Presentation Details
| Towards AI-Driven Behavioral Phenotyping & Pesticide Risk Assessment in Bumblebee Colonies August C Easton-Calabria1, Shelby A Loebertman1, Colson S Tidikis1, Roberta E Hunt2, James D Crall1. 1University of Wisconsin-Madison, Madison, Wisconsin, USA.2University of Copenhagen, Copenhagen, Øresund Region, Denmark |
Abstract
Social insect colonies depend on complex, coordinated behaviors, but many remain difficult to quantify at scale. In bumblebees, little is known about how brood incubation varies across individuals, or how that variation shapes colony responses to sublethal stressors such as pesticide exposure. This variation may be especially important because colony responses to stress emerge not from average disruption, but from how disruption is distributed across workers. Here, we present work on an AI-driven behavioral phenotyping tool to enable scalable quantification of key bumblebee behaviors like incubation while remaining accessible to users without coding experience. By integrating tag tracking with AI approaches like instance segmentation, we can improve individual detection, tracking coverage, and behavior classification within dense nest environments. Using these tools, we ask how pesticide exposure alters incubation across individuals and how that may drive colony-level stress sensitivity. Broadly, this work highlights how AI can reveal important structure in the behavior of biological systems, with implications for pollinator health research and sublethal risk assessment regulation.
No part of this publication may be reproduced, distributed, or transmitted in any form or by any means, including photocopying, recording, or other electronic or mechanical methods, without the prior written permission of the author.
No part of this publication may be reproduced, distributed, or transmitted in any form or by any means, including photocopying, recording, or other electronic or mechanical methods, without the prior written permission of the author.
Content Locked. Log into a registered attendee account to access this presentation.