Floodplain forest on the upper Mississippi River (UMR), a unique habitat in the Midwest that is important for many bird species, has been reduced and is undergoing continued reduction and changes in structure and species diversity because of river engineering and invasive species. Hydrological changes are causing tree diversity to decline favoring Acer saccharinum (silver maple) and Fraxinus pennsylvanica (green ash). Invasive Phalaris arundinacea (reed canary grass, Phalaris) threatens tree regeneration, and recent Agrilus planipennis (emerald ash borer) arrival threatens to decimate the important ash component of the forest canopy. During the 1990s, virtually no information was available about breeding songbird species and abundances on the UMR floodplain forest from along many river miles and a broad range of forest situations (for example, mainland, island, edge, interior). From 1994 to 1997, we surveyed breeding birds and sampled vegetation at 391 random points on UMR floodplain forest along a latitudinal gradient from Red Wing, Minnesota, to Clinton, Iowa, to characterize bird assemblages and associations with gradients in forest structure at survey points (local scale) and land cover composition within a 200-meter radius of survey points (landscape scale).

Eighty-six bird species were detected during the study, but 28 species comprised 90 percent of all detections. Species that are typically associated with woodland edge or are tolerant of fragmentation were the most common: Setophaga ruticilla (American Redstart), Troglodytes aedon (House Wren), Turdus migratorius (American Robin), Quiscalus quiscula (Common Grackle), and Vireo gilvus (Warbling Vireo). Species typically associated with large forest patches—Setophaga cerulea (Cerulean Warbler), Hylocichla mustelina (Wood Thrush), and Dryocopus pileatus (Pileated Woodpecker)—were rare. Principal components analyses consistently described local habitat gradients related to canopy cover and Phalaris presence and described landscape gradients related to forest area and areas of open land cover types. However, nonmetric multidimensional scaling revealed no pattern in bird assemblages. Canonical correspondence analyses with local habitat variables for each year revealed that bird assemblages were affected by canopy cover, the presence of Phalaris, and the number of tree species. Four bird species were consistently associated with Phalaris presence or negatively with canopy cover, and no species were associated with the number of tree species variable. Although landscape variables were significantly related to the bird assemblage in canonical correspondence analyses, no bird species were consistently related to any landscape variable. These results indicate that there is one assemblage of forest birds on the UMR composed mainly of edge-tolerant species. Species associated with lower canopy cover and Phalaris presence may be favored to increase in abundance as canopy cover opens as trees die and Phalaris becomes more prevalent.