Remote sensing of biological soil crust under simulated climate change manipulations in the Mojave Desert

Abstract

Earth's arid and semiarid ecosystems are subject to novel combinations of disruptive factors and unprecedented rates of change. Biotic soil crust is believed to be sensitive to impacts caused by land use and climate changes. This study examined the potential for spectral detection of different biological soil crusts (BSC: cyanobacteria, moss and lichen) and bare soil components at a long-term manipulative experiment at the Mojave Global Change Facility (MGCF) in southwestern Nevada. We evaluated the potential for spectral detection of experimental treatments using laboratory and field measured reflectance spectra in the second and third year of the experiment, and airborne hyperspectral data obtained in the third year of the manipulations for soil disturbance, increased summer rainfall, and dry nitrogen deposition. Laboratory spectra of individual components of biological soil crust and bare soil measured under controlled laboratory conditions were spectrally different over much of the spectrum and exhibited features at 0.42, 0.50, and 0.68 μm, which could differentiate these materials. Field measured spectra were more similar in overall shape in each of the MGCF treatments and individual BSC could not be distinguished. The field spectra most closely resemble cyanobacteria from laboratory measurements, which are known to cover up to 60% of the inter-shrub spaces. There were significant treatment differences between control, soil disturbance, and irrigation treatments in field spectral measurements and a spectral feature in the 2.00–2.08 μm region could distinguish these treatments. The treatments were also apparent in high spatial resolution (~ 4 m ground IFOV) airborne hyperspectral imagery using a minimum noise fraction (MNF) analysis, although treatments were not distinct in terms of laboratory or field-based specific features. Disturbance treatments were easily apparent in color-infrared imagery although other treatments were not distinguished. Three-band composites from a MNF analysis and classification images of the six most significant MNF bands for treatment differences, revealed disturbed and irrigation treatments and combinations of these with nitrogen treatments can be observed but control treatments were not separated from the untreated background. Nitrogen treatments were generally not significantly different from controls unless combined with irrigation or disturbance treatments. These data suggest that hyperspectral imagery could be used to monitor local and perhaps regional changes in biological soil crust in the southwestern deserts of the United States, even if crust components are not individually detected.