Wetlands

, Volume 32, Issue 3, pp 411–422

Multiple Scales of Influence on Wetland Vegetation Associated with Headwater Streams in Alaska, USA

Authors

    • Smithsonian Environmental Research Center
  • Coowe M. Walker
    • Kachemak Bay Research Reserve
  • Ryan S. King
    • Center for Reservoir and Aquatic Systems Research, Department of BiologyBaylor University
  • Steven J. Baird
    • Kachemak Bay Research Reserve
Article

DOI: 10.1007/s13157-012-0274-z

Cite this article as:
Whigham, D.F., Walker, C.M., King, R.S. et al. Wetlands (2012) 32: 411. doi:10.1007/s13157-012-0274-z

Abstract

Vegetation of wetlands adjacent to headwater streams on the Kenai Lowlands was dominated by Calamagrostis canadensis, indicating that it is a keystone species that influences stream-wetland interactions across a wide range of geomorphic settings from which headwater streams have their origin. We sampled 30 sites as part of a project to determine the relationships between landscape features and the biological and chemical characteristics of headwater streams and their associated wetlands. In this paper we consider vegetation in wetlands adjacent to headwater streams. Calamagrostis canadensis was the only species that occurred at all sites and only a few species were widespread and abundant across the range of sites sampled. Nonmetric multidimensional scaling of species importance values indicated that the distribution of sites and species was primarily related to stream-reach scale environmental and biological factors. Sixteen stream-reach factors were significantly correlated with the distribution of sites and species on one axis of the ordination. Headwater streams that were located in relatively flat areas with extensive wetlands had species characteristic of nutrient poor wetlands and sites located in steep valleys with narrow wetlands had species characteristic of uplands and wetlands on mineral soils. The distribution of sites and species on the second ordination axis was interpreted to be a response to biological interactions; primarily the negative relationship between C. canadensis and the diversity of other species. We concluded that large-scale watershed features of the landscape are less important than local scale factors in determining the characteristics of vegetation in headwater stream-wetland complexes in the Kenai Lowlands. There was no evidence, however, that differences in the stream-reach scale conditions across the study sites resulted in distinct plant communities associated with the headwater wetlands even though the headwater streams had their origin in different landscape settings.

Keywords

AlaskaWetlandStreamsideCalamagrostis canadensisKenai PeninsulaHeadwater streams

Introduction

Headwater streams account for most of the length in stream networks (Leopold et al. 1964; Naiman 1983; Benda et al. 2005) and because of a high degree of connectivity with surrounding landscapes they have been referred to as the equivalent of the fine branches of human lungs, the primary system across which important exchanges occur (Lowe and Likens 2005). Headwater streams are crucial for maintaining the integrity and ecological function of river networks for many reasons (Meyer et al. 2007), including the transport of biota (e.g., invertebrates) and organic matter and the influence that they have on water quality (Wipfli et al. 2007).

Lateral linkages between headwater streams and adjacent uplands and wetlands are also important and influence the structure and function of headwater streams (Ward 1989; Wallace et al. 1997; Meyer et al. 2007). Wetlands immediately adjacent to headwater streams influence water quality by serving as buffers that intercept and store or process nutrients from adjacent upland habitats (e.g., Brinson 1993; Weller et al. 1996). Adjacent wetlands are also important sources of allochthonous inputs to headwater streams such as plant litter, woody debris, and terrestrial invertebrates (Richardson et al. 2005; Meyer et al. 2007; Shaftel et al. 2011), and they are part of the continuum of hydrologic connectivity between uplands, wetlands, and stream networks (Nadeau and Rains 2007).

Most studies of streamside wetlands have focused on the impacts that disturbances (e.g., fire, logging, vegetation removal) have on the quality and quantity of litter, nutrients, and sediment inputs to headwater streams (Jackson and Sullivan 2009; Muto et al. 2009; Binckley et al. 2010; Kiffney and Richardson 2010). A few studies have focused on the extent and functions of headwater wetlands in different geologic and geomorphic settings. Janisch et al. (2011) studied 30 headwater catchments in Washington that were distributed across two different lithographic settings. They found that wetlands associated with headwater streams were common and extensive and had the potential to play a key role in controlling ecological processes in headwater streams, especially in landscapes disturbed by logging. Devito et al. (1999) found that the export of sulfate from wetlands associated with headwater streams varied with geomorphic conditions (i.e., lower export from wetlands that had developed on till >1 m deep). Rheinhardt et al. (1998) was the only study that we are aware of that examined differences in the plant species composition of forested wetlands associated with headwater streams. They found differences in the tree species associated with floodplains of different ordered streams but little variation within each stream order sampled.

Alaska is an ideal and important environment to study the relationships between landscape settings and the linkages between headwater streams and adjacent wetlands because they are linked to streams and rivers that support sustainable salmon fisheries. Recently, significant effort has been directed towards developing a GIS-based wetland classification, called the Kenai Lowlands Wetland Management Tool (KWMT), with the intention that this spatially explicit wetland map could be used as the framework for wetland management decisions in the Kenai Lowlands area (Gracz et al. 2008). This tool identifies and maps eight different wetland types at a scale of 1:25,000 on the Kenai Lowlands, on the southern half of the Kenai Peninsula. Ongoing work is being devoted to assigning functions to these wetlands (Homer Soil and Water District 2012) as an aid to management and conservation across such a vast area.

We conducted studies to quantify the relationships between landscape features and the physical, chemical and biotic characteristics of headwater streams and their associated wetlands on the Kenai Lowlands. Results from investigations into the effects of alder (Shaftel et al. 2011, 2012), fish and stream macroinvertebrate communities (Dekar et al. 2012; King et al. 2012) and stream chemistry (Walker et al. 2012) are reported separately. The work that we report here presents the results from our studies on the species composition of wetland plant communities adjacent to headwater streams. We predicted that the vegetation of wetlands associated with headwater streams would differ across the range of geomorphic settings through which headwater streams flowed. Wetland vegetation associated with headwater streams that originated in broad low nutrient wetlands with deep organic soils that formed in depressions would, for example, be dominated by poor-fen or bog related species. Species that were more typical of rich-fens would be associated with streams that originated in groundwater fed discharge slopes. We anticipated that understanding how different wetland plant communities affect headwater streams could provide insights into the ecological connections between adjacent landscapes and headwater streams, which in addition to being inherently valuable, provide rearing habitat for salmon (King et al 2012), which are an important economic and ecological force in the Kenai lowlands region.

Methods

Study Area and Site Selection

The Kenai Lowlands occupy approximately 9,400 km2 between Cook Inlet and the Kenai Mountains in south-central, Alaska (Fig. 1). Elevations vary from sea level to almost 1,000 m and the climate is transitional between maritime and continental. Climate varies across the study area, especially winter temperatures and annual precipitation patterns, as shown by a comparison (source: http://cdo.ncdc.noaa.gov/cgi-bin/climatenormals/climatenormals.pl ) of the two cities at the southern (Homer) and northern (Kenai) ends of the study area. Average winter (January) and summer (July) temperatures are −5.2°C and 11.9°C in Homer and −10.3°C and 12.8°C in Kenai. Mean annual precipitation in Homer is 61.7 cm compared to 48.1 cm in Kenai. The seasonal distribution of precipitation also differs. Mean monthly precipitation of more than 5 cm occurs over a three month period (August–October) in Kenai compared to a seven month (August–February) period in Homer. Our study area within the Kenai Lowlands included headwater streams that were distributed across five major drainages that support salmon runs: South and North Forks of the Anchor River, Deep Creek, Stariski Creek, and Ninilchik River.
https://static-content.springer.com/image/art%3A10.1007%2Fs13157-012-0274-z/MediaObjects/13157_2012_274_Fig1_HTML.gif
Fig. 1

Location of the Kenai Lowland study area (outlined in black) and the 30 sample site locations (white circles) in south-central Alaska

Our goal in selecting headwater streams to study was to distribute them among sites that were representative of the range of topographic and wetland conditions that have been described and mapped for the four river systems (Gracz et al. 2008). Detailed site selection procedures are described in Walker et al. (2012) and King et al. (2012) and are summarized here. ArcGIS Spatial analyst in ArcGIS 9.3 (ESRI, Inc., Redlands, CA) was used to identify first-order streams that were potential study sites, defined in terms of accessibility as streams within 3 km of a road, in the five drainage systems from USGS DLG data (1:63,360 topographic maps). The major wetland class or classes at each potential study site were classified based on wetland maps in Gracz et al. (2008). The series of analyses and graphical evaluations that were used resulted in the identification of four abundant and widespread wetland categories associated with headwater streams (King et al. 2012; Walker et al. 2012). Stratified random sampling, was used to select and distribute 30 study sites (Fig. 1) across the four wetland categories and five drainage basins. The sites were sampled between mid-May and late July 2006.

Wetland Vegetation Sampling

At each of the 30 sites we established a 200 m long by 100 m wide study area with the headwater stream positioned in the middle of the long axis. The 200 m long stream reach was divided into four 50 m segments. At the midpoint of each 50 m segment we estimated the cover of all herbaceous plants, defined as herbaceous species or woody plants <1 m in height, in two 1 × 1 m quadrats that were located 5 m from and perpendicular to the stream bank, one on either side of the stream. Importance Values (IVs) were calculated for each species in each study area using frequency and cover data. Species identifications in the field and laboratory were based on Hultén (1968) and several field manuals that included keys and descriptions of species that would occur in the Kenai Lowlands (e.g., Pojar and MacKinnon 1994a,b; Johnson et al. 1995). The names that are reported here are based on the current USDA plant list (http://plants.usda.gov/java/downloadData?fileName=plantlst.txt&static=true)

Habitat Characterization of Study Areas

As described in Walker et al. (2012), we expected that the species composition of the wetlands associated with the headwater streams would potentially be influenced by characteristics of the landscape at three scales: 1. the watershed upstream of the study area (hereafter referred to as watershed-scale), 2. habitats that the stream flowed through (hereafter referred to as plot-scale), 3. physical and chemical characteristics of the stream channel (hereafter referred to as stream-reach scale) .

Watershed-scale Characteristics

We used GIS to calculate 16 variables that characterized the drainage area upstream of the study area. Procedures used to calculate the watershed-scale variables that were used in the ordinations (described below) were described in Walker et al. (2012) and are listed in Online Resource 1.

Plot-scale Characteristics

Two types of data (habitat, woody vegetation) were compiled in the field to characterize conditions within each 200 × 100 m study area. Habitat metrics were the percentages of each study area occupied by wetland, open water (unvegetated), and upland habitats. We mapped the extent of the three habitat types in the field using a combination of meter tapes and hand-held range finders, resulting in a map of the 200 m stream reach that was drawn onto a gridded field-form. The maps were scanned and the percentages of the study area occupied by each habitat type were determined.

Woody vegetation (trees and shrubs) was sampled in eight (two per segment) nested transects in each study area. Three tree (defined as individuals with a DBH >10 cm) categories (living, standing dead, fallen) were sampled. Living trees in each segment were sampled in two 20 m wide by 50 m long transects. Each transect had the origin of the long axis positioned at the creek bank. We determined the identity of each tree and measured the DBH. The DBH of standing dead and fallen trees in each segment was measured in two 20 × 20 m transects that were nested within the 20 × 50 m transects and also had the origin of the long axis at the creek bank. The size of the transect used to measure the two dead tree categories was based on preliminary measurements of tree height which showed that trees within 20 m of the stream had the potential to fall across or into the stream and contribute large woody debris to the stream ecosystem. The number of individual stems of each shrub species (woody plants with a DBH <10 cm and a height >1.5 m) was determined in two 2 × 10 m plots that had their origin at the creek bank and were nested inside the 20 × 20 m plots.

DBH measurements were converted to basal area (m2ha−1). Basal area, density, and frequency data were used to calculate IVs for the three tree categories in each study area. Density and frequency data were used to calculate IVs for shrubs.

In addition to the habitat and woody plant variables, we included the type of wetland that had been mapped by Gracz et al. (2008) as a plot-scale variable. The plot-scale variables used in the ordination analysis are listed in Online Resource 1.

Stream-reach Characteristics

Chemical and physical characteristics of the stream that flowed through each study area were determined from water samples collected at the upstream end of the stream reach and physical measurements made at 11 evenly spaced locations along each stream. Procedures used to collect and analyze water samples are described in Walker et al. (2012) and procedures used to sample the physical characteristics of each stream (i.e., stream channel morphology, substrate composition, amount of woody debris, etc.) are described in King et al. (2012). The chemical and physical characteristics of the stream that were used in the ordination analyses (described below) are listed in Online Resource 1.

Ordination Analysis

Species IVs were used to compare the herbaceous wetland vegetation at the 30 sites using nonmetric multidimensional scaling (NMS) program in PC-ORD (version 5; McCune and Medford 2006). Species that were present in three or fewer plots (36 taxa) were excluded from the final analysis, resulting in a comparison of 48 species across the 30 study sites. The final version of the ordination that was selected was based on 58 iterations, a step length of 0.20, and using Sørensen (= Bray-Curtis) dissimilarity to calculate the distance matrix. The final stress for the 2-dimensionality solution was 13.7. The species and axis scores from the final version of the NMS ordination were used to compare relationships with the watershed-, plot-, and stream-reach variables (McCune and Grace 2002). A cutoff r2 value of 0.20 was used to select watershed-, plot- and stream-reach variables (based on Pearson and Kendall correlations with the ordination axes) that were correlated with the two ordination axes.

Regression Analysis

We regressed the number of plant species versus the IV of Calamagrostis canadensis, the only species present at all 30 sites (see Results). Calamagrostis canadensis was so prevalent that we hypothesized that factors positively influencing its cover were also important, negative drivers of species richness in wetlands associated with headwater streams. We used a generalized additive model (GAM) to fit the relationship between number of species and C. canadensis IV because the distribution of the response was negative binomial (O’Hara and Kotze 2010) and graphical analysis implied that the response was nonlinear and warranted a model that did not restrict the response to a linear model (Zuur et al. 2009). GAMs were fit using the mgcv library in R 2.9.2 (Wood 2008; R Development Core Team 2010). The fitted response (GAM smoother) was deemed significant if p < 0.001 (Zuur et al. 2009).

Results

Eighty-four taxa, including mosses and Cyperaceae that were not identified to species, were recorded in the 240 plots and 36 of the taxa were recorded at ≤3 of the 30 sites and were not included in the final NMS ordination. Calamagrostis canadensis was the only species that was recorded at all sites and 67% of the taxa occurred in less than 10% of the plots (Online Resource 2). Equisetum arvense (24), Polymonium acutiflorum (21),Chamerion angustifolium (20), Salix barclayi (19), Sanguisorba canadensis (17), Comarum palustre (16), Thalictrum sparsiflorum (16), and Trientalis europaea (15) were the only other species recorded at 50% of more of the sites (Online Resource 2). Only C. canadensis (214) and E. arvense (146) were present in ≥50% of the 240 plots sampled. Three species (Salix barclayi, Chamerion angustifolium, Sanguisorba canadensis) were recorded at 50% of more of the sites and were present in 25–50% of the plots (103, 93, and 86, respectively). Eleven species were tallied in 10–24.9% of the plots and they occurred at between 5 and 16 of the 30 sites (Online Resource 2).

The two-solution ordination of the IV data explained 86.9% of the total variance and the amounts of variance explained by the two axes was similar (44.5% = Axis 1; 42.4% = Axis 2). There was no clear separation of sites into distinct groupings and sites were distributed almost continuously along both axes (Fig. 2a). The distribution of species scores on the two axes (Table 1) and the correlations between plot-, watershed- and stream-reach variables (Table 2) indicated that Axis 1 was primarily related to environmental conditions and Axis 2 was most likely an expression of species interactions.
https://static-content.springer.com/image/art%3A10.1007%2Fs13157-012-0274-z/MediaObjects/13157_2012_274_Fig2_HTML.gif
Fig. 2

Results of NMS ordination of 30 study sites based on importance values of species recorded in 1 × 1 m herb plots (a) and biplots of plot- (b), watershed- (c) and stream-scale (d) variables that were measured at each of the study sites. The orientation (+ or −) and magnitude of the vectors is represented by the arrows in b–d and the correlations coefficients are provided in Table 2. Significant correlations (r2 ≥ 0.2) are: b: + on Axis 1 = Dead T, Upland % and Down T; c: - on Axis 2 Elev.m; d: + on Axis 1 = Sand, Cobb, Grav.log, DO.mg, Silt, pH, and Grav.sm; – on Axis 1 = Depth.xs, Thalweg, Fines, Embed; + on Axis 2 = Overveg, Canopy, SWD; - on Axis 2 = Fines

Table 1

Species scores sorted by decreasing axis scores for the first and second NMS ordination axes for 30 study sites in the Kenai Lowlands. Only species that were recorded in more than three plots are included in the Table

Species

Axis 1

Species

Axis 2

Urtica dioica

1.11

Rumex arcticus

0.88

Heracleum maximum

0.77

Spiraea stevenii

0.78

Rubus idaeus

0.71

Swertis perennis

0.78

Streptopus amplexifolius

0.69

Viola spp.

0.76

Athyrium filix-femina

0.62

Anemone Richardsonii

0.72

Alnus incana ssp. tenuifolia

0.54

Vaccinium oxycoccos

0.71

Dryopteris expansa

0.49

Achillea millefolium

0.69

Equisetum sylvaticum

0.43

Salix planifolia

0.67

Sambucus racemosa

0.40

Pyrola asarifolium

0.66

Chamerion angustifolium

0.36

Valeriana capitata

0.63

Stellaria borelis

0.31

Mosses

0.56

Gymnocarpium dryopteris

0.30

Sanguisorba canadensis

0.56

Geranium erianthum

0.29

Salix barclayi

0.55

Galium trifidum

0.29

Rubus arcticus

0.50

Ribes triste

0.27

Cardamine oligosperma

0.48

Angelica genuflexa

0.26

Streptopus amplexifolius

0.48

Calamagrostis canadensis

0.22

Angelica genuflexa

0.43

Cardamine oligosperma

0.22

Aconitum delphinifolium

0.43

Thalictrum sparsiflorum

0.21

Ribes laxiflorum

0.41

Viola spp.

0.21

Dryopteris expansa

0.38

Ribes laxiflorum

0.15

Equisetum arvense

0.31

Viola epipsila

0.15

Athyrium felix-femina

0.28

Aconitum delphinifolium

0.07

Equisetum sylvaticum

0.28

Equisetum arvense

0.07

Viola epipsila

0.26

Valeriana capitata

0.07

Thalictrum sparsiflorum

0.25

Trientalis europaea

0.04

Polymonium acutiflorum

0.22

Sanguisorba canadensis

0.01

Stellaria borealis

0.22

Salix barclayi

−0.02

Geranium erianthum

0.21

Anemone Richardsonii

−0.03

Ribes triste

0.21

Betula papyrifera

−0.05

Alnus incana spp. tenuifolia

0.18

Rumex arcticus

−0.06

Gymnocarpium dryopteris

0.16

Pyrola asarifolium

−0.09

Trientalis europaea

0.16

Spiraea stevenii

−0.09

Rubus idaeus

0.13

Equisetum palustre

−0.10

Galium trifidum

0.12

Polemonium acutiflorum

−0.14

Empetrum nigrum

0.12

Rubus arcticus

−0.19

Betula papyrifera

0.11

Equisetum fluviatile

−0.20

Chamerion angustifolium

0.09

Achillea millefolium

−0.25

Urtica dioica

0.08

Mosses

−0.33

Heracleum maximum

0.06

Carex aquatilis

−0.34

Sambucus racemosa

−0.02

Sewertia perennis

−0.46

Betula nana

−0.09

Empetrum nigrum

−0.52

Sphagnum spp.

−0.27

Vaccinium oxycoccus

−0.55

Comarum palustre

−0.38

Sphagnum spp.

−0.61

Calamagrostis canadensis

−0.29

Comarum palustre

−0.72

Equisetum fluviatile

−0.56

Salix planifolia spp. pulchra

−1.33

Carex aquatilis

−0.78

Betula nana

−1.28

Carex spp.

−0.98

Carex spp.

−1.31

Equisetum palustre

−1.39

Table 2

Correlations coefficients (r) between ordination axes and plot-, watershed -, and stream-reach variables for 30 headwater streams of the Lower Kenai Peninsula, Alaska. Variables with r2 > 0.2 are highlighted in bold. Variables are sorted in rank order along Axis 1 and they are described in the text and in more detail in King et al. (2012)

Scale

Variable

Axis 1

Axis 2

Plot

Dead trees (Dead T)

0.563

0.146

Percent upland (Upland %)

0.479

0.022

Fallen trees (Down T)

0.481

0.021

Live trees (Live T)

0.368

0.008

Shrubs (Shrub)

0.281

0.391

Percent water (Water %)

−0.339

−0.124

Percent wetland (Wet %)

−0.370

0.014

Wetland type (Wettyp)

−0.128

0.092

Watershed

% riparian wetland (RP)

0.382

0.083

Flow weighted slope (FWslope)

0.312

0.235

Mean slope of watershed (Wslope)

0.262

0.280

% discharge slope wetlands (DS)

0.258

−0.081

% kettle wetlands (KT)

0.160

0.035

Total wetland cover (Wet %)

0.118

−0.234

Site elevation (Elev.m)

0.109

−0.528

Watershed size (Shed.KM2)

0.067

0.025

% relict glacial lakebed wetlands (LB)

−0.077

−0.319

Topographic wetness index (Wetx)

−0.281

−0.342

% relict glacial drainageway wetlands (DW)

−0.352

−0.157

Stream-Reach

Substrate – sand (Sand)

0.606

−0.007

Substrate – cobble (Cobb)

0.591

0.442

Substrate – large gravel (Grav.lg)

0.564

0.206

Dissolved oxygen (DO.mgl)

0.552

0.289

Substrate – silt (Silt)

0.515

−0.346

pH (pH)

0.504

0.384

Substrate – small gravel (Grav.sm)

0.503

0.121

Overhanging vegetation (Overveg)

0.437

0.537

Substrate – small boulders (Bould.sm)

0.394

0.387

Canopy density (Canopy)

0.386

0.561

Periphyton (Peri)

0.341

0.368

Large woody debris (LWD)

0.338

−0.008

Small woody debris -(SWD)

0.249

0.585

Roots of living plants (Roots)

0.204

0.123

Substrate – large boulders (Bould lg)

0.066

0.435

Bank height (Bk.ht)

−0.206

−0.199

Stream width (Width)

−0.223

0.099

Stream cross section depth (Depth.xs)

−0.572

−0.310

Thalweg (Thalweg)

−0.598

−0.233

Substrate – fines (Fines)

−0.638

−0.536

Substrate - embedded (Embed)

−0.643

−0.192

Sediment – peat (Peat)

−0.705

−0.146

Species scores were continuously distributed on Axis 1 with 27 having positive scores and 21 with negative scores (Table 1). The distribution of the species scores, however, suggested a gradient of increasing wetness from right to left on Axis 1 was indicated by the indicator status and distribution of species scores. Species (e.g., Urtica dioica, Heracleum maximum, Rubus idaeus, Streptopus amplexifolius, Athyrium filix-femina, Alnus incana ssp. tenuifolia, Dryopteris expansa) with the highest positive Axis 1 scores (Table 1) are widely distributed throughout Alaska (Hultén 1968) and have a FAC or FACU indicator status (http://www.kenaiwetlands.net/AKWetlandIndicatorStatus.html). Species (e.g., Betula nana, Comarum palustre, Vaccinium oxycoccus) with the highest negative scores on Axis 1 (Table 1) are also widely distributed in Alaska (Hultén 1968) and all but B. nana have a FACW or OBL indicator status.

The distribution of species scores on Axis 2 (Table 1) did not seem to be related to any obvious environmental gradient (e.g., upland-wetland, high to low elevation) and the distribution pattern was very different from Axis 1. Only 9 species, including C. canadensis, had negative scores on Axis 2 compared to 39 species with positive scores. The most interesting patterns associated with Axis 2 were the relationships between Axis 2 scores and the IV of C. canadensis (Fig. 3) and the negative relationship between the number of species in plots and the IV of C. canadensis (Fig. 4). We interpreted Axis 2 to be primarily a gradient that was influenced by biological processes, as more fully described in the Discussion.
https://static-content.springer.com/image/art%3A10.1007%2Fs13157-012-0274-z/MediaObjects/13157_2012_274_Fig3_HTML.gif
Fig. 3

The distribution of Calamagrostis canadensis IV values on Axis 2

https://static-content.springer.com/image/art%3A10.1007%2Fs13157-012-0274-z/MediaObjects/13157_2012_274_Fig4_HTML.gif
Fig. 4

Plot of relationship between importance values of Calamagrostis canadensis and the number of plant species of each site (n = 30). Lines represent the fitted response (solid) and 95% confidence interval (CI; dotted) estimated using a negative binomial generalized additive model (GAM; p < 0.0001, variance explained = 54.9%)

The distribution of sites (Fig. 2a) was correlated with 2 plot-scale variables, 1 watershed-scale variable and 16 stream-reach variables (Table 2) and the direction and magnitude of the correlations are shown in Fig. 2b–d. When taken together, the plot- , watershed- and stream-reach correlations indicate that the distribution of sites on Axis 1 was primarily a gradient of the physical settings associated with the streams. Sites toward the right on Axis 1 were mostly better drained sites located in high gradient valleys where the streams and associated wetlands were narrower and the adjacent uplands had higher densities of living, dead, and downed trees (Fig. 2b). Sites toward the left of Axis 1, especially the four sites that are separated from the main distribution of sites (Fig. 2a), were located in broader valleys and the streams were associated with larger wetlands that had less woody vegetation near the streams.

The large number of stream-reach variables associated with Axis 1 and the directions of the correlations (Fig. 2d, Table 2) suggest that the streamside vegetation was primarily influenced by stream-wetland related processes. Sites toward the left of Axis 1 had characteristics of low gradient streams that flow through large wetlands with deep organic substrates. The streams were deeper, had smaller thalweg values, and the substrates were more embedded with higher percentages of fines and peat (Fig. 2d, Table 2). In contrast, streams associated with sites to the right of Axis 1 had characteristics of higher gradient systems such as substrates composed of cobbles, large and small gravel, small boulders and silt (Fig. 2d, Table 2). The streams toward the right of Axis 1 also had higher dissolved oxygen and pH, characteristics of fast flowing streams with significant groundwater contact.

Discussion

Wetlands in the Kenai Lowlands account for more than 40% of the total land area (Gracz et al. 2008) and the four most abundant and widespread wetland categories sampled in this project (relict glacial drainageways, relict glacial lakebeds, discharge slopes, riparian) are dominated by distinct plant communities that occur on different glacially influenced landforms. Given the extent, diversity and range of landforms (Karlstrom 1964) that support wetlands in the Kenai Lowlands, we expected that there would be differences in the chemistry and biota of headwater streams that reflected differences in the sources of surface and groundwater flows (King et al. 2012; Walker et al. 2012). Some of the headwater streams sampled in this study, for example, had their origin in large bog and fen complexes (e.g., relict glacial drainageways and relict glacial lakebeds) that were located in areas with almost level slopes and deep organic soils.

Analysis of the water quality and faunal characteristics of the 30 headwater streams resulted in two distinct categories of sites distributed on a wetness gradient (King et al. 2012; Walker et al. 2012). Headwater streams on sites with higher flow weighted slopes (higher gradient streams in steeper slope positions with lower amounts of wetland in the upstream watershed) had substrates with more gravel and cobble and water with higher dissolved oxygen, nitrate, and lower temperatures compared to sites with lower flow weighted slopes (Walker et al. 2012). Wetter sites associated with lower slopes had streams with higher levels of dissolved organic carbon, higher temperatures, more dissolved N in the form of ammonium and lower dissolved oxygen (Walker et al. 2012). Based on invertebrate and fish taxa, the 30 sites were similarly divided into two broad groupings. Streams at wetter sites with lower flow-weighted slopes had fewer but larger coho salmon and Dolly Varden and invertebrate taxa were dominated by species characteristic of lentic habitats (King et al. 2012). Streams with higher flow-weighted slopes had invertebrate taxa characteristics of lotic habitats and higher abundances of smaller coho and Dolly Varden.

We expected that there would be differences in the vegetation associated with the headwater wetlands associated with the flow-weighted slope gradient. We found that the 30 sites did not separate into the two distinct categories based on the wetland taxa and there was an almost continuous distribution of sites on ordination Axis 1 (Fig. 2a). The relationships between the 30 sites and the plot-, watershed- and stream-reach environmental variables (Fig. 2b–d, Table 2) supported the interpretation that the distribution of sites on Axis 1 was primarily an environmental gradient that integrated features of site wetness. Sites with increasingly positive Axis 1 scores were located in steeper valleys with narrower streamside wetlands in a matrix dominated by uplands with more woody vegetation and streams that had higher dissolved oxygen, higher pH, and substrates dominated by cobbles and boulders (Fig. 2b,d, Table 2). The two sites that were distributed at the far right of Axis 1 represent one end of the wetness gradient. Calamagrostis canadensis was a dominant species at both sites (IV = 55.4 and 46.2) but the most striking features of the two sites was the mix of upland and wetland species. Urtica dioica and Heracleum maximum were both listed as FACU species in Alaska (National List of Vascular Plant Species that Occur in Wetlands: 1996 National Summary (http://library.fws.gov/Pubs9/wetlands_plantlist96.pdf) and they had importance values of 37.6 and 31.7, respectively at one of the sites. The presence of the fern Phegopteris connectilis and seedlings of Populus balsamifera ssp. trichocarpa along with the relatively high IV of Alnus incana ssp. tenuifolia at those two sites also were indicative of the upland-wetland characteristics of streamside vegetation in steep narrow valleys. Athyrium felix-femina also occurred at both sites and Phegopteris and Populus were listed as facultative upland species in Alaska. Alnus and Athyrium are listed as facultative species and Gracz et al. (2008) found that they occurred equally in upland and wetland habitats throughout the Kenai Lowlands.

At the opposite end of Axis 1 and positioned to the left of the other sites in the ordination (Fig. 2a) were four sites that were located in broad valleys with low gradient streams embedded in large wetland complexes (Walker et al. 2012). Streams at those sites had deeper channels and the substrates were composed of organic matter and finer sediments (Fig. 2d, Table 2). Vegetation at the four sites was characterized by the presence and relatively high IVs of Betula nana, Comarum palustre, and Salix planifolia ssp. pulchra, the latter only occurring at 3 of the 30 study sites. While these three species had different indicator statuses in Alaska (i.e., B. nana is facultative, C. palustre is obligate and S. planifolia ssp. pulchra is facultative wet), they are all characteristic of poor fens in the Kenai Lowlands (Gracz et al. 2008). Low and wet areas in oligotrophic to mesotrophic poor fens and bogs are well known habitats for C. palustre across its range in the northern hemisphere (e.g., Pietsch 1991; Närhi et al. 2010). Salix planifolia ssp. pulchra, and uncommon species in AK (Gracz et al. 2008), only occurred at three of the four sites but it had relatively high importance values (46.4 and 16.3) at two of them.

Axis 2 accounted for as much of the variance in the NMS ordination as Axis 1 and the 30 sites were distributed continuously along Axis 2 (Fig. 2a). We interpreted the distribution of sites on Axis 2 to be primarily related to interactions between C. canadensis and other herbaceous species as few of the environmental variables (Table 2) were significantly related to Axis 2 (n = 5) compared to Axis 1 (n = 16). There was a clear trend of decreasing IVs of C. canadensis from the bottom to the top of Axis 2 (Fig. 3) and the decreasing IV of C. canadensis was correlated with an increase in the number of species in the study plots (Fig. 4). The relationship was non-linear and indicated that the decline in the number of species was almost linear up to a C. canadensis IV of approximately30 and then a sharp drop beyond an IV of approximately 75.

Most studies of the relationships between C. canadensis and other species have focused on the effects that increasing density of C. canadensis exerts on tree species (Lieffers et al. 1993; Hangs et al. 2002; Landhäusser et al. 2007; Matsushima and Chang 2007a,b) but it has also been shown to be a strong competitor of herbaceous species (Landhäusser and Lieffers 1994). The importance of C. canadensis has increased across the Kenai Lowlands in areas where trees have been killed by the spruce bark beetle (Boucher and Mead 2006; Werner et al. 2006; Boggs et al. 2008) as evidenced by the many instances where we found that C. canadensis had completely overgrown downed spruce trees and had formed vegetation covered bridges over streams (Fig. 5a, b).
https://static-content.springer.com/image/art%3A10.1007%2Fs13157-012-0274-z/MediaObjects/13157_2012_274_Fig5_HTML.gif
Fig. 5

Two headwater streams (a and b) in which herbaceous species (primarily Calamagrostis canadensis) had started to cover the streams. The process of overtopping the stream was often initiated by tree trunks that have fallen across the stream (b). The impact of high spring runoff on near stream habitats (c)

The importance of competition between C. canadensis and other species suggested by this study, and supported by studies of its interactions with trees and other herbaceous species (references in previous paragraph), was only correlative and additional research would be needed to confirm our interpretation. What was clear, however, is that C. canadensis is potentially a keystone species in many of the interactions that occur between wetlands associated with headwater streams and processes that occur in the streams.

Our initial research on the ecology of headwater streams system in the Kenai Lowlands has provided further evidence that the most of the nitrogen that occurs in stream biota had its origin in upland habitats dominated by species of Alnus (e.g., Helfield and Naiman 2002; Compton et al. 2003; Shaftel et al. 2012; Dekar et al. 2012) and that C. canadensis is an important source of organic matter to headwater streams (Shaftel et al. 2012). Questions remain about ecological functions that link headwater streams in the Kenai Lowlands with adjacent wetlands; especially as they relate to abundance of C. canadensis. It was not surprising, for example, that C. canadensis had high IVs in many of the sites such as riparian and discharge slope wetlands as mapped by Gracz et al. (2008). Calamagrostis canadensis is a common species in uplands and wetland habitats with organic soils over shallow mineral soils (Landhäusser and Lieffers 1999). What was surprising was its abundance and restriction to areas immediately adjacent to headwater streams in wetlands with deep organic soils such as kettles. Vegetation in those sites was dominated by species characteristic of low nutrient environments (e.g., mats of Sphagnum spp., Drosera rotundifolia, Rubus chamaemorus, Betula nana, Andromeda polifolia). In these settings, C. canadensis only occurred in restricted areas immediately adjacent to the headwater streams that we sampled, but we observed that soils in those areas had higher mineral content, including visible accumulations of ash from volcanic eruptions in the region. Mineral sediments were most likely delivered to the headwater stream habitats during flooding events associated with spring thaws that are clearly important hydrologic events associated with headwater streams (e.g., Fig. 5c). Gracz et al. (2008) also suggested that disturbances (e.g., roads and culverts associated with logging or recreation) across peat-dominated wetlands may be lead to sediment input and the establishment of C. canadensis. Other questions related to wetland functions and C. canadensis are also important. How productive are the wetlands over the range of wetness associated with headwater streams, how much of the C. canadensis production associated with creek bank habitats enters the stream as organic matter, and how important is Alnus as a source of nitrogen that supports production of the streamside wetlands, especially C. canadensis? These issues will be the focus of future articles based on our studies of headwater streams as critical habitats for anadromous fish in the Kenai Lowlands and the dynamic and critical linkages between headwater streams and adjacent habitats.

Acknowledgments

The project was funded through US EPA’s Wetland Program Development Grant program. Staff of the Kachemak Bay Research Reserve (Amy Alderfer, Kim Donohue, Conrad Field) and several volunteers (Jennifer Brewer, Shan Burson, Patrick Dougherty, Dwayne Evans, Rachel Hovel, Megan Murphy, Caitlin Schott, Simeon Smith, Scott Thompson) assisted with various parts of the project. Jeff Back deserves special recognition because of his good spirits and efforts to keep the field teams on course and on schedule under a wide range of climate conditions. We also thank the Ninilchik Tribal Association and private landowners for providing access to sampling sites. We especially thank Phil North for his encouragement, support and efforts to make the project a reality. We thank Jay O’Neill, Kathy Boomer, Mike Gracz and three anonymous reviewers for their editorial suggestions.

Supplementary material

13157_2012_274_MOESM1_ESM.doc (42 kb)
Online Resource 1(DOC 42 kb)
13157_2012_274_MOESM2_ESM.doc (93 kb)
Online Resource 2(DOC 93 kb)

Copyright information

© US Government 2012