A Spatial Analysis of the Great Lakes Ecoregions
George E. Host, Mark A. White,
and Philip L. Polzer
Natural Resources Research Institute
University of Minnesota
Duluth, MN 55811
This project was conducted under the auspices of the USFS Great Lakes Assessment
project. The Great Lakes Assessment is a large-scale interdisciplinary
project designed to develop databases and tools for assessing the ecological
and socioeconomic health of the Great Lakes Region. Partners in this project
include the US Forest Service, the Natural
Resources Research Institute, the University of Wisconsin-Madison,
Michigan State University, and numerous other federal and state agencies.
This project consists of two phases. The first phase will compile existing
information previously managed by different federal and state organizations.
This includes (1) environmental information on climatic
gradients, surficial geology, ecological units, soils, hydrography,
drainage patterns, and contaminants; (2) biological information on current
forest conditions, the biogeography of game, non-game, and threatened and
endangered species; (3) socioeconomic information on land use and ownership,
human demographics, recreational demand, and road densities; and (4) ecological
process information on the frequency and effects of natural and anthropogenic
disturbance associated with fire, wind, and flooding events, and resource
development and consumption.
The second phase of the project involves spatial analyses of geographic
information, and the development of data visualization/decision support
environment for natural resource planners and managers. This latter product
will allow complex information to be placed in a format that is easily
understood by the public, policy makers, and other non-scientific audiences.
As a first step in this project, we have conducted a spatial analysis
of the Lake States Forests based on a Landsat Thematic Mapper classification
of northern Minnesota and northwestern Wisconsin. This charts below briefly
summarize the composition and spatial structure of the Great Lakes region,
in the context of an ecological classification of this landscape.
Forest Composition
The type and distribution of land cover is one of the most fundamental
descriptors of the landscape. In this analysis, we describe the ecoregions
in terms of relative cover. Across the region, the dominant cover type
is Upland Grass/Agland,
which includes agricultural lands, grasslands, and grass/herbaceous mixtures;
this type constituted 18% of the total area. The next most dominant type
is the Aspen-birch forest
type, accounting for 13% of the total land area. Northern
Hardwoods, inland waters, and spruce-fir/hardwoods each account for
approximately 10% of the region. These five cover types thus account for
about 60% of the land area.
Cover types were not uniformly distributed across ecoregions; in fact,
there were wide variations in relative abundance by ecoregion. We use relative
rather than absolute abundance to account for differences in ecoregion
size and the fact that varying portions of each ecoregion were sampled.
The Upland Grass/Agland
category was the most variable cover type: ecoregions in the southern tier
of the study area (212Jd, 212Jg, 222Mc, and 212Kb) had 30 to 50% of their
total area in Grass/Agland. Ecoregions within the Superior scene, in contrast,
generally had < 3% in this category.
Northern Hardwoods
were the next most variable category. Ecoregions of the Chequamegon scene
(with the exception of 212Jg) ranged from 30 to 50% Northern Hardwoods.
All other ecoregions were 15% or less of this type. The Spruce-fir/hardwood
type was the inverse of the Grass/Agland distribution: 18% or more of the
Superior-scene ecoregions were in the Spruce-fir/hardwood type, whereas
the southern ecoregions generally had <3% in this type.
Finally, Pine
is most dominant on the sandy outwash materials of the Chequamegon National
Forest. This region hold the largest tracts of pine stands.
Landscape Pattern
Landscape diversity H, or Shannon's evenness index measures the evenness
of the proportional distribution of patch type area. H increases with increasing
evenness. Shannon's H' decreased from north to south reaching 2.53 in subsection
212Ma (Chippewa scene) and decreasing to 1.40 on 212Jd (St Croix). This
index is a function of both the absolute number of patch types, and their
relative distribution. A low H' indicates that a few cover types dominate
the map, whereas a high H' indicates a larger number of patches which are
more evenly distributed across the landscape.
Contagion and Angular Second Moment (ASM) describe the spatial component
of diversity. Specifically, they show the degree of aggregation or ‘clumping'
of pixels. The maps of contagion are quite different than those of H'.
The most highly aggregated patch structures were found in subsection 212Jd
(St Croix) and 212Lb on Superior's north shore. Two different cover types
account for these patterns. Agricultural lands are the dominant patches
in 212Jd, whereas boreal forest forms large patches in 212Lb. The lowest
degree of contagion (i.e. most dispersed patch structures) was found within
the Chippewa scene.
Conclusions
This type of information is useful for assessing the current state of
the landscape, and for future land management planning. Understanding the
composition and structure of the regional landscape allows effective planning
for timber supply, available habitat for birds, mammals, and herptiles,
recreational land use, and numerous other public objectives. Future work
will examine finer-scale spatial patterns, three-dimensional representations
of the landscape, and integration of geographic data with predictive models.
Collectively, the Great Lakes Assessment will provide tools and information
to assist land managers in ecological and economically-sound regional assessments
and planning.
|
