Antarctic Plants in Polynesia 305 



said to show any special kind of adaptation. The wings of the 

 pod, generally supposed to increase floating capacity, are absent 

 in the two Juan Fernandez species; furthermore, experience 

 shows that the ripe pods open on the tree, to which they are 

 firmly attached, and discharge the seeds.^' " According to Guppy, 

 the seeds are buoyant. However, Edwardsia is not a seaside 

 genus, and, although its seeds sometimes may be caught by the 

 current, this so far has not led to an extension of specific range, 

 because local endemism prevails throughout. 



Gunnera has small, bright red drupes. In the Falkland Islands 

 birds feed upon G. magellanica and thereby spread it. In Juan 

 Fernandez, Chile, and Hawaii, where the Gunnerae are fre- 

 quendy seen growing on the banks of streams, the drupes are 

 carried away by running water. G. macrophylla is probably dis- 

 tributed in the same way (at least in Java), but whether this 

 habit has anything to do with the wide range of this species, 

 I do not know; all the insular species of Pangue are restricted 

 within a small area. 



Astelia has a scarlet- or orange-colored berry and seeds with 

 a hard but sometimes britde testa. It is an established fact that 

 birds eat the berries in Hawaii and New Zealand and very likely 

 they do so wherever these plants grow, but endemism is absolute 

 in this genus and, when a species is found in two remote stations, 

 it is represented by different forms. 



Coprosma, Nertera, and Coriaria have drupes. However, 

 Coprosma shows pronounced local endemism. N altera, if really 

 the same species all over the southern hemisphere and beyond 

 the equator, would strongly support the theory of endozoic over- 

 sea migration, but it is better not to draw far-reaching conclu- 

 sions from the distribution of this genus; a careful revision of 

 it will probably permit us to distinguish a number of endemic 



