Associate Professor University of Minnesota St. Paul, Minnesota, United States
Background/Question/Methods
There is an increased interest in measuring and monitoring plant biodiversity using remote sensing because of its potential to capture changes in diversity and plant community function. Several previous studies in forests and a few in grasslands have shown that spectral diversity can serve as a proxy for plant species and phylogenetic diversity. It remains unclear, though, whether spectral diversity can also detect effects of global change drivers on biodiversity, a step necessary for spectral diversity to be useful for detecting and attributing anthropogenic-driven changes in biodiversity. Towards this goal, we estimated grassland biodiversity using hyperspectral remote sensing in the context of environmental change. This study was conducted in the BioCON experiment at Cedar Creek Ecosystem Science Reserve in Minnesota, USA, which fully crosses treatments of biodiversity (1, 4, 9 or 16 grassland plant species), CO2 (ambient or elevated), added nitrogen (ambient or enriched). We tested the ability of spectral diversity to estimate plant diversity and detect treatment effects compared to ground-based measurements from 2021. The coefficient of variation (CV) of spectral reflectance across space was used as the indicator of spectral diversity, and α-diversity metrics included inverse Simpson’s diversity, inverse Simpson's evenness, and observed plant species richness.
Results/Conclusions
Spectral diversity was highly and positively related to Inverse Simpson’s diversity (R2 = 0.90), observed plant richness (R2= 0.88), and to Inverse Simpson's evenness (R2 = 0.89). The biodiversity treatment significantly increased spectral diversity (p = < 0.0001), similarly as to what was observed for Inverse Simpson’s diversity and observed plant richness (p = < 0.0001) from ground-based measurements. In contrast, Inverse evenness decreased with the biodiversity treatment (p = < 0.0001). Further, nitrogen addition decreased observed plant richness (p = 0.007), an effect which was not detected by spectral diversity (p = 0.80). We did not detect any other correlations among diversity measures or main effects or interactions among experimental treatments. Overall, these results are consistent with previous studies showing that spectral diversity can serve as a proxy for plant diversity, but additionally indicate that more work is needed to allow spectral diversity measures to detect and attribute anthropogenic-driven changes in biodiversity. These findings can be used to guide airborne studies of biodiversity and develop more effective large-scale biodiversity sampling methods.
