Using multivariate techniques to quantitatively estimate ecological stages in a mixed grass prairie

Abstract

Cluster analysis followed by stepwise discriminant analysis was used to delineate ecological stages on a mixed grass prairie in western South Dakota. Forty-seven variables were analyzed for 48 sites ranging from potential vegetatlon to early seral stages. A cover-frequency index for western wheatgrass (Agropyron smithii), blue grama (Bouteloua gracilis), and buffalograss (Buchloe dactyloides) was the most valuable in identifying 4 different (P<0.0001) ecological stages. Ecological stage classification was estimated to be 95% accurate. The methods presented are quantitative, precise, easy, time-efficient, and meet the goals of resource managers with a minimum of bias.