these species, and in fact, all species, are equally different and discrete, 

 but this runs counter to the tenets of numerical taxonomy (Sokal and Sneath, 

 1963).* Thus, the continuum advocates find themselves in the inconsistent 

 position of criticizing the classification of vegetation and offering an al- 

 ternative which is premised on a classification of flora which suffers some of 

 the same criticisms. 



One solution to this problem would weight species for similarity calcu- 

 lations. However, this would require an exhaustive knowledge of species aut- 

 ecologies and some sort of weighting scheme that address the pertinent fac- 

 tors. The alternative used for this study is to look for anomalies resulting 

 from the multivariate analysis and try to resolve them on a case by case basis. 

 Consideration of synusiae is an example of this rectification. 



QUANTITATIVE VS. PRESENCE AND ABSENCE FLORISTIC DATA 



Lambert and Dale (1964) and Williams and Lambert (1965) discuss the two- 

 fold problem of quantitative data as opposed to data for recording floristic 

 presence or absence. Coverage, for example, measures the degree to which a 

 species is present, but not the degree to which it is absent. Coverage also 

 reduces both those plants that do not happen to be in the sample area and 

 those that cannot be in the sample area to a single measure of zero. 



For example, a plot in the Stco/Bogr-Caf i may fail, by chance, to contain 

 any Carex fi lifolia , although it was present in 90% of the samples of this 

 type, whereas Ginkgo biloba is also absent from the plot --though hardly by 

 chance or to the same degree. Yet, both species are represented in the data 

 by the identical coverage, namely zero. The aforementioned authors there- 

 fore argue for presence or absence values rather than quantitative presence 

 data. 



This agrument, though valid, imposes severe limitations in practice. 

 First, the floristic description can no longer convey an image of the samples. 

 Second, presence-absence data gives emphasis to ubiquitous species of typically 

 low coverage. For example, a Tragopogon dubius -Sp haeralceacoccinea com- 

 munity type could easily be deduced from data for the study area. Third, 

 absence and presence are not in any case equally informative. The absence of 

 a species from an area tells us nothing, whereas presence at least tells us 

 that the environment there is favorable to that species or ecotype (Dubenmire 

 1963). Finally, the agrument is partly overcome if a sampling is adequate, 

 and if records of the species associated with types resulting from classi- 

 fication are kept. As sample size increases, so does the chance of observing 

 the variety of species which can occupy similar sites. 



A possible solution to these problems has been offered by Pfisher and 

 Arnu, however, (personal communication, manuscript in progress) they suggest 

 the use of "weighted presence", which used D cover class numbers in similarity 

 indices. The T (< 1% coverage) class is assigned a 0.5 value. 



* Hull (1974) and Ruse (1973) discuss the questions of speciation from a 

 philcsophical standpoint. Some of their concerns also apply to phytosociology 

 and classification. 



112 



