For a story I'm writing, I'm looking for an undetectable plant based poison that if ingested in a miniscule amount, would make a ten year old boy sick: intrinsic muscle weakness, stunted growth, and loss of speech, slowly progressing over four years, but not kill him. Help.
By inventing it you can make it perfect for the story. Presumably this is not a story for biochemists or forensic pathologists, but of interest because of the doings surrounding this boy's slow demise. Benefits of invention:
1: You can make it do what you want it to do.
2: You can give the poisoner a backstory to explain how he/she came by knowledge of this exotic plant or mushroom, a species unknown to science. Perhaps she is a washed-up postdoctoral fellow who found the plant in the jungles of Albania. Or a devotee of occult lore who found the directions in medieval Tasmanian alchemical manuals. Good stuff like that is the Tabasco sauce of a story!
3: You do not need to worry that some malevolent soul will read your book and decide to follow your recipe for poisoning some unfortunate boy.
4: Worldbuilders do not need to worry that you are that malevolent soul who is looking for a recipe for poisoning some unfortunate boy.
If someone exclaims "There is no such plant as Venenum herba!" you can nod and stroke your chin knowingly.
You're basically looking for a derivative of a strychnos-type plant poison. The Strychnos genus of trees and woody vines contains some 100 distinct species, the most well-known of which are strychnos nux-vomica and strychnos toxifera. These two species produce the poisons strychnine and curare, respectively. Numerous other species produce related alkaloids, many of which are also extremely potent neurotoxins, like the less well-known but even more potent toxiferin.
These poisons are fatal in concentrations as low as 0.5 mg/kg, so a microdose of the poison would border on undetectable in modern tox screens unless you were specifically looking for it. So, you have plant-based, undetectable, and effective in small doses; we're off to a good start.
Of the two, curare has a specific mechanism of action closer to your needs. The poison blocks nicotinic acetylcholine receptors between motor neurons and muscles, which prevents the neurons being able to trigger the muscle to contract. This causes muscle weakness up to and including paralysis, and in the extreme the diaphragm is paralyzed leading to asphyxiation. This happens at an LD50 of about 750 mcg/kg (.75mg/kg), meaning that just 25mg of the poison would be lethal to a 30kg 10-year-old. A much smaller dose, say 2-5mg, would likely induce significant muscle weakness without full paralysis and asphyxiation.
Chronic toxic effects of curare and related compounds have not, to my knowledge, been extensively studied; the toxin is short-acting and metabolizes completely from the system in some hours, assuming the victim survives the ordeal. You can probably hand-wave the chronic effects of stunted growth and loss of speech as side-effects of sustained microdosing, based either on the chemical's actual effects or as a simple result of being bedridden with the muscle relaxer making any movement including of the jaw and tongue difficult.
Strychnine has an opposing effect, binding to glycine receptors in the motor neurons, which increases the neurons' sensitivity to triggering from other neurochemicals. The mechanism of death is again asphyxiation, but this time it's due to uncontrollable contraction of the diaphragm preventing the victim being able to breathe. In fact, curare was used medicinally as an emergency antidote to strychnine poisoning as it allows the locked-up diaphragm to relax (mechanical ventilation can then keep blood oxygenation at healthy levels while both drugs are metabolized out of the patient's system).
We now have better options for pharmaceutical muscle relaxers, many of them derivatives of curare's chemical mechanism, so curare (specifically its isolated principal toxin, d-tubocurarin) is not used medicinally anymore and it's highly unlikely to show in a routine tox panel.
Only one last problem for your desired route of administration; ingesting curare produces no ill effect. The toxin does not easily penetrate mucous membranes such as the intestinal wall (making it a perfect "arrow poison" for use in hunting, which was its original use). This typically requires administering curaroids intravenously or intramuscularly, requiring the poisoner to have some excuse to perform injections. A medical care provider, a so-called "angel of death" with Munchausen-by-proxy, would be an ideal candidate; failing that, tampering with the insulin supplies of a diabetic would be another viable route for pretty much any family member of the diabetic.
Regular small amounts of Tetrodotoxin. Maybe.
Tetrodotoxin is a nasty poison which evolved in poisonous sea life. This is how you know it's bad, because as a rule of thumb, aquatic poisons are always worse than land poison. The LD50 for this stuff isn't known for humans, but the estimate is around 1-2 mg.
Upon consuming it, tetrodotoxin will slowly spread throughout your body and shut down your nervous systems by blocking the voltage gate channels in your nerve cells. Now, the fun thing about this is that it can't pass the brain blood barrier, so it doesn't affect the brain, but affects the communication to muscles, and thus the subject will eventually die as their lungs shut down over the course of a few hours. Why being fully conscious the entire time, but unable to do anything. I believe the preferred treatment is by placing the subject on artificial life support until the poison wears off.
Now, if you poisoned someone will below lethal levels of TTX, and I mean in doses measured in micrograms, it won't kill them, but it'll stop some of the nerves firing to some of the muscles. Regular poisoning of such would effectively mean that a decent percentage of their muscles simply aren't getting used at any given time, which will lead to muscle weakness and muscle atrophy, as well as explain the other symptoms you're looking for. I'm not positive this will happen, as it's never been tested like that (for very obvious reasons), but it's a reasonable explanation in a fictional story.