Standardizing Selection Strengths to Study Selection in the Wild: A Critical Comparison and Suggestions for the Future

We critically review the main approaches for standardizing and comparing selection differentials and gradients among traits, populations, and species and assess their differential merits. In particular, we explain why the most widespread approach to standardizing selection, which measures selection differentials in units of a trait's phenotypic standard deviation, is less appropriate for characterizing the strengths of natural selection in wild populations. In contrast, an alternative standardization approach that uses a trait's phenotypic mean in addition to its standard deviation results in another dimensionless measure: the mean-standardized selection gradient. This measure offers some key benefits over alternative approaches, such as ease of interpretation, independence of a trait's variance, and the natural interplay of ecological and evolutionary dynamics, and yet it remains rarely used by evolutionary biologists. We explain how the more routine application of this measure will facilitate comparisons of selection strengths in the wild among traits, populations, and species.