ABSTRACT
Using publicly available data for contaminated sites, regression relationships between As or Pb concentrations in co-located soils and leaf, root, and fruit vegetables, were developed. The improvement of these regression relationships by additional independent variables known to influence bioavailability of these trace elements in soil (soil pH, available phosphorous, Fe oxide, total Fe, and organic carbon content) was tested. Soil pH, but not plant-available P, decreased unexplained variation in the model for As in leafy vegetables. Iron oxide concentration in soil reduced unexplained variability in As concentrations in root vegetables, but with a positive coefficient thus contradicting its anticipated role as a competitor for As uptake by plants. None of the soil characteristics beyond total Pb concentration reduced variability observed in Pb concentrations in leafy or root vegetables, and there was no model that predicted Pb concentrations in fruit. Predictions of tissue concentration from single-value Plant-Uptake Factors (PUFs) for As in leaf and root vegetables, and for Pb accumulation in root vegetables, deviated more from the observed values than predictions from the regression relationships. The FW PUF determined from this study was within an order of magnitude of that used by the United Kingdom for development of generic soil quality values.
ACKNOWLEDGMENTS
The authors gratefully acknowledge the financial support of Health Canada (Contaminated Sites Division) as well as the Metals in the Human Environment Strategic Network, a full list of the sponsors of which can be found at www.mithe-sn.org. The input of H. Renkema in the preparation of the text is also gratefully acknowledged.
Notes
List of Abbreviations: CI: Confidence Interval, CR: Concentration Ratio, Df: degrees of freedom, DW: Dry Weight, FW: Fresh Weight, GM: Geometric Mean, GSD: Geometric Standard Deviation, MS: Mean Square, OC: Organic Carbon, PD: Proportional Deviation, PUF: Plant Uptake Factor, SS: Sum of Squares, TBLM: Terrestrial Biotic Ligand Model
1Model: ln[plant] = β0 + β1 (ln[soil]); where concentrations are μg/g WW for plants and DW for soil, β0 = intercept of regression, β1 = slope of regression, and SE = Standard Error; a p < .001; b p < .05; NSNot Significant (p > .05).
aPlant concentrations reported as μg/g wet weight; soil concentrations reported as μg/g dry weight; OC reported as percent dry matter.
aPlant concentrations reported as μg/g wet weight; soil concentrations reported as μg/g dry weight; OC reported as percent dry matter.