Do animals as simple as flies have personalities?
Definitions of personality vary substantially across fields, and are often left implied. The heart of the phenomenon seems to be that individual organisms display idiosyncratic behavioral patterns that persist for a substantial amounts of time. By this definition, flies have abundant personality. The culmination of a couple years of work by Jamey Kain in our lab, our paper on fly personality was just published at PNAS.
The detection of such personality requires a method to statistically distinguish variability within the behavior of an individual from variability between individuals. If there is more variability between individuals than within (i.e. the observed behavioral distribution is over-dispersed compared to what would be expected from sampling error alone) – and the particular behavioral tendencies of individuals persist on subsequent re-evaluation then you have uncovered personality. There is mounting evidence in organisms ranging from pea aphids to trout exhibit personality in a wide variety of behaviors. We suspect that personality may be universal.
We came to work on this project while trying to map genetic differences between lab strains of Drosophila simulans that exhibit different light preferences. One runs toward light, while the other runs away from light, when startled. These strains are essentially isogenic, harboring no genetic diversity between individuals, and yet we found that the distribution of behavioral scores within each strain was broader than we’d expect based on sampling error alone (we gave each fly a choice to go toward or away from light 20 times, and thus had a fairly precise estimate of its preference).
This heterogeneity meant that mapping the genetic underpinnings of the difference in strain mean preference was going to require bigger sample sizes, and consequently be very labor intensive. That’s when we decided to build FlyVac, a platform for the autonomous manipulation of flies to measure phototaxis. That’s also when we realized that fly phototactic idiosyncrasy was an interesting phenomenon on its own.
So, after a couple years’ work, what have we learned about fly phototactic idiosyncrasy? A number of things:
• All fly strains seem to have it. Only when we put blind flies into the device, or gave them symmetrical light stimuli did we observe behavioral distributions matching what we’d expect based on sampling error alone. In fact, the phenomenon is not limited to flies. The white clover weevil Ischnopterapion virens performed very similarly.
• It is persistent. Flies recovered from FlyVac and tested up to 28 days later show significantly correlated phototactic preferences
• Genetic differences between animals cannot explain their behavioral differences. There isn’t much genetic diversity to begin with in most of the lines we examined, but additionally we inbred them for 10 generations, mating daughters back to fathers or grandfathers (sorry flies). This did not reduce the magnitude of personality; if anything, it amplified it.
• Personalities cannot be inherited (even by non-genetic means, such as epigenetics). Mating two flies with strong positive (or negative) light preferences had no effect on the behavioral distribution of their progeny. In all cases the progeny had behaviors identically distributed as the parents.
• The gene white, and the neurotransmitter serotonin (whose synthesis may depend on white) act in wild type flies to suppress personality, driving behavior toward homogeneity.
Please see our paper for all the gory details:
Why does non-heritable personality exist?
This will be the focus of forthcoming paper from our lab (hopefully soon). But there are two basic ideas. Personality could be essentially noise, i.e. non-adaptive, but tolerated for some other greater good. Possible greater goods include rapid development, or avoiding the metabolic costs of additional signaling pathways or neurons that would suppress the developmental stochasticity that generates the variability.
Alternatively, it might be inherently advantageous. Retaining a population with high diversity might mean that when a transient selective pressure arises that favors a subset of animals, they will successfully produce a subsequent generation. If that selective pressure goes away, then the non-heritability of the behaviors means that the subsequent generation will revert to the original (well adapted for typical conditions) distribution of behaviors immediately. Another consideration is that if behaviors are entirely predictable, they can be exploited. If flies always run toward the light, predators will learn this, and any fly that runs away from the light will have an advantage. This is an example of frequency-dependent selection, in which the fitness of a particular phenotype depends on the number of other individuals exhibiting it. Such dynamics often equilibrate to what economists and game-theorists call mixed strategies, when the optimal strategy entails acting randomly from instance to instance.
Jamey and I have a Q and A conversation about this work in the following video/podcast. If you stick around to the end of the video, we’ll take you on a walking tour of the lab, and show you FlyVac in operation: