There is widespread concern that reed canary grass (Phalaris arundinaceae) is adversely affecting floodplain forest regeneration in the upper Mississippi River. Phalaris appears to out-compete tree seedlings in high-light environments such as forest edges and in gap areas caused by natural senescence, disturbance, or timber harvest. In fall 2006, we established a forest restoration experiment near La Crosse, WI, in a large forest clearing dominated by Phalaris. We used fall applications of pre-emergent herbicide in areas pre-treated with a Fecon™ mower, combined with grass-specific herbicides during the growing season. Glyphosate was also applied late in the fall of the first year, after most plants had senesced but Phalaris was just entering dormancy.

Results/Conclusions Although we have detected no differences among the specific herbicide combinations we compared, the treatments overall have been successful. The pre-emergent herbicides, applied in the fall, significantly slowed the springtime growth of Phalaris. As a result, grass-specific herbicides were still effective when the site was dry enough to allow for their application. By the end of the second summer, we no longer detected surviving established Phalaris individuals; germinating seedlings have also, to date, been effectively controlled. By the end of the third growing season, there was an average 90% decrease in reed canary grass cover in treated areas relative to untreated controls. Other members of the plant community appear to have responded to reduced competition from Phalaris, and a diverse mix of 60 native woody and wet meadow herbaceous species now dominate the site. We augmented natural tree seed inputs with seed additions and planted cuttings, but the majority of surviving tree seedlings resulted from natural recruitment, and their density is strongly affected by the proximity of canopy tree seed sources. Tree seedling density is unrelated to elevation in our site. The growth of established tree seedlings appears to be limited, however, by intense herbivory from white-tailed deer (Odocoileus virginianus. In the second fall of the experiment, approximately 75% of tree seedlings in treated areas showed signs of herbivory, compared to 1% of those inside small (2.25 m²) deer exclosures. Protected seedlings are also approximately 80% taller than those outside the exclosures. We conclude, therefore, that the successful restoration of floodplain forest in this and similar sites with large deer populations will require a two-phase approach: reduction of Phalaris cover first, followed by protection of established tree seedlings from deer herbivory.