Coupling Remote Sensing Data with Ground Surveys of Grassland Plant Communities Following Experimental Nutrient Addition and Disturbance

Abstract

Anthropogenic nutrient deposition and large-scale disturbances are two prominent and pervasive drivers of global change in grassland ecosystems. Both of these drivers cause changes in grassland community dynamics and plant diversity that threaten to alter grassland ecosystem functions and processes. The emergence of unmanned aerial vehicles (UAVs; i.e., drones) and advancements in remote sensing (RS) technology to assess changes in grassland community composition is of increasing interest to ecologists and restoration practitioners alike. Here, I coupled plant community data collected from traditional ground surveys with UAV RS imagery to quantify plant community changes in a tallgrass prairie following experimental nutrient addition and physical soil disturbance. The field site consisted of 25, 5 m x 5 m plots, arranged in a 5 x 5 Latin Square design. Treatments included: (1) Control (C), (2) nitrogen (N), phosphorus (P), potassium (P), and micronutrients addition (µ) (NPKµ), (3) soil disturbance (D), (4) NPKµ + D, and (5) NPKµ cessation (not used in this analysis). Soil disturbance was applied in January 2022 by rototilling the soil surface to a depth of 15 cm. Nutrient addition was administered in April 2022 as 10 g m-2 of granular N, P, K, and micronutrients. Ground based species percent cover surveys were taken monthly between March – September 2022. Corresponding UAV RS surveys were taken monthly within seven days of each ground survey. Nonparametric multidimensional scaling of ground cover data indicated community composition was changed by nutrient addition and disturbance. Nutrient addition resulted in increased cover in C4 grasses - primarily invasive Sorghum halepense. Nutrient addition combined with disturbance resulted in the emergence of novel plant communities. C3 and C4 grasses and legume cover were significantly reduced, replaced by the cover of forbs, specifically the annual forb Ambrosia trifida. Grass and forb cover estimates from UAV RS data mirrored ground survey results, but the accuracy of results varied across sample dates. UAV RS was most accurate at estimating grass and forb cover during summer months coinciding with peak plant activity and expression of phenological traits including leaf area, leaf color, and leaf shape. UAV RS was least accurate at estimating grass and forb cover during the early growing season when plant activity and phenological trait expression was limited. Results of the study indicate that grassland plant communities may lack short-term resistance to nutrient addition and soil disturbances. Changes in community composition quantified and mapped using UAV RS surveys accurately identified the major functional group changes seen in ground surveys.