Does Landscape Structure Reveal Ecological Sustainability? - Landscape Modelling: Geographical Space, Transformation and Future Scenarios - page 154

was least different among the regions and the landscape metrics for non-irrigated
arable land show the most differences. The most contrasting regions were
Continental Forests and Mountains Open Spaces. Nearly every metric in all land-
cover classes differed significantly from each other. The most similar regions
were Central Atlantic Mixed Agricultural Activities - North Atlantic Arable Land,
Boreal Forest - Nemoral Mixed Agricultural Activities, Central Atlantic Mixed
Agricultural Activities - Continental Forest and Mediterranean Arable Land -
Nemoral Mixed Agricultural Activities (Fig. 12.2 as example).
Fig. 12.2 Boxplot of landscape metric Mean Patch Size (MPS) for the land-cover class “agricul-
tural land with natural vegetation” 2
The factor analysis revealed two different dimensions of landscape pattern, but
the loadings were rather different for the individual land-cover classes (Table 12.2).
Mean Patch Edge (MPE) had the highest loading for non-irrigated arable land, Mean
Patch Size in ha (MPS_ha) for pastures, Mean Shape Index (MSI) for complex
cultivation pattern and agricultural land with natural vegetation and Mean Fractal
Dimension (MFRACT) for natural grasslands. The QDA showed that the landscape-
cover classes could only poorly be predicted by its combination of multivariate
landscape metrics among the SRRF regions. In general, the error rate is rather high.
Spatial visualisation via Geographical Information Systems (GIS) gives a useful
tool for the communication of statistical results. For example, the elements that
match certain statistical results (above or below the Median etc.) can be marked
accordingly (Fig. 12.3) and gives a quick overview over a certain area of interest.