Historical changes in wetland classes in three shoreline fens, Baie du Dore, Oliphant, and Howdenvale, along the eastern shoreline of Lake Huron were analyzed to determine responses to fluctuating water levels. Wetland classes (lake, floating emergent, emergent marsh, meadow marsh, fen, and exposed substrate) were delineated through interpretation of aerial photographs for the period 1938 to 1995. Scan vectorization was used to develop a digital data base of wetland classes. Spatial and temporal analyses, conducted in a Geographic Information System (GIS), allowed comparison of trends among and within wetland classes which were linked to water level conditions. In periods with low water levels, overall wetland area increased primarily through expansion of the exposed substrate class lakeward; although meadow marsh also contributed to the increase. While wetland area increased during low water levels, exposed substrate did not markedly add to wetland habitat value but represents potential new wetland area that can be recolonized from seed banks. During high water levels, wetland area decreased as the lake class expanded inland and exposed substrate contracted or disappeared. Fen remained consistent in areal extent and location for most years at Howdenvale, but fen at Baie du Dore expanded with lower water levels, and contracted and became more fragmented with higher levels. At Oliphant, during low water levels fen area did not change, and with high levels the lake flooded the fen. With high water levels, lake area and wetter classes dominated and migrated inland. Under low water level conditions, lake was replaced by exposed substrate and drier wetland classes expanded, but did not necessarily colonize exposed substrate.
Great Lakes coastal wetlands are located in the dynamic shoreline environment between the open water of the Great Lakes and terrestrial uplands. The most common wetland type in the Great Lakes Basin is marsh, which supports a mix of floating, submerged, emergent, and meadow vegetation (Maynard and Wilcox, 1997). Other wetland types include swamps, which are dominated by tree and shrub species, and fens and bogs, which are peat-accumulating wetlands dominated by meadow and moss species, respectively. Water level fluctuations, along with climate variations, shoreline wave and sediment dynamics, and land use changes are key factors determining wetland distribution, area, and biotic communities (Mitsch, 1992). Numerous studies in the Great Lakes Basin (e.g., Lakes Ontario, Erie, St. Clair, and Michigan) have explored the influence of water level fluctuations on wetland extent, as well as vegetation and other wetland-dependent species (Jaworski et al., 1981; Keddy and Reznicek, 1986; Quinlan and Mulamoottil, 1987; Casanova and Brock, 2000; Environment Canada, 2002). However, few have focused on the response of Lake Huron wetlands (Fahselt and Maun, 1980; Wilcox, 1995). None have addressed coastal fens in particular. This study used air photo interpretation and spatial analysis in a Geographic Information System (GIS) to document and assess wetland areal changes and class responses to water level fluctuations in the 20th century. The assessment focused on three fen-dominated wetland complexes; Baie du Dore, Oliphant, and Howdenvale, along the eastern coast of Lake Huron. The objective was to determine whether the three fen-dominated wetland complexes on Lake Huron responded to water level changes in a manner similar to wetlands, typically marshes, in the lower Great Lakes (Herdendorf et al., 1981; Bolsenga and Herdendorf, 1993; Minc and Albert, 1998; Wilcox, 2004). This paper reviews the influence of water level fluctuations on Great Lakes coastal wetlands, describes the selected study sites, outlines the methods used to develop the digital wetland data base, and describes wetland area and class responses to water level conditions.
Coastal wetlands and water level fluctuations
Water levels in the Great Lakes fluctuate both in the short-term and over the long-term. Short-term meteorological events create storm surges, wind set-ups, and high waves that can have dramatic impacts on wetland geomorphic form and vegetation communities. Seasonally, water levels in Lake Huron fluctuate about 0.3 m from a high in summer to a low in winter, reflecting the influence of snow accumulation, melt and runoff, and evapotranspiration. Inter-annual and inter-decadal fluctuations in water levels are key perturbations affecting wetland ecosystems. Water levels are influenced by climatic variability related to precipitation, evapotranspiration and runoff; although human activity such as dredging and infilling in the connecting channels can also influence water levels (Derecki, 1985; Quinn, 1985; Quinn et al., 1993). For the period 1900 to 2005, the lowest water level on Lake Huron occurred in 1964 (175.7 m) while the highest water level occurred in 1986 (177.3 m). On longer time scales, a 4700-year lake-level record reconstructed for Lake Michigan (and also applicable to Lake Huron) shows paleo lake-level fluctuations ranging 0.5 to 0.6 m about every 33 years with longer-term fluctuations about every 160 years (Baedke and Thompson, 2000). In the near future, the effects of human-caused climate changes on the hydrology of the Great Lakes basin are likely to emerge with implications for coastal wetland ecosystems. Most but not all climate change impact assessments, project reductions in Great Lakes water levels for the 2050s; ranging from 0.10–1.00 m depending on the global climate model, emission scenario and lake examined (Mortsch, 1998; Mortsch et al., 2000; Lofgren et al., 2002; Mortsch et al., 2006a). Wetlands in Canada are defined as “land that is saturated with water long enough to promote wetland or aquatic processes as indicated by poorly drained soils, hydrophytic vegetation, and various kinds of biological activity which are adapted to a wet environment” (National Wetlands Working Group, 1997). Water level fluctuations through changes in water depth and duration of flooding influence wetland area, diversity of vegetation, and relative abundance of vegetation communities. Seasonal and inter-annual water level changes help maintain plant diversity in wetlands (Jaworski et al., 1979; Keddy and Reznicek, 1986; Quinlan and Mulamoottil, 1987; Williams and Lyons, 1991; Wilcox, 1993; Keough et al., 1999; Casanova and Brock, 2000).
An array of substrate types, shoreline gradients, water depths and flooding durations occurs within wetlands (Keddy, 2000; Mitsch and Gosselink, 2000). Many plants have unique characteristics that enable their growth and survival in wetland environments. Some species tolerate a range of environmental conditions, while others have a narrower niche (Keough et al., 1999; Meyer et al., 2006). Wetland plant species have been classified into vegetation communities that reflect similar tolerances; they, generally, grow at comparable elevations and moisture conditions (Wilcox, 2004). A plant community displacement model developed for Lakes Erie and St. Clair includes zonation of submergent macrophytes, surface floating macrophytes, emergent macrophytes, wet meadow, and trees and shrubs and reflects a progression from shallow water to the shore and upland (Herdendorf et al., 1981; Bolsenga and Herdendorf, 1993). Minc and Albert (1998) characterized specific vegetative zones for the Great Lakes that included submergent marsh, emergent marsh, shoreline or strand, herbaceous or wet meadow, shrub swamp and swamp forest; all these zones are not necessarily well-developed or present in every wetland.
Wetland vegetation communities have been shown to expand and contract along a moisture gradient associated with fluctuating lake levels on Lakes Ontario, Erie, and St. Clair (Harris et al., 1981; Burton, 1985; Keddy and Ellis, 1985; Herdendorf and Raphael, 1986; Bolsenga and Herdendorf, 1993; Maynard and Wilcox, 1997; Minc and Albert, 1998; Mortsch, 1998; Keough et al., 1999; Wilcox, 2004). In response to water level fluctuations, vegetation communities are displaced landward or lakeward and their areal extent contracts or expands. Total wetland area also changes. For example, during low water years the shoreline retreats lakeward, and substrate is exposed and landward margins of wetlands dry out; total wetland area increases. Areas typically dominated by submerged aquatic vegetation are replaced by emergent vegetation in shallow water. Emergent vegetation becomes drier and denser on the more upland portions, and meadow marsh as well as shrub communities expand. With higher water levels, wetland vegetation communities retreat landward or are flooded. Typically, open water, submerged aquatic vegetation and floating-leaved emergent vegetation classes expand. Total wetland area decreases due to flooding and the landscape becomes increasingly fragmented.
The three wetland sites along the eastern coast of Lake Huron, Baie du Dore, Oliphant, and Howdenvale (Figure 1) represent relatively large, lake-influenced wetlands that have a high proportion of unique, endangered fens. These rich northern fens show a somewhat different community composition than the marsh-dominated wetlands in lower Great Lakes, which have received much research attention (Natural Heritage Information Centre, 1995; Maynard and Wilcox, 1997; Minc and Albert, 1998). The Baie du Dore coastal wetland, an Area of Natural and Scientific Interest (ANSI), is located adjacent to the Bruce Nuclear Plant, north of the town of Kincardine. It is exposed to the direct influence of water levels in the lake, but is sufficiently sheltered to allow for the development of extensive, diverse coastal wetlands. The marshes in Baie du Dore are similar to marshes near Port Elgin and Oliphant (Fahselt and Maun, 1980).
The Oliphant coastal wetland complex is comprised of the Fishing Islands and mainland shoreline and inland wetlands situated north of the town of Sauble Beach. The wetland complex is located in the lee of the Fishing Islands on a flat very gently sloping dolostone plain that dips towards the southwest into the sheltered bay. Oliphant has the most widespread human influence of the three sites including an active cottage community, a public boat launch and marina with a dredged channel for boat traffic, and a shoreline road that runs along the inland edge of much of the wetland complex (Langille et al., 1984a; Kirk et al., 1985; Falls et al., 1990).
Howdenvale, the most northerly site, is located on the Bruce Peninsula, near Little Red Bay. The fen complex is situated on a wet sand plain behind a sheltered bay. Lake Huron and an abundant supply of groundwater maintain moist soil conditions, while areas of beach deposits create drier ‘islands’ in the wetland (Cuddy et al., 1976; Langille et al., 1984b; Johnson, 1990). The area consists primarily of natural areas with cottages and road development in the south-western portion of the wetland. The Howdenvale wetlands are designated as a Provincial Life Science Area of Natural and Scientific Interest (ANSI-LS). Many unusual, calciphilic plant species occupy the unique niche afforded by calcium-rich groundwater and soil in the shoreline fens (Ontario Nature, 2005).
Aerial photography has been used to show changes in Great Lakes wetlands due to fluctuations in water levels on Lakes Erie, St. Clair, Michigan, and the St. Mary's River shoreline (Jaworski et al., 1979; Harris et al., 1981; Whillans, 1985; Williams and Lyons, 1991; Lyon and Greene, 1992; Williams and Lyon, 1997; Gottgens et al., 1998) as well as to document changes in wetland extent due to human modification (Jaworski and Raphael, 1976; Bosley, 1978; Busch and Lewis, 1984). Historical aerial photographs and geospatial data have been used to guide wetland restoration (Kowalski and Wilcox, 1999).
For the Lake Huron study, the development of the digital wetland data base and the GIS-based spatial and temporal analyses of the wetland classes followed the same method as reported in Wilcox et al. (2003) and Hebb (2003):
– Black and white (and one colour infrared) airphoto coverages were interpreted, by one analyst, using a mirror stereoscope. The classification scheme included: open water, submergents, emergents, meadow, open bog, and trees and shrubs as well as adjacent land uses (e.g., built-up area, homes, agriculture, roads) (Snell and Cecile Environmental Research, 1992; Wilcox et al., 2003). For the Lake Huron wetlands, fen classifications consisting of fen, fen with sparse sand showing and fen with scattered trees were also interpreted but combined into one category for analysis here.
– The class boundaries and attributes of the wetland communities and associated land uses were outlined on the airphotos and transcribed onto Ontario Base maps (OBMs) (1:10,000 scale) (Snell and Cecile Environmental Research, 1992, 2001). Hebb (2003) describes the method for maintaining consistency between years with different scales of aerial photographs as this was critical for accurately assessing and inter-comparing temporal and spatial trends.
– The interpreted wetland classification maps were scan-vectorized into a GIS and georeferenced to real-world Universal Transverse Mercator (UTM) coordinates. The final output layers were polygon coverages containing a wetland class value for each polygon (see Hebb, 2003; Sabila, 2005).
– Each year of wetland data was clipped to a common study area boundary unique to each wetland site. All non-wetland classes, except exposed substrate, were not included in the analyses.
– The original wetland classifications were grouped into six wetland classes reflecting wetland communities related to water levels and the unique fens. The classes (tending from the lake to drier inland conditions) included: lake, floating emergent, emergent marsh, meadow marsh, fen, and exposed substrate.
– The polygon coverages of the historical wetland class layers were converted to grid format in ARC/INFO using a 10 m by 10 m cell resolution; the minimum mapping unit for this analysis was therefore 100 m2. The minimum mapping unit could have been smaller for some maps since the fen classification maps were developed from air photos ranging from 1:8000 to 1:17,000 in scale (see Table 1). When the data were converted to grids, any polygons smaller than half a cell size would have been incorporated into the surrounding or adjacent class.
– FRAGSTATS 3.0, a landscape and spatial pattern analysis program for categorical maps (Pacific Meridan Resources, 2000), was used to calculate a variety of landscape metrics for the classes in each wetland for each year of aerial photograph coverage. Temporal and spatial trend analyses were undertaken to determine how wetland classes changed over time (see Sabila (2005) for a detailed description of the methodology).
The three study sites are Ontario Ministry of Natural Resources evaluated wetlands with field assessments (Ball et al., 2000). Seven years of aerial photograph coverages (1938, 1954 or 1955, 1966, 1978, 1985, 1988, 1995) were used for the spatial/temporal analysis. There was full coverage for all three wetland sites (except where 1954 coverages had to be used for Howdenvale and Oliphant, and 1955 for Baie du Dore). The air photo years correspond with a range of water levels including low, medium and high water levels and rising and declining conditions (Table 1). The Lake Huron water levels used in this study were based on whole lake averages from a number of Canadian and U.S. water level gauging stations around the lake (R. Moulton, Environment Canada, Burlington, ON, Canada, pers. comm.). The air photos were a suitable scale. Generally, they were also taken during the preferred season, summer, for interpretation. The 1988 aerial photos were from spring (April and May) which could reduce the differentiation of vegetation classes, particularly floating emergent and emergent marsh, or fen near the shoreline and exposed substrate.
The aerial photograph analysis key was originally developed for the Canadian Wildlife Service to document wetland vegetation change over time on Lakes Ontario, Erie, and St. Clair for the International Joint Commission (IJC) Water Level Reference (Snell and Cecile Environmental Research, 1992; Working Committee 2 Land Use and Management, 1993). Subsequently, more air photo analysis extended the time series for wetlands in Lakes Ontario, St. Clair and Erie (Snell and Cecile Environmental Research, 2001). This also allowed for the first assessment of Lake Huron fens reported in this study and an assessment of the historical distribution of Phragmites australis at Long Point, Lake Erie (Wilcox et al., 2003). The classification, reflecting a sequence from wet to dry conditions, allowed for consistent interpretation of historical aerial photographs to understand long-term ecological change in coastal wetlands with an emphasis on understanding lake level effects on coastal wetlands. The categories were selected to allow as much delineation as possible but also allow for combining of classes for analyses, comparisons, and modelling. Subsequently, the Ecological Land Classification (Lee et al., 1998) was developed; the two classification systems correspond quite closely at the community class level (Snell and Cecile Environmental Research, 2001). For the three Lake Huron sites, the 1985 coverage was interpreted, mapped, and then calibrated with the 1984 Ontario Ministry of Natural Resources wetland evaluation maps and field notes (Ball et al., 2000; Snell and Cecile Environmental Research, 2001) in a similar manner as Chow-Fraser et al. (1998) and Gottgens et al. (1998). For Baie du Dore, the 1:2,000 floodplain mapping from the Saugeen Valley Conservation Authority based on 1988 air photos was compared with the wetland boundary interpretation. Generally, they matched well although some variation occurred, especially along the very gently sloping areas of the inland boundary (Snell and Cecile Environmental Research, 2001). The Lake Huron sites were visited in 2001 with the most recent air photos, topographic maps, surficial geology maps, and interpreted GIS-based maps to evaluate the general wetland class types. Subsequent field visits were made in 2002, 2003, and 2004.
FRAGSTATS calculates a variety of landscape metrics, based on concepts from landscape ecology, to describe patterns and underlying processes in the landscape by focusing on characteristics including structure, function, and change (McGarigal and Marks, 1995; McGarigal et al., 2001; McGarigal et al., 2002). For this paper, metrics such as class area and mean patch size are discussed although edge, nearest-neighbour, shape, diversity, contagion, and interspersion characteristics were computed at the landscape and class level. These simple metrics were linked with mean annual water level conditions to determine if patterns detected in the landscape structure (e.g. the dominant wetland classes) and any changes in area of those classes were related to water levels.
Wetland vegetation community changes
Baie du Dore
Both the total wetland area (excluding lake) of Baie du Dore and various wetland classes responded to water level fluctuations (Figure 2a). As expected from the conceptualization of wetland response to water level change, total wetland area increased during low water levels (1938 and 1966) with exposed substrate comprising a large proportion of the landscape along with fen, meadow marsh and emergent marsh. With higher water levels (1955, 1985) inundating the wetland, total wetland area decreased and exposed substrate also decreased dramatically or disappeared. While fen was a dominant wetland class in the landscape in all years (ranging from 5.4% to 10.4% coverage), its area expanded during low water years and contracted, due to flooding, during high water years. The maps, in Figure 2b and c, illustrate the wetland areal extent and wetland class distribution in relation to two extreme water level conditions in 1966 (low) and 1985 (high). Most evident during the low water period was the larger overall wetland area, the expanse of exposed substrate, and the expanded area in fen, meadow marsh and emergent marsh. During the high water level period, total wetland area decreased. Emergent marsh and fen were the dominant communities but all wetland classes declined in area. The average size of the class polygons (mean patch size) for meadow marsh and fen classes decreased with rising water levels and increased with declining water levels. Overall the lowest mean patch size values corresponded to high water levels with higher mean patch sizes during low water levels. This suggests a more fragmented wetland landscape in high water years and larger expanses of wetland classes during low water years.
At Oliphant, total wetland area and selected wetland classes responded to water level conditions over time (see Figure 3a). Lake, floating emergent, meadow marsh and exposed substrate classes followed the patterns conceptualized for the lower Great Lakes. Lake and floating emergent expanded and meadow marsh and substrate area contracted with high water levels and the opposite occurred with low water levels. Fen did not exhibit a consistent pattern related to water level change. In the 1966 low water level case (Figure 3b), total wetland area increased. Exposed substrate covered a large area, and meadow marsh (the dry wetland class), emergent marsh and fen were dominant wetland vegetation classes. During high water in 1985 (Figure 3c), total wetland area decreased due to flooding. Lake and floating emergent classes expanded. Emergent marsh remained a key component of the wetland landscape; it did not change markedly in areal extent but changed in location. The meadow marsh class contracted and exposed substrate disappeared. Mean patch size for meadow marsh, fen and exposed substrate classes decreased as water levels rose, while mean patch size for these classes increased with lower water levels. With low water levels, these wetland classes became larger, contiguous expanses.
Exposed substrate and lake classes responded to water level changes with large changes in area while the less well-developed floating emergent, emergent marsh, and meadow marsh classes also responded but their overall contribution to the wetland area was small (Figure 4a, Figure 4b, Figure 4c). The exposed substrate class expanded with lower water levels and contributed greatly to the overall increase in wetland area yet it is not quality habitat. The key is what vegetation communities colonize the exposed area. The area of fen in Howdenvale remained fairly stable throughout all water level conditions comprising 5.4% to 6.3% of the total landscape. Emergent marsh showed a modest increase in area with low levels. Meadow marsh was not an important class in the Howdenvale. Mean patch size decreased for exposed substrate, emergent marsh and meadow marsh with high water levels and increased as water levels declined. Also, aggregation of patches (exposed substrate, emergent marsh and meadow marsh) in the landscape decreased as water levels rose and increased with low water levels. Both of these metrics indicate more fragmentation in the landscape as water levels increased.
Discussion and Summary
This is the only study that maps and analyses historical changes in area and wetland classes of fens in Lake Huron. Total wetland area of the three Lake Huron wetlands responded to water level fluctuations - increasing with low water levels and decreasing with high water levels (Table 2). This response is also typical of coastal lacustrine-type wetlands in the lower Great Lakes. As expected, area of the lake class increased with high water levels and decreased with low water levels. Exposed substrate class emerged or expanded during low water level conditions and contracted or disappeared during high water levels. Although it contributed to overall wetland area, exposed substrate does not contribute especially to wetland habitat quality; but it represents potential wetland area that can be recolonized from seed banks.
Meadow marsh was the only wetland vegetation community that exhibited consistent responses to high and low water levels across all three wetlands (Table 2). As a drier wetland vegetation community, meadow marsh expanded during low water levels and decreased during high levels in the fens. This was similar to responses in the lower Great Lakes wetlands. The emergent marsh wetland class, located in shallow water, was expected to expand in area with lower water levels and contract with higher levels, which occurred in Baie du Dore and Howdenvale. At Oliphant, the emergent marsh class area decreased over time. The floating emergent wetland community in the lower Great Lakes typically decreases during low water levels because of drying and increases during high water levels due to flooding and creation of standing water habitat. Oliphant and Howdenvale exhibited this pattern of change.
The fen class did not exhibit consistent response to water level fluctuations across all three wetlands. These coastal wetlands are located in gently sloping coastal terrain (elevation changes of less than two metres) and are exposed to lake conditions, although the Ontario Wetland Evaluations indicate that regional groundwater input may be important. The fen class in Howdenvale was the dominant wetland class and remained relatively stable in location and areal extent irrespective of lake level changes. This suggests less lake influence and the important role of regional groundwater input. The Howdenvale fens are located further inland away from the shore and a large lake, Sky Lake, located up slope may be an important source of groundwater particularly during low water level periods. Baie du Dore's fen area increased with low water levels and decreased with high water levels. A core, inland area of fen remained, but fen located near the shoreline was influenced by the lake. In this zone it expanded and contracted with water level changes. At Oliphant, fen response to water level change did not exhibit a consistent pattern. During low water levels other wetland communities colonized the exposed areas while fen area did not change. With high water levels, the lake flooded the wetland communities and also affected the fen (particularly those along the northern portion of the wetland complex close to the shoreline). Human influence is high in the Oliphant wetland area. In particular, roads such as Shoreline Avenue, which runs along the inland edge of the wetland, may act as a barrier to vegetation colonization and water movement.
The purpose of this research was to use aerial photos and GIS to document the response of three Lake Huron fens, in terms of changes in total area and wetland classes, over time, while relating these changes to water levels. The variable response to water level change in these fen communities raises further research questions that are beyond the scope of this paper. For example, since fens are reliant on groundwater discharge, what is the role of regional groundwater hydrology in mitigating the effects of changing water levels, particularly in low water years? To what extent does groundwater hydrology influence the vegetation response that was observed in the fens? To what extent does human disturbance, such as nearby roads, influence ground water movement and vegetation responses?
Looking beyond the fen class, the response to changing water levels of the other wetland classes in these Lake Huron wetlands is similar in some ways to observations on the lower Great Lakes. For example, the increase in exposed substrate and decrease in the fragmentation of vegetation classes is associated with declining water levels. Also, it is important to note that this study corroborates other findings that a decline in water levels does not result in a decrease in overall wetland area. The observations for these Lake Huron fens reflect the variable responses of wetland vegetation classes to changing water levels (Hebb et al., 2006) seen elsewhere. This variability raises important challenges when developing models of wetland vegetation class dynamics. Some success has been achieved in modelling wetland class responses to water levels by determining vegetation classes based on rules regarding water depth and duration of inundation and/or exposure (Hebb, 2003; Sabila, 2005; Mortsch et al., 2006b). The unique conditions (geomorphology, topography, groundwater hydrology, and human influences) at each wetland highlight the data base and modelling challenges facing this endeavour. In addition, the limited availability of air photos and the long lag between some photos makes it difficult to account for events that influence wetland vegetation response between air photo years such as the patterns of rising or falling water level conditions. However, new remote sensing technologies present opportunities for improving the data bases upon which model rules are developed and simulations are based. Improving the temporal resolution of wetland observations through the acquisition of high-resolution satellite imagery will allow us to more thoroughly analyse and understand vegetation class responses to changing water levels and the influence of human activities in specific case studies. Improving the accuracy, spatial resolution, and temporal resolution of digital elevation models (DEMs) for wetland areas through the use of LiDAR technology will also help to better understand the responses of wetland vegetation communities to changing water levels. More accurate DEMs will provide an important foundation for wetland analysis and model development.
Through air photo interpretation and GIS-base spatio-temporal analysis we demonstrated that three Lake Huron coastal fens responded to different lake level conditions over time. Indicators such as the total wetland area, the predominance of selected wetland classes, and the size of individual wetland class units (mean patch size) were used to document the responses. Key findings included:
– Total wetland area increased with low water levels. With high water levels, total wetland area decreased.
– During low water level conditions there was an expansion of the “drier” wetland community meadow marsh as well as an increase in exposed substrate. Exposed substrate per se does not have high habitat value. However, over time these exposed reaches can be recolonized through germination from seed banks and increase wetland area.
– Inundation of shoreline areas, during high water levels, resulted in a decrease in area of most wetland classes except lake and some cases the floating emergent class.
– Wetland classes became larger, contiguous expanses with lower water levels while the wetland landscape became more fragmented with a smaller mean size of wetland classes during high water years.
While many questions remain, this study has shown how Lake Huron wetlands containing fens compare in their responses to changing water levels with wetlands in other parts of the Great Lakes.
Funding from the Government of Canada's Great Lakes 2000 Program allowed aerial photograph interpretation of the Lake Huron fens, and the Climate Change Impacts and Adaptation Program supported scan-vectorization and GIS analysis of the wetland class data. The invaluable GIS expertise and graphic assistance of Andrea Hebb is gratefully acknowledged.