| |
Landscape Characterization / Forest Fragmentation / Quantifying forest fragmentation with indices / A simple study from Baltimore County
Quantifying forest fragmentation with indices
A simple study from Baltimore County
As an example of how landscape indices can be used to quantify landscape structure
this section presents an analysis of changes in forest fragmentation along an
urban-rural gradient running from the water front area in Baltimore City out through
the rural areas of Carroll County (Figure 7). Nine hexagons of
5000 ha in size were constructed and aligned along the transect. The following
indices were calculated for each landscape: total area of forest, total number
of patches, average patch size, patch size standard deviation, total length of
edge, average shape index, average patch fractal dimension, average nearest neighbor
distance and average proximity index.
Because indices often provide redundant information, a correlation analysis
of the indices was preformed to determine a subset of indices that could be
used to describe forest fragmentation across the urban-rural gradient. Two set
of indices were highly correlated; one related to composition and configuration,
which include total area of forest, average patch size, patch size standard
deviation, total length of edge, average nearest neighbor distance and average
proximity index, and one related to patch shape, which included total number
of patches, average shape index and average patch fractal dimension (Table
1). Two indices, total forest area and total number of patches, were chosen
from each group to represent forest fragmentation.
Figure 6. Two landscape with identical
composition (i.e. areas) of forest, but with very different configurations.
Note that the average distance between patches is approximately three times
greater in landscape 2 than in landscape 1.
Figure 7. A sample study from Baltimore County.
Table 1. Pearson correlation matrix showing the relationships among nine
landscape structure indices measures for the nine 5000 ha landscape across
an urban-rural gradient. High correlation values (> 0.60) are highlighted
to show the two major groups of indices with close relationships. Abbreviations
are: CA = total area of forest, NUMP = total number of patches, MPS =
average patch size, PSSD = patch size standard deviation, TE = total length
of edge, MSI = average shape index, MPFD = average patch fractal dimension,
MNN = average nearest neighbor distance and MPI = average proximity index
|
|
CA
|
MPS
|
PSSD
|
TE
|
MNN
|
MPI
|
NUMP
|
MSI
|
MPFD
|
|
CA
|
1.00
|
|
|
|
|
|
|
|
|
|
MPS
|
0.83
|
1.00
|
|
|
|
|
|
|
|
|
PSSD
|
0.91
|
0.83
|
1.00
|
|
|
|
|
|
|
|
TE
|
0.93
|
0.73
|
0.77
|
1.00
|
|
|
|
|
|
|
MNN
|
-0.96
|
-0.76
|
-0.86
|
-0.96
|
1.00
|
|
|
|
|
|
MPI
|
0.88
|
0.82
|
0.98
|
0.79
|
-0.87
|
1.00
|
|
|
|
|
NUMP
|
0.06
|
-0.42
|
-0.08
|
0.24
|
-0.25
|
-0.04
|
1.00
|
|
|
|
MSI
|
0.28
|
0.66
|
0.21
|
0.32
|
-0.20
|
0.27
|
-0.69
|
1.00
|
|
|
MPFD
|
-0.14
|
0.18
|
-0.28
|
-0.07
|
0.27
|
-0.27
|
-0.66
|
0.80
|
1.00
|
|
