Introduction

A large amount of radiocesium (137Cs + 134Cs) was released into terrestrial and marine environments following the Fukushima Dai-ichi Nuclear Power Plant (FDNPP) accident in 20111. Consequently, freshwater environments are contaminated, and freshwater fish have accumulated radiocesium into their bodies2,3,4. The activity concentrations of 137Cs in freshwater fishes are relatively high compared to those in marine fishes due to the balance of osmoregulation and excretion of hypotonic urine5. Even now, more than 9 years after the accident, some of freshwater fish still contains a higher concentration of total radiocesium than the Japanese standard limit for general foods of 100 Bq kg−1 which have been in force since April 1, 2012. In fact, government restrictions and self-imposed restrictions of distribution for freshwater fish have been continued in some regions in Japan. Fish is one of the sources of dietary protein, and the importance of inland water fisheries has globally increased6. This raises concerns about internal exposure of humans through consumption of contaminated freshwater fish.

Food cultures have developed all over the world, and edible parts and cooking methods differ among localities. The common edible part of most fish is muscle, and thus muscle tissue is often selected for the measurement of radiocesium activity concentrations7. There is also a custom, however, to eat a whole fish, without removing the bones, skins, internal organs and so on. Because the internal radiation dose of humans is influenced by their dietary habits, information on the distribution of radiocesium in fish bodies is useful for improving the internal dose estimation.

There is some interest in environmental protection from ionizing radiation due to the FDNPP accident. The ICRP has said that the purpose of the radiation protection of the environment is to maintain biological diversity, to conserve species, or to keep the health and status of natural habitats, communities, and ecosystems8. In order to achieve this purpose, the estimation of dose–response relationship is essential, and this relationship is built on dose estimation of environmental organisms. In most cases, fish muscle tissue is selected for the measurement of radiocesium activity concentrations because of its consumption by the people7, but a homogeneous distribution of radiocesium within a fish body is unlikely. Therefore, data on distribution of radiocesium in various tissues and organs will also provide valuable information regarding dose estimation of freshwater fish species.

Lake Inba in Chiba Prefecture, Japan, was contaminated with radioactive fallout from the FDNPP accident9. Fish living in Lake Inba have been mainly contaminated with 137Cs released by the accident. In this study, the radioactivity concentrations of 137Cs were measured in muscle, bone, internal organ, spawn, and milt of 8 kinds of freshwater fishes which were caught in Lake Inba. The amounts of 137Cs in each tissue and organ were calculated from the activity concentrations of 137Cs and weights of tissues and organs, and then the distributions of 137Cs were determined as the ratio of 137Cs amount in each part relative to its amount in the whole body. In addition, factors influencing the 137Cs distribution were discussed.

Results and discussion

Status of 137Cs contamination in Lake Inba

Since all the fish samples were from Lake Inba, status of 137Cs contamination in this lake is described first. Water from Lake Inba is used as raw water for drinking-water, and the quality is of a concern to local residents. The Ministry of the Environment (MOE) has investigated the activity concentration of 137Cs in water of this lake10. Since the ministry’s investigation started, 137Cs concentrations in the water column have all been below the screening levels for drinking-water for gross beta activity, 1 Bq L−111. The MOE results gave arithmetical mean values of the concentration as 0.029 ± 0.012 Bq L−1 in the surface (top 0.1 m) and 0.031 ± 0.017 Bq L−1 in the bottom (0.5 m from the surface) layers. It should be noted that these values of 137Cs were obtained from unfiltered water of the surface and bottom layers. Lake Inba is shallow, and mixing of the water body by winds can cause almost the same concentration of 137Cs to be reached in both layers. Although the sediment in Lake Inba is a sink for 137Cs12, there might be little elution of 137Cs from the sediment to the water column. Little variation in the 137Cs concentrations after 2015 suggested that fish caught in this study had been living under an equilibrium or an apparent equilibrium condition for 137Cs in the water column.

Distribution of 137Cs in freshwater fishes

Distributions of 137Cs were significantly different among tissues and organs (Fig. 1, P < 0.01). The Steel–Dwass post-hoc test showed significant differences except between bone and milt, internal organs and spawn, and spawn and milt. The highest distribution of 137Cs was found in muscle. The arithmetic mean values of the distribution were decreased in the order of muscle tissue (54 ± 12%), internal organs (7.8 ± 4.6%), spawn (7.4 ± 5.4%), milt (3.2 ± 2.1%) and bone (1.2 ± 0.58%). The highest concentration of 137Cs or stable Cs being in muscle tissue has also been observed in various organisms such as humans13, mice14, rats15, cattle16 and wild boars17. Accumulation of 137Cs in muscle seems reasonably common in fish and other animals. Although the sum of values for the distribution was less than 100%, the remaining 137Cs, approximately 26%, might be distributed to other parts such as the head, gills, skin and fins.

Figure 1
figure 1

Boxplot showing the distribution of 137Cs in muscle (n = 308), bone (n = 28), internal organs (n = 94), spawn (n = 38) and milt (n = 7) of all freshwater fishes caught in this study. Each box indicates the inter-quartile range. The line inside the box shows the median. Marks “×” and “open circles” are mean values and outliers, respectively. Significant difference was found for the Kruskal–Wallis test (P < 0.01). The Steel–Dwass post-hoc test showed significant differences except between bone and milt, internal organ and spawn, and spawn and milt.

Full size image

Although the proportion of 137Cs in milt was low at 3.2% of the total body burden, the activity concentration of 137Cs in this organ was found to be 1.5 times higher on average compared to the whole body. Saxén and Koskelainen18 reported that the activity concentration of 137Cs in sperm of gutted fish was higher than that in the whole fish, but this finding was based on limited data (n = 5). The present study also had a similar result, and the relatively small value of the 137Cs distribution in milt (3.2%) must have been due to the small weight of milt compared to body weight. Since germ cells are sensitive to ionizing radiations19, accumulation of 137Cs in milt may affect reproduction. This study, however, examined only 7 milt samples, and detailed studies on the distribution and accumulation of 137Cs in milt are required in heavily contaminated areas.

Physiology and ecology of fish vary by species, and thus differences in fish species may affect the level of the 137Cs distribution. Since it was shown that 137Cs taken in by all freshwater fishes studied was mainly distributed in muscle (Fig. 1), levels of the 137Cs distribution in muscle were compared among the species (Fig. 2). A significant difference was found (p < 0.01). The highest mean value was 71 ± 13% for largemouth bass, and then the level decreased in the order of 69 ± 14% for snakehead, 63 ± 17% for channel catfish, 62 ± 14% for common carp, 58 ± 6.5% for barbel steed, 57 ± 7.7% for silver carp, 53 ± 4.7% for bluegill, and 50 ± 10% for crucian carp. Similar results have been reported for perch and pike from lakes in Finland, and the distribution of 137Cs in muscle of these fish species were 68%-75%. The proportion of 137Cs in muscle of crucian carp was significantly lower than that of all other species except bluegill (Steel–Dwass’s test, p < 0.05). Similarly, the 137Cs distribution level for bluegill was significantly lower compared with levels for largemouth bass and snakehead, but no differences were observed among other species.

Figure 2
figure 2

Boxplot showing the proportion of 137Cs in muscle of 8 kinds of fishes. Abbreviations of SH, LB, CF, BG, CC, BS, SR, and CR are snakehead, largemouth bass, channel catfish, bluegill, common carp, barbel steed, silver carp, and crucian carp, respectively. The numbers in parentheses are the number of samples. Significant difference was found for the Kruskal–Wallis test (P < 0.01). Lowercase letters above boxes indicate significant differences from the post-hoc Steel–Dwass test.

Full size image

These differences in the 137Cs distribution among fish species can be explained by food habits. Trophic levels (TLs) of bluegill and crucian carp are 3.2 and 3.1, respectively (FishBase, http://www.fishbase.org), and these values are lower than those of snakehead (TL = 4.4) and largemouth bass (TL = 3.8). Because differences in trophic levels mean differences in food habits, these results suggest that food habits affect the proportion of 137Cs in muscle. In addition, piscivorous fish muscle is known to have a higher activity concentration of 137Cs compared to non-piscivorous fish muscle20,21,22, suggesting the former easily accumulate 137Cs in muscle tissue. Food habits of fish change during their growth process. For example, juvenile snakehead mainly consume zooplankton, and the mature feed on insects, shrimps and fish23. Although there was no direct evidence for the difference in food habits between omnivorous bluegill and crucian carp and piscivorous snakehead and largemouth bass in this study, their dietary foods might have been different judging from their body size and physical characteristics. The minimum body length of snakehead, Channa striata, at maturation is estimated to be from 17 to 19 cm24. In this study, the minimum body length was 24.5 cm. Thus, all of the snakehead caught in Lake Inba must have been mature although it is necessary to consider differences in habitat and species. Mature snakehead feed mostly on other fish, frogs, crustaceans and some insects25. Similarly, the minimum body length of largemouth bass was 25.2 cm (29.4 cm of total length) in this study, and this largemouth bass was holding eggs. Paragamian26 reported that largemouth bass with a total length of above 20 cm consumed smaller fish more frequently. On the other hand, crucian carp are omnivorous. C. auratus langsdorfii, a kind of crucian carp, feed on plant fragments and zooplankton27, and C. cuvieri, also a kind of crucian carp, have pharyngeal teeth which are adaptive for phytoplankton feeding28.

The digestive system might also play an important role as a factor affecting the 137Cs distribution in fish bodies. A carnivorous digestive system differs from an omnivorous one. It is known that intestine lengths of omnivorous fish are longer than carnivorous fish29,30. The longer digestive tract is thought to have a higher surface area and allow a longer retention time of the food31. Consequently, the longer digestive system is useful for enhancing nutrient absorption from indigestible food. The intestines of omnivorous and herbivorous fishes often exhibit higher carbohydrase activities to digest lower quality food such as fiber-rich diets32, whereas abilities to digest more easily digestible food such as protein-rich diets are low compared to carnivore fish33. The proteins are the most important muscle constituents34, and efficient digestion of muscle by carnivore fish cause efficient nutrient and mineral absorption. Muscle also contains a constant level of potassium35, and this mineral has important physiological roles36,37. Therefore, potassium is an essential element for muscle function. Potassium in muscle is more efficiently absorbed by piscivorous fish than omnivorous fish, and it is immediately distributed to the fish's own muscle tissue. Similarly, 137Cs which is a homologous element of potassium may be absorbed and distributed to muscle tissue through digestion. This may be the reason that the difference in digestive system causes differences in the distribution of 137Cs among species.

Factors affecting the proportion of 137Cs in muscle

Wide range values of the proportion of 137Cs in muscle were observed for crucian carp (Fig. 2, 22.8–91.2%). To determine the factors affecting the proportion of 137Cs in muscle, effects of season, length of exposure time, and physical characteristics (body length, fresh weight, and condition factor) on the 137Cs proportion were investigated.

The activity concentrations of radionuclides in the muscle tissues vary seasonally for wild animals such as roe deer38,39, wild boar38,40, and black bear40, and it is considered that changes in food diets of these animals caused the seasonal variation. However, limited data are available for the seasonal variation in the 137Cs concentration and distribution of freshwater fish. To fill this gap, both the activity concentration of 137Cs and the proportion of 137Cs in muscle tissue were compared among the population subgroups of crucian carp caught each month (Fig. 3). The results showed no differences for both the concentration and the distribution. Although these data included values of both male and female crucian carp, different results may obtain when using only female data. Eggs of crucian carp develop from autumn to spring in Lake Inba, and the proportion of 137Cs in muscle may decrease during the incubation period by a dilution effect. In addition, a decrease in the 137Cs proportion was also expected after the spawning. However, no seasonal variation of the proportion of 137Cs even in female muscle was observed (Fig. S2). These results suggested that seasonal changes in physiology, food habits and spawning did not affect the proportion of 137Cs in muscle for crucian carp inhabiting Lake Inba.

Figure 3
figure 3

Boxplots showing the seasonal variations of the 137Cs activity concentrations (a) and the proportion of 137Cs in the muscle tissues of crucian carp (b) which were caught each month and assayed. Both the concentration and distribution had no difference.

Full size image

The effects of the length of exposure time and body size (length and weight) on the proportion of 137Cs in muscle tissue were examined. The length of exposure time is known to affect the accumulation of heavy metals by fish41. Together with length of exposure time, fish size must also have an effect, since fish grow over time. In fact, the activity concentration of 137Cs in muscle increases with fish weight42 and total length43, and this is known as ‘size effect’. In this study, estimated age, body length, and fresh weight were used as indexes of the time of exposure. The estimated ages were determined from the growth curve of crucian carp in Lake Inba. The relationships between these three indexes and the proportion of 137Cs in muscle of crucian carp are shown in Fig. 4. No correlations were observed for any indexes. This might be partly because crucian carp are not predatory, and no size effect for 137Cs accumulation has been found in predatory fish44. In addition, the individuals examined were large (> 13.1 cm in body length), and they must have undergone the ontogenetic diet shift. Food and feeding habits of freshwater fish shift ontogenically23,45,46, and the major diet component of crucian carp changed changes from plankton to insects with increasing body size47. Changes in food habits occur in individuals at a body length of approximately 2 cm48, which was much smaller than the examined individuals (> 13.1 cm in body length). Furthermore, crucian carp examined in this study include both C. auratus langsdorfii and C. cuvieri. Despite being the same genus, food habits of C. auratus langsdorfii and C. cuvieri are different, which might have caused unclear relationships between the 137Cs distribution and size.

Figure 4
figure 4

Relationships between the indexes of the growth and the proportion of 137Cs in muscle of crucian carp. The estimated (a) age, (b) body length, and (c) fresh weight were used as the indexes. The number of samples was 197. No correlations were observed for any indexes.

Full size image

Finally, the condition factor, which is an indicator of nutritional status, was found to influence 137Cs distribution (Fig. 5). Even if fish species live in the same environment, nutritional status varies from individual to individual. However, there are few data on the relationship between nutritional status and the 137Cs distribution. Based on the assumption that a heavier fish of a given length is in better condition49, the condition factor expressed as the ratio of total body weight and the cube of body length is often used as the indicator of nutritional status and relative health of fish. Therefore, the condition factor was used as a way to estimate nutritional status, and the relationship between condition factor and the proportion of 137Cs in muscle of crucian carp was determined (Fig. 5). Spearman’s correlation coefficient showed a positive correlation between the two variables (P < 0.01). When the data for the condition factor were divided into quartiles, it was found that the population classified into the first quartile (Q1) had a lower 137Cs distribution than the other three populations (Fig. S3). The condition factor is known to correlate positively to total lipid content of fish50. That is, fish with low condition factor values have small amounts of lipids in their muscle tissues. Fats which are a type of lipid are not a reservoir for 137Cs in wild boar51. Although there are differences in fat components between wild boar and fish, the accumulation of 137Cs in fish lipids can be considered as hard to achieve. The fact that a small amount of 137Cs was distributed in muscle despite the low amount of lipids means that it is hard for 137Cs to accumulate in muscle tissue of fish with poor nutritional status. Since crucian carp were caught in a certain area during the entire study period, there would not be much difference in the quality of the food. Quantity of food intake, therefore, is probably an important factor affecting the proportion of 137Cs in muscle at least for crucian carp in Lake Inba.

Figure 5
figure 5

Relationship between the condition factor and the proportion of 137Cs in muscle of crucian carp. Positive correlation was observed between the two variables (n = 197).

Full size image

Conclusions

This is the first study to document that nutritional status of fish is an important factor affecting the proportion of 137Cs in muscle tissue in addition to differences in fish species and trophic level. Previously there were only limited data on the distribution of 137Cs in each part of a fish body compared to data on activity concentrations of 137Cs in fish muscle. In this study, the 137Cs distribution within freshwater fish bodies were investigated in Lake Inba. The obtained results showed that the 137Cs distribution differed among body parts. The highest proportion was found in muscle, and the proportion of 137Cs decreased in the order of internal organs, spawn, milt, and bone. Moreover, the proportion of 137Cs in muscle differed among fish species. Fish with high trophic levels easily distributed more 137Cs in their muscle tissue than fish with low trophic levels. To determine factors affecting the 137Cs distribution in muscle, the effects of season, time of exposure and physical characteristics on the proportion of 137Cs in muscle were investigated using crucian carp. Only the condition factor as an indicator of nutritional status affected the 137Cs distribution. Fish with poor nutritional status had less 137Cs distributed in their muscle tissue than fish with good nutritional status. The 137Cs distribution within fish bodies is potentially useful for assessing human risks through freshwater fish consumption and for estimating dose of the fish themselves.

Materials and methods

Study area

The study area was Lake Inba which is located in the northwest part of Chiba Prefecture, Japan (latitude 35° 48′ north and longitude 140° 15′ east) and monthly sampling of fish was done from September 2016 until March 2020. This lake is surrounded by paddy fields, and its water has agricultural, industrial and drinking water uses. This lake is a polymictic lake with a surface area of 11.55 km2. The water depth is 1.7 m on average (maximum 2.5 m) and residence time is about 0.08 years52. The water depth is controlled by the local government. The lake is about 220 km away from FDNPP. Lake Inba and its surrounding areas were contaminated with fallout radionuclides after the FDNPP accident. According to the report from Ministry of Education, Culture, Sports, Science and Technology (MEXT), deposition densities of radiocesium (137Cs + 134Cs) in the surrounding environment was from 1.0 × 104 to 3.0 × 104 Bq m−2 in September 20119.

Fish samples and their processing

Eight kinds of freshwater fishes were caught using stake nets placed close to the shore of Lake Inba starting in September 2016. A maximum of 4 nets were set up for one monthly sampling. Each net was put in a separate fixed position from the previous one-day-sampling period. Both surface and bottom fishes were caught at the same time because the lake is shallow. The kinds caught were: snakehead (Channa argus), largemouth bass (Micropterus salmoides), channel catfish (Ictalurus punctatus), bluegill (Lepomis macrochirus), common carp (Cyprinus carpio), barbel steed (Hemibarbus barbus), silver carp (Hypophthalmichthys molitrix), and crucian carp (Carassius spp.: C. auratus langsdorfii and C. cuvieri).

All fish were caught by a researcher who is a member of a local fishery cooperative having license of a class 5 common fishery. The experimental protocols followed the regulations concerning animal experiments of National Institutes for Quantum and Radiological Science and Technology (QST) and were approved by the Institutional Animal Care and Use Ethics Committee of QST. The study was approved by the President of QST. All experiments were performed in accordance with the ARRIVE guidelines (https://www.nc3rs.org.uk/arrive-guidelines) and relevant regulations.

The fish were stored onto the ice cubes in a fishing cooler until they were brought back to the laboratory. In accordance with the Invasive Alien Species Act in Japan (Law No. 78, 2004) largemouth bass, channel catfish, and bluegill were transported after the kill. Total length, body length, body depth, body width, and fresh weight were measured for most fish samples in the laboratory. The condition factor (K) of crucian carp was calculated from the following equation [Eq. (1)]:

$$K = W/L^{b} \times 100$$
(1)

where W is the fresh weight of crucian carp in grams, and L is the body length in centimeter. The b value is a factor in the standard allometric equation

$$W = aL^{b}$$
(2)

and it was determined empirically using data collected in this study (Fig. S1). These fish samples were kept at 4 °C in a laboratory refrigerator until the dissection in the following days.

Fish bodies were washed with a brush and then dissected to separate muscle, bone, internal organs, spawn, milt and the remaining parts including the head, gills, skin fins, and muscle residues. Contents of their digestive organ were removed during the dissection period. However, fish having no stomach, i.e. common carp (Cyprinus carpio) and crucian carp (Carassius spp.: C. auratus langsdorfii and C. cuvieri), were processed without the removal of undigested contents. Bony parts excluding the head were immersed in 80˚C water until the soft tissues were discolored. After that, tissues attached on bones were careful removed by using a nylon brush. We did not consider loss of 137Cs from bone parts by this heating process. Finally, dissected parts were lyophilized. To obtain water contents, the weight of these parts was measured before and after lyophilization. The lyophilized samples were powdered with a grinder (Labo Milser LM-PLUS, Osaka Chemical Co., Ltd.), and the powdered samples were packed into a U8 polypropylene container (D: 55 mm × H: 64 mm, 100 mL) for the analysis of 137Cs. The mass of the measured samples ranged from 0.72 to 98 g-dry with an average of 20 g-dry. All parts of an individual fish were combined together and mixed well to determine the total 137Cs in whole-body after the measurement of 137Cs activity in each part.

Growth curve of crucian carp

The growth curve of crucian carp was determined from the relationship between age and body length of some specimens. The age of crucian carp was determined with otoliths (ear stones), collected from 62 individuals, by counting annuli on the otolith. Otoliths were removed from head parts and enclosed in a plastic resin. Thin sections were prepared, and otolith rings were counted by Marino Research Co., Ltd. (Mie, Japan). Then, ages of all the captured crucian carp in this study were estimated using the obtained growth curve. These fish samples for age determination were not used for the analysis of the distribution of 137Cs because the amount of 137Cs in whole-body could not be determined due to the lack of the head part.

The von Bertalanffy growth function (VBGF) was used for modeled fish growth. This model is given by the following equation:

$$L \, = \, L_{\infty } \left[ {{1} - {\text{exp }}\left\{ { - k\left( {t - t_{0} } \right)} \right\}} \right]$$
(3)

where L is the expected body length at age t, L∞ is the asymptotic length, k is the growth rate coefficient, and t0 is age at which the fish would have had body length of zero (that is, t0 is a modeling artifact to adjust the equation for the initial size of the fish). Three parameters, L∞, k, and t0, in VBGF were obtained by non-linear least squares and maximum likelihood estimation. This calculation was carried out by the solver add-in in Microsoft Excel for Microsoft 365. The representative body length of each age was calculated from the obtained model formula.

Analysis of 137Cs activity concentrations

The activity concentrations of 137Cs were determined using high-purity germanium detectors (GMX- and GEM-types, ORTEC, SEIKO EG&G Co., Ltd.; GC4018 and GX-4018, CANBERRA Industries Inc.) with Gamma Station (SEIKO EG&G) or Spectrum Explorer software (CANBERRA). The relative efficiencies of the germanium detectors used were 25.0%, 35.0%, 45.9% and ≥ 40.0%, respectively. Each detector was calibrated for energy and detection efficiency with volume radioactivity standard gamma source, MX033U8PP (Japan Radioisotope Association). The measurement accuracy was confirmed using the standard reference material JSAC-047 (Japanese Society for Analytical Chemistry) approximately once a month. Self-absorption correction factors were obtained using a mathematical equation equipped in each software. The measurement times ranged from 10,000 to 227,000 s, and the activity concentrations with relative errors of < 5% were obtained from these measurement times in most cases. The values of measurements were decay-corrected for radioactive decay to the sample collection day. The obtained concentrations on a dry-weight basis were converted to those on a fresh-weight basis with water content values.

Distribution of 137Cs in fish body

Distribution of 137Cs in fish body was shown as the percent of the body burden contained in dissected parts using the following equation [Eq. (4)]:

$${\text{Distribution }}\;{\text{of}}\;^{{{137}}} {\text{Cs }}\left( \% \right) \, = Q_{{{\text{tissue}}}} /Q_{{\text{whole - body}}} \times { 1}00$$
(4)

where Qtissue is the amount of 137Cs in each tissue and organ, and Qwhole-body is that in whole body. The amounts of 137Cs were determined as the product of the fresh weights and activity concentrations of 137Cs.

Data analysis

The Bartlett test was conducted to check homogeneity of variance before the analysis of variance. As a result of that test, the Kruskal–Wallis test was adopted to check if the means of several groups were significantly different from each other, unless otherwise noted. When a significant difference was confirmed, differences between multiple group means were explored using Steel–Dwass post hoc comparisons. When the p-value of the Bartlett test was more than 0.05, one-way analysis of variance (ANOVA) and the Tukey‒Kramer test were conducted. Regression analysis was carried out to test the strength of the association between the two quantitative variables. These analyses were performed using R software version 3.3.353, but all figures were generated by Microsoft Excel.