Those meshes look ok, from what is visible in these views. There's a few tri's at the ends of the mesh sections that appear to be modified spheres, but they aren't going to cause any major problems as it doesn't look like there will be any morphing happening in these areas. Subdivision works best on quads (4 sided faces) and can go a little odd on 3 sided faces (tri's), if you convert a basic Anim8or sphere into a subdivision and look at the 'polar regions' of the sphere you'll see what I mean. In wireframe view you can see how the subdivisions create odd shapes in tri's. Anyway, the bottom line in all things graphic is that if it looks right, it is right, and I think you've done a great job with these models.
NOTE: Re the walk cycles, as you mention the body is a little rigid, which is a very common problem when animating things that are static in real life (ie. plastic toys), since it is not always obvious how things should move. One simple method you might want to consider is to twist the body at the 'waist' in the walk cycle, so the shoulders pivot one way while the hips pivot the other way. You'll need to adjust the leg motion in a further axis to stop the them swinging out during the cycle, but I suspect it would be simpler than attempting to get the haunches themselves moving. This body-twist is actually quite natural and would definitely be a part of a detailed quadruped walk cycle, I am just suggesting that in this instance it might be an effective replacement for haunch movement, which looks like it could be quite tricky to accomplish.
