Figure 2. K-function envelopes and Diggle-Cressie-Loosmore-Ford (DCLF) test p-values weighted by tree diameter for select sites in the Midwestern point-pattern dataset. Dashed red lines represent the theoretical Poisson distribution. Grey shaded sections are 95% acceptance envelopes. The black line illustrates the empirical K-function. Departure of the black line from the envelope signify a non-random distribution where points above indicate clustering and below indicate overdispersion.
Examination of Dixon contingency tables of points marked either by species (redcedar vs deciduous trees) or by size of trees (DBH < 2.5 cm vs DBH > 2.5 cm) revealed segregation in many of study sites in the Midwest all trees dataset andMidwest living trees datasets (Table S3). When considering Dixon contingency tests of only two classes (e.g. small vs large), significant findings indicate the nearest neighbor is likely to be of adifferent type. At Lakeside 1 in the Midwest all trees andMidwest living trees datasets the nearest neighbor of deciduous species was more likely to be a redcedar tree than another deciduous tree. In both datasets, the redcedars at Edge 5 were segregated from other redcedar trees. Instances of segregation according to size in theMidwest all trees dataset was mixed. In Edge 4 and Gettysburg 1, large trees were segregated from each other.