On the Use of the Jackknife to Compare R_XY Statistics in Conservation Genomics Research

Many useful quantities in conservation genomics are point estimates that summarise genetic variation at the population or species level. This presents a challenge to the application of null hypothesis testing frameworks, which require some measure of uncertainty to test the authenticity of measured group differences. The jackknife is a convenient statistical resampling tool which sees regular use to this end: producing confidence intervals for point estimates. Unfortunately, the slightly obscure nature of the jackknife coupled with its superficial resemblance to the bootstrap has led to its misuse, specifically in the analysis of R_XY statistics, a recently formulated quantity designed to measure relative mutation load. In this review, I summarise the application of the jackknife as a method to produce confidence intervals and review its uses in the conservation/population genomics literature. I outline how it is similar to and different from the bootstrap, and how the treatment of jackknife pseudo-values as bootstrap pseudo-values produces inflated confidence in otherwise negligible group differences. I then provide suggestions for the future use of the jackknife in conservation genomics.