Keeping as a border that can only be defended — clonal groups of an aggregating sea anemone holding a bare strip of rock between them by repeated fighting, with no allele, no vessel and no offer that could ever turn the neighbour into self
R950 closed with two colonies that could, on one shared allele, dissolve their border into a single bloodstream. This page takes the turn it pointed to, to a shore where that move does not exist. The aggregating sea anemone Anthopleura elegantissima spreads asexually by longitudinal fission: one polyp stretches and splits into two genetically identical ones, and repeated splitting builds a densely packed clonal group across a boulder or rock face. Each polyp is a separate animal. Unlike the ascidian’s zooids, clonemates here share no tunic, no vasculature and no body at all — only a genotype, and a neighbourhood.
First, recognition is by contact, and the answer is tolerance or attack. Tentacle contact between clonemates is tolerated passively. The same contact with a genetically different conspecific can elicit aggression in one or both animals. There is no fusion outcome waiting on the far side of a compatible reading, and no shared circulation to gain: a non-clonemate recognised correctly is not admitted, it is fought. So the border between two expanding clones is not a decision made once at a vessel tip. It is a behaviour that has to be performed again every time the two sides touch.
Second, the weapon is kept only for this. Below the outer tentacles sit the acrorhagi: blunt, pale-tipped, inflatable structures loaded with large stinging capsules. They are not feeding tentacles and are not used on prey. In aggression an anemone inflates them, stretches and bends its column and oral disc to bring them against the adversary, and leaves something behind on contact: a peel — a scrap of acrorhagial surface tissue studded with nematocysts that adheres to the other animal and goes on firing there. The receiving tissue can die in patches. The attacker, in other words, does not merely sting and withdraw; it deposits a slowly discharging piece of its own surface and lets that piece keep working after the animals have separated.
Third, what the fighting holds is empty ground. Where two clones meet, their border polyps’ repeated attacks maintain a distinct anemone-free zone, in places several centimetres across — bare rock that neither side durably occupies. Scouts, in the later whole-clone work, are small polyps that push into that border ground and are killed or driven back; warriors set a few rows behind the front can swell, lengthen strikingly as the tide comes in, and reach over their clonemates to strike an intruder. Bouts are tidal and temporary: fighting flares when the animals are covered and active, then settles, and has to be renewed at the next tide. The strip is therefore not a wall either side built. It is the standing result of a standoff nobody wins outright.
Fourth, the border is paid for by the polyps on it. Francis’s classic comparison found the interclonal warriors carrying more and larger acrorhagi than clonemates elsewhere, and — in June, when many mid-clone clonemates carried ripe gonads — the warriors had none. They were smaller on average than centre animals and split no faster, and plausibly slower, a pattern read as less energy left for growth and reproduction after the costs of armament and fighting. Later work complicated the castes without removing the cost: up to five intergrading kinds have been described, from border scouts and the warriors behind them through free-edge polyps to large reproductives and small reserves at the centre, and the warrior form is partly inducible — enemy contact and enemy stings can raise acrorhagi numbers even in reproductive polyps, by amounts that vary from clone to clone. These are positions and responses, not five fixed kinds of animal. One caution more: a heavily stung scout returning with enemy peel on it may be attacked by its own clonemates, and that return has been offered as a way news of a neighbour could reach a clone’s interior in animals with no physical connection between members. It is a suggested reading of an observed attack, not an established signalling system, and it is kept that way here.
Last, set this beside R950 and the inversion is exact. Botryllus had a border that a single shared allele could open, and opening it only moved the contest inside, into stem cells and gonads. Anthopleura has no opening on offer at all. Sameness is recognised well enough to leave clonemates alone, but it buys a neighbour nothing: no fusion, no chimera, no second contest behind a dissolved line. Keeping here is maintenance, repeated at tidal rhythm, by unconnected polyps whose only coordination is that each answers its own contact — and the visible record of all that work is not a scar along a seam or a shared bloodstream, but a strip of rock with nobody on it.
A walk from a split to a standoff. Five stages: one polyp splitting into a clonal group of separate animals; two clones expanding until their edges touch; contact telling clonemate from non-clonemate; acrorhagi inflated and their peel left on the adversary; and the result that has to be renewed each tide, a bare border strip defended by armed, non-reproducing edge polyps. Schematic, compressed from the published accounts.
Walk it in order.
Schematic only: the stages compress work across aggregations, laboratory boulders and pairwise assays, so no single border runs this exact account. Labelled cautions: do not say clonemates are physically joined — they are separate polyps sharing a genotype, with no common circulation to open or close. Do not call acrorhagi feeding tentacles — they are specialised structures deployed in intraspecific aggression. Do not present the five kinds as fixed castes — they intergrade, and warrior armament is partly induced by contact and stings, varying by clone. Do not call the bare zone a built wall or a settled boundary — it is the renewed outcome of repeated, tidal fighting. Do not present a returning scout’s reception as a proven message system — the communication reading is a hypothesis offered for an observed attack on a peel-carrying returner. And do not import R950’s rule: no shared allele opens anything here.