Towards automated vision-based monitoring of insects in high temporal resolution – a pilot project in the Swiss Alps.
Abstract ID: 3.10178 | Accepted as Talk | Talk/Oral | TBA | TBA
Nils Roling (0)
Gossner, Martin (2), Volpi, Michele (3), Kempel, Anne (0,1)
Nils Roling ((0) WSL Institute for Snow and Avalanche Research SLF, Flüelastrasse, 7260, Davos, Graubünden, CH)
Gossner, Martin (2), Volpi, Michele (3), Kempel, Anne (0,1)
(0) WSL Institute for Snow and Avalanche Research SLF, Flüelastrasse, 7260, Davos, Graubünden, CH
(1) Climate Change, Extremes and Natural Hazards in Alpine Regions Research Centre CERC, Flüelastrasse, 7260, Davos, Graubünden, CH
(2) Swiss Federal Institute for Forest, Snow and Landscape Research WSL, Zürcherstrasse, 8903, Birmensdorf, Zürich, CH
(3) Swiss Data Science Center, ETH Zurich and EPFL, Andreasstrasse, 8050, Zürich, Zürich, CH
(2) Swiss Federal Institute for Forest, Snow and Landscape Research WSL, Zürcherstrasse, 8903, Birmensdorf, Zürich, CH
(3) Swiss Data Science Center, ETH Zurich and EPFL, Andreasstrasse, 8050, Zürich, Zürich, CH
Insects play a crucial role in mountain ecosystems, contributing to key ecological functions such as plant pollination and serving as an essential food source for various animals. Despite their importance, global declines in insect diversity, abundance, and biomass are well-documented. However, mountain ecosystems remain underrepresented in insect monitoring efforts due to the logistical challenges posed by steep terrain and limited accessibility, though they are ideal for studying how insect communities respond to temperature variations because of their steep climatic gradients. A novel approach to monitoring insect communities involves automatic insect cameras. Compared to conventional monitoring methods, insect cameras provide a higher temporal resolution for assessing abundance, do not require manual sorting or classification, and avoid the need to kill insects. Here, we present the first results of a pilot study in which we installed nine insect cameras (Diopsis system) along three elevational gradients in the Swiss Alps, capturing over 47,000 images of individual insects. We developed a computational pipeline to count insects, measure their body size, and classify them into taxonomic groups using both existing and custom-trained algorithms. We present preliminary findings on (1) insect abundance, biomass, and taxonomic composition along elevational gradients, (2) insect temporal dynamics, and (3) the dependencies of observed patterns on local weather conditions. Finally, we discuss the ecological implications of our results, as well as the limitations and potential of using automated insect monitoring systems in mountain ecosystems.
N/A | ||||||||
|