Stability and predictability of diameter distributions in a managed uneven-aged oak forest

A privately-owned 156,000-acre oak forest in the Ozark Highlands of southern Missouri has been managed using the single-tree selection method of uneven-aged silviculture since 1954. Based on 40 years of forest inventory records, the diameter distributions of trees were analyzed at the forest-wide scale. Two diameter distribution models were compared: (1) the negative exponential model (N = {dollar}a{lcub}rm e{rcub}sp{lcub}b{lcub}rm D{rcub}{rcub}),{dollar} and (2) the power function model (N = {dollar}a{lcub}rm D{rcub}sp{lcub}b{rcub}),{dollar} where N is the frequency of occurrence of trees in 1-inch diameter classes (D). Results showed that diameter distributions closely conformed to the negative exponential model for all major species groups and stand density (stocking) classes based on 1962, 1977, and 1992 data. Most coefficients of determination for the associated regression models were {dollar}ge{dollar}0.95.; The stability of diameter distributions was evaluated by species groups and stocking classes. Slope coefficients for the negative exponential function were stable over 15- and 30-year intervals at high stocking ({dollar}>{dollar}57%), but tended to be unstable at lower stocking levels {dollar}({lcub}<{rcub}40{dollar}% and 40-57%). Among the species groups compared, white oaks (subgenus Leucobalanus) were least stable and the all-species group most stable. Statistical tests of the equality of the two parameters of the negative exponential function indicated there was significant discontinuity between diameter distributions above and below the minimum cutting diameter. The overall instability of diameter distributions appears to be related to self-limiting oscillations in slope shape and discontinuities that, within a cutting cycle, pivot about the minimum cutting DBH of 10 inches.; Predictability of diameter distributions by species groups and stocking class was assessed using the TWIGS model. Predictions were accurate only for red oaks at medium stocking (40-57%) and the other-species group at low stocking({dollar}<{dollar}40%). For the oaks, DBH was usually underestimated and for the other-species group was overestimated. Nevertheless, the slope shape parameter b, and thus q, was accurately predicted for both oak groups at medium and high stocking. TWIGS accurately predicted the stand density parameter, a, for white oaks and the other-species group at low and medium stocking, and for red oaks at medium and high stocking.; A linear regression model for adjusting TWIGS predictions of DBH indicated that prediction error was related to model deficiencies associated with stand stocking, species differences, and diameter growth rates. The adjusting model may be instructive in modifying TWIGS parameter estimates for specific application to the Ozark Highlands.