Multiannual dust transport simulations have been performed using a Mars general circulation model containing a dust transport scheme which responds to changes in the atmospheric state. If the dust transport is radiatively active, the atmospheric state also responds to changes in the dust distribution. This paper examines the suspended dust distribution obtained using different lifting parameterizations, including an analysis of dust storms produced spontaneously during these simulations. The lifting mechanisms selected are lifting by (1) near-surface wind stress and (2) convective vortices known as dust devils. Each mechanism is separated into two types of parameterization: threshold-sensitive and -insensitive. The latter produce largely unrealistic annual dust cycles and storms, and no significant interannual variability. The threshold-sensitive parameterizations produce more realistic annual and interannual behavior, as well as storms with similarities to observed events, thus providing insight into how real Martian dust storms may develop. Simulations for which dust devil lifting dominates are too dusty during northern summer. This suggests either that a removal mechanism (such as dust scavenging by water ice) reduces opacities at this time or that dust devils are not the primary mechanism for storm production. Simulations for which near-surface wind stress lifting dominates produce the observed low opacities during northern spring/summer, yet appear unable to produce realistic global storms without storm decay being prevented by the occurrence of large-scale positive feedbacks on further lifting. Simulated dust levels are generally linked closely to the seasonal state of the atmosphere, and no simulation produces the observed amount of interannual variability.