Multivariate characterization of Sonoran Desert vegetation in southwest Arizona using US Army field data

Abstract

This paper describes the classification and ordination of Sonoran Desert vegetation using systematically collected data from the US Army Yuma Proving Ground (USYPG). Two classifications were created, one based upon relative plant cover derived from 100 m line transect data and one based upon relative plant density derived from 6 m . × 100 m belt transect data, with the belt transect being a lateral extension of the line transect. Both cluster analysis using Ward’s Method and TWINSPAN were used for classifying the data while Principal Component Analysis, Correspondence Analysis, Detrended Correspondence Analysis, and Non-Metric Multidimensional Scaling were used as ordination methods. Cluster analysis was superior to TWINSPAN in creating logical classifications comparable to published descriptions of vegetation communities found in the Lower Colorado Subdivision of the Sonoran Desert. Together, the ordination methods served to accentuate different aspects of the data including main gradients of species composition, in particular a gradient separating plots with riparian-af-finities from the main data set, a Larrea tridentata-Ambrosia dumosa gradient, and a gradient separating Encelia farinosa from the main data set. The main difference between the relative cover and relative density classifications was that the former under-represented cacti such as Opuntia bigelovii and the latter under-represented such as Parkinsonia microphylla and Olneya tesota. The classification methodology used in this study is useful for evaluating resource sampling strategies on U.S. Army bases in sparsely vegetated areas and the classifications could be used as a baseline for monitoring changes in vegetation communities.