Spatial and temporal variability of nitrate concentration in seepage water under a mature Norway spruce [Picea abies (L.) Karst] stand before and after clear cut

The spatial and temporal variability of the nitrate (NO3 −) concentration in seepage water below the main rooting zone of a mature spruce forest was investigated using 121 suction cups which were implemented in a 2 × 2 m grid at 40 cm depth in the mineral soil. Seepage was collected at least monthly during the vegetation period in 2005. In January 2006 a clear cut was performed on the plot and monthly seepage water was collected with 118 suction cups. Conventional and geostatistical methods were used to investigate the spatial and temporal variability in NO3 −. We tried to explain the observed variability with multiple classification analysis (MCA) and multiple linear regression models (MLR) based on vegetation, stand and soil parameters, which could be measured without disturbance of the ecosystem. The NO3 − concentration in the mature stand reached mean values up to 35.2 mg l−1 per sampling date and the distributions were positively skewed. The temporal variability was much lower than the spatial variability. NO3 − in seepage water showed clear structural patterns over the whole vegetation period. Spatial autocorrelation ranged between 16 and 19 m and structural variance was between 65% and 80% of the whole model semivariance. However, for practical purposes it should be sufficient to consider an autocorrelation range of about 12 m as the last 5–6.5 m only explained 5% of the total structural variance of NO3 − in seepage water. Vegetation and stand parameters such as distance and size of the trees surrounding the measuring points explained about 40% of total variability in the MCA and MLR models. After clear cut, concentration means per sampling date were below 10 mg l−1 in spring but increased to more then 150 mg l−1 until December 2006. The distribution was nearly normal. The patterns of spatial and temporal variability were reversed compared to the mature stand. Structuring was less pronounced with autocorrelation ranges decreasing from 10 m in April 2006 to values below 3 m during the following vegetation period. The structuring variance was below 30% at most sampling dates. Thickness of the humus layer and amount of water in the sampling bottles, which were not considered in the mature stand, contributed significantly to the explanation of variability after the clear cut. Together with vegetation coverage and stand parameters 30% of the total NO3 − variability could be explained in the MCA and MLR models. The consequences for future seepage water investigations are discussed.