Symbiosis

How can two unrelated species come to depend on each other for survival?

A fig cannot set seed without its wasp, and the wasp cannot breed without the fig. Neither planned that, and evolution has no interest in the welfare of another species. So how does a relationship of that depth arise between organisms sharing no goal — and, more awkwardly, how does it stay stable when each side would do better by taking the benefit and skipping the payment?

Start where the arrangement starts: not with cooperation but with exploitation. One organism uses another — eats its waste, occupies its tissue, robs its nectar. Suppose that in taking, it happens to do something the other benefits from; a gut microbe fermenting your food for its own sake releases compounds you can use. Nothing generous has occurred. But selection now favours the host that tolerates the microbe, and the microbe that lands in a tolerant host.

Follow that for a few thousand generations. Each side gets better at exploiting a partner now reliably present — and, crucially, each stops maintaining what the partner supplies. Why keep an expensive capacity you are handed for free? That is the engine of dependence: not deepening affection but accumulating loss. Aphids depend utterly on bacteria inside their cells for amino acids they no longer make, and those bacteria have shed so much of their genome that they can live nowhere else. Neither can leave, because each has spent the machinery required to.

So dependence explains itself. Does stability? What does a partner gain by cheating — taking the resource and withholding the service? Usually a great deal, which is why cheats turn up in essentially every mutualism examined. Some fig wasps lay eggs without pollinating. Some rhizobia occupy a legume's root nodules and fix no nitrogen. Cleaner fish, which live by picking parasites off larger fish, will sometimes take a bite of the client's protective mucus instead, which they prefer.

If cheating pays, why does the arrangement not collapse? Because the partner is not passive. Legumes cut the oxygen supply to nodules that are not delivering nitrogen, and the bacteria in them reproduce far less well — an experiment feeding nodules argon instead of nitrogen gas, so their occupants could not fix, showed exactly that penalty. Yuccas abort flowers carrying too many moth eggs, killing the larvae with them. Client fish chase a cleaner that bites, and take their custom elsewhere. What holds these together is not goodwill but the cost of being caught.

So what is a mutualism? Two lineages exploiting each other, each restrained by the other's capacity to withhold or punish. Change the conditions and it moves: where soil phosphorus is abundant, a plant gets little from its mycorrhizal fungi while still paying them in sugar, so the same fungi that read as partners in poor soil read as a drain in rich.

"Reciprocal exploitation" is a claim about the accounting, not about motive: a mutualism persists because each partner's net return is positive, and it drifts whenever that ledger changes. Nothing in it requires the partners to give equally, or even to give much.

Two honest qualifications. Sanctioning is well demonstrated in several systems but is not universal — in many mutualisms it is unclear whether stability comes from punishment, from partner choice before the fact, or simply from interests aligned by circumstance, and the weighting is still argued. And the boundary between mutualism, commensalism, and parasitism describes a moment, not a kind of organism: the same pairing can sit on either side of it depending on soil, season, or how much of the resource is already free.

• How the ancestors of mitochondria and chloroplasts crossed from symbiont to organelle.
• Why symbionts inherited directly from a parent evolve differently from those picked up fresh each generation.

Every term the collection defines is gathered in the glossary.