A Jurassic giant with its mouthparts showing

A giant Jurassic dipluran shows that hidden mouthparts evolved more than once.

The specimen was collected about twenty years ago in Daohugou, Inner Mongolia, but it was many years before we really looked at it closely.

The Daohugou Biota has yielded hundreds of thousands of insect fossils, and our team has worked there for decades. After this specimen was collected, it sat quietly in the collections until we systematically screened Daohugou's hexapod material and noticed it again. Under the microscope, at that first close look, the most immediate impression was "big"—over 5 cm in body length, extremely rare for a dipluran. Next came the stout, pincer-like cerci at the rear, and the powerful mandibles with strongly curved, coarse teeth at the front. But what really made us stop was the mouthparts: part of the mandibles sat outside the head capsule. For a dipluran, that is something that almost should not happen.

The hardest decision was where to place it on the evolutionary tree.

We ran it through a phylogenetic matrix for diplurans. Its body features—especially the shape of the cerci—placed it clearly in the stem group of Japygoidea: the pincers are symmetrical, smooth and toothless on the inner side, and the rear abdomen is neither obviously shortened nor heavily sclerotized. These are "transitional" traits that living japygoids lack. The hardest call was this: is the ectognathous mouthpart a unique derived trait of this fossil, or a retained ancestral state of diplurans? We concluded the latter, because hexapods are themselves a specialised group of crustaceans, so their ancestral state should be ectognathous, and this specimen's ectognathous mandibles are structurally closer to the general hexapod condition than to any specialisation. That judgment directly supports the conclusion of "multiple origins of entognathy," and it was the step the whole paper had to handle most carefully: the mouthpart state of a single specimen would have to carry the weight of rewriting the ancestral state of an entire group. In the end, we settled its position through the morphological matrix and item-by-item comparison with living groups.

Ongoing thread. More from Yehao Wang and Diying Huang to follow.