Abstract: Intro/Methods Grasslands provide valuable ecosystem services, have spatial-temporal widespread distribution, and are increasingly vulnerable to climate extremes. Precipitation variability, including drought events, as well as changes in precipitation can contribute towards shifts in seasonal patterns of soil water availability. Ecological responses, such as shifts in plant composition and phenology of grasslands can be affected by drought and overall changes in the seasonal timing of precipitation between spring and summer due to an increase in summer drought events. Multi-scale monitoring using remote sensing can enables us to document the spatial-temporal variation in species distribution allowing us to scale up and predict responses to global changes like shifts in plant communities or changes in phenological responses to extreme weather events. We examined the response of plant species to experimentally manipulated precipitation in a mixed-grass prairie at both leaf and canopy levels. We used a randomized block split-plot design with seven precipitation treatments (five precipitation exclusion levels [-20 %, -40 %, -60 %, -80%, and -100%], precipitation addition [+50%], and control [0%, no change in precipitation]) replicated three times for a total of 21, 2 x 2 m plots. The rain-out shelters for this study were established in spring 2016, these shelters reduced precipitation but not sunlight. We collected leaf traits, canopy level spectral reflectance and calculated multiple vegetation indices. This study aimed to answer the following questions: (1) Can canopy spectral measurements capture differences in vegetation responses at the leaf level in a precipitation gradient? (2) If functional groups (e.g. C3, C4 and N-fixers) respond differently to drought, are there noticeable differences in their spectral properties and functional traits?
Results/Discussion We observed drought-induced shifts in plant composition with non-drought plots having a higher C4:C3 proportion than those experiencing drought. Using remote sensing approaches, we have successfully detected canopy level hyperspectral differences across a manipulated precipitation gradient indicating canopy/vegetation changes in pigmentation, potential plant stress and water content. We found that non-drought plots have higher Normalized Difference Vegetation Index (NDVI) at canopy levels and higher Specific Leaf Area (SLA) at leaf levels than those experiencing drought. Our data show a significant relationship between NDVI and the month it was taken, meaning that NDVI differs greatly throughout the growing season. These results confirm that climatic changes such as drought are causing a shift in plant community composition and such stressors can be observed at both leaf and canopy levels.
