Wetland ecosystems are three-dimensional, volumetric units of the earth’s surface, including climate, physiography, hydrology, soil and biota, that occur in a hierarchy of spatial sizes. Wetland ecosystems are found in certain hydrologic settings within specific physiographic features. Furthermore, local wetland ecosystems are integral components of characteristic landscapes within a given region and are inextricably linked to their therrestrial and aquatic counterparts. The wetlands of northern Lower Michigan, including UMBS, were first described by Gates in 1942. Numerous additional studies have been conducted on vegetation-environment relationships in the area, primarily focusing on correlations between specific, individual, hydrological components and vegetation composition. However, there has been no integrated discussion of the wetland ecology of UMBS at local, landscape and regional scales in the intervening half-century, despite the proliferation of studies in the region and numerous theoretical advances in the field. Furthermore, techniques of canonical correspondence analysis have been recently developed which allow for multivariate, direct gradient analysis of vegetation-environment relationships. The research described herein was part of a project begun in 1987 to develop an inventory and map of upland and wetland landscape ecosystem types at UMBS, to be used as a baseline for assessing climate-induced changes and biodiversity at the landscape level. The primary goal of my study was to develop an ecological classification and analysis of the wetland ecosystems of UMBS. Specific objectives were to: i) identify and describe wetland ecosystem types using a multi-factor approach ii) classify the ecosystem types in three different ways: a physiographic classification that emphasizes landscape linkages; a hydrologic classification that reflects similarities in source of inputs, water regime, and water chemistry; and a classification that depicts commonality in the physiognomy and composition of dominant vegetation iii) develop ecological species groups for the area iv) determine the degree of distinctness of the ecosystem types, and indirectly evaluate the multi-factor approach, using multivariate techniques v) identify the abiotic parameters that have significant influence on patterns on groundflora composition. My approach was to use a landscape ecology and ecosystem perspective, emphasizing the interactive role of climate, glacial geology, landscape linkages, and physiographic-hydrological characteristics in determining patterns and processes. The study was conducted in an ca 4000-ha tract located in northern Lower Michigan. The region is characterized by a cool, moist, temperate climate and is covered in glacial deposits. The wetland ecosystem types were identified using a multi-factor, field-based approach that integrates the combined effects of climate, physiography, hydrology, soil and vegetation. Field reconnaissance and transects were used initially to characterize the diversity of wetlands in the area. Forty permanent sample plots were randomly established in eight major, frequently occurring ecosystem types. Extensive vegetative, hydrological and soil data were collected to produce descriptive and tabular comparisons among ecosystems. Groups of groundflora species that indicate a specific range of environmental conditions, ecological species groups, were developed from reconnaissance and field sampling, and corroborated by ordination and polythetic, divisive clustering. In addition, multivariate techniques of canonical variates analysis and discriminant analysis were used to examine the distinctness of the ecosystem types relative to dominance type, indicator species, hydrologic, soil and combined variable sets; these analyses were also used to evaluate the multi-factor approach, in comparison to single-factor approaches. Relationships between groundlfora vegetation composition and environmental gradients of hydrology, soil, and light were investigated by both indirect (detrended correspondence analys s) and direct (canonical correspondence analysis) methods. The results and general conclusions from the study were as follows: 1. 18 wetland ecosystem types were identified and described, including nutrient-rich swamps, ombrotrophic bogs, and a host of intermediate ecosystems. The ecosystem types occured differentially within specific landscape features. For example, swamp ecosystems were found in broad, level outwash and glacial lake plains; bogs were found in trapped depressions of ice-contact origin. 2. The ecosystem types were classified in three alternate ways. In the physiographic classification, ecosystems were distinguished on the basis of major physiographic feature (outwash-lakeplain, ice-contact, or moraine), specific landform, linkages to other systems, and hydrologic-soil-vegetative characteristics. The hydrologic classification grouped ecosystems on the basis of physiographic setting and associated hyrologic source (trapped depressions, level plain or riparian/lakeside), water chemistry and water regime. Neither landscape linkages nor functional similarities were reflected in the physiognomic dominance-type classification based solely on vegetative characteristics; there were several examples of hydrologically and physiographically distinct wetland types falling under the same physiognomic category. 3. Seven ecological species groups were identified in the wetlands, tolerant of a range of water chemistry values from extremely acid to alkaline. The groups reflected three categories of moisture conditions (very moist to wet, and very wet). Five of the seven species groups were characteristic of shaded light environments, beneath moderately to completely closed canopies; two were found in open, high light conditions. Although no quantitative analyses were performed, patterns of species groups occurrence within specific ecosystem types were observed. Two or more species groups typically characterized a given ecosystem. 4. There were clear differences in various vegetative, hydrologic and soil characteristics among the eight major ecosystem types. The best separation among wetland ecosystem types was obtained using either a single-factor variable set based on groundlflora species, or a combined data set, reflecting multiple factors. Ecosystems were very poorly differentiated on the basis of dominance type, despite its prevalent use in wetland classification systems. Ecosystems were also somewhat poorly separated by soil or hydrological variable sets, two other commonly used factors in wetland classification. 5. Substrate characteristics (primarily organic matter type, and to a lesser degree depth of organic matter) were the most important variables explaining species-environment relationships of groundflora vegetation. Light conditions, as reflected in tree and shrub coverage, were also important. In addition, depth and pH of water had an influence on community composition. The field-based, multi-factor approach to wetland identification used in this study is an appropriate framework for both management and research applications. Wetlands identified in this manner can subsequently be compared on the basis of any number of biotic and abiotic factors, and considered in the context of other common approaches to wetland identification and classification. The three separate wetland classifications derived from the comprehensive inventory and description of ecosystem types at UMBS demonstrate the flexibility of such an approach. For example, multivariate comparisons and analyses indicated that wetlands were poorly separated when classified on the basis of dominance type, hydrologic, or soil characteristics alone. In the range of variation in wetlands at UMBS these factors appear to work well only in certain situations, such as in the separation of types dominated by northern white-cedar from the others. These results suggest that a multi-factor approach can be valuable in small scale, local areas. The fundamental importance of groundflora indicator species to wetland classification was also recognized. Results from th study of vegetation-environment relationships, using the multi-factor framework as a basis for sampling, were equally promising. In addition to several hydrological parameters previously identified in the literature as significant factors effecting vegetation composition in northern wetlands, multivariate direct gradient analysis techniques revealed the importance of substrate characteristics. Additional studies on the role of these factors in influencing vegetation dynamics are warranted; particularly in regards to the complex interaction among hydrology, soils and the microbial-mediated nutrient pathways essential to plant growth and establishment.