A Method for Mapping Vegetation Utilizing Multivariate Statistical Techniques

Abstract

Principal component analysis and stepwise discriminant analysis were used to map the vegetation of the Pat Welder Ranch on the Texas Coastal Plains into 5 vegetation types based on relative frequency data of the major plant species. These 5 types were shown to be subdivisions of 2 plant communities previously reported for the area by researchers utilizing conventional mapping techniques. In addition to separating the 5 types and classifying the 140 sample points into their respective types, the technique provided a key for field separation of the types based on 3 vegetative variables, and provided indicator species values for rapid field identification of the types. The technique is presented as a means of mapping vegetation with a minimum of human bias, a maximum of repeatability and information content, and maximum applicability.