165 14 Evaluation of Rapid Assessment Techniques for Establishing Wetland Condition on a Watershed Scale Vanessa L. Lougheed, Christian A. Parker, and R. Jan Stevenson 14.1 INTRODUCTION Recently, the U.S. National Research Council (2001) recommended utilizing a watershed perspective together with science-based, rapid assessment procedures to track wetland mitigation and restoration. Rapid assessment tools can be used as a warning sign to give a quick idea of wetland condition and determine sites in need of further assessment or immediate protection. Many U.S. states have or are developing three-tiered assessment procedures that include an initial landscape-scale assess- ment using aerial imagery (tier 1), followed by a rapid condition assessment (tier 2), and a more intensive monitoring program (tier 3) (e.g., Miller and Gunsalus 1999, Mack 2001, Fennessy et al. 2004). Wetlands can be signicantly impacted by a variety of physical, chemical, and biological factors, and although a single environmental factor can sometimes be implicated as the primary stressor to a wetland ecosystem (King and Richardson 2003), it is more likely that a combination of factors result in wetland degradation on a landscape scale (Danielson 2001). Furthermore, spatial and temporal variability in chemical stressor levels can make it difcult to diagnose one specic nutrient causing impairment, especially for sites sampled just once in landscape-scale assessments. In such cases, multistressor axes can be used to ensure assessments reect a greater number of stressors (e.g., Mack 2001, Lougheed et al. 2001). In particular, one encounters a variety of wetland classes in a single watershed (e.g., lacustrine, riverine, and isolated wetlands) and these different classes may respond differently to a variety of stressors (Fennessy et al. 2004). Multistressor axes may therefore have a greater utility for a suite of wet- lands in a landscape setting than does any one individual measure. Existing rapid assessment methods generally combine various measures of hydrology, water quality, soils, landscape setting, and vegetation (Fennessy et al. © 2008 by Taylor & Francis Group, LLC 166 Wetland and Water Resource Modeling and Assessment 2004). Fennessy et al. (2004) reviewed 16 different rapid assessment methods that met 4 criteria they deemed to be important for successful rapid assessment. They concluded that the best methods should: 1. describe the condition along a single continuum ranging from least to most impacted 2. provide an accurate assessment of conditions in a relatively short time period (e.g., 1 day total for both eld and lab components) 3. include an onsite assessment 4. be capable of onsite verication using more comprehensive ecological assessment data (tier 3) Using these guidelines, the goal of this study was to develop a suite of rapid assessment techniques and examine their utility in evaluating wetland condition in a single large watershed in Michigan. In particular: We compare a eld-based estimate of riparian land use to actual land use val- ues determined from GIS (geographic information system) maps in a 1-km buffer around each wetland. We create a multimetric wetland disturbance axis (WDA) that incorporates rapid measures of hydrology, water quality, and land use. As a rapid assessment of biological condition, we compare an estimate of epi- phytic algal thickness against epiphytic chlorophyll biomass values and percent cover of epiphytic macroalgae. To verify the utility of the WDA in reecting biological condition, we determine whether plant community composition responds along the WDA. 14.2 METHODS The Muskegon River drains a 7,000-km 2 watershed that ows into Lake Michigan on its eastern shore and is dominated by forested land in the upstream regions and agricultural and small urban areas (e.g., Muskegon, population 40,000) in the down- stream region. We visited 85 wetlands in the Muskegon River watershed (MRW in Michigan) during the summers of 2001 through 2003. This included 35 isolated depressions, 25 lacustrine and 25 riverine wetlands. Fifty-two (52) sites were selected randomly based on a numbered grid overlaid on GIS-based wetland maps, while the remaining 33 sites were purposely selected to represent a gradient of disturbance. Approximately half (18) of the randomly selected wetlands were outside the MRW and in the upstream reaches of immediately adjacent watersheds (e.g., Chippewa River, Grand River, Pere Marquette River). For determination of water chemistry, water was collected from an open water area in each wetland in 250-mL, acid-washed bottles. Total phosphorus (TP), total nitrogen (TN), nitrate + nitrite (NOx), ammonia (NH 3 ), silica (Si), soluble reac- tive phosphorus (SRP), and chloride (Cl) were determined using standard methods (American Public Health Association [APHA] 1998) on a Skalar auto-analyzer. Conductivity was measured in the eld using a YSI 556 multiprobe. Sediment was collected from 3 random locations in the wetland using a 5-cm corer; the 3 samples © 2008 by Taylor & Francis Group, LLC Evaluation of Rapid Assessment Techniques 167 were combined and frozen until analysis. C:N was determined using a Perkin-Elmer 2400 Series II CHN analyzer, while percent organic matter was determined follow- ing loss-on-ignition at 500°C. We did not measure contaminant levels in this study; however, local public health departments had identied several areas with contami- nated sediments at a level of concern and these were noted. We constructed a multimetric stressor axis designed to integrate and give equal weight to measurements in 3 primary stressor categories: land use, hydrological modication, and water quality. Unlike many other rapid assessment methods (see Fennessy et al. 2004), we did not include plant habitat variables, as we felt that this would create circular relationships with our plant community metrics. This wetland disturbance axis (WDA) included 3 metrics indicative of land use and land cover change (riparian land use, buffer width, distance to nearest wetland), 2 metrics indicative of hydrology (hydrological modication, water source), as well as 2 water quality metrics (conductivity, contaminants) (Table 14.1). Some of these metrics were loosely based on those used in the Ohio Rapid Assessment Method (ORAM) (Mack 2001), while new metrics were also included to reect different data collection methods in this study. We assigned scores to some of the metrics by placing the “answers” to assessment questions into different categories and then assigning a score by category (Fennessy et al. 2004). For example, hydrological modication was categorized using questions such as: Are there roads along the wetland edge? Is there evidence of dams, dredging, or ditching? Then, each hydro- logical stressor answer was assigned a score, which was summed to achieve a metric indicative of all hydrological modications. Most metrics were scaled using a 1-3-5 scaling system, where a value of 0 or 1 was given to the least impacted wetlands and a value of 5 was given to the most degraded sites. For example, average buffer width around wetlands was categorized in the eld in 4 categories (0 = >50 m; 1 = 25–50 m; 3 = 10–25 m; 5 = <10 m). Similarly, water source was characterized as year round (0), intermittent (3), or none visible (5), and contaminants were classi- ed as none (0), low levels (3), or level of concern (5). In the eld, riparian land use was categorized as either agricultural, fallow pasture, urban, suburban, parkland, or forested on a scale from 0 to 4 (sum total of all categories = 4). For inclusion in the WDA, the proportion (out of 4) for each of these land use categories was mul- tiplied by 5 (for high-impact land categories such as urban and agricultural land), by 3 (for moderate land use impacts such as fallow pasture, park, and suburban residential), whereas forested land was multiplied by zero. Two metrics (nearest neighbor, conductivity) were scaled based on the frequency distribution of values observed for all wetlands in this study. One of these, conductivity, was scaled from 0 to 10 in order to increase the weight of this metric in the overall WDA calcula- tion. Finally, all individual scores from each metric were added together. Although the maximum WDA in this study was 75, the WDA was scaled from 0 to 100, to allow its use in more degraded watersheds in the region. Low value of the WDA indicate higher-quality wetlands. Land use and distance between wetlands were determined in ESRI ArcMap (ver- sion 9.0) using land use maps current to 1998. Using these data, we determined lin- ear distance to the nearest wetland (nearest neighbor), as well as riparian land use in a 1-km buffer around each wetland. Nearest neighbor is the only metric included in © 2008 by Taylor & Francis Group, LLC 168 Wetland and Water Resource Modeling and Assessment TABLE 14.1 Description of metrics used in the wetland disturbance axis (WDA). Sum of all metrics is 45, but is scaled out of 100 to get final WDA. Score and range of values MAX Land use and habitat fragmentation (MAX: 15) Average buffer width (score 1 value only) 0: >50 m 1: 25–50 m 3: 10–25 m 5: <10 m 5 Surrounding land use (calculate and add) 0: multiply 0x proportion forested land 3: multiply 3x sum of proportion park, fallow pasture, and suburban residential land 5: multiply 5x sum of proportion urban, industrial, and agricultural land 5 Nearest neighbor a (score 1 value only) 0: <0.13 km 1: 0.13–0.33 km 2: 0.33–0.66 km 3: 0.66–0.92 km 4: 0.92–1.64 km 5: >1.64 km 5 Hydrology (MAX: 15) Water source (score 1 value only) 0: year-round inputs (river, lake, groundwater) 3: seasonally intermittent 5: no visible inputs 5 Hydrological modication (add all visible modications together to maximum of 10) 0: none 1: road along less than 1/4 of wetland edge 1: human dam (pre-1980) 3: human dams (post-1980) or natural dams (beaver, clogged culvert) 3: road along >1/4 of wetland edge 5: high impact (ditching, dredging, culverts) 10 Water quality (MAX: 15) Conductivity a (score 1 value only) 0: <85 μS/cm 2: 85–159 μS/cm 4: 159–289 μS/cm 6: 289–386 μS/cm 8: 386–498 μS/cm 10: >498 μS/cm 10 Contaminants (score 1 value only) 0: None 3: Present at low levels 5: Level of concern 5 a Ranges included in metric based on frequency distribution. © 2008 by Taylor & Francis Group, LLC Evaluation of Rapid Assessment Techniques 169 the WDA that was not estimated in the eld; however, it may be possible to estimate this variable more rapidly using aerial photos or topographic maps if GIS is not available. Macrophyte and epiphytic algae communities were surveyed using a stratied random design. We established 3 regularly spaced parallel transects, perpendicular to the shore, and randomly placed 1-m 2 rectangular quadrats along each transect according to a random numbers table. In each quadrat, we recorded relative cover of each plant species using a modied Braun-Blanquet scale, estimated the percent cover of lamentous macroalgae, and classied epiphyte thickness on a semiquan- titative scale (rapid epiphyton survey [RES]: 0 = no growth; 1 = thin lm, tracks can be drawn with your ngernail; 2 = 1 to 5 mm; 3 = >5 mm). These were visual estimations of epiphytic thickness, and did not represent precise measurements. Epi- phytic algae were collected from cuttings of the dominant vegetation type in each wetland selected from random locations along each transect; we avoided collecting plants with macroalgal growth. Algae was removed from the plants with a com- bination of gentle rubbing from emergent stems and shaking of submerged plant stems in distilled water. Cleaned plants were placed in zipper bags and refrigerated so that surface area could be determined using image analysis software (ImageJ, NIH). Subsamples of the resulting algal suspension were frozen and analyzed for chlorophyll-a within 2 months of collection. Chlorophyll-a was extracted with 90% ethanol for 24 hours in the dark at 4°C; samples were then sonicated for 15 minutes and chlorophyll uorescence determined on a Turner Designs uorometer. Chloro- phyll concentration was expressed per surface area of plant. Results presented are not corrected for phaeophytin because our RES could not distinguish between live and dead epiphytes. We selected the Floristic Quality Assessment Index (FQAI) for Michigan (Her- man et al. 2001) and its related coefcient of conservatism (CofC) to describe the wetland condition represented by the plant communities. The FQAI indicates the extent to which the community is dominated by sensitive wetland plants. The CofC is the sensitivity value given to each plant and we used the average CofC calculated for all plant species in each wetland. To explain structure in the biological communi- ties of the wetlands, independent of any preconceived environmental preferences or gradients, we used nonmetric multidimensional scaling (NMDS). NMDS analysis identies axes that describe biologically meaningful, multivariate gradients in the community data (McCune and Grace 2002). We selected the Bray-Curtis distance measure and used the rst NMDS axis identied by PC-ORD (version 4.10) as an indicator of plant community structure. The NMDS, FQAI, and CofC were deter- mined from previous analyses (Lougheed et al. 2007) to respond strongly to envi- ronmental gradients in the MRW. Relationships between the rapid assessment variables and more detailed mea- surements of land use and epiphytic chlorophyll-a were studied in the large dataset of 85 wetlands, regardless of wetland class. In studying the responses of the plant communities, we divided the data into wetland classes (depressions, lacustrine, riv- erine) because biological communities in differing classes may respond uniquely to differing stressors. © 2008 by Taylor & Francis Group, LLC 170 Wetland and Water Resource Modeling and Assessment 14.3 RESULTS Actual land use in a 1-km buffer around each wetland was well represented by the estimated land use categories (Figure 14.1); however, our estimates of land use more accurately reected urban and agricultural land use. For both these land use types, we were able to distinguish among 3 separate categories and the 0 category had an average of 4% developed land in both cases. Our measurements of forested land dif- fered between the lowest (0 and 1) and highest categories (3 and 4); however, the 0 Agriculture Category 0 Percent Agriculture 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 Forest Category Percent Forested Urban Category Percent Urban 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 A AA A AB AB B B B C C C BC 1 2 3 01 2 34 0 123 FIGURE 14.1 Comparison of GIS-calculated percent land use values determined in a 1-km buffer around each wetland in 4 to 5 land use categories estimated in the eld. Letters indi- cate statistical similarities (Tukey multiple comparisons; p < 0.05). © 2008 by Taylor & Francis Group, LLC Evaluation of Rapid Assessment Techniques 171 category had an average of 26% forested land, which was not signicantly different from category 1 at 33% forested land. We took the average rapid epiphyton survey (RES) values from each wetland and rounded the value up to the nearest 0.5. Epiphytic chlorophyll-a was signi- cantly different between sites with a “thin lm” (category 1) of algae, relative to sites with approximately 1 to 5 mm of growth (category 2) (Figure 14.2). There was no signicant increase in category 3, likely because it included sites with increased macroalgal cover, which we excluded from our epiphyte samples. This is supported by comparisons of macroalgal cover, expressed as relative dominance of macroalgal cover (per m 2 ) relative to total plant species cover (per m 2 ), which was signicantly higher in sites with an average RES value of 3. We used principal components analysis to determine which rapid assessment metrics explained the greatest amount of variation in the dataset. The rst 3 PCA axes together accounted for 68% of the variation among sites. The rst principal component (PC1) explained 34% of the variation in the dataset, and was most highly Rapid Epiphyton Survey 1.0 1.5 2.0 2.5 3.0 1.0 1.5 2.0 2.5 3.0 0 200 400 600 800 1000 1200 1400 Rapid Epiph y ton Surve y Macroalgal Dominance 0.00 0.05 0.10 0.15 0.20 0.25 A AA AB AB AB ABB B B FIGURE 14.2 Comparison of epiphytic chlorophyll-a biomass (top) and macroalgal domi- nance (bottom) in 5 rapid epiphyton survey (RES) categories estimated in the eld. Letters indicate statistical similarities (Tukey multiple comparisons; p < 0.05). © 2008 by Taylor & Francis Group, LLC Epiphyton CHL ( g/cm ) + 2 172 Wetland and Water Resource Modeling and Assessment correlated with land use and fragmentation variables (buffer width, r = 0.82; ripar- ian land use, r = 0.83, nearest neighbor, r = 0.53) and water conductivity (r = 0.64); all other metrics were also signicantly correlated with this axis, but at much lower levels (r = 0.24–0.36). The second axis (PC2; 18% of variation in dataset) was most highly correlated with hydrological variables (modication, r = 0.73; water source, r = 0.60), as well as a negative correlation with nearest neighbor (r = −0.59). The third axis (PC3; 16% of variation in dataset) was most highly correlated with con- taminants (r = 0.80). There was no signicant difference in the location of different wetland classes along PC1; however, PC2 values were signicantly higher in depres- sional wetlands, likely because fewer of these had year-round inputs of water (water source) as opposed to all riverine and lacustrine sites. As an indicator of disturbance, the WDA correlated strongly with many mea- sured land use and water chemistry variables (p < 0.05). In particular, it was highly correlated with land use variables (r = 0.54–0.60), water chemistry measures (r = 0.3–0.36), and sediment characteristics (r = 0.23–0.25) (Table 14.2). For subsequent analyses, we separated the data into 3 sections, representing the different wetland classes (depressions, lacustrine, riverine). A signicant amount of the variation in a measure of plant community structure (NMDS) and the extent to which the community was dominated by native, sensitive taxa (FQAI and CofC) could be explained by the WDA (Figure 14.3). The FQAI was strongly correlated with the WDA for riverine sites, whereas the CofC was a better metric for depres- sions and lacustrine wetlands. Overall, these relationships were strongest for depres- sional and lacustrine wetlands, and lower for riverine sites. In many cases, the WDA explained more variation in the biological metrics than did any individual environ- mental variable (Table 14.3); however, forested land explained slightly more of the variation in the NMDS values for depressional wetlands, and variation in riverine plant communities was explained slightly better by TP and conductivity. It is inter- esting to note, however, that using a suite of 120 plant metrics calculated for all TABLE 14.2 Significant correlations between WDA and environmental variables (p < 0.10; Bonferoni corrected). Variable r p % Urban 0.54 0.0000 % Agriculture 0.43 0.0000 % Forest –0.60 0.0000 TP 0.30 0.0061 NO X 0.36 0.0008 SRP 0.21 0.0496 NH 3 0.35 0.0011 Cl 0.71 0.0000 Sediment: %organic –0.23 0.0477 Sediment C:N 0.25 0.0356 © 2008 by Taylor & Francis Group, LLC Evaluation of Rapid Assessment Techniques 173 NMDS axis -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 WDA log CofC 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 WDA WDA 0102 03 04050607080 0102 03 04050607080 0102 03 04050607080 log FQAI 1.20 1.25 1.30 1.35 1.40 1.45 1.50 1.55 Depressional r 2 = 0.46 p<0.001 Lacustrine r 2 = 0.51 p<0.001 Riverine r 2 = 0.21 p = 0.02 Depressional r 2 = 0.38 p<0.001 Lacustrine r 2 = 0.30 p<0.00 1 Riverine r 2 = 0.20 p = 0.03 FIGURE 14.3 Relationship between plant community metrics and the WDA for depressional (left), lacustrine (middle), and riverine (right) wetlands in the MRW. © 2008 by Taylor & Francis Group, LLC 174 Wetland and Water Resource Modeling and Assessment site types, including measures of species richness and plant community composition (Lougheed, unpublished data), more of these metrics were correlated to the WDA (26 metrics; Bonferoni corrected; p<0.05), than the next most commonly correlated environmental variables: developed land (21 metrics), TP (12 metrics), and Cl (5 metrics). 14.4 DISCUSSION This study provides evidence that eld-based estimates of algal cover and land use can accurately reect more detailed measures requiring increased lab processing time and technical skills. In addition, we present the development and verication of a multimetric wetland disturbance axis (WDA) that successfully integrates stressors from 3 categories: land use, hydrological modication, and water quality. The WDA is highly correlated with a variety of land use and water chemistry measures, as well as several measures of plant community composition. Rapid epiphyton assessment can be highly useful because it enables the determi- nation of algal biomass over larger spatial scales than sampling algae off individual substrates followed by lab analysis (Stevenson and Bahls 1999). We provide evidence that an estimate of epiphyte cover using a rapid epiphyton survey can be a good sur- rogate for more detailed measures of epiphytic and macro-algal biomass. Despite its accuracy, both the rapid and more detailed measurements of algal biomass were not correlated to any rapid or detailed measures of wetland condition, including the WDA or nutrient levels. Wetlands are complex environments, where both vascular plants and algae compete for nutrients and light. Measures of diatom community composition (Lougheed et al. 2007) or trophic state indices (e.g., Van Dam et al. 1994) may be more sensitive indicators of algal responses to nutrient enrichment in wetlands than more simple measures of algal biomass. In particular, Lougheed et al. (2007) found that diatom community composition (as indicated by NMDS) was a TABLE 14.3 Significant correlations between biological metrics & environmental variables (p < 0.10; Bonferoni corrected). Depressions Lacustrine Riverine NMDS CofC NMS CofC NMS FQAI WDA 0.68 –0.61 0.77 –0.62 a 0.50 a –0.55 Agriculture 0.61 –0.56 — –0.54 a 0.44 a — Urban 0.39 a — 0.63 –0.41 a —— Forest –0.72 0.55 –0.61 0.55 — — TP 0.36 — 0.501 –0.45 a 0.54 a — Cl 0.50 –0.39 0.62 –0.51 a — –0.52 a COND 0.60 –0.46 0.65 — 0.41 a –0.63 NO X ————0.41 a –0.58 a Not signicant when Bonferoni corrected at p < 0.05. © 2008 by Taylor & Francis Group, LLC [...]... Pijanowski This project was funded by the Great Lakes Fisheries Trust as part of the Muskegon River Initiative REFERENCES American Public Health Association (APHA) 1998 Standard methods for the examination of water and wastewater 20th ed Washington, DC: American Public Health Association Crosbie, B., and P Chow-Fraser 1999 Percentage land use in the watershed determines the water and sediment quality... structure of plants, diatoms, and zooplankton occurred Given these analyses for depressional wetlands in the MRW and nearby watersheds, the WDA rapid assessment tool, which has been verified using more comprehensive biological data, can now be used to categorize additional wetlands in the watershed as well as track the state of wetlands that were identified as needing remedial action In an era of reduced funding... GIS layers are not available or experience using GIS programs is limited; however, this study indicates that riparian land use estimates can be a good approximation of actual riparian land use calculated from GIS layers In this study, our estimates may have underestimated forested land in some cases, likely because many of our wetlands were accessible by roads or tracks and thus may have been biased... wetland watersheds is a highly significant predictor of reduced water quality in wetlands (Crosbie and Chow-Fraser 1999, Lougheed et al 2001), while an increased proportion of forested land, including forested buffer strips along streams (e.g., Crosbie and Chow-Fraser 1999) in wetland watersheds, can be beneficial in improving water quality Land use covers can be time-consuming to determine, especially... 22 marshes in the Great Lakes basin Canadian Journal of Fisheries and Aquatic Science 56:1781–1791 Danielson, T J 2001 Methods for evaluating wetland condition: Introduction to wetland biological assessment EPA 822-R-0 1-0 0 7a Washington, DC: Office of Water, U.S Environmental Protection Agency Fennessy, M S., A D Jacobs, and M E Kentula 2004 Review of rapid methods for assessing wetland condition EPA/620/R-04/009... use and nearest neighbor) and water conductivity As verification of its utility, the WDA was highly correlated with detailed land use and water quality measures, as well as measures of plant community composition (NMDS) and dominance by sensitive plants (CofC, FQAI) In addition, Lougheed et al (2007) showed that the WDA could be used as a rapid assessment tool for categorizing depressional wetlands...Evaluation of Rapid Assessment Techniques 175 highly sensitive measure of disturbance in depressional wetlands Early changes in algal species composition, as opposed to changes in algal biomass, may result from minor changes in nutrient availability and may be a better indicator of alterations in fundamental microbial processes (Stevenson et al 2002) The proportion of agricultural and urban land in wetland. .. immediate protection In addition, we have provided evidence that field-based estimates of algal cover and land use can accurately reflect more detailed measures requiring increased lab processing time and technical skills ACKNOWLEDGMENTS We greatly appreciate field assistance from Mollie McIntosh, Sarah Wolf, Alyson Yagiela, Nicole Behnke, and James Montante Land use shapefiles were provided by Brian... macroinvertebrates and fish, 2nd ed., EPA 841-B-9 9-0 02, ed M T Barbour et al Washington, DC: U.S Environmental Protection Agency; Office of Water Stevenson, R J., P V McCormick, and R Frydenborg 2002 Methods for evaluating wetland condition: Using algae to assess environmental conditions in wetlands, EPA-822-R-02021 Washington, DC: Office of Water, U.S Environmental Protection Agency Van Dam, H., A Mertenes, and. .. Chow-Fraser 2001 Primary determinants of macrophyte community structure in 62 marshes across the Great Lakes basin: Latitude, land use, and water quality effects Canadian Journal of Fisheries and Aquatic Sciences 58:1603–1612 Mack, J J 2001 Ohio rapid assessment method for wetlands, manual for using version 5.0 Ohio EPA Technical Bulletin Wetland/ 200 1-1 -1 Columbus: Ohio Environmental Protection Agency, Division . maps in a 1-km buffer around each wetland. We create a multimetric wetland disturbance axis (WDA) that incorporates rapid measures of hydrology, water quality, and land use. As a rapid assessment. (for high-impact land categories such as urban and agricultural land), by 3 (for moderate land use impacts such as fallow pasture, park, and suburban residential), whereas forested land was multiplied. Francis Group, LLC Epiphyton CHL ( g/cm ) + 2 172 Wetland and Water Resource Modeling and Assessment correlated with land use and fragmentation variables (buffer width, r = 0.82; ripar- ian land