Area of competitive influence and tree density scale with tree diameter in a central hardwood old-growth forest

Background/Question/Methods Examining tree spacing can give insight into competitive filtering of forest composition and illuminate the processes that shape forest dynamics. Success in plant competition has been experimentally shown to positively correlate with plant biomass. However, spacing between trees in temperate forest communities has received little attention in the literature. Most previous research has investigated plantation spacing with little attention given to natural communities. In contrast, there have been multiple studies of spatial pattern and community dynamics in tropical forests where it has been found that area of influence does not scale with tree diameter at breast height (DBH). We present an analysis of an old-growth temperate deciduous forest to test the hypothesis that area of influence of a tree scales with DBH and to provide comparative data with tropical forests.

Utilizing mapped tree census data from a 4-ha area in a 25 ha plot, we analyzed the spatial pattern using Thiessen polygons as a proxy for the area of influence. We regressed polygon area against DBH. We analyzed the data with the spatial statistics Ripley’s L-function and the pair-correlation function to examine competitive interactions between forest tree species and to generate hypotheses on the processes that created the observed patterns.

Results/Conclusions

The results indicate that polygon area scales with DBH in OLS regression for all trees ≥10cm DBH. Mean polygon area was 29.42 m² and the regression of polygon area on DBH was significantly positive (F_1,1213 = 6.26, P = 0.012). Two dominant species were analyzed independently. Acer saccarum polygon area was strongly positively correlated with DBH (F_1,617 = 145.33, P < 0.0001). In contrast, Quercus prinus polygon area was not correlated with DBH (F_1,295 = 2.89, P < 0.09). Analysis of tree density with Ripley’s L function indicates that trees were clustered at scales ≤ 8m while they displayed dispersion at ≥ 12m. Results from the pair-correlation function indicate that small trees have a dispersed pattern at distances less than 3m and otherwise are randomly arranged with respect to larger trees. These results are compared with tropical forest tree patterns. Several possible mechanisms that could generate these patterns will be discussed.