Biodiversity and Global Change: quantification of ecosystem (vegetation) fragmentation using fractal geometry and neural networks
start on 01/10/1996_____end on 30/09/2000
part of: Spatial Patterning of Terrestrial Ecosystems: Assessment of the Impact of Global Change on Land Cover Distribution Using Satellite Data
start on 01/10/2001_____end on 30/09/2004
Research done at:
Universitaire Instelling Antwerpen (UA-UIA)
= University of Antwerp-UIA campus (UA-UIA)
Department of Biology
Phone: +32-(0)3-820.22.56 ---------- Fax: +32-(0)3-820.22.71
Official language: Dutch
more information on this research:
In little more than a decade, fragmentation, the breaking up of large habitats or land areas into smaller parcels, has become a worldwide environmental issue. Ecologists agree that fragmentation changes the landscape regarding habitat area, interior-to-edge ratios, patch shape, total patch boundary length, isolation, connectivity, and patch number. Despite the increased use of measures of landscape structure, experts have not agreed on how to measure landscape patterns. Most effectsof fragmentation can be measured separately; there is, however, no single measure that captures all aspects of fragmentation. This fact can be considered as astonishing, in view of the abundance of proposed landscape indices and developed landscape analysis software. Two solutions can be proposed to this dilemma. The first is to select the single aspect of fragmentation that is of most concern to the question of interest, the second is to use several measures. Both solutions have their disadvantages: interpretation of the single-factor is denoted as `tricky' and combination of different measures cannot be carried out unambiguously.
The objectives of the research project can be summarized as: (1) to develop indices to describe patch shape, area, number and isolation; (2) to improve existing indices with regard to reference values, dimensionality, performance and ecological significance; (3) to develop a methodology to combine single-patch metrics for landscape fragmentation assessment; and (4) to develop an ecologically relevant comprehensive measure for landscape fragmentation. The research project used the patch-corridor-matrix landscape model to represent the landscape. Analysis and development of indices was consequently executed at two scalar levels: patch and landscape. Development of landscape metrics was based upon a mathematical and theoretical approach, and was tested with neutral models (percolation maps). Examples were elaborated using the Biological value map. Species were not considered in the methodology to enable the widest application of the developed metrics.
The results are all published in the principal scientist's PhD dissertation, and in peer-refereed journalsor books.
Reference: J. Bogaert, Quantifying habitat fragmentation as a spatial process in a patch-corridor-matrix landscape model. PhD dissertation (advisor: Prof. dr. ir. I. Impens), University of Antwerp, Faculty of Sciences, Department of Biology, 2000, 209 pp. [UIA BC-F 7:1167]
Habitat fragmentation is the process of breaking up continuous vegetations into dispersed remnants and thereby generating habitat loss, isolation of the habitats, and edge effects, the latter caused by interaction of the patches with their surroundings.Fragmentation -deservedly- has been denoted as a major threat to biological diversity. From the perspective of terrestrial ecosystems, fragmentation, inducing landcover change, is considered as an important component of global change. If compared with contiguous habitats, scattered fragments show more edges characterised by an altered microclimate and species diversity. This phenomenon, designated as the `edge effect', is observed at the patch boundary, and divides the patch in two zones: the edge and the undisturbed interior habitat. Spatial segregation of habitats influences their population dynamics: species become more prone to extinction and to inbreeding. Moreover, fragment size often doesn't exceed the minimum home range of species.This dissertation contributes to a quantitative approach towards fragmentation as a spatial process. Because of this, an improved assessment of theecological consequences and a comparison of different fragmentation patterns are enabled. Starting from the'patch-corridor-matrix' model, i.e. assuming that a landscape is composed of only patches, corridors and the matrix in which both are embedded, the quantification is firstly executed for the patch itself (e.g. size, perimeter, shape, interior-to-edge ratio). Secondly metrics are calculated at the landscape level (e.g. degree of isolation, number of patches). A key issue of this thesis is the development of reference values for fragment area, shape, number, isolation, perimeter and for the interior-to-edge ratio. Shape is assumed to be a combination of patch compactness and perimeter curvature. The reference values are based upon the comparison with a reference area and shape. For raster data, the reference values are also determined by the pixel geometry and by the resolution used (scale effect). Two methods are presented to quantify fragmentation at the landscape level. In the first method, different fragmentation descriptors are calculated. The vector length in a multi-dimensional space is used to represent the overall fragmentation status. The length is calculated as the Euclidean distance using the index value of each descriptor. The contribution of each fragmentation descriptor is then found by projecting the vector on the respective axis. In a second method, designed for raster images, different data layers are generated, each quantifyinga different aspect of fragmentation. A `moving window technique' is thereby applied. Regions, characterised by specific properties of fragmentation, can be identified using overlay operations. Spatial gradients of fragmentation can be detectedin this way.
Publications related to this research:
Academic Bibliography of the University of Antwerp