Proxies for census size are correlated with the estimated impact of selection on neutral diversity.

<p>For each species, we obtained estimates of size (in meters) and range (in square kilometers), and used those as predictors in a linear model with a measure of the impact of selection on neutral diversity as the response (see main text and <a href="http://www.plosbiology.org/article/info:doi/10.1371/journal.pbio.1002112#pbio.1002112.t002" target="_blank">Table 2</a> for full model information). Both range (part A) and size (part B) are significant predictors of the impact of selection on neutral diversity in the expected directions. Points are colored by kingdom (blue = animals, green = plants), and regression lines estimated independently for plants and animals are shown. Raw data underlying this figure can be found in <a href="http://www.plosbiology.org/article/info:doi/10.1371/journal.pbio.1002112#pbio.1002112.s002" target="_blank">S2 Data</a>. C) Robustness of our model fit. We tested our main model (see text and <a href="http://www.plosbiology.org/article/info:doi/10.1371/journal.pbio.1002112#pbio.1002112.t002" target="_blank">Table 2</a>) across a wide range of different analysis options, including different filtering options, different window sizes, and different population genetic parameters. Each point represents the adjusted R² of the full model for one set of parameter values, colored according to <i>p</i>-value. R code to replicate this analysis is available at: <a href="https://github.com/tsackton/linked-selection/blob/master/final_analysis/linear_models.R" target="_blank">https://github.com/tsackton/linked-selection/blob/master/final_analysis/linear_models.R</a>.