The question can be answered from a historical point of view. It concerns the stability of the solar system, and the question whether it does have long-term stability at all. This question has exercised the minds of theoretical astronomers for a long time.
The first convincing theoretical calculations on the question were made by Laplace and Lagrange, who "demonstrated that the planets' semi-major axes undergo only small oscillations, and do not have secular terms". This was "the first major result" on the question of stability for the Solar System". See (http://www.scholarpedia.org/article/Stability_of_the_solar_system).
On that basis, the mean axes of all the planets should indeed be constant.
But later studies of the 1890s by Poincaré qualified the results of Laplace and Lagrange (see citation above, again): he showed they could not be valid over indefinitely long periods, and the stability question therefore remained open. More recent results and simulations have tended to confirm that there are possibilities for instability, though at low probability.
In the meantime, the small time-dependent component in the case of Mars now detected might be due to either of both of two possible causes.
One such cause would be long-period oscillations of the Laplace-Lagrange type which have not yet been incorporated into the theory of Mars' motion (perhaps due to asteroid perturbations, which certainly do affect Mars and have not yet all been accounted for, because there are so many of them).
The second possible cause, perhaps for part of the time-variation, might be a small non-periodic drift resulting from the effects identified by Poincaré and the simulations. (I don't know whether these latter seeming possibilities have been eliminated from consideration.)