Natural Disturbance Regimes in Hemlock‐Hardwood Forests of the Upper Great Lakes Region

The frequency of natural disturbances and their influence on the forest landscape mosaic were investigated on three large tracts of primary forest in Upper Michigan. Seventy 0.5—ha plots were randomly distributed in a total forest area of 23 000 ha dominated by sugar maple (Acer saccharum) and eastern hemlock (Tsuga canadensis). Radial increment patterns were used to estimate canopy accession dates for each of a number of randomly selected overstory trees on each plot. From these data a disturbance chronology, representing the percentage of stand area occupied by cohorts originating during each decade over the last 130 yr, was compiled for each plot. Average rates of disturbance or canopy mortality are estimated at 5.7 to 6.9% per decade. The corresponding average canopy residence time of a tree is 145—175 yr. No significant differences were detected in average disturbance rates among the three study areas, between plots near the coast of Lake Superior and inland plots, among several different aspects, and among several different slope positions. Natural rotation periods increase exponentially with increasing disturbance intensity, which is defined as the approximate percentage of the plot area converted to gaps during a disturbance episode. Estimates of rotation periods range from 69 yr for ≥10% canopy removal to 1920 yr for ≥60% canopy removal. Spatial autocorrelation analysis indicated that plots with light and medium disturbances (<40%) are randomly distributed over the landscape. Plots wit heavy disturbances (≥40%) are clustered with a patch radius of ≈2 km, consistent with the sizes of thunderstorm downbursts. The data indicate that light and medium disturbances dominate the disturbance regime. The majority of stands on the landscape are composed of several major and many minor age classes. Even—aged stands with one predominant age class are uncommon. The age distribution of individual patches or cohorts in the two larger study areas (14 500 and 6073 ha) follows a nearly uniform distribution. None of the three study areas had more than 15% of the forest area converted to gaps in a single decade. The two larger areas meet most of the criteria that have been proposed for equilibrium landscapes.