Modeling bioaccumulation in small mammals
Environ. Toxicol. Chem. 20, 2001
2637
termine which individuals are transients and which are long-
term residents of a site and thus more appropriate for tissue
analysis sampling. Not all areas of a site are used by small
mammals because of disturbance or poor habitat quality, which
typically coincide with the most contaminated areas. Trapping
can therefore help identify areas that are relevant for sampling
media and biota for inclusion in food web modeling. Trapping
also facilitates collecting feces or stomach contents of small
mammals for diet analysis. Site-specific diet analyses are crit-
ical to improve bioaccumulation and exposure estimates since
much wildlife exposure modeling is plagued by lack of in-
formation about local food webs [10]. Diet analyses are es-
pecially critical for species whose diets are variable or op-
portunistic.
Predictive bioaccumulation and exposure models can in-
troduce considerable uncertainty into screening-level ecolog-
ical risk assessments, prompting the need for model validation
through tissue sampling. The soil–small mammal bioaccu-
mulation regression models developed by Sample et al. [4]
can adequately predict whole-body concentrations of chemi-
cals in small mammals, as illustrated by arsenic and lead in
this study. The accuracy of these models varies among metals
and among sites at which they are applied, but use of prediction
intervals can help ensure conservative estimates for screening-
level assessments. Modifying the models by incorporating ad-
ditional regression parameters, such as soil properties and dif-
ferentiating between metal species in soil, could reduce pre-
diction error.
The cumulative ingestion bioaccumulation model as param-
eterized in this study produced excessive estimates of body
burdens of most metals in house mice. Which parts of the
model contributed most to prediction error are unknown, but
certain refinements would increase the realism of the model.
Diet composition, body weight, and field metabolic rate can
be represented as factors that change over seasons and an
individual’s lifetime. Also, elements of mechanistic toxico-
kinetic models (e.g., absorption, metabolism, distribution, and
elimination kinetics) may be applied in place of constant AE
factors. Along with the these improvements to the structure
of wildlife exposure models, more experimental data on gas-
trointestinal absorption and assimilation of chemicals from
various soil and food matrices (e.g., plant parts, soft animal
tissues, exoskeletons, bone, scales, hair, and feathers) are need-
ed. Simpler statistic-based models, such as empirical regres-
sions or BAFs, may generate more accurate predictions of
chemical doses and bioaccumulation in many cases and are
easier to parameterize. However, mechanistic models can be
promising tools for increasing the understanding of contami-
nants in wildlife as they are tested and refined.
upper probabilistic percentiles. Predictions of metal concen-
trations in individual mice were not significantly more accurate
than sitewide predictions and failed to justify the additional
effort of individual-level modeling.
Acknowledgement—We thank Daniel W. Anderson and Roger L.
Hothem for their comments on the manuscript. The assistance of
Cuong Pham, Wallace Neville, John Randell, Robin Leong, and others
at Mare Island was also appreciated.
REFERENCES
1. Baes CF III, Sharp RD, Sjoreen AL, Schor RW. 1984. A review
and analysis of parameters for assessing transport of environ-
mentally released radionuclides through agriculture. ORNL-5786.
U.S. Department of Energy, Oak Ridge National Laboratory, Oak
Ridge, TN.
2. Travis CC, Arms AD. 1988. Bioconcentration of organics in beef,
milk, and vegetation. Environ Sci Technol 22:271–274.
3. Bechtel Jacobs. 1998. Empirical models for the uptake of inor-
ganic chemicals from soil by plants. BJC/OR-133. U.S. Depart-
ment of Energy, Oak Ridge National Laboratory, Oak Ridge, TN.
4. Sample BE, Beauchamp JJ, Efroymson RA, Suter GW II. 1998.
Development and validation of bioaccumulation models for small
mammals. ES/ER/TM-219. U.S. Department of Energy, Oak
Ridge National Laboratory, Oak Ridge, TN.
5. Sample BE, Beauchamp JJ, Efroymson RA, Suter GW II. 1999.
Literature-derived bioaccumulation models for earthworms: De-
velopment and validation. Environ Toxicol Chem 18:2110–2120.
6. Fordham CL, Reagan DP. 1991. Pathways analysis method for
estimating water and sediment criteria at hazardous waste sites.
Environ Toxicol Chem 10:949–960.
7. MacIntosh DL, Suter GW, Hoffman FO. 1994. Uses of proba-
bilistic exposure models in ecological risk assessment of contam-
inated sites. Risk Anal 14:405–419.
8. Pascoe GA, Blanchet RJ, Linder G. 1994. Bioavailability of met-
als and arsenic to small mammals at a mining waste-contaminated
wetland. Arch Environ Contam Toxicol 27:44–50.
9. Alsop WR, Hawkins ET, Stelljes ME, Collins W. 1996. Com-
parison of modeled and measured tissue concentrations for eco-
logical receptors. Hum Ecol Risk Assess 2:539–557.
10. Pastorok RA, Butcher MK, Nielsen RD. 1996. Modeling wildlife
exposure to toxic chemicals: Trends and recent advances. Hum
Ecol Risk Assess 2:444–480.
11. Menzie CA, Burmaster DE, Freshman JS, Callahan CA. 1992.
Assessment of methods for estimating ecological risk in the ter-
restrial component: A case study at the Baird and McGuire Su-
perfund site in Holbrook, MA. Environ Toxicol Chem 11:245–
260.
12. Banton MI, Klingensmith JS, Barchers DE, Clifford PA, Ludwig
DF, Macrander AM, Sielken RL Jr, Valdez-Flores C. 1996. An
approach for estimating ecological risks from organochlorine pes-
ticides to terrestrial organisms at Rocky Mountain Arsenal. Hum
Ecol Risk Assess 2:499–526.
13. Gorree M, Tamis WLM, Traas TP, Elbers MA. 1995. BIOMAG:
A model for biomagnification in terrestrial food chains. The case
of cadmium in the Kempen, The Netherlands. Sci Total Environ
168:215–223.
14. Richardson GM. 1996. Deterministic versus probabilistic risk as-
sessment: Strengths and weaknesses in a regulatory context. Hum
Ecol Risk Assess 2:44–54.
SUMMARY
15. Smith RL. 1994. Use of Monte Carlo simulation for human ex-
posure assessment at a Superfund site. Risk Anal 14:433–439.
16. Clifford PA, Barchers DE, Ludwig DF, Sielken RL, Klingensmith
JS, Graham RV, Banton MI. 1995. An approach to quantifying
spatial components of exposure for ecological risk assessment.
Environ Toxicol Chem 14:895–906.
17. Maurer BA, Holt RD. 1996. Effects of chronic pesticide stress
on wildlife populations in complex landscapes: Processes at mul-
tiple scales. Environ Toxicol Chem 15:420–426.
18. Torres KC, Johnson ML. 2001. Bioaccumulation of metals in
plants, arthropods, and mice at a seasonal wetland. Environ Tox-
icol Chem 20:2617–2626.
19. U.S. Environmental Protection Agency. 1993. Wildlife exposure
factors handbook. EPA/600/R-93/187a. Office of Research and
Development, Washington, DC.
Soil–small mammal bioaccumulation regression models de-
veloped by Sample et al. [4] produced accurate estimates of
arsenic and lead body burdens in house mice but did not ad-
equately predict copper or nickel concentrations in mice. A
mechanistic ingestion-based model of bioaccumulation over-
predicted most concentrations of lead by less than one order
of magnitude but substantially overpredicted levels of arsenic,
copper, and nickel. Monte Carlo simulations of the regression
models generated probabilistic distributions of body burdens
that were consistent with deterministic results. However, de-
terministic minimum and maximum predictions of the inges-
tion model were excessively conservative relative to lower and