Constraints on the evolution of learning ability in Drosophila

Département de Biologie, Unité d’Ecologie & Evolution
Constraints on the evolution of learning ability in Drosophila melanogaster
Munjong Kolss
The principal aim of my PhD research was to find out more about the factors which may
constrain the evolution of higher learning ability. Learning allows animals to modify their
behavior in an adaptive way, based on previously obtained information. However, better
cognitive abilities are unlikely to come cost-free. Alleles positively affecting learning
ability may have negative pleiotropic effects on other fitness-related traits. Alternatively,
the relevant genes may be linked to other genes which affect different fitness components.
At a phenotypic level, better cognitive abilities may require greater allocation of resources
into the neural and sensory structures involved in the acquisition, processing, storage, and
retrieval of information. This would then leave less available for somatic maintenance,
growth, immune responses, or reproduction. Only if the benefits of learning outweigh these
potential fitness costs, higher learning ability may be expected to evolve. I addressed these
evolutionary issues using artificial selection experiments as main research tool and the fruit
fly Drosophila melanogaster as model organism.
First, I used Drosophila lines selected for better learning and memory performance
(“high-learning” lines) and unselected “low-learning” control lines (MERY & KAWECKI
2002, Proc. Natl. Acad. Sci. U. S. A. 99:14274-14279) to investigate whether selection for
improved learning ability had led to correlated changes in fitness-related traits. Flies from
“high-learning” lines were heavier than those from “low-learning” control lines. This may
suggest that the former are more constrained, since they need to reach a greater critical
mass to complete development. I found no constitutive costs of learning ability (i.e. costs
occurring even in the absence of learning) in terms of reduced desiccation resistance or
fecundity, although operational costs (which occur when learning is actually executed)
regarding these traits had previously been found. This indicates that different mechanisms
are involved in these two types of costs.
Second, I set up a new selection experiment, where populations of fruit flies were
selected both for better larval development on restricted food and for improved learning
ability. There was no detectable response to selection for higher learning ability within the
scope of my experiment. By contrast, selection on restricted food led to significantly
higher survival and faster development. Moreover, it led to a decrease in learning ability as
a correlated response. Since under natural conditions, there is probably high selection
pressure on flies to perform well under nutritional stress, this result indicates a serious
constraint on the evolution of learning ability.
Prof. Dr. Louis-Félix Bersier
Dr. Tadeusz J. Kawecki
Prof. Dr. Louise E. M. Vet
Dr. Mathias Kölliker