Many plants of fire-prone country will not germinate until they have burned, or nearly so, and throughout most of the last century the mechanism was assumed to be heat. The reasoning was straightforward: a seed lying in the litter is momentarily cooked as the combustion front passes, the coat fractures, water penetrates, and the seedling emerges into ground swept clear of competitors. Heat could be administered in a laboratory, and administered to certain species it succeeded. The mechanism possessed the merit of visibility: one could observe a coat split.
It did not work on others, and the failures were not negligible. Several genera whose seed germinates freely on burnt ground proved almost wholly indifferent to heating, at any temperature that left them alive. For a considerable time these were classified as exceptions awaiting a better oven. The explanation, when it ultimately arrived, was that the cue is not the fire but the smoke -- and specifically a small, stable molecule generated when cellulose undergoes combustion, which reaches the soil dissolved in rainwater and remains active there at concentrations of a few parts per billion.
The molecule accounts for a distribution that heat never could. Seeds in the uppermost centimetre of soil are the ones a passing fire genuinely cooks, and yet germination after fire is frequently heaviest considerably below that depth, where the temperature barely fluctuates. A soluble compound transported downward by the first rain reaches those seeds precisely as it reaches the surface ones, and does so according to a timetable that matters: rain after fire is when the cleared ground is transiently both bare and moist. Depth accounts additionally for a response that outlasts the fire by weeks rather than by minutes.
It furthermore accounts for something stranger. Seed of several such species can be stimulated to germinate by smoke from burning paper, which no plant in their range has ever encountered, and by smoke from vegetation types they do not grow among. A cue that operates indiscriminately resembles inadequate design until one recognises what the plant is actually detecting. It is not detecting fire, which it possesses no means of distinguishing from a bonfire; it is detecting the product of burnt cellulose, which in an unploughed landscape is generated by fire and by nothing else. The specificity lies in the world rather than in the receptor.
Whether the molecule qualifies as a signal in any richer sense is disputed. It is not manufactured by the plant, and the fire has no interest in being detected, consequently the arrangement lacks the two participants that a signal ordinarily requires. What it resembles more closely is an eavesdropping: a lineage that has discovered a reliable trace of an occurrence it cares about, and has constructed a switch around it. Whether that constitutes a distinction with consequences, or merely a manner of speaking, is not yet settled.
