Relating landscape to stream nitrate-N levels in a coastal eastern-Atlantic watershed (Portugal)

Abstract

We apply a linear regression mixed effects model to explore the influence of landscape factors on nitrate-N concentrations in a coastal watershed of Portugal. Landscape composition and configuration metrics, together with variables assessing the physical characteristics of the study area, were used. The analysis was performed using seasonal data from the years 2001 and 2006. The seasonal influence was included as a random effect to account for temporal correlations. Together, the fixed and the random factors explain 78% of the variance, whereas the fixed factors alone explain 10%. Urban, slope, elevation and aggregation index of urban class contribute to the differences found in the NO3−-N concentrations. Urban has the weakest effect, whereas slope and elevation show a conditioned negative effect on nitrate-N. The effect of slope gets stronger for higher standard deviations of elevation and the effect of the standard deviation of elevation, measuring the variation of elevation within a sub-watershed, gets stronger for steeper slopes. Of the configuration class level metrics included in the analysis, only aggregation index of urban played a significant role in the final model, and it revealed to be related to urban percentage. The influence of landscape configuration metrics, though observed by others, was not obvious in this study. Future analysis evaluating the effect of metrics selection could be performed.