Bez nadpisu

1
General strategy for extracting vegetation
classification from large phytosociological
databases
Milan Chytrý Dept. of Botany Masaryk
University Brno, Czech Republic
2
Step 1 Establishment of the database

• Example Czech National Phytosociological
  Database
• Started in 1996
• Current state

• 55,000 phytosociological relevés
• Sampled in 19222002
• Made by 332 authors
• 1.3 Million individual plant records

3
Step 2 Relevé selection

• Deletion of extreme plot sizes

4
Step 3 Geographical stratification
(Chytry Tichy 2003, Folia Fac. Sci. Univ.
Masar. Brun. 108, in press Kuzelova Tichy,
talk at this Symposium)
5
Step 3 Geographical stratification
(Chytry Tichy 2003, Folia Fac. Sci. Univ.
Masar. Brun. 108, in press Kuzelova Tichy,
talk at this Symposium)
6
Step 4 Identification of major gradients and
groups in the data set
7
Step 4 Identification of gradients and groups in
the data set
(Bruelheide Chytry 2000, J. Veg. Sci. 11
295306)
8
An alternative approach?

• Delimitation of vegetation units by formal
  definitions

(Bruelheide Chytry 2000, J. Veg. Sci. 11
295306)
9
Step 5 Evaluation of expert-based
phytosociological classification

• Calculation of diagnostic capacity of species
  for traditional phytosociological units

(Chytry et al. 2002, J. Veg. Sci. 13 7990)
10
Step 5 Evaluation of expert-based
phytosociological classification

• Calculation of diagnostic capacity of species
  for traditional phytosociological units

(Chytry et al. 2002, J. Veg. Sci. 13 7990)
11
Step 6 Reproduction of traditional syntaxa by
formal definitions

• Only well-defined syntaxa are reproduced
• Cocktail method, applied to a large database
  (Bruelheide 2000, J. Veg. Sci. 11 167178)
• Species co-occurring together are combined into
  sociological groups
• Sociological species groups are combined by
  logical operators to form definitions of
  vegetation units
• Example of association definition (Caltha
  palustris Group AND Cirsium rivulare Group) AND
  NOT (Carex echinata Group)
• Example with coverFilipendula ulmaria cover gt
  25 AND Chaerophyllum hirsutum Group

12
Step 6 Reproduction of traditional syntaxa by
formal definitions
13
Step 7 Fixing overlaps and unassigned relevés by
similarity criterion
(Koci et al. 2003, J. Veg. Sci. 14, in
press Tichy, poster at this Symposium)
14
Step 8 Parametrization of formally defined
vegetation units

• Diagnostic species statistical comparisons of
  species occurrences in the relevés of the
  vegetation unit and in the rest of the database
• Constant and dominant species

• Means and variances of measured vegetation and
  environmental variables

15
Step 8 Parametrization of formally defined
vegetation units

• Diagnostic species statistical comparisons of
  species occurrences in the relevés of the
  vegetation unit and in the rest of the database
• Constant and dominant species

• Means and variances of measured vegetation and
  environmental variables

• Ellenberg indicator values

16
Step 8 Parametrization of formally defined
vegetation units

• Diagnostic species statistical comparisons of
  species occurrences in the relevés of the
  vegetation unit and in the rest of the database
• Constant and dominant species

• Means and variances of measured vegetation and
  environmental variables

• Ellenberg indicator values

• GIS overlays

17
Step 9 Predictive distribution modeling

• Coincidence maps of diagnostic species
• GIS-based models