| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 6.69 MB | Adobe PDF | |||
| 9.86 KB | Internal XML |
Authors
Advisor(s)
Abstract(s)
Reconstructing the demographic history of populations with genetic data from present-day samples is a challenging inference problem. The general aim of this thesis is to determine whether major demographic events can be detected, quantified and dated using model-based inference approaches. The emphasis is on the study of admixture events and population size changes, which are relevant for conservation biology. Approximate Bayesian computation (ABC) methods were developed to make inference under models involving admixture. These methods were first implemented into a general and relatively simple model and later improved to deal with up to four populations and two admixture events. An ABC approach based onallele-frequencies was tested and compared in detail with a full-likelihood methods. Several aspects of theABC methodology were investigated in a simulation study, such as the choice of summary statistics anddistance metrics. The estimates obtained with the ABC approximated well the full-likelihood. Moreover, amodel choice procedure was developed to assess the relative probability of alternative admixture and populationsplit models. The results indicate that the ABC approach is able to identify with high probability thecorrect model. These methods have been implemented in a user-friendly software.The effect of population structure on estimates of population size change was also investigated. A simulationstudy was performed to assess the robustness of full-likelihood methods to deviations due to populationstructure. The results show a clear effect of population structure, leading to the detection of spurious bottlenecks,which depends on the sampling scheme and is stronger with limited gene-flow level, and higherscaled mutation rate.The methods developed and investigated here were applied to study two critically endangered freshwaterfish species, Iberochondrostoma lusitanicum and I. almacai. Results suggest that both species were highlystructured, and suffered recent population declines. The re-analysis of I. lusitanicum data with the ABCmethod developed suggested that the potential admixture events were likely due to shared polymorphism.Regarding the bottleneck signatures, the results suggest that the observed data cannot be explained by thepopulation structure alone, indicating that these species are undergoing a population decline.Overall, the results of this thesis may contribute to a better understanding of the potential and limitations of model-based inference methods using genetic data.
Description
Tese de doutoramento, Biologia (Biologia Evolutiva), 2010, Universidade de Lisboa, Faculdade de Ciências
Keywords
Estrutura populacional Mistura de populações Declínio populacional Approximate Bayesian computation Genética da conservação Iberochondrostoma Teses de doutoramento