Geoinformatics 2004 Proc. 12th Int. Conf. on Geoinformatics − Geospatial Information Research: Bridging the Pacific and Atlantic University of Gävle, Sweden, 7-9 June 2004
LANDSCAPE INDICES FOR COMPARISON OF SPATIAL FOREST PATTERNS IN DIFFERENT GEOGRAPHICAL REGIONS Eva M. De Clercq and Robert R. De Wulf Laboratory of Forest Management and Spatial Information Techniques, Ghent University, Coupure Links 653, 9000 Ghent,
[email protected], Tel + 32 9 264 61 01, Fax + 32 9 264 62 40
Abstract Present management, aimed at conservation and optimal use of forest resources, requires monitoring of environmental changes at the regional scale. The spatial pattern of land use, also named landscape structure, reflects underlying human processes as well as influences environmental parameters, such as species abundance, biodiversity, fire risk. The first step to understand highly human-altered landscapes and their functioning is the description of landscape structure. This can be done by means of quantitative indices. This allows to compare landscape structure objectively. It seems that landscape structure, as measured by these common landscape indices, is not significantly different in the respective geographical regions. The search for quantitative methods to analyze and describe the structure of landscapes has become a high priority in landscape ecology. This study assesses the effectiveness of commonly used measures to discern spatial forest cover patterns. INTRODUCTION Present environmental management considers forests as one of the elements of a landscape (Riitters et al., 1995; Forman and Godron, 1986). Spatial forest cover patterns are often the result of human processes, but they can also be used to explain or predict species abundance, biodiversity, fire risk, etc. (Forman and Godron, 1986; Riitters et al., 1995; Farina, 2000; Cumming and Vernier, 2002; Bogaert et al., 2000). This requires a methodology that describes forest cover patterns quantitatively (Hulshoff, 1995; Batistella and Robeson, 2003). A comprehensive list of landscape indices has been developped, a lots of these have been subject to severe criticism and thus new indices are introduced on a regular basis (Frohn, 1998; Luck and Wu, 2002). The relative spatial arrangement of forest patches is known as landscape structure. A set of landscape indices has been identified (these are quantitative variables that measure an aspect of spatial pattern), suitable for describing the forest cover pattern in Flanders. The ability of these indices to discriminate among landscape structures can be judged by examining their geographic distribution. One would expect structure indices to be different in forest-poor and forest-rich areas. If many landscapes have similar values, the index has little discriminating power (O’Neill et al., 1988). Different regions should thus have their own spatial signature. These spatial signatures are usually seen by the human eye. Comparison of these two methods reveils whether landscape indices can discern the same spatial patterns.
573
E.M. De Clercq and R.R. De Wulf
MATERIALS AND METHODS The study focuses on Flanders, which covers 13 750 km² and has generally low relief. The west is forest-poor, while in the east forest is abundant. The Flemish forest map is divided into equally sized hexagons, as shown in Figure 1. A set of landscape indices were calculated for each hexagon using ESRI’s ArcView 3.1 and Fragstats (McGarigal and Marks, 1994). Hexagons were attributed to different groups, first based on administrative units (called provinces) and afterwards based on landscape zones (Figure 2). These landscape zones were delimited using the digital landscape inventory by OC GIS-Vlaanderen (2001). The classification of these landscape zones was based upon environmental and historical criteria (Antrop, 1997). Hexagons having their center point outside the study area were not included in the analysis.
Figure 1: Forest map of Flanders (Bos & Groen, 2001).
Figure 2: Different subdivisions of Flanders.
These groups were screened for significant differences. In a second phase this research looked for spatially similar landscapes by means of cluster analysis. Statistical analysis was performed using S-Plus 6.1. Since the data does not follow a normal distribution, KruskallWallis non-parametrical multiple comparison test was performed. The used landscape metrics were: • TA: total forested area (in m²) • NP: number of forest patches 574
Landscape indices for comparison of spatial forest patterns in different geographical regions
• AREA.SD: standard deviation of the mean patch size • SHAPE.MN: mean patch shape index
pij min pij pij = perimeter of patch ij (m) min pij = minimum perimeter of patch ij (m) • PARA.MN: mean perimeter-area ratio per patch
pij aij pij = perimeter of patch ij (m) aij = area of patch ij (m²) RESULTS AND DISCUSSION Visual interpretation Most provinces seem to have different spatial forest cover patterns. Province W is extremely forest poor and where forest does occur, it is clumped together. Province O has a slightly higher forest cover and this seems to be randomly distributed. Both small and large patches occur. This pattern is similar to that of province V, were large forested complexes are found in the south. Provinces A and L have the highest forest cover and forest fragments are co-agulating. The landscape zones present a similar picture. Zone I is easily recognised since forest is nearly absent here and zone III is densily forested. Zone II and IV have an intermediate forest proportion. In zone IV the forest patches are slightly more randomly distributed than in zone II. Landscape indices The results of the non-parametrical multiple comparison test are displayed in Table 1. In the second column results for different provinces are shown, in the third column landscape zones are displayed. Symbols that are underlined by the same line do not differ significantly from each other. Table 1: Groups for various landscape indices. Provinces TA W O V A NP W O V L AREA.SD O V W A SHAPE.MN O A L V PARA.MN W L A V
L A L W O
Landscape zones I II IV I II IV II I IV II III IV IV I II
III III III I III
When looking at provinces, TA seems to have most discriminating power. Visually, province O and V could not easily be identified. Each province has a different amount of forest, only province A and L cannot be distinguished. All landscape zones have different TA’s. The number of patches is not significantly different for different provinces, but it is for the landscape zones. The standard deviation of mean patch size shows three groups in
575
E.M. De Clercq and R.R. De Wulf
the provinces, as well as for the landscape zones. The mean shape index has little discriminating power in both cases, as well as the perimeter-area ratio. These results suggest that the spatial forest cover patterns are not significantly different for different provinces. A characterization of landscape structure by means of landscape zones seems to be more persistent. Cluster analysis The clustering operation indicates that 4 groups of spatial structure can be distinguished. The groups are spatially packed (Figure 3), but they do not coincide with a known subdivision. It is quite stricking that this spatial pattern is not revealed by simple visual interpretation. Table 2 shows the mean landscape indices for the different clusters.
Figure 3: Spatial distribution of the clusters. Table 2: Means for the different groups as found with the clustering algorithm. Cluster TA NP AREA.SD 1 615 108 17 2 1953 281 32 3 398 128 8 4 3599 212 101
SHAPE.MN 1.69 1.72 1.76 1.79
PARA.MN 558 632 702 658
The total forested area and the number of patches seem to present the largest differences between clusters. The search for quantitative methods to analyze and compare the structure of landscapes has become a high priority in the field of landscape ecology. It remains a challenging task to select the indices most suitable for the pattern under study. Also regarding the interpretation of patterns in relation with management objectives, a lot of research is needed. REFERENCES Antrop, M., 1997: The concept of traditional landscapes as a base for landscape evaluation and planning. The example of Flanders Region. Landscape and Urban Planning 38, 105-117.
576
Landscape indices for comparison of spatial forest patterns in different geographical regions
Batistella, M. and Robeson, S., 2003: Settlement design, forest fragmentation and landscape change in Rondônia, Amazônia. Photogrammetric Engineering & Remote Sensing 69(7), 805-812. Bogaert, J., Van Hecke, P., Salvador-Van Eysenrode, D. and Impens, I., 2000: Landscape fragmentation assessment using a single measure. Wildlife Society Bulletin 28(4), 875881. Bos & Groen, 2001: Digital version of the forest map. MVG, LIN, AMINAL, department Bos & Groen, version 2001 (OC GIS-Vlaanderen). Cumming, S. and Vernier, P., 2002: Statistical models of landscape pattern metrics, with applications to regional scale dynamic forest simulations. Landscape Ecology 17, 433444. Farina, A., 2000: Principles and methods in landscape ecology. Kluwer Academic Publishers, The Netherlands. Forman, R. and Godron, M. 1986: Landscape Ecology. John Wiley, New York. p 83 – 222. Frohn, R.C. , 1998: Remote sensing for landscape ecology: New metrics for monitoring, Modeling and Assessment of Ecosystems. Lewis Publishers, Boca Raton, Florida, USA. Hulshoff, R.M., 1995: Landscape indices describing a Dutch landscape. Landscape Ecology 10(2), 101-111. Luck, M. and Wu J., 2002: A gradient analysis of urban landscape pattern: a case study from the Phoenix metropolitan region, Arizona, USA. Landscape Ecology 17, 327-339. McGarigal, K. and Marks B.J.., 1994: Fragstats: Spatial pattern analysis program for quantifying landscape structure. Forest Science Department, Oregon State University, USA. OC GIS-Vlaanderen, 2001: Digital version of the landscape inventory, MVG-LINAROHM-Monuments & Landscaps, OC GIS-Vlaanderen. O’Neill, R.V., Krummel, J.R., Gardner, R.H., Sugihara, G., Jackson, B., DeAngelis, D.L., Milne, B.T., Turner, M.G., Zygmunt, B., Christensen, S.W., Dale, V.H. and Graham, R.L., 1988: Indices of Landscape pattern. Landscape Ecology 1(2), 153-162. Riitters, K.H., O’Neill, R.V., Hunsaker, C.T., Wickham, J.D., Yankee, D.H., Timmins, S.P., Jones, K.B. and Jackson B.L., 1995: A factor analysis of landscape pattern and structure metrics. Landscape Ecology 10(1), 23-39.
577
