Infra-red sampling to assess Apis mellifera colony strength and improve crop pollination

Monday, November 14, 2022

9:00 AM – 9:12 AM PT

Location: Vancouver Convention Centre, Meeting Room 119/120

Presenting Author(s)

Lauren M. Goldstein

Master's Student
Michigan State University
East Lansing, Michigan

Co-Author(s)

DW

Dan Wyns

Michigan State University
East Lansing, Michigan
JR

Joselyn Ralph

Michigan State University
East Lansing, Michigan
RL

Rob Long

Director of Operations
The Bee Corp
Indianapolis, Indiana
KG

Kate Graney

Operations Manager
The Bee Corp
Indianapolis, Indiana
MM

Meghan Milbrath

Assistant Professor
Michigan State University
East Lansing, Michigan
JP

Jacquelyn Perkins

Michigan State University
East Lansing, Michigan
Rufus Isaacs

Professor and Extension Specialist
Michigan State University
East Lansing, Michigan

Measuring Apis mellifera colony strength is vital to running a successful commercial beekeeping operation, because this can provide beekeepers with a prediction of colony health and foragers available for pollination. Colony strength is typically evaluated through a manual in-hive inspection where size is estimated via counting the number of frames covered with bees, but this can be cumbersome, time-consuming, disruptive to the colony, and biased. Using a non-invasive and rapid method for assessment could improve revenues by allowing for identification of sub-standard colonies for remedial management and full-strength hives for pollination contracts. We tested Verifli technology to quantify colony size using a hand-held camera to capture thermal images of colonies, and these were processed to determine a frame count score. In 2021, we conducted a pilot study using 20 hives following a distribution of sizes commonly found during the blueberry pollination season, and compared the digital assessment to manual in-hive frame counts. In 2022, we tested the technology on colonies in commercial blueberry farms. Correlation analyses of the pilot study showed that Verifli assessment of individual colony size was positively correlated with manual estimation, although predictive power was relatively low. Further refinement of the algorithm to process images is expected to improve the predictive power of the experiments conducted in 2022, and will be reported in this presentation.