16
area, habitat range, specific structures and functions (incl. typical species), and
future prospects (ETC/BD 2011; see also Vanden Borre et al. 2011b). To assess the
specific structures and functions, several EU member states have taken an approach
to evaluate local habitat quality at (all or a sample of) individual habitat occurrences, and identified relevant indicators to grade habitat quality in the field (e.g.,
North Rhine-Westphalia, Germany: Verbücheln et al. 2002; Austria: Ellmauer 2005;
Flanders, Belgium: T’jollyn et al. 2009). Such an approach benefits local conservation managers, by providing data on where habitats are in good or poor condition.
This may be directly used to prioritize the areas where conservation actions are
needed most.
Generally, the indicators cover floristic and/or fauna composition (e.g., number
of key species present), vegetation structure (e.g., height, cover, proportion of dead
wood), disturbances (e.g., invasive species), and landscape configuration (e.g., connectivity and isolation). All these indicators relate to specific properties of the habitat (Bock et al. 2005), and hence contain useful information for managers and
policy-makers (Spanhove et al. 2012; T’jollyn et al. 2009). However, their relative
importance in the eventual conservation status assessment strongly depends on the
context (geographically, socio-economically, etc.). Adding to that the large variation
between and within habitat types, it becomes clear that ‘default’ remote sensing
methods will only be able to address a small number of indicators. A degree of
adaptation of the method will generally be unavoidable.
As an example, we evaluated the indicators that are used for the assessment of
natural habitats in Flanders (T’jollyn et al. 2009) and Germany (PAN & ILÖK
2010). In Flanders, 120 indicators have been defined for 43 habitat types (69 subtypes). Frequently used indicators are the number and spatial coverage of key species, and some widespread threats such as forest, grass and tall herb encroachment
and alien invasive species. The majority of the indicators, however, are only relevant
for a few habitats, and almost half of them (48%) are used for only one habitat type
or subtype (Fig. 2). Figure 3 shows the results of a ‘rarefaction’-like analysis, where
the number of conservation status indicators or key species is modelled as a function
of the number of habitat (sub) types evaluated, for Flanders (T’jollyn et al. 2009)
and Germany (PAN & ILÖK 2010). None of the curves shows signs of nearing a
plateau value, indicating that any habitat type added will further raise the total number of indicators/species on which data need to be gathered.
Available Ground Reference Data Differ
Ground reference data, if available, come in many forms, both sampling-wise and
content-wise (Stehman and Czaplewski 1998). Depending on the applied sampling
design, samples can be provided as points (but referring to a spatial sample unit on
the ground of all possible sizes and shapes) or as objects (polygons). Their selection
can have followed random or non-random sampling principles, and they can have
been collected in the field, or deduced from other (usually older) map information.
J. Vanden Borre et al.
area, habitat range, specific structures and functions (incl. typical species), and
future prospects (ETC/BD 2011; see also Vanden Borre et al. 2011b). To assess the
specific structures and functions, several EU member states have taken an approach
to evaluate local habitat quality at (all or a sample of) individual habitat occurrences, and identified relevant indicators to grade habitat quality in the field (e.g.,
North Rhine-Westphalia, Germany: Verbücheln et al. 2002; Austria: Ellmauer 2005;
Flanders, Belgium: T’jollyn et al. 2009). Such an approach benefits local conservation managers, by providing data on where habitats are in good or poor condition.
This may be directly used to prioritize the areas where conservation actions are
needed most.
Generally, the indicators cover floristic and/or fauna composition (e.g., number
of key species present), vegetation structure (e.g., height, cover, proportion of dead
wood), disturbances (e.g., invasive species), and landscape configuration (e.g., connectivity and isolation). All these indicators relate to specific properties of the habitat (Bock et al. 2005), and hence contain useful information for managers and
policy-makers (Spanhove et al. 2012; T’jollyn et al. 2009). However, their relative
importance in the eventual conservation status assessment strongly depends on the
context (geographically, socio-economically, etc.). Adding to that the large variation
between and within habitat types, it becomes clear that ‘default’ remote sensing
methods will only be able to address a small number of indicators. A degree of
adaptation of the method will generally be unavoidable.
As an example, we evaluated the indicators that are used for the assessment of
natural habitats in Flanders (T’jollyn et al. 2009) and Germany (PAN & ILÖK
2010). In Flanders, 120 indicators have been defined for 43 habitat types (69 subtypes). Frequently used indicators are the number and spatial coverage of key species, and some widespread threats such as forest, grass and tall herb encroachment
and alien invasive species. The majority of the indicators, however, are only relevant
for a few habitats, and almost half of them (48%) are used for only one habitat type
or subtype (Fig. 2). Figure 3 shows the results of a ‘rarefaction’-like analysis, where
the number of conservation status indicators or key species is modelled as a function
of the number of habitat (sub) types evaluated, for Flanders (T’jollyn et al. 2009)
and Germany (PAN & ILÖK 2010). None of the curves shows signs of nearing a
plateau value, indicating that any habitat type added will further raise the total number of indicators/species on which data need to be gathered.
Available Ground Reference Data Differ
Ground reference data, if available, come in many forms, both sampling-wise and
content-wise (Stehman and Czaplewski 1998). Depending on the applied sampling
design, samples can be provided as points (but referring to a spatial sample unit on
the ground of all possible sizes and shapes) or as objects (polygons). Their selection
can have followed random or non-random sampling principles, and they can have
been collected in the field, or deduced from other (usually older) map information.
J. Vanden Borre et al.