A multifactor ecological classification of the northern hardwood and conifer ecosystems of Sylvania Recreation Area, Upper Peninsula, Michigan

An ecological method of multifactor ecosystem classification was applied in the Sylvania Recreation Area, an 8500-ha tract of old-growth northern hardwood – conifer forests in upper Michigan. The uplands and wetlands were subjectively classified into 25 ecosystems by a method combining reconnaissance, plot sampling, data analysis, and ecosystem mapping. Each ecosystem was a characteristic combination of physiography, ecological species groups (ground vegetation), and soil. Discriminant analysis was used to evaluate the distinctness of the upland ecosystems and to compare the discriminating abilities of different ecosystem components (physiography, ground vegetation, and soil). The classification was corroborated in the multivariate analyses. The lowest estimated error rate (9.4%) in discriminant analysis was obtained by a model based on a combination of physiographic and soil characteristics and ecological species groups. The estimated error rates based on the species groups alone and physiography and soil alone were 42.2 and 25.0%, respectively. The discriminant analyses indicate that neither vegetation alone nor physiography and soil alone could be used with high reliability in classifying and mapping ecosystems. An additional discriminant analysis of the three ecosystem components indicated that the ecosystems could be distinguished by field characteristics without information from soil laboratory analyses. This analysis also demonstrated the particular value of the vegetation component as a readily observed, acceptable substitute for soil laboratory data in identifying and mapping ecosystem units.