Characterization and multivariate analysis of floodplain forest communities in Mississippi River Pool 24

Background/Question/Methods

Flood control structures (e.g., levees) and navigation dams on the Upper Mississippi River modify river hydrology, potentially changing the composition of floodplain forests. Channel constriction and impoundment contribute to differing patterns of water level variation within navigation pools.  We hypothesized lowland forest community composition responds to the hydrologic gradient within the navigation pools. This study is intended to (1) quantify differing patterns in forest community within Navigation Pool 24 and (2) describe the differential responses of species in relation to the varying hydrology. Forest survey, river level (1993-2011), and elevation data were integrated using a GIS. Non-metric multidimensional scaling was used to ordinate floodplain forest stands defined by combinations of river mile and elevation classes.

Results/Conclusions

Ordination separated stands on 2 axes. The first separated along a gradient from small to large boxelder (Acer negundo; r_s(n=15) = 0.9429, p < 0.01), silver maple (Acer saccharinum; r_s(n=15) = 0.925, p < 0.05), and green ash (Fraxinus pennsylvanica; r_s(n=15) = 0.9571, p < 0.01). The second axis separated along a gradient of small to large eastern cottonwood (Populus deltoides; r_s(n=15) = 0.6679, p < 0.01) and willow (Salix spp.; r_s(n=15) = 0.9214, p < 0.05). River level variability (SD of water level) positively correlated (r_s(n=15) = 0.67,  p < 0.01) with the first axis. Permutation tests (MRPP) separated all elevation groups of lowest river mile group (274-278) from all other river mile groups (p < 0.05), and separated the middle group (284-288) from the fourth (289-293; p = 0.03) and fifth (294-298; p = 0.04). These results suggest that Pool 24 floodplain forest communities differ in species dominance patterns as a result of their relative position within the hydrologic gradient between locks and dams.