Trait covariation implies the existence of shared genetic mediation by common polymorphisms. It is therefore expected that some strains will be extreme on a set of correlated measures, and as a group, may be useful as a starting point in selected breeding or as research models for examination of the biological substrates of extreme phenotypic deviation. The Mahalanobis distance is one such measure that can be used to detect multi-variate outliers. For example, we have detected extreme strains for composite traits spanning measures of anxiety and stress, namely, percent time spent in open quadrant (μ = 17.92 ± 8.84), percent entries in open arms (μ = 37.69 ± 12.18), percent time in center of the open field (μ = 15.75 ± 6.21) and percent time in light in light–dark box (μ = 33.64 ± 8.37). For the multi-variate set of traits analyzed, BXD 1 (high for percent time in center of open field and percent entries in open arms, intermediate for percent time spent in open quadrant, low for percent time in light in light–dark box), BXD 11 (high for
