Applying Objective Data for a Multi Temporal Analysis of Habitat Suitability Indices to Monitor Biodiversity : A Case Study for the Example Key Species Red Kite (Milvus milvus) and Black Stork (Ciconia nigra) : Dissertation, Universität Hamburg, Fakultät für Mathematik, Informatik und Naturwissenschaften
The study describes the potential of various habitat suitability indices (HSI) using remotely sensed data (Landsat 5 and 7) and other mapped information in digital format for the characterization and monitoring of rare species habitats at the landscape level over time. The main focus was to develop a flexible and open system for habitat monitoring which allows a pragmatic overview of habitat development without field assessments, or with very limited field assessments. The potential HSI models for the exemplary key species Red Kite (Milvus milvus) and Black Stork (Ciconia nigra) are analysed with regard to their sensitivity to changing environmental conditions, and also with regard to the influences of individual attributes used as input for the models. The Moritzburg area located close to the city of Dresden, Germany was selected as the study site. It is characterised by a pronounced heterogeneity of landscape elements such as forests, meadows and lakes. The remote sensing data for the year 2000 were combined with ground data collected in the field campaign of the EU research project “MNTFR”. In addition, the database “Datenspeicher Wald” provided forest information for the year 1989 based on the forest inventories at the company level. Attributes, based on Natura 2000, such as food supply or nesting resources, were utilised as input for HSI models. The in situ data were combined with satellite data using a spatial statistic approach called kNN method for extending in situ attributes to the entire area of interest. Habitat suitability maps for both occasions (1989 and 2000) were compared for the individual key species. The methods described underlay the three HSI models tested in this study: a) The Habitat Suitability Index (HSI) with binary attribute maps, b) The Enhanced Habitat Suitability Index (EHSI) applying binary attribute maps enhanced with fuzzy sets and c) The Habitat Suitability Index with Home Rage Aspect (HR-HSI) applying recalculated attribute maps with an activity radius of 200 m for each pixel. Each of these HSI models includes two levels of consideration: the attribute level and the life requisite level. The multiplicative approach with multiplicative combination of life requisites resulted in the original models HSI, the EHSI and HR-HSI and the summation approach with additive recombination of life requisites resulted in the models HSI+, EHSI+ and the HR-HSI+. While all six HSI models are able to detect habitat changes and to predict future habitat development, the EHSI model proved to be efficient to enhance purely binary data into discrete transition probabilities along suitable pixel with a decreasing probability within a distance of 150 m. The HR-HSI model proved to be useful in describing neighbourhood relations of habitat attributes. It offers a graduation of habitat potentials calculating continuous transition probabilities. The HR-HSI model is sensitive for areas of minimum 25 hectares indicating potential habitat loss or gain in the test site. The model approaches can support decision- and policy-making concerning landscape management, as well as enabling simulation of changing individual attributes. The main obstacle to a successive implementation of the HSI models is a comprehensive description of factors driving habitat suitability that have hardly been presented in quantitative terms. Therefore an interdisciplinary knowledge transfer is recommended to realise the implementation of quantitative information of species specific requirements.
