| dc.description.abstract | Invasive species are a significant economic and ecological threat, which is growing with climate change, land use change, and increasing global trade and travel. Hyperspectral remote sensing can be used to estimate the increasing ecosystem impacts of invasions, enabling early detection and prompt management. Plants traits, measurable characteristics connected to plant growth, survival and reproduction, have been linked to both invasiveness and ecosystem service provision. Optical traits, those measurable with spectroscopy, were used to assess the effects of invasion on native ecosystem functional trait values. We investigate invasion-induced trait replacement on the landscape scale in the Sonoran Desert, Arizona, US, by assessing if optical traits can be differentiated between invasive tamarisk (Tamarix spp.) and native palo verde (Parkinsonia spp.).
Field data and predictive algorithms of species presence were combined with handheld and airborne spectroscopy to measure trait values of each species and test whether they varied significantly by species. Tamarix spp. was consistently differentiable from Parkinsonia spp. at reflectances below 1350 nm, and in the short-wave infrared (>1940 nm). Given this spectral separability, vegetation indices were then used to measure each trait. Tamarix spp. showed significantly higher vegetation index values for all measured traits, namely chlorophyll, leaf area, water content, lignin-cellulose, and salinity tolerance, although less strongly for water content.
Indicators were constructed using principal component analysis to combine and weight the most robust indices for each trait. Tamarix spp. was significantly associated with every indicator of invasiveness and the summed invasiveness score for the ground-truthed data and the combined species presence map using both algorithms. However, the Mixture Tuned Matched Filtering classification showed Parkinsonia spp. to have significantly higher invasive trait scores.
Finally, indicators were mapped to illustrate trait hotspots, using total invasiveness score maps, and using indicators as RGB bands. Highest invasive trait values tended to correlate with larger patches of vegetation. The difference between the species, however, was not as strong as expected – vegetation sparseness and drought conditions may have reduced reflectance and trait values. Therefore we can only make general, directional assessments of the impact of invasion on optical traits and in turn ecosystem effects. However, the study demonstrates a transferable method with the potential to track non-species specific invasion effects in native populations, using traits that can be linked to ecosystem processes and services, providing another tool in the management of invasives. | |
