HomeMy WebLinkAboutDERR-2024-007807SITE-SPECIFIC APPROACH FOR SETTING WATER QUALITY CRITERIA FOR
SELENIUM: DIFFERENCES BETWEEN LOTIC AND LENTIC SYSTEMS
W.J. Adams, 1
J.E. Toll, 2
K.V. Brix, 2
L.M. Tear, 2
D.K. DeForest, 2
1 Kennecott Utah Copper
P.O. Box 6001
Magna, UT 84044
2 Parametrix, Inc.
5808 Lake Washington Blvd. NE, Suite 200
Kirkland, WA
ABSTRACT
Results of an in-depth review of the literature indicates there are significant differences in the
bioaccumulation of selenium by fishes and invertebrates from lotic (flowing) and lentic (standing) water
bodies and that selenate is much less bioaccumulative than selenite. Bioaccumulation in fish is a factor of
10 or more higher in lentic systems as compared to lotic systems. These differences are a function of
selenium’s site-specific biogeochemical cycling. Further, we observed considerable variation in bird
accumulation of selenium from site to site. To account for differences in bioaccumulation potential of
selenium we developed a residue-based Bayesian Monte Carlo model to derive site-specific selenium
water quality criteria protective of fish and sensitive avian species.
The approach uses data from a given site of interest to calibrate a model based on data from several other
similar sites. When evaluating a specific site, the range of water and tissue concentrations is typically
limited. This makes it difficult to use site-specific data to identify a water concentration sufficiently low
that tissue concentrations do not exceed the tissue-effect threshold. Data from several similar sites
provide a broader range of water and tissue residue concentrations that allow for an appropriate statistical
extrapolation of the data to the site of interest. The Bayesian Monte Carlo model accounts for the
significant site-to-site variability that exists in the relationship between water selenium and the mean
tissue residue. In practice, data from similar sites are pooled to define a set of possible water and mean
tissue residue relationships. This set of possible relationships is then used with data from the site of
interest to determine which relationships, from the set of possibilities, best fit the specific site. Once we
have determined which set of possible relationships fit the specific site, we extrapolate from the observed
water concentration to a water concentration that results in a tissue residue concentration less than or
equal to a chronic effect threshold. This value becomes the chronic water quality criterion.
Adams, W.J., J.E. Toll, K.V. Brix, L.M. Tear and D.K. DeForest. 2000. Site-specific approach for setting water
quality criteria for selenium: differences between lotic and lentic systems. Proceedings Mine Reclamation
Symposium: Selenium Session; Sponsored by Ministry of Energy and Mines, Williams Lake, British Columbia,
Canada, June 21-22, 2000.
INTRODUCTION
Selenium contains properties that make it unique relative to other metals and metalloids. It occurs in both
organic and inorganic forms which are differentially toxic and is an essential element for most organisms.
The selenium forms present in aquatic systems are controlled by the biogeochemical cycling of selenium
which is strongly influenced by site-specific environmental factors such as redox, pH, and biological
productivity (Lemly and Smith, 1987; Bowie and Grieb, 1991; Porcella et al., 1991).
Reduction of inorganic selenium species tends to immobilize selenium in an aquatic system, while other
processes, such as oxidation and biotransformation tend to make selenium bioavailable to aquatic
organisms. Biological mechanisms such as uptake of sediment selenium by rooted plants, benthic
invertebrates, and detritus-eating invertebrates, can act to re-mobilize selenium into the aquatic food web.
Accordingly, lentic systems tend to bioaccumulate selenium much more than lotic systems that have
higher flushing rates and lower productivity (Lillebo et al., 1988; Van Derveer and Canton, 1997). For
example, Lillebo et al. (1988) demonstrated this by plotting bioaccumulation data for impounded and
flowing waters; fish selenium residues were approximately six times greater in impounded waters than in
flowing waters at a water selenium concentration of 10 µg/L.
Based on increased awareness of the ecotoxicological effects of selenium, a number of water quality
monitoring programs have been implemented to evaluate potential selenium contamination at freshwater
sites. In order to interpret the significance of the selenium concentrations measured under these
programs, several authors have proposed selenium guidelines for various environmental compartments
(e.g., Lemly, 1993a; Skorupa et al., 1996). Specifically, guidelines have been proposed for surface water,
sediment and various tissues, including ovaries, whole body, diet, liver, eggs and testes based on the
authors’ reviews of published and unpublished literature.
These guidelines include recommended toxicity thresholds for abiotic (water, sediment) and biotic
(various fish tissues and diet) compartments. Given the site-specific factors that influence selenium
bioavailability, bioaccumulation, and toxicity in aquatic systems, we believe that the use of a single
guideline value for selenium in surface waters is inappropriate. Different sites will require different
selenium water concentrations to ensure that concentrations in tissues such as fish and bird ovaries do not
exceed a toxic threshold. Site-to-site variability has been demonstrated for fish by Van Derveer and
Canton (1997) and for birds by Adams et al. (1998). Van Derveer and Canton (1997) used a sediment-
based bioaccumulation model to demonstrate that fish in a lotic system in Colorado were not at risk at
water selenium concentrations of approximately 30 µg/L, three times higher than concentrations at which
effects were observed in Belews Lake.
Adams et al. (1998) evaluated the differences in bioaccumulation in shorebirds (Figure 1). They used
probabilistic regression models to assess water, food chain and bird egg residues from fifteen lentic sites
in the western United States. Uncertainty analysis of the regression models provided a probability
distribution of waterborne selenium concentrations associated with bird egg tissue residues. The 90th and
10th percentile water selenium concentrations associated with a selenium concentration of 20 mg/kg dw
in bird eggs ranged from 6.8 to 318 µg/L. These studies support the concept of the need for site-specific
water quality guidelines for selenium based on a bioaccumulation model and measurement of selenium
concentrations in critical tissues. The Adams et al. model is represented in this paper (Figure 2) using a
revised selenium concentration of 16 mg/kg dw in mallard duck eggs as a threshold for chronic effects,
i.e., EC10 for duckling hatchability and survival (Fairbrother et al. 1999). The 90th and 10th percentile
water selenium concentrations associated with a selenium concentration of 16 mg/kg dw in bird eggs
ranges from 4.6 and 213 µg/L, respectively. These data indicate a large degree of variability in selenium
concentrations that are potentially protective for sensitive bird species at different sites depending upon
site-specific biogeochemistry, bioaccumulation and bird feeding behavior.
SITE-SPECIFIC BIOACCUMULATION EVALUATION
The concept that site-specific differences in the biogeochemistry of selenium can significantly alter the
potential for toxicological effects is best demonstrated by site-specific differences in bioaccumulation of
selenium in sensitive species. We have previously examined this for birds (Adams et al., 1998). To
evaluate differences in selenium bioaccumulation in fish we evaluated published fish residue data for
essentially the same sites evaluated for birds by Adams et al. (1998). We separated the data into lotic
sites (flowing water; short hydraulic retention times, i.e., minutes to days) and lentic sites (standing water;
long hydraulic retention times, i.e., weeks to years). The concept of lotic versus lentic is one that is used
to typify extremes in biogeochemistry (oxygen content, redox, hydraulic retention time, carbon content)
that is important in terms of controlling the formation of reduced selenium forms including organo-
selenium compounds. In lotic environments, selenium in the water column is most often found in the
form of selenate and migration to sediments is limited. In lentic environments, selenate is less prevalent,
selenite is more common, and both forms are biologically and chemically reduced to elemental and
organo-selenium forms. These reduced forms are prevalent in lentic sediments and form the basis for
uptake by benthic invertebrates and subsequent food chain bioaccumulation.
The results of our bioaccumulation data analysis indicates that a clear and distinct difference between the
patterns of bioaccumulation by fish in lotic and lentic environments (Figure 3). The data presented here
are predominately for warm water fish species (e.g., centrarchids, cyprinids, and ictalurids). Selenium
bioaccumulation factors for fish from lentic environments are typically a factor of 10 or more higher than
those from comparable water concentrations in lotic environments. Additionally, these data indicate that
bioaccumulation factors for fish and selenium are inversely related to exposure concentration. This is
consistent with data reported for several other metals (Brix and Deforest, 2000). Recognition of this
relationship provides insight into the variability that exists in reported bioaccumulation factors (BAFs) for
selenium, i.e., highest BAFs occur at the lowest exposure levels.
Further analysis of the selenium fish residue data from lotic systems indicates that tissue selenium
concentrations remain fairly constant across a range of water concentrations up to about 13 µg/L. In
contrast, in lentic systems, selenium tissue levels begin to increase as selenium water concentrations
increase above 1.0 µg/L. The hockey-stick regressions presented in Figures 4 and 5 demonstrate distinct
differences in bioaccumulation between lotic and lentic systems. While the slopes of the upper parts of
the regressions are similar, the inflection points are more than a factor of ten different. Recognizing that
selenium is an essential element for fish, the lower part of each regression is thought to represent a range
of concentrations across which fish can actively regulate selenium uptake. Recognition of these
differences in bioaccumulation provides the basis for developing tools to assess site-specific
bioaccumulation.
SITE-SPECIFIC WATER QUALITY CRITERIA METHODOLOGY
Bioaccumulation data leave little doubt that water selenium concentrations protective of aquatic life and
wildlife differ from site to site as a function of selenium’s site-specific biogeochemical cycling.
Consequently, from a regulatory perspective to avoid over-regulation with associated costs, there is a
need for developing a site-specific water quality criteria methodology for selenium. Existing
methodologies for deriving site-specific water quality criteria such as water effect ratios are not applicable
to selenium because unlike most contaminants, for selenium, the diet is the critical exposure pathway.
Therefore, approaches for deriving site-specific water quality criteria must be based on the dietary
exposure pathway to be appropriately protective for both birds and aquatic life. Given this need,
identification and agreement on tissue toxicity thresholds for use in site-specific bioaccumulation models
are essential. To that end, thresholds for birds and aquatic organisms was recently reviewed and
summarized by DeForest et al. (1999) and Fairbrother et al. (1999).
Overall, Adams et al. (1998) found a high correlation between water and mean egg selenium
concentrations that is strongly influenced by site-specific factors. In light of the observed site variability,
the following was concluded: first, the numerical water quality criterion for selenium is best used as a
screening tool. Second, when waterborne selenium approaches or exceeds the criterion, a site-specific
assessment should be used to determine whether existing water concentrations pose risk, and if so,
identify a safe selenium water concentration for the site.
The conclusions of Adams et al. (1998) led us to develop a methodology for developing site-specific
water quality criteria using tissue residue concentration and effects threshold data. We have applied the
methodology to selenium, but it is applicable to any constituent for which tissue residue-based endpoints
are of concern. In general, setting water quality criteria protective of tissue residue-based endpoints for
metals including selenium poses problems because the bioaccumulation factor (BAF) is not constant and
is inversely related to the exposure concentration (Brix and DeForest, 2000). This must be taken into
account in the model. Factors influencing BAF include: site-specific water and sediment chemistry,
trophic relationships, and the degree of spatial and temporal co-occurrence of habitat, stressor and the
endpoint of concern.
The statistical technique we use in our methodology is Bayesian Monte Carlo analysis (BMC). Monte
Carlo methods are numerical techniques for generating a representative sample from a probability
distribution function (PDF). BMC evolved from earlier procedures used to ensure that the PDF of a
model’s predictions was consistent with observed data. These earlier acceptance/rejection procedures
involved deleting predictions that were inconsistent with observations (Hornberger and Spear 1980, Beck
1987, Woodruff et al. 1992). The difference between BMC and earlier procedures is that BMC is more
statistically rigorous. BMC defines the acceptance/ rejection procedure using the axioms of probability
theory, as expressed in Bayes’ theorem (Bayes, 1763). Additional details on the site-specific
methodology is reported by Toll et al. (1999). Only a brief synopsis of the method is presented here.
The model approach is one that uses a generic bioaccumulation model for fish or birds, such as that
developed by Adams et al. (1998). The purpose of the generic model is to describe bioaccumulation as
reported in the literature for a wide variety of sites. This model provides an a priori estimate of
bioaccumulation potential for a given site in the absence of any site-specific data. The generic model’s
prediction interval (Figure 6) is based on data from all sites in the data set. It describes the range of
possible site-specific relationships. The data set for the generic model can be updated as additional site
data are obtained. Tissue residue data from a given site can then be compared to the model’s estimated
value. A determination is then made as to whether or not the site-specific bioaccumulation is greater or
less than what would be predicted by the generic model. If the site specific-tissue residue, at a given
water concentration is less than predicted by the generic model, then the model calculates a water quality
criterion that is higher for the site. This becomes the site-specific concentration in water that is protective
of a given tissue threshold concentration. Likewise, if the tissue residue at the site is greater than the
generic model would predict, the site-specific water quality criterion would be revised downward to
insure that the tissue threshold is not exceeded.
CONCLUSIONS
(1) Literature data indicate that selenium bioaccumulation varies from site to site for both birds and
aquatic organisms.
(2) There is considerable evidence supporting that the conclusion that bioaccumulation of selenium is less
at lotic sites than at lentic site.
(3) There is a need for a methodology to derive site-specific water quality criteria for selenium.
(4) Using a residue-based Bayesian Monte Carlo model, site-specific selenium water quality criteria can
be calculated for sensitive avian and aquatic species.
REFERENCES
Adams, W.J., K.V. Brix, K.A. Cothern, L.M. Tear, R.D. Cardwell, A. Fairbrother, and J.E. Toll. 1998.
Assessment of selenium food chain transfer and critical exposure factors for avian wildlife species: Need
for site-specific data. Environmental Toxicology and Risk Assessment: Seventh Volume. ASTM STP
1333, E.E. Little, A.J. DeLonay, and B.M. Greenberg (Eds.). American Society of Testing and Materials,
Philadelphia, PA.
Bayes, Rev. T. 1763. An essay toward solving a problem in the doctrine of chances. Philos. Trans. R.
Soc. London 53: 370-418. Reprinted in Biometrika 45: 293-315 (1958).
Beck, M.B. 1987. Water Quality Modeling: A Review of the Analysis of Uncertainty. Water Res. Res.
23: 1393-1442.
Brix, K.V. and D. K. DeForest. 2000. Critical review of the use of bioconcentration factors for hazard
classification of metals and metal compounds. Parametrix, Inc., 5808 Lake Washington Blvd. NE, Suite
200, Kirkland WA 98033.
Bowie, G. L. and T. M. Grieb, 1991. A model framework for assessing the effects of selenium in aquatic
ecosystems. Water Air Soil Pollution 57-58, 13-22.
DeForest, D.K., K.V. Brix, and W.J. Adams. 1999. Critical review of proposed residue-based selenium
toxicity thresholds for freshwater fish. Hum. Ecol. Risk Assess. 5(6): 1187-1228.
Fairbrother, A., K.V. Brix, J.E. Toll, S. McKay, and W.J. Adams. 1999. Egg selenium concentrations as
predictors of avian toxicity. Hum. Ecol. Risk Assess. 5(6): 1229-1253.
Hornberger, G.M and R.C. Spear. 1980. Eutrophication in Peel Inlet II: Identification of Critical
Uncertainties via Generalized Sensitivity Analysis. Water Resources 14: 43-49.
Lemly, A. D. 1993a. Guidelines for evaluating selenium data from aquatic monitoring and assessment
studies. Environmental Monitoring Assessment 28, 83-100.
Lemly, A. D. and G. J. Smith, 1987. Aquatic cycling of selenium: implications for fish and wildlife. U.S.
Dept. of the Interior, Fish and Wildlife Service. Leaflet 12. Washington, D.C. 10 pp.
Lillebo, H. P., S. Shaner, D. Carlson, N. Richard, and P. DuBowy. 1988. Regulation of agricultural
drainage to the San Joaquin River. State Water Resources Control Board. SWRCB Order No. W.Q. 85-
1.
Porcella, D. B., G. L. Bowie, J. G. Sanders, and G. A. Cutter. 1991. Assessing Se cycling and toxicity in
aquatic ecosystems. Water Air Soil Pollution 57-58, 3-11.
Skorupa, J. P., S. P. Morman, and J. S. Sefchick-Edwards. 1996. Guidelines for interpreting selenium
exposure of biota associated with nonmarine aquatic habitats. U.S. Fish and Wildlife Service National
Irrigation Water Quality Program, Sacramento Field Office, March 1996. 74 pp.
Toll, J. E., L. M. Tear, K. V. Brix, D. K. DeForest, and W. J. Adams. 1999. A method for determining
site-specific water quality criteria protective of tissue residue-based endpoints, Society of Toxicology and
Chemistry presentation (manuscript in preparation).
Van Derveer, W. D. and S. P. Canton. 1997. Selenium sediment toxicity thresholds and derivation of
water quality criteria for freshwater biota of western streams. Environmental Toxicology and Chemistry
16(6), 1260-1268.
Woodruff, T.J., F.Y. Bois, D. Auslander and R.C. Spear. 1992. Structure and Parameterization of
Pharmacokinetic Models: Their Impact on Model Prediction. Risk Analysis 12: 189-210.
Figure 1. Relationship between water-selenium and mean bird-egg-selenium.
Fig ure 2. Percentiles of Log w ater selenium distribution that w ould
produce a mean avian egg selenium value of 16 mg/kg dw .
-1 -0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
W ater Selenium (ug/L)
5%
10%
25%75%
95%
50%
90%
0.1 0.3 1 3 10 30 100 300 1,000 3,000 10,000
Fig ure 3 . Lo g bio a c c umua ltio n f a c to r (B AF) v s . wa t e r s e le nium (WS ) fo r lo tic
a nd le nt ic a qua tic s y s t e ms
0
1
2
3
4
5
-10 10 30 50 70 90 110 130 150
WS
lo
g
F
I
S
H
/
l
o
g
W
S
(
B
A
F
)
BAF of log Lotic data BAF of log Lentic data
Log. (BAF of log Lotic data)Log. (BAF of log Lentic data)
Figure 4. Regression of water selenium versus whole fish selenium in lentic
environments
-0.50
0.00
0.50
1.00
1.50
2.00
2.50
3.00
-1.00 -0.50 0.00 0.50 1.00 1.50 2.00 2.50 3.00
Log Water Selenium (ppb)
Figure 5. Water selenium versus whole body fish selenium in lotic environments
-0.50
0.00
0.50
1.00
1.50
2.00
2.50
3.00
-0.50 0.00 0.50 1.00 1.50 2.00 2.50 3.00
Log Water Selenium (ppb)