Improving the precision of estimating nitrate (NO3−) concentration in seepage water of forests by prestratification with soil samples

Nitrate (NO3 −) concentrations in seepage water (\( {\text{SW}}_{{{\text{NO}}_{ 3} }} \)) of N saturated forests regularly show a high spatial variability. As seepage water sampling is expensive, most studies use small numbers of soil solution samplers. This implies high uncertainties of calculated concentration and flux data. At the N saturated Hoglwald site we investigated, whether a preselection of seepage water sampling places by a prior intensive soil sampling can improve the precision of the estimation of \( {\text{SW}}_{{{\text{NO}}_{ 3} }} \). We implemented 121 suction cups and analyzed the soil, removed with the soil auger during installation at the sampling depth of 40 cm, for water extractable NO3 −. The NO3 − content in soil (\( {\text{Soil}}_{{{\text{NO}}_{ 3} }} \)) was calculated and \( {\text{SW}}_{{{\text{NO}}_{ 3} }} \) was measured. With this data set we tested the correlation between \( {\text{Soil}}_{{{\text{NO}}_{ 3} }} \) and \( {\text{SW}}_{{{\text{NO}}_{ 3} }} \) and simulated a random selection of sampling places (RS) as well as a prestratified sampling based on \( {\text{Soil}}_{{{\text{NO}}_{ 3} }} \) (PS). With bootstrap statistics different numbers of replications (n) were tested and the 95% confidence interval (95% CI) was used to compare RS with PS. Highly significant correlations between \( {\text{Soil}}_{{{\text{NO}}_{ 3} }} \) and \( {\text{SW}}_{{{\text{NO}}_{ 3} }} \) were found (r = 0.99***). It was even possible to calculate \( {\text{SW}}_{{{\text{NO}}_{ 3} }} \) from \( {\text{Soil}}_{{{\text{NO}}_{ 3} }} \). Here best results where achieved when \( {\text{Soil}}_{{{\text{NO}}_{ 3} }} \) was determined from field moist soil samples. With PS the accuracy of the mean value of \( {\text{SW}}_{{{\text{NO}}_{ 3} }} \) improved considerably. To achieve a 95% CI lower than ±10%, 250 suction cups would be needed for RS compared to n = 20 for PS. Even 6 months after implementation, precision of mean \( {\text{SW}}_{{{\text{NO}}_{ 3} }} \) was higher for suction cup places selected with the PS method. A detailed description is given, how these findings may be applied in studies and monitoring programs focusing on NO3 − leaching in forest ecosystems.