Landscape ecosystems of disturbed oak forests of southeastern Michigan, U.S.A

An ecological multifactor approach was used to identify and describe oak ecosystem types in highly disturbed landscapes and fragmented forests in an area of over 19 000 km 2 in southeastern Michigan, United States. Eleven upland ecosystems and 1 wetland ecosystem were identified in the field using reconnaissance, plot sampling, and test mapping. Each ecosystem type was a characteristic combination of physiography, soil, and climax vegetation (overstory and ground-cover vegetation). The ecological approach emphasized physiographic and soil factors because of the disturbed state of the vegetation. Of 222 species of ground-cover vegetation, only 68 were used in forming the 13 ecological species groups. White oak (Quercusalba L.) exhibited the largest ecological amplitude of the three major oak species; it occurred on dry to mesic sites. Red oak (Q. rubra L.) occurred on dry-mesic to mesic sites, and black oak (Q. velutina Lam.) was restricted to dry sites. Discriminant analysis was used to examine the distinctness of the upland ecosystems and to compare the error rates of different ecosystem components. The misclassification rates obtained by using all ecosystem components (physiography, soil, ecological species groups, and overstory vegetation) were the lowest: 20% in highly dissected terrain and 34% in flat to gently rolling terrain. However, results obtained with physiography-soil and ecological species group variables were nearly as good as results that added the overstory vegetation. More overlap among ecosystem types and higher misclassification rates were found than in ecosystems of old-growth forests of northern Michigan and oak forests in southwestern Wisconsin where similar methods were used. Nevertheless, for the highly disturbed forests of southern Michigan, the ecological, multifactor landscape approach is a useful and effective method of identifying, describing, and mapping ecosystem types.