Changed management of semi-natural grasslands is thought to be one important factor for explaining the decline of butterfly populations in Sweden and the rest of Europe. This study explores how vegetation height, variation in vegetation height and flower abundance can help predict the occurrence of butterfly species in semi-natural grasslands in southern Sweden. My study is based on data collected by a national environmental monitoring programme (NILS) during 2006-2010. Generalized linear models showed that all three explanatory variables significantly affected 26 of the 41 species of butterflies studied. The study also investigated preferences for three different systems for grouping species but did not find differences among groups. Several of the species would most likely benefit from variation in grazing pressure or rotational grazing as well as flower-rich grasslands. The results may also help when aiming management at the preservation of individual species.

Semi-natural landscapes are recognized as suitable habitats for different plant species and provide ecosystem services that contribute to increased plant biodiversity. At the stand level, plant biodiversity is influenced by vegetation structure, of which vegetation height is an important parameter. Photogrammetry from drone captured images has the potential to provide a quick and cost-effective analysis of vegetation height. In addition, the relation between spectral signatures and species distribution can indicate where higher plant biodiversity can be found, as species can be identified based on their spectral signature. Spectral signatures are thus used in the current study in conjunction with vegetation height as a proxy for plant biodiversity in herbaceous-dominated patches. Two field surveys were conducted to collect drone data and reflectance data in July and August 2019. Twelve plots of ten metres diametre were delimited in the drone-derived orthophotos around the reflectance readings coordinates. In order to assess vegetation height, the difference between the digital surface model derived from the orthophotos and the national digital elevation model was determined. Two statistical indices were calculated: the modified soil-adjusted vegetation index (MSAVI) and the coefficient of variation of heights (CV). The relationship between the two indices was evaluated as a proxy for plant biodiversity. Drone-derived point clouds can be used to measure vegetation height in herbaceous-dominated environments due to the very fine scale of drone imagery. A possible negative correlation was found between MSAVI and CV on both surveyed months (July r2 = 0.675; August r2 = 0.401) if the outlier plots were removed from the analysis. There is not enough evidence to clearly explain the anomalous behaviour of the outlier plots. Further research is needed to confirm the use of the relationship between vegetation height variability and reflectance as a proxy for plant biodiversity assessment in herbaceous-dominated environments.