A regional landscape ecosystem classification of Michigan stressing physiographic, geologic, and soil factors.

Abstract

A regional, hierarchical, landscape ecosystem classification was developed for Michigan based on geology, physiography, and soils to provide an abiotic framework for the study and management of biota. The major objectives of the research were: (1) develop a preliminary classification and map of Michigan based on existing geological, topographic, and soils maps and publications, (2) describe each classification unit using published and field-collected abiotic data, (3) contrast units using statistical analyses of plot sampling data, (4) refine the classification on the basis of field data and statistical analyses, (5) evaluate the relationship of overstory and shrub species to classification units, and (6) determine the effectiveness of classification methods and suggest modifications for future research. Five regions were delineated on differences in bedrock geology and glacial landform, three within Lower Michigan and two within Upper Michigan. Regions were subdivided into districts and subdistricts on the basis of soils and site factors. Statistical comparisons of regions, districts, and subdistricts were conducted using soils and overstory vegetation data collected within 503 sampling plots from throughout the state. All regions and 97% of neighboring districts and subdistricts could be statistically differentiated with two pairwise univariate nonparametric statistical methods, the median and rank-sum tests; multivariate approaches, including canonical variates analysis and the Mahalanobis D$\sp2$ statistic, were less effective. Although point sampling data demonstrated differences between classification units, landscape patterning was more effectively evaluated through analysis of several-square-mile areas as mapped in published soil surveys. Clusters of co-occurring overstory tree species showed significantly non-uniform distribution patterns relative to classification units. At the regional level, spatial patterning of species primarily reflected macro-climatic differences; at the intra-regional or district level, drainage class differences were the primary factors affecting species' distributions. The study concluded: (1) existing geologic, physiographic, and soils publications were sufficient to develop a regional landscape ecosystem classification of Michigan, (2) abiotic and biotic plot data statistically differentiated most classification units, but (3) recurring landscape patterns of soil texture, drainage, and slope would be more effective than plot data for characterizing districts and subdistricts.