Fig. 1. Network age, a one-dimensional descriptor of an individual’s role within the colony, based on an individual’s interaction pattern.

Using the BeesBook automated tracking system, we obtain lifetime tracking data for individuals (a). These tracks are used to construct multiple weighted social interaction networks (b). We aggregate daily networks (c) to then extract embeddings that group bees together with similar interaction patterns, using spectral decomposition (d). Finally, we use a linear transformation (e; CCA canonical-correlation analysis) that maximizes correlation with the fraction of time spent in different nest areas (f) to compress them into a single number per day called “network age” (g).