Maturity 2

Maturity 2

5 MERCURY
Prior to the analysis in this study, it was uncertain whether a relationship between the concentrations of mercury in soil and plants from multiple studies would be significant. Both the speciation of mercury and the uptake route via air were expected to contribute large uncertainty bounds to any empirical relationship. In contrast to other metals, most mercury in above-ground plant tissue is taken up as volatile, elemental mercury through the leaves (Bysshe 1988, Siegel and Siegel 1988, Lindberg et al. 1979), with limited accumulation from the soil via the roots and transpiration stream. However, significant relationships between soil and plant mercury have been observed previously. For example, a significant correlation between soil mercury and tissue concentrations was observed for several plant
species found in mining areas (Siegel et al. 1987) and near chloralkali plants (Lenka et al. 1992 and Shaw and Panigrahi 1986).
6 NICKEL
In contrast to the results of this study, an association of nickel concentrations in plants and pH has
previously been observed. Sims and Kline (1991) found significant multiple regression models between nickel in wheat and soybean and soil metal concentrations and pH, but not with soil metal concentrations alone. Reducing the pH of soils led to increased uptake in several plant species (Sauerbeck and Hein 1991). Thus, it is surprising that pH did not contribute significantly to the variability in the present multiple regression model.
7 SELENIUM
Major determinants of the uptake of selenium include chemical form and soil properties. Selenate
is taken up more effectively than selenite (Banuelos 1996, Hamilton and Beath 1963, Gissel-Nielson and Bisbjerg 1970, Smith and Watkinson (1984)), and the uptake of organic selenium is lower than that of inorganic forms (Hamilton and Beath 1963). Banuelos (1996) suggests that soils of high redox in arid regions probably have selenate as the primary species in solution, whereas acid or neutral soils are not likely to have much selenate. Thus, because the present regression model was generated predominantly using data from western sites, the uptake of selenium by plants may be somewhat lower in non-arid environments.
8 ZINC
pH has commonly been observed to be a controlling variable in the uptake of zinc. An increase in
soil pH was associated with a decrease in the zinc content of radish tops (Lagerwerff 1971). Similarly, a decrease in soil pH was associated with an increase in the concentration of zinc in kidney bean (Phaseolus vulgaris), though the mass taken up was unchanged with pH, because the pH decrease was associated with a reduced yield (Xian and Shokohifard 1989). Both of these results are consistent with the relationship derived from data in this study. In contrast, in a study of the uptake of zinc by radish (Raphanus sativus), the regression was improved by including pH as a variable (Davies 1992). However, the positive value that was obtained for the pH term would suggest that raising soil pH increases accumulation of zinc, a result opposite to that found here.

RECOMMENDATIONS
Measurements of contaminant concentrations in plants at a specific waste site are always superior
to estimates of these concentrations for assessing risks to herbivorous or omnivorous wildlife. Even a small number of samples (e.g., 10 or 20) from which site-specific uptake factors can be developed would probably give more precise and accurate estimates of concentrations of chemicals in plants at the site than the use of models recommended below. However, in the absence of these data, regression models or uptake factors should be used. Our study demonstrates that regression models are generally superior to uptake factors for estimating concentrations of chemicals in plants from concentrations in soil.
Single-variable regressions of the natural log-transformed chemical concentration in plant on the
log-transformed concentration in soil are recommended as good tools for estimating concentrations of contaminants in plant tissues for all eight chemicals tested (Table 10). Multiple regressions with chemical concentration in soil and pH are recommended as good tools for estimating the uptake of cadmium, mercury, selenium, and zinc. Although multiple regressions were good predictors of plant concentrations of copper and lead in the validation dataset, pH was not a significant variable in the final combined models. For mercury, the multiple regression with pH was the best predictor of the plant concentrations in the validation data.
Both the 90th percentile uptake factor and the 95% upper prediction limit for the single-variable
regression were adequately conservative for screening ecological risk assessments. Indeed, for data from the two validation studies, these models were arguably too conservative, overpredicting 100% of the measured values for most chemicals. The appropriate level of conservatism should be agreed upon by regulatory agencies, risk assessors, and site managers in the DQO sessions and work plan approval process.
The 95% upper prediction limit for the single-variable regression is recommended as the better of
the two models for providing conservative estimates of plant uptake of contaminants. The method
provided the best, conservative estimate for four of eight chemicals. For three others, the 90th percentile uptake factor provided the best conservative estimate, though one of these comparisons (for mercury) was based on only three samples. The 95% upper prediction limit would be expected to be the better model for a wide range os soil concentrations. The log-transformed regression models consistently proved to be better than uptake factors for estimating chemical concentrations in plants, and the slopes were apparently different from one, indicating that uptake factors are not the best models to use.
Therefore, conservative bounds on the regression models should be better conservative estimates of uptake for most random datasets than the uptake factors.