222 L. HILLIS-COLESrVAUX 



colonize habitats which will certainly be unlike that familiar to the 

 parent clone. Finally, sex may well be the reproductive resort of indi- 

 viduals in the middle of the parent clone, even when the habitat 

 remains unperturbed. These individuals will gain no advantage in 

 fitness by producing copies of themselves since their surroundings are 

 already full of their own copies. For these crowded ones in the middle 

 of the clone the best hope of fitness will be to send out colonizing 

 propagules (made various by sex), or to flood the old clonal habitat with 

 new experimental genotypes, one of which just might be better than 

 the parental strain. 



It is important to notice that Williams' model is based on the postu- 

 late that fitness will be won in competition with other individuals or 

 clones. Diversity through sex is demanded, in the first instance, not by 

 the physical needs of new habitats but because a slight competitive 

 advantage in new or changed habitats may go to a fresh genotype. It is 

 on grounds of competitive advantage that sex should precede resting 

 and dispersal, or be the resort of individuals in the middle of clones. 



2. The Halimeda reproductive strategy summarized 



The commonest means of reproduction appears to be by vegetative 

 cloning. Most of this takes place by sending out "runners" of rhizoidal 

 filaments. Sand-dwelling forms thrust the filaments through the sand 

 so that each plant looks to be de novo and independent. Rock forms 

 spread more obviously from epicentres. Secondary forms of vegetative 

 reproduction are the sprouting of vegetative parts and the establish- 

 ment of pieces broken off from thalli. 



Sexual reproduction is sporadic and almost always involves the 

 death of the entire thallus. The usual pattern is that a sexual episode is 

 separated from the next by one or more vegetative generations. Sex 

 involves the production of an extremely large number of flagellated 

 gametes to the manufacture of which the entire resources of the parental 

 protoplasm have been devoted. This gamete swarm is to be expected to 

 yield a large swarm of zygotes. 



The data on what happens to the zygotes are extremely scanty, but 

 nonetheless suggestive. The zygote appears to develop very slowly, 

 taking months for the initial growth. And the result of this growth is a 

 filamentous mat in no way like the parental thallus. After 12 months in 

 a laboratory container no more development is observed. The develop- 

 ment time of a zygote, therefore, appears to be of the order of one to 

 two vegetative generations. It is also of a length to serve as a resting 

 stage to pass the adverse seasons of an annual cycle. 



