Developing a classifier for the Habitats Directive grassland types in Denmark using species lists for prediction

Abstract This paper describes the use of supervised methods for the classification of vegetation. The difference between supervised classification and clustering is outlined, with reference to their current use in vegetation science. In the paper we describe the classification of Danish grasslands according to the Habitats Directive of the European Union, and demonstrate how a supervised classification can be used to achieve a standardized and statistical interpretation within a local flora. We thereby offer a statistical solution to the legal problem of protection of certain selected habitat types. The Habitats Directive protects three types of Danish grassland habitats, whereas two remaining types fall outside protection. A classification model is developed, using available Danish grassland data, for the discrimination of these five types based on their species composition. This new Habitats Directive classification is compared to a previously published unsupervised classification of Danish grassland vegetation. An indicator species analysis is used to find significant indicator species for the three protected habitat types in Denmark, and these are compared to the characteristic species mentioned in the interpretation manual of the Habitats Directive. Eventually, we discuss the pros and cons of supervised and unsupervised classification and conclude that supervised methods deserve more attention in vegetation science. Abbreviations: DCA = Detrended Correspondence Analysis; CCA = Canonical Correspondence Analysis; CVA = Canonical Variates Analysis; HD = Habitats Directive; TWINSPAN = Two-way indicator species analysis. Nomenclature: Tutin et al. (1964–1993).

[1]  David M. Cairns,et al.  A comparison of methods for predicting vegetation type , 2001, Plant Ecology.

[2]  M. Hill,et al.  Computerized matching of relevés and association tables, with an application to the British National Vegetation Classification , 1989, Vegetatio.

[3]  P. Keddy,et al.  The individualistic and community-unit concepts as falsifiable hypotheses , 1987, Vegetatio.

[4]  E. Taylor,et al.  Population subdivision in westslope cutthroat trout (Oncorhynchus clarki lewisi) at the northern periphery of its range: evolutionary inferences and conservation implications , 2003, Molecular ecology.

[5]  Bernd Münier,et al.  PREDICTION OF HABITAT QUALITY USING ORDINATION AND NEURAL NETWORKS , 2002 .

[6]  Yudong D. He,et al.  Gene expression profiling predicts clinical outcome of breast cancer , 2002, Nature.

[7]  R. Barandela,et al.  Supervised classification of remotely sensed data with ongoing learning capability , 2002 .

[8]  D. Klein,et al.  Using GIS to predict potential wildlife habitat: A case study of muskoxen in northern Alaska , 2002 .

[9]  M. Barbour,et al.  Typology and ecology of Californian serpentine annual grasslands , 2001 .

[10]  Bernd Münier,et al.  A biotope landscape model for prediction of semi-natural vegetation in Denmark , 2001 .

[11]  V. Smith,et al.  Classification of the terrestrial habitats on Marion Island based on vegetation and soil chemistry , 2001 .

[12]  A. Freiburg.  Fagus sylvatica forest vegetation in Greece: Syntaxonomy and gradient analysis , 2001 .

[13]  Antoine Guisan,et al.  Predictive habitat distribution models in ecology , 2000 .

[14]  H. H. Bruun,et al.  Gradient analysis of dry grassland vegetation in Denmark. , 2000 .

[15]  R. Ejrnæs,et al.  Can we trust gradients extracted by Detrended Correspondence Analysis , 2000 .

[16]  H. H. Bruun,et al.  Classification of dry grassland vegetation in Denmark , 2000 .

[17]  Helge Bruelheide,et al.  Towards unification of national vegetation classifications: A comparison of two methods for analysis of large data sets , 2000 .

[18]  Helge Bruelheide,et al.  A new measure of fidelity and its application to defining species groups , 2000 .

[19]  G. Hewitt.  Post-glacial re-colonization of European biota , 1999 .

[20]  J. Vogel,et al.  Where are the glacial refugia in Europe? Evidence from pteridophytes , 1999 .

[21]  P. Racey,et al.  Post-glacial re-colonization of European biota , 1999 .

[22]  Daniel G. Brown,et al.  Supervised classification of types of glaciated landscapes using digital elevation data , 1998 .

[23]  Ladislav Mucina,et al.  Classification of vegetation: past, present and future , 1997 .

[24]  P. Legendre,et al.  SPECIES ASSEMBLAGES AND INDICATOR SPECIES:THE NEED FOR A FLEXIBLE ASYMMETRICAL APPROACH , 1997 .

[25]  B. Ripley,et al.  Modern Applied Statistics with S-Plus. , 1996 .

[26]  R. H. Økland.  Are ordination and constrained ordination alternative or complementary strategies in general ecological studies , 1996 .

[27]  C. Romão,et al.  Interpretation manual of European Union habitats. , 1996 .

[28]  B. Ripley.  Pattern Recognition and Neural Networks , 1996 .

[29]  J. Franklin.  Predictive vegetation mapping: geographic modelling of biospatial patterns in relation to environmental gradients , 1995 .

[30]  M. B. Dale,et al.  Evaluating classification strategies , 1995 .

[31]  J. Michaelsen,et al.  Regression Tree Analysis of satellite and terrain data to guide vegetation sampling and surveys , 1994 .

[32]  B. Mirkin.  Which plant communities do exist , 1994 .

[33]  P. White,et al.  On the existence of ecological communities , 1994 .

[34]  J. Rodwell,et al.  British Plant Communities. Vol. 3. Grasslands and Montane Communities. , 1994 .

[35]  Lee Belbin,et al.  Comparing three classification strategies for use in ecology , 1993 .

[36]  P. Keddy.  Do ecological communities exist? A reply to Bastow Wilson , 1993 .

[37]  H. J. B. Birks,et al.  Assessing the representativeness of nature reserves using multivariate analysis: Vascular plants and breeding birds in deciduous forests, western Norway , 1993 .

[38]  Ruprecht Düll,et al.  Zeigerwerte von Pflanzen in Mitteleuropa , 1992 .

[39]  Keith Bennett,et al.  QUATERNARY REFUGIA OF NORTH EUROPEAN TREES , 1991 .

[40]  R. H. Økland.  Vegetation ecology: theory, methods and applications with reference to Fennoscandia , 1990 .

[41]  M. O. Hill,et al.  TWINSPAN: a FORTRAN program of arranging multivariate data in an ordered two way table by classification of individual and attributes , 1979 .