This web page was produced as an assignment for an undergraduate course at Davidson College.
Mathematical modeling of network-specific ages of honey bees allows for the accurate prediction of their behaviors, including death
Scientists seek to better understand complex animal systems through the study of social interactions within them. More specifically research of these networks allows for the identification and understanding of the roles of individuals within their population. Insects, for example, have long been thought to have strong relationships between social interactions and roles that they play within their hive. Bees in particular are known for elaborate social signals and information exchanges between one another (Wild et al. 2021). Analysis of such networks allows for the understanding of social exchanges, cultural behaviors, and infectious diseases within species (Sah et al. 2019). Smaller life forms such as these insects can act as model organisms, and can even be utilized to understand human interactions including roles within the current pandemic and the spread of COVID-19 (Firth et al 2020). However, these incredibly complex behaviors are shaped by a variety of external and internal factors throughout one’s lifetime and thus are hard to pinpoint.
To further investigate the relationships between social networks and the roles individuals play within them, Wild et al. looked specifically at honey bees and recorded videos of one colony at 3 Hz over 25 days from August 1st, 2016 to August 25th, 2016. Overall, 1920 individuals ages 0 to 8 weeks were tracked through automation. Within the colony, nest areas were annotated for the tasks they were associated with each day (Wild et al. 2021). Time spent in these areas was then used as a way to estimate what tasks each bee was completing, along with each bee’s daily contact frequency, food exchange, distance, and changes in movement speed after contacts. Using all of these interaction types, the authors were able to map each bee to a single unitless value, called their network age, which represents the bee’s role in the greater social network.
In order to verify that this network age correctly identifies a bee’s role in its hive, the authors quantified the relationship between social interactions and an individual’s spatial preferences. This was accomplished using regression in order to predict where a bee spent time in the nest (Wild et al. 2021). The results indicated that network age is twice as effective as biological age at predicting location preferences of bees along with a generally better predictive ability of time spent in task-associated areas of the nest. Beyond regression tests, Wild et al. also experimentally demonstrated that network age accurately captured an individual’s role in the hive. Setting up sucrose feeders, these scientists identified workers that foraged, comparing their known biological age to their calculated network age. Overall, network age more accurately reflected the tasks of these bees, even when as little as 1% of the hive was tracked.
Differences in network age and biological age also become apparent when looking at developmental changes within bees over their lifetimes. After about six days of a bee’s biological age, Wild et al. saw a gradual divergence of network ages, where certain bees suddenly started aging much more rapidly in terms of their network age before flattening out while other bees maintained a lower network age for longer, aging gradually like their biological age. It was then observed that those individuals with lower network ages worked in honey storage while those with high network ages grouped on the dance floor (Wild et al. 2021).
Knowing that network age has been established to be a good predictor of where and what an individual will be doing in the nest, Wild et al. then showed this to be true for up to 10 days into the future. In fact, the authors determined that network age is a better predictor of tasks an individual will be completing a week out than presently (Wild et al. 2021). Network age even acted as a good predictor for behaviors such as death, with biologically young but network old bees to be more likely to die within the week than biologically old but network young individuals. The authors hypothesized that this was due to the environmental factors these different groups were exposed to in their roles within the hive. Other behaviors that were found to be predictable by network age included movement patterns, circadian rhythm, and time of an individual’s peak activity.
The authors’ mathematical modeling of network age and their verification of its success in predicting task allocation in the context of honey bees establish a potential model organism, which can be applied to the study of other complex animal social networks. Still, the weakness of this study lies in its lack of confirmation that this modeling method works within other species or environmental contexts. Considering network age was calculated by recording individual’s movements through an annotated hive, this could be hard to replicate with larger organisms. Moreover, a complete understanding of the influences on network age has yet to be established, and further research would need to include looking at the internal and external actors that drive the observed relationships and developmental transitions. Doing so becomes incredibly important in order to understand how disturbances in these social networks have lasting impacts particularly in consideration of disease. Understanding the spread of information within populations lays the foundation for studying a variety of fields, such as medicine.
Mimi Ughetta is a Genomics Major at Davidson College in Davidson, NC. Contact her at miughetta@davidson.edu.
References:
Dmitry Grigoriev, (2019). https://unsplash.com/photos/yxXpjF-RrnA
Firth J. A., J. Hellewell, P. Klepac, S. Kissler, A. J. Kucharski, et al., 2020 Using a real-world network to model localized COVID-19 control strategies. Nature Medicine 26: 1616–1622. https://pubmed.ncbi.nlm.nih.gov/32770169/
Sah P., J. D. Méndez, and S. Bansal, 2019 A multi-species repository of social networks. Scientific Data 6: 44. https://pubmed.ncbi.nlm.nih.gov/31036810/
Wild B., D. M. Dormagen, A. Zachariae, M. L. Smith, K. S. Traynor, et al., 2021 Social networks predict the life and death of honey bees. Nature Communications 12: 1110. https://pubmed.ncbi.nlm.nih.gov/33597518/
Back to home page here.
© Copyright 2020 Department of Biology, Davidson College, Davidson, NC 28036.
