We analyzed macroinvertebrate community composition from 19 sites, spanning 5 vegetation types, in the White River drowned river mouth wetland (eastern shore of Lake Michigan). Sites were distributed along a gradient of anthropogenic disturbance from the less-impacted upper wetland to the more-impacted lower wetland. The relative importance of surrounding land use and cover, water quality, and vegetation type in structuring macroinvertebrate communities was evaluated using correspondence analysis (CA) of biotic data, and principle components analysis (PCA) of chemical/physical water quality data. A significant correlation (p = 0.001) between site scores of the CA and PCA showed that ambient chemical/physical water quality was likely an important driver of macroinvertebrate community composition. Canonical correspondence analysis (CCA) was then used to corroborate this relationship. Vegetation type, did not appear to structure macroinvertebrate communities to a significant degree. Both water quality and macroinvertebrate community composition correlated with surrounding land use and cover. Therefore, we hypothesize that macroinvertebrate communities in the White River drowned river mouth wetland respond to surrounding land uses via its influence on the abiotic environment. This response occurred even at the within-system scale used in this study. Practitioners of bioassessment protocols should, therefore, be aware of spatial heterogeneity in both water quality and macroinvertebrate community composition in Great Lakes drowned river mouth wetlands.

Introduction

Great Lakes coastal wetlands serve as important interfaces between upland and pelagic habitats. They have been shown to be important habitat for waterfowl (Prince et al., 1992; Prince and Flegel, 1995; Whitt, 1996), passerine birds (Harris et al., 1983; Whitt, 1996; Riffell, 2000), fish (Liston and Chubb, 1985; Jude and Pappas, 1992; Brazner, 1997; Uzarski et al., 2005) and invertebrates (Krieger, 1992; Cardinale et al., 1998; Gathman et al., 1999; Burton et al., 1999; Uzarski et al., 2004). Despite their importance, Great Lakes coastal wetlands have suffered extensive degradation and continue to receive developmental pressures. Understanding biotic communities and their response to the abiotic environment is vital to our understanding of coastal wetland structure and function including their role as a buffer to upland constituents reaching the Great Lakes.

Drowned river mouth wetlands form at the riverine/lacustrine interface in the lower reaches of tributary rivers of the Great Lakes (Keough et al., 1999; Albert and Minc, 2001; Albert, 2003). These systems are influenced by fluctuations in Great Lakes water levels but are generally sheltered from the direct force of wave energy by sand dunes or bars (Wilcox et al., 2002; Albert and Minc, 2001). This geomorphology makes drowned river mouth systems subject to both riverine and lacustrine physical processes resulting in highly variable aquatic habitats (i.e., vegetation community composition and organic sediment depth) (Keough et al., 1999; Wilcox et al., 2002; MacKenzie et al., 2004).

Invertebrates in these systems form important links between trophic levels and play key roles in nutrient cycling. Benthic macroinvertebrates are also continually exposed to disturbances of natural and anthropogenic origin and often respond predictably to anthropogenic disturbance. Thus, they are valuable indicators of ecosystem health (Flint, 1979; Reynoldson and Zarull, 1989; Burton et al., 1999; Uzarski et al., 2004). Furthermore, macroinvertebrate community structure can be used to integrate time and space to detect both episodic and cumulative impacts to water quality (Plafkin et al., 1989; Barbour et al., 1992). In recent years the development of indicators of ecosystem health has received considerable attention throughout the Great Lakes community. Currently, invertebrate-based IBIs have been developed and are being tested for use in monitoring Great Lakes coastal wetlands (Burton et al., 1999; Kashian and Burton, 2000; Uzarski et al., 2004).

Detecting differences between natural ecosystem attributes (e.g., water level fluctuation, vegetation type, pelagic mixing) and anthropogenic stressors (e.g., nutrient loading, toxicants, etc.) on a system-scale has been among the greatest hurdles encountered during IBI development (Wilcox et al., 2002). Burton et al. (2002, 2004) found that effective fetch (as a measure of wave exposure) was a significant driver of macroinvertebrate community composition among marshes of Saginaw Bay and northern Lake Huron. They also concluded that plant community composition was important in structuring macroinvertebrate communities along exposure gradients within coastal wetlands. To account for the variability due to these natural drivers, Burton et al. (1999) and Uzarski et al. (2004) developed IBI metrics based on specific vegetation zones. Since vegetation communities generally respond to wave exposure and ice scour in Great Lakes open coastal wetlands (Minc, 1996; Minc and Albert, 1998), their approach accounts for much of this natural variability (the use of vegetation zones also allows the IBI to be used over a range of Great Lakes water levels).

The conceptual models proposed by Burton et al. (2002, 2004) to characterize the effect of vegetation type and wave exposure on coastal wetland invertebrate communities and the IBI of Burton et al. (1999) and Uzarski et al. (2004) focused mainly on open and protected embayments of Lake Huron (although two inland protected wetlands were included in Burton et al., 2002). Since habitat characteristics (i.e., organic sediment depth, vegetation community composition and habitat heterogeneity) vary among coastal wetland classes (Minc and Albert, 1998; Keough et al., 1999), characterizations of invertebrate communities in all types of Great Lakes coastal wetlands are needed to identify the most important natural and anthropogenic factors in structuring community composition. Since most drowned river mouth wetlands in the Great Lakes fall onto the “protected” or “very protected” end of the exposure gradient identified by Burton et al., (2002, 2004), it is unclear whether macroinvertebrate communities in these systems respond to vegetation type to the extent of those in Great Lakes embayment wetlands. Identifying the responses of invertebrate communities to natural and anthropogenic factors in drowned river mouth wetlands will provide information necessary for the further development of effective bioassessment protocols. Furthermore, understanding the response of wetland communities to environmental conditions will provide insight into the structure and function of these coastal ecosystems.

In this study we characterize macroinvertebrate assemblages within a moderately degraded Lake Michigan drowned river mouth wetland typical of other drowned river mouth systems in the region. We then examine the associations between macroinvertebrate communities and dominant vegetation type, water quality, and surrounding land use and cover. Understanding the relative influence of anthropogenic disturbance and habitat characteristics on invertebrate community composition will be valuable in future attempts to utilize macroinvertebrates for determining Great Lakes coastal wetland health and function.

Methods

Study area

The White is a fourth order river that lies on the western shore of the lower peninsula of Michigan. It drains a 1,370 km2 watershed and forms a freshwater estuary where it empties into Lake Michigan via White Lake (Muskegon County, N43.41° W86.35°). The confluence of the White River and White Lake, where the river enters the lake plain, forms a drowned river mouth wetland of approximately 350 ha. The wetland has three diked and drained agricultural areas adjacent to it that are used for row crop production (Figure 1). Runoff from these fields either drains or is pumped into the river. U.S. 31, a four-lane highway built on an earthen levee with a bridged opening over the main river channel, bisects the middle of the wetland. Business route U.S. 31, a two-lane road, also built on an earthen levee with a bridged opening, crosses the lower wetland and links the cities of Whitehall (pop. 3,403) and Montague (pop. 2,422) (1998 U.S. Census) (Figure 1). A closed municipal solid waste landfill is located on the upland area of the southern bank, west of U.S. 31. Remedial investigations at the landfill have reported a plume of contaminated groundwater migrating toward the White River (MDNR, 1987). The municipal wastewater treatment system serving Whitehall and Montage has an effluent discharge point in Silver Creek, east of U.S. 31 (Figure 1). The wastewater receives tertiary treatment by extended aeration and land application prior to discharge. The White River watershed contains 58% forested land, 22% agricultural land, 5% developed land, 5% surface water, and 10% undeveloped open land (MIRIS, 1978 with updates provided by the Information Service Center of the Annis Water Resources Institute-Grand Valley State University). White Lake is a 1040 ha. eutrophic drowned river mouth lake that has considerably degraded water quality resulting from many residential, industrial and municipal pollutants (Freedman et al., 1979; Evans, 1992; Rediske et al., 1998) and is designated as an Area of Concern (AOC) by the International Joint Commission (IJC, 1989). Specific Beneficial Use Impairments (BUIs) associated with the AOC designation include eutrophication, degradation of benthos and fish populations, contaminated sediments, and loss of habitat.

Sampling design

Sampling of the drowned river mouth wetland was conducted at 19 sites from 13 August through 15 August 2001 (Figure 1). The sites were selected to represent a gradient of anthropogenic disturbance, determined a priori from adjacent land use and cover and preliminary limnological parameters, from the relatively pristine upper wetland to the impacted lower wetland. Sampling locations within sites (henceforth referred to as ‘stations’) were chosen based on inundation of vegetation and access by boat. All available monodominant vegetation stands were sampled at every site, yielding 24 stations in total. Five plant community types were identified in the drowned river mouth and stations were classified as Typha- (mostly Typha latifolia L.), Sparganium-, Scirpus- (mostly Scirpus acutus Muhl), Pontederia-(mostly Pontederia cordata L.), or Nuphar and Nymphaea-dominated. All sites had relatively dense vegetation and little, if any, visible current. Depths ranged from 10 cm to 110 cm. To facilitate comparisons of the more pristine habitats of the upper wetland to the more impacted habitats of the lower wetland, we classified sites as either “upper,” “middle” or “lower” (Figure 1). This classification was based on upstream/downstream location within the drowned river mouth. Stations are referred to by their geographic classification (upper, middle or lower), dominant vegetation type, and site number. For example, station Upper-Lily-3 was located in the upper wetland, was dominated by lily (Nuphar or Nymphaea) and was at site #3.

Macroinvertebrate sampling

Samples were collected with a standard 0.5 mm mesh, D-frame dip net. Sampling consisted of sweeping through the water column from the sediment surface to the water surface. Net contents were then placed in a white pan and samples of 150 macroinvertebrates were collected using forceps. Three replicate samples were collected at each station from randomly chosen locations. To avoid biases toward larger and more mobile organisms the pan was divided into small sections (roughly 5 cm squares) and all specimens from one area of the pan were picked before moving on to the next area. While some organisms did have a tendency to move around in the sampling pan, most remained sessile enough to allow us to pick each area clean before moving on to the next. For more information on this sampling technique see Burton et al. (1999) and Uzarski et al. (2004).

Specimens were sorted to lowest operational taxonomic unit in the laboratory; this was usually family or genus for most insects, crustaceans, and gastropods. Some invertebrate groups including Oligochaetae, Hirudinea and Turbellaria, were identified to order level or, in a few cases, to class. Taxonomic keys such as Thorp and Covich (1991), Merritt and Cummins (1996), Pennak (1989), and Burch (1982) were used for identification.

Chemical/physical parameters

Basic chemical/physical parameters were collected in conjunction with each set of macroinvertebrate samples. Water samples for analysis of soluble reactive phosphorus (SRP), nitrate-N, ammonium-N, sulfate-S, chloride and total alkalinity were collected in 1 liter acid-washed polyethylene bottles. Temperature, dissolved oxygen (DO), %saturation of dissolved oxygen (%DO), chlorophyll a, oxidation-reduction potential, total dissolved solids, turbidity, pH and specific conductance were measured in situ using a HydroLab DataSonde 4a (calibrated 1 week prior to sampling following protocols recommended by the manufacturer). Sample collection and in situ measurements were taken at mid-depth prior to macroinvertebrate sampling. Analytical procedures and quality assurance/quality control procedures followed protocols recommended by APHA (1998).

Land use and cover parameters

Land use and cover parameters were calculated for a 1 km circular buffer around each site. Land use and cover data were obtained from the Michigan Resource Information System (MIRIS, 1978) with updates and ground-truthing conducted by the Information Service Center of the Annis Water Resources Institute-Grand Valley State University. Seven land use and cover parameters were calculated for each of the 19 sites including %agriculture, %barren field, %developed land, %forest, %wetland+open water and total road density (m km− 2). Arcview version 3.3 was used to calculate all land use and cover parameters within the buffer zones.

Statistical analysis

We conducted indirect gradient analyses on the chemical/physical, land use and cover, and macroinvertebrate datasets. Principal components analysis (PCA) was used to identify the most important gradients in the chemical/physical and surrounding land use and cover datasets. We then decomposed each principal component (PC) to determine which parameters they were most strongly associated with. We used correspondence analysis (CA) to identify the most important gradients in macroinvertebrate community composition. The gradients identified with each of these techniques were then compared to determine how macroinvertebrate community composition related to chemical/physical water quality and surrounding land use. Canonical correspondence analysis (CCA), a direct gradient analysis, was then used to corroborate this relationship (Ter Braak, 1986).

We conducted one PCA on the correlation matrix of seven land use and cover proportions and one PCA on the correlation matrix of 10 chemical/physical variables. Soluble reactive phosphorus and ammonium-N were excluded because a majority of stations had concentrations below analytical detection limits, and temperature was excluded because of its correlation with time of day. For the two stations that had nitrate-N concentrations below detection limit (< 0.01 mg l−1) we substituted values of half detection limit. Chemical/physical data were log-transformed and land use and cover proportions were arcsine-square root-transformed prior to PCA. To decompose PCs we used Pearson correlation between PC scores and individual parameters.

One-way ANOVAs were used to identify differences in chemical/physical, land use and cover, community indices (Shannon diversity, Pielou's evenness, and percent insects), and relative abundances of common taxa among wetland regions and vegetation types. Tukey's HSD tests were conducted post-hoc to identify specific differences. Pearson correlation was used to identify relationships between community indices and relative abundances of common taxa with individual chemical/physical parameters and PC scores. Differences and correlations were deemed significant at p < 0.05 without correction for multiple comparisons. Therefore, we recognize that marginally significant relationships could be due to chance alone. For most analyses mean taxon relative abundances (means of the three replicate samples from each station) were used as a measure of central tendency. However, diversity indices were calculated for each replicate sample then averaged. Macroinvertebrate relative abundances, percent insects, and land use and cover proportions were arcsine-square root-transformed prior to analysis to satisfy assumptions of normality.

Correspondence analysis was conducted on raw mean counts (of triplicate samples) of the 47 most abundant macroinvertebrate taxa as well as log- and square root-transformed datasets. The 47 taxa included those represented by 7 or more organisms or 0.05% of the total macroinvertebrates collected. Transformation of macroinvertebrate data did not increase the amount of variation explained in the first two CA dimensions or our interpretation of the resulting gradients. Thus, we elected to use the CA of untransformed mean counts for our analyses (Hair et al., 1998). To relate community composition to chemical/physical condition a Pearson correlation was calculated between CA dimension 1 scores and site scores from the first two chemical/physical PCs. Combining these two indirect gradient analyses allowed us to model the apparent macroinvertebrate community response to the abiotic environment without forcing the community gradient to ordinate based on our measured abiotic variables. Canonical correspondence analysis was conducted on mean counts of the 47 most abundant macroinvertebrate taxa and 10 chemical/physical variables (log-transformed) (Ter Braak, 1986).

Principal components and correspondence analyses were conducted using SAS version 8.0 (Cary, North Carolina). Canonical correspondence analysis was conducted using Canoco for Windows version 4.02 (Wageningen, The Netherlands). Tukey's HSD, ANOVA, and Pearson correlations were conducted with SYSTAT version 8.0 (Evanston, Illinois).

Results

Land use and cover

Principal components analysis of seven land use and cover parameters separated sites of the upper, middle and lower wetland (Figure 2). Principal component 1 explained 70.9% of the variability in the land use and cover data while PC 2 explained an additional 18.4%. Upper wetland sites plotted in the same direction as the %forest and %barren field eigenvectors. Middle wetland sites plotted in the same direction as the %agriculture and %wetland eigenvectors. Sites of the lower wetland plotted in the same direction as the eigenvectors for %open water, road density and %developed land. Significant differences between upper, middle and lower wetland sites were found for nearly all land use and cover parameters (p < 0.05, Tukey's HSD). An exception was between middle and upper wetland sites for %developed land (p = 0.92).

Chemical/physical conditions

Principal components analysis of 10 chemical/physical variables separated stations of the upper wetland from the lower wetland (Figure 3). Principal component 1 correlated positively with %DO (p < 0.001, r = 0.820), specific conductance (p = 0.010, r = 0.514), and pH (p < 0.001, r = 0.758), and negatively with chloride (p < 0.001, r = −0.806) and alkalinity (p < 0.001, r = −0.855). Principal component 2 correlated positively with sulfate-S (p < 0.001, r = 0.736), nitrate-N (p<0.001, r = 0.854), and oxidation-reduction potential (p = 0.001, r = 0.643). In the first two PCs (explaining 52.7% of the variability collectively), seven of the eight upper wetland stations were separated from the lower stations. Middle wetland stations showed no pattern in the first two PCs and were placed throughout the area occupied by upper and lower stations. The PCA showed most upper stations to be associated with higher %DO and pH (Figure 3). However, these parameters were not significantly different among wetland regions when means were compared using ANOVA.

While most upper wetland stations scored high in PC 1 and relatively low in PC 2, Upper-Lily-3 was the exception with a higher PC 2 score than the other upper wetland stations (Figure 3). Nitrate-N concentrations and turbidity at Upper-Lily-3 were higher than those of any other upper wetland station (0.16 mg l−1 and 38.1 NTU, respectively). Chloride concentration and specific conductance at Upper-Lily-3 were also slightly higher than that of the other upper stations (Table 1). Upper-Lily-2 and Upper-Lily-15 also had elevated nitrate-N concentrations relative to the other upper wetland stations.

In general, most lower wetland stations and half of the middle wetland stations were pulled away from upper stations in PC 1 and/or PC 2. Lower-Lily-7 and Middle-Lily-18 were the only stations with DO concentrations below 5 mg l−1 and these two stations also had the lowest PC 1 scores. Concentrations of chloride, SRP, and ammonium-N at Middle-Lily-18 were higher than those of any other station sampled. Middle-Typha-11, Middle-Scirpus-12 and Lower-Typha-13 scored highest in PC 2 because of their elevated nitrate-N concentrations; all being greater than 0.30 mg l−1.

The PCA was also used to search for patterns in water quality based on plant community type. However, our interpretation of the PCA based on vegetation type suffers from a lack of comparable vegetation zones throughout the wetland (no inundated Typha stands were found in the upper wetland and no inundated Pontederia or Sparganium stands were found in the lower wetland).

Macroinvertebrates

Ninety-nine macroinvertebrate taxa representing four phyla and eight classes were collected. Seventy-eight of the 99 taxa were insects representing nine orders. The 10 most-abundant taxa in the wetland made up 79.9% of the total collected and are listed in Table 2. Taxon richness ranged from 17 to 48 with a mean of 29.3± 1.3 (mean± standard error) (Table 2). Shannon diversity indices ranged from 0.33± 0.11 to 1.18± 0.01 and evenness values ranged from 0.35 ± 0.09 to 0.83 ± 0.01 (Table 2). Percent of macroinvertebrate assemblage as insects ranged from 15.9 ± 2.4 to 86.1 ± 3.3 (Table 2). No significant differences were found among upper, middle and lower stations or among vegetation types for taxon richness, evenness, Shannon diversity, or percent of assemblage as insects (p > 0.05, ANOVA). However, Shannon diversity and evenness were negatively correlated with PC 2 scores from the chemical/physical PCA (Shannon diversity: p = 0.016, r = −0.452 and evenness: p = 0.027, r − 0.452). Shannon diversity and evenness were also negatively correlated with nitrate-N concentration (Shannon diversity: p = 0.010, r = −0.517 and evenness: p = 0.011, r = −0.511).

Dimension 1 of the CA explained 23.7% of the variability in the invertebrate data and represents a gradient of stations from those of the upper wetland to those of the lower (Figure 4a). In dimension 1, upper and lower stations were completely separated while middle stations were spread throughout the area occupied by the upper and lower stations.

Gammarus was the dominant macroinvertebrate in the wetland. It comprised 19.8% of all organisms collected and was the most abundant taxon at eight of the 24 stations. No significant differences in Gammarus abundance was found between the upper, middle and lower wetland stations or between the different vegetation types (p > 0.05, ANOVA) (Table 2). In dimension 1 of the CA, Gammarus plotted in the range where upper and lower wetland stations converge, indicating its ubiquity in the wetland (Fig. 4b). Gammarus abundances did, however, correlate positively with nitrate-N (p = 0.006, r = 0.541), sulfate-S (p = 0.039, r = 0.424), and PC 2 (p = 0.009, r = 0.519).

Corixidae was the second most abundant taxon, representing 16.2% of all organisms collected. Corixidae plotted among the lower wetland stations in CA dimension 1 (Figure 4b) and Corixidae abundances were significantly greater in the lower wetland than the middle (p = 0.001, Tukey's HSD) and upper (p < 0.001, Tukey's HSD) wetland (Table 2). Corixidae was the dominant taxon at six of the seven lower wetland stations. Corixidae abundances correlated positively and significantly with nitrate-N (p = 0.005, r = 0.557), sulfate-S (p = 0.002, r = 0.597), and chemical/physical PC 2 (p = 0.004, r = 0.561). Significantly more corixids were found at Typha-dominated stations than either Lily- or Sparganium-dominated (p = 0.013 and 0.007, respectively, Tukey's HSD).

The location of Hyallela azteca in dimension 1 of the CA reflected the importance of this species in the upper wetland (Figure 4b). Hyallela azteca was among the two most abundant taxa at four of the eight upper wetland stations and two of the nine middle wetland stations (Table 2) and was the third most abundant macroinvertebrate in the wetland, comprising 6.4% of the organisms collected. Hyallela azteca was not found in large numbers at any lower wetland station (Table 2). Relative abundances of this taxon were negatively correlated with nitrate-N (p = 0.046, r = −0.411), sulfate-S (p = 0.003, r = −0.577) and chemical/physical PC 2 (p = 0.019, r = −0.473). However, no significant difference among vegetation types or regions was detected for Hyallela azteca (p > 0.05, ANOVA).

Coenagrionidae was the fourth most abundant taxon in the wetland, comprising 6.1% of the macroinvertebrates collected. Coenagrionidae appeared important to Upper-Lily-1, Upper-Sparganium-1, Upper-Scirpus-1, Upper-Pontedaria-1, and Middle-Lily-4 by its location in dimension 1 of the CA (Fig. 4b) and the relative abundances of Coenagrionidae found at these stations (Table 2). Coenagrionidae relative abundances were not significantly different among upper, middle and lower wetland stations or among vegetation types (p > 0.05, ANOVA). Coenagrionidae relative abundances were negatively correlated with nitrate-N (p = 0.003, r = −0.579), chemical/physical PC 2 (p = 0.001, r = −0.625), and redox potential (p = 0.004, r = −0.566), and positively correlated with %DO (p = 0.033, r = 0.436).

Caecidotea was the fifth most abundant taxon found, comprising 5.6% of the macroinvertebrates collected and was plotted among the upper wetland stations in dimension 1 of the CA (Figure 4b). Caecidotea relative abundances were negatively correlated with nitrate-N (p = 0.030, r = −0.444), sulfate-S (p = 0.033, r = −0.436), specific conductance (p = 0.031, r = −0.444), and chemical/physical PC 2 (p = 0.033, r = −0.437). Significantly more Caecidotea were found at upper stations than lower stations (p = 0.025, Tukey's HSD). Also, significantly more Caecidotea were found in Pontederia-dominated than Typha-dominated stations (p = 0.045, Tukey's HSD).

Mesoveliidae, Chironomini, and Neoplea, were also among the ten most abundant taxa in the wetland (Table 2) and the placement of these taxa in the middle of CA dimension 1 shows their cosmopolitan distribution (Figure 4b). No significant differences in relative abundances of any of these taxa were found among vegetation types or wetland regions (p > 0.05, ANOVA) and only one significant correlation (Mesoveliidae:nitrate-N) was found between relative abundances of these taxa and chemical/physical parameters. Turbellaria, on the other hand, was plotted among upper wetland stations in dimension 1 (Figure 4b). Turbellaria relative abundances were negatively correlated with sulfate-S (p = 0.006, r = −0.541), and chemical/physical PC 2 (p = 0.024, r = −0.458) and positively correlated with chloride (p=0.016, r=0.485), and SRP (p = 0.003, r = 0.575). No significant differences in Turbellaria relative abundances were found among wetland regions or vegetation types (p > 0.05, ANOVA).

Physidae was the tenth most abundant taxon in the wetland and appeared to be important in the lower wetland by its position in CA dimension 1 (Figure 4b). Physidae relative abundances were positively correlated with sulfate-S (p = 0.019, r = 0.476). Significantly more physids were found in the lower wetland than the upper (p = 0.005, Tukey's HSD) and no significant differences in Physidae relative abundances were found among vegetation types (p > 0.05, ANOVA).

Corduliidae contributed strongly to dimension 1 of the CA (Figure 4b). Corduliids, however, were only found at Middle-Lily-4, Middle-Sparganium-4, Middle-Lily-17 and Upper-Lily-3. Where it did occur, this taxon was not found in large numbers (relative abundance ≤ 3.0%).

Correspondence analysis was also used to search for broad patterns in macroinvertebrate communities based on vegetation type. The four Typha-dominated stations were placed within a fairly narrow range in dimension 1 while Lily-dominated stations formed the largest group and had the greatest range in both dimensions. Pontederia-, Scirpus- and Sparganium-dominated stations formed groups that overlapped almost entirely (Figure 4a).

A significant positive correlation was found between CA dimension 1 scores and chemical/physical PC 2 scores (p = 0.001, r = 0.635) (Figure 5). Significant positive correlations were also found between CA dimension 1 scores and nitrate-N (p = 0.001, r = 0.652) and sulfate-S (p = 0.002, r = 0.601). Dimension 1 scores were negatively correlated with pH (p = 0.036, r = −0.429) and %DO (p = 0.034, r = −0.434).

Canonical correspondence analysis revealed gradients similar to those found with PCA (chemical/physical data) and CA (macroinvertebrate data) (Figure 6). The first CCA axis explained 18.5% of the variance in macroinvertebrate data and 33.0% of the variance in the macroinvertebrate-environment relationship (eigenvalue = 0.350). The second CCA axis explained 12.1% of the variance in macroinvertebrate data and 21.5% of the variance in the macroinvertebrate-environment relationship (eigenvalue = 0.228). In general, upper and lower wetland stations separated along CCA axis 1.

Discussion

The PCA of land use and cover variables indicated a strong association between wetland region and adjacent land use and cover (Figure 2). As shown by both the PCA and Tukey's HSD tests, the upper wetland could be characterized by its adjacent forested land while the middle wetland could be characterized by its adjacent agricultural land and the lower wetland by its adjacent developed land. Therefore, our a priori designation of sites (either “upper,” “middle” or “lower”) seemed appropriate for testing the effects of surrounding land use and cover at a coarse scale.

Water quality, though quite variable, coincided with differences in adjacent land use and cover. Lower wetland stations had relatively degraded water quality, most likely due to the combined effects of adjacent developed land (cities of Whitehall and Montague), White Lake (an AOC), and the agricultural land located immediately upstream. The PCA of chemical/physical parameters (Figure 3) revealed that Lower-Typha-13 could be characterized by its relatively high nitrate-N concentration while most of the other lower stations had either elevated chloride concentration and/or lower dissolved oxygen. The lower wetland also had more stations with detectable ammonium-N concentrations than either the middle or upper wetland. The upper wetland appeared less impacted by anthropogenic disturbance and most upper wetland stations could be characterized by their lower concentrations of dissolved nutrients and higher dissolved oxygen and pH. We have found that chemical/physical parameters in Great Lakes drowned river mouth wetlands are often highly variable due to heterogeneous biological and hydrological conditions (i.e., water depth, organic sediment accumulation, P:R ratios, etc.) (Uzarski et al., 2005 and unpublished data). Thus, PCA was an important tool for identifying the meaningful patterns (through linear combinations of variables) in chemical/physical conditions of the White River drowned river mouth wetland.

While chemical/physical conditions seemed closely associated with adjacent land use and cover, they did not appear to correspond with vegetation type. One exception, however, was that three out of the four Typha-dominated stations had relatively high nitrate-N concentrations (> 0.30 mg l−1). While Typha seemed more common in the more-degraded lower wetland, we did not census vegetation and cannot say that the proliferation of Typha correlated significantly with the chemical/physical conditions found there. Our inability to locate and sample Typha-dominated stations throughout the White River drowned river mouth wetland represents a limitation of our study to conclusively partition the effects of vegetation and water quality on macroinvertebrate communities.

Macroinvertebrate communities in the White River drowned river mouth wetland appeared to be structured by both chemical/physical condition (which corresponded to adjacent land use and cover) and vegetation type. The relationship between invertebrate community composition and vegetation, however, was less apparent than previous work has shown for other Great Lakes coastal wetland types. For example, Burton et al. (2002, 2004) showed that vegetation type was among the most important factors in predicting macroinvertebrate community structure in embayment wetlands of Saginaw Bay and northern Lakes Huron and Michigan. These authors hypothesized that gradients in vegetation type also represent gradients in the chemical and physical environment (i.e., turbidity, dissolved oxygen, etc.) that further contribute to the structure of invertebrate communities in these systems.

In the White River drowned river mouth wetland we found that abundances of eight of the ten most common taxa did not correspond to vegetation type. Relationships between the remaining two taxa and vegetation type also corresponded to differences in water quality. Corixidae abundances were greater at Typha-dominated stations than both lily- or Sparganium-dominated stations. However, Typha-dominated stations were only found in areas with degraded water quality. Caecidotea abundances were significantly higher in Pontederia- than in Typha-dominated stations. However, water quality was less degraded in Pontederia- than in Typha-dominated stations. Therefore, the effect of vegetation structure was either masked by the over-riding influence of anthropogenic disturbance, or had an affect on community composition too minimal to be detected with our techniques.

The predominance of Gammarus, Hyallela azteca and Caecidotea in the White River drowned river mouth wetland is consistent with findings in other Great Lakes coastal wetlands. For example, in their study of macroinvertebrate communities of the Peshtigo River wetland, a flooded delta of Green Bay, Lake Michigan, MacKenzie et al. (2004) found Asellus sp. (Isopoda) and Gammarus to be among the most abundant taxa. Burton et al. (2004) found Gammarus, Hyallela azteca and Caecidotea to be characteristic of protected or very protected embayment wetlands of northern Lakes Michigan and Huron. These protected and very protected embayment wetlands would be similar to drowned river mouth wetlands in terms of their limited hydrologic mixing and relatively deep organic sediment accumulation.

Burton et al. (2004) found that Corixidae preferred exposed, wave swept wetlands. The high Corixidae abundance in the White River drowned river mouth wetland suggests that either this taxon can utilize a variety of habitats or that we collected different genera or species than those of Burton et al. (2004). Corixids occurred in greater abundances at stations of the lower and middle wetland and correlated well with elevated nitrate-N. In the upper wetland, Corixids were only found in substantial numbers at the Silver Creek station (Upper-Lily-3) where the municipal wastewater effluent discharge resulted in water quality similar to the lower wetland (elevated turbidity, nitrate-N and ammonium-N). Stations that had higher Corixidae abundances also seemed to have deeper organic sediments. While organic sediment depth was not included in our analyses, most lower wetland stations had over a half-meter of soft organic sediments while most upper and middle wetland stations had less accumulated organic sediment (personal observation). This habitat characteristic may be important in structuring community composition in Great Lakes coastal wetlands and should be included in future analyses.

Physidae abundances also appeared to be dictated by anthropogenic disturbance. Physids were found at all of the lower stations, but in the upper wetland, they were found only at Upper-Lily-2, Upper-Lily-3 and Upper-Pontederia-14. Station Upper-Pontederia-14 was immediately downstream of where Silver Creek (sewage effluent) empties into the main channel and had relatively high turbidity, most likely from Silver Creek. Upper-Lily-2 contained physids and had relatively high nitrate-N (though we could not identify the source of the nutrient). Upper-Lily-1, Upper-Pontederia-1, Upper-Scirpus-1, Upper-Sparganium-1 and Upper-Lily-15 had slightly better water quality and had no physids. Middle-Lily-4 had the lowest nitrate-N, chloride and turbidity of all middle wetland stations and had non-detectable ammonium-N and SRP. This was the only middle station without physids. Physids have been shown by others to be tolerant of pollution in Great Lakes coastal wetlands (Kashian and Burton, 2000). This tolerance may explain the higher abundance of this taxon at relatively degraded locations in the White River drowned river mouth wetland.

Stations that had the highest Hyallela azteca abundances were those that had the least anthropogenic disturbance. Most of the upper wetland stations as well as Middle-Lily-4, Middle-Sparganium-4 and Middle-Lily-17 had high abundances of Hyallela azteca and relatively low turbidity, sulfate-S, nitrate-N, ammonium-N, chloride and SRP. Hyallela azteca represented substantially less of the macroinvertebrate community at lower wetland stations and at stations of the middle wetland with degraded water quality.

The response of macroinvertebrate communities to abiotic conditions in the White River drowned river mouth wetland is best demonstrated by the relationship between CA dimension 1 scores and chemical/physical PC 2 (Figure 5). PC 2 represents increasing nitrate-N, sulfate-S, and oxidation-reduction potential while CA dimension 1 represents a shift in communities from Coenagrionidae-, Hyallela azteca-, and Caecidotea-dominated to Corixidae-dominated. Patterns revealed by canonical correspondence analysis agreed with these relationships. Negative correlation coefficients between Hyallela azteca, Coenagrionidae and Caecidotea relative abundances and chemical/physical disturbance parameters and positive correlation coefficients between Corixidae abundances and disturbance parameters further support these findings.

A number of taxa were cosmopolitan among our sampling stations. Gammarus, for instance, was the most abundant taxon in the wetland, was plotted in the center of CA dimension 1 and was common at nearly every station in the wetland (Table 2). Gammarus are known to be generalists, proliferating in areas with accumulated organic sediments. The weak association between Gammarus and degraded water quality suggests that this taxon is tolerant of disturbance (within the limits found in this wetland) and that the deeper organic sediments of the lower wetland may be its preferred habitat. The cosmopolitan nature of Chironomini, Mesoveliidae, and Neopleidae may also be an indication of the wide tolerance that these taxa have to moderately degraded water quality.

Macroinvertebrate community composition of the middle wetland stations was the most variable of the three regions yet corresponded predictably to water quality. Middle-Scirpus-12, Middle-Typha-11 and Middle-Sparganium-16, had high nitrate-N concentrations probably due to their proximity to agricultural fields. Macroinvertebrate communities at these three stations were similar to lower wetland stations and were characterized by their high abundance of Corixidae and low abundance of Hyallela azteca. Middle-Lily-4, Middle-Sparganium-19, Midddle-Sparganium-4 and Middle-Lily-17 were low in nutrients and had high pH and dissolved oxygen concentrations, making them more similar to the upper wetland stations in terms of water quality. Macroinvertebrate communities at these four middle stations were also similar to those of the upper wetland (low Corixidae abundance, high Hyallela azteca and Coenagrionidae abundances).

Conclusions

Water quality in the White River drowned river mouth wetland corresponded to adjacent land use and cover. In turn, macroinvertebrate community composition appeared to be structured more by water quality than by vegetation type. Macroinvertebrate communities in areas adjacent to agricultural and/or developed land tended to have higher abundances of Corixidae and Physidae whereas communities adjacent to forested land tended to have higher abundances of Coenagrionidae, Hyallela azteca, and Caecidotea.

The link between invertebrate community composition and anthropogenic disturbance among systems is becoming well established for Great Lakes coastal wetlands. The influence of vegetation type in structuring macroinvertebrate communities has also been established for Great Lakes embayment wetlands. Macroinvertebrate community composition did not appear to be strongly associated with vegetation type in the White River drowned river mouth wetland. We recommend that future studies investigate this relationship further by using a factorial approach to partition the affect of vegetation structure from that of water quality. We also recommend that future studies explore the relationship between faunal community composition, organic sediment accumulation, and hydrology in Great Lakes coastal wetlands. The current study demonstrates the relationship between macroinvertebrate community composition and chemical/physical condition, mediated by anthropogenic practices on the adjacent landscape, which can occur within a single coastal wetland system.

Acknowledgements

This work was supported with funds from the Alcoa Foundation and the Muskegon County Community Foundation. We thank Beau Braymer, Rochelle Heyboer, Scott Mueller, John Genet, Shawn Wessell, and Kelly Martin for field and laboratory assistance. We thank Rod Denning for assisting with land use and cover measurements. We thank Adam Bosch, Keto Gyekis, Aaron Parker and Michael Shoemaker for their assistance in preparing this manuscript. Drs. L.C. Grapentine, P. Chow-Fraser, and T.F. Nalepa, as well as two anonymous reviewers, greatly improved earlier versions of this manuscript.

The text of this article is only available as a PDF.

References

Albert, D. A.
2003
.
Between land and lake: Michigan's great lakes coastal wetlands
,
East Lansing, MI
:
Michigan Natural Features Inventory, Michigan State Univ.
.
Extension bulletin E-2902
Albert, D. A. and Minc, L. D.
2001
.
Abiotic and floristic characterization of Laurentian Great Lakes coastal wetlands
.
Verh. Internat. Verein. Limnol.
,
27
:
3413
3419
.
APHA (American Public Health Association)
.
1998
.
Standard Methods for the Evaluation of Water and Wastewater,
, 20th edition,
Washington, DC
:
APHA
.
Barbour, M. T., Plafkin, J. L., Bradley, B. P., Graves, C. G. and Wisseman, R. W.
1992
.
Evaluation of EPA's rapid bioassessment benthic metrics: Metric redundancy and variability among reference stream sites
.
Environ. Toxicol. Chem.
,
11
:
437
449
.
Brazner, J. C.
1997
.
Regional, habitat, and human development influences on coastal wetland and beach fish assemblages in Green Bay, Lake Michigan
.
J. Great Lakes Res.
,
23
:
36
51
.
Burch, J. B.
1982
.
North American freshwater snails
.
Walkerana
,
1
:
81
215
.
Burton, T. M., Uzarski, D. G. and Genet, J. A.
2004
.
Invertebrate habitat use in relation to fetch and plant zonation in northern Lake Huron coastal wetlands
.
Aquatic Ecosystem Health and Management
,
7
:
249
267
.
Burton, T. M., Stricker, C. A. and Uzarski, D. G.
2002
.
Effects of plant community composition and exposure to wave action on invertebrate habitat use of Lake Huron coastal wetlands
.
Lakes and Reservoirs: Research and Management
,
7
:
255
269
.
Burton, T. M., Uzarski, D. G., Gathman, J. P., Genet, J. A., Keas, B. E. and Stricker, C. A.
1999
.
Development of a preliminary invertebrate index of biotic integrity for Lake Huron Coastal wetlands
.
Wetlands
,
19
:
869
882
.
Cardinale, B. J., Brady, V. J. and Burton, T. M.
1998
.
Changes in the abundance and diversity of coastal wetland fauna from the open water/macrophyte edge towards shore
.
Wetland Ecol. Manage.
,
6
:
59
68
.
Evans, E. D.
1992
.
Mona, Muskegon, and White Lakes in Muskegon County, Michigan: the 1950s to the 1980s
,
Michigan Department of Natural Resources
.
MI/DNR/SWQ-92/261
Flint, R. W.
1979
.
Responses of freshwater benthos to open-lake dredge spoils in Lake Erie
.
J. Great Lakes Res.
,
5
:
264
275
.
Freedman, P. L., Canale, R. P. and Auer, M. T.
1979
.
Applicability of land treatment in the Great Lakes basin area impact of wastewater diversion, spray irrigation on water quality in the Muskegon County, Michigan Lakes
Washington, DC
U.S. EPA. EPA-905/9-79-006
Gathman, J. P., Burton, T. M. and Armitage, B. H.
1999
. “
Coastal wetlands of the upper Great Lakes: Distribution of invertebrate communities in response to environmental variation
”. In
Invertebrates in Freshwater Wetlands of North America
, Edited by: Batzer, D. P., Rader, R. B. and Wissinger, S. A.
949
994
.
New York, NY
:
John Wiley and Sons
.
Hair, J. F., Anderson, R. E., Tatham, R. L. and Black, W. C.
1998
.
Multivariate Data Analysis,
, 5th edition,
Upper Saddle River, NJ
:
Prentice Hall
.
Harris, H. J., Milligan, M. S. and Fewless, G. A.
1983
.
Diversity: Quantification and ecological evaluation in freshwater marshes
.
Biol. Conserv.
,
27
:
99
110
.
International Joint Commission.
1989
.
Revised water quality agreement of 1978
,
United States and Canada
:
International Joint Commission
.
Jude, D. J. and Pappas, J.
1992
.
Fish utilization of Great Lakes wetlands
.
J. Great Lakes Res.
,
18
:
651
672
.
Kashian, D. R. and Burton, T. M.
2000
.
A comparison of macroinvertebrates of two Great Lakes coastal wetlands: Testing potential metrics for an index of ecological integrity
.
J. Great Lakes Res.
,
26
:
460
481
.
Keough, J. R., Thompson, T. A., Guntenspergen, G. R. and Wilcox, D. A.
1999
.
Hydrogeomorphic factors and ecosystem responses in coastal wetlands of the Great Lakes
.
Wetlands
,
19
:
821
834
.
Krieger, K. A.
1992
.
The ecology of invertebrates in Great Lakes coastal wetlands: Current knowledge and research needs
.
J. Great Lakes Res.
,
18
:
634
650
.
Liston, C. R. and Chubb, S.
Relationships of water level fluctuations and fish
.
Coastal Wetlands: Proceedings of the First Great Lakes Coastal Wetlands Colloquium
. Edited by: Prince, H. H. and D'Itri, F. M. pp.
121
140
.
Chelsea, MI
:
Lewis Publishers
.
MacKenzie, R. A., Kaster, J. L. and Klump, J. V.
2004
.
The ecological patterns of benthic invertebrates in a Great Lakes coastal wetland
.
J. Great Lakes Res.
,
30
:
58
69
.
MDNR.
1987
.
Remedial action plan for White Lake
,
Michigan
:
Michigan Department of Natural Resources, Surface Water Quality Division
.
Merrit, R. W. and Cummins, K. W., eds.
1996
.
An Introduction to the Aquatic Insects of North America
,
Dubuque, IA
:
Kendall/Hunt Publishing Co.
.
Minc, L. D.
1996
.
Michigan's Great Lakes coastal wetlands, Part 1: an overview of abiotic variability
,
Chicago, IL
:
U.S. EPA Great Lakes National Program
.
Minc, L. D. and Albert, D. A.
1998
.
Great Lakes coastal wetlands: abiotic and floristic characterization
,
East Lansing, MI
:
Michigan Natural Features Inventory, Michigan State Univ.
.
MIRIS.
1978
.
Michigan Resource Information System
Pennak, R. W.
1989
.
Freshwater Invertebrates of the United States: Protozoa to Mollusca,
, 3rd edition,
New York, NY
:
John Wiley and Sons
.
Plafkin, J. L., Barbour, M. T., Porter, K. D., Gross, S. K. and Hughes, R. M.
1989
.
Rapid bioassessment protocols for use in streams and rivers: benthic macroinvertebrates and fish
,
Washington, DC
:
Office of Water, US Environmental Protection Agency
.
EPA/440/4-89-001
Prince, H. H. and Flegel, C. S.
1995
. “
Breeding avifauna of Lake Huron
”. In
The Lake Huron Ecosystems: Ecology, Fisheries and Management
, Edited by: Munawar, M., Edsall, T. and Leach, J.
247
272
.
Amsterdam
:
SPB Academic Publishing
.
Prince, H. H., Padding, P. I. and Knapton, R. W.
1992
.
Waterfowl use of the Laurentian Great Lakes
.
J. Great Lakes Res.
,
18
:
673
699
.
Rediske, R. R., Fahnensteil, G., Nalepa, T., Meier, P. and Schelske, C.
1998
.
Preliminary investigation of the extent and effects of sediment contamination in White Lake
Michigan
U.S. EPA-905-R-98-004
Reynoldson, T. B. and Zarull, M. A.
1989
. “
The biological assessment of contaminated sediments-the Detroit River example
”. In
Environmental Bioassay Techniques and Their Application
, Edited by: Munawar, M., Dixon, G., Mayfield, C. I., Reynoldson, T. and Sadar, M. H.
463
476
.
Belgium
:
Kluwer Academic Publishers
.
Riffell, S. K.
2000
.
The effects of landscape context and coastline complexity on birds in Great Lakes coastal wet meadows
,
Ph.D. dissertation
East Lansing, MI
:
Michigan State Univ.
.
Ter Braak, C. J. F.
1986
.
Canonical correspondence analysis: a new eigenvector technique for multivariate direct gradient analysis
.
Ecology.
,
67
:
1167
1179
.
Thorp, J. H. and Covich, A. P., eds.
1991
.
Ecology and classification of North American freshwater invertebrates
,
San Diego, CA
:
Academic Press Inc.
.
Uzarski, D. G., Burton, T. M., Cooper, M. J., Ingram, J. W. and Timmermans, S. T. A.
2005
.
Fish habitat use within and across wetland classes in coastal wetlands of the five Great Lakes: development of a fish-based index of biotic integrity
.
J. Great Lakes Res.
,
31
:
171
187
.
Uzarski, D. G., Burton, T. M. and Genet, J. A.
2004
.
Validation and performance of an invertebrate index of biotic integrity for Lakes Huron and Michigan fringing wetlands during a period of lake level decline
.
Aquatic Ecosystem Health and Management
,
7
(
2
):
269
288
.
Whitt, M. B.
1996
.
Avian breeding use of coastal wetlands on the Saginaw Bay of Lake Huron
,
M.S. thesis
East Lansing, MI
:
Michigan State Univ.
.
Wilcox, D. A., Meeker, J. E., Hudson, P. L., Armitage, B. J., Black, M. G. and Uzarski, D. G.
2002
.
Hydrologic variability and the application of index of biotic integrity metrics to wetlands: a Great Lakes evaluation
.
Wetlands
,
22
:
588
615
.