Michael C. Dietze and James S. Clark. 2008. Changing the gap dynamics paradigm: vegetative regeneration control on forest response to disturbance. Ecological Monographs 78:331–347.

Appendix A. Detailed description of stand selection, plot layout, and experimental gap creation.

Stand and plot selection

Stands were selected based on extensive consultation with the site directors at each research site in order to find a sufficiently large track of representative forest that was compatible with each site’s research goals and land-use designations (e.g., experimental vs. control). This experiment is expected to run for many decades and thus required incorporation into long-term site planning. Within each stand approximate future gap locations were chosen based on the constraints of maximizing separation among gaps while minimizing the total area of stand that needed to be mapped. Plot selection also involved ensuring that plots were close enough to dirt roads to allow access of the heavy machinery that would create the gaps while also avoiding the impacts of edge effects. Maps of the final plot locations are given in Fig. A1. The intervention design of the study, with collection of pretreatment data on the growth, survival and reproduction of trees across all their life history strategies, greatly helps to reduce the impact of any biases in site selection. Once the center of each future gap was located, a set of North-South and East-West transects was set up in each gap to measure seedlings, seed rain, soil seed bank, soil moisture and nutrients, and understory light. These transects extended 15–25 m beyond the anticipated gap edge depending on gap size. All trees within 10 m (small gaps) or 20 m (large gaps) of the future gap center that were visually determined to be dominant or codominant in the canopy were then marked to be pulled. All trees that were within a 20–30 m buffer around the transects and the anticipated gap edge were then mapped relative to a surveyed 10-m grid as noted in the Methods.

Gap creation

Gaps were created in March 2002 at both sites based on the methodology presented in Cooper-Ellis et al. 1999. The direction of treefall was determined primarily by access to the plot and all trees within a gap were pulled in the same direction. Mean treefall direction is depicted in Fig. A1 as is the shape of the gap. At both sites a 5-ton winch was used to pull the trees by attaching a 1" steel cable to the trunk near the base of the crown using either long ladders or climbing the trees with spikes. At Duke Forest the winch was attached to a logging skidder while at Coweeta it was attached to a bulldozer. Trees were pulled until either the stem snapped or the tree was either uprooted or down. All damaged trees were left in place. Falling trees created a substantial amount of damage to the understory; in terms of number of stems approximately twice as many trees were damaged indirectly as were pulled. While logistical constraints prevented pulling the trees during the hurricane season, all trees were pulled when the ground was wet and before leaf-out, thus we feel that we were still able to capture the correct phenology of large windthrow gap formation, which occurs mostly during the fall.

FIG. A1. Duke Forest and Coweeta stem map and gap polygons. Black points represent stem locations, while the red dot and line represent the center of each gap and the average treefall direction for each gap, with ray length equal to intended gap radius (10 m or 20 m). The polygons represent two extremes in gap definition, based either on a convex hull around all damaged stems (red in Duke, green in Coweeta) or a projected gap based on aerial imagery (available for Duke Forest only, green). Unmapped areas surrounding the stands consist primarily of similar forest types.