Comparing fixed- and variable-sized belt transects for quantifying woody structure in savanna ecosystems using BECVOL-3 models

Abstract

Accurate quantification of woody vegetation structure is essential for assessing browse availability and informing rangeland management in savanna ecosystems. This study compared fixed-size belt transects (FBT) with variable-size belt transects (VBT) for generating datasets to estimate browsing tree equivalents per hectare (BTE/ha) using BECVOL-3 models in Musina Mopane Bushveld and Limpopo Ridge Bushveld. Paired sampling at 12 sites enabled direct comparison of plant-level measurements, density estimates and derived biomass metrics. Generalised linear mixed models showed that VBT produced substantially lower BTE/ha values than FBT – approximately 45% lower after accounting for crown diameter and mean plant spacing. Plant-level volume results did not differ significantly between techniques, confirming that techniques did not differ in estimating plant volume. Instead, discrepancies arose from density estimation. A strong interaction between Technique and Mean Plant Spacing indicated that VBT yielded progressively lower density estimates as spacing increased, whereas FBT were not as sensitive to changes in plant spacing. These results demonstrate that VBT provides structurally responsive density estimates and may offer more ecologically realistic assessments of woody biomass in heterogeneous, low-density savanna systems, with important implications for browse capacity and rangeland management.