was within the range of that in the innermost sapwood and that in the outermost heartwood in most disks (Fig. 2a-f ).
The radial distribution of 137 Cs activity concentration within the heartwood showed an increasing pattern toward the pith or was almost uniform throughout the heartwood in the upper parts of the stems (at or above 15 m above the ground in the site-2 cedars, at 10 m in the site-5 cypresses, and at 20 m in the site-2 larch) and showed a decreasing pattern toward the pith in the lower parts of the stems (Fig. 2). However, in the site-1 cedars, the distribution within the heartwood was almost uniform, even in the lowest part of the stems (1 m above the ground). As a result of the statistical analysis of the SW-to-HW model, the LMM model including HWA (heartwood cross-sectional area) and LARCH as fixed effects was selected to explain the HW/SW CR of 137 Cs. The other effects (HWCL-per-HWA, CROWN, D-from-T, HWMC, SITE2, and CYPRESS-SITE5) were not included in the selected model. The regression coefficients of HWA and LARCH were both negative, meaning that the HW/SW CR of 137 Cs was lower in disks with larger HWA and was lower in the larch than in cedar and site-5 cypress ( Table 3). The variance of the random effect (trees) was 0.04. The coefficient of determination (R 2 ) of the simple regression analysis between the estimated and observed values was 0.91.
For the within-HW model, the LMM model including D-from-T (distance from the treetop) and HWMC (heartwood moisture content) as fixed effects was selected to explain the HW inner /HW outer CR of 137 Cs. The regression coefficient of D-from-T was negative and that of HWMC was positive, meaning that the HW inner / HW outer CR of 137 Cs was lower in the disk farther from the treetop and higher in trees with higher HWMC ( Table 4). The variance of the random effect (trees) was estimated to be 0, suggesting that there was no consistent random effect among trees. R 2 of the simple regression analysis between the estimated and observed values was 0.72.

Stable alkali metals ( 133 Cs, 85 Rb, and 39 K)
The radial and vertical distribution patterns of concentrations of the stable alkali metals in the stem wood were similar among the elements in all analyzed trees (Fig. 3). However, the HW/SW CRs of 133 Cs, 85 Rb, and 39 K (mean ± standard deviation) were 2.5 ± 0.8, 5.3 ± 1.1, and 6.6 ± 1.2 in the cedar; 1.4 ± 0.7, 2.1 ± 1.0, and 2.5 ± 1.1 in the cypress; and 0.5 ± 0.1, 0.4 ± 0.0, and 0.5 ± 0.0 in the larch, respectively; that is, the HW/SW CR was high in the order of K, Rb, and Cs in the cedar and cypress, and was similar among K, Rb, and Cs in the larch.

Comparison between radioactive and stable cesium
The ratio of 137 Cs activity concentration to 133 Cs concentration ( 137 Cs/ 133 Cs ratio) was calculated to compare the distribution patterns of 137 Cs and 133 Cs, and its radial distribution in each disk is shown in Fig. 4. The 137 Cs/ 133 Cs ratio in the heartwood showed decreased patterns toward the pith in three of four analyzed trees (site-2 cedar, site-5 cypress, and site-2 larch) ( Fig. 4c-i). On the other hand, the 137 Cs/ 133 Cs ratio was similar between the sapwood and the outer heartwood and was almost constant throughout the heartwood in the other tree (site-1 cedar) sampled at 1 m above the ground (Fig. 4b). Remarkably, in the disk from the site-1 cedar collected at 10 m above the ground, the 137 Cs/ 133 Cs ratio was higher in the heartwood than in the sapwood and showed an increased pattern toward the pith within the heartwood (Fig. 4a).

Vertical distribution of 137 Cs in the sapwood and heartwood
As previously reported for Japanese cedars and red pines (Pinus densiflora) [4,14,18,19], the vertical distribution of 137 Cs activity concentration in the sapwood was almost uniform in all three species examined in this study. Because 133 Cs, 85 Rb, and 39 K showed similar distributions, alkali metals are considered to be highly mobile and/or evenly bound within the sapwood of these species. However, we should note that such uniform vertical distributions in the sapwood may not be common in all species, because Japanese oak (Quercus serrata) had higher 137 Cs activity concentrations in the upper parts of the stems [14].
A trend toward a higher 137 Cs activity concentration in the heartwood of the upper stem was commonly reported for Japanese cedars after the FDNPP accident [4,18,19]; however, its mechanism is not fully understood. The results of this study suggest four possible factors causing this trend. The first possible factor is the dilution effect of heartwood volume (heartwood cross-sectional area) suggested by statistical analysis of the HW/SW CR of 137 Cs (SW-to-HW model). In case accident-derived

Table 3 Regression coefficients of the SW-to-HW model (after model selection)
The response variable is log 10  Cs transfers into the already-formed heartwood, the larger heartwood volume at the lower parts of the stem results in a relatively low 137 Cs activity concentration in the heartwood within a tree. The second possible factor is the initial foliar uptake of 137 Cs after the accident. As discussed in more detail later (see "Comparison between the distribution of 137 Cs and 133 Cs"), the fact that a higher 137 Cs/ 133 Cs ratio was found in the heartwood than in the sapwood of the upper stem (Fig. 4) suggests that the initial foliar uptake possibly increased the 137 Cs activity concentration in the heartwood of the upper stem. The third possible factor is slow downward diffusion of 137 Cs in the heartwood. Ogawa et al. [18] reported the possibility that downward diffusion homogenized the vertical distribution of 137 Cs activity concentration in the heartwood from 2011 to 2013. Because we still observed high heterogeneity in 2014, the downward diffusion of 137 Cs in the heartwood is likely to be slow. The fourth possible factor is the intrinsic tendency of the vertical distribution of alkali metals. Although the tendency is not as pronounced as that of 137 Cs, the concentrations of naturally distributed alkali metals also tended to be high in the upper stems (Fig. 3). This fact implies that the activity of the radial transport toward the heartwood might be higher at the upper stem; however, further studies are needed to understand such a physiological nature. We infer that these multiple factors cause the characteristic vertical distribution of 137 Cs in the heartwood in the nonequilibrium state.

Transfer of 137 Cs from sapwood to heartwood
It is known that the radial distributions of alkali metals in stem wood are species-dependent [23] and that an active transport by ray parenchyma cells can be involved in the radial transfer [11,12,30]. Statistical analysis using the SW-to-HW model demonstrated that the differences in the HW/SW CR of 137 Cs among the vertical positions of the stem within a tree, among trees of the same species, and between the cedar and site-5 cypress were explained efficiently by the HWA of the disk, suggesting that the HW/SW CR of 137 Cs was dependent on the dilution effect of heartwood volume as of 2014. At the same time, the statistical analysis detected a difference that was not explained sufficiently by the HWA, and the HW/SW CR of 137 Cs was lower in the larch than in the other species. A similar difference was seen in the HW/SW CRs of stable alkali metals; the CRs were ca. 0.5 and did not differ among the elements in the larch, whereas the CRs were more than 1.0 and differed among the elements (high in order of K, Rb, and Cs) in the other species. Such differences between the stable alkali metals imply that there is an active transport toward the heartwood ND letters indicate that 137 Cs activity in the samples was under the detection limit and show the detection limit values. SW sapwood, HW heartwood targeting mainly K in the cedar and cypress and no active transport or rather an active transport toward the outer sapwood in the larch. Consequently, we infer that the difference between the larch and the other species in the SW-to-HW model arises from the different activity levels of radial transport toward the heartwood by ray parenchyma cells.

Radial distribution of 137 Cs within the heartwood
In contrast to the SW-to-HW model, the within-HW model selected in this study explained the differences in the HW inner /HW outer CR of 137 Cs among the species by the HWMC, without adding a species effect. This is quite a convincing result, because 137 Cs transfer within the heartwood must be caused only by diffusion [12], and the diffusibility obviously depends on the moisture content. Iizuka et al. [20] demonstrated that Japanese cedars with high HWMC (around 200%) had higher 137 Cs activity concentrations in the heartwood near the pith than did those with low HWMC (around 80%). Although the HWMC of the cedar samples in this study was relatively low (below 100%), it was found to be a useful variable to explain differences not only within the cedars, but also among the species. In addition, the HWMC value used in this study is the representative value for each individual and not the value for each disk. Because the HWMC can vary among vertical positions within a tree [31], using the HWMC value of each disk may improve the explanatory power of the model. The interpretation of the other fixed effect, D-from-T, which explains the differences in the HW inner /HW outer CR of 137 Cs among trees and among vertical positions within trees, remains unclear. Because D-from-T partly includes the effects of HWA, HWCLper-HWA, and CROWN, the selection of D-from-T in preference to the other effects suggests that the differences in the HW inner /HW outer CR of 137 Cs were likely to be due to multiple factors rather than a single factor.

Comparison between the distributions of 137 Cs and 133 Cs
The radial distribution of the 137 Cs/ 133 Cs ratio in stem wood was investigated to clarify how different the distribution of accident-derived 137 Cs was from that of the naturally existing 133 Cs as of 2014. Assuming that 137 Cs and 133 Cs in the outermost sapwood are well mixed and their ratio represents the latest value at the time of the sampling, we categorized the radial distribution patterns of the 137 Cs/ 133 Cs ratio into three types to discuss the characteristics of 137 Cs distribution. In type 1, the 137 Cs/ 133 Cs ratios in the heartwood are lower than those in the sapwood and decrease toward the pith; in type 2, the ratios are similar from the outer sapwood to the inner heartwood; in type 3, the ratios in the heartwood are higher than those in the sapwood and increase toward the pith (i.e., the reverse pattern to type 1).
The type 1 radial distribution pattern indicates that not as much 137 Cs had transferred to the inner parts of the heartwood compared with the 133 Cs distribution pattern. Thus, it seems to take a longer time or be more difficult for 137 Cs to achieve the same distribution pattern as 133 Cs in stem wood of this type. Three out of four trees analyzed in this study were categorized as type 1. Although data on the radial distribution of both 137 Cs and 133 Cs are scarce, from analysis of the data reported by Mahara et al. [17], the 137 Cs/ 133 Cs ratio patterns in a Japanese cedar and oak (Quercus serrata) collected in 2012 were found to be type 1. Therefore, several years after the accident, type 1 was probably the common pattern in mature trees with a substantial amount of heartwood, suggesting that 137 Cs distribution was still in the non-equilibrium state. The type 2 pattern suggests that 137 Cs had already transferred to the inner parts of the heartwood as much as the 133 Cs distribution pattern. Because the 137 Cs/ 133 Cs ratio pattern in a Scots pine collected 12 years after the CNPP accident was type 2 [21], 137 Cs distribution in the stem wood of type 2 is considered to be in an equilibrium state, and its pattern is expected to be unchanged in the future. This type was found in the site-1 cedar, which had relatively high HWMC (97%; Fig. 4b) compared with the other trees (HWMC 35-53%). Thus, as suggested by the within-HW model, high HWMC is likely to be one of the factors causing type 2.
Type 3, which was found in the disk of the site-1 cedar collected at 10 m above the ground (Fig. 4a), is a remarkable pattern, indicating that the 137 Cs activity concentration in the sapwood was considerably higher in the past than at the time of sampling (in other words, 137 Cs concentration decreased at a greater rate than radioactive decay of 137 Cs). This pattern could have happened when the initial foliar uptake of 137 Cs shortly after the accident was dominant and the subsequent root uptake was not dominant. In addition, the relatively high concentration of 137 Cs obtained by the initial foliar uptake must have transferred to the inner heartwood sufficiently to form the pattern of type 3. Although it is difficult to estimate the initial foliar uptake, the subsequent root uptake is inferred to be not dominant in the site-1 cedars from the observation that the 137 Cs activity concentration in the sapwood of the site-1 cedars did not change significantly from 2011 to 2016 [24]. From this point of view, the site-2 cedar, in which root uptake of 137 Cs is inferred to be dominant [24], and the larch, which did not have foliar uptake because of the absence of foliage at the time of the deposition event, are expected to have no type 3 pattern; and actually the site-2 cedar and the larch were not categorized as type 3 but as type 1. This study showed that the type 3 pattern persisted for at least 3 years after the accident, although it may become indistinct as the fraction of newly formed heartwood increases. Consequently, the distribution of 137 Cs in stem wood was suggested to be affected not only by the tree's internal factors (e.g., HWMC and active transport), but also by environmental factors (foliar and root uptake) and the course of time.

Conclusions
In mature trees that had a substantial amount of heartwood at the time of the FDNPP accident in 2011, the vertical distribution of 137 Cs activity concentration in the sapwood was relatively uniform for all species, while the vertical and radial distributions in the heartwood were heterogeneous as of 2014, and the radial distribution pattern varied among species, individuals, and vertical positions within individuals. Statistical analysis suggested that the differences in radial distribution patterns within the heartwood among species can be explained by heartwood moisture content. On the other hand, there was an unexplained difference between the larch and the other two species in the radial distribution patterns between sapwood and heartwood. Taking the distribution patterns of stable alkali metals ( 133 Cs, 85 Rb, and 39 K) into account, such species dependency in the radial distributions between sapwood and heartwood was inferred to be due to the different activity levels of the radial transport of alkali metals toward the heartwood by ray parenchyma cells.
The radial distribution pattern of the 137 Cs/ 133 Cs ratio showed that not as much 137 Cs had transferred to the inner parts of the heartwood compared with the 133 Cs distribution pattern in the majority of analyzed trees as of 2014. However, there was also a tree in which 137 Cs was transferred to the inner heartwood excessively compared with the 133 Cs distribution pattern. We deduced that such patterns could be found in a tree that had significant foliar uptake of 137 Cs initially and poor root uptake subsequently after the accident, in addition to having high heartwood moisture content. Although analysis of the 137 Cs/ 133 Cs ratio is useful to understand the current situation of 137 Cs distribution and to predict possible temporal changes in the future, follow-up monitoring and studies will be necessary to determine how rapidly transfer of 137 Cs to the heartwood progresses and when it reaches an equilibrium state. Heartwood circumferential length; HWMC: Heartwood moisture content; VIF: Variance inflation factor; AIC: Akaike's information criterion.