Information and allometry

Many life-history parameters have condition-dependent optima, but individuals are often required to set the values of such parameters relatively early in development, before the relevant conditions can be assessed with full accuracy. If cues are available that predict such future conditions, then the condition-dependent parameter should evolve to assume values that deviate from the mean in the direction implied by the cues, but these deviations should regress towards the mean to the degree that the cues are less than fully reliable. Under mild assumptions, the slopes of the resulting relationships between condition-dependent life-history parameters and the variable conditions on which their optima depend will be the 'ideal' slopes (those that would maximize fitness if the parameter could be chosen on the basis of full information) devalued by the squared correlation between the condition and the parameter.