1/1
7 files

Novel approaches to sampling pollinators in whole landscapes: a lesson for landscape-wide biodiversity monitoring

dataset
posted on 28.11.2018, 18:29 by Christoph Scherber, Tatiane Beduschi, Teja Tscharntke

Context

Biodiversity monitoring programs require fast, reliable and cost-effective methods for biodiversity assessment in landscapes. Sampling pollinators across entire landscapes is challenging, as trapping needs to cover many habitat types.

Objectives

We developed and tested a landscape-wide sampling design for pollinators. We assessed the predictability and stability of pollinator biodiversity estimates in agricultural landscapes, and tested how estimates were affected by sampled habitat, landscape composition and spatial scale.

Methods

We sampled pollinators using pan traps at 250 locations in 10 replicated landscapes measuring 1 × 1 km and calculated bee richness predictions based on different sample sizes. Traps were placed regularly in each landscape, sampling each habitat proportionally to its area. Landscapes contained semi-natural habitats, crop fields and forests and differed in the amount of a mass-flowering crop (oilseed rape).

Results

Regular sampling reflected local habitat amount. Compared with cereal fields, significantly more pollinators occurred in oilseed rape, and fewer in forests. Sampling in only one habitat type led to biased estimates of landscape-wide bee species richness, even when sample size was increased. The spatial scale of best predictions depended on the sampled habitat. Species richness was overestimated when sampling was limited to semi-natural habitats and underestimated in oilseed rape fields. Precision increased with the number of sampling points per landscape.

Conclusions

To study landscape-wide pollinator biodiversity, we suggest to sample multiple sites per landscape in a broad range of resource-providing habitat types, with sample sizes proportional to habitat amount. Our approach will also be useful for biodiversity monitoring programs in general.

Funding

We are grateful to the Landesamt für Geoinformation und Landentwicklung Niedersachsen for providing information on land-use and to the farmers for allowing us to perform this study on their fields. Funding was provided by the Deutsche Forschungsgemeinschaft (DFG) within the frame of the Research Training Group 1644 “Scaling Problems in Statistics”. RapidEye satellite imagery was kindly provided by the RapidEye Science Archive (RESA), grant number RESA 464, funded by the German BMBF (Federal Ministry of Education and Research).

History

Licence

Exports