The major difference is that the ferrous materials are harder and more brittle. Non-ferrous materials have a greater plasticity than ferrous materials. Another reason is related to the structure of the material, which significantly affects the fatigue life.

I believe your question is about why steels show a horizontal S-N curve (called the fatigue limit) while the non-ferrous alloys show a continuously decreasing S-N curve (called endurance limit).

This is something to do with "strain ageing or age hardening" behaviour of a material. Refer to some standard book that deals with metallurgical aspects of deformation. In brief, strain ageing is a process in which dislocation will be pinned down by solute atoms thereby restricting its movement. If dislocation movement is restricted, deformation does not take place and you require higher stresses to free these dislocations and restore the deformation. In other words, we can say that "strain ageing" increases the flow stress  of the material. After every fatigue cycle, you are therefore dealing with a situation as if you are testing a material whose yield stress is different (read increased). When your applied stress amplitude is below this enhanced stress, obviously, it is not sufficient to create enough damage leading to failure. That is why you get a horizontal life, as if the material possesses an infinite fatigue life. You call this as fatigue limit (meaning no fatigue damage possible below this stress level!)

Non-strain ageing materials (those whose flow stress does not change with time) do not enjoy such an advantage. For every magnitude of applied stress, there will be some fatigue damage (magnitude of which decrease with decrease in applied stress level) and you get a continuously decreasing S-N curve. Such materials, do not have a definite fatigue limit but you can assign an endurance limit (meaning how long will it take to withstand a particular level of fatigue damage).

The answer to your question therefore is majority of the steels are strain ageing type and majority of non-ferrous alloys are non-strain ageing type.

Hope that helps 

S. Sivaprasad ,thanks a lot for your detailed answer

Thanks Sivaprasad Sir for this detailed answer. Now I also have a question that can we co relate this with some general equations? For example with Basquin's law?

In materials, the motion of dislocations is a discontinuous process. When dislocation meets obstacles (like forest dislocations) they are temporary arrested for a certain time. During this time solutes (such as interstitial particles) diffuse around the dislocations further strengthening the obstacles held on the dislocations. Eventually these dislocations will overcome these obstacles with sufficient stress and will quickly move to the next obstacle where they are stopped and the process can repeat. This process's most well-known manifestations are Lüders bands and the Portevin–Le Chatelier effect. Though the mechanism is known to affect materials without these physical observations

The dislocation movement applies to both ferrous and nonferrous materials. There are also other causes of accumulation of defects (cracks growth) not related to dislocations.

As for the S-N curves, stress and strain influence the behavior of the graph. In the range of large plastic deformations, the strain has a greater impact than stress.