Charles Darwin's The Origin of Species (sixth edition) does not explain how systems in organisms (such as for reproduction) could have come about through evolution. However, an excellent model for exploring this matter is the genus coryanthes (bucket orchid), which Darwin characterizes as an "extraordinary adaptation."
I will denote as "pre-cory," coryanthes' most recent ancestor genus that had not yet begun forming the reproductive structures that would ultimately become those of coryanthes. For pre-cory to have evolved to coryanthes means that the former gave rise to a series of descendent varieties, each possibly many generations removed from its nearest ancestor variety, whose structures reflected gradual change until they became those of coryanthes. Because these structures would be unable to carry out reproduction until they had fully formed, the descendent varieties in this series would need to have retained the pre-cory reproductive system.
Darwin states that a given generation of a particular variety is highly unlikely to yield a new variety, which implies a negligible probability that a new variety would differ in more than one respect from its nearest ancestor variety. This and the previous discussion imply that in the first descendent variety of pre-cory that would have a fully formed and functioning coryanthes-type reproductive system, which variety I will call mezzo-cory, the pre-cory reproductive system would have remained intact. For mezzo-cory to evolve to coryanthes would entail the gradual receding of the pre-cory reproductive system, thereby leaving only the coryanthes-type reproductive system.
Let's assume that the evolutionary chain from pre-cory to mezzo-cory comprised 100 distinct varieties, which doesn't seem unreasonably high, given the elaborate structure of coryanthes. The first of these 100 varieties will be denoted here as pre-cory-1. During the second step in this evolutionary path, pre-cory would give rise to a new variety I'll denote as pre-cory-2 (to distinguish it from pre-cory-1: the first "child" variety of pre-cory), and pre-cory-1 would give rise to a new variety I'll denote as pre-cory-1a. In the third step, pre-cory-1a would give rise to pre-cory-1aa, pre-cory-2 to pre-cory-2a, pre-cory-1 to pre-cory-1b (being that pre-cory-1a was used to denote the first "child" variety of pre-cory-1), and pre-cory to pre-cory-3. These steps are depicted in the diagram below.
START STEP ONE STEP TWO STEP THREE |- pre-cory | |- pre-cory ----| | | | |- pre-cory-3 | |- pre-cory ---| | | | | |- pre-cory-2 | | | | |- pre-cory-2 --| | | | |- pre-cory-2a | pre-cory -| | | |- pre-cory-1 | | | |- pre-cory-1 --| | | | | | |- pre-cory-1b | | |- pre-cory-1 -| | | |- pre-cory-1a | | |- pre-cory-1a -| | |- pre-cory-1aa
Darwin notes that a variety that has a partially formed adaptive structure that does not benefit the plant, will be at a competitive disadvantage because the structure will nonetheless require resources to form. This implies that of the varieties in the above chart, pre-cory would be the fittest to survive; pre-cory-1, pre-cory-2, and pre-cory-3 would be the next-fittest; pre-cory-1a, pre-cory-1b, and pre-cory-2a would be still less fit; and pre-cory-1aa would be the least fit. A generalization of this is that at any given step, the most recently formed variety in the chain from pre-cory to mezzo-cory would be the least fit variety that descended from pre-cory.
However, in order for mezzo-cory to have eventually formed, we have to assume that each of its predecessor varieties, despite their progressive lessening of fitness as they approached mezzo-cory, would have survived for enough generations to have brought about the next variety in the evolutionary chain to mezzo-cory. This resilience would apply also to the varieties descending from pre-cory that are not part of the chain that culminates in mezzo-cory, because each of those "spur" varieties would have more modestly developed (and consequently less resource-intensive) adaptive structures than would the variety in the chain from pre-cory to mezzo-cory that had formed during the same step.
Implicit in the above diagram is that with each step in the evolutionary chain from pre-cory to mezzo-cory, there is a doubling of the number of descendent varieties of pre-cory, including pre-cory itself. The number of existing pre-cory descendent varieties at the 100th step would thus be
2100 = (210)10 = 102410,
100010 = (103)10 = 1030 = one nonillion.
Being that this number of varieties would far exceed the planet's capacity, the vast majority of these hypothetical varieties would never come about, and the descendent varieties of pre-cory that are closest to mezzo-cory would seem the least likely to reproduce (because of the high resources required by their mostly formed but nonfunctional coryanthes-type reproductive system).
If despite the fierce competition faced by its near-ancestors, mezzo-cory were to come into existence, the competition faced by its near-descendent varieties, each of which would have a nearly fully formed but nonfunctional reproductive system in addition to its functional coryanthes-type reproductive system, would be even more intense. This would be partly due to the procreative advantage held by mezzo-cory, owing to its two working reproductive systems. But if despite the enormous odds against it, coryanthes came about in this way, one might well have expected to also find surviving instances of mezzo-cory, which would seem poised to compete at least equally against coryanthes.