Estimating cavity tree abundance by stand age and basal area, Missouri, USA

Abstract

We analyzed cavity tree distribution among Missouri Forest Inventory and Analysis (FIA) plots using the nonparametric classification and regression tree (CART) model and Weibull probability density function (pdf). Fifty-nine per cent (2370) of the 4052 Forest Inventory and Analysis plots (aged 1–160 years) had at least one cavity tree. The overall odds ratio of a plot with cavity trees (odds of a plot having cavity trees/odds of a plot with no cavity trees) across the five survey units of the entire state was 1.4. Three and four disjoint clusters (nodes) which differ significantly in cavity tree distribution were identified by CART using the two most discriminating stand level indicator variables: age and basal area, respectively. Cavity tree density distribution within each cluster was further described by the Weibull pdf. Cavity tree density per hectares varied considerably among stands (plots) of the same age or density, and the number of cavities for a given size or age class was distributed in an asymmetric form (primarily reverse-J shape). CART partitioning and Weibull fitting, in combination, provide an intuitive way to depict cavity tree distribution (variation) by important stand indicator variables such as age and basal area. This information can help forest managers and planners formulate management guidelines and results can be linked with forest landscape planning efforts, regional inventories, wildlife habitat modeling, and landscape simulation to evaluate or predict the consequences of different management alternatives.