Evolutionary genetics

Differentiation by dispersal

Gene flow between populations — caused by migration, for instance — is most often viewed as a homogenizing force in evolution. But two studies of wild birds and non-random dispersal find otherwise.

Whether or not two separate populations of a species become genetically different is thought to depend largely on gene flow. Classical population-genetics theory predicts that populations that frequently exchange individuals through dispersal will remain genetically similar1. Disconnected populations, by contrast, have a greater capacity to become distinct through forces such as genetic drift and adaptation to local conditions. In population genetics, dispersal is often viewed as a diffusion-like, random process, and selection and genetic variation are assumed to be locally homogeneous. Populations of organisms with high rates of dispersal — such as songbirds — are therefore expected to be fairly genetically alike at small spatial scales. But two new independent studies of wild great tits, Parus major, challenge this assumption: they show that when dispersal is non-random, genetic differentiation can be produced at surprisingly fine spatial scales (see pages 60 and 65 of this issue2,3).

Postma and van Noordwijk3 studied clutch size in great tits (Fig. 1) on the tiny — 4,022-hectare — island of Vlieland in the Netherlands from 1975 to 1995. They first found that birds that bred in the western part of the island laid, on average, 1.15 more eggs than birds from the eastern part. How much of this difference is determined by the environment, and how much is genetically controlled? Fortunately, 10% of the females born on one side of Vlieland disperse to breed on the other, and this allowed genetic and environmental effects to be teased apart. The authors' analysis showed that birds of eastern ancestry produced consistently smaller clutches in either environment — so there is clearly a large genetic component to the difference in clutch size between the regions. In fact, genetic effects accounted for about 40% of this difference. But, given that the western and eastern regions are separated by only a few kilometres, and they exchange migrants and receive immigrants from outside Vlieland, why does this genetic difference persist?

Figure 1


The great tit: challenging assumptions about gene flow and genetic differentiation.

To answer this, Postma and van Noordwijk examined the viability and fecundity of birds born in the east or west that breed in the other region, and precisely quantified levels of immigration from outside Vlieland. Immigrants and birds born in the west tended to have larger clutches than birds born in the east, regardless of where they bred. However, female birds born in the east seemed to be better adapted to life on Vlieland, because they were twice as likely to survive as birds born elsewhere (perhaps allowing them to have smaller clutch sizes). So, from the standpoint of clutch size and survival, immigrants seemed most closely related to the birds born in the west. Most interestingly, 43% of first-time breeders in the west were immigrants to Vlieland, compared with only 13% in the east.

There has thus been an influx of genes for relatively large clutches, and the higher rate of immigration to the west has resulted in bigger clutches there. Birds in the east have maintained their locally adapted smaller clutch sizes against an influx of 13% immigration. But the west is swamped by immigrant genotypes. So differing levels of gene flow have maintained large genetic differences at a very fine spatial scale.

Postma and van Noordwijk suggest3 that differentiation at this scale because of processes such as these may not be rare (as refs 4 and 5 also suggest), but may remain undetected in other populations because few researchers have appropriate long-term data. Results from a 36-year study of nestling body mass in the same species living in woodlands at Wytham in Oxfordshire, UK, would seem to support this suggestion.

Garant et al.2 found that the mean mass of birds in the eastern block of Wytham woods has decreased since 1965, whereas that of birds in the northern block has remained constant. Quantitative genetic analyses demonstrated a genetic component to these trends, and overall there was 50% more genetic variance in the northern population than in the eastern population. These regions are separated by only about 2 kilometres, and on average are both composed of more than 50% immigrants from other parts of the woods or from outside. How has this differentiation persisted under such high rates of potentially homogenizing gene flow?

Heterogeneous dispersal again seems to be the key. Birds from central regions of Wytham woods and immigrants showed non-random dispersal patterns with respect to their weight. Emigrants to the north tended to be larger over time, whereas emigrants to the east tended to be smaller. Even within families, heavier offspring showed an increased tendency to settle in the north.

So, a markedly non-random settlement pattern drives fine-scale genetic differentiation in Wytham woods. But why is there this non-random settlement? Garant et al. suggest that temporal and spatial variations in bird density provide the mechanism. Local density has been, on average, twice as high in the east as in the north, producing greater pressures on habitat in the east. However, density has been increasing in the north, as birds there are more likely to survive and be reproductively successful. Taken together, the results suggest that individuals that are genetically predisposed to be larger and heavier have preferentially settled in the lower-density habitat in the north, and this has driven population differentiation.

These studies2,3 document remarkable — and quantitatively similar — levels of genetic differentiation at a very fine scale relative to the birds' dispersal capability. In both cases, differentiation is maintained primarily by non-random dispersal and settlement, and in the absence of major spatial differences in selection. The findings shed new light on the mechanisms of microevolution, because there is no reason to suspect that these phenomena are unique to great tits4,5. But the ability to detect such phenomena depends on the availability of long-term data from continuously monitored populations of marked individuals, and on the existence of biotic or abiotic gradients that may drive the underlying microevolutionary processes. In both of these studies there was sufficient pedigree information to be able to compare the performance of individuals with common genetic backgrounds but inhabiting different environments.

With the increasing use of quantitative genetics analyses such as these2,3 in evolutionary studies of wild animals6, we ought to be paying more attention to how quantitative genetic variation is spatially and temporally structured. For example, studies that integrate fine-scaled maps of habitat quality with complex pedigrees can look at how genes are distributed spatially, and how they may interact with the environment.

The great advantage of the quantitative genetic approach is that microevolution can be studied in the wild without knowing exactly which genes are responsible for the variation in the physical traits in question. But this is also a great disadvantage, because it would be ideal to see evidence for spatially associated differences in the genetic loci involved7. Indeed, a truly mechanistic understanding of microevolution requires an understanding of genetic architecture (the properties of the individual genes underlying variation)8. One way of gaining such an understanding of microevolution in nature will be to apply genomics9,10 to ecological and evolutionary studies in non-model species, using comparative approaches11,12,13.


  1. 1

    Slatkin, M. Science 236, 787–792 (1987).

    ADS  CAS  Article  Google Scholar 

  2. 2

    Garant, D., Kruuk, L. E. B., Wilkin, T. A., McCleery, R. H. & Sheldon, B. C. Nature 433, 60–65 (2005).

    ADS  CAS  Article  Google Scholar 

  3. 3

    Postma, E. & van Noordwijk, A. J. Nature 433, 65–68 (2005).

    ADS  CAS  Article  Google Scholar 

  4. 4

    Blondel, J., Dias, P. C., Ferret, P., Maistre, M. & Lambrechts, M. M. Science 285, 1399–1402 (1999).

    CAS  Article  Google Scholar 

  5. 5

    Chan, Y. & Arcese, P. J. Evol. Biol. 16, 939–947 (2003).

    CAS  Article  Google Scholar 

  6. 6

    Kruuk, L. E. B. Phil. Trans. R. Soc. Lond. B 359, 873–890 (2004).

    Article  Google Scholar 

  7. 7

    Purugganan, M. & Gibson, G. Mol. Ecol. 12, 1109–1112 (2003).

    CAS  Article  Google Scholar 

  8. 8

    Mackay, T. F. C. Annu. Rev. Genet. 35, 303–339 (2001).

    CAS  Article  Google Scholar 

  9. 9

    Feder, M. E. & Mitchell-Olds, T. Nature Rev. Genet. 4, 651–657 (2003).

    CAS  Article  Google Scholar 

  10. 10

    Thomas, M. A. & Klaper, R. Trends Ecol. Evol. 19, 439–445 (2004).

    Article  Google Scholar 

  11. 11

    Primmer, C. R., Borge, T., Lindell, J. & Saetre, G. P. Mol. Ecol. 11, 603–612 (2002).

    CAS  Article  Google Scholar 

  12. 12

    O'Brien, S. J. et al. Science 286, 458–481 (1999).

    CAS  Article  Google Scholar 

  13. 13

    Aitken, N., Smith, S., Schwarz, C. & Morin, P. A. Mol. Ecol. 13, 1423–1431 (2004).

    CAS  Article  Google Scholar 

Download references

Author information



Rights and permissions

Reprints and Permissions

About this article

Cite this article

Coltman, D. Differentiation by dispersal. Nature 433, 23–24 (2005). https://doi.org/10.1038/433023a

Download citation

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.