31 



the Drcpaiiid birds thernselvos, indicating' likewise an extremely ancient oecnpa- 

 tiou of the islands, and as the Drcprniid l)irds are the pride of the Hawaiian 

 ornithologist, so are the Lobelioidcac of the Hawaiian botanist." Perkins again 

 says: "the development of the extreme forms of these birds is not comprehen- 

 sible without a knowledge of the island flora." 



"A series of observations made on one of the must siiperl) of the Lobelioideae 

 showed that it could only be fertilized by these highly specialized birds. In this 

 species the pollen is mature before the stigma is exserted, by which time the 

 pollen h.as vanished. The latter cannot be wind-borne because it is shed in a 

 viscid ma.ss on contact and so is constantly deposited on the bird's forehead, from 

 which it is difficult to remove it." To this the v.-riter would remark that birds 

 are not essential to the pollination of Hawaiian Lobelioideae though no doubt 

 they are important factors in pollination. In the Lobelioideae especially in the 

 Hawaiian specie.s, we find at the apex of the style immediately below the stig- 

 matie lobes and usually on the lower surface of the stigmatic lobes, rows of hair, 

 which are on a level with the base of the tube of the anthers at the time of the 

 pollen-discharge. The pollen remains in tiie tube of the anthers as the anthers 

 are united and thus the pollen cannot be disseminated. At the time of the 

 pollen discharge within the tulie. the style has usually not reached its full length, 

 but it continues to grow and as it pushes through the anther-tube it brushes the 

 pollen with its rows of hair out of the tube, and there the pollen remains adhering 

 to the Iiristles or bunches of hair, with which usually the two lower, or in some 

 cases all five anthers are fitted out at their apices. It can be seen that dichogamy 

 is not so pronounced that autogamy is not possible. The proterandrous anthers 

 are often eaten by insects or liii'ds ( .' ) and the pnllen is llins premjiturely dis- 

 charged. 



The base of the corolla tubes, especially those of the large flowering species, 

 is usually fillrd with great iiiunbrrs of Bnicln/jK /iliis (small insects with short 

 wings) and the Dii panid birds may find in them a source of insect food. Dr. Per- 

 kins seems not to have mentioned that fact, though he has often, on being invited, 

 picked off specimens of lirarlniix pUix from dried herliarium siH'ciniens of the 

 rarer and new Hawaiian Lobelioideae. When one l)reaks flowering branch(>s of 

 Ci/aveae or Cleriiioiitiae especially those with large flowers, the inflorescences 

 become alive with Braehijjicplus as well as small Carabidae. 



We can judge from these remarks that the T^obelioideae nnist certainly belong 

 to one of the oldest grouiis of jilanfs inhabiting this archipelago. Long before 

 there were any DrepanUl birds the Lobelioideae must have t>ecurred in these 

 islands, to lie sure not in such numbers, but perhaps in isolated individuals with 

 structural characters probably ditt'erent from those which are now existing. 

 Since the Drepanid birds themselves show a relationship to American birds, we 

 must look to the ancestor of the Brepanideae of today as the possible agent of dis- 

 persal of tjie baccate Lobelioideae. That the Drepatiid birds and I^obelioideae 

 had a more or less simultaneous development can again be surmised by the fact 

 that, in some of the Dv poiiid biivls we slill find individual variation as regards 

 the length of the bill. 



