CH. XV] ENDEMISM AND DISTRIBUTION: SPECIES 167 



required for dying out), or advantageous (as required for local 

 adaptation), these facts, we repeat, are very much against any 

 explanation that is based, as are the two first named, upon 

 natural selection. Further, upon these suppositions it is impos- 

 sible to make any of the predictions that have already been so 

 successful!)^ made. 



There remains the third hypothesis, that in general endemics 

 are species so young that they have not yet had time to spread 

 to any great extent, or in other words that they are in general the 

 most recent appearances of species in the genera to which they 

 belong. Only in some such way can one explain the appearance 

 of such maps as those given above for Doona or Ranunculus, or 

 the "hollow curves" of distribution. No valid evidence has yet 

 been brought up to show that this is not the correct view to take 

 of the existence of the majority of endemics. There can be little 

 doubt, however, that quite an appreciable number of existing 

 species must be looked upon either as relics, or as local adapta*^ 

 tions. The relics may or may not be dying out (cf. rephes to 

 objections, pp. 88 to 94). The local adaptations must, of course, 

 be looked upon as simply a special case, i.e. as species which 

 appeared at first (as all species, to survive at all, must do) as 

 eminently suited to the local conditions that obtained at their 

 birthplace, but which have not been able to spread far, by reason 

 of ecological boundaries caused by changes of conditions at a 

 very short distance. 



There are many points in fa^'our of tliis third hypothesis. It 

 explains as well as the other two all the phenomena that they 

 were able to account for, and also very many to which they were 

 quite inapplicable, as, for example, the eleven given on p. 166. 

 It also enables us to make predictions about distribution, Avhich 

 an examination of the facts shows to be justified, and it has 

 already been successfully employed in this way nearly a hundred 

 times. Under these circumstances, Age and Area may perhaps 

 be regarded as at any rate possessing a greater basis of probability 

 than either of the two hypotheses based upon natural selection. 



Summary 



It is shown that no real difference can be pointed out between 

 endemic and non-endemic species (or genera). The former are 

 frequent upon mountains, upon islands, and in isolated pieces 

 of country, or in regions in which dispersal is very slow, or 



