Effect of selfing on the local population dynamic
Mating system is known to affect the level and organization of genetic diversity in populations and thus influences their ability to respond to selection. A large number of plant species are preferentially self-fertilizers. This mode of reproduction is associated with high levels of homozygosity and reduced pollen flow which leads to decreased effective population sizes compared to outcrossing species. Even though, a certain number of theoretical expectations have been verified experimentally, our fine understanding of the local dynamics of genetic diversity within population is still limited. The purpose of my PhD thesis was to give insights into the processes at work at the within-population scale in wild populations of the legume species Medicago truncatula. This has been done using two kinds of approaches, namely population and quantitative genetics approaches.
In the first part of my PhD work, I studied the fine-scale population structure of a population of Medicago truncatula using microsatellite markers. My work outlined a characteristic organization of highly selfing populations. These are composed by a set of inbred lines, each of which can be represented by several individuals, more or less dispersed throughout the population. The occurrence of some individuals exhibiting some intermediates multilocus genotypes between the most frequent lines of the population is a strong signature of the rare outcrossing events. These recombinant genotypes indeed represent some more or less advanced stages of what we call ‘recombinant inbred lines’ in plant breeding. The recurrent production of such recombinant genotypes in highly selfing populations may be of great potential adaptive significance, although this remains to be tested formally.
The magnitude of Ne determines the amount of gene sampling errors between generations that causes genetic drift and thus allows predictions about the rate of genetic differentiation between populations as well as inferences about the relative effects of genetic drift and selection at nonneutral loci. The second part of my work has been to estimate the effective size of two populations of Medicago truncatula using temporal variation in allele frequencies. The main result of this work was that the low value of the N e/N ratio was at least partly accounted for by the large variance in reproductive success. Furthermore, in one population, this variation in reproductive success seems to be heritable. Another feature is that the major part of the variation in allele frequencies is related to the line dynamic itself. Therefore, it is likely, that the estimated N e values estimated reflect a combination of neutral and selective processes.
The third part of my work has been devoted to the estimation of quantitative genetic parameters in wild populations under natural conditions. Such estimations are crucial to our understanding of the evolution of natural populations, since the adaptive potential depends primarily on the balance between genetic over environmental variation. The central idea of this work was to take advantage of the structure just described – namely the co-occurrence of several inbred lines each of which can be represented by several individuals. This offers a unique opportunity of partitioning the phenotypic variance into its causal components in a natural setting without the need of generating families. The main problem to be faced is the non-random distribution of different lines in the population. To alleviate this problem, I implanted all over the population three “control” lines whose phenotypic values were used as environmental covariates. Although the results of this study are somewhat exploratory, they show that this approach is very promising for future studies in highly selfing plant species.
Evolutionary genetics of transitions from outcrossing to selfing
Shifts in mating systems constitute major evolutionary transitions. Among these shifts, one of the most prominent is the transition from outcrossing to selfing which has occurred in numerous lineages of vascular plants. I intend to examine the consequences of the transition from outcrossing to selfing at the genomic level using the species Eichhornia paniculata. Particularly, because of multiple transitions to selfing in this species and the occurrence of both monostylous selfing populations and di or tristylous outcrossing populations, there is an outstanding opportunity to investigate the replicated effect of this key transition at the genomic level. Further, it provides a particularly sound model, since it does not suffer from the flaws inherent to studies comparing related outcrossing and selfing species (like very different life histories).
Transitions from outcrossing to selfing are expected to reduce the effective population size. The increased homozygosity reduces the effective recombination rate thus exaggerating the effects of hitchhiking and background selection at neutral sites flanking regions under selection. Furthermore, because of their capacity of establish new populations from a very small number of founders (technically even one is enough), self-fertilizing species are said to be more prone to recurrent bottlenecks. One thus expects an increased drift in selfing populations compared to outcrossing populations and a comparatively lower genetic diversity.
Furthermore, the increased drift combined to the reduced recombination efficiency are expected to cause an increase in fixation rates of slightly deleterious mutations and a decrease in fixation rates of advantageous mutations (since the efficiency of selection depends primarily on the product N es). We thus expect to detect this relaxed selection at the molecular level.
Evidence from genomic data collected on selfing species and related outcrossing species confirms the loss of diversity in selfing lineages. However to date, studies aimed at detecting a reduced selection efficiency among selfing lineages yielded inconclusive results, some detecting a small effect whereas others do not. So I intend to test these predictions using multilocus DNA sequence data in collaboration with Rob Ness and Dr Stephen Wright.