90 C. J. HUMPHRIES 



genera and the southern hemisphere Cotuleae), but since there is no evidence that Anacyclus is 

 unnatural in its present circumscription, this observation must, for the time being, be considered 

 a uniquely derived condition and serve to unite the two genera. The monotypic Leucocyclus 

 formosus is a herbaceous endemic perennial from Turkey which resembles Anacyclus in most 

 respects except that the leaves are almost vermiform, with the small segments suboppositely 

 connected to the rhachis and divided into spinulose-dentate lobes. These genera have approxi- 

 mately vicarious distributions: Anacyclus occupies the southern, western and south-eastern 

 Mediterranean areas, particularly in the mountains and the dry, disturbed lowlands, while 

 Leucocyclus grows in the lowland montane areas in south central Turkey (Grierson, \915b). 



The tribe Anthemideae is usually divided into two subtribes; Anthemidinae Dumort. and 

 Chrysantheminae Less. The Anthemidinae normally have chaff-like, scarious receptacular scales 

 invariably subtending the ovary and, to some degree, the floret as well. The Chrysantheminae 

 by contrast lack scales. Since this division is clearly artificial (Humphries, 1976a), it is inapprop- 

 riate to become involved here in a detailed discussion of the subtribal classification of the 

 Anthemideae, as all taxa allied to Anacyclus have receptacular scales and do not have any close 

 allies without scales. It is therefore more appropriate to discuss the genera with which Anacyclus 

 has been allied from one time to another and consider its relationships on more recent evidence. 



As described in the historical section, species of Anacyclus have tended to be confused with taxa 

 of the Anthemis assemblage rather than with any other group. Recently, Grierson (\915a) has 

 suggested that Anacyclus valentinus L. (incl. A. homogamos), A. pyrethrum and A. monanthos are 

 similar to the species of Anthemis section Cota in having a subterminal corolla. Also, the per- 

 sistent tubular part of the ray floret corolla on the ripe cypselas of Anthemis arenicola Boiss. and 

 A. davisii Yavin, a rare feature in Anthemis, are also found in the radiate taxa of Anacyclus. 

 These trivial convergences or parallelisms are of little consequence in the formulation of phylo- 

 genetic hypotheses. Detailed studies of the fruits in genera of the Anthemideae also reveal a 

 number of parallel morphological trends which tend to obscure the genealogical relationships in 

 the group. As pointed out by Humphries (1977), such features include the slightly compressed 

 cypselas of Anthemis section Cota, which differ during their development in having 7-22 ribs in 

 the pericarp wall (Wagenitz, 1968; Kynclova, 1970; Reitbrecht, 1974) and 5 vascular bundles 

 (Humphries & Innes, unpublished), rather than 7-22 bundles as was wrongly assumed by the 

 above authors. 



Examination of fruit and corolla characters on a wider scale demonstrate that no case exists 

 for making Anthemis, itself a polyphyletic assemblage, the sister group of Anacyclus. An alterna- 

 tive hypothesis originates from the phytochemical work of Greger (1977, 1978), where he suggests 

 that Anacyclus has a cyanogenic glycoside and flavonoid phytochemical profile more closely 

 related to Achillea and its allies than to species of Anthemis. Genera of the Achillea group regularly 

 have tiny cypselas with distinct lateral ribs rather than wings, as well as unusually thin pericarp 

 walls. It is tempting to offer a hypothesis that a group comprising Anacyclus, Leucocyclus, Helio- 

 cauta Humphries, Sclerorhachis Rech. fil. and Achillea L. may be recognized, as these genera all 

 have a reduced vascular system with two lateral vascular bundles in the pericarp wall (Fig. 6), 

 but studies on this aspect are in a preliminary state. 



It is fairly clear that difficulties in forming evolutionary hypotheses of relationship at the generic 

 level and above will persist until such a time as monophyletic groups are identified. The only 

 recent attempts at forming natural generic groupings can be found in the work of Reitbrecht 

 (1974; see also Hey wood & Humphries, 1977), who considers that the Anthemideae consists of 

 seven provisional groups. His 'Matricaria-gruppe' is taken to comprise Anthemis, Anacyclus, 

 Chamaemelum Miller, Cladanthus Cass., Matricaria L., Tripleurospermum Schultz Bip., Oto- 

 spermum Willk. and Daveaua Willk. ex Mariz. There seems to be no improvement in the above 

 classification over the former groupings of Bentham (1873) and Hoffman (1894) since it is a 

 paraphyletic assemblage without any definite character states which adequately link the genera. 

 Such genera as Achillea are excluded, as they are considered to have affinities with the wind- 

 pollinated structurally reduced members of the Artemisia group, as also are the southern hemis- 

 phere Sphaeroclinium and its allies, which Mitsuoka & Ehrendorfer (1972) have managed to 

 hybridize with northern hemisphere members of Matricaria. Nevertheless, it seems probable that 



