Location: Carl Hayden Bee Research CenterTitle: Breakfast Canyon discovered in honey bee hive weight curves
|HOLST, NIELS - Aarhus University|
Submitted to: Insects
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 11/24/2018
Publication Date: 12/1/2018
Citation: Holst, N., Meikle, W.G. 2018. Breakfast Canyon discovered in honey bee hive weight curves. Insects. 9(4):176. https://doi.org/10.3390/insects9040176.
Interpretive Summary: The use of sensors in monitoring bee hives is increasing, and hive scales are being used to monitor not just hive weight gain from day to day, but also colony growth and activity. Those data are rich in information. What is needed are novel methods to extract that information, which increases the value of the data and increases the scientific yield from field experiments. Here we analyzed published raw within-day weight changes every 15 minutes for many hives over two different experiments in which bee colonies were given sublethal doses of a pesticide, imidacloprid. Hive weights change because of forager departure and return, water gain and loss, and nectar collection (and consumption) and other reasons. We fit a “broken line” regression to the data, which was essentially a line with several segments. We found that a line with 5 segments and thus 4 break points had the best fit. By examining the break points and segments, we linked the lengths and slopes of the first two segments to nightly hive weight change and estimated the relative size of the foraging population. These are useful response variables in field experiments and indeed revealed important treatment differences, which is critical information for the industry.
Technical Abstract: The supply of electronic devices to sense, store and transmit data is rapidly increasing, providing a comprehensive toolbox for inventive minds. In apiculture, both researchers and practitioners have welcomed the opportunity to equip bee hives with a variety of sensors to monitor hive weight, temperature, forager traffic and more, resulting in huge amounts of accumulated data. The problem remains how to distill biological meaning out of these data. In this paper we address the analysis of bee hive weight monitored at a 15-minute resolution over several months. Inspired by an overlooked, classic study on such weight curves we derive algorithms and statistical procedures to allow biological interpretation of the data. Our primary finding was that an early morning dip in the weight curve ('Breakfast Canyon') could be extracted from the data and also provide information on bee colony performance, in terms of foraging effort. We include the data set used in the study, together with R scripts that will allow other researchers to replicate our method.