Probabilistic classification offers various advantages in its application to vegetation studies: it can use data in the form of ordered as well as quantitative values; it can use a range of values for each attribute (species) in each relevé or group of relevés; it can use incomplete data sets; it takes account of all species, rather than only characteristic species; and it enables a null hypothesis of random distribution of species among relevés to be tested.
The procedure is here explained in some detail, and its application is illustrated, first with a classical data set from the Alps, and second with an extract from the extensive Netherlands national data base.
It is shown that the presence or absence of species is often more informative about the relationships between relevés than the quantities in which they are present. The results do not support the concept of discrete and uniform vegetation units, but rather of vegetation composition varying around centres of concentration.
Braun-Blanquet, J. 1928. Pflanzensoziologie. Grundzüge der Vegetationskunde. Biologische Studienbücher 7. Walter Schoenichen, Berlin.
Braun-Blanquet, J. 1936. Über die Trockenrasengesellschaften des Festucion valesiacae in den Ostalpen. Comm. Stat. Intern. Geobot. Med. Alp. 49:169–189.
Fisher, R.A. 1934. Statistical Methods for Research Workers. 5th edition. Oliver and Boyd, Edinburgh & London
Goodall, D.W. 1964. A probabilistic similarity index. Nature 203: 1098
Goodall, D.W. 1966. Deviant Index: a new tool for numerical taxonomy. Nature 210: 216.
Goodall, D.W. 1966a. A new similarity index based on probability Biometrics 22: 882–907.
Goodall, D.W. 1968. Affinity between an individual and a cluster in numerical taxonomy. Biom. Prax. 9: 52–55.
Goodall, D.W. 1993. Probabilistic indices for classification – some extensions. Abstracta Botanica 17: 125–132
Goodall, D.W. 1994. The treatment of spatial data in probabilistic classification. Abstracta Botanica 18: 45–47.
Goodall, D.W. and E., Feoli. 1988. Application of probabilistic methods in the analysis of phytosociological data. Coenoses 3: 1–10. Reprinted in 1991 as Chapter 13 in E. Feoli and L. Orlóci (eds.), Computer Assisted Vegetation Analysis. Kluwer, Amsterdam, pp. 137–146.
Kershaw, K.A. 1964. Quantitative and Dynamic Ecology. E. Arnold, London.
Lancaster, H.O. 1949. The combination of probabilities arising from data in discrete distributions. Biometrika 36: 30–382.
Moore, J. J. 1962. The Braun-Blanquet system: a reassessment. J.Ecol.. 50: 761–769.
Moore, P.D. and S. B. Chapman (eds.). 1986 Methods in Plant Ecology. 2nd edition, Blackwell, Oxford.
Mueller-Dombois, D. and H. Ellenberg. 1974. Aims and Methods of Vegetation Ecology. Wiley, New York.
Poore, M.E.D. 1955a. The use of phytosociological methods in ecological investigations. 1. The Braun-Blanquet system. J. Ecol. 43: 226–244.
Poore, M.E.D. 1955b. The use of phytosociological methods in ecological investigations. 2. Practical issues involved in an attempt to apply the Braun-Blanquet system. J. Ecol. 43: 226–244.
I am beholden to the ecologists at Wageningen in The Netherlands, particularly Dr. Stephan Hennekens, for making available to me a sample from their enormous data bank for Netherlands vegetation.
About this article
Cite this article
Goodall, D.W. Probabilistic classification and its application to vegetation science. COMMUNITY ECOLOGY 3, 147–157 (2002). https://doi.org/10.1556/ComEc.3.2002.2.2
- Festucion valesiacae
- Hypothesis testing
- Ordinal data
- Vegetation classification