Meredith Thomsen

Biology Department


The floodplain forests of the upper Mississippi River represent a valuable resource from multiple perspectives. They provide critical habitat and food resources for threatened songbird species using the river valley flyway during their northern migration (Knutson & Klaas 1998). Tree roots hold sediment in place, improving river water quality and riverbank stability, thereby decreasing the amount of work required to maintain navigation channels of adequate depth (Urich et al. 2002). Finally, the floodplain forests provide aesthetic value to the recreational users of the river corridor, including those camping on the forested islands.

Given the value of the floodplain forest, it is of considerable concern to many parties that the long-term stability of the ecosystem is uncertain. When the lock and dam system was built in the late 1930s, timber was harvested from river islands and other floodplain areas (Urich et al. 2002). These sites now contain even-aged stands dominated by silver maple, a relatively short-lived tree species (Knutson & Klaas 1998). As silver maple stands reach maturity and individual trees begin dying, gaps are created in the forest canopy. In most forest types, the growth of young trees is stimulated by the increased light availability within gaps, causing the trees to grow up from the sub-canopy and fill the gaps. New tree species may enter the canopy in this way, altering forest community composition, but the forest ecosystem type is maintained.

The usual pattern of forest regeneration appears to have been disrupted in many upper Mississippi River floodplain forests by the presence of reed canary grass, an invasive species that has become widely distributed in various wetland habitats throughout the upper Midwest and elsewhere (Adams and Galatowitsch 2005).  Reed canary grass has been linked to decreased biodiversity in much of its range, due in large part to its tremendous competitive ability. High rates of seedling growth allow it to dominate an area rapidly each spring, leaving little room for other plant species (Lavergne & Molovsky 2004). Furthermore, it is a perennial species, meaning that individual plants survive many years. Multiple years of reed canary grass growth results in a thick accumulation of dead plant material, which further impedes the growth of neighboring plants. It is hypothesized that in floodplain forest treefall gaps, reed canary grass seeds in the soil germinate in response to the increased light availability, just as the tree seeds do. The vigorous growth of reed canary grass seedlings may then impedes that of the tree seedlings, and even of saplings present before gap creation. Ultimately, gaps may become dominated by the grass instead of young trees, preventing forest regeneration (Urich et al. 2002).

Land managers are now seeking strategies to decrease competition from reed canary grass in treefall gaps and thereby increase tree establishment. One method is to remove standing vegetation, including any reed canary grass, exposing the mineral soil and giving newly germinated tree seedlings the best chance to get their roots into the ground. Bulldozers and heavy-duty mowing machines are commonly used to accomplish this goal. Additionally, herbicides can be used both to kill back existing reed canary grass patches and to prevent the germination of seeds and regrowth of underground stems.  Because of the location of floodplain forest sites near open water, it is a major priority to identify herbicides that break down into non-harmful compounds relatively quickly, and to determine the lowest concentrations of these substances that significantly reduce reed canary grass populations. Restoration projects carried out in the La Crosse area by the US Army Corps of Engineers (see below) have varied the timing and application rate of various types of herbicides (US Army Corps of Engineers 2006). To date, however, there have been no replicated experiments comparing the efficacy of various herbicide combinations on the ultimate success of tree seedling establishment in local floodplain forest gaps.

Reed canary grass is present throughout large portions of the upper Mississippi River valley, and thus has the potential to alter forest dynamics over a wide area. Reed canary grass spreads rapidly among sites via numerous seeds that are easily transported by water, germinating in new locations (Lavergne & Molovsky 2004). It can also spread when fragments of its underground stems are moved by water and re-sprout in a new location. Any forested area that flood waters can reach, then, may contain reed canary grass seed, even if adult plants are not present in the area. Since reed canary grass seeds can only survive in the soil for a limited period of time, the frequency of flood events will also have an effect on the amount of viable seed present in a site: the more frequently an area floods, the more frequently fresh seed will be deposited, and the larger the proportion of viable reed canary grass seed in the soil. Flood duration is also likely to affect the amount of seed present, since longer floods allow for a longer period of seed delivery. Since seed density and viability will affect the intensity of competition from reed canary grass in future treefall gaps, we need information about exactly how seed abundance and condition varies throughout the floodplain forest. At the current time, such information is not available. 



Project Description

In the summer of 1998, straight-line winds associated with a strong thunderstorm blew down extensive areas of the floodplain forest on Goose Island in the Mississippi River, near La Crosse, WI.  Foresters with the St. Paul District of the US Army Corps of Engineers (US ACE) have been working to establish floodplain tree species in two sites on Goose Island, using a combination of physical and chemical treatments. They have seen good regeneration in some areas, but plan to re-treat other areas in Fall 2006 (US Army Corps of Engineers 2006).

In collaboration with Eileen Kirsch at the Upper Mississippi Environmental Science Center (UMESC) and Randy Urich at the US ACE, I am establishing a restoration experiment as part of the re-treatment of the Goose Island sites. In October 2006, we established 20 plots in the site, with 5 replicates each for controls and three treatment combinations. In November 2006, all treatment plots will be mowed using a Fecon Mower®, which will chop up the standing vegetation and break ground, thereby improving germination conditions for tree seeds next spring. Furthermore, three herbicide treatments will be used in the mowed areas (see Table 1). We will test the effect of a high vs. low concentration of a pre-emergence herbicide (Oust®, sulfometuron methyl) that has proven effective against reed canary grass in other habitat types. We will also include a treatment of the low concentration of Oust® used in combination with another herbicide (Pendulum®, pendimethalin), that may provide additional grass control. Our experimental design does not allow the evaluation of either mowing or herbicide treatments alone, because the mowing is regarded as a necessary precondition for any successful restoration in sites already dominated by reed canary grass (R. Urich, personal communication). Our full control treatment will allow us to determine the overall effectiveness of each mowing/herbicide treatment relative to no management activity.

At the time of plot establishment, we conducted sampling at five randomly-selected points per plot to document pre-treatment plant community composition. We identified all plant species present and visually estimated their percent cover in two- meter-square quadrats. Pre-treatment data will be useful in determining how plant community composition (e.g. grass density) interacts with experimental treatments. We also measured the distance from each sampling location to the nearest mature tree (likely to affect tree seed availability) and the distance and direction of the gap edge (which will determine the degree of shading at each location). At a later date, we will determine the elevation of each sampling location, to determine how elevational changes across the site may affect plant competitive interactions, environmental conditions, and treatment effectiveness. Tree seedlings may grow especially poorly in lower, wetter areas, where we suspect the strength of competition from reed canary grass is the greatest.

In late October 2006, several University of Wisconsin – La Crosse students will collect soil cores to be used as part of their undergraduate research projects at the same five sampling locations per plot. One student will use the Cowley Hall Greenhouse to germinate seeds present in the cores to determine the abundance of reed canary grass and other seeds (including tree species) present in the soil across the site. Re-sampling in spring 2007 will allow us to determine if the herbicide treatments decrease the seed viability as intended. A second will extract reed canary grass underground stems from soil cores to generate an estimate of the density of these reproductive structures at our sampling locations.  A third student will determine pre-treatment nutrient availability across the site, and follow up these measurements with post-treatment measurements early in the spring growing season of 2007. I will continue to monitor soil nutrient availability in all plots throughout the experiment. The same student will also conduct bi-weekly monitoring of the site during winter term 2007, to document water levels, soil moisture availability, and the timing of plant germination across sampling locations. I will continue to monitor patterns soil moisture availability for the duration of the experiment.


Table 1: Experimental treatments and replicate numbers for

the Goose Island restoration project.



Sampling points

Additional experiments




Grass – tree seedling competition

High Oust®




Low Oust®




Low Oust® plus Pendulum®





Proposed research

Seedling establishment

In the summer of 2007, I propose to document patterns of tree seedling establishment across experimental treatments in the Goose Island restoration experiment.  At the same sampling locations as the pre-treatment measurements I will count, identify, and mark all tree seedlings present, and visually estimate the percent cover of reed canary grass and other species in each quadrat. I will continue to monitor the plots on a monthly basis during the summer and fall of 2007, and again during the 2008 and 2009 growing season. This will allow us to evaluate the ultimate effectiveness of the restoration treatments relative to controls and to determine whether or not any follow-up treatment (e.g. application of the grass-specific herbicide Sethoxydim G-Pro®) is necessary within treatment plots. I will also evaluate how pre-treatment plant and seed community composition, reed canary grass underground stem density, and nutrient and water availability data help explain treatment effects on seedling establishment.


Competition experiment

Within the control (completely untreated) plots, I propose to establish additional sampling points in which I will manipulate the litter depth and the density of reed canary grass and tree seedlings. This will allow me to generate experimental data on the precise interactions between the two groups of plants. Such data will aid in the evaluation of the restoration experiment, by demonstrating the effect of reduced density of reed canary grass litter vs. growing individuals on tree seedling establishment. It will also serve as preliminary data in future grants focused on the larger question of how reed canary grass is affecting floodplain forest regeneration throughout the river corridor (see below).


Reed canary grass seed density survey

The Goose Island restoration experiment provides an excellent opportunity for me to begin studying the interactions between reed canary grass and floodplain forest species at the local scale, and will generate valuable information about best management practices. I am also interested in pursuing the broad question of how reed canary grass is likely to influence floodplain forest regeneration at the landscape scale. Ultimately, I will seek collaborators at UMESC and at UW-L to generate a geographic information system with aerial photography, elevation and flooding frequency data, which would provide an estimate of relative risk among sites for reed canary grass delivery to the forest understory. Combined with information about the effect of reed canary grass on tree seedling growth, this model will identify areas where future treefall gaps are likely to be invaded by reed canary grass, and thus guide management priorities.

To generate preliminary data for a grant application for this project, I propose to survey the amount and viability of reed canary grass seed in the soil within several forested sites in the summer of 2007. I will use aerial photographs to randomly select transects crossing the elevational gradient in contrasting sites, including islands. I will collect location and elevation data across the transects, and take soil cores to germinate at UMESC. By counting the number of reed canary grass seedlings emerging from each core, I will be able to estimate the number of viable seeds per unit area at each forest site. I can use the elevation and location information for each point to generate a simple model of how reed canary grass seed density varies across the landscape, and make preliminary predictions about the relative risk for invasion in treefall gaps in different areas.


Anticipated products

As mentioned at several points above, the proposed project will generate preliminary data I can use in applications to external funding sources. Future projects will be carried out in collaboration with researchers at UW-L, UMESC and US ACE, and will expand the scope of my research to the broader question of how reed canary grass affects floodplain forest regeneration. I intend to continue involving UW-L undergraduates in my research, and to seek external funding for more extensive summer projects. Furthermore, well- replicated restoration experiments are unfortunately something of a rarity, so the work proposed here will almost certainly generate a publishable manuscript. It will also provide valuable information to land managers in the area, and as such is receiving enthusiastic support at the US ACE.  Finally, any undergraduates involved in the project will present their results at the UW-L Undergraduate Research Day. 

References Cited

Adams, C.R., & Galatowitsch S.M. 2006. Increasing the effectiveness of reed canary

grass (Phalaris arundinacea L.) control in wet meadow restorations. Restoration Ecology 14: 441-451.

Adams, C.R. and S.M. Galatowitsch. 2005. Phalaris arundinacea (reed canary grass):

rapid growth and growth pattern in conditions approximating newly restored wetlands.  Ecoscience 12: 569-573.

Knutson, M.G., & Klaas, E.E. 1998. Floodplain forest loss and changes in forest

community composition and structure in the Upper Mississippi river: A wildlife habitat at risk. Natural Areas Journal 18: 138-150.

Lavergne, S., & Molofsky, J. 2004. Reed canary grass (Phalaris arundinacea) as a

biological model in the study of plant invasions. Critical Reviews in Plant Sciences 23: 415-429.

Urich, R., Swenson, G., & Nelson, E. 2002. Upper Mississippi and Illinois River

Floodplain Forests: Desired Future and Recommended actions. Upper Mississippi River Conservation Committee: Rock Island, Illinois.

United States Army Corps of Engineers, St. Paul District. 2006. Mississippi River

Operation Management Plan, Pool 8, Compartment 12, Management Units 1, 3, and 40. September 26, 2006.

Form II




Project Title   Promoting the regeneration of floodplain forest tree species in gaps invaded by reed canary grass


Project Director(s)        Meredith Thomsen                                                                               


I.    Salaries and Wages


      A.  Faculty Stipend(s)


            1.   Name         Meredith Thomsen                                $ 6,614.00                  

            2.   Name                                                                     $                     



      B.   Student Help                                                                $ 780 (half time for 2 months)


                                                                        Salaries & Wages Subtotal       $ 7680


II.   Travel                                                                                 $          500


      A.  Meals                                                               $          0         


      B.   Transportation                                                  $  500  (to study site and for survey)    


      C. Lodging                                                             $          0         


                                                                        Travel Subtotal             $          500     


III. Supplies & Services


      A.  Consumable Supplies                                                    $  500 


      B.   Duplicating, postage, communications, etc.                    $   2500 (soil analysis)


                                                                        Supplies & Services Subtotal    $          3000   


IV. Equipment                                                                          $1500 (soil moisture sensors)   


                                                                        Equipment Subtotal                   $          1500   



                                                                        TOTAL REQUESTED          $          12,394