Hi, All.
AO is my clear preference, and is a great illustration of why I prefer head-turns towards the camera, rather than parallel to the sensor.
For me, the difference is about engagement, which is a concept dominated to such an extent by our human-to-human interactions that we respond to creatures anthropomorphically. Even where creatures have different structures to humans, e.g. having eyes on the sides of their heads, we still measure degrees of engagement in terms of human structure. A sideways-facing posture will therefore seem relatively uninterested - or even 'shifty' - compared with one in which the subject is turning to face us.
In this case, the Spoonbill can see the photographer equally well in both shots, since the head-turn in AO is not great enough to bring binocular vision into play. I this were a human subject, though, particularly since the body is angled away from the photographer, there would be a real sense that the subject is making an effort to connect with the photographer. Perhaps this is why over-the-shoulder poses are so popular in portrait photography.
Also, the combination of angled-towards-us head and angled-away-from-us body creates a curve in the horizontal plane that adds interest compared with the 'flat', 2-dimensional subject in AP.
Finally, I think Charles raised a key question about which is more important, HA relative to the bird or HA relative to the camera/viewer. I agree with Art that these combine in a situation-dependent manner but, if I had to choose one or the other, I would definitely go for HA relative to the camera/viewer. Many photographers would delete images that have a HA relative to the camera/viewer off by only one degree, but tolerances for HA relative to the bird are invariably much greater. With a great head like the one in image AO, I'd put up with almost any body angle!
Mike.