Data from: Diel activity, frequency and visit duration of pollinators in focal plants: in situ automatic camera monitoring and data processing

Data collection on interactions between organisms and their environment has traditionally been conducted by on-site human observations, a time-consuming enterprise that could explain the shortage of around-the-clock observations of free-ranging wild animals. In this paper, I outline a time-efficient procedure to collect data on flower-visiting animals. The objectives were, first, to model diel activity rhythms by using cosine-based mixed-effects regression models (cosinor method) on data from an established automatic video monitoring system and, secondly, to test the use of a cheap off-the-shelf digital camera modified for automated monitoring of flower visitors. Two different model systems were studied: foraging bumblebees visiting focal white clovers, monitored around-the-clock (193 h) to model diel activity; and honeybees visiting thistles, monitored over a shorter period (5 h) to test the applicability and reliability of a new method for monitoring pollinators. The data were automatically entered and processed using R-scripts after manual filtering of the images, obviating the need for manual data entry prior to analysis. For diel activity in bumblebees, the model that gave the best fit included the 24-h fundamental period and one harmonic, a 12-h period to modulate the signal, together with temperature. The bumblebees were exclusive diurnal, with activity starting about 5 h after sunrise, peaking sharply in the afternoon and ending about 1 h before sunset. In addition to time of day, activity also increased with temperature. The off-the-shelf digital camera, Canon PowerShot®, with motion detection script, was triggered by every flower-visiting honeybee. In addition to recorded visitor frequency and visitor duration, it enabled high-resolution images, which could be important for species identification. Automatic camera recording is advantageous for close-up monitoring, compared with continuous video recording, because the latter demands more time and effort in reviewing the material. It could be used to study a range of different species such as pollinators, on-plant behaviour of herbivorous animals, cavity dwellers or cavity breeders. Moreover, the procedures for automatic data entry, data processing and statistical analysis for modelling diel activity rhythms could have great relevance for researchers using other types of camera monitoring systems operating 24 h per day.传统上，生物与环境间交互作用的数据采集均依赖人工现场观测，该过程耗时漫长，这也是自由活动的野生动物全天候观测数据匮乏的重要诱因。 本文提出了一种高效的访花动物数据采集流程。本研究的目标有二：其一，基于已建成的自动视频监测系统获取的数据，采用基于余弦的混合效应回归模型（余弦法，cosinor method）构建昼夜活动节律（diel activity rhythms）模型；其二，测试一款经改装以实现访花动物自动监测的低成本市售数码相机的实用性。本研究选取两类模式系统开展实验：一是对访指定白三叶的觅食熊蜂开展全天候（193小时）监测，以构建其昼夜活动节律模型；二是对访蓟类植物的蜜蜂开展5小时的短期监测，以验证新型传粉者监测方法的适用性与可靠性。所有数据经人工筛选影像后，通过R脚本自动录入并处理，无需在分析前手动录入数据。 针对熊蜂的昼夜活动节律，拟合优度最优的模型包含24小时基础周期、一个用于调节信号的12小时谐波项，以及温度变量。熊蜂为严格昼行性物种，其活动于日出后约5小时启动，午后达到活动峰值，并于日落前约1小时结束。除当日时段因素外，活动强度亦随温度升高而增强。 本次使用的市售数码相机为佳能PowerShot®，搭载运动检测脚本，可被每一只访花蜜蜂触发拍摄。该设备除可记录访花频次与访花时长外，还能获取高分辨率影像，这对物种鉴定具有重要意义。 相较于连续视频录制，自动相机录制更适用于近距离监测，因为后者需要耗费大量时间与精力审阅录制素材。该方案可用于研究多个类群，包括传粉者、植食性动物的植株行为、洞栖生物以及洞筑巢生物。此外，本研究中用于自动数据录入、数据处理及昼夜活动节律建模的统计分析流程，对于采用其他24小时运行的相机监测系统的研究者而言，亦具备重要参考价值。