Classification and Ordination of Nigerian Savanna Vegetation

A variety of techniques is now available for the classification or ordination of vegetational units. Classification methods range from the subjective approach of the continental plant sociologists (Braun-Blanquet 1951) to the computerized methods of associationanalysis and information-analysis (Williams & Lambert 1959, 1960, 1961; Williams, Lambert & Lance 1966; Lambert & Williams 1966). On the other hand an approximation to ordination can be achieved simply by the use of x2 constellation diagrams, or by the more exact method of component analysis. Much of the published work using association-analysis or component analysis, is related to vegetation which is floristically well known and where the important environment controlling factors are also already known. Interpretation of the statistical results is thus relatively straightforward and the method employed is often the feature of interest rather than the results obtained. Unless these methods can be employed in a situation where there is no prior knowledge of the affinities of various vegetational fragments, it is difficult to assess their relative merits as ecological tools. Data collected in the Northern Guinea savanna of Nigeria presented an ideal situation for such a comparison and a range of these methods has been used in an attempt to delimit the more useful of them. The choice of classification as opposed to ordination depends on the ecological question to be answered. Classification in an ecological context gives arbitrary units which can be named, and which are discontinuous by virtue of the fact that the method of extraction has imposed a discontinuity where none necessarily existed. Thus the BraunBlanquet classificatory approach tends to ignore intermediate stands on the basis of their lack of homogeneity and by the careful selection of the stands in the first instance. In an association-analysis, discontinuity is introduced by assigning plots to a particular group, on the presence (or complete absence) of a species. This is an artificial discontinuity since the degree of abundance of a species is an integral part of the concept of a plant association. Ordination on the other hand produces no defined units but merely relates all plots to a series of axes which reflect the inter-relationships of the plots themselves, and also the environmental factors controlling the plot inter-relationships. The axes in fact allow a more objective assessment of controlling factors to be made. Both approaches have been made to the savanna data. Within classification or ordination it is possible to consider either the classification/ ordination of the sample plots in relation to their species spectrum, or the inverse of this, the classification/ordination of the species themselves in relation to their distribution throughout the sample. Both techniques have been employed. The data were collected from six areas in Northern Nigeria, from Maska in the north of Zaria Province down to Idon in the south (Fig. 1) and near the southern limits of the