Paige and Whitham (1987) showed that browsed plants of Ipomopsis aggregate had greater seed production than unbrowsed plants. This study has been hailed as the first unequivocal support for Owen and Wiegert's (1976) prediction that consumers can increase plant fitness (Owen 1990). These surprising results touched off a number of recent commentaries about the interpretation of selective forces exerted by herbivores on plants (Crawley 1987, Leather 1988, Fitter 1989, Westoby 1989, Owen 1990). Most of this discussion has centered on the interpretation of yield increase or compensation in response to herbivory as an adaptive response that has evolved as a consequence of natural selection due to herbivory. In wrestling with this interpretation, Paige and Whitham (1987) write: That plants should evolve to negate the deleterious effects of herbivory is obvious; that they should evolve to depend on herbivores to achieve their greatest fitness through overcompensation is less clear (p. 414). In addressing this quandary, Crawley (1987) suggested that, the predictability of herbivore attack on the primary shoots is sufficiently high, it is not difficult to imagine that an evolutionary advantage might accrue to a strategy of restrained early reproduction, leading to the production of single, primary shoots, followed by vigorous multistemmed regrowth following herbivore attack. However, this argument requires that attack on the regrowth shoots is predictably low. Even with this plethora of 'ifs' the story fails to explain why ungrazed plants do not put on a second burst of shooting once the risk of herbivory has passed, for, if it is advantageous to produce new flowering shoots following defloration of the primary shoot, then surely it is just as advantageous to produce them when ungrazed? (p. 168). According to Crawley (1987), one of the most puzzling questions here is, Why do we not witness the prospering of plants that branch from the base when they are not grazed? (p. 168). We think that the answer to this question requires a shift in focus concerning the role of natural selection. Several studies have shown that plants which have had the shoot apex removed produce greater yields than plants with the shoot apex left intact (Harris 1974, Clifford 1979, Binnie and Clifford 1980, Tayo 1980, 1982, Inouye 1982, Argall and Stewart 1984, Sheldon 1986, Benner 1988, Wein and Minotti 1988). The reason for these yield increases appears simply to involve release from the constraints of apical dominance which, when left uninterupted, can be regarded as an adaptive consequence of selection from competition for light. In a recent paper, Maschinski and Whitham (1989) recognized the important role of the disruption of apical dominance in producing the overcompensatory response in Ipomopsis. The critical question then becomes, does yield stimulation following herbivory represent an adaptive consequence of selection pressures from herbivory or an indirect consequence of selection from competition? The mechanism of apical dominance has been studied extensively by physiologists (Phillips 1975, Hillman 1984). Surprisingly however, the ecological and evolutionary implications of apical dominance have been scarcely addressed. Presumably, the evolutionary advantage of apical dominance is conferred by a disproportionate meristem commitment to vertical extension because, under crowded conditions, dominance is commonly achieved by those neighbours that can get the tallest, the fastest. Since lateral meristems are suppressed however, apical dominance may incur an overall 'cost' in terms of reduced maximum potential yield for plants growing free from the suppressive effects of neighbours. Harris (1974) writes, dominant plants remain programmed for apical dominance even when growing without competition and with non-limiting resources. The consequence is that in these situations, which are mostly man-made, the growth of the apically dominant plant is restricted to the maximum that can be achieved by the terminal. Removal of the terminal releases the lateral buds from correlative inhib-