Conservation Evidence strives to be as useful to conservationists as possible. Please take our survey to help the team improve our resource.

Providing evidence to improve practice

Individual study: Comparison of manual and herbicide control methods for Japanese, giant, and bohemian knotweeds (Fallopia japonica, F. sachalinense and F. X bohemicum) in the Sandy River Watershed, Oregon, USA

Published source details

Soll J., Kreuzer D., Strauss K., Dumont J., Jeidy L., Krass M., Aldassy C. & Nemens D. (2006) Sandy River Riparian Habitat Protection Project Report 2006. The Nature Conservancy report.


The Sandy River Watershed, a sub-basin of the Columbia River Basin, drains an area of 508 square miles (131,600 ha) within Multnomah and Clackamas Counties in Oregon. Despite its location near Portland, Oregon's largest population center, it supports rare and characteristic wildlife including twenty-two species of state or federal concern, such as Chinook Onchorhnychus tshawytscha and Coho salmon O.kisutch and steelhead O.mykiss). Two major sections of the Sandy River Watershed have federal 'Wild and Scenic River' and/or an 'Oregon State Scenic Waterway' designation.

The Sandy River Watershed's tendency towards catastrophic flooding and its proximity to developed landscapes (Portland, Gresham, Sandy, the Hoodland Corridor and the growing urban/suburban fringe), active nurseries and farms make it particularly vulnerable to water quality issues and invasions of well known noxious weeds such as Japanese knotweed Fallopia japonica (syn. Polygonum cuspidatum), giant knotweed F.sachalinense and bohemian knotweed F. X bohemicum, English ivy Hedera helix, Himalayan blackberry Rubus discolor, Scots broom Cytisus scoparius and new and previously unknown species of horticultural origin.

Invasive varieties of knotweed (F.japonica, F.sachalinense and F. X bohemicum) are able to rapidly colonize fresh sediment deposits and other low nutrient, disturbed sites, reproduce vegetatively, grow rapidly, form dense monocultures, and tolerate poor soils and long periods of submersion. These characteristics adapt it perfectly for life in the dynamic riparian and floodplain systems of the Pacific Northwest. They also enable knotweed to fundamentally alter the function of riparian systems by contributing to stream bank erosion, which can result in more sediment in the water and broader, shallower, and warmer waterways. Knotweed can also out-compete native vegetation including trees (through suppression of seedlings and saplings), which can result in an overall decrease in stream shading, decrease in woody debris, loss of appropriate riparian wildlife habitat, and ultimately a loss of aquatic invertebrate biodiversity.

Study area: The experiments took place within Oxbow Park on the Sandy River, Multnomah County, Oregon at river mile 13. Soils at the study site are sandy and the entire site is subject to inundation during major flood events. In April 2000, 45 individual knotweed patches were identified within a 0.5 square mile area. Each patch contained between 20 and 239 stems. Patches were numbered, permanently marked and their locations were recorded using a global positioning system (GPS). Each patch was randomly assigned to one of 15 treatment groups. Two sites were replaced in June, and 6 sites (summer treatments) were added in August 2000, for a total of 51 plots and 17 treatments. All sites are in full sun prior to bud break / leaf-out of deciduous trees in the spring. The degree of mid-summer shading varies between sites.

Treatment experiments: During 2000 through 2003 field seasons (May 2000 – June 2003), 17 treatment combinations were used and compared for controlling Japanese and giant knotweed. Treatments were: manual control, 2 herbicides (glyphosate and triclopyr), 2 application methods (foliar spray and wick), 3 application timings (spring and fall, summer only, fall only), in combinations of manual treatment with herbicides (spring cut and fall herbicide, early fall cut and herbicide treatment of resprouting stems, late fall (autumn) cutting and wicking, and late fall cutting to 1.5 m tall and foliar herbicide).

For manual control, each stem was cut monthly, at the top of root crown (if visible) or at the soil surface using loppers or pruning shears. For foliar spray application, upper leaf surfaces were sprayed using a low-pressure spray unit to "just wet" with a 5% solution of either glyphosate (Rodeo) or triclopyr (Garlon 3a, reduced to 3% after year 1.) A non-ionic surfactant (R-11 for Glyphosate in 2000 and 2001, Li-700 in 2002), Hasten for Garlon3a) was added at a rate of 1 ounce per gallon. A small amount of herbicide dye was also added to highlight areas that had been treated. For cut-stem (wicking) application, a 50% solution of triclopyr or glyphosate in water was applied to the stem surface immediately following cutting, using a weed wand (Ben Meadows) in 2000 and a hand type plant mister in 2001 and 2002.

This experiment resulted in an overall reduction in stem count of 80% (see Figure 1 for comparison of all 17 treatment combinations tested). However, subsequent monitoring at other sites in the watershed indicates that some knotweed plants are able to persist below the surface of the soil, indicating that monitoring of aboveground biomass alone may not be a sufficient measure of the effectiveness of knotweed treatments. For more details see Soll et. al 2006.

The results of this experiment suggest the following:

i) For the reasonably small patches that were tested, knotweed can be effectively controlled within two field seasons of foliar spray treatment with Garlon 3a, but may require three or more seasons of treatment with Rodeo. In many cases small stems with irregular growth may persist, especially, but not only with glyphosate. These conclusions have been born out by additional field trials over several years with the following additional caveats. Patches with initial stem counts > 50 appear to be unlikely to be eradicated with glyphosate treatment alone even with 3 or 4 years of treatment. Garlon used at rate of 2% rather than 5% appears to perform slightly better at achieving full eradication of both smaller and larger patches, but also appears to rarely provide full eradication of patches with initial stem numbers > 500;

ii) Although wicking type, cut-stem treatments can provide effective control, they are less effective and more time consuming than foliar type applications, and do not appear to completely control knotweed, even after three field seasons;

iii) Late summer/ early fall foliar herbicide treatment can be combined with spring manual control without loss of treatment effectiveness, as compared to two herbicide treatments. For Rodeo herbicide, a late season cutting to 1.5 m followed by foliar spray can deliver effective control if repeated for several seasons;

iv) Successful control based on cutting alone appears to require more than 3 years, and/ or involve cutting stems more than monthly. Three years of monthly cutting resulted in 80% stem reduction (healthy stems) and no epinastic growth. It was a useful method for small patches with easy access and a landowner committed to avoiding herbicides, but was not a viable landscape scale approach;

v) Timing of herbicide treatment is important. The failure of the spring-fall foliar herbicide treatment to deliver benefits beyond manual/herbicide combination is not surprising, since translocated herbicides generally do not give good control of deep-rooted perennial plants when applied during the early phase of rapid spring growth;

vi) The success of some of the cut-stem (wicking) treatment offers a middle ground for individuals with particularly strong objections to herbicide spraying, or for sites where herbicide spraying is not appropriate (i.e. presence of rare or sensitive species). Care must be taken, however, to treat every stem, and multiple treatments will be necessary.

For detailed results and analysis see Soll et al. 2006, which can be downloaded from: