Ecologists increasingly are recognizing the importance of studying geographically defined ecosystems in addition to their component species, populations, and communities. Results are presented from a long-term project undertaken to describe and classify the landscape ecosystems of the University of Michigan Biological Station (UMBS) to provide a baseline for future studies of ecosystem change. The objectives of the current study were: (1) to determine the extent to which the selected ecosystem types differ with regards to soil pH, soil texture, and ground-cover vegetation, and (2) to evaluate the distinctness of the selected ecosystem types and groups, and to determine which soil and vegetation variables best discriminate among them. The University of Michigan Biological Station is situated in a landscape characterized by a great diversity of landforms and soils owing to its history of glaciation and fluctuating Great Lake levels. More recent disturbances by logging and severe post-logging fires have complicated and obscured the effects of this ecosystem diversity on the existing vegetation. A landscape ecosystem classification and map were developed using a multi-factor approach incorporating landform, microclimate, soil, and vegetation. Sample plots were established and soil attributes and the species composition of the overstory, understory, and ground cover were determined. Ground-cover species observed to have similar indicator values were combined into ecological species groups. Twelve ecosystem types representing a diversity of landforms were chosen for further analysis and comparison. These 12 types were divided into 4 ecosystem groups for analysis –Total, Pellston Plain (low-level outwash), High-Level Outwash and Ice Margin, and Moraine. Soil texture, soil pH, and the coverage of 16 ecological species groups were compared for each ecosystem group using Kruskal-Wallis tests. Discriminant models were developed using soil variables only, species group variables only, and combined-variable sets for each of the four ecosystem groups. """ Significant differences were found among landscape ecosystem types in each of the four ecosystem groups. The 12 landscape ecosystem types of the Total ecosystem group exhibit significant differences in all 8 soil attributes and 15 of the 16 species group variables. The other three ecosystem groups, combining types from similar landforms, had fewer significant differences among ecosystem types. Ecosystem types of the Pellston Plain ecosystem group differed very little in soil pH or texture, but they differed greatly in occurrence of species groups. Ecosystem types of the High-Level Outwash and Icemargin ecosystem group differed significantly in both soil texture and soil pH, as well as in species-group coverages. Ecosystem types of the Moraine ecosystem group had""" significant differences in soil pH and in species-group coverages. """ Discriminant analyses of the Total ecosystem group effectively separated the moraine ecosystems from those of the Pellston Plain and the High-Level Outwash and Ice-Margin, illustrating the marked differences among these ecosystems. Discriminant analyses were successful in distinguishing among types in each of the other ecosystem groups, although some overlap among types occurred. Discriminant functions that combined both soil and species-group variables were usually better predictors of a sample plot’s correct ecosystem type than functions using either soil or species-group variables alone.""" """ The landscape ecosystem approach to classifying and mapping University of Michigan Biological Station lands was successful in identifying units which differ in the important soil and ground-cover vegetation variables. Multivariate discriminant analyses were useful in distinguishing the ecosystem types and ecosystem groups and identifying the specific factors that best distinguished them. Despite the fact that UMBS landscape has been highly disturbed by logging and burning and that the analyses focused on relatively similar ecosystem types within groups, no compelling reason was found for combining any of the ecosystem types. Thus quantiative analyses using relatively few variables serves to confirm the classification and field mapping processes. Furthermore, additional ecological factors not quantified in this study, such as microclimate, landform traits, physiographic position, and overstory composition, would provide additional evidence for differences among the ecosystem types.""" Abstract - Part II Populations of many neotropical migrant bird species are in decline. Habitat degradation and destruction is implicated in this decline, but the specific habitat requirements and preferences of many neotropical migrants are poorly understood. The classification of landscape ecosystems at the University of Michigan Biological Station, Northern Lower Michigan was used as a framework to compare the breeding bird communities of four ecosystem types similar in overstory composition but differing in other vegetative and abiotic parameters. Four ecosystem types were chosen, representing a range of forested habitats for breeding birds. Data on bird occurrences in these four ecosystem types were gathered by point-count censuses. Each sample point was censused repeatedly throughout the 1993 breeding season. At each census point, all birds observed by sight, song, or call note were recorded, and the maximum number of individuals of a given species found on a single census was used as the estimate of that species+ abundance at a census point. """ Kruskal-Wallis and nonparametric least significant difference tests were used to compare the abundances of the most common bird species in each of the four ecosystem types. The species richness and abundance (number of individuals) of each census point were also compared among ecosystem types. Total Bird Diversity, a modification of the Shannon -Weaver Index, was calculated for each census point as well as for each ecosystem type as a whole. Qualitative (Jaccard’s) and quantitative (Sorenson+s) similarity indices were also calculated for each type.""" """ A total of 43 bird species were observed in the 23 census points. Ecosystem type 1, on the Pellston Plain, was the least similar to the other types in the composition of its breeding bird community according to both qualitative and quantitative measures. The Pellston Plain is distinct from the other three ecosystem types sampled in its climate and in the species composition, structure, and stature of its forest. Species richness, abundance, and TBD were not significantly different among types. Ten of the most common bird species had significantly different abundances among ecosystem types.""" """ The units identified by the landscape ecosystem classification of UMBS differ in""" the composition of their breeding bird communities as well as in their physiography, soils, and vegetation. Landscape ecosystem classification in disturbed areas may be useful for identifying variation across the landscape that effects breeding bird communities but which may be overlooked by habitat evaluations based on overstory tree community types.