If something requires, say, five individually-deleterious mutations, each of low probability, it's quite plausible that these will never align in a single organism.

There may be a more esoteric explanation... Imagine higher efficiency has a price : plant lives a shorter life, and reproduces less. Then you want to fine tune for the right amount of efficiency in your environment. Then it's advantageous to have a single efficiency lever, instead of a dozen systems to fine tune jointly... Just a guess - I know nothing about bio. :)

FWIW I'm quite bullish on the type of research in the Fine Article; humans have the ability to provide choice conditions to our crops, and so we can maximize yield without (much) concern for the tradeoffs experienced in the state of nature.

I'd predict, for example, that we'll have a C4 rice crop out of the lab and in the field in ten years or less.

I expect us to find out why evolution didn’t do it by doing it to the plant genome ourselves — and that either it requires too many simultaneous changes for evolution, and/or it turns the plant into something irresistible to insects or fungi.