Study site
The study sites were located approximately 20–75 km from the Fukushima Daiichi Nuclear Power Plant in Fukushima Prefecture, Japan (Fig. 1). According to an aircraft radioactivity survey reported by the Ministry of Education, Culture, Sports, Sciences, and Technology of Japan19, the air dose rate in this region was 0.3–3.2 μSv/h, and the deposition of cesium-134 and cesium-137 ranged from less than 64,000 to 940,000 Bq/m2 (Table 1) in June 2011. The study catchment area is mostly forested and dominated with deciduous trees. Other areas in the region are also forested as well, with Japanese cedar and cypress plantations used for timber production. A field survey was conducted at one headwater tributary (A) of the Nagase River and three headwater tributaries (B, C, and D) of the Kido River. The substrate of these sites was consisted with sand, cobble and rocks. Geological feature of the soil on all the sites was the same, biotite granite. Streams at sites B, C and D were covered with riparian forests and it was difficult for sunlight to penetrate directly. Stream width of site A was wider than sites B, C and D, so sunlight could penetrate through the forest cover and contact the stream surface only along the middle of the stream.
Study site in Fukushima Prefecture, Japan. Square: sampling sites, circle: FDNPP (Fukushima Daiichi Nuclear Power Plant). This map was generated by using software program Microsoft Paint Windows 10.
Sampling
The air dose rate at 1- m above the ground was measured with a γ survey meter at the sampling site (TCS-172 NaI scintillation counter; ALOKA). The electrical conductivity (EC) of the streams was measured using a portable compact twin conductivity meter (B-173; Horiba); pH was measured using a portable compact twin pH meter (B-212; Horiba), and the dissolved oxygen (DO) was measured using a portable DO meter (DO-5509; Lutron). Stream velocity was measured using a portable meter (V-303, VC-301, KENEK). All parameters were measured at all sites on all sampling dates.
Sand substrate, litter and algae were sampled from stream riffles at a depth of 10-15 cm from July 2013 to April 2019, as was reported in previous studies13. The sand substrate was sampled in each riffle to a depth of 5- cm. When sand was not immediately visible in the stream substrate, stones were removed and the sand underneath the stones was sampled. Litter shed in the water was collected after gentle hand-rinsing. Leaf litter forms the base of stream food webs. Periphytic algae were collected by brushing the pebbles or rocks with a toothbrush. These algae are also primary producers at the base of stream food webs. Prior to brushing, we gently hand-rinsed the stone surface to remove other organic matter and aquatic invertebrates in the periphyton.
Aquatic invertebrates from thirteen groups (Perlidae Gen. spp., Nemouridae Gen. spp., Ephemera japonica, Ephemerellidae Gen. spp., Heptageniidae Gen. spp., Hydropsychidae Gen. spp., Stenopsychi spp., Rhyacophilidae Gen. spp., Epiophlebia superstes, Lanthus fujiacus, Tipulidae Gen. spp., and Corydalidae Gen. spp., Geothelphusa dehaani,) were qualitatively sampled from riffles at a depth of 10-15 cm at the four sites from July 2013 to April 2019. At each site, a D-frame net with a 1-mm mesh was placed downstream of the sampling area on the substrate in water. We then disturbed the substrate upstream of the net, allowing insects to drift into the D-frame net. The sampled aquatic invertebrates were identified to family level in the field and then frozen.
Three bricks (210 × 100 × 60 mm) were placed separately within the stream riffle at a depth of 10–20 cm on August 25, 2014 at each of the four sites. Then, periphytic algae growing on the bricks were collected by brushing the substrate with a toothbrush. Before brushing, we gently hand-rinsed the brick surface in running water to remove other organic matter from the periphytic algae. The sampling was carried out eight times: in October and December 2014; March, May, June, July and November 2015; and April 2016. Stream velocity of right side, upper reaches side and left side of each brick were measured and averaged. This averaged value was used as the stream velocity of each periphytic algae sample.
Radiocesium analysis
Radiocesium was analysed according to the methods in previous studies10,20. Samples of sand substrate and litter were dried at 75 °C in an oven. Thereafter, samples of sand were placed in a sieve (mesh size 2 mm; Iida, Japan), and the sand that passed through the sieve was used, meaning that the sand substrate in this study included silt granules. Samples of algae were concentrated via evaporation and dried in an oven at 75 °C. Samples of aquatic invertebrates were also dried in an oven at 75 °C. All samples were homogenized and packed into 100-ml polystyrene containers (U-8). Gamma-ray spectrometric measurements were performed on each sample. The radioactive concentrations of cesium-134 (604 keV) and cesium-137 (662 keV) were measured using an HPGe coaxial detector system (GEM40P4-76, Seiko EG and G, Tokyo, Japan) at the Forestry and Forest Products Research Institute (FFPRI) with a time of 36,000 s or longer. Data with a standard error of < 10% were used for analysis. Gamma-ray peaks at 604.66 and 661.64 keV were used to identify cesium-134 and cesium-137, respectively. The measurement system was calibrated using a standard gamma-ray source (MX033U8PP; Japan Radioisotope Association, Tokyo, Japan), and a standard soil material (IAEA-444) was used to check the measurement accuracy (relative expanded uncertainty value: 4.6%).
The homogenized samples of aquatic invertebrates were further homogenized with Na2SO4 for radiocesium analysis. The processed aquatic invertebrates were packed into 20-ml plastic tubes. Gamma-ray spectrometric measurements were performed on each tube. Total concentrations of cesium-134 and cesium-137 were measured using a gamma particle counter with a NaI crystal detector (2480WIZARD2, Perkin Elmer, Downers Grove, IL, USA) and a sampling time of 21,600 s at the FFPRI. The measurement system was calibrated with a standard gamma-ray source (E-265/107F16-3, Canberra Industries, Oak Ridge, TN, USA) and a standard soil material (IAEA-444) was used to check the measurement accuracy (relative expanded uncertainty value: 2.9%). A strong correlation between the data obtained with the HPGe coaxial detector system and the data obtained with the NaI crystal detector was observed (cesium-134: τ = 0.75, z = 5.61, n = 33, P < 0.001; cesium-137: τ = 0.76, z = 8.43, n = 63, P < 0.001; Kendall test). As the detection limit of the HPGe coaxial detector system was higher than that of the gamma particle counter with the NaI crystal detector, the data obtained with the gamma particle counter with the NaI crystal detector were used for the analysis of aquatic invertebrates because we could then use smaller numbers of aquatic invertebrates. Only samples with a dry mass > 50 mg were included in the analysis to avoid measurement errors due to the low quantity.
The relation between radiocesium concentration and the days after accident
The data measured for sand substrate, litter, algae and aquatic invertebrates were plotted on a logarithmic scale against the days after accident occurred, and correlation coefficients were calculated. The significance of the correlations was statistically assessed using the Pearson’s product-moment correlation coefficients. Radiocesium concentrations that were below the detection level on a dry weight basis because of insufficient mass for radiocesium analysis were excluded from the analysis. Using the regression curves, the radiocesium decay was corrected to September 1, 2013 and the differences in the values among sites were statistically tested using the Friedman-test.
The ecological half-life and the transfer coefficient
Ecological half-life was estimated by fitting the measured data to an exponential curve with the equation:
$${text{Ln}},{text{Y}} = {text{Ln}},{text{A}}{-}{text{Ln}},2*left( {{text{t}}/{text{T}}_{{{text{ecol}}}} } right),$$
where the activity concentration in a sample group is plotted as a function of time. A is a constant, Tecol is the ecological half-life, and t is time after the accident. The ecological half-life of cesium-134 and cesium-137 was estimated through exponential curve fitting. A positive relationship between the radiocesium concentration and the days after the accident indicates the increase of radiocesium according to the time, thus the ecological half-life cannot be calculated under this scenario. It would be better to include the physical decay in calculations of the biological half-life using long-term data21. However the research of this study started 2 years after the accident and the research period was only 6 years. Moreover, we did not need to establish the biological half-life, but rather the ecological half-life. For this reason, in this study, we did not consider the physical decay of a radionuclide to calculate the ecological half-life, which enabled us to integrate the actual radionuclide transition in the field. Differences in ecological half-life among sites were statistically tested using the Friedman-test.
The transfer coefficient (T) is an empirical coefficient that can be used in predictive models for natural ecosystems. It is defined by the following expression:
$${text{T}} = {text{Activity}},{text{concentration}},{text{of}},{text{the}},{text{invertebrate}},{text{samples}},left( {{text{Bq}}/{text{kg}}} right)/{text{Activity}},{text{concentration}},{text{of}},{text{the}},{text{algae}},{text{or}},{text{litter}},left( {{text{Bq}}/{text{kg}}} right)$$
The diet of some aquatic invertebrate species comprises litter rather than algae, whereas the diet of other species comprises both algae and litter. In this analysis, all the invertebrate groups were analysed under the assumption that their diets were dependent on algae or litter. Differences in the T value among sites were statistically tested using the Friedman-test.
The relation between radiocesium concentration and current velocity
The radiocesium concentrations based on dry weight were estimated for September 1, 2013. Radiocesium concentrations that were below the detection level on a dry weight basis, because of insufficient mass for radiocesium analysis, were excluded from statistical analysis. Nonparametric tests for statistical analysis were conducted. The Kendall rank correlation test was used to clarify the relationships between stream velocity and concentrations of radioactive cesium-134 and cesium-137. To examine the differences in the decline between sites A and D, the comparison of two regression slopes test was conducted.
Source: Ecology - nature.com