Understanding how and why animals choose what to eat is one of the fundamental goals of nutritional and behavioral biology. This question can be scaled to animals that live in social groups, including eusocial insects. One of the factors that plays an important role in foraging decisions is the prevalence of specific nutrients and their relative balance. This dissertation explores the role of relative nutrient content in the food selection decisions of a species that is eusocial and also agricultural, the desert leafcutter ant Acromyrmex versicolor. A dietary choice assay, in which the relative amount of protein and carbohydrates in the available diets was varied, demonstrated that A. versicolor colonies regulate relative collection of protein and carbohydrates. Tracking the foraging behavior of individual workers revelaed that foragers vary in their relative collection of experimental diets and in their foraging frequency, but that there is no relationship between these key factors of foraging behavior. The high proportion of carbohydrates preferred by lab colonies suggests that they forage to nutritionally support the fungus rather than brood and workers. To test this, the relative amounts of 1) fungus, and 2) brood (larvae) was manipulated and foraging response was measured. Changing the amount of brood had no effect on foraging. Although decreasing the size of fungus gardens did not change relative P:C collection, it produced significant increases in caloric intake, supporting the assertion that the fungus is the main driver of colony nutrient regulation. The nutritional content of naturally harvested forage material collected from field colonies was measured, as was recruitment to experimental diets with varying relative macronutrient content. Field results confirmed a strong colony preference for high carbohydrate diets. They also indicated that this species may, at times, be limited in its ability to collect sufficiently high levels of carbohydrates to meet optimal intake. This dissertation provides important insights about fundamental aspects of leafcutter ant biology and extends our understanding of the role of relative nutrient content in foraging decisions to systems that span multiple trophic levels.

A swarm describes a group of interacting agents exhibiting complex collective behaviors. Higher-level behavioral patterns of the group are believed to emerge from simple low-level rules of decision making at the agent-level. With the potential application of swarms of aerial drones, underwater robots, and other multi-robot systems, there has been increasing interest in approaches for specifying complex, collective behavior for artificial swarms. Traditional methods for creating artificial multi-agent behaviors inspired by known swarms analyze the underlying dynamics and hand craft low-level control logics that constitute the emerging behaviors. Deep learning methods offered an approach to approximate the behaviors through optimization without much human intervention.

This thesis proposes a graph based neural network architecture, SwarmNet, for learning the swarming behaviors of multi-agent systems. Given observation of only the trajectories of an expert multi-agent system, the SwarmNet is able to learn sensible representations of the internal low-level interactions on top of being able to approximate the high-level behaviors and make long-term prediction of the motion of the system. Challenges in scaling the SwarmNet and graph neural networks in general are discussed in detail, along with measures to alleviate the scaling issue in generalization is proposed. Using the trained network as a control policy, it is shown that the combination of imitation learning and reinforcement learning improves the policy more efficiently. To some extent, it is shown that the low-level interactions are successfully identified and separated and that the separated functionality enables fine controlled custom training.