I think you need to be very clear why you are undertaking the conservation program. Are you doing it to ensure the species is not lost from the hobby or are you doing it to possibly release them back into their natural habitats? The way I see it you can't have it both ways.
If your goal is to prevent a species from being lost to the hobby then the idea is to follow the guidelines set down by Juan (BTW Juan - long time no see - nice to see you again) and get a group of like minded hobbyists around you who are also dedicated to preserving the species and once you breed them spread them around so that if something happens to your fish the species isn't doomed because you have them safely located in many places. Only the strongest individuals need be bred and culling can take place to ensure quality can be maintained. Much like any serious fish breeder would do anyway.
On the other hand, if your goal was to carry out a breeding program to restock habitats from which the species has disappeared and was
found in naturally, then that is a different kettle of fish so to speak
The best way to do that would be to select your starting pair or pairs and to place them in a very large pond or even a dam and allow them to breed unrestricted to choose their own mates and allow natural selection to play a part in determining who gets to breed and who gets selected against and therefore removed from the gene pool. This helps to overcome something called a bottleneck where populations originating from a small genetic base experience reduced genetic variability and become genetically homogenous. By allowing large scale and unrestricted breeding you are allowing all the genes the fish has to be, in effect, reshuffled into millions of different combinations and in doing so create genetic variability.
To illustrate this this point take humans, who have 23 pairs of chromosomes. These are sorted randomly when eggs and sperm are produced. The number of different combinations possible from a single man or woman is 8,388,608 different genetic combinations. And that is from a single person. Now when you take into account fertilisation still has to occur the probability of any two individuals having the same genetic makeup sky rockets to 1 in 64 trillion. In fish, where chromosome numbers vary from that of humans, you still have a similar pattern occuring AND if this is allowed to occur in tandem with random mating with the added selective pressures involved in finding a mate and surviving in the open for several generations etc then you could reasonably expect to create stock that is more suited to being released, with a 'deep' gene pool, and in numbers that will ensure the survival of the species in the wild.
It has been shown that when breeding is allowed to take place unrestricted in confined waterways, the population often appears to become stunted so tests would have to be done to ensure the mean size of the fish did not decrease and when it showed signs of this a harvest and release would have to occur.
Anyway, just food for thought or lively discussion lol
ciao - Simon