Modeling spatial patterns in soil arsenic to estimate natural baseline concentrations

Arsenic in soil is an important public health concern, but risk-based toxicity regulatory standards derived from laboratory studies should also consider concentrations measured away from obvious contamination (i.e., baseline concentrations that approximate natural background) to avoid unnecessary remediation burdens on society. We used soil and stream sediment samples from the USGS National Geochemical Survey to assess the spatial distribution of As over a 1.16 × 10 km area corresponding to the state of Ohio. Samples were collected at 348 soil and 144 stream sites at locations selected to minimize anthropogenic inputs. Total As was measured by sodium peroxide fusion with subsequent dissolution using concentrated HCl and analysis using hydride-generation atomic absorption spectrometry. Arsenic in the soil and streambed samples ranged from 2.0 to 45.6 mg kg. Sequential Gaussian simulation was used to map the expected concentration of As and its uncertainty. Five areas of elevated concentration, greater than the median of 10 mg kg, were identified, and relationships to geologic parent materials, glacial sedimentation, and soil conditions interpreted. Arsenic concentrations <4 mg kg were rare, >10 mg kg common, and >20 mg kg not unusual for the central and west central portions of Ohio. Concentrations typically exceeded the soil As human generic screening level of 0.39 mg kg, a value corresponding to an increase in cancer risk of 1 in 1,000,000 for soil ingestion. Such results call into question the utility of the USEPA and similarly low soil screening levels. The contrast between laboratory screens and concentrations occurring in nature argue for risk assessment on the basis of baseline concentrations.