A summary of the relative (%) distribution of the total area of this landcover type among the four classes of clearing disturbance is reported in the following table.
Table 13: A summary of the areas (km2) in each of the three key disturbance classes for each of the landcover types of the ILZ. The index Loss was calculated as the percentage of the total area of each landcover type that was thinned and cleared. The disturbance score was assigned thus: no clearing (-); <10% (*); 11-30% (**); >31% (***).
Code Uncleared Thinned Cleared Loss (%) Clearing Fragmentation ______________________________________________________________________________ xTML4 32517 7585 3929 26 ** * eTML3 271355 75008 88498 37 *** ** wTML3 21269 39595 47097 80 *** *** xTML3 5641 3284 10429 71 *** *** eM2 206336 174296 331281 65 *** ** wM2 200 145 676 80 *** - xM2 6056 3445 4470 57 *** ** eL2 29405 16232 46344 40 *** *** wL2 20126 31524 96081 86 *** *** xL2 107518 2321 9164 10 * * eM1 13441 74262 83840 92 *** *** eL1 30646 16519 32633 33 *** *** wL1 40962 8 1491 4 * * xL1 41454 2834 24005 39 *** * eS3 205 300 19388 99 *** *** wS3 12944 6381 15188 62 *** * xS3 8513 597 10310 56 *** *** eS2 41926 45516 151005 82 *** *** wS2 25798 593 174 3 * * xS2 17338 8247 19130 61 *** *** eS1 4123 3422 21898 86 *** *** wS1 21279 0 0 0 - - xS1 5632 0 0 0 - - xZ3 3135 1321 3644 61 *** *** xZ2 80747 0 0 0 - -
Only the three critical classes, uncleared, thinned and cleared are displayed. Where all one landcover was classified as indeterminate, such as for the grasslands (xG), its disturbance rating was determined by visual interpretation and reported in Disturbance by Grazing.
For each type, the simplest index of Loss was calculated as the aggregate area that was either thinned or cleared. This index was arbitrarily rated and scored thus: < 10% was low (* ); 10-30% was moderate (** ); > 30% was substantial ( ***).
A second index of habitat loss was derived from a measure of the fragmentation of the remaining uncleared landcover patches. There are many published indices of landscape pattern and fragmentation; eg Turner (1990). However, since robust relationships between the rate of biotic erosion and landcover type and degree of fragmentation are lacking, there was little point in computing complicated measures of landscape pattern; eg Cale and Hobbs (1994), Hobbs (1994), Murcia (1995).
The calculation of the index of fragmentation comprised both objective and subjective parts. The objective basis was, for each landcover type, a graphical comparison of the cumulative distribution of patch size between that mapped in Natural Vegetation, by definition uncleared, with its contemporary fragmented condition measured by satellite. This specimen fragmentation chart illustrates the principles involved.
A comparison of the cumulative distribution of proportional uncleared patch size for the baseline (undisturbed) and contemporary (fragmented) states. The degree of fragmentation was rated and scored by the comparison of three curve attributes: ordinate, slope and asymptote. This example is for the closed forests.
For this landcover type, there are 40 separate occurrences within the ILZ. These patches are sorted by size, converted to a proportion of the total area and plotted as a cumulative baseline curve. The largest patch size alone constituted approximately 25% of the total area (the ordinate); the next largest patch adds approximately 10% and so forth until the asymptote of 100% is reached with the last and smallest patch.
The ordinate (proportion occupied by the largest patch) and the shape of the curve as it approaches the asymptote express the size ranks within the ILZ. In this example there are many small patches; the last twenty patches accounting for just the remaining 20% of the total area.
The process was repeated for the contemporary uncleared patches. With clearing, there will be more patches of smaller size than the baseline set. Sorting these by size and plotting only the first 40 patches, the same as the undisturbed yields a very different shaped curve.
Three comparisons of the two curves indicate the degree of loss and fragmentation: the ordinate (comparative size of the largest uncleared patch); the shape of the curve; and the difference in asymptotes, in this example 70% compared with 100%.
This simple objective comparison was quite insightful. A highly cleared and fragmented landcover type will show large ordinate differences; a flat curve generated by many small patches; and a large disparity between the endpoints of the two curves. Conversely, a little fragmented landcover type would return a curve similar to the baseline. The first three case studies in the Case Studies are good illustrations of the nature of fragmentation. The graphical comparison was subjectively rated and scored as either low ( *), moderate (** ) or substantial (*** ).
