Very high resolution Earth Observation features for monitoring plant and animal community structure across multiple spatial scales in protected areas

Monitoring the status and future trends in biodiversity can be prohibitively expensive using ground-based surveys. Consequently, significant effort is being invested in the use of satellite remote sensingto represent aspects of the proximate mechanisms (e.g., resource availability) that can be related tobiodiversity surrogates (BS) such as species community descriptors. We explored the potential of veryhigh resolution (VHR) satellite Earth observation (EO) features as proxies for habitat structural attributesthat influence spatial variation in habitat quality and biodiversity change. In a semi-natural grasslandmosaic of conservation concern in southern Italy, we employed a hierarchical nested sampling strategyto collect field and VHR-EO data across three spatial extent levels (landscape, patch and plot). Speciesincidence and abundance data were collected at the plot level for plant, insect and bird functional groups.Spectral and textural VHR-EO image features were derived from a Worldview-2 image. Three windowsizes (grains) were tested for analysis and computation of textural features, guided by the perceptionlimits of different organisms. The modelled relationships between VHR-EO features and BS responsesdiffered across scales, suggesting that landscape, patch and plot levels are respectively most appropriatewhen dealing with birds, plants and insects. This research demonstrates the potential of VHR-EO forbiodiversity mapping and habitat modelling, and highlights the importance of identifying the appropriatescale of analysis for specific taxonomic groups of interest. Further, textural features are important in themodelling of functional group-specific indices which represent BS in high conservation value habitattypes, and provide a more direct link to species interaction networks and ecosystem functioning, thanprovided by traditional taxonomic diversity indices.