The ecology of evolution: the role of environmental heterogeneity in evolutionary dynamics

Environmental heterogeneity, in the form of temporal and spatial variation in different variables such as weather and food availability, has the potential to have a substantial impact on the quantitative genetic architecture of a population, and hence on the course of evolutionary change, the maintenance of genetic variation and the extent to which organisms trade off different activities. However, no studies have yet investigated this topic in depth in a natural population. Using an exceptional long-term data-set on the unmanaged Soay sheep population on St. Kilda and measures of four kinds of environmental variation (weather, population density, food availability and parasite abundance), we will provide a comprehensive analysis of the impact of environmental variation on evolutionary dynamics. We will address three broad questions, each with a number of subsidiary questions:

(1) How does environmental variation affect the multivariate genetic architecture?
(a) How do genetic variances and covariances between traits (i.e. the G matrix) change with varying environmental conditions (indicating genotype-environment (GxE) interactions)?
(b) Are GxE patterns consistent across traits and environmental variables, for example is expression of genetic variance greater under more favourable environmental conditions as some authors suggest?
(c) What are the impacts of environmental variation on other components of phenotypic variance and covariance (e.g., cohort, maternal and environmental sources of trait (co)variance)?
(2) How does environmental variation affect the multivariate selection regime?
(a) Does the strength of directional selection on individual traits change systematically with environmental conditions?
(b) Does correlational selection change with environment such that different trait combinations are preferred under different conditions?
(c) Do selective trade-offs result from fluctuating selection pressures across environments?
(d) To what extent does observed selection on phenotype reflect genetic associations between a trait and fitness?
(3) How does environmental variation impact on expectations of evolutionary change?
(a) Do the major axes of selection align with those of genetic variance for the multivariate phenotype (facilitating rapid evolution), or are they orthogonal (constraining evolution)?
(b) Is there negative genetic covariance between fitness expressed in different environments?
(c) How do predictions of phenotypic response to selection compare if generated by the breeder's equation using phenotypic selection gradients versus the genetic covariance between trait and fitness (2d)?

Although we have a good general understanding of how evolution operates, we have only limited insight into its detailed workings in natural populations. A prominent challenge is to understand the extent to which variation in environmental conditions affects evolutionary change within natural populations. Improved knowledge in this area will greatly advance our understanding of how genetic diversity is maintained in natural populations and of how populations are likely to respond to predicted changes in climate.



The characteristics of individuals in a population are a consequence of their genes, the environment in which they live and the evolution that has occurred up to that point. So much is clear, but this simple statement conceals much complexity. Specifically, the consequences of carrying particular genes can vary with environmental conditions; it is possible (but not generally proven) that organisms express more of their genetic potential under good rather than bad environmental conditions. Similarly, the strength of natural selection, whereby some individuals have higher survival or reproductive success than others, may vary with environmental conditions. Since evolutionary change occurs when natural selection acts on inherited characters, environmental fluctuations can therefore have a substantial impact on evolution. Finally, because some of the genes underlying a particular character (e.g. body size) may be the same as, or associated with, the genes underlying another character (e.g. parasite resistance), characters are not free to evolve independently of one another. Instead the effect of natural selection on one character will be constrained by any selection that is also occurring on other characters to which it is genetically tied. Recent research suggests that these constraints also vary with environmental conditions. Currently, we have very little understanding of how these various processes interact to shape evolution in natural populations.



In this study we will investigate the effect of environmental conditions on evolutionary processes using data from an unmanaged study population of Soay sheep on the island of St Kilda, NW Scotland. The data set is particularly useful for this study because, firstly, we have records on the individual life histories of several thousand sheep measured across 23 years. DNA profiling has been used to determine paternity and provide a family tree suitable for the kind of genetic analyses proposed here. Secondly, our previous research has demonstrated that there are four kinds environmental variation which show strong temporal variation affecting sheep performance, and we will be investigating all four kinds of variation in this study: the weather, sheep population density, food availability and parasite abundance.



We will address the effect of the environmental conditions described above on the expression of genetic variation, the strength of natural selection and the genetic relationships between characters. Specifically we will ask how the environmental affects first, the expression of genetic variation in single characters like body weight: is more genetic variation always expressed when conditions are good? Second does environmental variation affect genetic relationships between characters? For example, are body size and parasite resistance freer to evolve separately under good or bad conditions? Third, is selection stronger or weaker under good environmental conditions? Together these analyses will allow us to predict the course of evolution under specific time series of environmental conditions - for example systematic climate change. Long-term studies of individually-recognised wild animals such as these are an important source of information on many aspects of ecology and evolutionary biology, and the proposed grant would guarantee continuity of data collection for one of the UK's most valuable ecological and evolutionary field data sets.

We have investigated predictions of phenotypic response by the breeder's equation (univariate and multivariate forms) and from the genetic covariance of trait and fitness for four morphological traits in Soay sheep. There are appreciable differences in the predictions, with the breeder's equation consistently predicting increase in body size while the covariance measure does not. The genetic covariance predictions are closer to observation (the sheep are actually getting smaller). These results suggest that predictions of selection from the breeder's equation are upwardly biased by unmeasured factors in natural populations.