In 1996, 400 tree-centered plots were established by first randomly choosing x- and y- coordinates from an imaginary grid overlaying the study area. Each random point was also randomly assigned a tree-size category from a pre-determined sampling scheme. The scheme was to include 20 trees from each of 5 size categories. Size/height categories were: less than 0.1m, 0.1-1m, 1-2m, 2-3m, and greater than 3.0m. To avoid excessive aggregation of samples, no more than 5 trees in a size class could be chosen within a 100-meter sub-area of the sampling grid. The nearest tree to each random point that met the pre-determined size requirement was identified and marked with a permanent aluminum tag. For the size category less than 0.1m tall, field-collected seedlings were transplanted into plots. Transplanted seedlings were monitored and watered for 2 months prior to pre-burn measurements. Pre-burn measurements included: maximum tree height (using telescoping measuring poles), non-destructive estimation of total community biomass (Whitbeck and Grace 2006), and herbaceous community species composition. Sub-areas within the study area were randomly chosen for burning in either the growing or dormant season. Dormant-season burns were conducted in January 1997 and growing-season burns in August 1997. Prescribed burns were conducted by USFWS fire crews according to normal prescriptions for the area. Plots were resampled immediately following fire to determine burn completeness of plots, as well as for general observations. Effectiveness of fire treatments were assessed at the end of the subsequent growing season (September 1998). At that time, maximum height of trees was remeasured, along with general community characteristics.