Evolution of species with separate sapient and non-sapient forms?

This question is inspired by the last paragraph of my answer here: Negating the negative effects of childbearing

(A follow up question to this can be found here: Methods to ensure gene flow in species with seperate sapient and non-sapient lifecycle)

I mentioned a concept for a species which I think would be interesting to explore, but which would require some creativity to have evolve logically. The idea is that the species combines R and K reproduction strategies, having many (hundreds) of young originally, but only investing energy into raising a very small number of them to adulthood as sapient creatures. This is done by imagining that the species start out as non-sapient and only some will later metamorph into full sapience. I'm trying to figure out how such a creature would evolve, if it could happen, and perhaps some of the other biological requirements to make it realistic.

for this example imagine a creature that's mating starts out similar to frogs. When two sapient adults reproduce the produce hundreds, or thousands, of eggs which are left unprotected by the parents to develop into a sort of tadpole like creature on their one. These creature are not sapient. These non-sapient 'tadpoles' (they don't have to look like tadpoles, just an analogy) will try to survive on their own and many will die to predators or even competition with siblings quickly in the first few weeks. Eventually the parents will come back to the breeding spot where they left the tadpoles and will pick a small number of the strongest/fittest tadpoles that survived to raise as their children.

The few tadpoles that are chosen by the parents (probably only 2-5 in total) will undergo some sort of triggered metamorphosis to start developing more complex brains and begin the growth into true sapience, a growth that would require years of rearing and upbringing (you can't have true sapience without learning almost everything from scratch). Call those that go through the metamorphosis as being in their S phase, for sapient.

However, in addition to the few young the parents choose to raise there would be other tadpoles that survived and were not chosen by the parents. Those tadpoles not picked by their parents stay in their original phase without the metamorphosis into sapience, call this phase T for tadpole. Lets say that the unpicked T phase tadpoles will continue to grow and develop without parental support. Those T phase young that survive long enough (and very very few will) eventually develop into full T phase adults and be capable of reproducing on their own. These T phase adults can mate, with both T and S phase adults, to produce young of their own, all while still not being sapient.

The idea is to have both R and K reproductive strategies in one species, so I'm imagining the parents put extensive effort into raising the few sapient phase children they choose, and put little or no effort into the remaining T phase children, instead leaving them to strive to survive on their own. Thus the parents benefit from both lots of young with little effort on their part, some of whom may manage to beat the odds and survive to reproduce, and also have a few carefully groomed children they will raise and ensure reach adulthood. They get the benefits of both reproductive strategies.

For that matter S phase parents who are not ready to raise children would presumably still mate and simply leave all of their children as T phase without choosing any to raise any to S phase. They get the freedom of choosing rather to favor R or K reproductive strategies based off of their situation (though if not enough choose to raise young as sapient phase your eventually end up with only T phase young).

The question is how to make such a species work. My biggest question is how to justify their evolving (the sapient phase being the harder part to justify), and justifying their having an evolutionary niche which causes this to be a stable reproductive strategy (ie they don't split into separate populations or evolve to be entirely S or T phase individuals). Any other suggestions about the physical and evolutionary implications of such a species, or even their evolutionary psychology, is always welcome but isn't required.

For the sake of this question lets assume that the creatures are sapient but have not reached the level of culture where they would start to use technology to alter or modify their evolved breeding behavior. For instance they won't yet have excess resources which they could use to take care of T phase children or be domesticating/raising T phase individuals, just letting them run wild.

Some things I can already guess are likely for what I have already described, you can skip the below if you want, or argue with any of it...

1. There would have to be a large 'feral' group of T phase creatures. They are capable of breeding and expanding on their own even if they didn't have an influx of young coming from the sapient adults breeding. Many of the current T phase adults are not children of sapient adults, but of mated T phase individuals.

2. Sapient adults must have a significantly higher reproductive success rate then non-sapient adults, to justify the massive energy and expense parents put into them when the parents could be investing in simply having more T phase children.

3. There must be a way for the genetics of the T phase creatures to reenter the gene-pool of the sapient phase adults, otherwise speciation would occur quickly. I'm open to any suggestions for this to happen. The most obvious one that comes to mind is that sapient males that are unable to find sapient females to mate may choose to mate with non-sapient females and raise one or two of those as sapient children as a single father. Likewise sapient females could mate with T phase males, but that seems less likely. Presumably it would be considered preferable to mate with sapient phase adult since they were the pick of the litter as T phase children, and a mate would be able to better judge their biological fitness for producing sapient children if they are also sapient. Any other way to increase the gene flow between the sapient and non-sapient phase would be interesting to know, as I see this being the biggest hurdle in justifying such a mating strategy.

4. They would almost have to have an evolutionary ingrained method of identifying their offspring, by sent, sight, sound whatever, in order to ensure they only pick children that are genetically 'theirs' to raise into sapient.

5. They almost certainly started as R strategy non-sapient creatures, with the sapient phase coming later. Though how you justify a metamorphosis into sapience evolving is the real question.

6. The non-sapient creatures are likely not very intelligent, R strategy animals are usually mostly instinctual and not quick learners. Despite the potential for metamorphosis into more intelligent species this does not mean T phase are all that bright. There likely also smaller for much the same reason.

edit:

One other complication to this being a stable strategy comes up. How do we keep the sapient males from constantly mating with T phase females and out competing their t phase counterparts? Presumably their sapient is a sufficient advantage as to make it fairly easy to compete with non-sapient males for reproductive rights. Two approaches come to mind.

First, that T phase males grow to be physically stronger then their sapient counterparts (possible because they invest less nutrients on brain power). This means it is not as easy for a sapient male to out compete with T male for mating rights. I don't like this as much it seems that a R strategy creature doesn't have any right to grow larger then a K strategy one. More importantly this would make it harder for T and S phase to mate and mating between them have to be at least semi-common for gene-flow reasons.

A solution I like better is to imply that males actually have a heavy investment in reproduction, equal to that of females. Perhaps in addition to fertilizing eggs they secrete some sort of nutrient rich substance which provides nutrition for the young during their early development. Thus both males and females have a non-trivial caloric investment in reproducing. They can still reproduce semi-often, perhaps as often as every few months, but both males and females have reason to be selective about mates because both have equally high investments into child production and limited mating opportunities.