Flowers for bees: using a drone to assess floral resource cover in agricultural landscapes

The current loss of semi-natural habitats providing vital resources for organisms threatens pollinator diversity and pollination services. Studies of landscape effects on biodiversity and ecosystem services mostly characterize landscapes according to broad habitat categories such as “non-crop” or “semi-natural habitat”. However, the beneficial effect of these habitats may depend on the spatio- temporal availability of flowering plant resources for pollinators. Quantifying food resources rather than broader habitat types may better predict pollination services in agricultural landscapes.

Our study aims at comparing the predictive power of floral-resource based (hereafter “functional”) versus habitat-type based landscape maps on the pollination of field beans. By means of a moving windows analysis and initial screening of aerial photographs, we selected 24 landscape sectors of 500 m radius in north-eastern Switzerland that present a gradient of flower resource availability. For each of the landscapes, we produced three types of maps:

1) Functional map at a temporal resolution of 3-4 weeks between April and June, mapping flower resources through detailed field surveys;
2) Functional map at a temporal resolution of 3-4 weeks between April and June, mapping flower resources by means of a custom made drone (Copting Hexacopter, Germany) equipped with a Sony camera (α7R II).
3) Classical habitat map, detailing crop types and semi-natural habitats (e.g. woodland, hedgerow, extensive grassland) based on a single survey in May/June.

We compare the three mapping approaches regarding effectiveness and efficiency. Effectiveness is evaluated by the predictive power of the three types of maps to explain flower visitation by pollinators and seed set of field beans. Efficiency is evaluated by comparing the predictive power of the maps to the cost and effort required to generate them. Our results present insights on how functional habitat mapping can be improved through the targeted application of remote sensing and drone technology.