Remote sensing of grassland communities : integrated effects of soil nutrients and habitat age

Abstract: Semi-natural grasslands are characterized by high fine-scale plant species richness. The richest grassland communities, with many habitat-specialist species, are found in ancient grassland sites that have a long continuity of grazing management and low levels of soil nutrients. Grazed grasslands were widespread in the historical landscape. Agricultural intensification over the last two centuries has led to a reduction of grassland area: ancient, species rich grasslands now occur only as small and isolated fragments in the landscape. Young grasslands may also develop, under grazing management, on previously arable fields. These younger sites have high soil nutrient levels, making them unsuitable for grassland specialist species. However, leaching and biomass-removal by grazing management mean that, over time, there is a progressive reduction of nutrient levels that is tracked by a succession of plant communities – as the habitat conditions become more favourable for nutrient intolerant species. The aim of this thesis was to investigate the ability of spectral remote sensing to capture variation in plant community composition in dry, grazed grasslands. The study system consists of differently-aged grassland sites, within a succession from former arable land to ancient semi-natural grasslands, on the Baltic island of Öland, Sweden. Analyses of the relationships between field-collected data on plant community composition, and data on spectral reflectance were based on regression methods such as PLSR, and individual species’ responses to (field-measured) soil nutrient concentrations and reflectance were analysed using Bayesian joint species distribution modelling. Spectral data were acquired using airborne hyperspectral sensors and the multispectral satellite WorldView-2. The spectral reflectance of heterogeneous grassland canopies represents whole plant communities, consisting of multiple individuals and species. Grassland canopy reflectance can be indirectly related to the community composition, assuming that the spectral reflectance can detect variation in the environmental conditions that drive the species assembly. Species that are adapted to different habitats are characterized by differences in the structural and functional properties that determine their physical appearance and spectral characteristics. Responses of individual species to particular environmental conditions result in plant communities with similar preferences, and similar spectral characteristics. The strongest gradient of canopy reflectance in the studied grasslands represented variation in photosynthetic absorption in the blue and red wavelengths, and reflectance in NIR and SWIR. This gradient was associated with species’ preferences for ammonium availability and soil quality, and was largely characterized by the NDVI. However, the fact that the environmental preferences of individual species are not fully described by single gradients means that single gradients of reflectance will have a limited capability to capture patterns in grassland plant communities. Analysis of soil data showed that species’ distributions were explained primarily by their different preferences for soil phosphorus concentrations. Species responses to soil phosphorus were not associated with the main spectral gradient (or the NDVI) but were, instead, associated with reflectance in the green and red-edge regions. Low reflectance in the green regions, and high reflectance in SWIR, are likely to be useful spectral characteristics for identifying old, species rich and phosphorus-poor grasslands. Although the grassland canopy reflectance can explain variation in important environmental gradients, grassland community composition is not only dependent on the habitat conditions. Plant communities of old grasslands may include species that rarely occur in younger grasslands as a result of ecological processes, such as dispersal, that may not contribute to the spectral characteristics of the vegetation canopy.