Identification of the honey bee swarming process by analysing the time course of hive vibrations

Abstract

14 The amplitudes of the independent signals form a multi-dimensional time-varying 15 vector which was logged continuously for eight months. We found that combined 16 with specifically tailored weighting factors, this vector provides a signature highly 17 specific to the swarming process and its build up in time, thereby shedding new 18 light on it and allowing its prediction several days in advance. 19The output of our monitoring method could be used to provide other signatures 20 highly specific to other physiological processes in honey bees, and applied to better 21 understand health issues recently encountered by pollinators.22 23

Application of continuous monitoring of honeybee colonies

“…In early applications of continuous monitoring, Gates (1914) and Hambleton (1925) placed 10- Bencsik et al (2011) frame Langstroth hives on mechanical balances in sheltered locations. About every 3 weeks from October 1907 to September 1908, Gates (1914 recorded "by means of assistance" weight and temperature data hourly for 2 to 3 days but presented data on daily weight change only for the winter, including changes caused by rainfall.…”

“…Vibration of the hive substrate is an important form of communication among bees (Nieh and Tautz 2000;Sandeman et al 1996;Schneider et al 1986), which have specialized receptors in their legs for receiving low-frequency signals, generally <300 Hz (Sandeman et al 1996), and the signals are thought to involve recruitment for foraging (Nieh and Tautz 2000). However, bees in a hive produce many frequencies of vibrations, from <10 to >1,000 Hz (Bencsik et al 2011), and how much of this spectrum is used by bees is unknown. Several methods have been used to gather vibration data.…”

“…Researchers have used two main approaches to process vibration and sound data: (1) apply principal components analysis (PCA) to the entire range of resulting frequencies, and then evaluate the components that capture most of the information, and (2) focus on the amplitudes of a few frequency bands identified beforehand. Bencsik et al (2011) embedded a vibration sensor, an accelerometer, in the wall of each of two hives, linking the sensor to a signal conditioner and a computer with considerable disk space. They collected data almost continuously from November 2008 to June 2009.…”

“…They collected data almost continuously from November 2008 to June 2009. Individual bees can be expected to vibrate at different frequencies and without phase coherence, so to resolve that problem, Bencsik et al (2011) maintained a frequency resolution of 20 Hz and examined relatively short data samples of 510 s. They found that those samples produced power spectra with a satisfactory signal to noise ratio to determine frequency peaks, and they filtered much of the data noise by first expressing data in the frequency domain PCA and then selecting the components with eigenvalues in the top 5 %. Principal components are orthogonal to each other, so each component represents a unique source of variance.…”

“…Principal components are orthogonal to each other, so each component represents a unique source of variance. Bencsik et al (2011) were thus able to focus on the strongest patterns without losing appreciable resolution; they used the PCA-filtered data to examine vibration frequencies prior to bee colony swarming.…”