Geoscience Reference
In-Depth Information
12.4 Discussion and Conclusions
Various authors have criticised approaches (Estregueil et al., 2001, Haines-Young
& Chopping, 1996; Kozak, Estreguil, & Ostapowicz, 2008) where land-cover
information with relatively coarse spatial resolution is used for large-scale land-
scape studies. Corine land-cover classes reflect only the dominant land use but
contain also other land-cover types and therefore, important structural features
are not captured at this resolution. In contrast, high-resolution images and fine
grain habitat and land-use maps are used to explain small-scale ecological phe-
nomena on the landscape level (Lee, Ellis, Kweon, & Hong, 2008; Schindler
et al., 2007, 2008). Consequently, large scale and widely used datasets such
as Corine land cover should be supplemented and combined with an innova-
tive derivation of homogeneous spatial units by segmentation as presented in our
study.
Statistical analysis showed contradicting results concerning the differentiation of
landscape metrics for the individual SRRF regions. Univariate analysis revealed sig-
nificant differences between many regions for several land-cover classes (Fig. 12.2),
multivariate analysis did not. Overlaying different distributions of the individual
landscape metrics may have contributed to blurring of individual differences. The
use of administrative units (NUTS 2/3 implemented into the SRRF) has weakened
the desirable homogeneity in terms of primary and to some extent also secondary
landscape structure, by resulting in lesser meaningful units for ecological assess-
ment as confirmed by the poor performance of the set of metrics in the QDA.
Further improvement of results may also be achieved by a higher number of sam-
pling sites per region. Higher spatial resolutions could lead to a better discrimination
of size and shape of patches. But at the same time higher resolutions will hamper
the analysis for vast areas across Europe.
The use of landscape metrics as indices for Sustainability Impact Assessment
is not very common (Graymore, Sipe, & Rickson, 2008) although certain aspects
of ecological sustainability like biodiversity (Moser et al., 2002; Schindler et al.,
2007) or naturalness (Peterseil et al., 2004) were investigated. Already Odum and
Turner have shown in their classical work (1989) how increasing consumption of
fossil energy and agrochemicals are coupled with a geometrical simplification of
landscapes expressed by a decrease in fractal dimension. We have shown that even
at the European scale, single landscape metrics react differently depending on land
cover, suggesting that such sensitive indices could be utilised as indicators for eco-
logical sustainability of land use, when applied together with land-cover types in a
spatial regional reference. The elaboration of a scientifically sound knowledge-base
for any impact assessment requires the establishment of statistically valid relation-
ships between observed pattern and a particular ecological process of interest. This
is especially true for biodiversity and sustainability related studies at the landscape
level (Bunce et al., 2008) and calls for not only the combined use of geodatasets
with different spatial resolution, but also for the inclusion of empirical data derived
by representative field observations. As at European level, there is no consistent
Search WWH ::
Custom Search