This is textbook and uncontroversial. Every tetrapod, meaning amphibians, reptiles, birds and mammals, descends from lobe finned fish that were evolving limb like appendages in the Devonian. The fossil sequence from Eusthenopteron to Tiktaalik to Acanthostega is one of the best documented transitions in vertebrate paleontology.

A 2026 review in Frontiers in Cell and Developmental Biology reaffirms the modern consensus, including the point that limbs largely evolved in water, before anything walked on land. Limbs came first. Terrestriality came after.

The 2016 Nature paper is real and it is good. Using CRISPR/Cas9 and cell fate mapping in zebrafish, the team knocked out hox13 genes and watched fin rays shrink and disappear while endochondral distal radials, the bone type that becomes wrist and digits, increased. Cells marked by the hoxa13 autopodial enhancer went on to build fin rays, including the osteoblasts that mineralize them. Same genes, same enhancer machinery, different downstream anatomy.

The article presents "fingers came from fin rays" as settled. It is not. Duboule's group at Geneva published a competing view in PLOS Biology in 2014: when you drop fish Hox regulatory landscapes into transgenic mice, they drive proximal limb expression but fail to produce digit specific patterns. Their conclusion is that digits are a tetrapod novelty, built by retrofitting old regulatory hardware rather than inherited wholesale.

Both can be partly right. Shared genetic toolkit, novel structure. But a reader of the popular version would not know a live argument exists.

Correct, and it is one of the great "wait, what?" facts in comparative anatomy. Coates and Clack reported it in Nature in 1990: Acanthostega had eight digits on the forelimb, Ichthyostega had seven on the hindlimb, Tulerpeton had six. Five was not the starting condition. Five was the survivor.

The consolidation appears to be a single event along the tetrapod stem in the Early Carboniferous, around 340 million years ago, not repeated independent reductions. Michael Coates has argued the shift tracks the move from paddle to weight bearing foot, with real wrists and ankles arriving at the same time.

This is the weakest line in the piece and it is doing a lot of work. Taken literally it is wrong. A horse walks on one digit. A bird wing has three. Most salamanders have four on the forelimb. Skinks and other squamates have reduced and lost digits repeatedly. Digit loss is one of the most common things limbs do.

What the sentence probably means is that the pentadactyl ground plan is near universal, and that is defensible. But the accurate and far more interesting version of the claim is asymmetric: digit number goes down all the time, and essentially never goes up. That asymmetry is the actual finding. The article flattens it into a tidy 99 percent.

There is no gene for five. Hox genes pattern the limb axis, and Shh signaling from the zone of polarizing activity sets digit identity across it, but nothing in the genome specifies the integer. "Encoded blueprint" implies a spec sheet that does not exist.

The leading explanation is not a blueprint but a trap. Galis, van Alphen and Metz argued in Trends in Ecology & Evolution in 2001 that amniote limbs form during the phylotypic stage, when developmental signaling is running body wide and modularity is low. Mutations that change digit number therefore drag along pleiotropic damage elsewhere. The mutation is not lethal to the limb. It is costly to the organism. Their supporting prediction is elegant: amphibians with aquatic larvae build limbs after that vulnerable window, and they show more digit variability. Decouple the development, loosen the constraint.

Nakamura himself concedes the honest position in the article. Why five and not six or four is, in his words, quite a difficult question. That sentence deserved to be the headline rather than a mid piece aside.