REVISION OF ANACYCLUS 101 



satellited chromosomes. This increase has been correlated with increases in the degree of chromo- 

 some banding (Schweizer & Ehrendorfer, 1977) and the self-compatible breeding system (Uitz, 

 1970; Humphries, in press, a). 



Phytogeography 



The approximate total range of Anacyclus is illustrated in Fig. li and the distribution of all the 

 species on the maps in Figs 13, 15, 18, 20, 22, 23, 25 and 27 respectively. The main concentration of 

 species is in the Maghreb region of north Africa, particularly in Morocco. It is not the concentra- 

 tion of species that is the main point, however, but rather how the present species relate to one 

 another in terms of their distribution. It can be seen that several taxa, e.g. A. radiatus subsp. 

 radiatus, A. clavatus, A. homogamos and A. x valentinus, are pernicious weeds with an apparently 

 sympatric distribution in parts of their range, particularly in Morocco, Spain and France. How- 

 ever, for the most part, is it obvious that some species e.g. A. linearilobus, A. latealatus and 

 A. nigellifolius, have a discrete, sometimes small vicarious pattern of distribution. It is not 

 possible for this pattern to correspond with a general concept of a centre of origin with subsequent 

 dispersal; but instead it indicates an allopatric process of gradual migration of ancestors with 

 repeated isolation and vicariance which has eventually led to the present-day pattern of distribu- 

 tion. The importance of this concept, originally proposed by Croizat et al. (1974) and elaborated 

 by Platnick & Nelson (1978) and Rosen (1978), has already been elegantly described by Bremer 

 (1976, 1978) with reference to South African plant distributions; and it seems that a hypothesis 

 that sister monophyletic groups will have vicariant distribution patterns is a general concept 

 applicable to divergent evolutionary situations. 



Some examples of vicariance are more obvious than others. A. radiatus subsp. radiatus and 

 A. radiatus subsp. coronatus provide us with a good example of two evolving, vicarious sister taxa 

 (Fig. 18). Their ancestor probably had a continuous distribution from north Morocco right down 

 to the Ifni gap on the south-west Atlantic coast. Subsp. coronatus, the white-liguled form, has a 

 south-west-Moroccan distribution and is particularly prominent in the Sous valley and the Atlantic 

 seaboard from Ifni to Mogador. At the north end of its distribution there is a quite marked 

 transition to the yellow-liguled form - the more widespread north Moroccan and western 

 European/Mediterranean subsp. radiatus. It is interesting that the localized intermediates in the 

 Mogador region have been given a range of different names reflecting their transitional nature 

 (A. medians Murbeck, A. submedians Maire, A. radiatus var. ochroleucus Ball and A. radiatus var. 

 typicus subvar. concolor Maire). Their ancestor may well have had a continuous distribution 

 throughout their distribution range. Another example, this time of two vicarious species, is 

 provided by A. latealatus and A. nigellifolius. They are both annuals growing on the rocky 

 hillsides in the eastern Mediterranean (Fig. 27). A. latealatus is known only from the southern 

 Turkish steppe near Tefenni, whereas A. nigellifolius has a more widespread distribution in 

 southern Anatolia, Persia and the Lebanon. Such a pattern might be interpreted as quantum 

 speciation (Grant, 1971), where a small peripheral population has budded off from the ancestral 

 species. A similar interpretation might be applied to A. linearilobus, a narrow endemic from 

 northern Algeria (Fig. 25) which is the sister species to the more widespread weeds A. homogamos 

 and A. clavatus. 



A third and distinct vicariance pattern at a higher group rank within the genus is that of the 

 perennial varieties of A. pyrethrum (sect. Pyrethrarid) and all the annuals, which together form a 

 monophyletic group (sect. Anacyclus). A. pyrethrum is an upland species and a characteristic 

 component of the natural treeless, subalpine meadows and rocky habitats of the Atlas mountain 

 ranges of central and north Morocco and in Algeria (Fig. 13). By contrast, all the annuals occupy 

 disturbed, open, lowland habitats except those few species, e.g. A. homogamos, which have 

 invaded mountain habitats as roadside weeds. There is therefore a significant distribution pattern 

 resulting from the different ecological requirements of the sister groups. 



The answer to the question of the origin of the genus Anacyclus will not be found by a search 

 for its centre of distribution. Instead it is much more worthwhile to ask the question: what was the 



