Vascular plants in modern agricultural landscapes of SE Norway : spatial distribution and temporal dynamics

This thesis addresses the spatial distribution and dynamics of vascular plant species in modern agricultural landscapes in SE Norway. This is done by analysing several data sets for species composition and environmental factors; the most important are 2201 patch elements located in 16 1-km2 squares in SE Norway; data from farm ponds; and two data sets from present and former semi-natural grasslands, respectively. Eventually, results are compared to a companion study in a more traditional agricultural landscape of W Norway. Spatial and temporal patterns in species richness and composition are analysed by habitat specificity measures, ordination analyses and generalised linear modelling. A semi-natural affiliation index is used to assess the affiliation of each patch's flora to long-term extensively managed semi-natural grasslands versus intensively used agricultural land and waste ground.<br> This thesis also includes a theoretical evaluation of habitat specificity indices, which forms the basis for the empirical studies. The theoretical study provides arguments for use of gamma diversity contribution and core habitat specificity measures in parallel and the empirical studies show that the context used for quantifying habitat specificity strongly influences the results. High values for both measures were observed for woodland, pastures and road verges, whereas midfield islets, boundaries and ploughed landtypes were in general ranked low, in both modern and traditionally used landscapes. These findings were also supported by the semi-natural affiliation index. Semi-natural grassland species were scatteredly distributed throughout whole landscapes but peaked in woodlands in addition to semi-natural grasslands. Both information on past land-use derived from an old cadastral map in addition to results from the study of modern agricultural landscape patches characterised by the semi-natural species Scorzonera humilis, indicate that many semi-natural affiliated species persist long after management practices has come to an end, and that an extinction debt may be present in the flora of abandoned grassland patches. Floristic gradients of agricultural landscape patch types identified by ordination methods were related to regrowth succession (reflecting long-termmanagement and land-use intensity) and nutrient status. On the other hand, farm pond ecosystems showed few patterns in species compositional change over time, even though a slight increase in the nutrient content of pond waters was observed. The results indicate that plant species composition in ponds will largely be determined by historical, idiosyncratic events in addition to current environmental conditions.<br> To acquire a better basis for assessment of the status for rare, red-listed species, mapping and monitoring is necessary. Predictive habitat distribution models, which relate the geographical distribution of a species to its environment, are shown to be a promising tool in ecology. By using spatial predictions, sampling efficiency of rare species may be increased and finding new occurrences of rare species may be simplified. The applied use of habitat suitability models has been exploredby making fine-scaled spatial predictions of a red-listed species, Scorzonera humilis, in Norway. Use of independent evaluation data adds strongly to credibility of habitat suitability models of rare species.