Monday, August 12, 2019
Exhibit Hall, Kentucky International Convention Center
Background/Question/Methods
Many animals are capable of associative learning, yet very little is known about how learning affects the fitness of a forager. The complexity, temporal variability, and stochasticity of an environment determine the adaptive value of learning; therefore, experimental and theoretical work should address the ecological setting in which an animal learns.
To investigate the role of learning in making foraging decisions, we model an insect forager which dynamically updates its foraging strategy as it samples the environment. Foraging strategies are tested across combinations of ecological variables, including the total number of resource types, the reward, encounter rate, and variance of each resource type, and the rate and manner of change of the environment. We compare learning with fixed and optimal foraging strategies to quantify its costs and benefits.
In constant environments, we hypothesize that learning will be most beneficial when resources occur in relatively similar abundance, differ significantly in quality, and show little variation in quality within a resource type. In temporally variable environments, learning should be most valuable if the time scale of variation is comparable to the rates of resource discovery and collection.
Results/Conclusions
The fitness benefit of learning follows a non-linear relationship between encounter rates, reward amounts, and the rate of temporal variation. Learning offers the greatest fitness benefit in environments where significantly higher-quality resources occur occasionally, but not too rarely. For example, in an environment with only two resources, if one resource offers four times the reward as the other, but occurs 1/4 as often, learning provides a 50% fitness increase over a fixed foraging strategy; for the same rewards, if the better resource occurs 1/20 as often as the other, the fitness increase disappears, and learning may actually be detrimental. Contrary to our original hypothesis, we find little effect of variability in reward quantity for a particular resource.
Many animals are capable of associative learning, yet very little is known about how learning affects the fitness of a forager. The complexity, temporal variability, and stochasticity of an environment determine the adaptive value of learning; therefore, experimental and theoretical work should address the ecological setting in which an animal learns.
To investigate the role of learning in making foraging decisions, we model an insect forager which dynamically updates its foraging strategy as it samples the environment. Foraging strategies are tested across combinations of ecological variables, including the total number of resource types, the reward, encounter rate, and variance of each resource type, and the rate and manner of change of the environment. We compare learning with fixed and optimal foraging strategies to quantify its costs and benefits.
In constant environments, we hypothesize that learning will be most beneficial when resources occur in relatively similar abundance, differ significantly in quality, and show little variation in quality within a resource type. In temporally variable environments, learning should be most valuable if the time scale of variation is comparable to the rates of resource discovery and collection.
Results/Conclusions
The fitness benefit of learning follows a non-linear relationship between encounter rates, reward amounts, and the rate of temporal variation. Learning offers the greatest fitness benefit in environments where significantly higher-quality resources occur occasionally, but not too rarely. For example, in an environment with only two resources, if one resource offers four times the reward as the other, but occurs 1/4 as often, learning provides a 50% fitness increase over a fixed foraging strategy; for the same rewards, if the better resource occurs 1/20 as often as the other, the fitness increase disappears, and learning may actually be detrimental. Contrary to our original hypothesis, we find little effect of variability in reward quantity for a particular resource.
Foraging strategies that exhibit optimal performance in stable environments may be extremely sensitive to ecological change. Foraging strategies which are robust to environmental stochasticity allow insects to adapt to changes in resource availability, including those caused by human interventions. Our results, coupled with ongoing experimental work, help explain why some insect species are better able to adapt to urban or agricultural habitats.