04 - Studia Geobotanica - An International Journal - Vol. 4 (1984)

In the course of numerical ordinations of Apennines woodlands, the problem arose of different performance of ordination methods, using different character sets and methods. To reveal underlying trends and relationships, 47 ordinations are jointly examinated, corresponding to combinations of 6 methods and 8 character types. The Spearman formula has been used to measure the similarity between different ordinations. The ordinations based on families are, on the average, the most similar to the other ones irrespective to the method used. The ordination based on detailed life forms (growth forms) are the least similar to the other ordinations. Ordinations based on the same character set and different methods could be different as well as ordinations based on different character sets and the same method. The suggested procedure can be the basis to select among different ordinations those explaining complementary informations.

Data source: Centrai Databank of the ltalian flora and vegetation. Database: phanerogamic flora of Sicily; percents of species with similar distribution patterns, subdivided into 49 phytogeographical categories, in 23 elevation intervals of 100m each. Methods: Complete Linkage Clustering with Correlation Coefficient far the classification of categories and of elevation intervals; Concentration Analysis far the ordination. Results: species with similar distribution tend to be most frequent along given sections of the elevation gradient. The degree of correlation between phytogeographical categories and elevation has been quantified.

The vegetation of an alluvial forest of the Po plain, known as the Forest of S. Agostino, has been surveyed using the Braun-Blanquet method. Analysis of the data revealed two main ecological situations characterized chiefly by differences in pedological factors identified by the cover values of Carex pendula and Brachypodium syluaticum.

The efficiency of Intersection Analysis in producing species groups at different noise and sampling intensity levels has been tested on the basis of a simulated coenocline. The results suggest that lntersection Analysis is a robust method for detecting species groups and that it could be used in field surveys to find out the sampling intensity sufficient to describe the vegetation under study.

Numerical classifications of peatland communities from the Dolomites are compared by means of information functions. Classifications were evaluated considering both the structure of contingency tables (species groups x relevé groups) and the correlation between vegetation types and peat chemistry. A remarkable result is the coincidence of the results obtained hy internal and external criteria.

Six community types of epiphytic lichen vegetation are reported from the Calamone Lake area (N-Apennines). Data analysis is based on multivariate methods. The ecological interpretation of the compositional variation was based on indirect gradient analysis. Each of the communities is well characterized in floristical. ecologica! and phytogeographical terms.

A discussion on the complementarity of classification and ordination methods in vegetation ecology is given. The need to work with different vegetation spaces and hierarchical processes is stressed.

A coenocline simulator is described, which allows one to calibrate programs or chains of programs before using them on survey data. An example is given, and the listings of two versions of the simulator, in BASIC and FORTRAN respectively.

Two examples of application of numerical classification and ordination methods to textural and chorological data of vegetation types defined with the Braun-Blanquet approach are presented.