Landscape potential for animal species colonization, dispersal and survival
For more information on the whole project, see http://www.unil.ch/ecospat/landspot
We propose a new approach to test a series of hypotheses related to the spatial distribution of animal species in Switzerland. Based on modeling the distribution of habitat units (in the sense of biota) from GIS analyses and statistical analyses, the proposed approach is original in the sense that it will also allow the modeling of the distribution of a group of species of similar ecological requirements (e.g. guilds). Furthermore, it is flexible and powerful, as it does not require species' absence data, and is fast and can be run easily for many species at once.
Most previous modeling studies took a species-specific approach to habitat, fitting a model between occurrences of a species and a set of environmental explanatory predictors, thus modeling the species' ecological niche. The approach we propose here is more habitat-specific, since the distribution of pre-defined habitat units is modeled first, and the simulated habitat map is then used to predict animal distribution.
The development of the simulated map of habitats will be based on a set of available data, among which the most important are: (1) the GEOSTAT land-use information package developed by the Swiss Federal Office of Statistics (OFS 1999); (2) The Swiss vegetation map of Hegg et al. (1993); (3) The Digital Elevation Model (DEM) and related data; and (4) remotely sensed data: MODIS and LANDSAT satellite scenes. The latter data will be used as a potential substitute for vegetation data, which should allow for the delineation of a new map of habitats, independently of any previous vegetation mapping.
Using this map, we will predict the spatial distribution of animal species (i) from field observations, (ii) from theoretical ecological profiles of species, and (iii) using classical predictive distribution models. We shall evaluate the impact of the use of these several modeling approaches on the study of animal distribution. Model predictions will be evaluated using traditional methods as far as accuracy and error propagation assessment are concerned. As a side application in conservation biology, modeled species' habitat maps will be used to assess habitat connectedness and long-term species survival for those species for which distribution was adequately predicted and which distribution exhibits a critical level of fragmentation.