Grace C. McKenzie-Smith,Scott W. Wolf,Julien F. Ayroles,Joshua W. Shaevitz
Abstract:Animal behavior spans many timescales, from short, seconds-scale actions to circadian rhythms over many hours to life-long changes during aging. Most quantitative behavior studies have focused on short-timescale behaviors such as locomotion and grooming. Analysis of these data suggests there exists a hierarchy of timescales; however, the limited duration of these experiments prevents the investigation of the full temporal structure. To access longer timescales of behavior, we continuously recorded individual $\textit{Drosophila melanogaster}$ at 100 frames per second for up to 7 days at a time in featureless arenas on sucrose-agarose media. We use the deep learning framework SLEAP to produce a full-body postural data set for 47 individuals resulting in nearly 2 billion pose instances. We identify stereotyped behaviors such as grooming, proboscis extension, and locomotion and use the resulting ethograms to explore how the flies' behavior varies across time of day and days in the experiment. We find distinct circadian patterns in all of our stereotyped behavior and also see changes in behavior over the course of the experiment as the flies weaken and die.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is: How to capture and analyze the continuous behavioral changes of Drosophila melanogaster on a long - time scale, especially the behavioral patterns across minutes, hours and days. Specifically, the researchers hope to reveal the behavioral characteristics of Drosophila melanogaster in different time periods through high - resolution continuous recordings, including circadian rhythms, behavioral changes during the aging process, etc.
### Research Background and Problems
1. **Behavioral Research on Multiple Time Scales**
- Animal behavior involves multiple time scales, from second - level actions (such as grooming, feeding) to hour - level circadian rhythms, and then to changes throughout the life span (such as aging). Most quantitative behavioral studies focus on short - time - scale behaviors, such as movement and grooming, and less is known about long - time - scale behavioral changes.
2. **Experimental Challenges**
- Traditional experimental methods are difficult to meet the needs of long - time continuous recording and high - resolution data simultaneously. For example, a simple light - beam - interruption experiment can record the activity level for several days, but it cannot capture fine - grained behavioral details. And high - resolution data collection is usually limited by computing resources and can only be carried out in a short time.
3. **Research Objectives**
- This study aims to capture the behavioral changes of Drosophila melanogaster through long - time (4 - 7 days) high - resolution continuous recordings, especially the behavioral changes during circadian rhythms and the aging process. The researchers hope to reveal the behavioral patterns of Drosophila melanogaster in different time periods through these data and provide new insights into understanding the temporal structure of animal behavior.
### Method Overview
- **Experimental Design**
- Use infrared illumination and infrared - sensitive cameras to record 47 freely - moving Drosophila melanogaster at a speed of 100 frames per second for 4 - 7 days.
- The experimental environment simulates a 12 - hour day - night cycle, providing enough space for Drosophila melanogaster to move freely, but restricting flight and other behaviors that affect tracking quality.
- **Data Analysis**
- Use the deep - learning framework SLEAP to extract the whole - body pose data of Drosophila melanogaster, generating nearly 2 billion pose instances.
- Analyze the behavioral data through methods such as principal component analysis (PCA) to identify different behavioral patterns, such as grooming, antenna extension and movement.
- The researchers also used combinatorial data analysis methods to process closed data (that is, the sum of the time proportions of all behaviors is 1) to better describe behavioral changes.
### Main Findings
1. **Circadian Rhythms**
- Drosophila melanogaster shows more movement and grooming behaviors during the day, and is more in a static state at night.
- In the first hour after the lights are turned on, Drosophila melanogaster shows an obvious peak in movement and grooming, showing an anticipatory response to light changes.
2. **Behavioral Changes during the Aging Process**
- As the experiment progresses, the behavior of Drosophila melanogaster gradually weakens, especially after the 4th day, the proportion of static behavior increases significantly.
- This behavioral change is similar to the process of accelerated aging, and the difference in circadian rhythms also gradually decreases.
3. **Unique Morning Behavior**
- The first hour after the lights are turned on is a unique time point. Drosophila melanogaster shows a higher frequency of grooming and movement during this period, which may be to restore the body parts that were not groomed at night.
### Conclusion
This study successfully captured the behavioral changes of Drosophila melanogaster on multiple time scales through long - time high - resolution continuous recordings, and revealed the behavioral patterns during circadian rhythms and the aging process. These findings not only deepen our understanding of Drosophila melanogaster behavior, but also provide valuable data and methods for further research on the behavior of other animals.
### Formula Examples
- **Geometric Mean** (used to describe behavioral components):
\[
\text{Geometric Mean}=\left(\prod_{i = 1}^{n}x_{i}\right)^{\frac{1}{n}}
\]
where \(x_{i}\) represents the time proportion of each behavior.
- **Principal Component Analysis (PCA)**
\[
\mathbf{X}_{\text{centered}}=\mathbf{X}-\mathbf{\mu}
\]
where \(\mathbf{X}\)