Seasonal variation in radiocesium levels of largemouth bass (Micropterus salmoides): Implications for humans and sensitive wildlife species

INTRODUCTION

Contaminants released into the environment invariably move to and become concentrated within aquatic ecosystems. Thus, the consumption of fish is a potentially important mechanism by which environmental contaminants may enter food chains leading to humans or wildlife species. Because of the widespread occurrence of radiocesium (¹³⁷Cs) due to global fallout and accidental releases at nuclear facilities, knowledge regarding the bioaccumulation of this contaminant in fish has applications for biomonitoring, environmental remediaton, and risk assessment studies [1]. The process of ¹³⁷Cs bioaccumulation in fish has been the subject of numerous investigations during the past three decades [2]. Recently, these types of investigations have received increased attention following the Chernobyl nuclear accident [3-5].

One relevant aspect of ¹³⁷Cs bioaccumulation in fish, demonstrated by Kolehmainen [6], was that levels of this contaminant varied seasonally in bluegill (Lepomis marcrochirus) from a lake in the southeastern United States. Seasonal changes in ¹³⁷Cs levels have also been observed for several fish species from a lake in southern Sweden [7]. Understanding such seasonal changes in ¹³⁷Cs levels of fish may be relevant to risk assessments for wildlife that utilize fish on a seasonal basis for activities such as the feeding of offspring. However, despite the numerous investigations regarding ¹³⁷Cs bioaccumulation in fish, seasonal differences in levels of this contaminant have seldom been reported or evaluated.

Concentrations of ¹³⁷Cs in the water column are known to vary seasonally in a system of former nuclear reactor cooling reservoirs located on the Savannah River Site (SRS), near Aiken, South Carolina [8]. These reservoirs represent an excellent opportunity to address questions concerning the seasonal pattern of ¹³⁷Cs bioaccumulation in fish. Moreover, information concerning seasonal patterns of ¹³⁷Cs in fish from these SRS reservoirs has immediate application to the making of management decisions. For example, fish in these reservoirs are used seasonally by sensitive wildlife species, such as the threatened bald eagle (Haliaeetus leucocephalus), to feed nestlings. In addition, since production activities have ceased at the SRS, there is now interest in the possibility of opening the site's reservoirs to public recreational fishing.

Previous investigations in SRS reservoir systems have demonstrated seasonal differences in ¹³⁷Cs levels in a fish species functioning at an intermediate trophic level (mosquitofish, Gambusia holbrooki; [9]) as well as in a species of migratory waterfowl (American coots, Fulica americana; [10]). Based on these observations, we hypothesized that ¹³⁷Cs levels would also vary seasonally in other species within this ecosystem. The purpose of this investigation was to examine seasonal changes in ¹³⁷Cs levels of largemouth bass (Micropterus salmoides) collected over a 12-month period. The relatively high trophic position at which largemouth bass feed, along with its popularity as a sport fish and its importance as a food item for wildlife (e.g., bald eagles), makes this species particularly important to study at the SRS site and elsewhere throughout its range.

MATERIALS AND METHODS

Data analysis

The effects of month of sampling, sex, age, and log-transformed body mass on total body levels of ¹³⁷Cs were analyzed with analysis of covariance. Because few very young or very old bass were collected (Fig. 1A), and an age effect may not necessarily be linear [16], age was coded into three roughly equal classes (≤3, 4, and ≥5 years old). Month, sex, and age were tested as fixed effects, and body mass (log wet mass) was included as a covariate (SAS Ver 6.12 proc. GLM [17]). Because the catchability of fish varied by sex, age, and month, the analysis of covariance was highly unbalanced. In particular, sex varied greatly among months (likelihood ratio χ² = 43.3, df = 11, P = 0.001; Fig. 1B). Similarly, a larger number of older fish were female, while young fish were disproportionately male (χ² = 56.7, df = 2, P < 0.0000001; Fig. 1A). Females also tended to be larger compared to males (Kruskal-Wallis χ² = 132.5, df = 1, P < 0.0000001). Thus, the main effects of month, sex, age, and size could not be independently estimated. For example, a portion of the variance in ¹³⁷Cs concentration could be ascribed to either mass or sex. Furthermore, the imbalance in the design was strong enough that Type III tests would have no simple interpretation. Thus, we conducted two sequential tests for the effect of each factor. These tests included Type I tests, which were conducted with each factor entered first into the model, and a Type II test for that same factor after first including the other three factors. Post hoc comparisons of least-square means (correcting for body mass) among groups were performed with Bonferroni adjustments [17].

The previous analysis of covariance treated month of sampling as 12 independent categories rather than as a temporally ordered variable. We expected a possible seasonal effect of date rather than a linear trend, so we fit three forms of cyclic models to the ¹³⁷Cs concentrations. First, following Kolehmainen [6] and Brisbin et al. [10], an annual sine wave was fit. The period was fixed at 365 d (not the 360 d in Kolehmainen [6] and Brisbin et al. [10]). An overall mean ¹³⁷Cs concentration, amplitude of seasonal variation, and offset (date of peak) were fit with nonlinear least squares (SAS proc. NLIN [17]). Because the effects of sex, log of body mass, and age on ¹³⁷Cs concentrations were small relative to that of month (see the following discussion), this and the subsequent two models were fit without including the effects of sex, size, and age.

There is no theoretical reason why seasonal variation in ¹³⁷Cs concentrations necessarily needs to follow a symmetrical (six months of increase and six months of decrease) such as that described by a sine wave. To allow for the possibility of temporal asymmetry, we fit a second model (asymmetrical sawtooth)—sing the ¹³⁷Cs concentrations of the peak and the trough, the date of the peak, and the number of days between the peak and trough—with linear changes between the peak and trough. Because the second model had four parameters while the first had only three, a third model was fit (symmetrical sawtooth). This third model was a subset of the second model, fitting the ¹³⁷Cs concentrations of the peak and trough and the date of the peak but forcing a symmetric 182.5 d between peak and trough. The second and third models, along with the analysis of variance model fitting a separate mean for each month, formed a nested series of models with 3, 4, and 12 degrees of freedom. Thus, the significance of the additional terms could be tested with analysis of deviance. Under the null hypothesis of no effect of a term, the deviance (difference in lack of fit, in this case SS error) follows a χ² distribution with degrees of freedom (df) equal to the difference in df between the models compared [18].

Table Table 1.. Monthly variation in stomach contents of largemouth bass collected from Pond B on the U.S. Department of Energy's Savannah River Site from November 1973 to October 1974

Month	Sample size	% Stomachs with food	% With fisha	% With invertebrates
January	47	11.8	0	100
February	51	10.6	0	100
March	49	16.3	25	75
April	47	8.1	50	50
May	50	20.0	100	0
June	46	13.0	100	0
July	50	40.0	100	10
August	49	38.8	100	0
September	52	25.0	100	0
October	50	18.0	56	44
November	49	20.4	20	80
December	50	20.0	10	100

• ^a Frequencies of stomachs containing invertebrates and fish are calculated as percentages of all stomachs containing food.

RESULTS

Frequency distribution of ¹³⁷Cs levels

The overall distribution of ¹³⁷Cs values is illustrated in Figure 2. Whole-body concentrations of ¹³⁷Cs ranged from 2.95 to 12.60 Bq/g, with a mean of 7.09 and median of 7.01 Bq/g. In contrast to the asymmetric (often lognormal) distributions reported for other species [19], the distribution of ¹³⁷Cs in bass was relatively symmetrical with more observations in the tails of the distribution than in a corresponding normal distribution. Separate distributions for each month (not shown) were similarly symmetric.

DISCUSSION

Levels of ¹³⁷Cs in bass differed between sexes (males > females), even after correcting for differences in body size. Previous investigations conducted at SRS have also observed a significant effect of sex on ¹³⁷Cs bioaccumulation in fish [9, 20] whereas McCreedy et al. [1] failed to demonstrate this effect. Interpretation of these findings is confounded by the differences in sex ratios within age classes and months (Fig. 1) and by the fact that females were much heavier compared to males (337.22 g ± 6.06 vs 251.41 ± 4.35 g; mean ± SE). However, in our investigation, sex, age, and body mass were much less important than the month of sample collection since these factors accounted for only a small percentage of the variability in ¹³⁷Cs concentrations of bass (Table 1).

Results supported our hypothesis that ¹³⁷Cs concentrations in largemouth bass from Pond B would vary over the course of a year. To date, only one other investigation has examined ¹³⁷Cs levels in a species of fish throughout the year in the southeastern United States [6]. Similar to our findings, Kolehmainen [6] demonstrated that levels of ¹³⁷Cs in bluegill from a lake in Tennessee varied seasonally and also reached a peak in February. The percent increase between the minimum and maximum monthly values was also similar for bass in our investigation (33%) compared to bluegill (approx. 37%). Kolehmainen [6] estimated that minimum monthly values for ¹³⁷Cs in bluegill occur in August, whereas minimum ¹³⁷Cs levels in bass from our investigation did not occur until October.

Holloman et al. [9] examined concentrations of ¹³⁷Cs in mosquitofish collected from Pond B during four months (July and November 1991 and February and April 1992). In that investigation, the lowest ¹³⁷Cs levels in mosquitofish for the four monthly samples occurred in July, and the highest ¹³⁷Cs levels occurred in February. Concentrations of ¹³⁷Cs in mosquitofish increased steadily from July through February before decreasing rapidly from February to April. Unfortunately, it is not possible to determine the exact months during which maximum and minimum body concentrations of ¹³⁷Cs occur in mosquitofish from Pond B. However, it appears that ¹³⁷Cs levels in this species peak at about the same time as in largemouth bass (February) but decrease more rapidly compared to bass (minimum in July for mosquitofish vs October for bass).

Kolehmainen [6] used a sinusoidal curve to describe the observed seasonal changes in ¹³⁷Cs levels in bluegill. Our analyses demonstrate that seasonal variation in ¹³⁷Cs levels in bass do not follow an extrinsically driven sinusoidal pattern. Instead, ¹³⁷Cs levels in bass vary in an asymmetrical manner that, in our investigation, was described by an asymmetrical sawtooth model. An asymmetrical pattern of fluctuation in ¹³⁷Cs levels has also been reported for yellow-bellied turtles (Trachemys scripta) from Pond B [21]. The asymmetric seasonal pattern in ¹³⁷Cs concentrations of bass likely reflects a balance between uptake and excretion processes, each of which vary seasonally. Thus, factors that may potentially influence the uptake or excretion of ¹³⁷Cs from bass warrant further consideration.

Concentrations of ¹³⁷Cs in the water of Pond B [22] and the nearby Par Pond reservoir [8] reach a maximum in early to mid-fall and are lowest in early to mid-spring. This seasonal change in ¹³⁷Cs availability in water results from the remobilization of ¹³⁷Cs via ion exchange displacement from the sediments during thermal stratification in the summer months [13]. Thus, ¹³⁷Cs concentrations in bass reach maximum levels at a time when levels in water are at or near their minimum values. This demonstrates that direct exposure to ¹³⁷Cs in the water column is not an important factor in causing seasonal variation in levels of this contaminant in bass.

Table Table 3.. Analysis of deviance for sequential models for intra-annual variation in ¹³⁷Cs concentrations in largemouth bass

• ^a The base model for each series of comparisons is a single parameter of the overall mean.
• ^b Within each series, the reduction in the error sum of squares (lack of fit) for each sequential addition to the model is tested against a χ² distribution.

Numerous investigations have demonstrated that increasing water temperatures result in decreasing ¹³⁷Cs levels [7, 23-26] via the influence of temperature on metabolism. Rowan et al. [23] developed a bioenergetics model to explain the seasonality of ¹³⁷Cs levels in fish using their own data for rock bass (Ambloplites rupestris) and data for bluegills reported by Kolehmainen [6]. This model indicated that temperature was the most important factor determining the seasonality of ¹³⁷Cs levels in these two species, primarily because the effects of increased temperature on metabolism results in increased rates of ¹³⁷Cs elimination. Peters and Newman [24] showed that the elimination rate of ¹³⁷Cs in chronically contaminated large-mouth bass (i.e., fish living in Pond B) increased with temperature. In that investigation, elimination rates were roughly 2.5 times higher in bass at 26°C than at 15°C.

Although monthly water temperatures were not recorded in Pond B during the time of our investigation, water temperatures in nearby Par Pond peak in June-August and are lowest during January-March [27]. The fact that ¹³⁷Cs levels of bass from Pond B peak when water temperatures are likely lowest suggests that the increase phase of the seasonal cycle in ¹³⁷Cs levels in bass is resulting from reduced elimination of this contaminant due to lower water temperatures. However, the lowest ¹³⁷Cs levels in bass from Pond B were observed in October, at least two months following the peak in water temperature, when elimination of ¹³⁷Cs would be expected to be at a maximum. Thus, while water temperature appears to be important, this factor alone does not explain seasonal changes in ¹³⁷Cs levels of bass, and other physiological and ecological factors likely influence these changes.

We hypothesized that seasonal variation in ¹³⁷Cs levels of bass in Pond B could be related to changes in feeding habits. However, our data regarding stomach contents of bass, in conjunction with known ¹³⁷Cs concentrations in prey items, do not support this hypothesis. For example, bass from Pond B were bottom feeders with a diet consisting primarily of benthic invertebrates during the colder months (November-March) and were piscivorous during warmer months (May-September; Table 1). Concentrations of ¹³⁷Cs in invertebrates from Pond B have been shown to be much lower than for any fish species examined [11]. Thus, bass were consuming prey items that likely had the lowest levels of ¹³⁷Cs during the time in which levels of this contaminant reached a peak.

Regardless of the causes, seasonal differences in ¹³⁷Cs levels of bass at SRS have important implications for humans and wildlife and for the long-term management of radionuclide-contaminated habitats at this site. Although Pond B is currently not accessible to the public, there is continuing interest in making some of the areas on the SRS accessible for recreational fishing on a limited basis. Our results demonstrate that risk assessments for humans harvesting fish from areas with a history of ¹³⁷Cs contamination should take into account the season in which fish are harvested. Perhaps more relevant to the present situation, resident bald eagles are known to utilize fish, including bass, for feeding hatchlings from late December through late April (A.L. Bryan, pers. comm.). Consequently, hatchling eagles on the SRS are being fed bass at the time that ¹³⁷Cs levels in these fish reach their annual maximum.

It is important to note that all bass sampled in our investigation, regardless of month, sex, age, or body mass, were well above the European Economic Community [28] limits for human consumption (0.60 Bq/g wet muscle mass). Furthermore, ¹³⁷Cs levels for bass from our investigation were determined for whole body mass, and levels in consumable muscle mass would be even higher [29, 30]. Based on the known relationship between whole-body and muscle concentrations of ¹³⁷Cs in bass [31], ¹³⁷Cs levels in muscle are expected to be ˜31% higher than in whole-body determinations. Because of the relatively high ¹³⁷Cs levels in bass from Pond B, season of harvest is likely to become even more important as ¹³⁷Cs declines to lower levels (due to physical and biological processes) and approaches levels that may be acceptable for consumption by humans and wildlife at certain times of the year.

Although our data were collected only 10 years after the release of radionuclides into Pond B had ceased, ¹³⁷Cs in the reservoir has now completed one physical half-life (˜30 years) since the initial release. Despite the completion of a physical half-life, there is substantial evidence to suggest that relatively high levels of ¹³⁷Cs still remain in Pond B. For example, a survey conducted during 1994 concluded that the total ¹³⁷Cs inventory of Pond B was approx. 2.3 × 10¹¹ Bq [14]. In this same study, the total amount of ¹³⁷Cs within the standing crop of bass in the reservoir in 1994 was estimated at 5.2 × 10⁶ Bq, and the mean concentration in bass was found to be 4.4 Bq/g wet muscle mass [14]. Other recent estimates of ¹³⁷Cs concentrations in fish from Pond B include ˜4.24 Bq/g wet muscle mass (based on 21.22 Bq/g dry mass reported by Sugg et al. [32]) in bass collected during 1993 and 3.71 Bq/g wet muscle mass for yellow bullhead catfish (Ameiurus natalis) collected during 1995 [1].

Ecological half-life has been defined as “the amount of time required for a given level of isotope, once established and at equilibrium within a given ecosystem component, to decrease by 50% as a result of being either physically removed or rendered biologically unavailable in the system” [33]. Based on 22 years of data, Paller et al. [31] estimated the ecological half-life of ¹³⁷Cs in bass from Pond B to be ˜16.7 years. This estimate of ecological half-life can be used to make predictions regarding the amount of time necessary for ¹³⁷Cs levels in Pond B bass to reach acceptable levels for human consumption (≤0.60 Bq/g wet muscle mass). For example, mean ¹³⁷Cs levels during February (peak) and October (trough) would reach acceptable levels for human consumption in ˜69 and ˜63 years, respectively. Using the maximum ¹³⁷Cs value for all bass collected during our investigation (12.6 Bq/g whole-body mass or 16.5 Bq/g muscle mass) as a worst-case scenario, all bass would be expected to be safe for human consumption by the year 2054 (˜80 years from the time of this study).

Radiocesium is known to be more persistent in the Pond B reservoir than in other aquatic systems at SRS. For example, Paller et al. [31] demonstrated that the ecological half-life of ¹³⁷Cs in bass from Pond B was three times greater than in those from the nearby Par Pond reservoir and five times greater than in those from Steel Creek, a lotic system, on the SRS. Similar observations have also been made for Lepomis spp. from these systems [31]. Possible reasons for these differences include water turnover rate, water chemistry, presence of aquatic macrophytes, rates of groundwater discharge, and burial of ¹³⁷Cs by sedimentation [31, 33]. Thus, ¹³⁷Cs is lost relatively slowly from Pond B and should remain an important issue for any assessment of risk to humans and wildlife.

Our results demonstrate the need to consider a number of factors when monitoring levels of a contaminant such as ¹³⁷Cs. This is especially true when using the results of such monitoring to assess potential health risks to both humans and wildlife. Monitoring of ¹³⁷Cs levels in contaminated habitats at sites such as the SRS should be done over a temporal scale that encompasses all seasons, and risk assessments should be based on data collected during relevant time periods. Future investigations should also take into account factors such as seasonal changes in diet that may potentially influence the bioaccumulation of ¹³⁷Cs in bass.