The evolution of animal species living on islands has given rise to spectacular examples of adaptive radiation in response to new physical and biotic environments1,2. Natural selection is the main mechanism responsible for shaping these radiations, leading to accelerated rates of speciation, but non-adaptive processes may also play a part1,3. We suggest that the extreme topography of Madeira has restricted the movement of mouse populations between valleys, setting the stage for extensive chromosomal radiation.

We found six distinct chromosomal races of the house mouse on Madeira, separated from each other by mountain barriers (Fig. 1). The races have severely reduced diploid numbers relative to the standard for the species (2N=40) owing to the presence of numerous robertsonian chromosomal fusions4, each of which results from the joining of a pair of the standard chromosomes. These fusions have been recorded in wild house mice4, but the six races on Madeira that carry robertsonian fusions differ from each other and from races elsewhere as a result of the particular combinations of fusions that characterize them ( Fig. 2).

Figure 1: Distribution and composition of chromosomal races in Madeira (32.75° N, 16.97° W).
figure 1

Map source: Atlas Digital do Ambiente–DGA.

Of the 20 robertsonian (Rb) fusions observed, 13 have not previously been recorded. The presence of two fusions involving chromosome 19 is exceptional: they are the first observed in 91 fusions in wild house mice4,9. Diploid numbers (N) and sample sizes (n) with ranges are as follows: red dot, 2N=22, n=43 (1–21); red rectangle, 2 N=23–24, n=5 (1–2); red star, 2N=24–40, n=38 (2–19); yellow dot, 2N=28–30, n=5 (1–4); blue dot, 2N=25–27, n=10 (2–8); white dot, 2 N=24–26, n=11 (1–4); green dot, 2N=24–27, n=25 (1–7); black dot, 2N=24, n=6.

Figure 2: Karyotypes of female mice from the two most differentiated chromosomal races.
figure 2

a, 2N=22 (race identified by red dots in Fig. 1); b, 2N=24 (green dots in Fig. 1).

House mice are thought to have been introduced onto Madeira following the first Portuguese settlement during the fifteenth century, although fourteenth-century charts indicate that the island was known about before then5. The extent of their chromosomal differentiation indicates that populations of house mice were isolated in different valleys for long enough to allow fusions to accumulate. Because each successive fusion as it appears is associated with only a slight underdominance4, its fixation can most parsimoniously be explained by population processes such as genetic drift, as inferred for other chromosomal races of house mice6.

This example of chromosomal variation among mice on Madeira suggests that human movements were similarly restricted, limiting the passive migration of commensal house mice between valleys. So far, not a single hybrid has been recorded between races carrying robertsonian fusions among the 143 mice that we have karyotyped. We found extensive chromosomal heterozygosity in Funchal, however, probably because of the recent immigration of standard diploid mice through the commercial port.

Hybrids between any of the Madeiran mouse races carrying robertsonian fusions would be sterile or infertile owing to the complex chromosomal configurations that would be produced at meiosis7,8. Our results indicate not only that accelerated rates of radiation can occur without involving adaptive processes, but also that chromosomal evolution can be an efficient mechanism of isolation, as several reproductively isolated chromosomal races have appeared in less than 500 years.