Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Doctorat → États-Unis → 1973 → PHYTOSOCIOLOGY STUDIES OF A DESERT GRASSLAND SITE

University of Arizona (1973)


Fish, Ernest Bertly


Auteur : Fish, Ernest Bertly

Université de soutenance : University of Arizona

Grade : Doctor of Philosophy (PhD) 1973

Various methods for summarizing and ordering phytosociological data have been developed and utilized on a wide range of vegetation types during the past 25 years. The common objective shared by all methods is to produce meaningful groupings of stands of vegetation which approach a natural classification scheme by creating groups which are internally homogeneous with respect to a wide variety of variables. Supposedly, therefore, the different methods are only unique approaches to obtain essentially identical end results. Only in rare instances have two or more methods been concurrently applied to the same commun ity or communities to compare the results of the methods for consist ency and/or efficiency. The objective of this study was to compare the results from several quantitative descriptive methods involving association analy ses, linear ordination and factor/cluster analyses. The stands utilized in the study were located in a relatively uniform desert grassland com munity in southeastern Arizona. The groupings of stands derived from the application of the several methods to a common group of stands were evaluated for the "best" method. The "best" method was defined as the method which resulted in significant (p < .01) F statistics for the greatest number of ecologically important variables from one-way anal yses of variance for among group and within group mean squares. The results indicate that the methods used were not simply dif ferent approaches to obtaining the same final objective of producing groups with the greatest internal homogeneity for a maximum number of variables. The groups of stands produced by any particular method were highly dependent upon the exact criteria utilized to perform the group ings. Furthermore, the exact type and amount of data which were input to the method affected the final results. In general, the use of a greater variety of types of vegetative data and the inclusion of en vironmental variables as used in the factor/cluster analysis method re sulted in groups that more nearly fulfilled the objectives of producing groups of maximum internal homogeneity for a wide variety of variables. The primary objection to the use of the factor/cluster analysis method concerns the time and effort required to collect the input data. This objection can be partly overcome by the judicious use of more rapid qualitative methods to initially stratify the study area permit ting optimum allocation of samples within the delineated areas. The stratification should result in fewer total stands being required, thus making the factor/cluster analysis method more competitive in terms of time and cost relations.


Version intégrale (1,53 Mb)

Page publiée le 7 mars 2018