I've tried searching for literature references that explain this, finding only a single reference which refers to the two signals as originating from phosphine environments cis/trans to carbonyls, however I'm having trouble rationalising why this occurs.
In this complex there are two different 31P environments which are not related by symmetry:
The two green phosphorus nuclei can be interconverted by a $C_2$ rotation (the rotation axis bisects the OC–Mo–CO angle), and so can the purple ones, but green and purple cannot be interconverted.
As extra proof, consider that the green P is cis to both carbonyl ligands whereas the purple P is cis to one and trans to the other.
They therefore have different chemical shifts and in the spectrum you would expect to see two different peaks. Presumably they would show coupling to each other, so I would expect two triplets, if we ignore satellites from all other nuclei (e.g. 13C).