Social networks predict the life and death of honey bees

Abstract

In many social systems, an individual's role is reflected by its interactions with other members of the group (Gordon 2010, Pinter-Wollmann et al. 2014, Krause 2015, Farine & Whitehead 2015. In honey bee colonies ( Apis mellifera ), workers generally perform different tasks as they age, yet there is high behavioral variation in same-aged bees (Seeley 1982, Robinson 1992, Huang and Robinson 1996, Johnson 2010. It is unknown how social interactions within the colony relate to an individual's tasks throughout her life. We propose a new method to extract a single number from each individual's interaction patterns in multimodal social networks that captures her current role in the colony. This "network age" is better than biological age at predicting task allocation (+99%), survival (+157%), and activity patterns (+44-108%) and even predicts task allocation up to one week (around a sixth of her typical lifespan) into the future. Network age identifies distinct developmental paths and task changes throughout a bee's life: We show that individuals change tasks gradually and exhibit high task repeatability, and that same aged bees form stable behavioral subgroups in which they predominantly interact with one another. While we derived interaction networks by automatically tracking a fully tagged colony, we show that tracking only 5% of the bees is sufficient to extract a meaningful representation of the individuals' interaction patterns, demonstrating the feasibility of our method for detecting complex social structures with reduced experimental effort. Since network age more accurately predicts task allocation than biological age, it could be used in experimental manipulations to quantify shifts in the timing of task transitions as a response. We extend our method to extract interaction patterns relevant to other attributes of the individuals, such as their mortality, opening up a broad range of possible applications. Our approach is a scalable instrument to study individual behavior through the lens of social interactions over time in honey bees and other complex social systems.

Collective Regulation of Metabolism, Development and Longevity Through a Socially Exchanged Fluid

“…In ant trophallactic fluid, there are core proteins that are always present but can change in abundance, and proteins whose presence varies based on colony life cycle and environmental context. At the individual level, certain proteins correlate with behavioral caste within the colony, a trait known to encompass both individual task requirements and age (34)(35)(36). Certain overarching patterns across our comparisons are indicative of function: Proteins abundant in mature colonies and nurses are likely factors required to build large workers and sexuals.…”

TRex, a fast multi-animal tracking system with markerless identification, and 2D estimation of posture and visual fields

“…RFID tags are useful in many situations, but are also limited since individuals have to be in very close proximity to a sensor in order to be detected ( Bonter and Bridge, 2011 ). Attaching fiducial markers (such as QR codes) to animals allows for a very large number (thousands) of individuals to be uniquely identified at the same time (see Gernat et al, 2018 ; Wild et al, 2020 ; Mersch et al, 2013 ; Crall et al, 2015 ) – and over a much greater distance than RFID tags. Generating codes can also be automated, generating tags with optimal visual inter-marker distances ( Garrido-Jurado et al, 2016 ), making it feasible to identify a large number of individuals with minimal tracking mistakes.…”

TRex, a fast multi-animal tracking system with markerless identification, and 2D estimation of posture and visual fields

“…RFID tags are useful in many situations, but are also limited since individuals have to be in very close proximity to a sensor in order to be detected ( Bonter and Bridge, 2011 ). Attaching fiducial markers (such as QR codes) to animals allows for a very large number (thousands) of individuals to be uniquely identified at the same time (see Gernat et al 2018, Wild et al 2020, Mersch et al 2013, Crall et al 2015 ) – and over a much greater distance than RFID tags. Generating codes can also be automated, generating tags with optimal visual inter-marker distances ( Garrido-Jurado et al, 2016 ), making it feasible to identify a large number of individuals with minimal tracking mistakes.…”