Allophones or Separate Phonemes?

I thought I had solved all my orthographical problems when I decided
to transcribe Boreanesian in the IPA for the grammar I'm preparing.
But I just discovered a problem. There are a group of sounds I'm not
sure of how to transcribe. They all have that in common that they
evolved from the voiced and voiceless approximants of the
Proto-language, but both their similarity/dissimilarity to each
other and their distrubution is such that I can't figure out whether
they are allophones of the same phoneme or different phonemes
altogether.

One pair of approximantal proto-phonemes are the voiced and
voiceless pharyngeal approximants. Both can be represented in the
IPA by a turned m with a long right leg, the voiceless one marked by
an under-ring. In this post I'll have to represent them radically as
*R and *H respectively.

In the Boreanesian dialect I'm working on, syllable initial *R is
now realized as [?], while syllable initial *H is now realized as
[h]. In syllable final position, *R and *H are now respectively
realized as creaky voiced and voiceless [@] after [i] and [u], and
as creaky voiced and voiceless vowel length [:] after [@] and [a].
(Note that I have only used brackets [] as I'm not sure whether they
are allophones or separate phonemes in the modern language). These
are summarized in the table a) below:

a)
PROTO-B.  ->  MODERN-B.
  *Ra-    ->    [?a-]
  *R@-    ->    [?@-]
  *Ri-    ->    [?i-]
  *Ru-    ->    [?u-]
  *-aR    ->    [-a:?]
  *-@R    ->    [-@:?]
  *-iR    ->    [-i@?]
  *-uR    ->    [-u@?]

  *Ha-    ->    [ha-]
  *H@-    ->    [h@-]
  *Hi-    ->    [hi-]
  *Hu-    ->    [hu-]
  *-aH    ->    [-a:h]
  *-@H    ->    [-@:h]
  *-iH    ->    [-i@h]
  *-uH    ->    [-u@h]

As of now, I have been transcribing the sounds that evolved from
syllable-final *R and *H by IPA's turned-m-with-a-right-long-leg in
the grammar. The difference in phonation is marked by following this
symbol by either a glottal stop [?] for creaky voiced or [h] for
voiceless. In doing so, I have assumed that the various syllable
final sounds that evolved from *R and *H are allophones of each
other. The new syllable-initial sounds are simply [?] or [h], and
are represented likewise. In doing so, the
turned-m-with-a-right-long-leg only occurs in syllable final
position, while [?] and [h] can occur in both syllable initial
position and syllable final position.

The fact that all the new sounds that derive from *R and *H have [?]
and [h] respectively prompted me to wonder if the new sounds are in
fact allophones of each other. I mean, [?-] [-:?] and [-@?] all
could be allophones of the same phoneme, while [h-] [-:h] and [-@h]
all could be allophones of another. But on the other hand, one could
say that [-:h] and [-@h] are allophones of an approximant that
occurs only in syllable-final position while [h-] is an independant
phoneme on its own right since it is not an approximant.

If indeed they are allophones of each other, then it would be
difficult justify (for me at least) a similar conclusion for the new
sounds that derived from the other approximants of Proto-B. There
are similar diachronic sound changes that have occured to the other
approximant phonemes of Proto-B. For instance, table b) below
summarizes how *j (voiced palatal approximant) and *hj/jh (voiceless
palatal approximant) evolved in similar environments to table a):

b)
PROTO-B.  ->  MODERN-B.
  *ja-    ->    [ja-]
  *j@-    ->    [j@-]
  *ji-    ->    [ji-]
  *ju-    ->    [ju-]
  *-aj    ->    [-aj?]
  *-@j    ->    [-@j?]
  *-ij    ->    [-i:?]
  *-uj    ->    [-uj?]

  *hja-    ->    [sa-]
  *hj@-    ->    [s@-]
  *hji-    ->    [si-]
  *hju-    ->    [su-]
  *-ajh    ->    [-ajh]
  *-@jh    ->    [-@jh]
  *-ijh    ->    [-i:h]
  *-ujh    ->    [-ujh]

One could say for instance that [s-] [-:h] and [-jh] are all
allophones of the same phoneme. But somehow, it seems more
appropriate to assume that the new sound [s] is indeed a separate
phoneme since it bares little resemblance to either [-:h] or [-jh].
Like [?] and [h] in syllable-initial position, [s] is not an
approximant. So if [s] is indeed a separate phoneme from [-:h] and
[-jh], then [h] must be a separate phoneme from [-:h] and [-@h],
right? Hey wait a sec, what about [-:h]??!! Is it an allophone with
both [s][-jh] and [h][-@h]? Oh this is sooo confusing!!

Comments, please!

-kristian- 8-)