Evolution and Genetics

The individually-monitored Soay sheep on St Kilda provide an excellent opportunity to study genetics and evolution. The Soay sheep pedigree is a starting point for studies of the mating system and variation in breeding success, the genetic basis of trait variation and understanding microevolution in the population.


1) The Soay sheep pedigree

Not only do we know every sheep in the study population, but we also know who their relatives are. We can identify a lamb’s mother from observations of suckling in the field. Assigning paternity is more difficult, because males provide no parental care: however, we can use information from DNA samples from a lamb, its mother and candidate fathers (those males that were alive and in the study area in the previous rut) to identify the father amongst possible candidates. Fig 1 shows the current (2012) pedigree of study area sheep.

Figure 1. The Village Bay Soay sheep population pedigree (courtesy Camillo Berenos). Red lines show links to mothers; blue lines to fathers. The many blue lines entering from the right of the figure indicate rams which enter the study area from other parts of the island to breed.


2. The mating system and variation in breeding success

During the rut, ewes mate repeatedly during an oestrus lasting between one and four days, and most ewes mate with more than one male; occasionally they are mated by many males. DNA analysis has revealed a number of interesting patterns within this system.

First, and as expected from such systems, male mating success is very uneven, with many rams having low lifetime breeding success and few having high breeding success (Figure 2; Coltman et al 1999). A key determinant of lifetime reproductive success is longevity, and longevity, in turn, is largely a matter of luck: a ram born just after a population crash has a much better chance of surviving to a ripe age than one born other circumstances, since he is in his prime when the next population crash hits, and he is more likely to survive it (Figure 3; Coltman et al 1999).

Figure 2. Lifetime breeding success of all study area males born between 1986 and 1996.  Close bars and symbols are for all males, while open bars and symbols are for those males that had died at the time of analysis (i.e. their life histories were complete). From Coltman et al, 1999.


Figure 3. Mean lifetime breeding success of cohorts of Soay males is determined by density in the year of birth.  The 1986 and 1989 cohorts were born just after population crashes, while the 1988 and 1991 cohorts were born just before crashes. From Coltman et al 1999.

Second, in contrast to other ungulates, very young males participate in the rut and do occasionally obtain matings. Ram lambs (aged seven months by the time of the rut) are sexually mature and compete as best they can for matings. Although their breeding success is generally low, in years after a population crash which has removed a lot of older males, it can rise to as much as 0.6 offspring per ram lamb. This behaviour is consistent with life history theory, in that individuals with low life expectancy should attempt to breed early.

The attentions of ram lambs contribute to an overall high level of multiple mating by females. Just how high is apparent when considering the paternity of twins: around 75% of twins actually have a different genetic father!

Variance in female breeding success is lower than in males: virtually all females over the age of one attempt to breed each year they are alive, with the main variation in fecundity being cause by increased levels of twinning and female lambs lambing (i.e. on their first birthday) following years of good feeding conditions.


3. The genetic basis of trait variation

We can also use pedigree information to see how much of the observed variance (differences) between individuals is due to genetic versus environmental effects. Relatives share genes – therefore if a particular characteristic or trait such as body size or fertility is strongly determined by genetic effects, relatives should have similar values for that trait and the heritability of that trait is greater than zero. In principle the higher the heritability, the greater the potential that the trait has to respond to selection on that trait.

Figure 4 shows the heritability of Soay sheep body weight at different ages. The heritability of body weight is low at birth, but increases with age to a plateau at about 2 years of age.  Over much of the lifespan the heritability of body weight is around 0.4, indicating that around 40% of the variance is genetic in origin and could respond to selection. While the heritability of birth weight is very low,, in other studies we have shown that there is appreciable maternal genetic variation for birth weight – i.e., birth weight is partly determined by the genetics of the mother and there is inherited variation for offspring birth weight.

Figure 4. Heritability of Soay sheep body weight at different ages (line labelled Model 2.11). From Wilson et al, 2007.


The genes underlying variation in particular characteristics can sometimes be located and identified by DNA profiling members of a populations and searching for DNA variation that correlates with variation in the characteristics.  We have used this approach to genetically map the genes underlying polymorphisms in coat colour and pattern and in horns (see ‘Meet the Sheep’ for examples). For example, as Fig 5 shows, we have mapped the gene underlying both the major varation (normal, scurred and polled) and more subtle continuous variation (length and base circumference) in horn variation to chromosome 10. This is in the same spot as other sheep geneticists have found, suggesting that horn polymorphism in sheep is all due to variation in one particular gene.

Figure 5. A plot showing that the Soay sheep horn polymorphism, and much of the quantitative variation in normal-horned males as well, is determined by a gene located on chromosome 10. (a) horn type in all sheep; (b) horn type in males only (c) horn length in normal-horned males; (d) horn base circumference in normal-horned males. Each dot represent the association (measured as a probability) that a single nucleotide polymorphism in a known location in the sheep genome is associated with the (unknown) locus controlling horn variation. From Johnston et al (2013).


4. Understanding microevolution.

A major component of our work on St Kilda is to try and understand the extent to which selection (acting via population crashes or variation in breeding success) on inherited traits results in change (or lack of it) over time. A recent example is our analysis of why the genetic variant (allele) causing scurred or polled horns persists in the population, considering that a male with small horns seems unable to mate many oestrous ewes. To do this we used all the information on female and male breeding success embodied in the pedigree shown above as well as all our detailed information on the lifespan of each individual to investigate the fitness of the three genotypes at the horns locus.

In females, horn type is immaterial to either the annual breeding success or lifespan components of lifetime fitness.  However, in males we found a most interesting pattern. It is indeed true that males with normal horns, which are either homozygous for the normal-horned allele or heterozygous for the normal-horned and polled allele have higher annual breeding success than scurred males which are homozygous for the polled allele (Figure 6a). However, the scurred allele is associated with a longer lifespan, and this includes the heterozygous males (Figure 6b). Heterozygous males thus end up with the best of both worlds: high breeding success and long life (Figure 6c).  As long as this situation persists, the allele causing scurred males will persist.  It’s an example of ‘heterozygote advantage’, a phenomenon which has not often been observed but is a potentially strong force in maintaining variation in nature.

Figure 6. Annual reproductive success (a), annual survival (b) and lifetime fitness (c)  for males with different genotypes at the locus underlying horn type. From Johnston et al (2013).