Measuring factors affecting honey bee attraction to soybeans using bioacoustics monitoring

Honey bee foraging activity in four soybean fields during the blooming period (R2-R3), as well as before or after bloom, was recorded continuously using Sony ICD-PX370 audio recorders affixed to a wooden stake and placed below the soybean canopy, at the height of the highest flowers, approximately 35 m from the field edge. Continuous recordings were broken into 1 hour chunks and recording sequences made at night were discarded. The wingbeat frequency of honey bees (234±13.9 Hz) was identified first using an automated method implemented in Python and using the package scikit-maad. Potential bee sounds were defined as anything within the range of 370-570 Hz with a duration of 1 sec. and a threshold intensity of 0.0001 using the “find_rois_cwt” function. Second, all automated detections were manually corrected by importing label files into the audio software Audacity.

本研究对4块大豆田在开花期（R2-R3）及花前、花后的蜜蜂觅食活动开展了连续录音监测：采用索尼ICD-PX370音频录音机，将其固定于木质桩上，置于大豆冠层下方、与最高花齐平的高度，距田边约35米处。连续录音被分割为1小时的片段，夜间录制的序列予以舍弃。首先，借助Python实现的自动化方法并调用scikit-maad工具包，识别蜜蜂的振翅频率（234±13.9 Hz）。通过"find_rois_cwt"函数，将频率范围370~570 Hz、时长1秒、强度阈值为0.0001的声音定义为潜在蜜蜂声响。其次，将标签文件导入音频编辑软件Audacity，对所有自动化检测结果开展人工校正。