USGS - science for a changing world

Northern Prairie Wildlife Research Center

  Home About NPWRC Our Science Staff Employment Contacts Common Questions About the Site

Regional Landscape Ecosystems of Michigan, Minnesota, and Wisconsin


The aim of landscape ecosystem classification and mapping is to distinguish appropriately sized ecosystems—useful and functional land units that differ significantly from one another in abiotic characteristics as well as in their related biotic components. The subdivision of a large area into distinctive landscape ecosystems provides a much-needed framework for integrated resource management and planning; for biological conservation; and for comparsion of differences in composition, occurrence, interactions, and productivity of plants and animals among ecosystems.

This publication provides such a regional landscape ecosystem classification for Michigan, Wisconsin, and Minnesota. Based on differences in climate, bedrock geology, glacial landform, and soils, this classification delineates and describes map units at the Section, Subsection, and Sub-subsection levels that represent areas with distinctive natural conditions affecting species composition and productivity. Macroclimate and physiography were the major components used to distinguish sections and subsections; differences in local physiography and soil were used primarily to delineate sub-sections. Vegetation was used wherever possible to validate climatic and geomorphological boundaries. Further, by drawing on the expertise of numerous members of the scientific and conservation communities, I have incorporated specific information on rare species distributions, adequacy of existing preserves, and management concerns relative to the ecosystem mapping units delineated. The result is a product that expresses the interactive character of landscape ecosystems and their components of climate, geological parent material, physiography (landform and waterform), soil, plants, and animals that will prove useful for resource management, conservation, and study.

This project is a direct outgrowth of the earlier Regional Landscape Ecosystems of Michigan (Albert et al. 1986), and the units described here are the equivalent of the terms Region, District, and Subdistrict previously defined in that work. These two ecosystem classifications share a conceptual and methodological approach that can be characterized as multifactor and multilevel in orientation. A key aspect of the multifactor approach is the delineation of map units based on the evaluation of multiple abiotic factors (including bedrock geology, glacial landform, soils, hydrology, and regional climatic regimes). Because these abiotic factors represent the more constant and enduring features of the natural landscape, the mapping of ecosystems based on abiotic characteristics establishes a relatively stable framework for understanding and managing biological diversity. The three-State map and classification, by identifying significant environmental differences among areas along multiple abiotic dimensions, provides a basis for understanding patterns of species distribution, natural disturbance regimes, and human land use, as well as the natural processes responsible for these patterns.

A second distinguishing feature of the classification approach is the use of a multilevel spatial hierarchy. Following Rowe and Sheard (1981), the landscape is conceived here as a series of ecosystems, large and small, nested within one another in a hierarchy of spatial sizes. A hierarchical classification reflects the degree of relatedness or similarity among adjacent units and allows scientists and managers to move between larger or more local scales by working at higher or lower levels of the classification hierarchy. Because natural processes occur at various levels, from continent-wide to site-specific, ecosystems need to be delineated at a scale (regional or local) that is appropriate to the management intensity.

The three-level classification of Michigan, Wisconsin, and Minnesota is expressly hierarchical in organization. The largest units (sections) delineated here fit within macrolevel units defined by continent-wide classifications; at the other extreme, local delineation of ecosystem types or ecosystem components (such as landform, soil, vegetation, and/or animals) may proceed within the lower hierarchical levels as needed by different users.

Conceptual Approach

Ecosystems are the natural holistic units of the landscape that can be identified and mapped over wide areas or locally (from large regional geographic units such as outwash plains or till plains to small units such as rocky knobs or marsh-filled depressions). Each ecosystem, of whatever size, can be conceptualized as a layered volumetric segment of the biosphere (Rowe 1984b), consisting of an air layer over a landform or water layer, with organisms sandwiched near the sun-energized interface. The largely invisible climatic and hydrologic dimension, imposed on the various Earth surface substrates, produces the land's ecological mosaics (Rowe and Sheard 1981). The perceived abiotic and biotic components, in dynamic interaction with one another, are illustrated in the diagram below (fig. 1).

GIF-Biotic and abiotic components of the landscape ecosystem
Figure 1.—Diagrammatic illustration of the biotic and abiotic components of the landscape ecosystem. The dynamic interactions among components are not shown. (Courtesy of B.V. Barnes.)

Ecosystems clearly have a geographic dimension as well. Rowe (1961) stresses this geographic nature in his definition of an ecosystem as: "...a topographic unit, a volume of land and air plus organic contents extended areally for a certain time." It is this spatial component of ecosystems that provides the hierarchical context for the three-State classification developed here:

	The Ecosphere is the largest system of immediate interest and
	nested within it, composing it, are the macro-level
	ecosystems that we distinguish as seas and continents.  Within
	the continents, regional meso-level ecosystems constitute
	a lower sub-set of systems, while still lower are local micro-
	level ecosystems such as three-dimensional tracts of forest
	land" (Rowe 1992).

Within the macrolevel of the north American Continent, the three-State classification is part of the regional meso-level of landscape ecosystems. Bailey's (1976, 1980; Bailey and Cushwas 1981) categories of Domain, Division, and Province provide the upper hierarchical level, and the sections, subsections, and sub-subsections presented here provide the lower levels for the three-State geographic area. Below the subsection or sub-subsection levels is the mosaic of microlevel ecosystems—the local three-dimensional tracts of land that natural resource professionals seek to manage and conserve.

Factors in the Hierarchy

In the study, the ecosystem components used to distinguish major landscapes are macroclimate, physiography, soil, and vegetation. Climate and physiography strongly influence the regimes of energy and moisture that affect soil development and largely determine the structure and composition of vegetation and the occurrence of animal communities.

In Michigan, long-term climatic records were a primary component in delineating the larger hierarchical units — sections and subsections. Physiography was used in conjunction with climatic data to refine section and subsection boundaries because of its significant influence on the climatic regime. At the sub-subsection level, the primary determinants of boundaries were physiography (because it controls fluxes of radiation and moisture and thereby strongly determines the pattern of soil, microclimate, and vegetation [Rowe 1984b]) and soil conditions. In Minnesota and Wisconsin, climatic data were not analyzed in detail, but were taken from published climatic studies of the two States and the Midwest.

I must stress, however, that the hierarchy developed for this study is a flexible one based on several abiotic factors rather than a rigid hierarchy. In the complex glaciated landscape of Michigan, Minnesota, and Wisconsin, for a given mapping scale, different factors may be important within adjacent map units. For example, at the subsection level of the hierarchy, differences in landform are typically important, but the presence and type of large bedrock exposures may be equally important elsewhere.

Hierarchical Levels: Section, Subsection, Sub-subsection

The hierarchical levels of the section and subsection are recognized by the USDA Forest Service in the National Hierarchical Framework of Ecological Units (ECOMAP 1993), along with broader mapping units (Domain, Division, and Province) and finer mapping units (Landtype Association, Landtype, and Landtype Phase) (table 1). As can be seen from the general size range presented in table 1, the scales of section, subsection, and landtype association are broadly defined, and may overlap. For example, the scale of tens of square miles of the subsections overlaps with the scale of thousands of acres for the landtype association.

Table 1.--National Hierarchy of Ecological Units1

Planning and Analysis Scale
Ecological Units
Purpose, Objectives, and General Use
Broad applicability for modeling and sampling. Strategic planning and assessment. International planning. Millions to tens of thousands of square miles
Subregion Section
Strategic; multi-forest, statewide and multi-agency analysis and assessment. Thousands to tens of square miles.
Landscape Landtype Association Forest or area-wide planning, and water-shed analysis. Thousands to hundreds of acres.
Land Unit Landtype
Landtype Phase
Project and management area planning and analysis Hundreds to less than 10 acres

1From ECOMAP, USDA Forest Service (1993).

In this study, the sub-subsection has been introduced as an additional hierarchical level where it is necessary to divide a subsection, but where treatment as a landtype association (LTA) is not adequate. For example, Subsection VII.2 in Michigan is a high, sandy plateau, which has been subdivided into Sub-subsection VII.2.1, an area of sandy ridges; Sub-subsection VII.2.2, an area of extensive sand outwash plain, and Sub-subsection VII.2.3, another area of steep, sand ridges. The sub-subsection level is needed here to show the relatedness of the three units as subdivisions of the high, sandy plateau (Subsection VII.2) long-recognized as an important physiographic feature in Michigan (Veatch 1953). Treating all three sub-subsections as unrelated subsections would result in the loss of important ecological information. Nor is treatment as LTA's an appropriate alternative in this case for several reasons. First, the scale of the LTA is generally much smaller than the sizes of these units, usually thousands or hundreds of acres compared with the several thousand square miles encompassed by Sub-sections VII.2.1-VII.2.3. Accordingly, the LTA is defined as a unit for "project and management area planning," not "forest or area-wide planning" (table 1) as needed here. Second, the LTA has been used by the Forest Service for mapping individual landforms or groups of repeated landforms, such as drumlins in a drumlin field (but not the adjacent outwash deposits, which would be another LTA), rather than for mapping a mosaic of interdigitated landforms, such as both the drumlins and surrounding outwash deposits of a drumlin field. In contrast, the sub-subsection level allows for the identification of several "mosaics of similar landforms" within a larger subsection.

The amount of detail included in the three-State map varies geographically, depending on available information. In both Michigan and Wisconsin, prior regional mapping was done to the level of the sub-subsection, and all three hierarchical levels (section, subsection, and sub-subsection) were identified on existing maps. As a result, sub-subsections are typically described here for Michigan and Wisconsin. In contrast, prior work in Minnesota was restricted primarily to section and subsection levels. Accordingly, only a few subsections were divided here into sub-subsections, because information was inadequate or because contributors strongly disagreed on the need for further subdivisions or on the locations of proposed boundaries. However, sub-subsections are now being delineated for the entire State.

Further, because the nature of the landscape of a geographic area dictates the size and number of hierarchical levels, subsections do not always have natural sub-subsection divisions. The landscape must speak for itself; a neat, tidy, completely orthogonal classification with exactly the same number of units and levels in each hierarchy is not possible.

Maps Units as Hypotheses for Testing

The regional ecosystems delineated here are hypotheses developed from ecological theory and knowledge of what is ecologically important to the landscape. By delineating climatic and structural differences in the landscapes of Michigan, Minnesota, and Wisconsin, I hope to identify units that are functionally different in important and useful ways. The units are hypotheses for testing: similar local ecosystems within a given subsection or sub-subsection should respond to natural disturbances and to management in similar ways.

Relationship to Other Approaches and Other Regional Ecosystem Maps

The methodology employed in developing the three-State map and classification is modeled after the multifactor, multilevel integrated field approach to understanding ecosystem structure and function pioneered in the southwestern German state of Baden-Württemberg (Schlenker 1964, Spurr and Barnes 1980, and Barnes 1984). Their approach and regional level classification, in turn, stem from the ideas and publications of G.A. Krauss and R. Gradman, who were noted for their early 20th century contributions in landscape ecology and biogeography. The approach of integrating multiple ecosystem components to generate regional and local ecosystem classifications and maps has been the basis of their integrated resource management for more than 50 years. The integrated ecosystem approach used here is also similar in holistic concept to that of the forest ecosystem region (site region) and total site methods pioneered by Hills (1960) in Ontario. Rowe's (1979) ecological land classification for Canada similarly stresses the integrated approach and four hierarchical levels.

Some of the major ecological maps used in developing this map and classification are discussed below. These maps were used as the starting point for this study, but other publications and data were studied and incorporated, resulting in modifications of the original ecological maps.

Regional Landscape Ecosystems of Michigan

This map and classification developed by Albert, Denton, and Barnes (1986) provided the basis for the Michigan part of this study. Modifications of the initial Michigan map were based on further field work (especially in the wetlands of the State), recent geological and climatic publications, and ongoing ecological classification work on Michigan's National Forests. The Michigan Heritage Program data base provided additional information on rare plant and animal distribution as well as interpretation of the original vegetation of the State. The hierarchical approach developed for Michigan was extended to Minnesota and Wisconsin, where natural division maps did not use a multilevel hierarchical approach.

Natural Division Maps of Minnesota and Wisconsin

The present publication is based partially on modifications of the natural division maps of Minnesota (Kratz and Jensen 1983) and Wisconsin (Hole and Germain 1994). Because the Minnesota map was not hierarchical, it was necessary to determine hierarchial relationships among map units. In addition, some of the new map units were also introduced. Most of the natural divisions that were eliminated will be mapped on more detailed maps of Minnesota. For Wisconsin, I subdivided recognized natural divisions of the State into smaller map units. I further developed the hierarchy in portions of Wisconsin, but have also eliminated some of the smaller mapping units, which were too detailed to include at the scale of the three States.

Bailey's Climatic Ecoregion Map

Bailey's (1976, 1980; Bailey and Cushwa 1981) ecoregions map of North America, a part of which is reproduced in figure 2, delineates large geographic ecosystems at the scale of 1:7,500,000. Bailey's map is based on distinctive climates (following Köppen 1931), potential natural vegetation (after Küchler 1964), and soils (according to the classification of Crowley 1967). Bailey uses a hierarchical classification: domains describe subcontinental areas of broad climatic similarity; divisions describe subdivisions of the domain that are determined by isolating areas of differing vegetation and regional climates; and provinces are broad vegetation regions that have uniform regional climate and the same type or types of zonal soils).

According to Bailey's treatment, all of Minnesota, Wisconsin, and Michigan fall within the Humid Temperate Domain (unit 200 in figure 2). At the next lower hierarchical level, three divisions are represented in the three-State area: the Humid Warm-Summer Continental Division (210) (including the Laurentian Mixed Forest Province (211)); the Humid Hot-Summer Continental Division (220) (including the Eastern Deciduous Forest Province (221)); and the Subhumid Prairie Division (250) (including the Tall-grass Prairie Province (253) and the Aspen Parkland Province (254)).

GIF-Part of Bailey and Cushwa's map of ecoregions
Figure 2.--A portion of Bailey and Cushwa's (1981) map of ecoregions of North America.

In my treatment of the three-State area, each of Bailey's provinces is further subdivided into two or more sections, except for Bailey's Aspen Parkland Province, which is treated as a single section. The sections represent a further climatic subdivision of the province. Although each section generally has climatic conditions that differ distinctly from those of adjacent sections, a major change in bedrock, landform, or elevation may be at least partially responsible for the climatic change.

Although Bailey's ecoregion maps were not used to develop this three-State map and classification, the map boundaries of the two studies agree relatively well at the province/section level; and Bailey's corresponding classification units (province, division, and domain) are cited in the descriptions of the sections delineated in this study.1 A comparison of Bailey's ecoregion map (fig. 2) and the Michigan-Minnesota-Wisconsin landscape ecosystem map demonstrates this close correspondence between the boundaries of the two classifications at the province/section level. Because of the differences in scale, however, a province-level boundary may correspond to the boundaries of more than one section. For example, the southern boundary of Bailey's Laurentian Mixed Forest Province is the southern boundary of Sections VII, VIII, IX and X.


In Michigan, boundaries of the regional ecosystems were determined from analyses of meteorological data recorded by the National Weather Service network of first order and cooperative weather stations, field surveys, and existing publications of geology, physiography, and soil. In Minnesota and Wisconsin, boundaries were based largely on existing publications on climate, geology, physiography, and soil, and on modified natural division maps of Minnesota (Kratz and Jensen 1983) and Wisconsin (Hole and Germain 1994).

To develop the climatic boundaries in Michigan, daily weather data for 1951 to 1980 were obtained for 125 weather stations (Albert et al. 1986, Denton 1985). Analytical methods were chosen to minimize the effect of understated microclimatic and measurement idiosyncracies (Denton 1985; Denton and Barnes 1987, 1988). These data were used to compute a large number of statistics describing weather in Michigan; and principal component analyses were used to summarize most of the statistics into a smaller number that were used as variables in subsequent analytic comparisons of regions, districts, and subdistricts (sections, subsections, and sub-subsections). Contour maps were drawn for each major climatic statistic. Cluster analyses were used to classify weather stations into groups with similar climatic patterns. The contour maps, classifications, and major physiographic features in Michigan were used to identify climatically distinctive sections and subsections, and discriminant analyses were used to confirm the final classification. In the present publication, climatic maps from The Climatic Atlas of Michigan (Eichenlaub et al. 1990) were used to provide climatic statistics of Michigan; more detailed data from Shirley Denton's work (Denton 1985; Denton and Barnes 1987, 1988) are also included where they provide additional insights.

For Minnesota and Wisconsin, climatic data were not analyzed in detail, as in Michigan. Instead, published State (University of Minnesota 1969, 1971, 1973, 1977, 1979, 1980a-c, 1981a,b; Wisconsin Statistical Reporting Service 1967; Wisconsin Agricultural Statistics Service 1987, 1989) and midwestern climatic studies (Müller 1982, Wendland et al. 1992, Reinke et al. 1993) were examined to compare adjacent map units. Climatic statistics included in this study are growing season length, extreme minimum temperatures, average annual precipitation, and average annual snowfall.

For all three States, information from topographic maps, geological maps and publications, and soil maps and surveys was integrated to provide a preliminary physiographic and soil classification at section and subsection levels. In Michigan, to assess the accuracy and applicability of each map, forest vegetation and soils were sampled, and topographic data and notes on wetland vegetation were recorded. In Minnesota and Wisconsin, no field sampling was conducted. Instead, the presettlement vegetation maps were used to assist in developing and evaluating map boundaries, and scientists and managers from both States were asked to comment on the appropriateness of map boundaries.

Geologic variables of primary importance were bedrock type, landforms, glacial drift thickness, and hydrology. The many geologic references consulted are cited within the text. Also used were maps of bedrock geology, Quaternary geology, and glacial drift thickness.

Soils maps used include national maps of soil Orders, Suborders, and Great Groups (USDA Soil Conservation Service 1967), soil association maps of all three States, and numerous soil surveys from all three States. Information on soil texture, drainage class, topographic-soil relationships, and land use was available from these sources.

Topographic maps (scales 1:250,000; 1:62,500; and 1:24,000) were used to characterize the size and pattern of landforms, slope, and drainage class. The 1:250,000 maps were useful for recognizing topographically distinct areas at the section and subsection levels. The 1:62,500 and 1:24,000 maps were used more extensively to delineate more exact physiographic boundaries at the subsection and sub-subsection levels.

Integration of Climatic and Physiographic Classifications

In Michigan, although separate classifications of climate and physiography were made, an integrated classification and map were planned from the outset. Boundaries of the climatic map units were purposely placed along physiographic boundaries so that a minimum of change would be necessary when the physiographic classification was completed. When both classifications and maps were ready, the differences between them were carefully examined and boundaries were adjusted as necessary. For further discussion of the integration of the climatic and physiographic classifications, see Regional Landscape Ecosystems of Michigan (Albert et al. 1986).

In Minnesota and Wisconsin, climatic data were not directly used to generate section boundaries; instead, a combination of bedrock features, glacial landform, soils, and the response of vegetation to these abiotic factors was more important for delineating section boundaries, but climatic boundaries proved to be recognizable and important as well. For example, a strong precipitation gradient forms the boundary between Sections IX and X.


Vegetation played a varying but critical role in validating climatic-physiographic boundaries. In Michigan, pre-European settlement vegetation had not been adequately mapped for large portions of the State; accordingly, this information was not used initially for developing or validating boundaries. In contrast, both Minnesota and Wisconsin have presettlement vegetation maps (Marschner 1974, Finley 1976) that provided useful insights into boundary delineation. The General Land Office (GLO) surveyor's notes are now being interpreted for Michigan by Michigan Natural Features Inventory (Comer et al. 1993a, 1993b, 1994, and work in progress) and are the source for the detailed description of presettlement vegetation found within this text.

For further detail on the theory and practice of regional and local ecosystem classification, consult papers by Rowe (1971, 1979, 1984a, 1984b, 1992), Rowe and Sheard (1981), Bailey (1983, 1985), Barnes et al. 1982, and Barnes (1986, 1993).

1 Recent publication of an eco-subregion map by the USDA Forest Service (Bailey et al. 1994, McNab and Avers 1994), part of an ongoing Forest Service project to delineate subregions throughout the United States, have modified Bailey's province boundaries to correspond to those deliniated in earlier drafts of this publication (Regional Landscape Ecosystems of Michigan, Minnesota, and Wisconsin). The ecosubregion map and its concept of hierarchy are not widely incorporated into this classification.
Previous Section -- Acknowledgments
Return to Contents
Next Section -- General Setting

Accessibility FOIA Privacy Policies and Notices

Take Pride in America logo logo U.S. Department of the Interior | U.S. Geological Survey
Page Contact Information: Webmaster
Page Last Modified: Saturday, 02-Feb-2013 04:56:33 EST
Reston, VA [vaww54]