In population and conservation genetics it is widely accepted that small, isolated populations rapidly lose genetic diversity as a result of genetic drift (Falconer & Mackay 1996
; Frankham 1996
; Keller & Waller 2002
) that could drive the population to extinction (Spielman et al. 2004
). The theoretical reduction in heterozygosity (H
) in a closed random mating population is predicted by the classical formula (Hedrick 2000
is the effective population size, H0
is the heterozygosity at generation 0, Ht
is the heterozygosity expected at generation t
, and t
is the number of generations. Since the loss of genetic diversity is faster in small compared with large populations, a population experiencing a bottleneck or fluctuating dynamics is assumed to be subject to stronger drift than a stable population with a similar average population size (Nei et al. 1975
; Motro & Thomson 1982
; Vucetich & Waite 1999
; Groombridge et al. 2000
; England et al. 2003
). The loss of H
can be predicted in a cyclic fluctuating population using a more complex model (Motro & Thomson 1982
In support of these population genetic models, numerous empirical studies have reported the loss of genetic diversity following a bottleneck (England et al. 2003
), a positive relationship between diversity and population size (Frankham 1996
) and the lower diversity in insular compared with mainland populations (Frankham 1997
). This support generally comes from transversal comparative studies of populations that may be characterized by different demographic characteristics and are often interpreted in the absence of detailed historical demographic data. Such ‘snapshot’ studies may illustrate the global effect of drift on genetic diversity, but they do not capture the temporal dynamics of genetic diversity. A more appropriate way of testing for the loss of genetic diversity by genetic drift would be to use changes in H
observed over time within the same population. However, longitudinal genetic data of natural populations are very scarce. Furthermore, migration or population demography may interfere with drift making genetic changes in natural populations difficult to predict. An insular population with a well-known history therefore provides an appropriate model for testing the predictions of population genetic theory with empirical data on gene dynamics.
In 1957, one yearling male and one yearling female mouflon (Ovis aries
) were introduced onto Haute Island (6.5
; Chapuis et al. 1994
), one of the islands of the very remote Kerguelen archipelago located in the Sub-Antarctic Indian Ocean. The demographic history of the Kerguelen mouflon population is well documented from the introduction to the mid-1990s (Chapuis et al. 1994
). The two founders originated from a captive population at Vincennes Zoo (Paris, France). The Kerguelen population reached 100 individuals at the beginning of the 1970s (), and then grew exponentially to culminate with 700 individuals in 1977. Since then the population has been characterized by cyclical dynamics, fluctuating between 250 and 700 individuals (Chapuis et al. 1994
), with winter crashes occurring at a periodicity of 3–5 years after the number of individuals exceeds about 600. Crash survival is female-biased due to the costs of inter-male competition during the pre-winter rut (Boussès et al. 1994
). Given the strong founder effect, the cyclical population dynamics and the total isolation of the population, we expected very low heterozygosity in this population.
Number of individuals estimated on the Kerguelen mouflon population.
Here, we combine longitudinal genetic data and detailed information on demographic history from an island population and its ancestral source to study changes in genetic diversity over the time. Using simulations we also explore in more detail what would be the expected change of heterozygosity in the Kerguelen mouflon population under neutrality and the observed demographic history.