Multi-temporal assessment of grassland plant diversity using imaging spectroscopy

Grasslands, a.k.a prairie, form the most extensive biome in North America. Most of these prairies have been altered, mainly by conversion to agricultural and urban lands. The destruction of grasslands is significantly impacting the vital ecosystem services that grasslands provide, leading to ongoing biodiversity loss with attendant loss of ecosystem function. Especially in an era when climate change is impinging on land use change/habitat destruction, developing an operational and cost-effective global biodiversity monitoring system for grasslands and other ecosystems is fundamental. Remote sensing is emerging as one of the main sources for monitoring biodiversity within and across different ecosystems. An important aspect of remote sensing of plant diversity, which has been overlooked in many studies, is the temporal dynamics of plant communities. Although there are studies investigating multi-temporal aspect of ecosystem function especially productivity, to our knowledge, not many studies have investigated the capability of remote sensing to assess biodiversity over time. In this work, we asked how well remote sensing can detect biodiversity (α- and β-diversity) in a restored prairie across time using airborne hyperspectral data collected in two successive years (2017 and 2018) at different periods in the growing season. The results showed that our ability to map biodiversity through spectral diversity and spectral species indeed varies significantly over time. Our findings also showed that, unlike 2017, the relationship between field-based and remote sensing-based α- and β-diversity in 2018 weakened towards the end of the growing season. This pattern was attributed to shifts in species composition, and the convergence in the measured levels of diversity caused by contrasting management regimes (prescribed fire) in the two years. Additional tests using cubesats at this site demonstrated that current spaceborne multispectral sensors are not optimized for mapping biodiversity in grasslands. These findings indicate that remote sensing of biodiversity in grasslands should be multi-temporal, employing data with high spectral, spatial, and temporal resolution, and demonstrate an essential role for airborne platforms in developing operational approaches to biodiversity monitoring.