Fig. 2

Illustration of how different patterns of GxA lead to different patterns of additive genetic variance in behavior and cross-age correlations. For simplicity, changes in the breeding values with age are depicted as lines, but qualitatively the same patterns may arise under different functional forms. For 30 genotypes, the reaction norm elevation (defined here as the expected behavior at age 3) and the plasticity (slope of how the deviation relative to age-specific mean behavior is expected to change across ages) were randomly drawn from a multivariate normal distribution. Panel (a), (b) and (c) denote the situation where reaction norm elevation and the plasticity are correlated (variance in elevation = 4, variance in slope = 1, r = 0.65), which leads to (a) reaction norms mostly “fanning out”, producing (b) an increase in additive genetic variance with age, and (c) a positive genetic correlation between behaviors at age 2 and 4. Panels (d), (e), and (f) display when reaction norm elevation and slope are not correlated (variance in elevation = 1, variance in slope = 4, r = 0), which leads to (a) crossing reaction norms which (e) lead to a non-linear pattern in additive genetic variance over ages and (f) a negative genetic correlation between behavior at age 2 and 4.