ESTIMATING PESTICIDE EXPOSURE IN TIDAL STREAMS OF LEADENWAH CREEK, SOUTH CAROLINA

Abstract

This article estimates the potential exposure ofestuarine organisms to two pesticides (azin-phosmethyl and fenvalerate) in a tidal stream of Leadenwah Creek near the Edisto River, South Carolina, during four runoff episodes. Exposure is calculated from simulation runs of the one-dimensional transport equation solved by an implicit finite difference method. Calibration was done for each episode by adjusting three conditions (runoff starting time, duration, and flow) and a correction to the dispersion coefficient in order to match the continuously measured salinity transients. First-order rate constants used by the fate component were calculated from half-life values reported in the literature. Baseline scenarios for each episode and each pesticide were derived by using the same conditions obtained in the salinity runs and adjusting the pesticide loading in order to mimic the few data points of measured pesticide concentrations. In all baseline scenarios, pesticide concentration rises following the initial burst of runoff (also noticeable as an abrupt drop in salinity) and then oscillates, forced by the tidal cycle. These oscillations are dominated by transport, while fate imposes a secular decaying trend. Ten additional scenarios for each episode were obtained from the baseline scenario by randomly varying three pesticide load parameters (starting time and duration of runoff, and pesticide discharge) using a Latin hypercubes design. Two exposure metrics were calculated from the simulated and the measured pesticide concentration: maximum and time average, which was obtained by integrating the curve and dividing by the time period. The metrics calculated from the baseline runs are relatively close to the data-derived metrics, because the baseline runs attempted to mimic the data. For each one of the two metrics and all pesticide-episode combinations, several statistics of the set of 1 I scenarios were also calculated: minimum and maximum, mid-range, mean, standard deviation, and median. The mean ± standard deviation interval of the simulation-derived value consistently brackets the data-derived value for the maximum metric, but not for the time-average metric. This may indicate that even if the maximum value is correctly captured in the field sample, the time-average exposure could be in error when calculated directly from the field data due to undersampling of the pesticide time series. The methodology developed here attempts to reconstruct the possible exposure from the sparse sampling of the pesticide concentration during the runoff episodes; only when the number of field samples is high and regularly spaced is it possible to have confidence in the reconstruction.of the curve. The shape of the curve cannot be inferred from the field measurements alone; as expected, tidal movement makes the pesticide concentration swing up and down. This result has important implications because the biological community would be subject to repetitive pulses of exposure to the chemicals. The baseline simulations can be used to derive a pulse-exposure metric by calculating the sum of ratios of the time average of the threshold-exceeding concentrations to the time average of the toxic threshold during intervals of above-threshold concentration. This metric is species specific and extrapolates laboratory toxicity data in order to compare pulse exposure to mortality rates measured in the field.