Evolution of enlarged body size of coal tits Parus ater in geographic isolation from two larger competitors, the crested tit Parus cristatus and the willow tit Parus montanus, on six Scandinavian islands.

Here, we report that on six widely separated Scandinavian islands, the coal tit Parus ater has evolved morphologically in the direction of two absent competitors, the crested tit P. cristatus and the willow tit P. montanus, to the effect that it is up to 10% larger in linear dimensions than conspecifics on the adjacent Swedish mainland, where all three species coexist. The large size is genetically determined, as ascertained by clutch exchange experiments between island and mainland nests. We conclude that the increased size of P. ater in places where it is geographically isolated from its larger congeners is the result of evolutionary adaptation, due ultimately to relaxed interspecific competition. On the islands, P. ater has evolved into a medium-sized generalist, with selection pressures likely governed by the following causal relationships. When competitors are lacking, P. ater takes over the foraging space of the absentees. The enlarged food base allows higher population densities, which intensifies intraspecific interference competition. This, in turn, selects for increased body size. When P. ater coexists with its larger congeners, it occupies peripheral foraging sites in trees, which requires excellent manoeuvrability and energy-expensive locomotion modes. Reduction of body size increases locomotor capacity for mechanical and aerodynamic reasons and lowers energy consumption, so small size is favoured in sympatry. But in geographic isolation, P. ater exploits the tree periphery less and the inner tree regions more, and it also adopts the easier locomotion modes of the absent species. Therefore, selection for manoeuvrability and a small body size is relaxed. The new selection regime shifts the balance between opposing selection forces towards a larger body size. We were able to test 11 alternative hypotheses and available evidence conclusively eliminates them all. As a result, here, evolution could be predicted regarding both direction and amount of change.