Drones for butterfly conservation: larval habitat assessment with an unmanned aerial vehicle

Context Evidence-based nature conservation focuses on ecological facts and the incorporation of knowledge on the ecology of species, including its entire life cycle. In butterflies, imagos and its larvae often demand specific and diverging micro-habitat structures and resources. In consequence, ecological requirements of the imaginal and pre-imaginal stage have to be taken into consideration to conduct effective conservation management. Objective Here we analyse ecological pre-requisites of imagos and larvae for two lycaenid butterfly species, the common blue Polyommatus icarus and the adonis blue Polyommatus bellargus . Both butterfly species occur in calcareous grasslands and mainly depend on two plant species at our study site, the horseshoe vetch Hippocrepis comosa and bird’s-foot trefoil Lotus corniculatus . These plant species serve as nectar sources and larval host plants for the two butterfly species. Methods First, we assessed the occurrence of imagines and larvae of the two butterfly species and recorded various micro-habitat characteristics, like the number of flower buds of the two main host plants, the surrounding vegetation height, percentage of bare soil, availability of shadow, and the distance to and geographic direction of thickets at respective sites. In a second step we took high resolution aerial pictures from our study area using an unmanned aerial vehicle (drone). Based on these aerial pictures and the information on the larvae´s habitat preference from our field observations, we trained a habitat suitability model to identify micro-habitat structures suitable for larvae of the two butterfly species. Results We found that abundance of imagos is positively correlated with flower bud density of the two host plants. Low vegetation height and high proportion of bare soil (but not flower bud density) positively influence egg oviposition. The calculated habitat suitability models predict the occurrence of high quality larval habitats with high prediction power (AUC = 0.72). Conclusions This combined data set consisting of field observations, high resolution aerial pictures taken from an unmanned aerial vehicle, and models underline that (1) species with complex life cycles may request more than one habitat niche, depending its stage of development, and (2) high resolution aerial pictures taken from drones provide valuable background data to generate habitat suitability models—even on a micro scale but covering larger parts of a landscape.