Aquatic Botany 104 (2013) 111–120
Contents lists available at SciVerse ScienceDirect
Aquatic Botany journal homepage: www.elsevier.com/locate/aquabot
Water chemistry distinguishes wetland plant communities of the Great Lakes coast Carol A. Johnston a,∗ , Terry N. Brown b a b
Department of Natural Resource Management, Box 2104A, South Dakota State University, Brookings, SD 57007, USA Natural Resources Research Institute, 5013 Miller Trunk Highway, University of Minnesota-Duluth, Duluth, MN 55811-1442, USA
a r t i c l e
i n f o
Article history: Received 20 March 2012 Received in revised form 17 August 2012 Accepted 20 August 2012 Available online 1 September 2012 Key words: Wetland Plant community Water chemistry Phosphorus Chloride Conductivity pH Great Lakes coast
a b s t r a c t U.S. coastal wetlands of the Laurentian Great Lakes span a north-to-south gradient (latitude 47–41◦ N) of increasing human population and agricultural intensity that alters their water chemistry and vegetation. We related field-measured water chemistry to vegetation condition and composition using data from 48 freshwater coastal wetlands along this sub-continental gradient, building upon previous findings that GIS-derived landscape descriptors could adequately predict vegetation condition in Great Lakes coastal wetlands. Our aim was to determine (1) whether plant communities could be differentiated by their surface water chemistry, and (2) if water chemistry could better predict wetland vegetation condition than GIS-derived variables. Seven distinct plant communities were identified by agglomerative hierarchical clustering and non-metric multidimensional scaling of vegetation cover data: Sphagnumcarpeted poor fens, Sparganium eurycarpum marshes, Calamagrostis canadensis wetlands, Schoenoplectus pungens marshes, Phragmites australis marshes, and two floristically distinct Typha-dominated marshes. There were significant differences (ANOVA) among the seven plant assemblages identified for most water chemistry metrics (Cl− , chlorophyll a, conductivity, NO3 -N, pH, total N, total P, total suspended solids), but dissolved oxygen, dissolved organic carbon, and NH4 -N did not vary significantly across the assemblages. The two different Typha-dominated plant communities were chemically distinct from each other in chlorophyll a, conductivity, NO3 -N, pH and total suspended solids concentrations, and we recommend that they be separated into distinct associations: Typha spp. – Thelypteris palustris – Hydrocharis morsus-ranae and Typha spp. – C. canadensis – Leersia oryzoides. Plant communities tended to be geographically clustered, but wetlands that were geographic outliers of their floristic type were chemically similar to other wetlands in their plant community grouping despite being on different lakes. When offered both GIS-derived and field-measured potential predictor variables, a regression tree model of wetland condition chose only GIS-derived variables. However, a classification tree model derived solely from field-measured water chemistry variables correctly classified 79% of the sites into four plant community groupings based on total N, conductivity, and pH. Grouping wetlands by plant communities could provide a scientifically-defensible basis for stricter water quality standards to protect sensitive wetland types. © 2012 Elsevier B.V. All rights reserved.
1. Introduction Lake-edge plant communities are known indicators of watershed stressors, such that their vegetation composition can often be predicted by GIS-derived anthropogenic correlates (Houlahan et al., 2006; Sass et al., 2010). In a previous study of 90 wetlands along the Laurentian Great Lakes coast, we showed that 82% of the variance in wetland vegetation data was explained by five GIS-derived variables, most of them related to watershed anthropogenic stress
∗ Corresponding author. Tel.: +1 605 688 6464; fax: +1 605 688 6677. E-mail addresses:
[email protected] (C.A. Johnston),
[email protected] (T.N. Brown). 0304-3770/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.aquabot.2012.08.005
(Johnston et al., 2009b). We conjectured that “waterborne nutrient and/or sediment inputs to the wetlands are undoubtedly the real causes of degradation, rather than land use per se.” In this paper, we test that idea with water chemistry data for a subset of the wetlands originally studied. The use of biota to determine ecological status is of worldwide interest. The ECOFRAME system tested by the European Water Framework Directive uses biological measures as indicators of water quality so as to better reflect the ecological consequences of pollutant discharges (Moss et al., 2003). Biological assessment using the presence, condition, and variety of aquatic life is commonly used by the U.S. Environmental Protection Agency and Environment Canada. Aquatic macrophytes are increasingly being used to assess the health of shallow lakes (Toivonen and Huttunen,
112
C.A. Johnston, T.N. Brown / Aquatic Botany 104 (2013) 111–120
1995; Heegaard et al., 2001; Declerck et al., 2005; Penning et al., 2008; del Pozo et al., 2010; Sass et al., 2010) and wetlands (Murphy, 2002; King and Richardson, 2003; Rolon and Maltchik, 2006; Craft et al., 2007; Croft and Chow-Fraser, 2007). Relating vegetation characteristics to water chemistry could strengthen the use of plants as biological indicators by identifying more mechanistic relationships with stressors. This analysis utilized three datasets that we and others collected for the Great Lakes Environmental Indicators project (USEPA, 2006) (1) a vegetation dataset collected at 90 coastal wetlands by botanists (Johnston et al., 2009b), (2) a water chemistry dataset collected at 100 coastal wetlands by scientists studying fish and diatoms (Reavie et al., 2006; Trebitz et al., 2007; Morrice et al., 2008), and (3) GIS-derived data summarized for the entire U.S. Great Lakes basin (Danz et al., 2007; Johnston et al., 2009a). We added no new data, but performed new statistical analyses to link the vegetation data with the water chemistry data at the 48 sites that were visited in common by the two field groups. The purpose of this paper is to determine (1) if plant communities of Great Lakes coastal wetlands can be differentiated by their surface water chemistry, and (2) if water chemistry is a better predictor of wetland vegetation condition than GIS-derived variables. The chemistry data included metrics commonly used in water quality studies, some more relevant to vegetation than others. Based on previous studies, we expected the plant nutrients nitrogen and phosphorus to influence vegetation communities (Olde Venterink et al., 2003; Edvardsen and Økland, 2006; Verhoeven et al., 2006; del Pozo et al., 2010). Phytoplankton chlorophyll a (Chl a) was also expected to differ across wetland plant communities because of its close relationship to nutrient availability (Moss et al., 2003). Wetland vegetation is sensitive to pH (Sjörs, 1950; Vitt and Chee, 1990; Nicolet et al., 2004), and certain wetland plants (e.g., Sphagnum moss) release organic acids that acidify their environment (van Breemen, 1995). Total suspended solids (TSS) were expected to indicate sediment delivery to wetlands, which can alter wetland vegetation by affecting germination and species composition (Johnston et al., 2001; Werner and Zedler, 2002; Mahaney et al., 2005). Decomposition of wetland plant detritus could consume dissolved oxygen (DO) and/or augment dissolved organic carbon (DOC) (Arrigoni et al., 2008; Tuchman et al., 2009). Chloride and conductivity were also in the suite of variables measured, but were not expected to be related to wetland vegetation. Although the field sites are U.S. coastal wetlands of the Laurentian Great Lakes, the results are applicable to a wide spectrum of freshwater wetlands due to the variety of wetland plant communities and water chemistry conditions encountered. The Great Lakes coastline spans 16.2◦ longitude by 5.5◦ latitude, encompassing diverse climate, geology, population density, and land use.
2. Methods 2.1. Study sites and sampling Study sites were initially selected to represent the range of anthropogenic stresses impacting Great Lakes coastal wetlands (Danz et al., 2007). The Great Lakes span a north-to-south gradient of increasing human population and agricultural intensity that alters their water chemistry: wetlands in Lakes Superior and Huron have the lowest nutrient levels and clearest water, whereas wetlands in Lake Erie have the highest nutrient levels and the lowest water clarity (Lougheed et al., 2001; Morrice et al., 2008). Floristic condition also varies along this latitudinal gradient, with exotic or invasive plants substantially more prevalent in wetlands of the southern Great Lakes (i.e., Lakes Erie and Ontario) than those in the north (i.e. Lakes Superior and Huron) (Johnston et al., 2010). The
Fig. 1. Location of study sites, classified by plant community hierarchical cluster. FEN = northern poor fen, BLU = bluejoint, BUR = burreed, REE = reed, 3SQ = threesquare rush, WCA = western cattail, ECA = eastern cattail. Sites identified with 4-digit numbers are discussed in the text.
forty-eight wetlands (“sites”) in which both vegetation data and water chemistry data were collected are well-distributed across the region (Fig. 1). Sampling details are provided by the parent studies (Reavie et al., 2006; Danz et al., 2007; Trebitz et al., 2007; Morrice et al., 2008; Johnston et al., 2009b), but are summarized below. Vegetation sampling focused on herbaceous emergent wetland plants growing in shallow coastal waters that could be accessed by canoe or wading, usually <1 m deep, and associated submergent aquatic plants. Sampling was conducted by visual observation of plant taxa in 1 m × 1 m plots distributed along randomly placed transects. All vascular plant species and large non-vascular plants such as Chara and Sphagnum were identified to the lowest taxonomic division possible by trained botanists who were tested annually to ensure consistency of visual observations. Plants were identified using published taxonomic manuals, and the Interagency Taxonomic Information System (http://www.itis.gov/) was used to provide consistent taxonomic identities which can be easily compared with other past and future projects. Of the 138 taxa used, 120 were identified to species and 18 were identified to genus (see Appendix). Percent cover was estimated visually for each taxon according to modified Braun–Blanquet cover class ranges: <1%, 1 to <5%, 5 to <25%, 25 to <50%, 50 to <75%, 75 to 100%. The following taxa were considered to be invasive: Hydrocharis morsus-ranae, Lythrum salicaria, Myriophyllum spicatum, Phalaris arundinacea, Phragmites australis, Stuckenia pectinata, Typha angustifolia and its hybrid Typha × glauca. Due to their frequent co-occurrence and genetic relatedness, T. angustifolia and Typha × glauca were grouped into a single category, “invasive Typha.” Typha latifolia, a native noninvasive species, was identified and treated separately. Water chemistry data were collected by staff from the University of Minnesota Duluth and the U.S. Environmental Protection Agency-Mid-Continent Ecology Division (Reavie et al., 2006; Trebitz et al., 2007; Morrice et al., 2008). Hydrolab or YSI multimeter sensors were used for in situ measurements of dissolved oxygen (DO), conductivity (COND), and pH. Water was collected and composited from three to six subsites within each wetland, and transported on ice for laboratory analysis within 24 h for: total nitrogen (TN), total phosphorus (TP), ammonium nitrogen (NH4 N), nitrite + nitrate nitrogen (hereafter referred to collectively as NO3 -N), total suspended solids (TSS), chlorophyll a (Chl a), chloride (Cl− ), and dissolved organic carbon (DOC). Laboratory analysis followed standard methods for low-level detection (USEPA, 1983, 1991, 1993; APHA, 2000). Quality assurance/quality control procedures followed a Quality Assurance Project Plan (QAPP) submitted to EPA that followed EPA Guidance for Quality Assurance. Twenty-four environmental variables derived from existing geospatial data sources were summarized for each site, all
C.A. Johnston, T.N. Brown / Aquatic Botany 104 (2013) 111–120
113
computed for our prior publications (see Table 1 in Johnston et al., 2009b). These included several integrated measures of watershed anthropogenic stress, derived by principal components analysis of multiple stressors of a common anthropogenic origin (Danz et al., 2007): agriculture (PC1 AG), human population and development (PC1 URB), atmospheric deposition (PC1 ATDEP), point source pollution (PC1 NPDES), and the cumulative stress index (CSI), which was a generalized stress gradient derived from the preceding four principal components and land cover. Each wetland was characterized by its latitude (LAT), growing degree days (GDD), and the Great Lake on which it occurred (LAKE). Areal fraction of land uses within the catchments draining into each wetland and in surrounding concentric buffers (100 m, 500 m, 1000 m, 5000 m) were computed for row crops (RC), forest (FOR), development (DEV), and wetland and inland water cover adjacent to the target wetland site (WETW).
2.2. Statistical analysis Plant community analyses were conducted with PRIMER version 6.1.13 (Clarke and Gorley, 2006). The data for all plots within a site were pooled to compute average site cover by species. Square root transformation was done to downweight high abundance species, and a resemblance matrix was computed from Bray–Curtis similarity values. Plant communities were classified by agglomerative hierarchical clustering with group-average linking based on the Bray–Curtis similarities, using the SIMPER procedure to determine species contributions to the average similarity within each cluster and the SIMPROF procedure to verify that the communities identified were statistically distinct (Clarke and Gorley, 2006). Non-metric multidimensional scaling (NMDS) was used with the Bray–Curtis similarity data to ordinate sites by their vegetation, using 25 restarts and a minimum stress of 0.01. Plant communities identified were compared with associations (Jennings et al., 2009) described by the U.S. National Vegetation Classification (http://usnvc.org/explore-classification/). Water chemistry variables were examined using summary statistics and normality probability plots. All water chemistry variables except pH were log10 -transformed to meet normality assumptions, and means reported are back-transformed geometric means. Mean pH was calculated as arithmetic mean of pH values (equal to the negative log of the geometric mean of [H+ ]). One-way analysis of variance (ANOVA) and boxplots were used on pH and log10 -transformed values. Significant differences (P < 0.05) among plant communities and among lakes were assessed with ANOVA a posteriori comparisons using Tukey’s HSD test. The R software, version 2.13.1, was used for these univariate statistics (The R Foundation for Statistical Computing, Vienna, Austria, http://www.r-project.org/). Multivariate analysis of water chemistry was performed using PRIMER. A resemblance matrix was constructed for water chemistry after first normalizing the data (i.e., mean = 0, standard deviation = 1) then computing Euclidean distance, the preferred similarity measure for environmental data. A principal components analysis was performed on the resemblance data, retaining all principal components (PCs) with eigenvalues greater than 1.0. Interpretations of the PCs were based on their correlations with individual chemistry variables. The strength of the relationship between vegetation and water chemistry was analyzed using the RELATE permutation test, applied to the matching coefficient between the vegetation and water chemistry resemblance matrices. Under the null hypothesis that there is no relationship between two similarity matrices, will be approximately zero. PRIMER’s analysis of similarity (ANOSIM) procedure, which tests group structure with a nonparametric ANOVA analog, was also used with the water chemistry
Fig. 2. Dendrogram showing results of hierarchical clustering. The dashed line at 27% similarity is used to define the seven plant communities.
and vegetation resemblance matrices to determine if pairs of plant communities were distinguished by their water chemistry. Classification and regression trees were constructed to relate environmental and vegetation data using the recursive partitioning and regression trees package for the statistical software R (RPART version 3.1–39) (Therneau and Atkinson, 1997). RPART allows cross-validation runs on the data to determine the optimally sized tree, selected as the subtree that performs best on a validation set. We performed a 10-fold cross-validation, where each run consisted of 10 random divisions of the data into 90% learning and 10% test sets. For each split we computed the complexity parameter (cp), a measure of how much additional accuracy a split must add to the entire tree to warrant the additional complexity. Any split that did not decrease the overall lack of fit by a factor of cp was not attempted, meaning that the overall r2 must increase by cp at each step. We determined the optimal tree size from the table of cross-validation error values for various tree lengths as the number of splits corresponding to the minimum cross-validation error, and pruned the tree to the cp value for this optimum. Output from the summary function of RPART was examined to evaluate alternative primary split variables. Two trees were constructed (1) a regression tree that used scores from the vegetation first NMDS axis as the endpoint and all GIS-derived and chemistry variables as potential predictor variables, and (2) a classification tree that used the identified plant communities as the endpoint and the eleven chemical parameters as potential predictor variables. 3. Results 3.1. Plant communities Hierachical clustering of vegetation in the 48 wetlands yielded seven plant communities that were distinct at a similarity of 27% (SIMPROF, P < 0.01), defined by the species that contributed most to each cluster’s similarity (Table 1, Fig. 2): burreed marshes (Sparganium eurycarpum, BUR), bluejoint grass wetlands (Calamagrostis canadensis, BLU), three-square rush marshes (Schoenoplectus pungens var. pungens, 3SQ), reed marshes (P. australis, REE), eastern cattail marshes (ECA), western cattail marshes (WCA), and poor fens (FEN). The ECA and WCA marshes were both dominated by invasive Typha, but had different understory species (Table 1). Sphagnum contributed most to the similarity of the FEN wetlands, which were also vegetated by Carex lasiocarpa var. americana, Myrica gale, and two species of ericaceous shrubs. Invasive plant
114
C.A. Johnston, T.N. Brown / Aquatic Botany 104 (2013) 111–120
Table 1 Percent average similarity across sites for taxa in plant communities. All 138 taxa were used in the analysis, but only taxa contributing 4.0% or more to the average similarity of one or more communities shown here. FEN Andromeda polifolia var. glaucophylla Calamagrostis canadensis Carex lacustris Carex lasiocarpa var. americana Carex stricta Carex utriculata Ceratophyllum demersum Chamaedaphne calyculata Chara vulgaris Decodon verticillatus Drosera rotundifolia a Hydrocharis morsus–ranae Impatiens capensis Juncus sp. Juncus balticus var. littoralis Juncus nodosus Leersia oryzoides Lemna minor Menyanthes trifoliata Myrica gale a Phragmites australis Sagittaria latifolia Sarracenia purpurea Schoenoplectus pungens Schoenoplectus tabernaemontani Sparganium eurycarpum Sphagnum spp. Thelypteris palustris Typha latifolia a Invasive Typha Urtica dioica Utricularia macrorhiza Number of sites Average similarity, % a
BUR
BLU
3SQ
ECA
WCA
REE
6.8 25.3
10.3
4.3 12.7 4.3 5.1
14.2 6.0
8.5 14.3 4.3 4.3 7.3 11.9
5.6
5.9 6.2 6.9 8.3 10.0
6.2 11.5 4.8
84.2
8.2 5.2 37.9 5.6 17.5
7.1 4.0
18.7 5.0 4.8
5 60.3
5.6 8 35.0
8 32.8
4.7
4.6 4.2 36.2
4 39.5
9 37.1
19.5 4.7
4.8
11 39.0
3 36.1
Invasive taxon.
Table 2 Associations within the U.S. National Vegetation Classification corresponding to the plant communities identified. All associations are in subclass 2.C Temperate & Boreal Shrubland & Grassland. Code
Hierarchy
Description
FEN
Formation Division Macrogroup Group Association Formation Division Macrogroup Group Association Formation Division Macrogroup Group Association Formation Division Macrogroup Group Association
2.C.4 Temperate & Boreal Bog & Fen D029 North American Bog & Fen M062 North American Boreal Bog & Fen G185 Eastern North American Boreal Alkaline Fen Group CEGL005228 Chamaedaphne calyculata – Myrica gale/Carex lasiocarpa Dwarf-shrubland 2.C.4 Temperate & Boreal Bog & Fen D029 North American Bog & Fen M062 North American Boreal Bog & Fen G185 Eastern North American Boreal Alkaline Fen Group CEGL005115 Calamagrostis canadensis – Carex viridula – Cladium mariscoides – Lobelia kalmii Herbaceous Vegetation 2.C.5 Temperate & Boreal Freshwater Wet Meadow & Marsh D030 Eastern North American Freshwater Wet Meadow, Riparian & Marsh M069 Eastern North American Wet Meadow & Marsh G125 Eastern North American Freshwater Marsh Group CEGL002026 Schoenoplectus tabernaemontani – Typha spp. – (Sparganium spp., Juncus spp.) Herbaceous Vegetation 2.C.5 Temperate & Boreal Freshwater Wet Meadow & Marsh D030 Eastern North American Freshwater Wet Meadow, Riparian & Marsh M069 Eastern North American Wet Meadow & Marsh G125 Eastern North American Freshwater Marsh Group CEGL005274 Schoenoplectus acutus – Schoenoplectus subterminalis – Eleocharis palustris – (Schoenoplectus americanus) Northern Great Lakes Shore Herbaceous Vegetation 2.C.5 Temperate & Boreal Freshwater Wet Meadow & Marsh D030 Eastern North American Freshwater Wet Meadow, Riparian & Marsh M069 Eastern North American Wet Meadow & Marsh G125 Eastern North American Freshwater Marsh Group CEGL005112 Typha spp. – Schoenoplectus tabernaemontani – Mixed Herbs Southern Great Lakes Shore Herbaceous Vegetation 2.C.5 Temperate & Boreal Freshwater Wet Meadow & Marsh D030 Eastern North American Freshwater Wet Meadow, Riparian & Marsh M303 Eastern North American Ruderal Wet Meadow & Marsh G556 Northern & Central Ruderal Wet Meadow & Marsh Group CEGL004141 Phragmites australis Eastern North America Temperate Semi-natural Herbaceous Vegetation
BLU
BUR
3SQ
ECA, WCA
Formation Division Macrogroup Group Association
REE
Formation Division Macrogroup Group Association
A
FEN BUR BLU 3SQ ECA W CA REE
1 .0
0 .0
c
c
ab bc
abc a
8.0
NM D S 2
0 .5
pH
115
8.5
1 .5
9.0
C.A. Johnston, T.N. Brown / Aquatic Botany 104 (2013) 111–120
7.5
-0 .5 -1 .0
a
7.0
-1 .5 -2 .0
1 .0
-1 .5
-1 .0
-0 .5
0 .0
0 .5
NM D S 1
1 .0
1 .5
2 .0
2.5
-2 .0
B
c
logTP
c
0 .5
2.0
bc
NM D S 3
0 .0
ab
1.5
-0 .5
a a
ab
-0 .5
0 .0
0 .5
1 .0
1 .5
NM D S 2 Fig. 3. Plots of sites from three-dimensional NMDS analyses of vegetation data, sites coded by plant community hierarchical cluster. (a) Sites plotted relative to NMDS axes 1 and 2 and (b) sites plotted relative to NMDS axes 2 and 3.
taxa (P. australis, H. morsus-ranae, invasive Typha) contributed to the similarity of the REE, ECA, and WCA wetlands (Table 1). The plant communities corresponded well with associations described by the U.S. National Vegetation Classification, except that ECA and WCA wetlands were not distinguished from each other (Table 2). Site location was not an input to the hierarchical clustering, but plant communities tended to be geographically clustered. In the northern Great Lakes, the FEN wetlands were on Lake Superior, and the BUR wetlands were on Lakes Superior and northern Lake Huron (Fig. 1). In the southern Great Lakes, the REE, 3SQ, and WCA marshes were on Lakes Erie, Huron, and Michigan, respectively. The ECA and BLU plant communities were more widely distributed, however. The ECA wetlands occurred primarily on Lake Ontario, but had geographic outliers on Lake Michigan (site 1152) and Lake Erie (site 1241). A single BLU wetland occurred at Presque Isle on Lake Erie (site 1260), 414 km southeast of the other BLU wetlands on Lakes Huron and Michigan (Fig. 1). The NMDS analysis of the vegetation data achieved a stable three-dimensional configuration with a stress of 0.15. The first NMDS axis (NMDS1) defined two north–south groupings of plant communities (Fig. 3a). The FEN, BLU wetlands, and BUR wetlands had generally negative NMDS1 scores, whereas the four southern plant communities had generally positive NMDS1 scores. The NMDS2 axis distinguished 3SQ, BLU, and REE marshes (negative NMDS2) from ECA and FEN wetlands (positive NMDS2). The NMDS3 axis distinguished BLU and WCA wetlands (positive NMDS3) from BUR marshes (negative NMDS3) (Fig. 3b). Thus, both the hierarchical cluster and NMDS analyses supported the seven-community classification of the 48 sites.
logCOND
c bc
a
bc
bc
b
2.4
-1 .0
2.2
-1 .5
a
2.0
-2 .0
1.8
-1 .5
2.6 2.8 3.0 1.0
-1 .0
FEN BUR BLU 3SQ ECA WCA REE Fig. 4. Boxplots for selected water chemistry variables. (a) pH, (b) log total phosphorus and (c) log conductivity. Circles are outliers with values greater or less than 1.5 times the inter-quartile range. Clusters with the same letter code are not significantly different (Tukey multiple comparison of means, P < 0.05).
3.2. Water chemistry of plant communities The RELATE analysis yielded a matching coefficient that was significantly different than zero ( = 0.171, P < 0.003), indicating a significant relationship between the vegetation and water chemistry resemblance matrices. Plant communities differed significantly for all water chemistry variables except DO, DOC, and NH4 -N (ANOVA, Table 3). Mean water chemistry values were generally least for the FEN and greatest for the WCA and REE communities, although patterns differed by chemical variable (Fig. 4). Comparison of water chemistry values for pairs of plant communities showed a significant overall relationship (ANOSIM, global
116
C.A. Johnston, T.N. Brown / Aquatic Botany 104 (2013) 111–120
Table 3 Means of water chemistry variables by plant community. All variables except pH were log-transformed, and values reported are back-transformed geometric means. Letter codes show Tukey’s HSD comparisons for ANOVA; values followed by the same letter are not statistically different (P < 0.05).
Chl a Cl− COND DO DOC NH4 -N NO3 -N pH TN TP TSS
Units
F6,41
FEN
BUR
g/L mg/L S/cm mg/L mg/L g/L g/L – g/L g/L mg/L
9.4*** 23.5*** 19.8*** 1.9 0.3 0.5 3.9** 10.2*** 8.6*** 9.9*** 7.5***
4.1ab 1.8a 92a 7.5 10.6 10.1 11.7a 7.4a 457a 18.3a 4.1a
2.9a 4.2ab 159a 7.7 7.1 9.6 23.1a 7.8ab 411a 32.0ab 7.7ab
BLU
3SQ
1.9a 7.6b 336bc 9.0 8.9 12.8 24.0a 8.2bc 586ab 21.0a 4.6a
4.8ab 32.2c 306bc 8.5 7.1 15.6 43.2ab 8.7c 823abc 23.6ab 6.5ab
ECA
WCA
REE
6.0ab 18.0c 300b 6.2 9.2 11.1 17.0a 7.5a 783abc 54.8bc 4.2a
24.1c 26.3c 489c 8.1 8.0 14.8 191.7b 8.3c 1498c 92.9c 17.8bc
29.6bc 37.6c 432bc 6.8 7.5 23.0 113.2ab 8.2abc 1561bc 107.4c 36.2c
N.S. = not significant. ** P < 0.01. *** P < 0.001.
Table 5 Eigenvalues for chemistry principal components analysis. Bolded variables have correlations >0.5 with PCs. Variable
PC1 CHEM
PC2 CHEM
PC3 CHEM
log Chl a log Cl log COND log DO log DOC log NH4 -N log NO3 -N pH log TN log TP log TSS
−0.386 −0.374 −0.358 −0.006 −0.077 −0.147 −0.268 −0.226 −0.406 −0.377 −0.354
−0.180 0.022 0.110 0.510 −0.506 0.171 0.388 0.410 −0.189 −0.237 0.007
−0.156 −0.194 −0.181 0.363 0.392 0.694 0.144 −0.297 0.161 0.055 0.015
Eigenvalue Percent variation Cumulative percent variation
4.68 42.5 42.5
1.93 17.6 60.1
1.23 11.2 71.3
to plant communities (Table 3): TN, TP, COND, TSS, Cl− , and Chl a. Plant communities were arrayed along this gradient from FEN and BUR (most negative PC1 CHEM score) to WCA and REE (most positive PC1 CHEM scores, Fig. 5). PC2 CHEM was positively correlated with DO, pH, and NO3 -N, and negatively correlated with DOC, and
Table 4 Matrix of R-statistic values from ANOSIM analysis comparing water chemistry of plant community pairs. Significance determined by 999 random permutations based on the water chemistry resemblance matrix (fewer possible permutations for comparisons with 3SQ and REE communities).
BUR BLU 3SQ ECA WCA REE * **
P < 0.05. P < 0.01.
FEN
BUR
0.020 0.726** 1.000* 0.591** 0.778** 1.000*
0.204* 0.219 0.276** 0.542** 0.672*
BLU
3SQ
ECA
C om m unity
4
FEN BUR BLU 3SQ ECA WCA REE
DO pH
2 1260
PC 2
R = 0.397, P < 0.001), and that 16 of the 21 possible pairs were statistically distinct based on their water chemistry (Table 4). The two cattail communities (ECA, WCA) were chemically dissimilar, which substantiates their independently determined floristic separation. Water pH values were circumneutral to slightly alkaline (Table 3, Fig. 4a). The pH values of FEN waters were the lowest of all wetland types, consistent with the presence of Sphagnum. The pH of the ECA wetlands was also low, but more variable. The 3SQ wetlands, all of which contained the macroalga Chara, had the highest average pH values. There were mixed results for expected relationships with nutrients, chlorophyll a, and TSS. All nutrient variables except NH4 -N had a significant effect on plant communities, but only total phosphorus distinguished all three of the plant communities dominated by invasive plants (ECA, WCA, REE) from other plant communities (Fig. 4b). The ECA wetlands had much lower concentrations of TSS, NO3 -N, and Chl a than those in the WCA and REE wetlands. Thus, although two of the plant communities dominated by invasive species (WCA, REE) were consistently associated with poorer water quality, the third (ECA) was not. An unanticipated finding was that the two metrics that best differentiated the plant communities were chloride and conductivity (Table 3). The chloride concentrations were lowest in the FEN wetlands, intermediate in the BUR and BLU wetlands, and highest in the 3SQ, REE, and cattail communities of the southern Great Lakes. Conductivity exhibited a similar north-to-south increase, except that ECA wetlands had significantly lower conductivity than did WCA wetlands (Fig. 4c). The principal components analysis of the eleven water chemistry variables yielded three PCs with eigenvalues > 1.0 that explained a cumulative 71.3% of the variance (Table 5). The first principal component (PC1 CHEM) was highly correlated with six water chemistry variables, all of which were significantly related
NO3
NH4
COND
1152
0
TSS
Cl C h la TN TP
-2 DOC
WCA 1241
0.233 0.383** 0.514** 0.938*
-4
0.389* 0.042 0.685*
-4
0.423** 0.770*
−0.233
-2
0
PC1
2
4
6
Fig. 5. Principal components analysis of water chemistry data, sites coded by plant community hierarchical cluster. Length of lines within the circle of the bi-plot overlay shows correlation between original water chemistry variables and PC axes.
C.A. Johnston, T.N. Brown / Aquatic Botany 104 (2013) 111–120
117
Table 6 Top five primary splits for nodes shown in Fig. 5, and their improvement values. Split threshold shows variables that would go to the left side of the split. CSI = cumulative stress index, PC1 AG = agriculture principal component, GDD = growing degree days, LAT = latitude, log COND = log of conductivity, PC1 CHEM = chemistry first principal component, log TN = log of total nitrogen, WETW 500, WETW 1000 = wetland & inland water within 500 m or 1000 m of the sampled wetland, DEV SHED = urban development in the catchment draining to the wetland. Node
Primary splits
Improvement
Node 1 (n = 48)
CSI < 1.535 PC1 AG < 0.1 GDD < 1730 LAT ≥ 45.065 LAKE (Huron, Superior) GDD < 2243.5 PC1 AG < 0.845 logCOND < 2.56 PC1 CHEM ≥ −1.15 logTN WETW 500 ≥ 0.085 WETW 1000 ≥ 0.065 PC1 AG < 0.26 CSI < 2.15 DEV shed < 0.055
0.549 0.509 0.507 0.507 0.475 0.560 0.500 0.404 0.401 0.397 0.542 0.542 0.448 0.385 0.366
Node 3 (n = 35)
Node 6 (n = 28)
distinguished only high pH wetlands (e.g., 3SQ) from lower pH wetlands (e.g., ECA). PC3 CHEM was related primarily to NH4 -N (Table 5), a variable that was poorly related to plant communities (Table 3). As noted, several sites were geographic outliers of their plant community type, occurring long distances from others of their floristic type. However, inspection of their location in the multivariate space of the principal components ordination showed that they were chemically similar to other wetlands in their plant community grouping (Fig. 5). The Presque Isle BLU wetland (site 1260) was located on Lake Erie, which generally had the worst water quality of the five Great Lakes, yet its water chemistry was consistent with that of the other BLU wetlands located on the northern Great Lakes. Site 1260 had a PC1 CHEM score of 0.76, well within the average PC1 CHEM score of other BLU wetlands (1.05 ± 0.78, mean ± SD) and much higher than the average PC1 CHEM score of other Lake Erie wetlands (−2.64 ± 0.63). Similarly, the Lake Michigan ECA wetland (site 1152) also clustered well chemically with the ECA marshes on Lake Ontario, despite being a geographic outlier. The other ECA geographic outlier (site 1241) had the lowest PC2 CHEM score of any site because of its extremely low DO concentration and high DOC, more consistent with the average PC2 CHEM score of ECA wetlands (−1.13 ± 0.43) than WCA wetlands (0.31 ± 0.31). A regression tree that used the NMDS1 values as the endpoint and considered as predictors all GIS-derived and chemistry variables (including the chemistry PC scores) had four terminal nodes and explained 75.7% of the variance (Fig. 6a). The final solution used a combination of GIS-derived variables representing overall anthropogenic stress (CSI), growing degree days (GDD), and the proportion of wetland & inland lake area within a 500 m buffer surrounding the site (WETL 500). The terminal node with the lowest average NMDS1 scores (−0.83, best condition) included all five FEN sites plus four each of BLU and BUR sites. The terminal node with the highest average NMDS1 scores (+0.93, worst condition) included all three REE sites, three WCA sites, and one ECA site. Chemical variables (COND, PC1 CHEM, logTN) were considered as primary splits for node 3 (Table 6), but the GDD variable was more explanatory. A previous regression tree analysis using the full 90-wetland dataset also selected CSI and GDD for the first two splits (Johnston et al., 2009b).
Fig. 6. Classification and regression trees using environmental variables to explain vegetation characteristics. Split nodes indicated by ovals; terminal nodes indicated by rectangles. (a) Classification tree using NMDS1 scores as the endpoint and all chemical and GIS-derived environmental variables as potential predictors. Upper value within rectangles is mean NMDS1 score for each terminal node. CSI = cumulative stress index, GDD = growing degree days, WETW 500 = proportion of wetland & water land area within a 500 m buffer surrounding the site and (b) regression tree using plant community classes as the endpoint and the eleven chemistry variables as potential predictors. Upper value within rectangles is predominant plant community of each terminal node; lower values show number of sites assigned to the terminal node, separated by plant community class.
A classification tree that used the seven plant communities as the endpoint and the eleven chemical parameters as predictors correctly assigned 79% of the sites into four classes based on total N, conductivity, and pH: FEN & BUR, BLU & 3SQ, ECA, and WCA (Fig. 6b). The REE sites were distributed across the WCA and BLU & 3SQ classes.
118
C.A. Johnston, T.N. Brown / Aquatic Botany 104 (2013) 111–120
4. Discussion 4.1. Large lake wetlands This study illustrates that wetlands along the coast of a large lake may have significantly different water chemistry, and that vegetation characteristics can reveal those chemical differences. Geographic outliers of plant community types such as the Presque Isle BLU wetland on Lake Erie demonstrated that regionally anomalous vegetation was an indicator of unusual water chemistry. Classifications of ecological status by whole lakes may not adequately represent lakeside wetlands because of landward and autochthonous wetland influences on water chemistry. Our analysis of cattail wetlands on the Great Lakes showed two distinct types that differed floristically and chemically, illustrating the need to look beneath the over-story to the species that are associated with the dominant emergent. We recommend that the U.S. National Vegetation Classification cattail association for the Great Lakes (CEGL005112) be split into a Typha spp.–Thelypteris palustris–H. morsus-ranae association (our ECA community) and a Typha spp.–C. canadensis–Leersia oryzoides association (our WCA community). 4.2. Differentiating wetlands by surface water chemistry Surface water chemistry differentiated plant communities adequately, but not perfectly. The RELATE and ANOSIM analyses found highly significant relationships between vegetation and water chemistry. Sixteen of the 21 possible vegetation community pairs were statistically distinct based on their water chemistry. We were able to model a simplified community grouping using a threevariable water chemistry classification tree. Thus, this approach was generally successful. Our findings show that certain water chemistry parameters such as conductivity and pH may be important to consider when setting standards to support wetland plant life, although care must always be taken to avoid oversimplifying wetland systems which are also influenced by multi-year water level fluctuations, historical land use and land cover, and many other factors with complex interactions. In this study, the Sphagnum-containing FEN wetlands and Characontaining 3SQ wetlands apparently altered the pH of wetland waters (autochthonous influences). Water pH is affected by alkalinity, which is largely produced by the dissolution of carbon dioxide in water (Hem, 1970). Any removal of inorganic carbon (bicarbonate, CO2 ) from the water column will increase pH. The macroalga Chara can utilize bicarbonate as a carbon source, thereby increasing pH (Kufel and Kufel, 2002). The lower pH values associated with Sphagnum-containing FEN communities may have been due in part to the releases of organic acids. However, pH was low in ECA wetlands that lacked Sphagnum and high in WCA wetlands that lacked Chara, so the presence of these plant species alone was insufficient to explain differences in average pH concentrations across communities. Our result that invasive plant species abundance increased with increasing total phosphorus concentrations was consistent with the generally accepted paradigm that nutrient enrichment accelerates plant invasion (Rejmánek et al., 2005). The lack of relationship between plant community and NH4 -N was unexpected based on its observed importance elsewhere (Craft et al., 2007; Croft and Chow-Fraser, 2007), but phosphorus is often the limiting nutrient in wetlands (Verhoeven and Schmitz, 1991; Miao et al., 2000). Conductivity and chloride, the two chemical properties that best defined Great Lakes wetland plant communities, are not in themselves drivers of plant growth. Conductivity, the ability of water to conduct an electrical current, is strongly related to ionic solutes,
particularly calcium and magnesium (Hem, 1970). Chloride is a biologically conservative element, often used as a tracer in freshwater wetlands and streams (Turner and Townley, 2006). These measures of major ions are controlled primarily by the physicochemical environment, implying that vegetation is responding to rather than driving biogeochemical differences among plant communities. Although biologically conservative, chloride is augmented by human sources such as road salt, agricultural chemicals, and sewage discharge, and is considered to be a surrogate measure of anthropogenic watershed stress (Herlihy et al., 1998). A predictive model for Great Lakes coastal wetlands showed that watershed agricultural and human population land uses explained 58% of the variance in the concentration of chloride, the strongest predictive model for any of the seven water chemistry parameters tested (Morrice et al., 2008). Chloride concentrations were also significantly greater in the southern Eastern Broadleaf Forest ecoprovince than in the northern Laurentian Mixed Forest (Trebitz et al., 2007), consistent with the north-south wetland plant community differences observed here. Conductivity has been shown to discriminate wetland plant community types in comparative studies throughout North America and Europe (Vitt and Chee, 1990; Rey Benayas and Scheiner, 1993; Rolon and Maltchik, 2006; Sass et al., 2010), and is used to define ecotypes in the ECOFRAME scheme (Moss et al., 2003). In Canadian Great Lakes coastal marshes, conductivity was the water quality variable most highly correlated with the first axis of a canonical correspondence analysis of plant and environmental data (Croft and Chow-Fraser, 2007). Conductivity is an indicator of wetland water source because it tends to be greater in groundwater than in surface water or precipitation (Rey Benayas et al., 1990), particularly where the groundwater emanates from calcareous bedrock (Heegaard et al., 2001). Carbonate aquifers abut all of the Great Lakes except Lake Superior (Grannemann et al., 2000), which is consistent with the significantly lower conductivity in wetlands on that lake (Trebitz et al., 2007). The significantly higher conductivity concentration in WCA versus ECA wetlands may be due to their location relative to carbonate aquifers: there is no carbonate bedrock on the eastern end of Lake Ontario, where most of the ECA wetlands are located. 4.3. Water chemistry vs. GIS-derived variables Field-collected water chemistry was expected to outperform GIS-derived variables as a predictor of wetland vegetation condition, but that expectation was not supported by the regression tree analysis. Our finding implies that vegetation condition is impacted by a complexity of anthropogenic stressors (represented by the cumulative stress index in this study) and climatic conditions (represented by GDD in this study) that cannot be simplified to a single chemical indicator. The ECOFRAME classification of the European Water Framework Directive already recognizes the importance of climatic conditions in its scheme by grouping lakes into cold, cool, temperate, and warm classes before classifying ecological status (Moss et al., 2003). GIS-derived variables may better integrate stressors over time and space than would a one-time field sampling of water chemistry. Vegetation similarly integrates responses over time, an attribute that makes it a suitable ecological indicator (King and Richardson, 2003; Chow-Fraser, 2006; Penning et al., 2008; del Pozo et al., 2010). 4.4. Management implications and transferability of approach These results may aid water quality certification under §401 of the Clean Water Act, a tool being actively promoted by the U.S. Environmental Protection Agency that gives U.S. states and tribes a great deal of discretion to protect wetland resources and
C.A. Johnston, T.N. Brown / Aquatic Botany 104 (2013) 111–120
their valuable functions (USEPA, 2010; ASWM, 2012). All seven Great Lakes states have §401 programs, and Wisconsin’s includes wetland-specific water quality standards that prohibit “concentrations or combinations of substances which are toxic or harmful to human, animal or plant life. . . in amounts which individually or cumulatively may cause significant adverse impacts to wetlands” (http://legis.wisconsin.gov/rsb/code/nr/nr103.pdf). The findings of this research could strengthen such §401 programs by identifying water quality parameters associated with particular plant communities. Grouping wetlands by plant communities could provide a scientifically-defensible basis for stricter water quality standards to protect sensitive wetland types. This study used surface water chemistry because those were the available data, a situation that is not uncommon in aquatic botany field studies. We would have preferred to use sediment chemistry data because emergent vascular plants take up nutrients from the soils in which they are rooted, and element concentrations in the overlying surface water may not mirror those in the soil (Johnston et al., 2001). Fortunately, our results show that water chemistry is valuable for distinguishing floristic assemblages at the wetland scale. Advantages of using water samples are that they can be easily collected and are less spatially variable than sediment samples. Given that we examined differences between wetlands rather than within wetlands, water samples probably provided a more representative indicator of overall water chemistry within the wetland as a whole. As was demonstrated here, innovative approaches for combining and reanalyzing datasets can provide meaningful new results, and we encourage others to utilize existing datasets to explore aquatic plant–environmental relationships. Acknowledgments Anett Trebitz and John Morrice of the U.S. Environmental Protection Agency Mid-Continent Ecology Division and Richard Axler of the Natural Resources Research Institute of the University of Minnesota-Duluth were responsible for water sampling and analysis. Nicholas Danz managed the chemistry data. M. Aho, A. Boers, K. Bailey Boomer, M. Bourdaghs, K. Cappillino, R. Clark, S. Cronk, A. Freeman, C. Frieswyk, D. James, C. Johnson, L. Ladwig, A. Marsh, M. Tittler, L. Vaccaro, and C. Williams collected vegetation field data. This research was partially supported (2001–2006) by a grant from the United States Environmental Protection Agency’s Science to Achieve Results (STAR) Estuarine and Great Lakes (EaGLe) program through funding to the Great Lakes Environmental Indicators (GLEI) Project, US EPA Agreement EPA/R-828675. Although the research described in this article has been funded wholly or in part by the U.S. Environmental Protection Agency, it has not been subjected to the Agency’s required peer and policy review and therefore does not necessarily reflect the views of the Agency and no official endorsement should be inferred. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.aquabot. 2012.08.005. References APHA, 2000. Standard Methods for the Examination of Water and Wastewater. American Public Health Association, Washington, DC. Arrigoni, A., Findlay, S., Fischer, D., Tockner, K., 2008. Predicting carbon and nutrient transformations in tidal freshwater wetlands of the Hudson River. Ecosystems 11, 790–802. ASWM, 2012. Section 401 Certification Best Practices in Dredge and Fill Permit Programs. Association of State Wetland Managers, Windham, ME, p. 15. Chow-Fraser, P., 2006. Development of the water quality index (WQI) to assess effects of basin-wide land-use alteration on coastal marshes of the Laurentian
119
Great Lakes. In: Simon, T.P., Stewart, P.M. (Eds.), Coastal Wetlands of the Laurentian Great Lakes: Health, Habitat, and Indicators. Indiana Biological Survey, Bloomington, IN, pp. 137–167. Clarke, K.R., Gorley, R.N., 2006. PRIMER v6: User Manual/Tutorial. PRIMER-E, Plymouth, UK. Craft, C., Krull, K., Graham, S., 2007. Ecological indicators of nutrient enrichment, freshwater wetlands, Midwestern United States (U.S.). Ecological Indicators 7, 733–750. Croft, M.V., Chow-Fraser, P., 2007. Use and development of the wetland macrophyte index to detect water quality impairment in fish habitat of Great Lakes coastal marshes. Journal of Great Lakes Research 33 (3), 172–197 (special issue). Danz, N., Niemi, G., Regal, R., Hollenhorst, T., Johnson, L., Hanowski, J., Axler, R., Ciborowski, J., Hrabik, T., Brady, V., Kelly, J., Morrice, J., Brazner, J., Howe, R., Johnston, C., Host, G., 2007. Integrated measures of anthropogenic stress in the U.S Great Lakes basin. Environmental Management 39, 631–647. Declerck, S., Vandekerkhove, J., Johansson, L., Muylaert, K., Conde-Porcuna, J.M., Van der Gucht, K., Perez-Martinez, C., Lauridsen, T., Schwenk, K., Zwart, G., Rommens, W., Lopez-Ramos, J., Jeppesen, E., Vyverman, W., Brendonck, L., De Meester, L., 2005. Multi-group biodiversity in shallow lakes along gradients of phosphorus and water plant cover. Ecology 86, 1905–1915. del Pozo, R., Fernández-Aláez, C., Fernández-Aláez, M., 2010. An assessment of macrophyte community metrics in the determination of the ecological condition and total phosphorus concentration of Mediterranean ponds. Aquatic Botany 92, 55–62. Edvardsen, A., Økland, R.H., 2006. Variation in plant species richness in and adjacent to 64 ponds in SE Norwegian agricultural landscapes. Aquatic Botany 85, 79–91. Grannemann, N.G., Hunt, R.J., Nicholas, J.R., Reilly, T.E., Winter, T.C., 2000. The Importance of Ground Water in the Great Lakes Region. U.S. Geological Survey Lansing, MI, p. 19. Heegaard, E., Birks, H.H., Gibson, C.E., Smith, S.J., Wolfe-Murphy, S., 2001. Species–environmental relationships of aquatic macrophytes in Northern Ireland. Aquatic Botany 70, 175–223. Hem, J.D., 1970. Study and Interpretation of the Chemical Characteristics of Natural Water. U.S. Government Printing Office, Washington, DC, p. 363. Herlihy, A., Stoddard, J., Johnson, C.B., 1998. The relationship between stream chemistry and watershed land cover data in the Mid-Atlantic Region, U.S. Water Air and Soil Pollution 105, 377–386. Houlahan, J.E., Keddy, P.A., Makkay, K., Findlay, C.S., 2006. The effects of adjacent land use on wetland species richness and community composition. Wetlands 26, 79–96. Jennings, M.D., Faber-Langendoen, D., Loucks, O.L., Peet, R.K., Roberts, D., 2009. Standards for associations and alliances of the U.S. National Vegetation Classification. Ecological Monographs 79, 173–199. Johnston, C.A., Bridgham, S.D., Schubauer-Berigan, J.P., 2001. Nutrient dynamics in relation to geomorphology of riverine wetlands. Soil Science Society of America Journal 65, 557–577. Johnston, C.A., Brown, T., Hollenhorst, T., Wolter, P., Danz, N., Niemi, G., 2009a. GIS in support of ecological indicator development. In: Madden, M. (Ed.), Manual of Geographic Information Systems. American Society for Photogrammetry and Remote Sensing, Bethesda, MD, pp. 1095–1113. Johnston, C.A., Zedler, J.B., Tulbure, M.G., Frieswyk, C.B., Bedford, B.L., Vaccaro, L., 2009b. A unifying approach for evaluating the condition of wetland plant communities and identifying related stressors. Ecological Applications 19, 1739–1757. Johnston, C.A., Zedler, J.B., Tulbure, M.G., 2010. Latitudinal gradient of floristic condition among Great Lakes coastal wetlands. Journal of Great Lakes Research 36, 772–779. King, R.S., Richardson, C.J., 2003. Integrating bioassessment and ecological risk assessment: an approach to developing numerical water-quality criteria. Environmental Management 31, 795–809. Kufel, L., Kufel, I., 2002. Chara beds acting as nutrient sinks in shallow lakes—a review. Aquatic Botany 72, 249–260. Lougheed, V.L., Crosbie, B., Chow-Fraser, P., 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. Mahaney, W.M., Wardrop, D.H., Brooks, R.P., 2005. Impacts of sedimentation and nitrogen enrichment on wetland plant community development. Plant Ecology 175, 227–243. Miao, S., Newman, S., Sklar, F.H., 2000. Effects of habitat nutrients and seed sources on growth and expansion of Typha domingensis. Aquatic Botany 68, 297–311. Morrice, J.A., Danz, N.P., Regal, R.R., Kelly, J.R., Niemi, G.J., Reavie, E.D., Hollenhorst, T., Axler, R.P., Trebitz, A.S., Cotter, A.M., Peterson, G.S., 2008. Human influences on water quality in Great Lakes coastal wetlands. Environmental Management 41, 347–357. Moss, B., Stephen, D., Alvarez, C., Becares, E., Bund, W.V.D., Collings, S.E., Donk, E.V., Eyto, E.D., Feldmann, T., Fernández-Aláez, C., Fernández-Aláez, M., Franken, R.J.M., García-Criado, F., Gross, E.M., Gyllström, M., Hansson, L.-A., Irvine, K., Järvalt, A., Jensen, J.-P., Jeppesen, E., Kairesalo, T., Kornijów, R., Krause, T., Künnap, H., Laas, A., Lill, E., Lorens, B., Luup, H., Rosa Miracle, M., Nõges, P., Nõges, T., Nykänen, M., Ott, I., Peczula, W., Peeters, E.T.H.M., Phillips, G., Romo, S., Russell, V., Salujõe, J., Scheffer, M., Siewertsen, K., Smal, H., Tesch, C., Timm, H., Tuvikene, L., Tonno, I., Virro, T., Vicente, E., Wilson, D., 2003. The determination of ecological status in shallow lakes — a tested system (ECOFRAME) for implementation of the European Water Framework Directive. Aquatic Conservation: Marine and Freshwater Ecosystems 13, 507–549.
120
C.A. Johnston, T.N. Brown / Aquatic Botany 104 (2013) 111–120
Murphy, K.J., 2002. Plant communities and plant diversity in softwater lakes of northern Europe. Aquatic Botany 73, 287–324. Nicolet, P., Biggs, J., Fox, G., Hodson, M.J., Reynolds, C., Whitfield, M., Williams, P., 2004. The wetland plant and macroinvertebrate assemblages of temporary ponds in England and Wales. Biological Conservation 120, 261–278. Olde Venterink, H., Wassen, M.J., Verkroost, A.W.M., De Ruiter, P.C., 2003. Species richness–productivity patterns differ between N-, P-, and K-limited wetlands. Ecology 84, 2191–2199. Penning, W., Dudley, B., Mjelde, M., Hellsten, S., Hanganu, J., Kolada, A., van den Berg, M., Poikane, S., Phillips, G., Willby, N., Ecke, F., 2008. Using aquatic macrophyte community indices to define the ecological status of European lakes. Aquatic Ecology 42, 253–264. Reavie, E.D., Axler, R.P., Sgro, G.V., Danz, N.P., Kingston, J.C., Kireta, A.R., Brown, T.N., Hollenhorst, T.P., Ferguson, M.J., 2006. Diatom-based weighted-averaging transfer functions for Great Lakes coastal water quality: relationships to watershed characteristics. Journal of Great Lakes Research 32, 321–347. Rejmánek, M., Richardson, D.M., Pyˇsek, P., 2005. Plant invasions and invasibility of plant communities. In: van der Maarel, E. (Ed.), Vegetation Ecology. Blackwell, Oxford, UK, pp. 332–355. Rey Benayas, J.M., Bernáldez, F.G., Levassor, C., Peco, B., 1990. Vegetation of groundwater discharge sites in the Douro basin, central Spain. Journal of Vegetation Science 1, 461–466. Rey Benayas, J.M., Scheiner, S.M., 1993. Diversity patterns of wet meadows along geochemical gradients in central Spain. Journal of Vegetation Science 4, 103–108. Rolon, A., Maltchik, L., 2006. Environmental factors as predictors of aquatic macrophyte richness and composition in wetlands of southern Brazil. Hydrobiologia 556, 221–231. Sass, L.L., Bozek, M.A., Hauxwell, J.A., Wagner, K., Knight, S., 2010. Response of aquatic macrophytes to human land use perturbations in the watersheds of Wisconsin lakes, U.S.A. Aquatic Botany 93, 1–8. Sjörs, H., 1950. On the relation between vegetation and electrolytes in north Swedish mire waters. Oikos 2, 241–258. Therneau, T.M., Atkinson, E.J., 1997. An Introduction to Recursive Partitioning Using the RPART Routines. Mayo Clinic, Rochester, MN, USA.
Toivonen, H., Huttunen, P., 1995. Aquatic macrophytes and ecological gradients in 57 small lakes in southern Finland. Aquatic Botany 51, 197–221. Trebitz, A.S., Brazner, J.C., Cotter, A.M., Knuth, M.L., Morrice, J.A., Peterson, G.S., Sierszen, M.E., Thompson, J.A., Kelly, J.R., 2007. Water quality in Great Lakes coastal wetlands: basin-wide patterns and responses to an anthropogenic disturbance gradient. Journal of Great Lakes Research 33 (3), 67–85 (special issue). Tuchman, N., Larkin, D., Geddes, P., Wildova, R., Jankowski, K., Goldberg, D., 2009. Patterns of environmental change associated with Typha × glauca invasion in a Great Lakes coastal wetland. Wetlands 29, 964–975. Turner, J.V., Townley, L.R., 2006. Determination of groundwater flow-through regimes of shallow lakes and wetlands from numerical analysis of stable isotope and chloride tracer distribution patterns. Journal of Hydrology 320, 451–483. USEPA, 1983. Methods for Chemical Analyses of Waters and Wastes. U.S. Environmental Protection Agency, Cincinnati, OH. USEPA, 1991. The Determination of Inorganic Anions in Water by Ion Chromatography. U.S. Environmental Protection Agency, Cincinnati, OH. USEPA, 1993. Methods for Determination of Inorganic Substances in Environmental Samples. U.S. Environmental Protection Agency, Cincinnati, OH. USEPA, 2006. Research Centers: Great Lakes Environmental Indicators Project. http://cfpub.epa.gov/ncer abstracts/index.cfm/fuseaction/outlinks.centers/ center/125. USEPA, 2010. Clean Water Act Section 401 Water Quality Certification: A Water Quality Protection Tool for States and Tribes. U.S. EPA Office of Wetlands, Oceans, and Watersheds, Washington, DC, p. 45. van Breemen, N., 1995. How Sphagnum bogs down other plants. Trends in Ecology & Evolution 10, 270–275. Verhoeven, J.T.A., Arheimer, B., Yin, C., Hefting, M.M., 2006. Regional and global concerns over wetlands and water quality. Trends in Ecology & Evolution 21, 96–103. Verhoeven, J.T.A., Schmitz, M.B., 1991. Control of plant growth by nitrogen and phosphorus in mesotrophic fens. Biogeochemistry 12, 135–148. Vitt, D.H., Chee, W.-L., 1990. The relationships of vegetation to surface water chemistry and peat chemistry in fens of Alberta, Canada. Plant Ecology 89, 87–106. Werner, K., Zedler, J., 2002. How sedge meadow soils, microtopography, and vegetation respond to sedimentation. Wetlands 22, 451–466.