Spectral remote sensing provides tools for biological monitoring and can be used to characterize habitat quality in grasslands. These data have been used to study the relationships between plant community composition and remotely sensed canopy reflectance in grazed grasslands. The study area is located on the island of Öland, Sweden, and the data were collected in grazed grasslands that represent a succession from previously arable fields to old semi-natural pastures. All included grassland sites have been assigned to three different classes of grassland age, defined by the grazing continuity in years: young (5–14 years), intermediate-aged (15–49 years), and old (>50 years).

The plant community data consist of presences/absences for 100 vascular plant species in 104 (4 m × 4 m) sample plots positioned in open grassland vegetation. Information on species’ habitat preferences is included and has been used to explain the associations between plant species’ occurrences and the variation in community-level canopy reflectance.

The remote sensing data consist of 317 hyperspectral bands in the wavelength regions 414–1322 nm, 1496–1797 nm, and 2050–2351 nm, and show the mean reflectance in each spectral band for the 104 sample plots. The main gradient in the hyperspectral data is characterized by contrasting reflectance values between bands located in the NIR spectra and bands located in the red, blue, and SWIR spectra.

Grassland canopy reflectance was able to explain variation in the occurrences of individual plant species, particularly those with distinct habitat preferences. Species' habitat preferences indicated that vegetation reflectance in the red, blue, SWIR, and NIR spectra was linked to the plant-availability of mineral nitrogen. In contrast, species’ phosphorus preferences showed stronger associations with reflectance in the green and red-edge spectra.

The study area (centred on 56°40'49.0"N 16°33'58.0"E) measures approximately 5 km × 5 km and represents an agricultural mosaic landscape near the village of Algutsrum on the Baltic island of Öland, Sweden. The overall topography is flat, and the soils are neutral to basic (mean soil pH=7.0). The studied grassland sites were grazed by cattle and represent a succession from previously arable land to old semi-natural grassland. The ages of the grasslands have been determined with the help of historical map data (see Johansson et al., 2008), and the included sites are assigned to three categories of grassland age (defined by the number of years since the first historical map where the site was classified as grassland): young (5–14 years), intermediate-aged (15–49 years), and old (>50 years).

Data are presented for 104 (4 m × 4 m) sampling plots that were positioned in open and relatively homogeneous grassland vegetation. Each plot belongs to a pair of plots in a total of 52 grassland sites (17 young, 18 intermediate-aged, and 17 old). Grassland site selection and the positioning of sampling plots were carried out to minimize edge effects and shadows – detailed descriptions of the procedure are given in Möckel et al. (2014).

Data on the grassland canopy reflectance were acquired on 9 July 2011 by two airborne hyperspectral sensors (Norsk Elektro Optikk AS (NEO), Lörenskog, Norway). The flight height was 1500 m and the spatial resolution in the raw data was 0.5–1.0 m. The spectral data consist of 317 narrowbands: 160 bands from the first sensor (3.6 nm bandwidth; 414–991 nm), and 156 bands from the second sensor (6.0 nm bandwidth; 991–1322 nm, 1496–1797 nm, and 2050–2351 nm). The spectral data were pre-processed following the descriptions given in the section “2.4.1 Remote Sensing Data” in Möckel et al. (2014), and included corrections of atmospheric disturbance and topographic illumination effects, random noise reduction, and conversion of digital values to spectral reflectance. For each sampling plot, mean reflectance for each spectral band was extracted using a 4 m × 4 m pixel window.

Vegetation data were collected between 15 May and 15 July 2011. The data represent the presence/absence of 100 vascular plant species in the 104 sampling plots. Information on species’ habitat preferences was extracted from a previous study (Löfgren et al., 2020), and was used to explain species’ relationships with grassland canopy reflectance. The earlier study analysed species’ relationships with variation in soil concentrations of phosphorus, ammonium, and nitrate, and also used latent variable modelling to capture spatially dependent patterns in species occurrences at four fine-to-coarse spatial scales (S1, S2, S3, and S4). The species’ habitat preference values (Phosphorus, Ammonium, Nitrate, S1, S2, S3, and S4) consist of the species’ mean model-response parameters from the analysis in Löfgren et al. (2020).

References

• Johansson, L.J., Hall, K., Prentice, H.C., Ihse, M., Reitalu, T., Sykes, M.T., & Kindström, M. (2008). Semi-natural grassland continuity, long-term land-use change and plant species richness in an agricultural landscape on Öland, Sweden. Landscape and Urban Planning, 84(3-4), 200–211. https://doi.org/10.1016/j.landurbplan.2007.08.001
• Löfgren, O., Hall, K., Schmid, B.C., & Prentice, H.C. (2020). Grasslands ancient and modern: soil nutrients, habitat age and their relation to Ellenberg N. Journal of Vegetation Science, 31(3), 367-379. https://doi.org/10.1111/jvs.12856
• Möckel (Astor), T., Dalmayne, J., Prentice, H.C., Eklundh, L., Purschke, O., Schmidtlein, S., & Hall, K. (2014). Classification of grassland successional stages using airborne hyperspectral imagery. Remote Sensing, 6(8), 7732-7761. https://doi.org/10.3390/rs6087732