Carpenter Ant Trophallaxis as Modeled through Continuous-time Markov Chains
Open Access
Author:
Mattiace, John Michael
Area of Honors:
Statistics
Degree:
Bachelor of Science
Document Type:
Thesis
Thesis Supervisors:
Ephraim Mont Hanks, Thesis Supervisor Naomi S Altman, Thesis Honors Advisor
Keywords:
markov chain trophallaxis carpenter ant exponential distribution
Abstract:
Recent research has revealed that in ant colonies, forager ants tend to engage in more trophallaxis events when compared to the rest of the colony. Though the biological mechanism itself is well understood, there remain many variables and factors yet uncovered that influence it. The aim of this work is to investigate the probabilistic structure of trophallactic behavior in ants. In particular, this work explores data from one colony of Camponotus pennsylvanicus and proposes stochastic models for its behavior to better understand the nature of these events through the models’ properties. Using maximum likelihood estimators and Kolmogorov-Smirnov tests, we show that the distribution of trophallaxis event times for the entire colony are not well-modeled by an exponential distribution, but that for individual ants and pairwise combinations of ants by type, both time in trophallaxis and wait times between trophallaxis events were not significantly different from exponential distributions. This allows us to model individual and pairwise trophallaxis behavior using continuous-time Markov chain models. Both individual ants and pairwise type combinations were simulated to construct histograms and continuous-time Markov Chain approximations as a visual comparison of the data. All data analyses were conducted using the software R and publicly available packages.