High Reynolds number wall-bounded turbulent flows subject to buoyancy forces are fraught with complex dynamics originating from the interplay between shear generation of turbulence (
$S$
) and its production or destruction by density gradients (
$B$
). For horizontal walls,
$S$
augments the energy budget of the streamwise fluctuations, while
$B$
influences the energy contained in the vertical fluctuations. Yet, return to isotropy remains a tendency of such flows where pressure–strain interaction redistributes turbulent energy among all three velocity components and thus limits, but cannot fully eliminate, the anisotropy of the velocity fluctuations. A reduced model of this energy redistribution in the inertial (logarithmic) sublayer, with no tuneable constants, is introduced and tested against large eddy and direct numerical simulations under both stable (
$B<0$
) and unstable (
$B>0$
) conditions. The model links key transitions in turbulence statistics with flux Richardson number (at
$Ri_{f}=-B/S\approx$
$-2$
,
$-1$
and
$-0.5$
) to shifts in the direction of energy redistribution. Furthermore, when coupled to a linear Rotta-type closure, an extended version of the model can predict individual variance components, as well as the degree of turbulence anisotropy. The extended model indicates a regime transition under stable conditions when
$Ri_{f}$
approaches
$Ri_{f,max}\approx +0.21$
. Buoyant destruction
$B$
increases with increasing stabilizing density gradients when
$Ri_{f}<Ri_{f,max}$
, while at
$Ri_{f}\geqslant Ri_{f,max}$
limitations on the redistribution into the vertical component throttle the highest attainable rate of buoyant destruction, explaining the ‘self-preservation’ of turbulence at large positive gradient Richardson numbers. Despite adopting a ‘framework of maximum simplicity’, the model results in novel and insightful findings on how the interacting roles of energy redistribution and buoyancy modulate the variance budgets and the energy exchange among the components.