Spatial resolution, spectral metrics and biomass are key aspects in estimating plant species richness from spectral diversity in species‐rich grasslands

Abstract

Abstract Increasing evidence suggests that remotely sensed spectral diversity is linked to plant species richness. However, a conflicting spectral diversity–biodiversity relationship in grasslands has been found in previous studies. In particular, it remains unclear how well the spectral diversity–biodiversity relationship holds in naturally assembled species‐rich grasslands. To address the linkage between spectral diversity and plant species richness in a species‐rich alpine grassland ecosystem, we investigated (i) the trade‐off between spectral and spatial resolution in remote sensing data; (ii) the suitability of three different spectral metrics to describe spectral diversity (coefficient of variation, convex hull volume and spectral species richness) and (iii) the importance of confounding effects of live plant biomass, dead plant biomass and plant life forms on the spectral diversity–biodiversity relationship. We addressed these questions using remote sensing data collected with consumer‐grade cameras with four spectral bands and 10 cm spatial resolution on an unmanned aerial vehicle (UAV), airborne imaging spectrometer data (AVIRIS‐NG) with 372 bands and 2.5 m spatial resolution, and a fused data product of both datasets. Our findings suggest that a fused dataset can cope with the requirement of both high spatial‐ and spectral resolution to remotely measure biodiversity. However, in contrast to several previous studies, we found a negative correlation between plant species richness and spectral metrics based on the spectral information content (i.e. spectral complexity). The spectral diversity calculated based on the spectral complexity was sensitive to live and dead plant biomass. Overall, our results suggest that remote sensing of plant species diversity requires a high spatial resolution, the use of classification‐based spectral metrics, such as spectral species richness, and awareness of confounding factors (e.g. plant biomass), which may be ecosystem specific.

Keywords

• Biodiversity
• coefficient of variation
• data fusion
• imaging spectroscopy
• spectral diversity
• spectral species richness