Help if you can!

I have this problem where I have been given three sounds, two of which I have found to be separate phonemes. However both of these sounds are in complimentary distribution with the third sound when found in the environment: __V. How do I decide which one of the two phonemes is more likely to be allophone with the third sound.
This is the distribution table I have drawn up:
Note: [x] and [?] are the phonemes.
| ?
| i
| ?
| a
| u
| o
|
[x]
| ?
| ?
| -
| ?
| -
| -
|
[?]
| ?
| ?
| ?
| ?
| -
| -
|
[?]
| -
| -
| -
| -
| ?
| ?
|
Is it plausible for me to say the [?] and [?] are allophones of the sound [?] purely based on the fact that there is more evidence for it in the given data?
Thanks for any help you can give me!