Data Preparation
Data were combined into Midwestern (Table S1) and Oklahoma (Table S2) datasets. Each dataset was formatted and analyzed using tools in thespatstat (Baddeley et al. 2015) package in R (R Core Team 2021). For each sample plot, the coordinates and measurable attributes (deemed marks ) of each point were converted into marked multivariate point-pattern data (ppp objects). The pppobjects were combined into a two multivariate hyperframes(the Midwestern and Oklahoma, respectively), wherein each row corresponds with one sample site. Working in a hyperframe allowed analytical operations to be performed on an entire dataset at a time.
Prior to analyses, the unmarked point-pattern data associated with each site were checked for homogeneity. The Midwestern and Oklahoma hyperframes were subset into groups based on the area of the sampling window of each plot (see Table S1 and Table S2). A quadrat-count test was conducted to establish whether a given point-pattern had a homogeneous distribution of points. In the Midwestern dataset, certain plots were inhomogeneous, and thus, subsequent analyses conducted on those plots accounted for that characteristic. All 13 point-patterns in the Oklahoma dataset were homogeneous.