Environ. Sci. Technol. 2009, 43, 6684–6690
A Probabilistic Assessment of the Chemical and Radiological Risks of Chronic Exposure to Uranium in Freshwater Ecosystems T E R E S A M A T H E W S , * ,†,‡,⊥ K A R I N E B E A U G E L I N - S E I L L E R , †,‡ JACQUELINE GARNIER-LAPLACE,† R O D O L P H E G I L B I N , †,§ C H R I S T E L L E A D A M , †,§ A N D CLAIRE DELLA-VEDOVA| Institut de Radioprotection et de Su ˆ rete´ Nucle´aire, Service d’E´tude du Comportement des Radionucle´ides dans les E´cosyste`mes, Cadarache, 13115 Saint-Paul-les-Durance, France and magelis, 6 rue F. Mistral 84160, Cadenet, France
Received March 16, 2009. Revised manuscript received June 26, 2009. Accepted July 9, 2009.
Uranium (U) presents a unique challenge for ecological risk assessments (ERA) because it induces both chemical and radiological toxicity, and the relative importance of these two toxicities differs among the various U source terms (i.e., natural, enriched, depleted). We present a method for the conversion between chemical concentrations (µg L-1) and radiological dose rates (µGy h-1) for a defined set of reference organisms, and apply this conversion method to previously derived chemical and radiological benchmarks to determine the extent to which these benchmarks ensure radiological and chemical protection, respectively, for U in freshwater ecosystems. Results show that the percentage of species radiologically protected by the chemical benchmark decreases with increasing degrees of U enrichment and with increasing periods of radioactive decay. In contrast, the freshwater ecosystem is almost never chemically protected by the radiological benchmark, regardless of the source term or decay period considered, confirming that the risks to the environment from uranium’s chemical toxicity generally outweigh those of its radiological toxicity. These results are relevant to developing water quality criteria that protect freshwater ecosystems from the various risks associated with the nuclear applications of U exploitation, and highlight the need for (1) further research on the speciation, bioavailability, and toxicity of U-series radionuclides under different environmental conditions, and (2) the adoption of both chemical and radiological benchmarks for coherent ERAs to be conducted in U-contaminated freshwater ecosystems.
* Corresponding author tel: (865) 241-9405; fax: (865)576-9938; e-mail:
[email protected]. † ´ tude du Comportement des Radionucle´ides dans les Service d’E ´ cosyste`mes. E ‡ Laboratoire de Mode´lisation Environnementale (baˆt. 159). § ´ cotoxicologie (baˆt. 186). Laboratoire de Radioe´cologie et d’E | magelis. ⊥ Current address: Oak Ridge National Laboratory, Environmental Sciences Division, Building 1504, Oak Ridge, TN 37831-6351. 6684
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 17, 2009
Introduction As world nuclear capacity is expected to increase by 80% in the next two decades (1), the demand for uranium (U) production is projected to grow, and the impacts of U development are raising public health as well as environmental concerns. Uranium is naturally found in the environment as a mixture of three radioisotopes: 238U, 235U, and 234 U. These isotopes differ both in their specific activities (radioactivity per mass isotope; Table S1) and in their relative proportions in the environment (Table S2). 238U and 235U are primordial radionuclides that form the start of radioactive decay chains, producing a long series of radioactive daughter products (including 234U; Figure S1; Table S1). The anthropogenic enrichment of 235U, the only naturally fissile isotope, relative to the other U isotopes, is a critical process in nuclear power generation and nuclear weapons production, and creates mixtures of U isotopes which may be more or less radioactive (Table S2); depleted uranium (DU) is a byproduct of the enrichment process. Ecological Risk Assessments (ERA) for U are confronted with a unique challenge in that U induces both chemical and radiological toxicity, and the relative importance of these toxicities depends on the contribution of the different isotopes considered. Diverse nuclear applications (nuclear fuel cycle, military use, etc.) take advantage of the properties of different isotopic compositions of U, and consequently discharges from these applications represent different radiological and chemical risks for the environment. A number of studies have considered the toxicological risks of DU due to its use in various military applications (2, 3). Such studies, as well as those on natural uranium (NU) (4), suggest that there are greater risks due to the chemical rather than radiological toxicity of U, although to our knowledge there has been no explicit or systematic quantification of these risks between the various source terms of U. Furthermore, until now, ERAs have considered the chemical and radiological toxicity of U separately (5, 6) but it is now recognized that the risk assessment methods generally recommended for nonradioactive chemicals are also applicable to radionuclides (7, 8). In this study, we present the results of a classic four-step screening level ERA (9, 10) to quantitatively characterize the relative chemical and radiological risks of chronic exposure by using benchmark levels of natural, depleted, and enriched U source terms for nonhuman biota in aquatic ecosystems. Assuming that background exposure for a naturally occurring substance poses no risk to the ecosystems, this method concerns added (and not total) risks (11). Because of U’s high solubility and therefore high mobility in the environment (12), freshwater ecosystems may be especially susceptible to the chronic impacts of U exposure. We describe methods for the conversion between aqueous uranium concentrations (µg L-1) and radiological dose rates (µGy h-1) and apply these methods to the Predicted No Effect Concentration (PNEC) and Predicted No Effect Dose Rate (PNEDR) for U, which are benchmarks that aim to protect freshwater ecosystems from the chemical and radiological toxicity of U, respectively. The conversion between chemical and radiological values allowed us to determine which benchmark value would be most protective for the freshwater ecosystem for each of the selected isotopic compositions of U. These results quantify, for the first time, the relative ecological risks of chemical vs radiological toxicity of the different compositions of U. The results are therefore relevant to developing water quality guidelines/criteria for protecting the environment from risks associated with the use of U in nuclear applications. 10.1021/es9005288 CCC: $40.75
2009 American Chemical Society
Published on Web 08/05/2009
FIGURE 1. Graphical depiction of reference organisms chosen to represent a temperate freshwater ecosystem.
Materials and Methods We adopted the U.S. EPA framework for Screening Level Ecological Risk Assessments (SLERA) (9, 10), which we have slightly modified to address the goals of the present study. The framework consists of four steps (Problem Formulation, Effects Analysis, Exposure Analysis, and Risk Characterization) which are treated in detail below. Step 1: Problem Formulation. Table S2 lists the four source terms, or isotopic compositions, of uranium considered in this study: NU, DU, LEU (Low Enriched Uranium), and HEU (Highly Enriched Uranium). These four source terms can be considered equivalent in terms of chemical toxicity, since the three U isotopes exhibit nearly identical electronic and chemical behavior. Because the effects due to chemical toxicity of a contaminant can be directly related to its aqueous concentration (µg L-1), the chemical toxicity of U in this study is related to the total U concentration in water (i.e., the sum of the concentrations of all U isotopes). In contrast, radiological toxicity is related to the amount of energy deposited in an organism over time (radiological dose rate, expressed in Gray (Gy) h-1), which has three important implications for the radiological risk assessment of U. First, the amount of energy emitted by the three different U isotopes varies as a function of their specific activity (Table S1). Dose rates received from a given concentration of U therefore depend on the isotopic composition of U, and are different for the four source terms considered. Second, because all radionuclides have the same mode of toxicity, dose rates are additive. The total dose rate received by an organism is thus the sum of the dose rates received from all radioisotopes in the source term considered. Because the radioactive decay of U always invokes a contribution from its daughter products, which themselves are radioactive and emit energy over time, the calculation of dose rates due to exposure to any of the U source terms must take into consideration the energy emitted by all isotopes of U and their daughter products (Figure S1, Table S1). For the purposes of this SLERA, we consider that radionuclides that have low rates of formation have a negligible contribution
to the overall environmental risk, and only consider isotopes formed through reactions having branching ratios (probabilities of decay) higher than 0.9 (Figure S1). In addition, the radioactive concentrations (Bq L-1) of all isotopes in both U-series decay chains starting with radon (Rn) are adjusted by a factor of 0.4 to account for the solubility of gaseous Rn in freshwater at 8 °C (13). The environmental chemical concentrations of daughter products due to the decay of U are vanishingly low, so we consider that their contribution to the overall chemical toxicity of the source terms is negligible even if their contribution to radiological toxicity is not. Finally, radiological exposure differs from chemical exposure in that organisms may be externally irradiated, even at distance. The geometry of an organism (i.e., size, shape) thus plays an important role in determining the dose rate received from a given radioactive source. The calculation of dose rates is therefore not only isotope- or source-term specific, but also species-specific, and requires information about an organism’s size and tissue density (Tables S3 and S4), the fraction of time spent in each compartment of the environment (water, sediment), and its interactions with the environment in relation to internal and external radiological exposure (trophic relationships, habitat, etc). The species composition in aquatic ecosystems is diverse and varies from one environment to another, and as it is not feasible to develop species-specific dosimetric calculations for every species in every system, a generalized representation of the biota in a hypothetical aquatic ecosystem was developed (14). A set of reference flora and fauna was defined to represent the benthic and pelagic guilds of a temperate freshwater ecosystem. The reference organisms, shown in Figure 1, have been characterized in terms of their geometric relationships with radiation sources as well as their potential exposure pathways (internal vs external; Table S4) (15). The specific activities of the radioisotopes in the U-series decay chains range over some 20 orders of magnitude (Table S1). Because daughter products are created and decay over time, the radiological dose rates received by organisms for a given initial concentration of U change with time; as we VOL. 43, NO. 17, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6685
are concerned with chronic toxicity in this study, we consider the radiation doses at t ) 1 month, 1 year, 100 years, and 106 years after decay of the source terms. We assume that secular equilibrium (i.e., when the radioactivity of each daughter product is equal to that of the parent; (16)) is achieved at 106 years. Note that the timeframes discussed here refer to decay, rather than exposure periods. These timeframes are therefore discrete time points at which we qualitatively and quantitatively describe all the daughter products in a source term, rather than continuous periods of exposure to radiation. The assumption that U-series radionuclides are in secular equilibrium is often made when considering NU, as half of the NU originally present on Earth has decayed to its various daughter products, all of which have significantly shorter half-lives such that the radioactivity of each of the radionuclides in the U decay series is approximately equal (17). However, the process of separating U from its ores during the enrichment process breaks the decay chain, so it is unlikely that exploited U is in secular equilibrium with its daughter products. Step 2: Effects Analysis. The goal of the Effects Analysis part of this study was to derive chemical and radiological benchmark levels that protect the freshwater ecosystem. Such a benchmark for radiological exposure, the Predicted No Effect Dose Rate (PNEDR) was recently published (10 µGy h-1 7, 18). In the absence of an equivalent chemical benchmark, we applied the same approach to derive the Predicted No Effect Concentration (PNEC, µg L-1). A Species Sensitivity Distribution (SSD) (19) was constructed from chronic ecotoxicity data (Table S5, Figure S2). A log-normal distribution was fitted and median bias-corrected bootstrap resampling was used to construct confidence intervals (20). The number of samples was set to 1000 and the number of data drawn for each sample corresponds to that of the initial data set. The goodness of fit was evaluated with a Kolmogorov-Smirnov test and by the multiple R-square coefficient (R2) between theoretical and empirical distributions. We consider that the experimental data used for the chemical SSD were sufficiently complete to justify an Assessment Factor of 1 (21). The PNEC thus derived was 3.2 µg L-1 (Figure S2). This value is comparable to previously reported benchmarks for uranium derived with different methods (2.6-5 µg L-1 for different freshwater organisms and exposure conditions, see SI). Step 3: Exposure Analysis. In deterministic screeninglevel risk assessments for contaminants in the environment, risk quotients (RQ) are used to identify situations of concern, and are calculated by dividing a point estimate of exposure by a point estimate of effects (i.e., the benchmark PNEC or PNEDR) (22). In this study, the point estimate of radiological exposure that we are interested in is the radiological dose rate that organisms would receive when exposed to the PNEC for U (3.2 µg L-1), which we call the Predicted Environmental Dose Rate (PEDR). Likewise, the point estimate of chemical exposure that we are interested in is the chemical concentration of U that organisms would be exposed to when receiving a dose rate at the PNEDR (10 µGy h-1), which we call the Predicted Environmental Concentration (PEC). To calculate the PNEC-derived PEDR and the PNEDR-derived PEC, it is necessary to consider the specific activity of each of the isotopes considered, the partitioning of each isotope in the environment over time, and the amount of time each organism spends in contact with the isotope in each compartment of the environment. We give details of these calculations in the Supporting Information. When the estimate of exposure is greater than that of effects, i.e., the RQ > 1 (PEC > PNEC or PEDR > PNEDR), the contaminant may cause harmful effects for the system considered. Step 4: Risk Characterization. The representation of results in Figure 2 can be considered analogous to the 6686
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 17, 2009
deterministic RQ approach in that we can compare each organism’s PEDR or PEC to the corresponding benchmark value. However, recall that the chemical and radiological benchmarks were originally obtained via multispecies SSDs and are intended to be ecosystem-wide benchmark values, protective of the entire ecosystem, rather than speciesspecific benchmarks. To conserve the variability among species in exposure levels and because it was not possible to fit a known statistical distribution to our data, we adopted a probabilistic approach where we constructed empirical cumulative distribution functions using the PEC and PEDR values obtained for each reference organism, for each source term, and at each decay period considered. This representation permits us to visualize the percentage of reference organisms that are protected by the chemical benchmark when exposed to “safe” radiological levels (Figure S3), and the percentage of reference organisms protected by radiological benchmark when exposed to “safe” chemical levels of U (Figure S4).
Results and Discussion If the chemical toxicity of U indeed poses a greater environmental risk than the radiological toxicity, as often suggested in the literature (4, 23, 24), then the PNEC which is aimed at protecting the environment from this “stronger” form of toxicity, should also protect the environment from the radiological toxicity of U. Figure 2a shows the calculated dose rates to which each reference organism would be exposed if exposed to the PNEC (the PNEC-derived PEDR values) for each source term, assuming that isotopes are at secular equilibrium (t ) 106 years). The dose rates for each organism increase with increasing degree of 235U enrichment, such that the dose rate received from DU is always lower than that for natural or enriched U (Figure 2a). In reality, this observed increase in dose rates is not so much due to the enrichment of 235U as it is to the concurrent enrichment of 234U whose specific activity is 4 orders of magnitude higher than those of the other two U isotopes (Table S1). Indeed, Table S2 shows that although 234 U accounts for
