Selection in a complex world: reconciling predicted and observed responses to natural selection

Project Details


Expectations of evolution in natural populations consistently fail to match observations, a failure that stands in stark contrast to the continued success of artificial selection. The discrepancy is presumably because the key ingredients of evolution (selection and heritability) are being wrongly measured, and specifically because estimates fail to account for the complexity of the natural environment. Here, we will use an extensive data set on a wild mammal population, the Soay sheep on St. Kilda, and a suite of more sophisticated or novel analytical approaches in a comprehensive exploration of patterns of natural selection. In doing so, we hope to reconcile predictions and observations of microevolution in the wild via one or more of the following hypotheses: (1) An evolutionary response in a focal trait is constrained by genetic correlations with other measurable traits, either through negative genetic correlations or antagonistic selection pressures; (2) Traits do not respond as expected because their heritabilities vary with environmental conditions, and heritabilities are systematically lowest during the periods when selection is strongest; (3) Estimates of selection are erroneous either because the fitness measure used fails to account for population dynamics, or because it fails to consider simultaneous selection through all stages of the life cycle; (4) The appearance of phenotypic selection is generated by an environmentally-determined covariance between trait and fitness; (5) Traits are indeed changing as expected, but the evolutionary response is masked by a concurrent change in environmental conditions with an opposing effect on the focal trait.

Layman's description

Observations ranging from Darwin’s pigeons to the latest elite Holstein dairy cow clearly demonstrate the power of artificial selection. Responses to selection in nature are much less predictable. In fact, in a series of long-term studies of individuals in natural animal populations, strong selection on inherited characteristics often suggests that there should be a substantial response to selection, but we see no change or even change in the opposite direction to that predicted. In recent years, several explanations for this failure have emerged, many of which invoke environmental variation as the main complicating factor. Specifically, environmental variation may be causing us to overestimate the extent to which characteristics are inherited, it may be causing us to overestimate the strength of selection and it may be masking the response to selection. In addition, there may be further complexities that we have as yet failed to take into account due to the fact that the different characteristics of an organism are under simultaneous selection and may also be genetically correlated, due to the possibility that the expression of genetic variation varies with environmental variation, and due to the complexities of measuring selection when the dynamics of a population are taken into account. We plan to investigate these issues in an individuallly-monitored population of Soay sheep on St. Kilda, where the population size, and hence food availability and parasite load, vary from year to year. We will be using analysis techniques which have only recently been developed or applied to natural populations.

Key findings

Please see appended final report files for this grant which ended in 2010. It was succeeded by another grant (see related grant below) under which we are reporting all new developments.
Effective start/end date1/02/0531/01/10


  • NERC: £442,332.00


Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.