Building models : developing the behavioural model of Temnothorax collective wall building to study the evolutionary robustness of self-organised algorithms

Abstract

This thesis focuses on wall building behaviour in Temnothorax ants as a case study of self-organised collective behaviour. It contains a progressing series of research packages, building towards one evolutionary question: how eusocial insect nest building algorithms successfully make the transition between two rule variants. I start by revising the existent behavioural model of Temnothorax wall building. By replicating the original agent-based model, I identify two issues: 1. the behavioural model performs poorly in conditions of low building material availability; and 2. the behavioural model lacks behavioural termination. I then introduce a revised version of the behavioural model (the gradual model) in which high stone density at building sites triggers a decrease in building activity, eventually leading to behavioural termination. I then compare the fit of both models to empirical data using laboratory observations of T. rugatulus wall building, applying a hidden Markov model framework to interpret the data. The gradual model provides the best match to the observed data. Finally, I use the revised model to test, in an agent-based model setting, how wall quality responds to different types of inter-worker variation in the building rule used: the presence of a mutant variant spreading within the colony; the co-existence of multiple variants; and widespread epigenetic individual variation. I find wall quality to be very robust to nearly any degree and frequency of variants. With additional simulations, I identify the two key elements of the building algorithm that provide robustness: the positive feedback effect, co-localising worker effort despite starting individual variation; and the existence of an area of overlap where activity occurs with high frequency under all variants (a buffer zone). I predict that these two components have been under selection for evolvability in wall building Temnothorax ants.

Date of Award	29 Nov 2022
Original language	English
Awarding Institution	University of St Andrews
Supervisor	Graeme Douglas Ruxton (Supervisor) & V Anne Smith (Supervisor)