News & Views | Published:

Developmental biology

Hotspots for evolution

Two studies of fruitflies suggest that although development relies on a diverse toolkit of genes, the evolution of physical characteristics might be powered by variation in just a few of these tools.

The physical characteristics of animals evolve because their genes change over successive generations. It is not always clear, though, which genes are involved1,2. The genes that regulate embryonic or larval development are likely candidates, because they control how the animal develops its characteristic form and features. It is possible that natural selection might produce evolutionary change after the adjustment of just a few such switches on the genetic control panel of development. Writing in this issue, Sucena et al.3 and Gompel and Carroll 4 provide evidence that this can indeed happen. They show that modifications at just a few developmental hotspots underlie 'parallel' evolutionary changes that occurred independently in different species.

Tinkering with developmental genes is not necessarily an easy route to evolutionary change. For example, mutations in the antennapedia gene — an important regulator of development in the fruitfly Drosophila — can produce a fly with legs growing on its head5. This may be fascinating to developmental geneticists, but from the fly's point of view it is not helpful.

So, how are developmental genes altered during the normal course of evolution in natural populations? To answer this question, researchers need to marry a knowledge of evolutionary changes with developmental genetics — as Sucena et al.3 and Gompel and Carroll4 have now done. Both groups had already made preliminary studies with a single species of the fruitfly Drosophila melanogaster (one of the key model species in developmental genetics). But model species alone cannot tell us everything about evolution, and so each team broadened their investigations to include other fruitflies with distinctive physical characteristics (morphology). They then searched for changes in genetic pathways that might account for the differences.

Sucena and colleagues3 examined the development of hairs — called trichomes — in the larvae of different Drosophila species. They found that hairless patches on the young larvae had evolved independently in three of the lineages included in the study (Fig. 1). If something evolves once, it can be difficult to find out why, but if it evolves three times independently within a species group, we can look for correlations by mapping developmental changes and trait evolution onto a 'phylogenetic' tree (a sort of family tree)6. And Sucena and colleagues found such a correlation: the activity of a gene called shavenbaby (svb) was absent from the naked areas of all three lineages that showed hair loss.

Figure 1: Shavenbabies are hairy.

As described by Sucena et al.3, the correlation between the loss of trichomes (hairs) in different fruitfly species and the loss of the activity of the shavenbaby gene (svb) is revealed when the data are mapped onto a family tree (a phylogeny). The boxes represent trichome coverage and svb expression (assessed through measuring levels of messenger RNA, mRNA) in an anterior segment of the fruitfly's abdomen. In all three lineages in which trichome loss arose independently, svb expression is also lost (vertical red and grey bars, respectively, on the tree). Gompel and Carroll4 offer a similar but more detailed picture of their results in their Fig. 2 (page 933).

These findings could mean that a loss of svb expression was directly responsible for the trichome loss. Alternatively, the loss of svb could simply be a consequence of another, more important, change that occurred earlier in development. But, as Sucena and colleagues note, previous studies have shown that several 'upstream' genes involved in trichome patterning, including wingless and engrailed, are not altered in certain species, related to Drosophila virilis, that show trichome loss7,8. To determine exactly how the svb gene affected trichome development, the authors used another powerful tool — they crossed certain species of Drosophila and looked at gene activity in hybrid animals. These additional data enabled them to infer that a loss of svb expression itself was probably directly responsible for the evolution of hair loss.

Meanwhile, Gompel and Carroll4 surveyed variation across species in the abdominal pigmentation pattern of adult fruitflies. Each segment of the abdomen in D. melanogaster bears a dark band. The sexes also differ: the two terminal segments are darkly pigmented in males, but not in females. This patterning is controlled by the products of two bric-a-brac genes, bab1 and bab2. The Bab proteins repress pigmentation and are produced in segment-specific and sex-specific patterns in the pupal abdomen during development. In D. melanogaster, Bab function is required not only for pigmentation but also for trichome development.

The authors analysed 13 species of Drosophila and found several different patterns of bab2 expression. In most of the species, patterns of bab2 activity were well correlated with the pigmentation of the adult flies, suggesting that, in these animals, Bab2 does indeed function as a regulator of pigmentation and is involved in the evolution of similar pigment changes in different lineages. But in some species, bab2 activity was correlated not with pigmentation but instead with the pattern of trichomes on the abdomen. So although pigmentation and trichome patterning changes are often coupled, their evolution has become uncoupled in some groups. Gompel and Carroll suggest that the two traits become uncoupled when they are under different selective pressures. For example, when pigmentation changes, but trichomes do not, the biological function of the latter might somehow be important for survival.

Both studies suggest that although many genes are involved in the development of physical characteristics, some evolutionary changes — including examples of convergence, in which independent evolutionary events in different species result in similar physical characteristics — involve key regulatory points. The bab2 and svb genes might each function as convenient switches for modifying, or turning on and off, a given trait during evolution. Each gene might represent a developmental 'hotspot' for evolution — or, as Sucena et al. put it, a regulator that preferentially accumulates evolutionary change. Another possible example of a developmental hotspot can be envisaged during vertebrate limb evolution: changes to genes that determine the duration of limb growth in the embryo might have been important in the evolution of specialized limb types, such as the dolphin flipper6.

Although not all instances of trait convergence rely on the same genetic mechanism, these studies have uncovered several instances that do. And this creates a headache for biologists struggling with the concept of homology (in developmental genetics, homologous characteristics are defined as being 'identical by descent', which means they are derived from equivalent genetic networks in common ancestors rather than arising independently in separate lineages). Loss of trichomes in different lineages is traditionally considered an example of parallel evolution and not true homology, because the loss was not inherited — instead, it arose independently in the different species. But if changes in the same gene are involved in each lineage, we could perhaps say that trichome loss is homologous at the level of the underlying process.

The activity of bab2 and svb has been modified repeatedly in evolution, probably largely through changes in the regulatory sequences that switch the genes on or off. Understanding how such modifications have occurred might reveal why some genes and not others are developmental hotspots for driving evolutionary change. Sucena et al. suggest that svb might be a hotspot because it integrates numerous genetic inputs to control the final output — the pattern of hairs5.

The two new studies3,4 have revealed some of the genetic mechanisms that might underlie the evolution of hair loss and pigmentation. But they also point to a future challenge — understanding the function of the physical characteristics, and the causes of the evolutionary change. The integration of evolution with development is beginning to encompass the broad tools of quantitative genetics and functional genomics. This approach will show us how physical changes are generated, but it will tell us nothing about why evolutionary change has taken place in natural environments9. The brightest future for evolutionary developmental biology might lie instead with the study of systems in which we can analyse the cause as well as the function of evolutionary change.


  1. 1

    Stern, D. L. Evolution 54, 1079–1091 (2000).

  2. 2

    Tautz, D. & Schmid, J. Phil. Trans. R. Soc. Lond. B 353, 231–240 (1998).

  3. 3

    Sucena, E., Delon, I., Jones, I., Payre, F. & Stern, D. L. Nature 424, 935–938 (2003).

  4. 4

    Gompel, N. & Carroll, S. B. Nature 424, 931–935 (2003).

  5. 5

    Struhl, G. A Nature 292, 635–638 (1981).

  6. 6

    Richardson, M. K. & Chipman, A. D. J. Exp. Zool. 296B, 8–22 (2003).

  7. 7

    Bokor, P. & DiNardo, S. Development 122, 1083–1092 (1996).

  8. 8

    Hatini,V. et al. Genes Dev. 14, 1364–1376 (2000).

  9. 9

    Beldade, P. & Brakefield, P. M. Nature Rev. Genet. 3, 442–452 (2002).

Download references

Author information

Correspondence to Michael K. Richardson or Paul M. Brakefield.

Rights and permissions

Reprints and Permissions

About this article

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.