It's really fairly simple: there's a direct trade-off between color accuracy and apparent noise.

To sense color, the sensor has a filter in front of each sensel. As it happens, the filter for the red channel cuts out the most light. This means when you're doing the conversion, you have to multiply the red channel by the largest factor1 to achieve a particular color balance. When you multiply the data, however, you're increasing the apparent noise right along with the intended data.

As such, to minimize apparent noise, the raw converter software wants to use the weakest multiplier on the red channel that it can get away with.

It looks to me like Aperture is simply using a somewhat smaller multiplier on the red channel to reduce the appearance of noise. ACR apparently uses a slightly larger multiplier to give more accurate reds. Though it appears unlikely to matter much in this particular picture, it's probably fair to guess that when you have pictures taken in relatively low light that Aperture will do a slightly better job of suppressing noise without losing detail.

1 In case you care about the details, the green channel is normally the "baseline" so we'll treat its factor as 1.0. The blue channel is multiplied by a factor of around 1.3 to 1.4, and the red by a factor of about 2 to 2.3. This varies with the camera though -- Sony, for one example, uses much stronger red filters than Canon, so their cameras reproduce saturated reds more easily, but also do relatively poorly in noise tests.