Case studies for a simulation workflow Team 3

Please add any type of power analysis, performance analysis, and null model construction that you might think would be useful to add to a simulation workflow in R

Example 1 (Sean): Power analysis type (sampling). I wanted to compare the power of different sampling strategies (different number of markers and different number of samples) to detect a bottleneck using a simple statistic (M ratio). So I wanted to, in the end, identify a sampling strategy with high power for this type of study, but not to waste sampling effort (not to sample too TOO much). I used a simulation-based power analysis (described in Hoban et al 2013 Methods in Ecol & Evol) to test the sampling strategies. I found that, for species with different sizes bottlenecks (specifically European bison and Iberian lynx), very different sampling strategies were appropriate.

Example 2 (Sean): Power analysis type/ method evaluation type. I wanted to compare the power of two bottleneck-detection methods and different sampling strategies, under different models of population decline. I found that for two different models of population decline (instantaneous and gradual), the two methods (M ratio and heterozygote excess) performed differently. I wanted to, in the end, identify a method that would work well in different situations. Overall the heterozygote excess test was more effective, but both methods were poor under gradual declines. Increasing markers showed increasing power but with somewhat diminishing returns. See Hoban et al 2013 Molecular Ecology

Example 3 (Sean): Power analysis type (sampling). I wanted to compare power of different sampling protocols, but in this case the main variable was timing of samples for temporal monitoring. Similar to example 2, I looked at different models of population decline, and compared SNPs and microsatellites (so, the question of deciding the appropriate genetic marker). See Hoban et al 2014 Evolutionary Applications

Example 4 (Sean): ABC inference. Some colleagues and I wanted to infer the timing of a hybridization event- did hybridization between species occur in recent times, in ancient times, or both recent and ancient times? Has hybridization been continuous or is a more sporadic phenomena? We simulated many scenarios of hybridization rates and timing using ABC, and matched the observed dataset to a simulated model. We found that hybridization occurred in ancient and recent times, but only during inter-glacial (non Ice Age) periods of time.

Example 5 (Sean): prediction of management outcome. I am working with a dataset of hybrids. I am considering different levels of culling (lethal removal) of animals (e.g. culling animals that are admixed above a given structure Q value). I would like to simulate what will happen to the overall population admixture (and population size) if I implement a given culling policy. It would be predicted that the higher the threshold, the more individuals will be removed and the less admixed the population will be in the future. The goal might be to establish a culling policy that results, in ten years time, a population that is only 1% admixed.