Ordination, direct and indirect gradient analysis, and cluster analysis were 

 considered. 



Ordination is not well suited to data containing many stands with no 

 species in common. Gauch and Whittaker (1972) compared several ordination 

 techniques and the effect of beta diversity (between habitat diversity), and 

 found that none of the techniques examined worked wery well beyond about five 

 half changes in beta diversity. The data from the reconnaissance exhibited 

 over twelve full changes in beta diversity. Ordination endpoint selection 

 would also be difficult for this data. 



Finally, it is not at all clear that the axes of such an ordination could 

 be defined, e.g. several axes representing various soil properties might be 

 required. Although, an ordination of the plots of a type, or several closely 

 related types, might be enlightening, time did not allow this in this study. 



Direct gradient analysis is best suited to areas where the primary grad- 

 ients can be rather easily recognized and measured, as they can be in mount- 

 ains. For the most part, light, temperature, and moisture gradients are not 

 readily discerned in the study area, while other gradients such as soil salts 

 are important only occasionally. 



Cluster analysis, having been devised to answer questions of speciation, 

 was chosen as the appropriate multivariate technique due to its suitability 

 for classification and its ability to handle diversity and large amounts of 

 data. 



Bonham (1974) proposed a classification of grassland vegetation with an 

 aggloinerative cluster analysis using changes in diversity rather than the 

 usual similarity indices. Lambert and Dale (1964), Williams and Dale (1965), 

 and Williams, Lambert and Lange (1966) have analyzed numerical classifica- 

 tions emphasizing cluster anslysis. Everitt (1974) has reviewed the subject 

 of cluster analysis and its application to a wide variety of subject matter. 



SIMILARITY INDICES AND WEIGHTING: FLORISTICS AND SPECIATION 



Similarity indices use species as the basic unit for comparing samples. 

 The idea behind this is that stands having similar species composition can 

 be expected to have similar ecologic sums of habitat factors as perceived by 

 plants. 



As Dubenmire (1968) points out, this notion is dependent on the species 

 being an ecologic unit. He notes that even in adjacent habitats a species 

 may be represented by ecotypes having mutually exclusive ecologic amplitudes. 

 To the degree to which ecotypes exist, the species is not suitable for deter- 

 mining the similarity of stands. 



The use of species as units in similarity indices introduces another pro- 

 blem parallel to that of continuum vs. typo controversy which is perhaps un- 

 recognized by continuum advocates. Just as types or classes are not equally 

 different, so it is with species. For example, Agropyron smithii is consid- 

 ered to be different from Agropyron dasystacly um, and also from Cornus stolon - 

 ifera. The assumption implicit in an equation such as Sorensen's is that 



111 



