Details

Autor(en) / Beteiligte

• Feilhauer, Hannes
• Doktor, Daniel
• Schmidtlein, Sebastian
• Skidmore, Andrew K.
• Prinzing, Andreas

Titel

Mapping pollination types with remote sensing

Ist Teil von

• Journal of vegetation science, 2016-09, Vol.27 (5), p.999-1011

Ort / Verlag

Blackwell Publishing Ltd

Erscheinungsjahr

2016

Link zum Volltext

Quelle

Wiley Online Library

Beschreibungen/Notizen

• Questions: Pollination is an ecosystem function that varies at local spatial scales. Remote sensing may help to quantify and map the resulting patterns for a better understanding of ecosystem functioning. This task is challenging because the signal measured by sensors is dominated by leaf and canopy optical traits that determine the vegetation spectrum. No direct influence of pollination types (i.e. wind, insect and self-pollination) can be expected. We thus ask: (1) how strongly are pollination types at the community level linked with optical traits; and (2) do these links allow us to map spatial patterns of pollination types with remote sensing data? Location: Mires and Molinia grasslands in Bavaria, southern Germany. Methods: We sampled vascular plant species composition, as well as traits related to optical spectral signals, for 100 plots. Information on pollination types and additional optical traits were compiled from established plant trait databases. Simultaneously with the field sampling, image data of the study site were acquired using the airborne sensor AISA Dual. We tested for correlations between pollination types and optical traits. Based on these results, we regressed the plot-based information on pollination types against the corresponding spectral signal extracted from the image data. We inverted the models using the image in order to obtain maps of the pollination type distribution across the study site. Results: In our study site, pollination types and optical traits were significantly correlated, up to R = -0.813 in the case of wind pollination and leaf dry matter content. These relations are, however, not necessarily transferable to other ecosystems, phenological stages and spatial scales, and their physical interpretation requires careful consideration. We were able to statistically model the spatial distribution of pollination types with a RMSE < 10.5%. The resulting maps provide detailed insights into the spatial distribution of pollination types at the community level. Conclusions: The results show that pollination types are indeed related to canopy reflectance in a way that allows their mapping using remote sensing. More research is needed in order to improve our knowledge about transferable relations between pollination types and plant traits in other ecosystems and at different phenological stages.