Vertical mixing model

In Appendix C, the Version 3.6 manual says, regarding Vertical Eddy Viscosity: "The formulations PARAB, NICK, and RNG are appropriate for riverine/estuarine sections in which shear due to friction is dominant. The W2 is usually the choice for reservoirs and lakes where wind shear is dominant. The k-ε turbulence formulation is though general to any waterbody and was added to W2 in order to eliminate the choice of turbulence scheme for the model user. The k- ε model though is computationally expensive and similar results can often be obtained at less computational cost using another formulation." Does this imply that the TKE model is sort of universal, and might be the preferred approach to use in general? I have characteristics of both rivers and lakes in my application. Thanks, Brian