Effects of an invasive cattail species (Typha×glauca) on sediment nitrogen and microbial community composition in a freshwater wetland


  • Editor: Hermann Bothe

Correspondence: John J. Kelly, Department of Biology, Loyola University Chicago, Chicago, IL 60626, USA. Tel.: +773 508 7097; fax: +773 508 3646; e-mail: jkelly7@luc.edu


Sediments from Cheboygan Marsh, a coastal freshwater wetland on Lake Huron that has been invaded by an emergent exotic plant, Typha×glauca, were examined to assess the effects of invasion on wetland nutrient levels and sediment microbial communities. Comparison of invaded and uninvaded zones of the marsh indicated that the invaded zone showed significantly lower plant diversity, as well as significantly higher aboveground plant biomass and soil organic matter. The sediments in the invaded zone also showed dramatically higher concentrations of soluble nutrients, including greater than 10-fold higher soluble ammonium, nitrate, and phosphate, which suggests that Typha×glauca invasion may be impacting the wetland's ability to remove nutrients. Terminal restriction fragment length polymorphism analyses revealed significant differences in the composition of total bacterial communities (based on 16S-rRNA genes) and denitrifier communities (based on nirS genes) between invaded and uninvaded zones. This shift in denitrifiers in the sediments may be ecologically significant due to the critical role that denitrifying bacteria play in removal of nitrogen by wetlands.


Wetlands conduct critical ecosystem functions, including providing habitats for plants and wildlife, storing storm waters (Mitsch & Gosselink, 2000), and serving as sinks for terrestrially derived nutrients such as carbon and nitrogen, thus preventing the release of nutrients into adjacent surface waters, which could lead to eutrophication (Vitousek et al., 1996). Wetland nutrient uptake is driven by the activities of wetland plants and microbial communities (Mitsch & Gosselink, 2000), and denitrifying bacteria play an especially significant role in the nutrient removal function of wetlands due to their ability to convert nitrate to gaseous N2 (Otto et al., 1999).

Throughout the United States, many ecosystems, including wetlands, are threatened by invasive species (Galatowitsch et al., 1999; Zedler & Kercher, 2004). In the Great Lakes region, over 162 exotic plant and animal species have become established to date, one-third of which were introduced within the last 30 years (Mills et al., 1993; Ricciardi 2001). Invasive plants can significantly reduce the diversity of native plant and animal communities by outcompeting native species (Detenbeck et al., 1999; Werner & Zedler, 2002; Zedler & Kercher, 2004), and they can also alter cycling of carbon and nitrogen (for a review, see Ehrenfeld, 2003).

Invasive plants can impact carbon and nitrogen cycles directly, as they typically show increases in net primary productivity and standing stock biomass compared with native plants (Ehrenfeld, 2003). Invasive plants can also impact nutrient cycling indirectly through their influence on microorganisms. Recent studies have shown that microbially driven nitrogen cycling processes in terrestrial ecosystems (Kourtev et al., 2003) and freshwater wetlands (Windham & Ehrenfeld, 2003) can be impacted by invasive plants. In addition, recent studies have demonstrated that invasive plants can alter microbial community structure in terrestrial soils (Kourtev et al., 2002, 2003; Duda et al., 2003) and brackish marsh sediments (Ravit et al., 2003). Such changes in microbial community structure may be significant, as Callaway et al. (2004) recently demonstrated that the invasive plant Centaurea maculosa cultivates a soil microbial community that aids its growth and thus may contribute to its invasive success. However, no studies to date have examined impacts of invasive plants on the composition of microbial communities in freshwater wetlands, and no studies have focused on denitrifiers.

The objective of this study was to determine whether invasion of a freshwater wetland by an aggressive, exotic plant species would alter sediment nutrient content and microbial community composition. Sediment physical and chemical properties were analyzed in Typha×glauca-dominated and uninvaded zones of a coastal freshwater marsh, and terminal restriction fragment length polymorphism (T-RFLP) analysis was used to examine total sediment bacterial communities using 16S rRNA genes and to examine specifically denitrifier communities using the functional genes nirS and nirK, which code for two variants of nitrite reductase, a key enzyme in the denitrification pathway (Braker et al., 1998). The study focused on denitrifiers due to the significant role they play in wetland function, and nirS and nirK because they are effective targets for assessing denitrifier community composition via T-RFLP (Braker et al., 2001; Avrahami et al., 2002; Wolsing & Priemé, 2004).

Materials and methods

Study site

Cheboygan Marsh is a freshwater wetland located in Michigan on the northwestern shore of Lake Huron. The marsh covers c. 150 ha and experiences daily seiche activity, with approximately one-third of the marsh being continuously inundated. The native plant community of this marsh is a mixture of sedges, rushes, and bulrushes (Fig. 1). Typha×glauca (hereafter referred to as Typha), a hybrid of a native cattail species, Typha latifolia, and an exotic, Typha angustifolia, invaded the marsh 30–40 years ago (F. Cuthbert, pers. commun.) and now forms a monoculture covering more than 60% of the marsh. There are currently three distinct vegetation zones in the marsh: a Typha zone, which is composed almost entirely of Typha; a transition zone, which includes Typha and native plant species; and a native zone, which contains a diverse native plant community and no Typha. The Typha‘front’ has been advancing at a rate of 3–5 m per year. All sampling for this study was conducted between July 15 and September 15, 2004.

Figure 1.

 Relative biomass of plant species in native zone (a), transition zone (b), and Typha zone (c). Genus names abbreviated as follows: J (Juncus), E (Eleocharis), S (Schoenoplectus), C (Carex).

Field measurements

Three 0.5 m2 plots were established in each of the three vegetation zones (total of nine plots). Stem counts and heights of all emergent plants were measured within each plot. The total biomass of each plant species was determined by converting stem height measurements to biomass using species-specific height–biomass regressions based on 50 specimens of each species. The biomass of each specimen was measured after drying at 60°C. Plant diversity was calculated using the Shannon–Weiner index (Krebs, 1989). Within each plot, water depth was measured, water temperature at the sediment–water interface was determined with a Hydrolab Scout 2 (Hach Environmental, Loveland, CO), and pH was determined with an Accumet AP61 (Fisher Scientific, Pittsburgh, PA).

Sediment physical and chemical analyses

Five replicate sediment samples were collected from each vegetation zone with a soil corer (4.7 cm diameter, 10 cm depth) (total of 15 cores). Cores were placed in ziploc bags and stored on ice for transport. Sediments were sieved using a 2-mm mesh sieve and stored at 4°C. Physical and chemical analyses were completed within 24 h of sampling. Extraction for soluble NO3 and NH4+ was performed as follows: 10 g sediment was extracted with 40 mL 2 M KCl and centrifuged at 134 g for 5 min. The supernatant was filtered through G8 glass fiber filters (Fisher Scientific). Nitrate concentration in extracts was measured by the Automated Cadmium reduction method (APHA, 2005) on Auto-Analyzer 3 (Bran Luebbe, Farmington, MI), and ammonium concentration by the Automated Phenate method (APHA, 2005) on Auto-Analyzer 3. Extraction for soluble PO43− was performed as described above but 2 M KCl was replaced with Truog's Extract (Mehlich, 1953). Phosphate concentration in extracts was determined by the automated Ascorbic Acid method (APHA, 2005) on Auto-Analyzer 3. Water content was determined by drying at 105°C for 24 h and calculated as (wet weight−dry weight)/wet weight (Gardner, 1986). Organic matter content was determined by loss on ignition at 550°C (Bear, 1955). Physical and chemical data were analyzed with anova using systat version 11 (Systat Software Inc., Point Richmond, CA).

Microbial community analyses

Three replicate sediment samples were collected from each plot with a soil corer (4.7 cm diameter, 10 cm depth) (total of 27 cores, nine from each vegetation zone). Intact cores were placed in ziploc bags and stored on ice for transport. Within two hours of collecting, each individual core was homogenized and subsamples (0.5 g) from each were transferred to 2 mL microcentrifuge tubes and stored at −80°C.

DNA was extracted from each sediment sample using the MoBio UltraClean Soil DNA Kit (MoBio Laboratories, Carlsbad, CA) and confirmed by agarose gel electrophoresis. 16S rRNA genes were amplified via PCR using bacterial domain primers 8F and 926R (Liu et al., 1997). 926R was obtained from Operon (Alameda, CA) and 8F (labeled at the 5′ end with IRD-800) from LI-COR Inc. (Lincoln, NE). PCR conditions and cycling parameters are described in Janus et al. (2005). Duplicate PCR reactions were run for each sample and pooled.

For nirS, primers nirS1F and nirS6R (Braker et al., 1998) were obtained from Operon. PCR reactions contained 0.4 μM of each primer, 200 μM deoxynucleoside triphosphates (Promega, Madison, WI), 1 × PCR buffer (Promega), 1.5 mM MgCl2 (Promega), 1.5 U of Taq DNA polymerase (Promega), and 1.0 μL of DNA template. The cycling parameters were as follows: 5 min at 95°C, followed by 30 cycles of 1 min at 95°C, 1 min at 57°C, and 3 min at 72°C, followed by 4 min extension at 72°C. For nirS, reamplification was used to increase product yield for digestion (Rösch & Bothe, 2005): 1.0 μL of the first PCR reaction was used as a template for a second round of PCR, which was identical to the first, with the exception that the forward primer was replaced by 0.04 μM of nirS1F labeled at the 5′ end with IRD-800 (LI-COR Inc.). Duplicate PCR reactions were run for each sample and pooled.

Amplification of nirK genes was attempted using primers nirK1F and nirK5R and the PCR cycling parameters specified by Braker et al. (1998). Although the nirK gene could be amplified from genomic DNA of reference strain Achromobacter xylosoxidans (ATCC 15173), nirK genes could not be amplified using DNA isolated from sediments despite repeated attempts to optimize PCR by modifying template concentrations, reagent concentrations, and cycling parameters (including using a touchdown protocol). The sensitivity of nirK amplification was determined by spiking environmental DNA extracts with A. xylosoxidans genomic DNA isolated with the MoBio DNA isolation kit (MoBio Laboratories).

16S and nirS PCR products from each sediment sample (total of 27 samples, nine from each vegetation zone) were purified with the UltraClean PCR Cleanup Kit (MoBio Laboratories) and analyzed by T-RFLP as described previously (Janus et al., 2005) with the following exceptions: for 16S amplicons, 30 ng of each sample was digested with MspI, AluI, and HaeIII (New England BioLabs, Beverly, MA). For nirS amplicons, 30 ng of each sample was digested with MspI, HhaI, and TaqI (New England BioLabs).

Resultant 16S and nirS T-RFLP data sets for all three digestions were analyzed by nonmetric multidimensional scaling (MDS) and analysis of similarity (ANOSIM) using primer V.5 software package (Primer-E Ltd., Plymouth, UK). For a full description of MDS and ANOSIM procedures, see Clarke & Warwick (2001). In order to avoid potential biases introduced by reamplification, T-RFLP data were analyzed based on the presence/absence of each terminal restriction fragment (TRF) peak. In order to eliminate minor peaks from analysis, any TRFs present in less than 15% of the total samples were excluded from subsequent analyses. T-RFLP data were then imported into primer V.5, and a similarity matrix was calculated using the Bray–Curtis coefficient (Bray & Curtis, 1957). MDS was used to ordinate the similarity data (after 100 random restarts), and ANOSIM was used to examine the statistical significance of differences between groups of samples.

Results and discussion

Typha invasion in Cheboygan Marsh has resulted in a clear shift in plant species composition between the three vegetation zones (Fig. 1), significantly decreasing plant diversity in the Typha zone compared with the native zone (Table 1). This decrease in plant diversity following invasion follows the trend seen for other invading exotic plants (Ehrenfeld, 2003). Invasive plants also frequently show higher levels of net primary productivity and standing stock biomass than native plants (Ehrenfeld, 2003), and in this study the Typha zone showed twice as much aboveground plant biomass as the native zone (Table 1). The Typha zone also exhibited a 14-fold increase in plant litter (Table 1) and a fourfold increase in soil organic matter compared with the native zone (Table 2). These dramatic increases in litter and soil organic matter are likely a function of high C-fixation and increased aboveground biomass of Typha. The increased aboveground biomass and litter mass have also likely led to the decreased water temperature in the Typha zone (Table 2) through increased shading.

Table 1.   Biological properties of 3 vegetation zones in Cheboygan Marsh*
Plant diversity
Aboveground plant
biomass (g m−2)
Plant litter biomass
(g m−2)
Bacterial species
nirS genotype
  • *

    Values given as mean ± standard error (n=9).

  • Shannon–Weiner Species Diversity Index.

  • Number of terminal restriction fragments produced by T-RFLP analysis using 16S rRNA gene primers.

  • §

    Number of terminal restriction fragments produced by T-RFLP analysis using nirS gene primers.

  • Values with different letters are significantly different as determined by anova, followed by a Tukey Test for pairwise differences (P<0.05) (systat v.11).

Native0.82a± 0.06290.0a± 7.13177.3a± 2.8154a± 3.3626.4a± 2.60
Transition1.25b± 0.13354.8a,b± 19.3571.1b± 27.767.5b± 1.5036.7b± 1.74
Typha0.14c± 0.08639.6b± 29.32470.5c± 40.568.4b± 1.7137.0b± 1.83
Table 2.   Physical and chemical properties of three vegetation zones in Cheboygan Marsh*
Vegetation zonepHWater depth (cm)Water temperature (°C)Soil organic matter (%)Soil water content (%)
  • *

    Values given as mean ± standard error (n=5).

  • Values with different letters are significantly different as determined by anova, followed by a Tukey Test for pairwise differences (P<0.05) (systat v.11).

Native8.15a± 0.0430.3a± 0.0227.5a± 0.131.98a± 0.55426.6a± 2.03
Transition7.73b± 0.04818.2b± 0.01626.6a± 0.5142.85a± 0.81133.9b± 2.76
Typha7.52c± 0.0449.2c± 0.13325.4b± 0.4968.08b± 2.1580.2c± 4.51

Sediments associated with Typha demonstrated large and statistically significant increases in soluble nutrients compared with the native and transition zones, including a 14-fold increase in ammonium, a 10-fold increase in nitrate, and a 10-fold increase in phosphate (Fig. 2). This trend of elevated nutrients associated with Typha has been consistent at this site over eight samplings between 2003 and 2005 (data not shown). Previous studies by other groups have shown mixed results for impacts of invasive plants on soil nitrogen levels (Ehrenfeld, 2003). The increases in soluble nutrients observed in this study may have ecological significance, as one important function of wetlands is removal of terrestrially derived nutrients, including nitrogen and phosphorus, before these nutrients can enter adjacent surface waters and lead to eutrophication (Vitousek et al., 1996). The data suggest that Typha invasion may be impacting the wetland's ability to remove nutrients from the water.

Figure 2.

 Total soluble inorganic NH4+ (a), NO3 (b), and PO43− (c) in soils of each vegetation zone. Bars represent means for each vegetation zone (n=5) and error bars represent±standard error. Values with different letters are significantly different based on anova followed by a Tukey test for pairwise differences (P<0.001) (systat v.11).

Sediment bacterial communities in the Typha zone were significantly different in composition from communities in the native zone based on MDS (Fig. 3a) and ANOSIM (Table 3) analyses. MDS was used for data analysis because it offers significant advantages over other, more widely used statistical methods (Clarke & Warwick, 2001) and is a powerful tool for analysis of T-RFLP data (Rees et al., 2004). The Typha zone sediments also showed higher bacterial species richness, based on total number of TRFs, as compared with native zone sediments (Table 1). Although the T-RFLP assay generally provides an underestimate of biodiversity as it is biased toward the numerically dominant organisms, the number of TRFs produced for a set of samples is a useful indicator of changes in species richness (Klamer et al., 2002). The difference in bacterial species composition between the Tyhpa and native zones was not surprising, given the significant differences in sediment physical and chemical characteristics (Table 2, Fig. 2), but the lower bacterial species richness in the native zone was surprising as the native zone had much higher plant diversity than the Typha zone (Fig. 1, Table 1). This result suggests that Typha invasion may be increasing the diversity of microniches available to the bacteria. The bacterial communities in the transition zone were intermediate in composition between the communities in the Typha and native zones (Fig. 3a, Table 3), but showed bacterial species richness levels that were equivalent to those associated with Typha (Table 1). This result again indicates a positive effect of Typha invasion on bacterial species richness.

Figure 3.

 Nonmetric multidimensional scaling (MDS) analysis of 16S T-RFLP data (a) and nirS T-RFLP data (b). • (native zone), □ (transition zone), ▴ (Typha zone).

Table 3.   ANOSIM R statistics and p values for 16S and nirS T-RFLP analyses
Vegetation type16SnirS
Native vs. Typha0.7950.0010.7380.001
Native vs. transition0.2910.0050.4650.005
Transition vs. Typha0.3640.0040.4520.001

Denitrifier communities in the Typha zone were significantly different in composition based on nirS gene sequences (Fig. 3b, Table 3) and showed higher nirS genotype richness as compared with the native zone (Table 1). Denitrifier communities in the transition zone were intermediate in composition between the communities in the Typha and native zones (Fig. 3b, Table 3), but showed nirS genotype richness levels that were equivalent to those in the Typha zone. This result indicates a positive effect of Typha invasion on nirS genotype richness.

In this study, nirK genes could not be amplified from DNA isolated from sediment samples, despite the fact that the same primer set has been used successfully by other researchers to amplify nirK genes from environmental samples (Prieme et al., 2002; Wolsing & Priemé, 2004). The failure to amplify nirK genes may have been caused by inhibition of PCR by inhibitory compounds in the sediment DNA extractions. However, when environmental DNA extracts were spiked with A. xylosoxidans genomic DNA, nirK genes were amplified from as few as 75 genome copies per reaction. Based on the DNA extraction protocol used, this corresponds to c. 4 × 103nirK copies per gram of sediment. These data suggest that nirK-containing organisms were either not present in the Cheboygan Marsh sediments or were present at a level below 4 × 103nirK gene copies per gram of sediment. As the two structurally different nitrite reductases that are encoded by nirK and nirS are found to be mutually exclusively among denitriers (Braker et al., 1998), these results suggest that nirS denitrifiers may have been dominant in Cheboygan Marsh. This conclusion is supported by several groups that have suggested that nirK denitrifiers are less abundant than nirS denitrifiers in estuarine sediments (Nogales et al., 2002) and marine sediments (Braker et al., 2000). However, Throback et al. (2004) demonstrated that the nirK1F/nirK5R primer set does not amplify all nirK-containing strains, so Cheboygan Marsh may have included nirK denitrifiers not targeted by this primer set.

The differences in denitrifier community composition that were observed based on nirS gene sequences may have ecological significance, as different denitrifying taxa are known to differ in their oxygen threshold, carbon requirements, and kinetic parameters (Tiedje, 1988), and the N2O : N2 ratio resulting from denitrication can be dependent on species composition of denitrifying communities (Munch, 1989). Few prior studies have examined the impacts of invasive plants on denitrification. Otto et al. (1999) found no change in denitrification rates in freshwater marsh sediments under invasive plants, while Bolton et al. (1990) found an increase in denitrification under an invasive plant in terrestrial soils. However, neither of these studies examined the impacts of invasive plants on the composition of denitrifier communities. The current study demonstrated that Typha supported a distinctly different community of denitrifiers with higher nirS genotype richness, as well as 10-fold higher levels of soluble nitrate, suggesting that Typha may be impacting the denitrification process.


The results of this study demonstrate that invasion of a Great Lakes freshwater wetland by an exotic plant species, Typha×glauca, had significant impacts on sediment physical, chemical, and biological characteristics. The significant increases in soluble ammonium, nitrate, and phosphorus in sediments associated with Typha suggest that Typha invasion is affecting the wetland's ability to remove nutrients from water. The shifts in total bacterial and denitrifier community composition and nirS genotype richness indicate that Typha is altering the microbial makeup of the wetland sediments, and the observed shift in denitrifiers in particular suggests that Typha may be affecting microbially catalyzed nutrient cycling processes such as denitrification. Further work is needed to determine whether Typha invasion has altered rates of microbially catalyzed nutrient cycling processes and whether Typha invasion has impacted the ecosystem's ability to prevent release of nutrients to the lake.


The authors thank M. Berg, C. Ishida, M. Hausner, and A. Morris for their assistance, Loyola University Chicago for support of N.L.A., and University of Michigan Biological Station for lab space and instrumentation. This work was supported by DEB-0108847 awarded to N.C.T., and by a supplement to grant DEB-0108847 awarded to N.C.T. and J.J.K. from the National Science Foundation.