Premium
Modelling vascular plant diversity at the landscape scale using systematic samples
Author(s) -
Wohlgemuth Thomas,
Nobis Michael P.,
Kienast Felix,
Plattner Matthias
Publication year - 2008
Publication title -
journal of biogeography
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.7
H-Index - 158
eISSN - 1365-2699
pISSN - 0305-0270
DOI - 10.1111/j.1365-2699.2008.01884.x
Subject(s) - species richness , akaike information criterion , transect , deviance information criterion , quadrat , geography , generalized linear model , ecology , vascular plant , deviance (statistics) , statistics , collinearity , environmental science , physical geography , mathematics , biology , bayesian probability , bayesian inference
Aim We predict fine‐scale species richness patterns at large spatial extents by linking a systematic sample of vascular plants with a multitude of independent environmental descriptors. Location Switzerland, covering 41,244 km 2 in central Europe. Methods Vascular plant species data were collected along transects of 2500‐m length within 1‐km 2 quadrats on a systematic national grid ( n = 354), using a standardized assessment method. Generalized linear models (GLM) were used to correlate species richness of vascular plants per transect (SR t ) with three sets of variables: topography, environment and land cover. Regression models were constructed by the following process: reduction of collinearity among variables, model selection based on Akaike’s information criterion (AIC), and the percentage of deviance explained ( D 2 ). A synthetic model was then built using the best variables from all three sets of variables. Finally, the best models were used in a predictive mode to generate maps of species richness (SR t ) at the landscape scale using the moving window approach based on 1‐km 2 moving windows with a resolution of 1 ha. Results The best explanatory model consisted of seven variables including 14 linear and quadratic parameters, and explained 74% of the deviance ( D 2 = 0.742). Used in a predictive mode, the model generated maps with distinctive horizontal belts of highest species richness at intermediate altitudes along valley slopes. Belts of higher richness were also present along rivers and around large forest patches and larger villages, as well as on mountains. Main conclusions The approach involved using consistent samples of species linked to information on the environment at a fine scale enabled landscapes to be compared in terms of predicted species richness. The results can therefore be applied to support the development of national nature conservation strategies. At the landscape scale, belts of high species richness correspond to steep environmental gradients and associated increases in local habitat diversity. In the mountains, the belts of increased species richness are at intermediate altitudes. These general belt‐like patterns at mid‐elevation are found in all model parameterizations. Other patterns, such as belts along rivers, are visible in specific parameterizations only. Thus we recommend using several sets of parameters in such modelling studies in order to capture the underlying spatial complexity of biodiversity.