Modification Results

Nitrate-N concentrations in underdrain outflow from the treatment garden were lower than roof runoff nitrate-N concentrations for the entire period of study. However, ANCOVA results showed no net change in outlet concentrations from the treatment garden during the treatment period as compared to the calibration period. Both the slope and the intercept were significantly (p=0.01) different for the treatment regression as compared to the calibration regression. Although the nitrate-N calibration regression was significant (p=0.001), the treatment regression was not significant (p=0.06). One explanation for the lack of change in nitrate-N concentrations in treatment garden underdrain outflow may be related to the detection limit for nitrate-N. During the calibration period, 19% of samples were at or below the detection limit of 0.2 mg/L for both the treatment and control gardens. However, during the treatment period, 56% of samples were at or below the detection limit for the treatment garden, whereas only 25% of samples from the control garden were at or below the detection limit. The existence of a large number of samples below detection limit can impact the regression relationship, thereby providing inaccurate estimates of percent change. When the frequencies of samples above and below detection were analyzed using Chi-square analysis, a significant (p=0.01) difference was detected in the saturated garden, whereas no difference was detected in the control garden. A greater proportion of underdrain outflow samples from the treatment garden were below detection in the saturated garden.

Due to the high percentage of samples below detection for ammonia-N (> 59% for both gardens during calibration and treatment), ANCOVA was not performed for ammonia-N. However, Chi-square frequency analysis revealed significantly (Chi-square=5.43, p=0.05) fewer ND values in underdrain outflow from the treatment garden during the treatment period as compared to the calibration period. No difference between periods was detected for the control garden. These results indicate that saturation caused more outflow samples to have concentrations above detection in the treatment garden. TN concentrations decreased by 18%, with significant (p=0.05) changes in both the slope and intercept of calibration and treatment regressions. No significant differences were found for TKN or ON concentrations between the calibration and treatment periods. Despite the export of phosphorus from the system noted in Table 5, TP concentrations were reduced by 88% due to the saturation treatment. Although the calibration period regression was highly significant for TP, the treatment period regression was not significant (R^2=0.04). Clausen and Spooner (1993) caution that in paired watershed analysis, both watersheds should be in steady-state conditions. It is possible that the decay trend noted for TP biased ANCOVA results, and caused an erroneous result.

Back to Top

Metals

Due to the large number of samples at or below detection (60%, 89%, and 43% for Cu, Pb, and Zn, respectively) ANCOVA was not performed. During the calibration period, 14% and 21% of Zn concentrations were below detection for the control and saturated gardens, respectively. However, during the treatment period, 17% of samples were below detection in the control garden, whereas 67% of samples in the saturated garden were below detection. Chi-square frequency analysis showed a significant increase in samples below detection for the saturated garden. No changes were detected for the control garden. It is possible that the reducing conditions in the treatment garden caused Zn to be complexed, and therefore less mobile. No significant differences were detected using Chi-square analysis for Cu and Pb.

During the treatment period, dark staining was observed on the tipping bucket measuring outflow from the treatment garden, whereas no staining was noted on the bucket for the control garden. It was suspected that manganese (Mn) was being released from the system, so the last five months of samples were analyzed for Mn in addition to Cu, Pb and Zn. The geometric mean Mn concentration (n=5) in both precipitation and roof runoff was 4 ug/L. The Mn concentration in underdrain outflow from the control garden was 13 ug/L. The Mn concentration in underdrain outflow from the treatment garden was 270 ug/L, and significantly (p=0.001) higher than precipitation, roof runoff, and control garden underdrain outflow Mn concentrations. Initial soil concentrations of Mn were 331 and 213 ug/g for the treatment garden and control garden, respectively. The reducing conditions noted in the treatment garden were likely the cause of the release of Mn. A rain garden with a saturated zone has the potential to release Mn to receiving waters if an underdrain is directly connected.

Back to Top