Sweet fruit, guarded seed

Why does a plant make its fruit delicious and the seed inside it bitter or poisonous?

A peach is built to be eaten and built to punish eating, in the same object. The obvious reading — the plant wants the flesh taken and the seed spared, so it poisons the seed — is close enough to feel finished, and it hides two things worth having: what the toxin actually does, and the fact that the same peach was bitter, green and hard three weeks ago, flesh and all.

Start with what the plant is buying. Not being eaten — being carried. A seed dropped under its parent lands in the parent's shade, among its parent's pests, competing with its siblings. The pulp is not food the plant lost; it is a fee it paid, and the animal is a courier that does not know it has been hired.

Now the awkward part. The animal that can carry a seed can also grind it, and has every reason to: the seed is the richest thing in the fruit — oil and protein, where the pulp is mostly sugar and water. The two interests agree about the pulp and disagree about the cargo.

Look at how the defence is actually built, because this is the step people skip. An apple or cherry seed carries amygdalin, a cyanogenic glycoside — and, in a separate compartment of the same seed, the enzyme that breaks it down. Intact, the two never meet: a seed swallowed whole is chemically inert and comes out the far end unharmed. Crush it, the compartments rupture, the enzyme finds its substrate, and cyanide is released. The plant is not carrying a poison; it is carrying a reaction that only fires on the specific act it wants to prevent. Do you see why that beats a poison? A poison would deter the courier too.

Then a sharper case. Wild chillies make capsaicin, which is agony to rodents — animals that mill seeds between molars — and nothing at all to birds, whose receptor simply does not respond to it. Birds swallow the fruit whole and pass the seeds intact. Tewksbury and Nabhan showed exactly this pattern in 2001: the compound is not a defence against consumers in general, it is a filter that sorts them. The plant is not saying "do not eat me." It is saying "not you."

Now the timing. Unripe fruit is astringent with tannins and sometimes frankly toxic — a green tomato carries alkaloids the red one does not. Those defences are withdrawn at ripening: acids fall, starch converts to sugar, pigments switch on, deterrents are broken down or diluted. The real architecture is a switch in time for the pulp and a fence in space for the seed, tuned separately.

One honest complication, and it is contested. It is tempting to read every compound in a fruit as an adaptation aimed at frugivores, and that reading is probably too tidy. A ripe fruit is a warm, wet, sugary lump in the open air; much of its chemistry plausibly answers fungi and bacteria rather than mammals, some of it deters insect seed predators, and some appears to alter how long a seed sits in a gut. Several researchers have argued the single-purpose story is the exception, not the rule. And the seed's own chemistry keeps working long after dispersal, against rodents and beetles in the soil, in a period no disperser ever sees.

That the trait is real, genetically controlled and costly has a simple demonstration: we removed it. Cucurbitacins were bred out of squash, and a single regulatory switch separates a bitter almond from the ones you eat.

Two corrections. First, "poisonous seed" oversells it: at any dose an animal realistically takes, most of these compounds are deterrent rather than dangerous. The job is the taste and the stomach-ache, not the kill — a dead courier disperses nothing. Second, the picture of an animal tasting bitterness and spitting the seed out is only sometimes right; small seeds are swallowed unnoticed and pass, and for those the armour and the gut, not the flavour, do the work.

Held together, the fruit is three separately-tuned things wearing one skin: a signal that advertises when the seed is ready, a payment defended until that moment and then surrendered, and a cargo armoured and booby-trapped throughout.

Where this sits next to its neighbours: the collection's piece on ripening by contagion explains how the ripening switch is thrown and why it spreads through a fruit bowl; this one asks why there is a switch to throw. And the piece on pollination specificity turns on the same sorting logic in a different organ — a flower filters visitors by shape and access, a fruit filters them by chemistry and by what a gut does to a seed.

• Why some plants pay in fat rather than sugar, and which dispersers that buys.
• What happens to a fruit whose disperser has gone extinct — the large-seeded American fruits with no living animal that swallows them.

Every term the collection defines is gathered in the glossary.