I would ignore our(scientific) definition of species for pokemon. They aren't all subspecies. Being Pokemon different rules apply-specifically that all so far there have been no hybrids. Personally if they add hybrids it should be for similar pokemon like Ekans/Seviper except that they are different types.
I know that I probably shouldn't try and apply real-world biology to Pokemon, but...
In a sense there already ARE hybrids. Pokemon which know egg moves. In Pokemon, we observe that (except when Ditto or the Nidos are involved), the morphology (physical shape) is always that of the mother. However, certain instinctive behaviours (egg moves) are inherited from the father. This must mean that in the Pokemon reproductive process, the genes that control morphology - which is probably the majority - are always inherited from the mother, while the genes that control instinctive behaviour are inherited from the father.
Pokemon thus probably wouldn't have paired chromosomes like most real animals. Assuming Pokemon have chromosomes in the first, I think they'd be a combination of haploid and diploid. p haploid chromosomes from the mother determine morphology, q haploid from the father determine egg moves, and r diploid chromosomes determine IVs. In reproduction, the female gamete contains none of the 'q' chromosomes, while the male gamete contains none of the 'p', and the 'r' chromosomes are probably recombined like in mammalian meiosis.
Maybe I'm building castles in the sky here...but hey, I for one find it interesting.
To try and pull back on topic - the postulated behaviour leads to an interesting possibility for gen 5 - egg moves that skip generations. A move could be unusable by a Pokemon, but it could still have the 'gene' for that move, and (if male) pass it on. That could be used as a route to get previously impossible egg move combinations. It would however require an overhaul of the data structure. It might fit in with the idea of having Pokemon know all their moves, but take four into battle - being able to change which four between battles. The 'generation skipping' egg moves would then simply have a bit set to hide them, but come into play in the breeding.
EDIT:Another thing that the hypothesis genetics is in accord with is the fact that tutor moves are not inherited, but TM moves are. A tutor merely teaches a Pokemon the move, but a TM could actually affect the Pokemon's genes.
I'm still not sure how the fact that level up moves are only inherited if both parents know them can be figured out though.