Plant–Soil Feedbacks Contribute to the Persistence of Bromus inermis in Tallgrass Prairie
- First Online:
- Cite this article as:
- Vinton, M.A. & Goergen, E.M. Ecosystems (2006) 9: 967. doi:10.1007/s10021-005-0107-5
- 364 Views
As invasive plants become a greater threat to native ecosystems, we need to improve our understanding of the factors underlying their success and persistence. Over the past 30 years, the C3 nonnative plant Bromus inermis (smooth brome) has been spreading throughout the central grasslands in North America. Invasion by this grass has resulted in the local displacement of natives, including the tallgrass species Panicum virgatum (switchgrass). To determine if factors related to resource availability and plant–soil interactions were conferring a competitive advantage on smooth brome, field plots were set up under varying nitrogen (N) levels. Plots composed of a 1:1 ratio of smooth brome and switchgrass were located in a restored tallgrass prairie and were randomly assigned one of the following three N levels: (a) NH4NO3 added to increase available N, (b) sucrose added to reduce available N, and (c) no additions to serve as control. In addition, soil N status, soil respiration rates, plant growth, and litter decomposition rates were monitored. Results indicate that by the 2nd year, the addition of sucrose significantly reduced available soil N and additions of NH4NO3 increased it. Further, smooth brome had greater tiller density, mass, and canopy interception of light on N-enriched soils, whereas none of these characteristics were stimulated by added N in the case of switchgrass. This suggests that smooth brome may have a competitive advantage on higher-N soils. Smooth-brome plant tissue also had a lower carbon–nitrogen (C:N) ratio and a higher decomposition rate than switchgrass and thus may cycle N more rapidly in the plant–soil system. These differences suggest a possible mechanism for the persistence of smooth brome in the tallgrass prairie: Efficient recycling of nutrient-rich litter under patches of smooth brome may confer a competitive advantage that enables it to persist in remnant or restored prairies. Increased N deposition associated with human activity and changing land use may play a critical role in the persistence of smooth brome and other N-philic exotic species.
Keywordsinvasiontallgrass prairieplant–soil feedbacknitrogensmooth bromeBromus inermiscarbon addition
The invasion and persistence of nonnative plants is a global problem that further complicates the ongoing challenge of preserving and managing native ecosystems (Vitousek and others 1997; Mack and others 2000). Invasive plants occupy over 40 million ha in the United States and are estimated to be spreading at the rate of over 1 million ha per year (National Invasive Species Council 2001). Numerous mechanisms have been proposed to explain the persistence of invasive plants, including the possibility that invasion alters competitive interactions (Suding and others 2004) and affects ecosystem and disturbance processes in ways that favor nonnative plants over native species (D’Antonio and Vitousek 1992; Ehrenfeld 2003).
Bromus inermis (smooth brome) is a cool season, perennial Eurasian grass that is now a common member of ecosystems throughout the central grassland region in North America. It was introduced in the 1880s and widely planted, both for forage as well as soil stabilization along roads and waterways (Carlson and Newell 1985). Smooth brome is now considered to be naturalized, and occurs throughout most of north and central North America (Stubbendieck and others 1992). In Weaver’s extensive observations of 135 North American tallgrass prairies ranging in size from 20 to 360 ha, smooth brome is not mentioned, yet his observations do emphasize the dominance of another cool-season alien, Poa pratensis (Weaver and Fitzpatrick 1934; Weaver 1954, 1965, 1968). In contrast, smooth brome was found in all sites in a more recent survey of 10 remnant eastern Nebraska tallgrass prairies; these sites ranged in size from 1 to 18 ha, and some were located near Weaver’s original observations (Boettcher and others 1993). Clearly, smooth brome has spread from pasture and roadway plantings into native ecosystems in a relatively short period of time.
Smooth brome is especially problematic in the upper Midwest and Northern Plains regions of the North American tallgrass prairie range, where it can make up more than 80% of the biomass and form monocultures that appear to resist recolonization by native species (Butterfield and others 1996; Larson and others 2001). Numerous control measures including burning, mowing, and herbicide application (Waller and Schmidt 1983; Dill and others 1986; Willson and Stubbendieck 1996, 2000), can reduce the abundance of smooth brome, but without sustained efforts, the species is remarkably persistent. One significant factor determining the success of these control measures is the presence of a codominant native C4 grass community. For example, burning had little negative effect on smooth brome in pure stands, but in mixed stands of smooth brome and the native grass big bluestem burning was a more effective means of control (Willson and Stubbendieck 2000).
The growth and success of native plants relative to smooth brome may be affected by the resource environment. Soil resources, particularly nitrogen (N), are known to alter the growth and competitive interactions of invasive plants in native grasslands (for example, Huenneke and others 1990; Wedin and Tilman 1996; Milchunas and Lauenroth 1995; Maron and Connors 1996; Suding and others 2004), sometimes resulting in a long-term shift to a nonnative community (Vinton and Burke 1995). Increased N inputs to ecosystems via human activities (Vitousek and others 1997; Smith and others 1999) may be a significant factor associated with the increased cover of exotic species in temperate grasslands. Native grasses seem to be well adapted to low soil N availability (Wedin and Tilman 1990) and may not have the capacity to respond as quickly as nonnative plants when this resource is increased (Suding and others 2005).
Several recent studies have used carbon (C) additions in an effort to lower soil N and provide native plants with a competitive advantage over nonnative plants (Wilson and Gerry 1995; Zink and Allen 1998; Reever Morghan and Seastedt 1999; Alpert and Maron 2000; Paschke and others 2000; Baer and others 2003; Blumenthal and others 2003; Corbin and D’Antonio 2004; Suding and others 2004). Presumably, C additions cause the soil microbial community to be N-limited, leading to the N depletion or immobilization of N. The timing, rate, and form of C application required to reduce soil N to a point that will stimulate native and inhibit invasive species growth is by no means straightforward. In reviewing previous studies that found conflicting effects of C addition on alien species, Blumenthal and others (2003) suggested that the C addition rate must be sufficient to effectively immobilize N to the extent that alien species are suppressed. Furthermore, C additions may have little effect on nonnative plants in ecosystems where productivity is more limited by water than soil nutrients. In our system, the tallgrass prairie, N is a key limiting factor to productivity (Blair 1997; Knapp and others 1998), suggesting that the application of C may be an effective method for the control of invasive species and the restoration of native prairie plants.
We studied the role of N fertilization and C additions on the growth of smooth brome versus the native species Panicum virgatum (switchgrass) in intact tallgrass prairie communities. We hypothesized that growth of the exotic species would be enhanced by N addition and reduced by C addition. Further, we hypothesized that the growth of the native species would not be as strongly affected by C and N applications. In addition, we compared the ability of both species to act as an N source to the soil by comparing their respective tissue chemistry and litter decomposition rates. Based on our observation that smooth-brome canopies have a noticeable lack of litter buildup, we hypothesized that smooth-brome litter decomposes faster than that of native prairie grasses. In turn, if smooth brome responds to increased soil N more than native grasses, the plant–soil feedback may enhance the success of smooth brome over that of native plants.
MATERIALS AND METHODS
Studies were conducted at Allwine Prairie Preserve, a 65-ha restored tallgrass prairie, 20 km northwest of Omaha in Douglas County, Nebraska, USA. Allwine is owned and managed by the University of Nebraska–Omaha. Climate data from a 1954–2003 record (North Omaha Station, High Plains Regional Climate Center) show an average annual precipitation of 766 mm. Temperatures from this time period ranged from a mean monthly low of −6°C in January to a mean monthly high of 24°C in July. The soils are silt to silty clay loams (Bragg 1978). Total soil N in the top 30 cm ranges from 3.47 to 5.53 T/ha, and soil organic matter is 94.3–116.3 T/ha (Schacht and others 1996). Assuming total soil C is 58% of the soil organic matter, the C:N ratio ranges from 12.20 to 15.76. In our plots, the average percentages of sand, silt, and clay were 8%, 61%, and 31%, respectively. The site was formerly cropped but was restored to tallgrass prairie in 1970 (Bragg 1978). Management consists of periodic mowing and spring burning every 3rd year. Our experimental plots were set up in April 1996 in a location that was burned the prior spring.
The dominant plant species include the warm-season native grasses Andropogon gerardii (big bluestem), A. scoparius (little bluestem), Bouteloua curtipendula (side oats grama), Sorghastrum nutans (Indian grass) and Panicum virgatum (switchgrass) (Bragg 1978). Collectively, these species comprise more than 80% of the total vegetative cover, whereas forbs typically comprise less than 5% cover. Also locally abundant are the nonnative cool-season grass species Bromus inermis (smooth brome) and, in the lowlands, Phalarus arundinacea (reed canary).
Experimental Treatments and Responses
The experimental area consisted of a series of 4 m2 plots along the southern edge of Allwine Prairie. Each plot was dominated by smooth brome on one half and switchgrass on the other half. Other plant species present in the plots included big bluestem, side oats grama, and Indian grass. This pattern appeared to be a result of smooth-brome encroachment onto the established prairie from a roadside ditch. The treatment for each plot was randomly assigned, with five replicates per treatment, and consisted of (a) control with no additions, (b) N addition (NH4NO3) with a total amount equaling 42 g N /m2 over 2 years, and (c) C addition (sucrose—C12H22O11) with a total amount of 400 g C/m2 over 2 years. The NH4NO3 and sucrose were applied manually to plots at three times during year 1: 13 April (100 g/m2 C and 10 g/m2 N), 19 June (100g/m2 C and 10 g/m2 N), and 7 September (100 g/m2 C and 12 g/m2 N). During year 2, NH4NO3 and sucrose were applied on 10 May (50 g/m2 C and 5 g/m2 N) and 2 June (50 g/m2 C and 5 g/m2 N).
The amount of extractable inorganic N (NH4-N and NO3-N) in soils was sampled five times during the study. Soil samples were taken on 13 April before treatments were applied to measure initial conditions. Soils were also collected on 22 May (about 5 weeks posttreatment) and on 16 July (about 4 weeks after the second treatment application) of year 1. In year 2, soils were collected on 9 June and 1 July, approximately 4 weeks after the first treatment application and 1 week after the second application. All soil samples were taken with a 5 cm diameter × 10 cm depth corer in random locations within plots, and two cores were composited for each treatment plot. Soil samples were kept on ice until analyses could be completed. Soils were sieved to separate plant material and fragments larger than 2 mm in diameter. Total dry mass of sieved soil was used as an estimate of bulk density. The sieved soil was weighed into 10-g subsamples to analyze percent moisture and inorganic N. Percent moisture was estimated by comparing the initial sample mass to that after drying overnight at 105°C. Inorganic N was extracted from soils by adding 50 ml of 2 M KCl to 10 g of fresh soil and shaking on an orbital shaker at 200 rpm for 30 min. Extracts were filtered through no. 40 Whatman filters and frozen until colorimetric analyses for NH4–N and NO3–N were done at the University of Nebraska soil testing laboratory.
Soil incubations were set up in year 2 to measure rates of potential net N mineralization and C respiration in the plots. Soils were collected from the field plots on 9 June (about 1 week after the fifth treatment application) and kept on ice until incubations were set up. Soils were sieved and weighed, and percent moisture and initial inorganic N levels were measured as described above. In addition, C respiration for soils was measured on a 20-g subsample, brought up to field capacity (35% moisture), and placed in a 0.95-L canning jar along with a base trap containing 3 ml of 1 mol/L NaOH in a scintillation vial (Schimel 1986; Snyder and Trofymow 1984). Jars were kept in the dark and sealed except when base traps were removed, titrated, and replaced on days 5, 13, and 31 of the 31-day incubation period. The amount of carbondioxide (CO2) absorbed in base traps was calculated using a Mettler DL 12 auto titrator (Mettler - Toledo, Inc., Columbus, OH) to back-titrate with 1 N HCl as a means of detecting NaOH after CO3 was precipitated with an excess of BaCl2 (Robertson and others 1999; Paul and others 1999). Soil respiration was expressed as the total amount of CO2 absorbed by base traps during the incubation divided by the number of days in the incubation period. Net N mineralization/immobilization was measured as the NH4–N plus NO3–N levels on day 1 minus the NH4–N plus NO3–N levels on day 31 of the incubation divided by the number of days in the incubation period.
Field measurements of soil respiration were done mid-growing season during year 2 on 14 July. Glass scintillation vials were filled with 5 ml of 1M NaOH and capped tightly. Vials were transported to the field and placed on random locations on bareground within the plots. The vials were then uncapped and covered with an inverted 120-ml specimen cup, which was pushed down into the soil until the cup rim was about 0.5 cm deep, thus forming a small incubation chamber with base trap. Blanks were set up by placing the scintillation vials of NaOH inside sealed specimen cups, not in contact with the soil. Soil temperature during the incubation ranged from 21.6 to 22.8°C. After 12 h (0700 to 1900), the vials were collected, sealed, and brought back to the laboratory for titration analyses as described above, to yield CO2 respired.
Mean Levels of Soil Inorganic Nitrogen (NH4–N plus NO3–N) Measured on Five Dates during the Experiment
Plant tiller density in the treatment plots was recorded on three dates, 2 May and 1 August in year 1 and 12 June in year 2. The corners of 0.25 × 0.25 m quadrats were marked with golf tees, and a three-sided frame was slid into the canopy to delineate the area within which tillers were counted. Within each plot, two quadrats, one on the smooth brome side and one on the switchgrass side, were randomly selected, and the number of tillers within the quadrat was recorded. In the spring of year 1, switchgrass (a warm-season species) was not yet active, and by fall of year 1, smooth brome had started to senesce, so we report data from only June of year 2, when both species had active tillers. To determine the effect of treatment on plant biomass, three to five plants of each species were randomly collected from field plots at the conclusion of the growing season (late September) in both year 1 and year 2. Individual plants were clipped at ground level and placed in a paper coin envelope. The samples were taken back to the lab, dried at 52°C for 48 h, and weighed.
To determine the effect of soil treatments on the canopy productivity of each species, we measured the percentage of photosynthetically active radiation (PAR) intercepted by the canopy. Measurements were taken during mid-July of year 2 to capture the approximate peak of both species. An AccuPAR ceptometer (Decagon Devices, Inc., Aullman, WA) was placed at the soil surface and directly over the canopy in each plot to obtain percent PAR interception. Five sets of measurements were averaged for each of the brome and switchgrass sides of the plots. Measurements were taken on clear days within 2 h of solar noon.
To set up the litter decomposition experiment, litter from pure stands of smooth brome and switchgrass near the experimental plots was collected on 3 June of year 1. The area had been burned in the previous spring; thus, litter from both species had been produced during the previous growing season. All litter was air-dried and a subsample was oven-dried (52°C for 48 h) to find the relationship between mass of air-dried litter and oven-dried litter. Fiberglass mesh bags were constructed, and approximately 10 g of air-dried litter (with exact mass recorded) was placed in each bag. Seventy bags were constructed for each species and placed on the litter/soil surface of a transect near the experimental plots on 9 June of year 1. Bags were attached to the litter/soil surface with 2-cm staples placed at two corners of the bag. Approximately 23 bags for each species were collected on 24 July and 7 September of year 1 and on 22 April of year 2. Bags were oven-dried at 52°C for 48 h and mass was compared to initial oven-dried mass to calculate the percent of litter remaining after decomposition.
To determine the role of plant tissue composition on decomposition rates, a subsample of the litter of each species that had been collected for the decomposition experiment was analyzed for C, N, and lignin content. Analyses of C and N were done with combustion methods using a Leco C/N analyzer (LECO Corporation, St. Joseph, MO), and lignin anlyses were done using the wet chemistry, acid detergent fiber technique. Samples were analyzed at the University of Nebraska–Lincoln soil testing laboratory.
Statistical analyses were conducted using SAS (SAS Institute, Cary, NC, USA). For soil variables, one-way analyses of variance (ANOVA) were used to detect differences between the control, N-amended, and C-amended plots. One-way ANOVA was also used to detect plant species differences in tissue quality (C:N and lignin:N). For plant growth variables (tiller density, mass, and PAR interception), data were analyzed as a split-plot ANOVA design with treatment as the whole-plot factor and species as the subplot factor. Data distributions were normalized before ANOVAs were applied. When ANOVA indicated a significant effect, Tukey’s studentized range test was applied to detect which means were significantly different from one another.
Tiller Density, Tiller Mass, and Percent PAR Interception in Canopies of Smooth Brome and Switchgrass on Plots that Received No Soil Amendment (Control), N Amendments, or Carbon Amendments
Sp × Treatment Effect
Tiller Density (#/m2)
P < 0.0031
P = 0.0107
P = 0.0173
Tiller Mass—Year 1 (g)
P < 0.0001
P = 0.3105
P = 0.0063
Tiller Mass—Year 2 (g)
P < 0.0001
P = 0.845
P = 0.009
PAR Interception (%)
P = 0.0644
P = 0.924
P = 0.0334
Our hypothesis that smooth brome benefits more than switchgrass from enhanced soil N is supported by the data. Smooth-brome tiller density and tiller mass were stimulated by soil N addition. In contrast, tiller density and tiller mass of switchgrass were not stimulated by N addition. In addition, smooth-brome canopy PAR interception responded differently to soil treatments than did switchgrass canopies. Smooth brome formed a denser, more productive canopy (as indexed by percent PAR interception) under enhanced soil N conditions, whereas switchgrass did not.
In terms of litter quality and decomposition, it was clear that smooth brome had higher-quality aboveground litter that decomposed more quickly than that of switchgrass. We did not measure the quality and decomposition of roots in this study. Smooth-brome and switchgrass root N concentrations tend to be similar (Reich and others 2001) or lower (Reich and others 2003) than that of shoots. It is possible that the species patterns seen in aboveground tissue quality and decomposition are not as pronounced belowground. Further, with frequent fire in grassland ecosystems, aboveground tissue does not always persist long enough to enter the decomposer pathway. However, in the case of smooth brome, both its C3 pathway (conferring a high proportion of green and therefore less flammable tissue during most of the year) and its lack of litter buildup make stands less likely to burn. Thus, the quality and decomposability of aboveground litter are key traits that have the capacity to affect nutrient cycling in soils under smooth brome.
The fact that smooth brome’s aboveground tissue decomposes faster than that of switchgrass supports our hypothesis that smooth brome may foster high soil N via litter breakdown, creating a positive feedback between growth and available soil N. Smooth brome begins growth earlier in the season than many of the native C4 grasses. This factor, combined with its quicker response to enhanced resources, would confer a competitive advantage on smooth brome under high-nutrient conditions. Further, because smooth brome has the potential ability to create its own elevated nutrient conditions via decomposition, we would expect smooth-brome soils to have consistently higher N availability than soils under switchgrass. Other studies indicate that co-occuring plant species with different litter characteristics can have differential effects on soil N cycling by changing microbial activity and biomass (Bowman and others 2004). Further, it is worth noting that switchgrass shows strong positive responses to increased soil N in prairie restorations (Baer and others 2004). Certainly, switchgrass may be one of the more responsive native grasses to additional N in prairie systems. But results from this study indicate that its ability to prolong enhanced-N conditions via decomposable tissue may be more limited than that of smooth brome.
Tallgrass prairies generally become more N-limited under frequent burning (Knapp and Seastedt 1986; Blair 1997). Because smooth-brome persistence may be linked to high levels of soil N, frequent burning may play a role in its control. Frequent burning (especially in the spring) is known to enhance growth of native C4 prairie grasses (Knapp and others 1998). Given that smooth-brome control is most successful in areas with co-existing native species (Willson and Stubbendieck 2000), frequent burning (via its negative effect on soil N and positive effect on native grasses) would seem to be a sound management practice for the control of smooth brome.
Smooth brome’s positive response to soil N and its rapidly decomposing tissue are traits that we focused on in this study. Certainly, other traits may also enhance smooth brome’s success in tallgrass prairie systems. For example, it is tolerant of heavy grazing and is highly palatable both early and later in the growing season (Miller 1984). In addition, smooth brome reproduces both by seedling recruitment and prolific rhizomes, which often form a dense sod with little open space to accommodate colonization by other species (Carlson and Newell 1985; Dill and others 1986). Further, the C3 photosynthetic pathway of smooth brome enables it to begin growth earlier in the season than the C4 prairie grasses. Moreover, grasses with the C3 pathway tend to be more responsive to fertilization than C4 grasses (Reich and others 2001). It is likely that smooth brome’s interactions with soil N play a role in a suite of traits that make it a successful and persistent plant in tallgrass prairies.
Our results are consistent with recent studies showing that C additions to soil decreases plant N availability (Baer and others 2003; Blumenthal and others 2003). However, depletion of soil N (via C additions) did not always reduce smooth-brome growth and density relative to controls. Similar results were found in California grasslands, where the growth of exotic perennials was unaffected by sawdust addition (Corbin and D’Antonio 2004). The results of this study further indicate that the control of exotics by C addition may be conditional on other factors. For example, a number of studies have found that C addition has a negative effect on annuals but a positive or neutral effect on perennials (Alpert and Maron 2000; Paschke and others 2000; Blumenthal and others 2003).
Taken together, our data suggest that N addition had stronger effects on smooth brome than C addition. One factor that may help to explain the rapid increase in smooth brome in remnant prairies (Boettcher and others 1993) is the increased atmospheric N deposition associated with human activity and changing land use. During the past century, atmospheric N deposition in the central grassland region of North America has increased nearly 10-fold (Galloway and others 2004). Although N fertilization rates in this study were higher than those occurring via annual atmospheric deposition (0.48 g/m2 in 2004) (NADP 2005; http://www.nadp.sws.uiuc.edu), the cumulative effects of such N deposition could have stimulated smooth-brome growth to a greater extent than that of native grasses. Further, native tallgrass prairies could be experiencing more N loading than atmospheric deposition data suggest, via exposure to point sources of atmospheric or surface-water N pollution such as urban centers, highways, fertilized agricultural fields, and intensive livestock operations. Finally, high localized rates of N input (approximately 50 g/m2) via urine deposition are not unprecedented in a grassland heavily used by grazing animals (Stillwell 1983; Steinauer and Collins 1995).
This study contributes to a larger picture of invasive plant strategy, in that many aliens accelerate N cycling. In addition, enhanced soil N can create such productive conditions for some species that other, rarer species are driven locally extinct (Suding and others 2005). In a recent review, Ehrenfeld (2003) noted that invasive plants caused increases in soil N mineralization and nitrification in 11 of 16 cases and increases in soil microbial biomass N in eight of 10 studies. However, these increases in plant-available N in invaded soils were not consistently associated with increases in total soil N or extractable inorganic N (NH4–N and NO3–N). The seasonal dynamics of these pools, together with the differing allocation patterns of invasive species, may contribute to the lack of correlation between measures of N cycling in invaded areas (Ehrenfeld 2003). Further, Ehrenfeld (2003) noted that most exotic species have more rapidly decomposing litter than the displaced natives (10 of 14 cases), thus setting up the potential for faster N cycling. Our data suggest that smooth brome is similar to a majority of exotics in its effects on soil C and N cycling. The fact that faster N cycling and enhanced soil N can lead to the loss of rare species (Suding and others 2005) further increases the potential for negative impacts of exotic species.
The differences in response to elevated N between smooth brome and switchgrass, together with the rapid decomposition of smooth-brome litter, indicate a possible mechanism for the persistence of smooth brome in tallgrass prairies. Efficient recycling of N-rich litter under patches of smooth brome may confer a competitive advantage and enable it to persist in remnant or restored prairies in the central grassland region of North America. Carbon applications may contribute to smooth-brome control in ecosystem restorations; however, application rates would have to be sufficient and prolonged to effectively depress soil N. Furthermore, our data do not show a compelling decrease in smooth-brome growth under C additions as used in this study. Finally, the increased, human-induced N deposition on terrestrial ecosystems will likely make control of N-philic exotics such as smooth brome a continuing challenge.
Financial support came from the Clare Boothe Luce Endowment at Creighton University and the National Science Foundation’s Long Term Ecological Research Program (DEB 96328510) at Konza Prairie Research Natural Area. We thank Mary Heffron, Corey Rife, Dan Deatsch, Emily Kathol, and Leanne Vigue Miranda for field and laboratory assistance.