Keeping when the gathering ends in fusion — compatible amoebae merging into a single plasmodium, thousands of nuclei sharing one cytoplasm inside one membrane, a body that is one cell and can still learn, remember, and find its way
R943 closed with a body assembled from strangers who kept their own membranes the whole way through — and could therefore be cheated, a fifth of them talked, in effect, into dying as a stalk. This page takes the turn it pointed to, to the slime mould of the other kind. In Physarum polycephalum the gathering does not stay cellular. Where the membranes meet compatibly, they open, and what was two becomes one continuous living material.
Alone first, as single nuclei. The cycle begins with haploid spores. In moist conditions a spore releases a small uninucleate amoeba — in water it can take a flagellated form instead, and the two forms convert back and forth. The amoebae feed on bacteria and other microbes, grow, and divide as ordinary single cells. Each has exactly one nucleus and one boundary. Nothing about this stage predicts the size of what it can become.
The fusion is real, and it is gated. When two amoebae of compatible mating types meet, cell fusion and nuclear fusion produce a diploid zygote. That zygote is the seed of the plasmodium. And the gating does not end at mating: when two grown plasmodia collide, whether they merge depends on their genetic similarity at fusion loci — similar plasmodia fuse into one, dissimilar ones stay separate, and the literature also describes incompatibility reactions appearing after fusion in some pairings. So the honest version of this page’s title is narrower than it sounds: inside a plasmodium there are no cell boundaries left to cheat across, but the boundary has not vanished from the life cycle. It has moved to the door, and it decides who may become one body at all.
Growth without division into cells. The zygote grows by nuclear division without cytokinesis: nuclei multiply, in a large plasmodium in striking synchrony, while no new cell walls are laid down between them. The result, a syncytium, is a network of interlaced tubes sheathed in slime, bright yellow, able to spread over tens of centimetres in culture and reported larger. Inside the tubes the cytoplasm streams rhythmically back and forth — shuttle streaming, driven by contractions of the tube walls — so nutrients, chemical signals, and nuclei themselves are carried through the whole body. One membrane, many nuclei, and a circulation with no heart: the flow is the body, moving.
What such a body can do. It has no brain, no neurons, and no centre, and the experimental record is still startling. Cut into pieces and set in a maze with food at two ends, the pieces first spread and re-fused into one plasmodium, then withdrew from every dead end, leaving a thick tube along the shortest of the available routes between the foods (Nakagaki and colleagues, Nature, 2000). Fed cold, dry pulses at a regular hourly interval, plasmodia slowed in anticipation of a pulse that then did not come, and a single renewed pulse could reinstate the timed response (Saigusa and colleagues, 2008; intervals from about 30 to 90 minutes were tested). Given food flakes laid out at 36 points standing for the cities around Tokyo, with light standing for the terrain it avoids, it pruned its first sprawl into a network the authors measured as comparable to the real rail system in cost, efficiency, and fault tolerance (Tero and colleagues, Science, 2010). And its memory, in one well-supported account, is written into its own plumbing: a food encounter releases a softening signal that is carried by the flow, tubes receiving more of it widen at the expense of others, and the resulting hierarchy of thick and thin tubes stores where food was found and biases later movement (Kramar & Alim, PNAS, 2021). The maze, notice, is solved by the same rule as the memory: tubes that carry more, widen; tubes that carry less, shrink. Nothing decides. What is left is the answer.
Learning that can be handed over by merging. The fusion at this page’s centre does one more thing, and it is the thing Dictyostelium structurally cannot do. Plasmodia can be habituated: made, over days, to cross a bridge dosed with a repellent but harmless substance — salt, in the transfer experiments; quinine and caffeine in the earlier habituation work — to reach food, until they cross at full speed. When a habituated plasmodium was fused with a naive one, the fused body crossed like a habituated one, and a single habituated partner sufficed even in fusions of three or four. Separated again afterwards, the naive partner kept the lesson only if the fusion had lasted about three hours — by which time a visible vein had formed at the junction — and not after only one hour (Vogel & Dussutour and Boisseau and colleagues, Proceedings of the Royal Society B, 2016; the samples ran to thousands of plasmodia in each arm). In the cellular mould, knowledge could at best be outvoted. Here it can literally flow into another body, because for three hours there was only one body. What physically carries the lesson is still not known, and this page does not pretend otherwise.
Two ways to stop. A starving plasmodium in the dark dries, reversibly, into a dormant sclerotium — a hardened knot that can wait out bad conditions and revive when moisture and food return; the one body can also fragment, and the fragments can grow and re-fuse. A starving plasmodium exposed to light takes the irreversible road instead: it builds sporangia, the many-headed fruiting bodies its species name (polycephalum, many heads) refers to, meiosis happens in the developing spores, and the cycle closes with haploid spores that wake, elsewhere, as single-nucleus amoebae again. Fusion built the largest single cell in this story; it ends, like the stalk-builders, by becoming many small separate things — but it becomes them by division of a shared cytoplasm, not by strangers agreeing to gather.
A walk through one plasmodium. Five stages: the spore waking as a one-nucleus amoeba; the gated fusion of compatible cells into a diploid zygote; growth by nuclear division without cell division into a streaming, multinucleate network; the network at work — pruning itself to a shortest route, holding a food’s place in its tube widths, passing a learned indifference across a fusion; and the two endings, dormancy in darkness or sporangia in light. Schematic, compressed from the published record.
Walk it in order.
Schematic only: the stages compress a life cycle and a set of experiments done across many strains and decades. Labelled cautions: “thousands of nuclei” is the usual order for a substantial plasmodium, not a census — very large cultures are described with far more. Fusion is compatibility-gated at both mating and plasmodial contact, and post-fusion incompatibility exists, so “no boundaries” describes the inside of an established plasmodium, not a promise that any two may merge. The anticipation result is a slowing timed to a learned interval under laboratory pulses, not foresight. The Tokyo comparison is the authors’ measurement on their three criteria, not a claim the mould designed a railway. What carries a transferred habituation through a fusion is unknown. This page is about the plasmodial cycle; the species’ other details are left where they lie.