Scale dependence of landscape metrics and their indicatory value for nutrient and organic matter losses from catchments

We investigated scale dependence of landscape metrics and the relationship between land use parameters and FRAGSTATS-based landscape metrics (edge density (ED), patch density (PD), mean shape index (SHAPE_MN), mean euclidean nearest neighbor index (ENN_MN), contagion (CONTAG), patch richness density (PRD), and Shannon's diversity index (SHDI)) and nutrient/organic-matter-based water quality indicators (BOD7 and CODKMnO4 values, total-N and total-P concentrations in water) in 24 catchments with various land use patterns in Estonia. We used the Basic Map of Estonia (1:10,000), the Base Map of Estonia (1:50,000) and the CORINE Land Cover Map (1:100,000). In scale analysis, we calculated landscape metrics on artificial and real landscapes. Scale analysis showed that responses of landscape metrics to changing grain size vary among landscapes and metrics. Analysis of artificial landscapes showed that mean euclidean nearest neighbor distance and contagion are directly dependent on grain size and should therefore be used carefully. When finding relationships between landscape metrics and water quality indicators, significant differences between the relationships derived from the Base Map and the CORINE Land Cover Map were found. In the case of the Base Map, landscape metrics correlated strongly with land use and showed no expected relationships with water quality data. This underlines the importance of land use classification in such kind of analysis. Correlation between the landscape metrics calculated on the basis of the CORINE Land Cover Map and water quality data was stronger than in the case of the Base Map. The CODKMnO4 value significantly correlated with all land use types. For instance, the CODKMnO4 values are higher when fens and natural areas form a higher proportion of the catchments' land use. Except for the BOD7 value, all the water quality indicators showed significant correlation with urban land use proportions. Strong relationship between the patch density and the CODKMnO4 value is most likely caused by the fact that both parameters were significantly correlated with the proportion of natural areas. As the landscape metrics depend on pixel size, topographic scale, and land use classification, and as the effect of land use on water quality in catchments is the most significant of the factors, it was impossible to separate the influence of land use pattern from the influence of FRAGSTATS-based landscape metrics.