The dragon ants of New Guinea, and what a 3D X-ray scan found inside

Naming a species after a dragon is the part that travels. In July 2016, Eli Sarnat, Georg Fischer and Evan Economo published a revision of the Pheidole cervicornis species group in PLOS ONE, and two of the species in it were new: Pheidole drogon and Pheidole viserion, from the rainforests of Papua New Guinea, named after dragons from Game of Thrones. The spines on the workers do genuinely suggest a wing and a claw, which is where the comparison came from.

The paper is titled Inordinate spinescence, which is a better summary of what is going on.

The problem with a very large head

Pheidole are big-headed ants. In many species the caste system produces workers whose heads are disproportionate to the rest of them, because the head houses the musculature for the mandibles that do the colony’s cutting and crushing.

A head that size is a mechanical liability. It has to be carried, held level, and moved, and the structures that do that work have to attach to something. This is where the spines stop being decoration and start being a hypothesis.

What microtomography showed

The team imaged the ants with microtomography, effectively a three-dimensional X-ray scan, which produces a model of internal structure without cutting anything open.

The finding is a division of labour written into the anatomy. In workers with large heads, the spines along the upper thorax contain muscle fibres, apparently helping to support the weight of the head. In workers with smaller heads, the same spines are practically hollow.

Two things follow from that. The spines are load-bearing where load exists, which is a functional explanation rather than a decorative one. And within a single species the structure changes according to the caste that carries it, which means anyone treating spine shape as a simple taxonomic character is comparing partly unlike things.

Why the technique is the story

A species description has to designate at least one type specimen and say where it is deposited, so that a later researcher who disputes the claim has physical material to return to. That obligation makes destructive analysis expensive. For a group known from very few collections, sometimes from a single locality, cutting a specimen open to look inside a spine is not a neutral act.

Non-destructive 3D imaging removes the trade-off. It also changes what counts as available evidence: internal characters that were previously either invisible or only obtainable at the cost of the specimen become ordinary parts of a description. This is one of the concrete reasons the annual rate of species description keeps climbing rather than falling, a pattern set out in our explainer on how many new species are discovered each year.

Where they live, and what is not known

Both species come from montane rainforest and were collected on low vegetation. Pheidole drogon is known only from its type locality. The cervicornis group as a whole is known only from New Guinea.

That is a thin evidence base, and it is worth saying so rather than implying a well-characterised animal. Nothing is established about colony size, foraging behaviour, diet or range beyond the collection records. The paper is a taxonomic revision with an anatomical finding attached, not a natural history.

The dragon names got the work onto news sites in a dozen countries. What the work actually contributed was a demonstration that a preserved specimen in a drawer still has information in it, provided you have a way to look inside without breaking it.

What “known only from the type locality” means

The phrase appears in the record for Pheidole drogon and it is worth unpacking, because it is the single most limiting fact about the species.

A type locality is the place the type specimens came from. When a species is known only from that place, it means nobody has collected it anywhere else — which is compatible with two very different situations. The animal may genuinely be confined to a small area of montane rainforest, in which case it is a narrow endemic and potentially vulnerable. Or it may be widespread and simply undersampled, because the surrounding terrain has never been surveyed by anyone with the training and the permits to do it.

Distinguishing those two possibilities requires fieldwork that has not happened. New Guinea is mountainous, densely forested, difficult to move around, and biologically among the least catalogued land surfaces on the planet. The cervicornis group as a whole is known only from the island, and contains four species.

So the honest summary of what was learned in 2016 is: two new species, one striking anatomical finding, and no information at all about how common either animal is.