Вы находитесь на странице: 1из 10

Landscape Ecol Eng (2009) 5:110

DOI 10.1007/s11355-008-0053-4

ORIGINAL PAPER

Multiscale delineation of landscape planning units based


on spatial variation of land-use patterns in Galicia, NW Spain
lvarez-Lopez
Emilio Daz-Varela Carlos Jose A
Manuel Francisco Marey-Perez

Received: 11 March 2008 / Revised: 20 September 2008 / Accepted: 12 November 2008 / Published online: 5 December 2008
 International Consortium of Landscape and Ecological Engineering and Springer 2008

Abstract The success of landscape planning and environmental management strategies depends largely on the
congruence between the operational scales of landscapes
and the spatial scope of the planning instruments. In order
to achieve good results, landscape planning units should be
designed to fit the structural and functional characteristics
of the landscape. This paper describes a simple, straightforward methodology for the delineation of landscape units
based on the differentiation among areas according to the
trend in heterogeneity of their spatial structure. A movingwindow analysis was performed with varying window
sizes. This analysis enabled modeling of the spatial
response of heterogeneity to multiple scales and detection
of two different domains of scale. The information
obtained was used to delineate a multiscale structure for
landscape units. A subsequent study of the composition and
configuration of landscape structure in these units verified
that they all have distinct characteristics and therefore
different planning needs. The areas thus differentiated can
be considered as targets of specific and differentiated
planning strategies, and can be seen as a preliminary
zonation in which the relevant planning methodologies can
be developed according to each case.

lvarez-Lopez  M. F. Marey-Perez
E. Daz-Varela  C. J. A
Research Group 1716-Projects and Planning,
Department of Agroforestry Engineering,
University of Santiago de Compostela,
Santiago de Compostela, Spain
lvarez-Lopez  M. F. Marey-Perez
E. Daz-Varela (&)  C. J. A
Higher Polytechnic School, Campus Universitario, s/n,
27002 Lugo, Spain
e-mail: emilio.diaz@usc.es

Keywords Landscape ecological planning 


Land-use planning  Scale  Heterogeneity 
Moving-window analysis  Landscape pattern analysis

Introduction
Landscape ecological planning seeks to achieve landscape
sustainability by integrating the ecological function of
landscapes with the development of human activities (Van
Langevelde 1994; Forman 1995; Rookwood 1995; Ahern
1999; Hawkins and Selman 2002). To this end, landscape
ecological planning must take into consideration the spatial
variation of landscape structure, i.e., its heterogeneity
(Dramstad et al. 2001) as well as the processes that
determine both the functioning and the dynamics of
change. Nevertheless, landscape heterogeneity operates at
characteristic scales (Carlile et al. 1989; Meentemeyer
1989; Wiens et al. 2002) and is hierarchically organized in
functional levels (Rowe 1961; Allen and Starr 1982;
ONeill et al. 1986; Kotliar and Wiens 1990; King 1997).
Therefore coherence between such a hierarchical organization and the geographical areas specified in the planning
instruments must be pursued in order for spatial matching
of the conflicts to be resolved (Wilson et al. 1999;
Schneider 2001; Saunders and Briggs 2002; Kamada 2005;
Bastian et al. 2006).
One possible method of achieving this objective is to
delineate landscape planning units in a hierarchy of complexity of space (Zonneveld 1995), by dividing the
horizontal pattern of the landscape into areas or units that
are embedded into other areas or units, each of which
present specific characteristics in terms of structure and
function that make them different from other nearby units.
This approach has been widely used in the delineation of

123

landscape and ecological units (e.g., Klijn 1991; Dramstad


et al. 1996; Cleland et al. 1997; Mander et al. 2003;
McMahon et al. 2004; Nakamura et al. 2005; Lowe et al.
2006). A variety of methods, classification schemes, and
nomenclatures have been used to distinguish among
different hierarchical levels. Nevertheless, landscape
heterogeneity is conditioned by forces that may differ
depending on the hierarchical level considered (Meisel and
Turner 1998); for instance, climate and geomorphology
may influence landscape pattern and processes at many
hierarchical levels (Bailey 2005; see also Xu et al. 2004),
and are critical at upper, broader levels. Both factors
have been used to delineate units at national, continental,
and global levels (Meeus et al. 1990; Milanova and
Kushlin 1993; Mata-Olmo and Sanz-Herraiz 2003;
Jongman et al. 2006). At intermediate or lower levels,
land use is the most important factor that structures
landscapes, together with landform, biotic community,
and population distribution (Meisel and Turner 1998).
Consequently, land use and cover patterns can be used
at meso- and microscales as spatial references to differentiate landscape units with different structural and
functional characteristics.
The aim of this study is twofold: firstly, to analyze the
response to scale of landscape metrics as regards landscape
heterogeneity. Secondly, to detect reference areas for
landscape planning units based on the previously identified
scale-response types of heterogeneity. Because of the
existing relation between structure and function in the
landscape (Gonzalez-Bernaldez 1981; Turner 1989), it was
assumed that each of the units differentiated by structural
characteristics will require different planning measures,
adapted to their specific characteristics and defined
according to different administrative planning levels.

Material and methods


Study area
The area chosen for study was A Marina Oriental, a region
in the northeast of the Autonomous Community of Galicia,
Spain. The study region covers a total area of 400 km2 and
has a population of 17,287, with irregular density and rural
depopulation in the mountainous areas of the interior
(Fig. 1). Altitude ranges from sea level to 950 m above sea
level. A Marina Oriental is a subregional administrative
division that includes four municipalities (Ribadeo,
Barreiros, Trabada, and A Pontenova), which are in turn
subdivided into 39 submunicipal units known as parishes.
Two of the municipalities in this region are coastal areas
where tourism is becoming increasingly important and
where the principal economic activities are changing

123

Landscape Ecol Eng (2009) 5:110

Fig. 1 Location of the study area

towards the service sector. Towards the interior, the relief


is progressively more mountainous, and agricultural and
forestry activities prevail. Such activities determine a land
cover consisting mainly of fields (predominantly pasture
land) and forestry plantations (predominantly Eucalyptus
globulus Labill., Pinus pinaster Ait., and Pinus radiata D.
Don). In addition, there are areas of considerable ecological interest in terms of land cover, such as patches of
Quercus robur L. native woodlands and wetlands associated with estuaries.
The choice of the study area was based on an apparent
diversity of landscape characteristics: the coast-inland
gradient was reflected, and imbalances between agricultural and forestry areas with different intensities of
production were observed in this area. We expect that the
diversity reflected in the landscape pattern will be an
interesting subject for analysis of planning at different
scales in heterogeneous areas.
Digital cartography
Digital raster maps of land cover were used. Such maps
were obtained from interpretation of 1:18,000 scale aerial
photography and subsequent ground-truthing (Marey-Perez
2003), processed with ArcView 3.3 software. Pixel
size was 10 m 9 10 m, and the minimum mapping unit
(MMU) was 400 m2. The legend was prepared with a
dynamic criterion (Kuchler 1988), for which a lack of
stable climax vegetation was assumed, and an attempt was
made to represent states of land cover transition associated
with the current land use, which generated 24 classes.

Landscape Ecol Eng (2009) 5:110

Heterogeneity analysis
Analysis of the differences in landscape structure was
approached through a quantitative study of heterogeneity.
This was carried out by applying indices or landscape
pattern metrics to a categorical map (Romme 1982; ONeill
et al. 1988; Hoover and Parker 1991; Botequilha-Leitao
and Ahern 2002; McGarigal et al. 2002; Botequilha-Leitao
et al. 2006).
Application of indices to an entire study area does not
reveal the spatial distribution of the variables studied.
Consequently, a system of subdivision into intermediate
scales of analysis becomes necessary (ONeill et al. 1996).
The solution adopted in this study was the application of
moving windows. A moving window consists of a subset
(or window) of the raster map analyzed, for which a
landscape metric is calculated; the result is summarized in
the central pixel of the window, and the whole process is
repeated for each pixel in the land cover map. The maps
obtained with this procedure show the spatial distribution
of landscape structure (Berry 2001) based on the results of
the indices used. When different window sizes are applied
to the same map, results can provide information about the
behavior of an index at different scales. The capacity of
variable-size moving windows to represent different scales
or spatial extents has been used in the analysis of habitat
suitability for species with different home ranges (Riitters
et al. 1997; Roshier et al. 2001; Gaucherel et al. 2007); in
the detection of landscape spatial variation (Eiden et al.
2000); in studies of scale dependence of the components of
forest fragmentation (Riitters et al. 2002); in land-cover
pattern analysis (Riitters et al. 2000); and in the analysis of
landscape pattern in urban environments (Kong and Nakagoshi 2006; Pham and Nakagoshi 2007). We expect that
the differences identified between scales will enable the
definition of representative levels that will help in the
delineation of landscape planning units.
In the present study, the analysis was conducted for six
different window sizes (with diameters of 250, 500, 750,
1,000, 1,250, and 1,500 m) with FRAGSTATS software
(McGarigal et al. 2002) on grid formats generated in the
ArcView 3.3 GIS. A circular window shape was used, so
that the distance to the central pixel was as similar as
possible throughout the window.
The choice of a minimum window size is essential to
avoid distortion of the results caused by pixelation effects
associated with excessively small windows. Hargis et al.
(1997) established a minimum area of 101 9 101 pixels,
after having found inaccuracies in the values obtained from
64 9 64 pixel maps. Nevertheless, the analysis of spatial
extent conducted by Saura and Martnez-Millan (2001)
revealed meaningful results from a minimum window size
of 50 pixels, and Riitters et al. (1997) started from far

smaller sizes (9 pixels). In this study, we started from a


relatively small window size (250 m, equivalent to 25
pixels) so as not to disregard possible results at small scale
ranges, while being careful in the subsequent interpretation
of results.
In order to select the most suitable index for the research
objectives, preliminary tests were performed at the landscape level (i.e., considering all landscape elements
without differentiating among classes), for a total of 19
landscape indices. The Shannon diversity index (SHDI)
was found to be the most appropriate, due to its consistency
across scales (Wu et al. 2002; Diaz-Varela et al. 2008), its
capacity to reflect both configuration and composition
components of heterogeneity (Eiden et al. 2000), and its
sensitivity to rare or low-abundance classes (McGarigal
et al. 2002; Nagendra 2002). Thus, the SHDI reflects basic
components of heterogeneity such as the number and
proportion of classes, their spatial distribution, and the
shape of patches (Li and Reynolds 1995).
Detection of scale effects
In order to differentiate representative scales with the
moving-window analysis, the dissimilarity (S) observed
between the maps obtained was analyzed with the
following expression (ONeill et al. 1996; Saura and
Martnez-Millan 2001):
S

M1500  Mi
 100;
SD1500

where M1500 is the average value of an index for a map


generated with a 1,500-m window, Mi is the average
value of an index for a map generated with other windows,
and SD1500 is the standard deviation of an index value for a
map generated with a 1,500-m window. Because S
expresses the difference among the results obtained, the
observation of marked changes in S indicates differences in
the response of the landscape metric to the scale of
analysis. Variation in S is expected to be progressive and
therefore, in order to quantify changes in the values of S,
the following expression was used:


DVi
 1;
pi
DWi
where DVi is the percentage increase in the values of S with
respect its maximum value with each change in window
size (thus change in scale), DWi is the percentual increase
in moving-window sizes, and i is each increment in scale.
pi shows the variation of the slope of the curve of S, so
pi [ 0 would imply a variation in dissimilarity more than
proportional with respect to scale (window size) and pi \ 0
a variation less than proportional with respect to scale.
Thus, at least two different domains of scale could be

123

defined: one in which pi [ 0, where the moving window


only detects local effects, and thus is highly scale dependent, and another one in which pi \ 0, where the
heterogeneity trend is progressively less dependent on
window size, thus showing a tendency to self-similarity in
the response of heterogeneity across scales. These two
main response groups were categorized respectively as
microscale (site scale) and mesoscale (landscape scale),
following the terminology used by other authors (Dramstad
et al. 1996; Mander et al. 2003).
For each different response group, a representative map
was chosen from among the maps obtained with the different window sizes. This representative map was then
used to define areas based on the spatial variability of
heterogeneity. The map was simplified by reclassifying the
heterogeneity values that were higher or lower than the
overall average. The zones with values higher than the
overall average were considered as tending to greater heterogeneity and the zones with lower values were
considered as tending to homogeneity (less heterogeneity).
For reasons of continuity, the zones of area smaller than the
calculation window were eliminated.
Finally, in order to test if the areas obtained define
homogeneous zones that are different from each other, the
landscape pattern of the resulting zones was characterized
by landscape composition analysis, i.e., analysis of the
landscape mosaic characteristics that were not linked to the
geometry or spatial position of the elements. ArcView 3.3
software was used to calculate number of classes,
percentage per class, and number and average size of
patches within each area, based on the assumption that the
differentiated units should present marked contrasts in
composition.

Results and discussion


Multiscale trends in landscape metrics
Six output maps were obtained with the spatial distribution
for each of the selected indices. The extent of the information that can be analyzed is smaller than the extent of
the original land cover map, which is reduced at the edges
by a distance equal to the radius of the window used. Such
a reduction occurs because the value of the central pixel is
annulled if the window covers one or more pixels with null
value within its radius of action.
Figure 2 shows examples of the output maps for SHDI,
reclassified according to the distance from the mean value
for the whole map, which allows for an intuitive observation
of variability in the spatial distribution of results. SHDI
maps provided useful spatial variation in heterogeneity
when calculated with moving windows, in accordance with

123

Landscape Ecol Eng (2009) 5:110

Fig. 2 Results obtained for the SHDI index in windows with


diameters of 250, 750, and 1,500 m. Each color change represents
one standard deviation above (darker) or below (lighter) the mean
(dotted line). Note the difference in the spatial distribution of the
index value and therefore in landscape heterogeneity

other studies (Burel and Baudry 1999; Eiden et al. 2000),


although references to such an application are still scarce.
Some studies consider the Shannon index as inadequate for
the study of landscape heterogeneity when the number of
classes is low (Herzog et al. 2001), or when specific richness is targeted (Fjellstad et al. 2001), but the capacity of
the index to reflect simultaneously configuration and composition of the landscape (Eiden et al. 2000) and its
robustness in response to changes in scale (Wu et al. 2002)
make it a better choice than other indices that are based
exclusively on the geometry of the landscape pattern.
The maps obtained for the smallest window (250 m)
reveal highly fragmented zones distributed throughout the
study area. The gradual increase in window size allows us
to observe the merging of areas with similar results,
showing similar spatial distributions (500 m), and the clear
differentiation among larger zones for calculations performed with larger windows (1,500 m). The observations
will be interpreted as samples of the spatial distribution of
heterogeneity at different scales.
As regards the effects of variation in the values obtained
for SHDI according to scale of observation (Fig. 3, a, b, c,
and d), the following situations can be distinguished:
1.

2.

Increase in value with increase in scale (e.g., in


Fig. 3a, SHDI increases from 0.32 to 1.63): areas in
which the results obtained for small windows show
low values of heterogeneity, which increase with the
increase in window size. This effect is caused by the
gradual incorporation of landscape elements with the
increase in window size.
Decrease in value with increase in scale (e.g., in
Fig. 3b, SHDI decreases from 1.52 to 0.86): areas in
which small scales show high values of heterogeneity,

Landscape Ecol Eng (2009) 5:110

Fig. 3 Types of effects of


variation in the SHDI index
value with scale. Three possible
effects can be compared:
increase (a), decrease (b), and
nearly constant heterogeneity
(c) or homogeneity (d) in the
value (see text for details).
Window sizes are represented as
circles for comparison

3.

which decrease with the increase in scale. In this case,


the smallest windows distinguish local effects that are
imperceptible with larger windows.
The value remains constant within a range, whether
heterogeneous (e.g., in Fig. 3c, SHDI ranges from 2.02
to 2.13) or homogeneous (e.g., in Fig. 3d, SHDI ranges
from 0.58 to 0.55): areas in which little variation is
detected in the values of heterogeneity metrics with
changes in scale, normally because of the presence of a
very continuous type of cover or a small number of
large patches, or a continuous, chaotic mosaic of tiny
patches.

As expected, the values of S decrease gradually as they


approach the value calculated for the 1,500-m window,

but in a nonconstant manner. pi values were 0.51 for the


interval DWi between 500 and 750 m , and -0.08 for the
interval between 750 and 1,000 m . Consequently, scale
threshold between micro- and mesoscale domains is
expected to be around 750 m (Fig. 4). Maps obtained
with moving windows below 750 m show a fragmented,
noncontinuous spatial distribution of heterogeneity. Size,
shape, and distribution of landscape patches have a major
impact on this. For instance, if the narrower dimension of
one patch is longer than the window diameter, the SHDI
value becomes zero, thus indicating complete homogeneity. When a large part of the landscape meets this
condition, only the local variation of heterogeneity in
highly fragmented spots can be indicated, interspersed

123

Fig. 4 Variation in S with window size. Detail of the results for each
window size in a specific location are shown for comparison (dotted
line corresponds with average value)

within wide homogeneous areas. Such spatial configuration


defines the microscale. Above 750 m , the effect due to
the patch/window size rate is overcome, and maps
progressively show a continuous less-scale-dependent distribution of heterogeneity. This defines the mesoscale.
Similar examples of this double scale differentiation can
be found in the literature. In the application of a boxcounting method for computing the fractal dimension of
hawk home ranges at several scales, Loehle (1990) differentiated two responses when plotting fractal dimension
against box length: fragmentation (smaller box sizes) and
self-similarity (larger box sizes). Similarly, Wilson and
Keeling (2000) used moving windows to analyze spatotemporal data on predator-prey and genetic models, and
differentiated random behavior in the results obtained with
small windows from deterministic behavior in the results
obtained with larger windows, because of the averaging
effect of the latter. A clear correspondence can be observed
with the results obtained in this research: both fragmentation and random effects are similar to the local
heterogeneity found at microscale, while the self-similarity
and deterministic behavior of large windows are parallel to
the progressive scale independence at mesoscale in the
present study.
Landscape planning units based on heterogeneity
response to scale
In order to make the findings regarding multiscale behavior
of indices useful for landscape planning, the response to
scale must be spatially categorized. Our aim was to classify
the study area into areas with a different response to the
scale. This was done by: choosing a representative map for

123

Landscape Ecol Eng (2009) 5:110

Fig. 5 Graphical summary of the methodology used and its results.


Upper left, tridimensional view of distribution of SHDI values with
moving window 1,500 m, and the binary map, obtained by cutting
at the average value. Lower left, the same for moving window
500 m. Right, scale response areas obtained after overlaying: Ho,
heterogeneity at microscale, homogeneity at mesoscale; Hc, high
heterogeneity, not scale dependent; Oh, homogeneity at microscale,
heterogeneity at mesoscale; Oc, low heterogeneity, not scale
dependent

each scale domain; simplifying the spatial distribution of


scale responses represented in each map; mapping the
different multiscale responses by overlaying the former
information; and finally characterizing each of the defined
areas, to check for differences in heterogeneity and to help
planning decisions.
To represent the microscale level, the map obtained with
500 m was chosen, as a compromise among the values
showing a more fragmented behavior of the landscape.
Nevertheless, to represent the mesoscale level, the map
farthest from the threshold (i.e., 1,500 m ) was preferred
as it was expected to represent more adequately scaleindependent variation in the trend in heterogeneity.
Representative maps were simplified in order to differentiate areas that tend to greater or less heterogeneity
within each map. This was done by reclassifying all the
SHDI values in each map into two classes: values above
the average (representing a trend to heterogeneity), and
values below the average (representing a trend to homogeneity). This procedure enables basic standardization of
values, making the spatial distribution of heterogeneity
independent from absolute values, and allowing the direct
comparison of heterogeneity trends at micro- and mesoscales. The results gave rise to nine areas at the mesoscale
level and 77 areas at the microscale level (Fig. 5).
Overlapping of the two levels can then be used to represent the effects of increase, decrease, and constancy of
heterogeneity in a map, with scale as described in the
section above. Consequently, four types of areas (two

Landscape Ecol Eng (2009) 5:110

Table 1 Landscape composition [No, number of patches; %, relative


proportion; Aa, average area (ha)] by type of heterogeneity:
Hc, constant heterogeneity; Oc, constant homogeneity (low
Land cover

Ho
No.

heterogeneity); Hm, heterogeneity at microscale, homogeneity at


mesoscale; Om, homogeneity at microscale, heterogeneity at
mesoscale
Hc

Aa (ha)

No.

Oh
%

Aa (ha)

No.

Oc
%

Aa (ha)

No.

Aa (ha)

Seminatural deciduous forest

57

2.83

2.29

285

8.23

4.40

47

8.28

7.57

57

3.53

9.79

Riparian vegetation

31

0.23

0.34

10

0.08

1.20

0.04

0.26

29

0.05

0.28

Intensive yield coniferous forest

89

4.24

2.20

266

5.15

2.95

57

5.43

4.10

73

1.89

4.10

298

32.74

5.07

958

25.07

3.99

234

31.49

5.79

293

38.32

20.70

Intensive yield forest; mixed species

35

0.82

1.08

215

3.14

2.23

30

0.72

1.03

34

0.19

0.86

Seminatural and yield forest, mixed

77

1.55

0.93

422

7.80

2.82

81

2.87

1.52

103

0.43

0.67

0.02

0.37

0.01

0.48

0.04

0.40

14

0.06

0.66

297

13.33

2.07

668

12.80

2.92

158

22.15

6.03

359

5.86

2.58

59

1.46

1.14

510

8.13

2.43

79

2.26

1.23

88

0.42

0.76

240

11.74

2.26

561

8.48

2.30

134

5.24

1.68

243

3.13

2.04

Intensive yield broadleaved forest

Orchards
Shrubland
Sparsely forested shrubland (seminatural)
Sparsely forested shrubland (yield)
Sparsely forested shrubland with rocks

0.00

0.00

0.03

0.78

0.06

2.60

0.04

5.92

Shrubland with rocks

0.07

1.08

0.00

0.66

0.12

4.99

0.03

0.71

68

1.57

1.06

109

0.58

0.82

45

0.69

0.65

206

1.19

0.91

206

25.39

5.69

369

16.93

6.99

117

17.30

6.36

170

37.24

34.66

65
24

1.68
0.96

1.19
1.84

151
120

0.84
1.31

0.84
1.67

26
28

0.16
1.25

0.27
1.91

126
39

0.73
0.38

0.92
1.56

Marsh

0.01

0.15

0.14

4.22

0.00

0.20

1.59

50.37

Coastal cliffs and beaches

0.05

0.35

0.00

0.00

0.00

0.00

0.00

0.06

Rocks

0.14

1.11

0.02

0.69

0.01

0.49

0.01

0.26

Road/railroad network

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

51

0.82

0.74

105

0.82

1.19

24

1.73

3.10

118

1.98

2.65

Mines and dumping areas

0.07

1.01

0.13

2.90

0.00

0.00

0.05

1.20

Lakes and estuaries

0.03

0.23

0.00

0.37

0.00

0.04

2.67

141.03

56

0.27

0.22

53

0.29

0.84

0.16

0.96

66

0.20

0.47

Crops
Grasslands
Wet meadows
Pastureland

Buildings

Rivers
Number of dominant classes

% Dominant classes
Total number of patches

71.46
1,682

79.64
4,830

70.94
1,085

75.56
2,051

Dominant classes are highlighted in bold-italic

trends in heterogeneity at microscale 9 two trends in heterogeneity at mesoscale) can finally be defined:

Areas of heterogeneity at microscale/homogeneity


mesoscale (Ho in Fig. 5)
Areas of heterogeneity at microscale/heterogeneity
mesoscale (Hc in Fig. 5)
Areas of homogeneity at microscale/heterogeneity
mesoscale (Oh in Fig. 5)
Areas of homogeneity at microscale/homogeneity
mesoscale (Oc in Fig. 5)

at
at
at
at

Correspondence between the detected heterogeneity


trend and landscape pattern was analyzed for each area
through the previously described composition metrics.
Studies found in the literature show how the simplicity and
straightforwardness of landscape composition metrics
performs well in the assessment of differences among

landscapes (Trani and Gilles 1999; Croissant 2004; CalvoIglesias et al. 2006). In the present study, it is supposed
that, if the classification of trend areas is adequate, substantial differences can be found in the results of landscape
composition metrics for each type. Such results are shown
in Table 1.
Thus, the areas of homogeneity at micro- and mesoscale
(Oc: 15,822 ha) comprise 23 cover classes and 2,051 patches. More than 75% of the total area is covered by the two
dominant classes, revealing a homogeneous, simplified
pattern. This contrasts with the areas of heterogeneity at
micro- and mesoscale (Hc: 15,242 ha), which comprise 22
cover classes and 4,830 patches. The six dominant classes
occupy around 79% of the total area. Together these findings
indicate a higher level of heterogeneity. We can find also
strong differences in the areas that are homogeneous at
microscale and heterogeneous at mesoscale (Oh: 4,299 ha),

123

Fig. 6 Visual example of differentiation among heterogeneous and


homogeneous areas. On the right, the area presenting mesoscale
heterogeneity is subdivided to the areas showing microscale homogeneity (upper frame), with a pattern of relatively large patches of
forest and shrubland, and microscale heterogeneity (bottom frame)
showing a fragmented arrangement of small patches. These areas can
be used as references for landscape planning units

comprising 21 classes and 1,085 patches. More than 70% of


the area is covered by three classes, with a mean patch area of
about 6 ha. Finally, the areas that are heterogeneous at
microscale and homogeneous at mesoscale (Ho: 4,612 ha)
include 22 cover types and 1,682 patches, with three of them
covering 71.46% of the total area, with a considerably
smaller average patch size (about 4 ha). Consequently, we
conclude from the composition results that the classification
in heterogeneity trend areas is satisfactory. In the example
shown in Fig. 6, there are apparent differences in landscape
pattern between the two trends in microscale within a
mesoscale heterogeneous unit.
The areas representing these four different responses can
be used as a decision support tool for the assignment of
specific planning and management strategies. Thus, areas
heterogeneous at microscale and homogeneous at mesoscale (e.g., urbanization processes inserted in agricultural
land) would require an increase in detail in the planning
efforts that is not necessary in areas with homogeneity at
both levels (e.g., forestry areas). Similarly, areas homogeneous at microscale and heterogeneous at mesoscale
enable identification of coarse-grained landscape elements
for a different planning treatment (e.g., large deciduous
forest patches inserted in heterogeneous agricultural areas).

Conclusions
This study proposes a straightforward methodology for the
identification of landscape planning units at different scales

123

Landscape Ecol Eng (2009) 5:110

within the framework of landscape ecological planning.


Land-use pattern heterogeneity is analyzed by calculating
landscape metrics with moving windows of different sizes.
The usefulness of the moving-window approach is reflected
in the ability to show the spatial distribution of heterogeneity across the landscape. In addition, the use of moving
windows of different sizes enables detection of variation in
heterogeneity across scales, and the identification of scale
domains for this variation. This enables the detection of
areas with a similar scale-response for the heterogeneity.
Subsequent comparison of landscape composition among
such areas allows for the observation of structural differences that suggest dissimilar ecological functioning in each
area and that will present different planning needs. The
methodology can thus be integrated in decision support
systems to enhance the possibilities of success of landscape
ecological planning processes, by adapting planning units
to changes in heterogeneity across scales.
Acknowledgments This research was partly funded by a grant for
research projects from the Galician government (Xunta de Galicia)
(PGIDT02RAG29103PR), and by a Research Contract within the
framework of the Isidro Parga Pondal research programme, from the
same institution. The authors thank two anonymous reviewers for
their helpful suggestions and comments. All the experiments developed in this work, including data acquisition and processing, comply
with the current regional, state, and international laws and agreements. We also thank Dr Christine Francis for correcting the English
grammar of the text.

References
Ahern J (1999) Spatial concepts, planning strategies, and future
scenarios: a framework method for integrating landscape ecology and landscape planning. In: Klopatek JM, Gardner RH (eds)
Landscape ecological analysis. Issues and applications. SpringerVerlag, New York, pp 175201
Allen TFH, Starr TB (1982) Hierarchy: perspectives for ecological
complexity. University of Chicago Press, Chicago
Bailey RG (2005) Identifying ecoregion boundaries. Environ Manage
34:1426. doi:10.1007/s00267-003-0163-6
Bastian O, Kronert R, Lipsky Z (2006) Landscape diagnosis on
different space and time scalesa challenge for landscape
planning. Landsc Ecol 21:359374
Berry JK (2001) Map analysis. Procedures and applications in GIS
modelling. Basis Press, USA. Available online, URL: http://
www.innovativegis.com/basis/MapAnalysis/Topic9/Topic9.html.
Accessed 1 Sept 2008
Botequilha-Leitao A, Ahern J (2002) Applying landscape ecological
concepts and metrics in sustainable landscape planning. Landsc
Urban Plan 59:6593
Botequilha-Leitao A, Miller J, Ahern J, McGarigal K (2006) Measuring
landscapes: a planners handbook. Island, Washington
Burel F, Baudry J (1999) Ecologie du paysage. Concepts, methodes et
applications. Editions Tec & Doc, Paris
Calvo-Iglesias MS, Crecente-Maseda R, Fra-Paleo U (2006) Exploring farmers knowledge as a source of information on past and
present cultural landscapesa case study from NW Spain.
Environ Manage 78:334343

Landscape Ecol Eng (2009) 5:110


Carlile DW, Skalski JR, Batker JE, Thomas JM, Cullinan VI (1989)
Determination of ecological scale. Landsc Ecol 2:203213
Cleland DT, Avers PE, McNab WH, Jensen ME, Bailey RG, King T,
Russell WE (1997) National hierarchical framework of ecological units. In: Boyce MS, Haney A (eds) Ecosystem management
applications for sustainable forest and wildlife resources. Yale
University Press, New Haven, pp 181200
Croissant C (2004) Landscape patterns and parcel boundaries: an
analysis of composition and configuration of land use and land
cover in south-central Indiana. Agric Ecosyst Environ 101:219
232
Diaz-Varela ER, Marey-Perez MF, Rigueiro-Rodriguez A, AlvarezAlvarez P (2008) Landscape metrics for characterization of
forest landscapes in a sustainable management framework:
potential application and prevention of misuse. Ann For Sci (in
press)
Dramstad WE, Olson JD, Forman RTT (1996) Landscape ecology
principles in landscape architecture and land-use planning.
Island, Washington
Dramstad WE, Fry G, Fjellstad WJ, Skar B, Helliksen W, Sollund
MLB, Tveit MS, Geelmuyden AK, Framstad E (2001) Integrating landscape-based values-Norwegian monitoring of
agricultural landscapes. Landsc Urban Plan 57:257268
Eiden G, Kayadjanian M, Vidal C (2000) Quantifying landscape
structures: spatial and temporal dimensions. In: European Commission, from land cover to landscape diversity. Available online,
URL: http://europa.eu.int/comm/agriculture/publi/landscape/
index.html. Accessed 1 Sept 2008
Fjellstad WJ, Dramstad WE, Strand GH, Fry GLA (2001) Heterogeneity as a measure of spatial pattern for monitoring
agricultural landscapes. Nor Geogr Tidsskr 55:7176
Forman RTT (1995) Land mosaics. The ecology of landscapes and
regions. Cambridge University Press, Cambridge
Gaucherel C, Burel F, Baudry J (2007) Multiscale and surface pattern
analysis of the effect of landscape pattern on carabid beetles
distribution. Ecol Indic 7:598609
Gonzalez-Bernaldez F (1981) Ecologa y Paisaje. Blume, Madrid
Hargis CD, Bissonette JA, David JL (1997) Understanding measures
of landscape pattern. In: Bissonette JA (ed) Wildlife and
landscape ecology: effects of pattern and scale. Springer-Verlag,
New York, pp 231261
Hawkins V, Selman P (2002) Landscape scale planning: exploring
alternative land use scenarios. Landsc Urban Plan 60:211224
Herzog F, Lausch A, Muller E, Thulke HH, Steinhardt U, Lehmann S
(2001) Landscape metrics for assessment of landscape destruction and rehabilitation. Environ Manage 27:91107
Hoover SR, Parker AJ (1991) Spatial components of biotic diversity
in landscapes of Georgia, USA. Landsc Ecol 5:125136
Jongman RHG, Bunce RGH, Metzger MJ, Mucher CA, Howard DC,
Mateus VL (2006) Objectives and applications of a statistical
environmental stratification of Europe. Landsc Ecol 21:409419.
doi:10.1007/s10980-005-6428-0
Kamada M (2005) Hierarchically structured approach for restoring
natural forest-trial in Tokushima Prefecture, Shikoku, Japan.
Landsc Ecol Eng 1:6170. doi:10.1007/s11355-005-0005-1
King AW (1997) Hierarchy theory: a guide to system structure for
wildlife biologists. In: Bissonette JA (ed) Wildlife and landscape
ecology: effects of pattern and scale. Springer-Verlag, New
York, pp 185212
Klijn F (1991) Hierarchical classification of ecosystems: a tool for
susceptibility analysis and quality evaluation for environmental
policy. In: Ravera O (ed) Terrestrial and aquatic ecosystems.
Perturbation and recovery. Ellis Horwood, London
Kong F, Nakagoshi N (2006) Spatial-temporal gradient analysis of
urban green spaces in Jinan, China. Landsc Urban Plan
78:147164

9
Kotliar NB, Wiens JA (1990) Multiple scales of patchiness and patch
structure: a hierarchical framework for the study of heterogeneity. Oikos 59:253260
Kuchler AW (1988) The classification of vegetation. In: Kuchler AW,
Zonneveld IS (eds) Vegetation mapping. Kluwer Academic,
Dordrecht, pp 6780
Li H, Reynolds JF (1995) On definition and quantification of
heterogeneity. Oikos 73:280284
Loehle C (1990) Home range. A fractal approach. Landsc Ecol 5:39
52
Lowe WH, Likens GE, Power ME (2006) Linking scales in stream
ecology. Bioscience 56:591597
, Kulvik M, Jongman RHG (2003) Scaling in territorial
Mander U
ecological networks. Landschap 20:113127
Marey-Perez MF (2003) Tenencia de la tierra en Galicia. Modelo para
la caracterizacion de los propietarios forestales. Ph.D. Thesis.
Digital edition (CD), Servicio de Publicaciones de la Universidad de Santiago de Compostela, Santiago de Compostela, Spain,
641 pp
Mata-Olmo R, Sanz-Herraiz C (2003) Atlas de los paisajes de Espana.
Ministerio de Medio Ambiente. Centro de Publicaciones, Madrid
McGarigal K, Cushman SA, Neel MC, Ene E (2002) FRAGSTATS:
Spatial Pattern Analysis Program for categorical maps. Available online, URL: www.umass.edu/landeco/research/fragstats/
fragstats.html. Accessed 1 Sept 2008
McMahon G, Wiken ED, Gauthier DA (2004) Toward a scientifically
rigorous basis for developing mapped ecological regions.
Environ Manage 34:111124. doi:10.1007/s00267-004-0170-2
Meentemeyer V (1989) Geographical perspectives of space, time and
scale. Landsc Ecol 3:163173
Meeus JHA, Wijermans MP, Vroom MJ (1990) Agricultural
landscapes in Europe and their transformation. Landsc Urban
Plan 18:289352
Meisel JE, Turner MG (1998) Scale detection in real and artificial
landscapes using semivariance analysis. Landsc Ecol 13:347
362
Milanova EV, Kushlin AV (1993) World map of present day
landscapes, an explanatory note. Department of World Physical
Geography and Geoecology. Moscow State University, UNEP
Nagendra H (2002) Opposite trends in response for the Shannon and
Simpson indices of landscape diversity. Appl Geogr 22:175186
Nakamura F, Inahara S, Kaneko M (2005) A hierarchical approach to
ecosystem assessment of restoration planning at regional,
catchment and local scales in Japan. Landsc Ecol Eng 1:43
52. doi:10.1007/s11355-005-0004-2
ONeill RV, De Angelis DL, Waide JB, Allen TFH (1986) A
hierarchical concept of ecosystems. Princeton University Press,
Princeton
ONeill RV, Krummel JR, Gardner RH, Sugihara G, Jackson B,
DeAngelis DL, Milne BT, Turner MG, Zygmunt B, Christensen
SW, Dale VH, Graham RL (1988) Indices of landscape pattern.
Landsc Ecol 1:153162
ONeill RV, Hunsaker CT, Timmins SP, Jackson BL, Jones KB,
Riitters KH, Wickham JD (1996) Scale problems in reporting landscape pattern at the regional scale. Landsc Ecol 11:169
180
Pham DU, Nakagoshi N (2007) Analyzing urban green space pattern
and eco-network in Hanoi, Vietnam. Landsc Ecol Eng 3:143
157. doi:10.1007/s11355-007-0030-3
Riitters KH, ONeill RV, Jones KB (1997) Assessing habitat
suitability at multiple scales: a landscape-level approach. Biol
Conserv 81:191202
Riitters KH, Wickham JD, Vogelmann JE, Jones KB (2000) National
land-cover pattern data. Ecology 81:604. Available online, URL:
http://www.esapubs.org/archive/ecol/E081/004/metadata.html.
Accessed 1 Sept 2008

123

10
Riitters KH, Wickham JD, ONeill RV, Jones KB, Smith ER,
Coulston JW, Wade TG, Smith JH (2002) Fragmentation of
continental United States forests. Ecosystems 5:815822
Romme WH (1982) Fire and landscape diversity in subalpine forests
of Yellowstone National Park. Ecol Monogr 52:199221
Rookwood P (1995) Landscape planning for biodiversity. Landsc
Urban Plan 31:379385
Roshier DA, Robertson AI, Kingsford RT, Green DG (2001)
Continental-scale interactions with temporary resources may
explain the paradox of large populations of desert waterbirds in
Australia. Landsc Ecol 16:547556
Rowe JS (1961) The level-of-integration concept and ecology.
Ecology 42:420427
Saunders DA, Briggs SV (2002) Nature grows in straight lines- or
does she? What are the consequences of the mismatch between
human-imposed linear boundaries and ecosystem boundaries?
An Australian example. Landsc Urban Plan 61:7182
Saura S, Martnez-Millan J (2001) Sensitivity of landscape pattern
metrics to map spatial extent. Photogramm Eng Remote Sens
67:10271036
Schneider DC (2001) The rise of the concept of scale in ecology.
Bioscience 51:545553
Trani MK, Giles RH (1999) An analysis of deforestation: metrics
used to describe pattern change. For Ecol Manag 114:459470
Turner MG (1989) Landscape ecology: the effect of pattern on
process. Annu Rev Ecol Syst 20:171197

123

Landscape Ecol Eng (2009) 5:110


Van Langevelde F (1994) Conceptual integration of landscape
planning and landscape ecology, with a focus on the Netherlands. In: Cook EA, Van Lier HN (eds) Landscape planning and
ecological networks. Elsevier, Amsterdam, pp 2769
Wiens JA, Van Horne B, Noon BR (2002) Integrating landscape
structure and scale into natural resource management. In: Liu J,
Taylor WW (eds) Integrating landscape ecology into natural
resource management. Cambridge University Press, Cambridge,
pp 2367
Wilson HB, Keeling MJ (2000) Spatial scales and low-dimensional
deterministic dynamics. In: Dieckman U, Law R, Metz JAJ (eds)
The geometry of ecological interactions. Cambridge University
Press, London
Wilson J, Low B, Constanza R, Ostrom E (1999) Scale misperceptions and the spatial dynamics of a social-ecological system.
Ecol Econ 31:243257
Wu J, Shen W, Sun W, Tueller PT (2002) Empirical patterns of the
effects of changing scale on landscape metrics. Landsc Ecol
17:761782
Xu M, Qi Y, Chen J, Song B (2004) Scale-dependent relationships
between landscape structure and microclimate. Plant Ecol
173:3957
Zonneveld IS (1995) Land ecology: an introduction of landscape
ecology as a base for land evaluation. land management and
conservation. SPB Academic, Amsterdam

Вам также может понравиться