En savoir plus

Pat Monaghan opened the series of keynote speeches with questions about the effect of early life stress on longevity and overall fitness. A very interesting walk through ecology of many bird species, ending with the significance of telomere length regulation. A truly multidisciplinary talk.

Conservation geneticist Richard Frankham discussed the 50/500 rule of thumb which he claims to need revision based on recent data: in some cases, an effective population size of over a million may be needed to guarantee viability. Census size may be 8-fold larger...

Florence Débarre asked how to describe population dynamics through dynamically sufficient variables. One way to achieve this is through the "weak selection approximation", e.g. for small distances in phenotype space. Nice work, it will be useful for us to study divergence at small geographical distances!

In Brassica rapa, Léa Frachon showed that bumblebee pollination induces selection for larger plant size in a nicely crafted experimental evolution study. In addition, she could show that this was associated to significant genomic change.

The first talk I've heard about multigenic selection (surely something I hoped to hear about) was from Kati Csilléry, who developed on the impact of epistatic interactions on the rate of approach to a phenotypic optimum. She showed that the optimum is approached faster with epistasis, at the price of lowering the power of outlier detection tests for stabilising selection. On the contrary, power is good for divergence outlier tests. 

On a more "technical" level, we heard about Ash dieback genomic predictions from Carey Metheringham. Only 3% of the trees are resistant, but heritability is high (0.37 - 0.52). With 9 million SNPs, accuracy of assignment is at 60-70%.

In one of the many outstanding plenary talks, Orly Razgour showed that rapid adaptation to arid environments in bats can occur both through selection at SNP loci and through changes in the gut microbiota. This reminds us that symbionts should not be forgotten when analysis adaptation.

Sinead Collins, another plenary speaker, looked for an explanation of the change in the direction of selection observed so often. 

This is due, in her study system (phytoplankton), to the complex interaction of growth rates, toxin production and the emergence of resistance to toxins. 

In a rare study of short-term evolution in natural populations with evolutionary inferences, Mijke Var Der Zee studied the demographic history & rapid adaptation in guppy populations. She used some basic statistics (delta allele frequencies, Fst outliers, and delta(pi)) as signatures of selection.

Back to multigenic selection, Nick Barton looked at simulations of evolution of traits due to optimum shifts. He observed that, even though major-effect genes vary quickly at the beginning, then they are pushed back by underdominance, and most of the action occurs at small-effect genes, until, a very long time afterwards (more than 10k generations), new substitutions kick in. He (quite sadly) concluded that multigenic selection may just plainly be undetectable.

On the other extreme of the spectrum, Célia Neto found that adaptation of A. thaliana in Cape Verde islands is tightly linked to variation in a single gene (FLC) controlling flowering time. For life-history traits, things are more polygenic (GWAS spotted 40-55 SNPs). 

In an overcrowed room (the overflow room was also overcrowded...) Jon Slate showed his latest results on the famous Soay sheep population. this time, it is about genomic prediction compared to pedigree-based prediction of traits. Method used was BayesR, and he showed that there was a good correlation (r = 0.61) between the two estimates; genetic architecture of traits did not make a big difference.

Benoit Pujol (with whom I published a paper on the same topic last year) admits that theoretical equations on the rate of phenotypic change based on the intensity of selection do not work. So where is all evolution gone? No answer to this question, but research is ongoing in wild populations of both animal and plants. Yet Maria Clara Castellanos found that invasive Digitalis populations have evolved for flower size under pressure from a shift in pollinators (apparently this works well! see Léa Frachon above...).

GenTree fellow Juanjo Robledo-Arnucio presented methods for the analysis of results of viability selection. Type I good, but power moderate to low, except for very large samples. See the SGSC software. 

Local star scientist Anna-Liisa Laine described a very neat and detailed study of how connectivity helps populations to resist pathogens in a Plantago / mildew metapopulation system of the Aland archipelago (just out in front of Turku, between Finland and Sweden). Beautiful and detailed study, in which host connectivity suppresses pathogen diversity.

Back to A. thaliana and salmons (shiploads of that, too, given the Northen location fo the conference) we heard more genome scans, and at this point we stared being vaguely weary.

I'll finish with the delightful plenary talk by Rasmus Nielsen, who, besides having - as usual - plenty of nice stories of human evolution to tell us, declared that "demography is a nuisance" (from the point of view of the study of selection...). And he also presented some nice methodological advances, such as: a nice multi-pop demographic model (Ohana, Cheng et al. (2017) Bioinformatics 15: 2148) and methods to circumvent the complexities of topological search in coalescent models (Rasmussen et al. PLoS Genetics; Stern et al. BioRxiv 592675). 

Well, it was a very good edition; perhaps too many talks replicated the genome-scan-GWAS scenario, and there were too many bad news about our (lack of) capacity to detect polygenic selection. Maybe we will be luckier next time...