Data was converted from long to wide format, and species names were standardized across the years. Species were then categorized into functional groups using vegetation type, life cycle, and status. Total cover of these categories was added to the data table (Table 1).

The sampling units are the plots, each represented by a unique plot ID which includes year, treatments, site, block, and plot. At KCPP, an additional feature (Type) was recorded to determine if sites were NG or PG type. The predictor variables are spray and burn. At KCPP, plots were sprayed with one of three treatments (control, HerbicideC, or HerbicideR). At YFN, plots were sprayed with one of four treatments (control, HerbicideR, HerbicideP, or HerbicideR + Herbicide P). At both sites, plots were either burned or unburned. These are manipulated variables, and are categorical in nature. The response variables are the species/ground cover collected. Originally, each species was collected using a seven-letter code made from the first four letters of the genus and the first three letters of the specific epithet. Species covers were categorized into functional groups. The functional groups chosen for analysis were Bunchgrasses, Invasive Forbs, Native Forbs, Non-bunchgrasses, Annual Invasive Grasses, and Shrubs.

Change from the first year of monitoring to the last year of monitoring was calculated for each desired category. At KCPP, this consisted of changes from 2023 to 2025, and at YFN from 2024 to 2025. This change represented the difference between average covers before and after treatments.

Exploratory Graphics

To check data for errors, boxplots were constructed. This allowed any outliers and data trends to be observed. At pinegrass sites most data appeared normal, with several outliers (Figure 1). Trends were similar for needlegrass sites (Figure 2).

Figure 1. Boxplots of changes in foliar cover of species functional groups as affected by treatments averaged over pinegrass sites at KCPP.  Average change in cover is represented on the Y axis. Treatments are represented on the X axis. Each graph represents a different functional group. 

Figure 2. Boxplots of changes in foliar cover of species functional groups as affected by treatments averaged over needlegrass sites at KCPP.  Average change in cover is represented on the Y axis. Treatments are represented on the X axis. Each graph represents a different functional group. 

At YFN, boxplots were constructed for each functional group. Similar to the KCPP sites, most data appeared nearly normal with several outliers (Figure 3). Because virtually no shrub data appeared in YFN, shrubs was removed from analysis at that location. No other data gaps were identified.

Figure 3. Boxplots of changes in foliar cover of species functional groups as affected by treatments averaged over all sites at YFN.  Average change in cover is represented on the Y axis. Treatments are represented on the X axis. Each graph represents a different functional group. 

The boxplots showed some deviations from normality, especially in the invasive species. These groups showed a negative skew in sprayed plots, and multiple outliers are observed in all categories. The negative skew was most pronounced in sites that had been sprayed with herbicide, indicating that sites treated with herbicide declined more uniformly than sites that were not treated. In the pinegrass sites at KCPP, the invasive species Hypericum perforatum and Potentilla inclinata showed zero-inflated values. Upon inspection, this is because the cover of the species was low in both 2023 and 2025. These departures are not unexpected, given the fact that the treatments are expected to cause decreases in cover of certain species. Because of the large sample size and nature of ecological cover data, especially following spraying, deviations from normality are expected and not considered to be problematic. Outliers were checked and determined to be real observations due to differences in site or treatment. Given these results, two-factor mixed-effects ANOVAs on mixed models were used, as well as contrasts using estimated marginal means. 

Data exploration showed that, following treatment, several of the functional groups showed changes. At KCPP, the variation between treatments was greatest in invasive species and shrubs, while at YFN differences between treatments appear greatest for bunchgrasses and invasive forbs.