Oliver. — Vegetation of the Kermadec Islands. 127 



Homolanthus polyandrus, which ph^nt Mr. Cheeseman in 1887 (1 ; p. 172) 

 recorded as " not uncommon." The species is now almost extinct. 



In Denham Bay sheep eat down Sccevola gracih's, and in its place Stel- 

 laria media and Cerastium viscosum have become abundant. 



The invasion of a number of European and other plants has not in itself 

 much altered the physit)gnomy of the plant formations, but where ground 

 has been disturbed introdueed plants are quieker than indigenous species in 

 taking possession. Introduced species do not readily spread far into the 

 forest except in such places where the larger plants have deHberately been 

 cleared away, and here three new and distinct formations have been created. 

 These are the mcadoAVS described below. 



Ageratum conyzoides was introduced thirty years ago, and is now abundant 

 in several parts of the island. Possibly goats have assisted in spreading 

 this species, which is the only introduced plant taking a prominent part 

 in several of the plant associations. 



VI. The Plant Formations. 



All the plant formations on Sunday and Macauley Islands are more 

 or less modified through the depredations of introduced goats, which have 

 not only directly expunged certain plants from the forest, but in so doing 

 have made openings which have been partly filled by introduced species. 

 Hence in places it is difficult to imagine what the original forest was like, 

 especially as it is known that a handsome tree, Homolanthus polyandrus, 

 now absent, was once abundant in many places. The undergrowth perhaps 

 suffers most from the introduction of goats, which search out particular 

 plants whose elimination changes its appearance ; also, by the extermi)ia- 

 tion of Homolanthus polyandrus, and the suppression of young plants of 

 Pisonia Brunoniana, Nothopanax arboreum, and Cyathea Jcermadecensis, the 

 physiognomy of the forest may be entirely altered. 



In the Kermadec Islands climatic conditions favour the growth of forest ; 

 all other plant formations except meadow occurring there may be classed 

 as edaphic. The meadow formations ow^e their existence to the interference 

 of man, and are gradually being superseded by forest ; the swamp in Denham 

 Bay, also, appears to be drying up, and forest taking its place. 



The plant formations, then, naturally fall into tw^o groups — the climatic, 

 the character of whose vegetation is governed by atmospheric precipitations, 

 and the edaphic, whose vegetation is chiefly determined by the nature of 

 the soil (16 ; p. 161). In this paper the plant formations of the Kermadecs 

 are arranged according to their possible evolution, and are divided for 

 purposes of description into five groups — (1) coastal, (2) inland edaphic, 

 (3) forest, (4) young, (5) introduced. The climatic formations are included 

 in the third group, while groups (1) and (2) are edaphic. The last two 

 groups may be termed unstable formations. 



Supposing we go back to early Pliocene times, when the Kermadecs 

 were just appearing above the sea-surface, and try to imagine what the 

 conditions would be. The first plants to gain a footing on the new land 

 would be those able to stand wetting with sea-water, such as are now found 

 among coastal rocks.* Thus the coastal formations would be the first 



* On Krakatoa the first plants to appear after the destruction of the vegetation in 

 1883 were chiefly ferns (Treiib.. quoted by Schimi)er, 1(5 ; p. 185). but the island which 

 remained after the eruption was of some considerable area. 



