We used a random sampling design of 21 focal tree plots and
indirect gradient analysis to determine site characteristics of plant
communities where D. oliveri occurs. Each focal tree plot was a
circular sample plot (r = 10 m) laid out around a D. oliveri tree,
which was randomly drawn from the population of D. oliveri trees
in the study area (Fig. 1). Plot coordinates and elevation were
recorded with a GPS (Eagle Expedition II, Lowrance Inc., Tulsa,
USA). In each plot, all woody species (excluding woody climbers)
with a girth at breast height (gbh, measured at 1.3 m) equal to or
exceeding 14 cm (corresponding to a diameter at breast height or
dbh 4.5 cm) were identified. Girth and height were measured
with a measuring tape and Suunto clinometer respectively. Stem
number and basal area per plot (314 m2) were recalculated to stem
number and basal area per hectare. Average, maximum and
dominant tree heights (mean height of the five tallest trees) were
calculated.