Oak mortality in sand savannas following drought in east-central Minnesota

Savanna plots that were established and sampled in 1984 at Cedar Creek Natural History Area in eastcentral Minnesota were resampled in 1989, towards the end of a major drought period. Tree diameter was measured and mortality was assessed for all stems ≥10 cm dbh in eleven 50 x 75 m plots. Quercus ellipsoidalis E. J. Hill (northern pin oak) and Q. macrocarpa Michx. (bur oak) dominate the savannas, accounting for >95% of all stems ≥10 cm dbh. In 1984, diameter distributions and species composition were significantly different on different soil types. Zimmerman soils had a larger proportion of stems in the 10-25 cm dbh size classes (62%) than did the Sartell soils (36-60%), and a significantly larger proportion overall of Q. ellipsoidalis stems (93%) than did the Sartells (51-84%). Plots burned during the previous 20 years had fewer stems in the 10-25 cm dbh size class (45%) compared to unburned plots (58%). These differences were not changed by mortality over the 1984-1989 period, nor did diameter distributions within a plot change significantly over this period. Average five-year mortality rates per plot from 1984-1989 for Q. ellipsoidalis (21.4%, range 0.0-60.0%) were significantly higher than for Q. macrocarpa (6.1%, range 0.0-33.3%). Five-year mortality rates of Q. ellipsoidalis declined from 60% to 15% with higher total stem basal area and density. Drought or large moisture changes may have been primary factors responsible for oak mortality. Other factors, such as percent organic matter, depth to water table, and fire frequency were not correlated with mortality rates. The inverse relation between mortality and stand density suggests that the possible effects of drought varied with stand structure. Closed savannas or woodlands had lower mortality rates than open savannas. These findings emphasize the role of climate in maintaining sand savannas in the prairie-forest border region.