Inland, freshwater wetlands of the Mid-Atlantic provide ecosystem islands within a matrix upland habitat which experiences human disturbance of varying intensities; they provide constrained models for ecosystem interactions. Disturbance in surrounding land use is generally executed through stressors such as hydrologic modification, accelerated sedimentation, and eutrophication, leading to biotic consequences such as shifts in vegetative and invertebrate community composition. While the impacts of anthropogenic disturbance on wetland ecological function have been described at specific scales, we have yet to provide a model of ecological response to human disturbance that is scale-invariant. We propose that variability and the scale of variance provides such a model, and examine its expression at spatial scales of sub-1 m2 to over 100,000 km2. At the largest spatial scale, we utilize a series of 235 wetland sites across the commonwealth of Pennsylvania to illustrate, via ordination and descriptive statistics, the variability of plant community composition and soil characteristics across and within wetland type, and the change in this variability with anthropogenic disturbance. At the site and sub-1m2 scale, we describe the spatial structure of the plant community and soil parameters (soil organic horizon depth, soil moisture, pH, nitrate/ammonium, and soil organic matter).
In a relatively unimpacted condition, wetland plant communities exhibit a high diversity both between and within individual wetland types (e.g., floodplains, slope wetlands, depressional wetlands). With disturbance, plant communities exhibit increasing homogenization across all wetland types to a small number of community types, dominated by a relatively small number of wetland plant generalist and invasive species. At the site and sub-1m2 scale, soil parameters and plant community similarities show a similar homogenization. Geostatisics such as semiovariograms with kriging are used to describe each plant/soil microhabitat parameter within 40m2 sites and 1m2 plots. Variables such as range, sill, and nugget are extracted from semivariogram models and along with descriptive statistics are utilized to describe changes across the human disturbance gradient. For example, disturbed sites lack a significant soil organic horizon, while unimpacted sites have high variblity in the depth of their organic horizons. Coefficients of variation for soil moisture and pH are generally higher in these relatively pristine sites across scales. An “homogenization”in plant community types and microbial habitat parameters at disturbed sites compared to sites within intact landcapes, across a range of spatial scales, could significantly impact the key ecosystem processes, such as habitat diversity and nutrient cycling, that these systems provide.