Chapter I
An Unexpected Unity
Darwin wrote in an 1860 letter that embryology was to him "by far strongest single class of facts in favour of change of form", and Haeckel built a theory on it. But in the twentieth century evolutionary biology became a science of gene frequencies in populations, and development dropped out of it. Most biologists assumed that a fly and a mouse, which share no body parts, are built by unrelated genes.
The homeobox overturned that. Within months of its discovery in fly genes by developmental genetics, William McGinnis and others found it in frogs, mice and humans. By 1989 the mouse Hox genes were known to sit in clusters, in the same order as the fly's, and to be switched on along the body in the same order, head to tail. The last common ancestor of flies and mice, a small animal living well over 550 million years ago, already used them to tell its front from its back.
Chapter II
Eyes and Limbs
Eyes seemed the clearest case of separate invention. The compound eye of a fly, with hundreds of facets, and the camera eye of a mouse, with one lens, have nothing in common. Yet Walter Gehring's laboratory found in 1994 that the fly's eyeless gene and the mouse's Pax6 gene are the same gene, needed for eyes in both. In 1995 Georg Halder and Patrick Callaerts switched eyeless on in the wrong places and grew fly eyes on legs, wings and antennae. The mouse gene did the same in a fly, making fly eyes, not mouse eyes.
Neil Shubin, Cliff Tabin and Sean Carroll named the pattern in 1997: deep homology. A fly's leg and a fish's fin are not descended from a common appendage, but the genes that pattern them are. Evolution builds new organs by reusing old instructions. Shubin tested this in the rocks as well. With Edward Daeschler and Farish Jenkins he searched Arctic rocks of the right age for a fish on its way to walking, and in 2004 found Tiktaalik. Its fin held the beginnings of a wrist.
Chapter III
A Closer Look: Counting Hox Genes
Why do mice have more Hox genes than flies? Count them:
| Animal | Hox clusters | Hox genes |
|---|---|---|
| Fruit fly | 1 (split into two parts) | 8 |
| Amphioxus (lancelet) | 1 | about 15 |
| Mouse or human | 4 | 39 |
| Zebrafish | 7 | about 49 |
Amphioxus is a small, fish-shaped relative of the vertebrates, and its single cluster looks like the ancestral vertebrate one. Early in vertebrate history the whole genome was copied twice, so one cluster became
clusters, now called A, B, C and D. In mammals the Hox genes fall into 13 numbered groups, by their position in the cluster. If all four clusters had kept a gene in every group, there would be . Humans have , so
three quarters of the possible slots are filled, and 13 copies were lost. The ancestors of zebrafish and most other ray-finned fish copied their genome a third time, making clusters and possible slots. Zebrafish have seven clusters left and about 49 Hox genes, fewer than half.
Most spare copies are lost. The ones that survive often take on new jobs, and some are later found at the heart of vertebrate novelties. The Hox genes of groups 9 to 13 in the A and D clusters pattern the limbs, and a late phase of Hoxd13 activity builds the fingers and toes. Duplication gives evolution spare copies to experiment with while the original keeps working.
Chapter IV
Switches
If the kit is shared, what makes animals differ? Mostly, it seems, when and where the genes are used. Each toolkit gene has many switches, stretches of nearby DNA that turn it on in particular tissues. Changing one switch alters one body part and leaves the rest alone, while changing the protein would alter every tissue that uses it. The threespine stickleback shows this at work. When sea fish colonised lakes after the last ice age, many populations lost their pelvic spines. David Kingsley's group traced the loss to a deleted switch for the gene Pitx1, which is also needed for the jaw and pituitary. The protein was intact. The same deletion had happened independently in lake after lake.
How much of evolution works this way is argued over. Carroll's 2005 book made the case that switches matter most, and others replied that changes to proteins explain many adaptations too. The field now lies between developmental genetics and population genetics, and the open question is how genuinely new structures, such as feathers or the turtle's shell, first arose.