Mississippi River Pool 3 Floodplain Forest Silvicultural Trials (US Army Corps of Engineers)

State or Province
Nearest city or town
Describe the location
E1/2 of Section 10, Township 114 North, Range 16 West Goodhue County, Minnesota. Located within the Mississippi River floodplain. The stand lies within Gores Wildlife Management Area between Mississippi river mile 805 and 806.
US Army Corps of Engineers - St. Paul District
Cover type
Plant community detail and growth stage
FFs68 (southern floodplain forest) and FFs59 (southern terrace forest). Mature forest - late stem exclusion/understory reinitiation.
Forest health threats
Adaptive silviculture options
Silviculture system
Estimated year of stand origin
Brief silvicultural objective
Promote regeneration of floodplain tree species and prevent conversion of site to reed canary grass. Provide wildlife habitat for the next 50 to 100 years.
Site preparation method
Soil texture
Soil details
Calco Silt loam, ponded, 0 to 1 percent slopes, frequently flooded (2% of area). Calco silt loam, 0 to 2 percent slopes, frequently flooded (91% of area). Calco-Udifluvents, loamy complex, 0 to 18 percent slopes, frequently flooded (7% of area).
Stand area
Treatment area

44.695476, -92.715319


The entire site is located within the Mississippi River floodplain to the west and south of the main river channel and encompasses most of the Army Corps of Engineers forest stand P03C03U046. Elevation is generally highest adjacent to the main channel and lowest to the south as the landscape transitions into a marshy wetland complex. This timber sale was conducted with a primary objective of testing different silvicultural approaches to prevent the conversion of floodplain forest to a reed canary grass dominated wet meadow. Before harvest, the site had aging silver maple-cottonwood overstory with younger age classes being almost entirely absent. As the canopy naturally opened due to canopy mortality, reed canary grass began to invade the site. The harvest was implemented to facilitate establishment and growth of young tree seedlings and to reduce the dominance of reed canary grass.

Bottomland hardwood stand data:

River: Mississippi River
River type: Major river
Floodplain topographic position: Bottom, slough, ridge

Figure 1: Reed canary grass remains a problem to seedling regeneration in treatment areas.  

The area was divided into 3 areas with different canopy removal treatments being applied: shelterwood, group selection (patch cuts) and clearcut. Site preparation occurred post-harvest through log skidding, slash manipulation, forestry mulching (Fecon) and herbicide work. Forestry mulching is a harvesting method that was utilized to cut, grind, and clear vegetation with the use of  mechanical equipment. Each of the mechanical treatments, including the forestry mulching, were intended to disturb the upper layers of the soil and expose bare mineral soil for seedling regeneration, while also breaking down residual slash. Site prep treatments were planned to coincide with tree seed drop to improve chances of natural regeneration. 

Silviculture Objective(s)

This area was used to test three different harvesting treatments along with four different site prep treatments (no site prep, natural regeneration only, forestry mulching with planting and herbicide, and forestry mulching with planting and no herbicide) in order to determine the best method for regenerating floodplain tree species.

Figure 2: Initial site prep prior to planting during the first growing season after harvest.

Pre-treatment stand description and condition

Pre-treatment species composition: 

The most recent inventory data at the time of harvest was collected in 1993. At that time, the site was dominated by silver maple and green ash (Table 1). Trees present but not considered dominant in the overstory included: black ash, American elm, hackberry, boxelder and American basswood. American elm and hackberry were the most common midstory species. No woody regeneration was recorded in the 1993 inventory. 

Pre-treatment growth and stocking: 

The mean diameter, BA, TPA and % BA for the dominant species can be seen in Table 1.

Table 1. Overstory data from 1993 forest inventory in P03C03U046. More recent data was not available at the time of harvest. 

Species Mean Diameter (QMD) BA TPA % BA
Silver Maple 24.1 67.5 21.3 70.6
Green Ash 18.3 15.0 8.2 15.7
American Elm 5.7 4.7 26.2 4.9
Boxelder 10.2 4.1 7.1 4.2
Hackberry 7.3 1.9 6.4 2.0
American Basswood 14.4 1.6 1.4 1.6
Eastern Cottonwood 41.3 0.9 0.1 1.0
Total 15.8 95.6 70.6   -

Pre-treatment forest health issues: 

Death of the forest canopy and low recruitment of new floodplain trees has allowed reed canary grass to encroach on sites suitable for floodplain forest regeneration. Ash is present as a component but not dominant, though all ash on the site is vulnerable to emerald ash borer mortality. 

Landowner objectives/situation: 

Establish a new cohort of floodplain forest trees to maintain forest canopy and improve wildlife habitat with a target of a minimum of 200 established stems of regeneration per acre across 75% of the harvest area. 

Silviculture Prescription

The upper ⅓ of the unit (11 acres) will use an extended shelterwood system (Figure 3, solid red polygon) in which residual canopy trees will be retained until natural mortality occurs. Trees selected for retention will be those most likely to survive and located at appropriate spacing throughout the stand. Up to ½ of the standing volume will be removed from this 11 acre area to increase light to promote seedling establishment. Site preparation shall be completed using log skidding, slash piling, and use of a Fecon forestry mulcher. No herbicide will be applied in this area and no tree planting will be implemented. 

Figure 3: Map of Gores Silvicultural Prescription and Treatment Areas.

The middle ⅓  of the unit (5 acres) will implement group selection (patch clearcut) harvest (Figure 3, yellow polygons). Across the group selection area, up to ½ of the standing volume will be removed by harvesting all trees in eleven different groups, ranging from 0.1 to 2.2 acres, with an average of 0.4 acres. Groups 2, 3, 5, 6, 9, 10, and 11 will be treated with the Fecon followed by herbicide application and tree planting. Tree species listed in Table 3 below will be planted in these groups. Follow up weed control will be performed for the next three growing seasons if needed. No post-harvest treatments are planned for groups 1,4,7 and 8 to determine the effectiveness of natural regeneration without site prep. 

In the lower ⅓ of the unit (19 acres), a clearcut will be implemented to remove all merchantable trees aside from snags and scattered leave trees selected from young growing stock of desirable species (primarily hackberry and basswood). The clearcut is divided into four post-harvest site prep treatments, each approximately five acres in size. One treatment (Figure 3, red-filled polygons) will be a control in which natural regeneration without site prep will be assessed. There will be two planting treatments, one with herbicide application but no forestry mulching (Figure 3, orange filled polygons) while the other will include both herbicide and forestry mulching. The final treatment is a natural regeneration without forestry mulching (Figure 3, green polygon). The first area outlined in blue will be 0.14 acres and have 85 enriched elm seedlings and 27 swamp white oak seedlings. The second area is outlined in orange and will be 0.08 acres. There will be 57 enriched elm seedlings and 18 swamp white oak seedlings planted in this area.  All seedlings planted in the clearcut area are listed in Table 2.

In all harvest areas, snags and other selected trees for wildlife will be left in place. Follow-up vegetation control treatments will be completed for three years post-harvest to promote establishment of tree seedlings. Regular monitoring of seedling survival and growth will also be conducted during this time.  

All bare root seedlings will be dipped in mycorrhizal fungi root dip.  

Table 2. Tree seedlings planted in the Gores timber harvest area. Seedlings were 1-0 bareroot stock. A total of 14 acres were planted. No planting occurred in the shelterwood. Stems per acre in 2016 includes natural seeding so results may be higher than the initial planting numbers. 

Tree Species Seedling Count, Groups Seedling Count, Clearcut Seedlings Planted Per Acre 2016 Stems per acre (>1 ft) 2016 % >1 ft
Black Walnut 400 0 29 0 0%
Swamp White Oak 1192 608 129 11 9%
American Elm (enriched) 1358 142 107 47 44%
Eastern Cottonwood 402 1400 129 0 0%
River Birch 0 700 50 0 0%
Silver Maple 0 700 50 105 210%
Total Seedlings 3352 3550 494 163 33%

What actually happened during the treatment

The timeline of events for the implementation of treatments is described below in Table 3.

Table 3. A timeline of events for follow-up treatments post-harvest in the Gores timber harvest area. 

Month and Year Activity Treatment Specifics
February 2013 Timber harvest Approx. 213,000 bd. ft.
June 2013 Forestry mulching Fecon Forestry Mulcher
June 2013 Post-emergent herbicide application Rodeo (glyphosate), 5% in water
June 2013 Planting of bareroot seedlings Species in Table 3
November 2013 Post-emergent herbicide application Oust (sulfometuron methyl), 1 oz/ac.
August 2014 Post-emergent herbicide application Cornerstone (glyphosate), 3% in water; Envoy (clethodim), 2% in water
November 2014 Post-emergent herbicide application Oust (sulfometuron methyl), 1 oz./ac.
August 2015 Post-emergent herbicide application Cornerstone (glyphosate), 3% in water; Envoy (clethodim), 2% in water
November 2015 Post-emergent herbicide application Oust (sulfometuron methyl), 1 oz./ac.
August 2016 Mechanical release around seedlings String trimmers

Figure 4: 2013 Winter harvest with harvesting equipment present in the background.

Post-treatment assessment

Plant regeneration surveys occurred in 2014 and 2015. A comprehensive survey of the entire area was completed in 2016. Because the entire site was not surveyed prior to 2016, it is challenging to make any statements about regeneration as a whole immediately post-harvest. Group 7 had no management (but was relatively open prior to the 2013 harvest) and had the highest number of stems per acre for ash, cottonwood and elm out of all the surveyed areas (Figure 5).

Figure 5: Gores Wildlife Management Area regen survey summary data excluding silver maple. All seedling size classes are included.  Not all areas were surveyed in both 2014 and 2015.

Across all the areas surveyed in 2014 and 2015, ash and cottonwood appeared to do very well, while hackberry also established well on some of the sites. With cottonwood the only one of these species that was planted, and with the high densities, it can be assumed that most of this was natural regeneration. Oak and elm were also present in 2014 and 2015 at densities similar to the planting densities, indicating that survival in the surveyed sites was good. In addition, silver maple established in very high densities across all surveyed areas (Figure 6), with the exception of the shelterwood. The 2014 and 2015 surveys did not account for seedlings size, but it can be assumed that most seedlings in those surveys were small (< 1 foot tall), unless they were planted. 

Figure 6: Gores Wildlife Management Area regen survey summary data including silver maple. All seedling size classes are included. Not all areas were surveyed in 2014 and 2015. 

By 2016, the total number of seedlings per acre had decreased to an average of about 15,000, with less than 1,000 of those being greater than 1 foot tall (Figure 7). Overall, planted survival was low, with only swamp white oak, elm, and silver maple having any surviving seedlings in 2016 surveys (Table 2). Much of the maple regen may have been natural, so it is likely that mortality for planted silver maple was also high. No surviving walnut, cottonwood or river birch seedlings were found. The clearcut areas with forestry mulching and natural regen had the highest density of taller seedlings, though the large majority of those were ash, which are unlikely to survive to maturity due to emerald ash borer. The group selection areas with forestry mulching, herbicide and planting had the highest density of large seedlings that was not dominated by ash. These same areas also had the highest density of small seedlings (<1 ft tall), and these were heavily dominated by silver maple. The clearcut areas with forestry mulching, herbicide and planting had the next highest density. 

Figure 7:  Gores Wildlife Management Area seedling summary 2016. . All 2013 harvest sites were surveyed. Large regen consists of tree seedlings 1 ft. tall and larger, small regen is <1 foot tall. 

Based on an assessment of all seedlings, the site is stocked at levels well above the target of 200 seedlings per acre. Larger seedlings provide a more reliable estimate of seedlings likely to survive, and when considering only large regeneration, the average across all plots is about 225 stems per acre (excluding ash). However, when considering planted species, only 163 stems per acre were noted in surveys. Regardless, due to the large cohort of natural silver maple and, to some extent, hackberry, the site is stocked above the target level as of the 2016 survey.

The data from the elm and swamp white oak comparison planting showed that there was no mortality in 2013. In 2014, mortality occurred in both species across all areas surveyed. Elm survival was higher compared to swamp white oak, even though more mortality occurred in elm seedlings. By 2016, elm survival was approximately 50% of planted seedlings, while oak survival was about 75%. Of the surviving seedlings, the average height was just above 1 foot tall for the elms and just under 1 foot for the oaks, indicating little to no growth in the three growing seasons following planting.  

Deer herbivory was influential to survival and growth, with evidence of browse present on a vast majority of seedlings (Figure 6). The average seedling was classified as 25-50% of the twigs showing evidence of browse. Browse was noticeably higher in the shelterwood and group selection areas than in the clearcut areas. Silver maple and elm had the highest browse overall, though large swamp white oak and large elm showed high levels of browse (>50%) on all seedlings inventoried. No particular species, however, showed distinctly higher or lower levels of browse, indicating that deer were having a relatively uniform impact across all species. 


Figure 8: Gores Wildlife Management Area seedling summary 2016. All 2013 harvest sites were surveyed. Large regeneration consists of tree seedlings 1 ft. tall and larger, small regen is <1 foot tall.

In areas treated with herbicide, reed canary grass has been very effectively controlled, and vegetation is currently dominated by broadleaf herbaceous plants, such as stinging nettle and pigweed. In clearcut and group selection areas not treated with herbicide, reed canary grass remains dominant over large areas. In the shelterwood area there are patches of reed canary grass mixed with larger areas of native herbaceous vegetation.

It should be noted that the site experienced high water levels during much of the growing season in 2013 and especially in 2014 (Figure 9). High water occurred at times when floodplain trees drop seed and almost certainly would have had an impact on survival and growth of tree seedlings. 

Figure 9: Water level elevations at Lock and Dam 2, Hastings, MN, for the growing season (Apr. 1 – Sept. 30) for the three growing seasons post-harvest. The mean elevation is calculated for each day between 1988 and 2018.

Plans for future treatments

An informal, small scale study was initiated in 2017 to test the effectiveness of Plantskydd deer repellant in improving silver maple growth. Six small, paired plots were installed in which Plantskydd was sprayed on one plot and not on the other in June and November 2017. Initial results were promising (Figure 10), with somewhat higher growth on sprayed seedlings than on controls. High water prevented spraying in June and July of 2018, but the plots will be sprayed again in fall of 2018 and monitored into 2019. Based on the apparent success of this application, broadcast spraying of Plantskydd was scheduled for June of 2017 but could not be implemented due to high water. 

Figure 10: Silver maple seedling in a PlantsSkydd spray plot. 

The no herbicide areas within the clearcut that reverted to dominance by reed canary grass will be treated beginning in summer 2018. A targeted planting within these areas using a technique of planting seedlings in blocks will be applied, as will a trial test of direct seeding by hand. All seedlings planted in these areas in 2018-19 will receive a minimum of two growing seasons of competing vegetation control and will be sprayed annually in June with Plantskydd.

Summary / lessons learned / additional thoughts

1. Artificial regeneration in the floodplain remains challenging, with small seedlings vulnerable to overtopping during unpredictable flood events. More focus needs to be put on utilizing larger stocktypes and targeting seedlings to microtopography in sites to increase planting success.

2. June planting is later than optimal, but access in the floodplain can be difficult earlier in the spring. Developing techniques that allow for flexibility in planting prescriptions (i.e. multiple potential locations) or seedling supply could help to mitigate this challenge.

3. Much higher seedling densities can be achieved with reliance on natural regeneration, but herbicide work (at a minimum) and (optimally) some amount of soil disturbance, is required to create site conditions necessary for seed to germinate and establish. Timing of this is likely important, with bare soil available in May and June when silver maple and cottonwood drop seeds.

4. Deer browse appears to have a substantial impact on reducing seedling height growth, making seedlings more vulnerable to overtopping during flood events. Deer browse most likely interacts with flooding to increase mortality.

5. Deer browse was lower in the clearcut than in the group selections and herbaceous competition was more vigorous.

6. Water levels are unpredictable and regeneration success may require multiple entries and/or patience over a period of 5-10 years. However, a do-nothing approach will eventually lead to the conversion of floodplain forest sites to reed canary grass or other non-woody herbaceous plants (e.g. nettle). If maintenance of floodplain forest is an objective, management is necessary.

This case study was developed with support from the United States Department of Agriculture's National Institute for Food and Agriculture (USDA-NIFA), Renewable Resources Extension Act (RREA). Project #MIN-44-E02, principal investigator Eli Sagor, University of Minnesota.