Modeling the demographic history of Drosophila melanogaster using Approximate Bayesian Computation and Next Generation Sequencing Data

The main goal of this thesis was to develop demographic models of
the fruit fly Drosophila melanogaster using Approximate Bayesian
Computation and Next Generation Sequencing Data. These models were
used to reconstruct the history of African, European, and North
American populations. Chapter 1 deals with the demographic history
of North American D. melanogaster. This project was motivated by
the release of full-genome sequences of a North American
population, which showed greater diversity than European D.
melanogaster although the introduction of the fruit fly to North
America dates back to only �200 years ago. Here, we tested di�erent
demographic models involving populations of Zimbabwe, The
Netherlands, and North Carolina (North America). Among the tested
models we included variants with and without migration, as well as
a model involving admixture between the population of Africa and
Europe that generated the population of North America. We found
that the admixture model �ts best the observed data and we
estimated the proportion of European and African admixture in the
North American population. This population has 85% European and 15%
African ancestry. We also estimated other population parameters
including population sizes (current and ancestral) and divergence
times. Con�cerning previous studies we also estimated the
divergence between African and European populations to be around
19,000 years ago. Chapter 2 deals with gene flow of D. melanogaster
between African and European populations. Gene flow in D.
melanogaster is well acknowledged but has not been quanti�ed using
DNA sequence data. Previous studies from the late 80's based on
allozymes found that the number of migrants per generation (Nm) was
around 2 between several populations distributed worldwide. Here we
used ABC methods and full-genome sequences to estimate the rate of
migration between a population from Rwanda in Africa and a
population from France. We found that Nm is around 10, which may
imply there was a signi�cant increase of gene flow in the last few
decades. Our estimates show that the migration rate between these
two populations is not necessarily symmetrical, with migration from
Europe to Africa being higher than the opposite, although the
di�erence does not seem to be significant. The study of gene flow
is relevant because it constitutes an important force in population
genetics. Theoretical studies have shown that, under neutrality, it
is enough to have one migrant per generation to stop two
populations from diverging and speciating, and if migration is
strong enough it can also overcome the e�ect of selection. Chapter
3 focuses on the sequencing of 130 full genomes of D. melanogaster
from Africa and 9 from France. This project made use of haploid
embryos, a new technique introduced in 2011 that allows the
development of haploid D. melanogaster, which is then used for
sequencing. The main goal of this project was to characterize these
populations in terms of their diversity, admixture, and
di�erentiation. We found that the most diverse population comes
from Zambia, which is now thought to be much closer to D.
melanogaster 's center of origin. We also found a signi�cant amount
of non-cosmopolitan admixture in several African populations,
meaning that there exists a signi�cant amount of back migration
from Europe to Africa (corroborating the fi�ndings of chapter 2).
In order to identify admixture tracts a new method was developed
for this purpose, which uses a hidden Markov model to locate
admixed regions along the genome. Admixed regions, as well as
regions showing high levels of identity by descent were masked for
downstream population genetics analyses. These full genomes
constitute the second e�ort of the Drosophila Population Genomics
Project (DPGP 2) and are now available for the scienti�c community.