264 



C. SKOTTSBERG 



surprisinj^ that the geographically isolated Hawaiian species are closely related to 

 the species of Juan I^>rnandez. We know other examples of this connection, but 

 to construct a route between Hawaii and Juan Fernandez meets with serious obsta- 

 cles, and Pa)igue may have reached Hawaii along a quite different route. The 

 evolution and differentiation of the subgenera very likely took place in Antarctica. 



Scilixni-KR (Pflanzenreich) expressed his opinion of the history of Halor- 

 rhagidaceac thus: "Aus der geographischen Verbreitungder ursprlinglichsten Halor- 

 rhagaceen, namlich der Gattung Halorrhagis, ist mit Sicherheit zu folgern, dass die 

 Familie antarktischen Ursprungs ist." It is not easy to see how he arrived at this 

 conclusion, though I think it is correct, because at that time the endemic species 

 of Juan Fernandez were unknown; they passed as //. erecta, a New Zealand en- 

 demic, and this, incorrectly attributed also to Juan Fernandez and Chile, had been 

 carried there on purpose: "Der Standort auf Juan Fernandez und in Chile ist kein 

 urspriinglicher, sondern durch die auch in Neuseeland erfolgende Verwendung der 

 Pflanze als Futterptianze erklart." But there was no Ilaloi'rJiagis on the mainland, 

 and the species indigenous in Juan Fernandez were not used as forage, let it be 

 that the introduced animals eat them. Halorrhagis, eminently Australian (59 out 

 of 80 species) and with 7 species in New Zealand, extends north to Indomalaya, 

 IVIicronesia, S.F. China and Japan and east to Rapa (not known to Schindler) 

 and Juan h^rnandez — see Ti'YAMa's map of distribution [2yi]. The genus "well 

 illustrates the not infrequent extension of an Australasian group far north of the 

 equator, and the much rarer condition of occurrence in Juan Fernandez but not 

 in continental America" {log. 108). HalorrJiagis shows the same distribution pat- 

 tern as ScDitaluDi. 



Coitclla is essentially African, see above p. 207; the distribution is tricentric 

 with some remote stations. The widespread C. asiatica is scattered over a broad 

 belt but not reported from America. Very likely the Hydrocotyloideae, a sepa- 

 rate family according to some authors, are of Antarctic origin, but C. triflora 

 may well have reached Juan Fernandez with the traffic from Chile. 



The peculiar endeinic species of Eryjigiuui differ so much in habit from all 

 other species of this large and widespread genus that they have been referred 

 to a se{)arate genus, a rank to which they are not entitled, liryvgiuni concen- 

 trates in tiie Mediterranean region and in tropical South America, where also the 

 island species Ijelong, in sj)ite of their arborescent habit; to quote TURMKI. [26g. 

 130): "Se rattachant a la Region chilienne, on cite les especes de Juan Fernandez 

 li. hupleuvoidcs, sai'copJiylluiii et hiaccessuni . . . , plantes, du moins pour les deux 

 premieres, arbustives s'o[)posant radicalement aux autres especes des territoires 

 voisins"— but viacccssuni also is a dwarf tree, more so I would say than sarco- 

 pliyllufti. I doubt that they descend from herbaceous continental forms; they be- 

 long to an ancient type and find their proper place with the neotropical element. 



The tribe Apioideae-Amnfineae is very widely spread, with a concentration 

 in the \orth hemisphere, Apiuni distributed also in the south temperate zone, and 

 A. gyai'eoh)is L. is frecjuently regarded as bipolar species. I don't believe that 

 any of the southern forms should be included; they are, however, in need of 

 revision. A. fcDicDidcziajiuin is a well-marked species, related, but not very closely, 



