With the intensification of agricultural practices, a lot of species originally widespread in the semi-natural landscape are now confined to small landscape elements. The first objective of this research is to find out under which circumstances the small landscape elements can function as a refuge for vulnerable plant species. The distribution ecology is considered as the basis for more detailed populational research. An investigation of the correlation between populational traits, habitat characteristics and isolation measures will be carried out. Secondly, the use of spatially-explicit models in this context will be investigated. The third objective is the elaboration of measures for management and land-use of the habitats and the surroundings through integration of the data in a GIS environment.

Problem definition and objectives:
With the intensification of agricultural practices, a lot of species originally widespread in the semi-natural landscape are now confined to small landscape elements. Little is known however about the dynamics and viability of populations under these circumstances.
The first objective is to find out under which circumstances the small landscape elements can function as a refuge for vulnerable plant species. An investigation of the correlation between populational traits, habitat characteristics and isolation measures will be carried out. Secondly, the use of matrix projection models and point pattern analysis for modelling small-scale spatio-temporal changes will be investigated. The third objective is the elaboration of measures for management and land-use of the habitats and their surroundings.

Material and methods:
The research will be conducted using a set of species of small-scale, semi-natural habitats embedded in an agricultural landscape. Five species (Primula vulgaris, P. veris, P. elatior, Arum maculatum and Geum urbanum) are suggested, based on a series of both practical and theoretical criteria. The different study-areas will be chosen on the basis of the presence of the studied species in in a more or less intact network of small landscape elements. For all the populations in the study-area, locality positions will be measured using a differential GPS, which allows to determine the surface of the populations and in combination with species-counts to calculate several measures for density. The distribution ecology is considered as the basis for more detailed populational research. For the considered populations relevés of the vegetation will be made in a stratified random way, estimating cover (Londo) and height of the vegetation. In order to evaluate the internal habitat characteristics and to couple these to the population and vegetation data, soil samples will be taken and an estimation wil be made of the slope, exposition and light conditions of the sites. For a selection of all the populations in the study-area, position and several characteristics (life-stage, number of flowers, leaf length, seed set) of all the individuals will be recorded. These recordings will be repeated in the following years in order to get an idea about the population growth rate and seedling recruitment.
Statistical data processing will include the application of a series of multivariate techniques. Ordination of the data can be done by correspondence analysis (CANOCO, DECORANA); clustering by TWINSPAN. Digitizing the data of the populations of the different species into a GIS should make it possible to calculate several measures of isolation and connectivity. Some basic maps with land-use characteristics, soils and landscape elements that can be considered as pontential habitats, will also be digitized.
Population data gathered in the earlier part of the research will be used to develop dynamic models. Matrix projection models describe the transition in species numbers per life-stage. The projection matrices offer the possibility of obtaining synthesized quantitative information about populations which makes it possible to compare themunder different circumstances. Several software packages (e.g. POPPROJ, VORTEX) are developed to predict population dynamics under alternative scenarioes based on the calculated transition matrices. To conclude, point pattern analysis will be applied on the fine-scale population maps to detect local differences in growth, fecundity and mortality.