A Test for Estimating the Relative Informativeness of Vegetation Gradient

Abstract

An estimate of informativeness is proposed which measures the degree to which a species predominates in a limited portion of an ordination and the consistency with which it decreases from the area of predominance. The estimate comprises order, or the difference between the maximum quantitative value and the minimum value or values to either side, disorder, or the amount by which the gradient must be changed to show a consistent trend, and information, or order minus disorder. This test was applied to the results of a transect survey of an open savanna in south-eastern Wisconsin(U.S.A.). Eight gradients were constructed: two vegetatlonal, two locatlonal, three environmental and one random. The test showed the vegetatlonal ordinations to be more informative than the locational and environmental gradients which were higher than the random gradient. The two locatlonal gradients had a higher mean informative-ness than the three environmental factors. As a technique of quantitative classification, the vegetational ordinations gave the most complex vegetation pattern. A correlation study indicated the possible existence of two factoral regimes each of which contained three inter-correlated gradients. The first included the two vegetatlonal gradients and an elevation regime and reflected factors of soil moisture and drainage. The second included gradients of distance from tree, light intensity, and soil water retaining capacity and reflected the effects of shading.