Whole Stand Variable Density Yield Equations for Oak-Gum-Cypress Bottomland Hardwood Forests

Abstract

Variable density yield equations were constructed using fuzzy linear regression for bottomland oak-gum-cypress forests along the US Gulf Coast and lower Mississippi River Delta region. The USDA Forest Service’s Forest Inventory and Analysis program provided plot data (n = 526). Predictors included stand age, growing stock basal area per acre, sweetgum site index (base age 50 years), and US Environmental Protection Agency ecoregion dummy variables located in Alabama, Arkansas, Louisiana, Mississippi, Tennessee, and Texas. Dependent variables were per acre growing stock cubic foot yield (GSV) and Doyle board foot sawlog yield (SLV). Plots averaged 58 years, 90 ft2/ac basal area, 79 ft site index, 2,556 ft3/ac GSV, and 11,183 Doyle bf/ac SLV. Adjusted R2 were 0.98 (GSV) and 0.77 (SLV). Basal area possessed fuzziness in the GSV model, whereas the SLV model’s intercept was fuzzy. Six ecoregions possessed fuzziness in each model, but these were not identical across models. Study Implications: Some forestry measures at the stand level can be inherently vague or not as crisp as their reported values may suggest. Fuzzy linear regression can help overcome these imprecisions. Basal area per acre, which depends on average tree size and stand density, was both fuzzy and the most critical predictor of growing stock volume. The study’s relevance for clientele in the region specifically includes inventorying and appraising lands of the oak-gum-cypress mix. Narrower yield prediction intervals permit focusing resources on minimizing other sources of error when setting a reservation price (for landowners) or formulating a bid price (for buyers).