Temperature Dependency of Insect's Wingbeat Frequencies: An Empirical Approach to Temperature Correction

Abstract

This study examines the relationship between the wingbeat frequency of flying insects and ambient temperature, leveraging data from over 302,000 insect observations obtained using a near-infrared optical sensor during an eight-month field experiment. By measuring the wingbeat frequency as well as wing and body optical cross-sections of each insect in conjunction with the ambient temperature, we identified five clusters of insects and analyzed how their average wingbeat frequencies evolved over temperatures ranging from 10 °C to 38 °C. Our findings reveal a positive correlation between temperature and wingbeat frequency, with a more pronounced increase observed at higher wingbeat frequencies. Frequencies increased on average by 2.02 Hz/°C at 50 Hz, and up to 9.63 Hz/°C at 525 Hz, and a general model is proposed. This model offers a valuable tool for correcting wingbeat frequencies with temperature, enhancing the accuracy of insect clustering by optical and acoustic sensors. While this approach does not account for species-specific responses to temperature changes, our research provides a general insight, based on all species present during the field experiment, into the intricate dynamics of insect flight behavior in relation to environmental factors.

Keywords: acoustic sensor; clustering; environmental factors; field experiment; monitoring; optical sensor; temperature; wingbeat frequency.

Figure 1

Optical layout of the eBoss instrument.

Figure 2

(a) An example of an optical signal caused by a flying insect; the dotted line at the top indicates the background signal, lower dotted lines represent the drop in intensity due to the insect’s body and wings, respectively. $V_0$ represents the background signal, while $V_B$ and $V_w$ represent the body and wing contributions to the measured voltage. (b) Frequency spectrum of the insect signal, showing the first and fundamental peak corresponding to the wingbeat frequency of the insect (~180 Hz); the following peaks are the harmonics.

Figure 3

(a) The wingbeat frequency of each event as a function of the wing-to-body cross-section ratio, the five identified clusters are labeled C1 to C5. (b) BIC values for both GMM (ref line) and K-mean (blue line) methods as a function of the number of clusters, showing the ideal number of clusters at the elbow point.

Figure 4

The wingbeat frequency of each event as a function of the temperature recorded at the time of the transit by the weather station.

Figure 5

Cluster 1 to Cluster 5 show the average wingbeat frequency per temperature range (0.5 degree Celsius per bin) for each of the five clusters as well as a linear fit of the resulting points (red dashed lines).

Figure 6

Second-degree polynomial fit of slope values obtained from each of the five individual clusters. The horizontal error bar represents the dispersion of wingbeat frequency, and the vertical error bar indicates the standard deviation of each slope value.

Figure 7

(a) Initial wingbeat frequency distribution; (b) wingbeat frequency distribution after correction for a reference temperature of 20 °C.