r/K theory classically predicts that offspring size should increase under density-dependent selection. However, this is questionable, being based on implicit rather than explicit assumption (the logistic model does not include offsring size as a parameter).
From recent models of optimal offspring size (Sibly & Calow, 1983; Taylor & Williams, 1984) it can be shown that density should select for larger offspring if density-dependence in the per capita rate of increase is mainly due to a reduction of the juvenile growth rate or survivorship. In contrast, density should select for smaller offspring if such density-dependence is mainly due to a reduction of adult fecundity or survivorship. Therfore, the outcome of selection cannot be predicted without precise knowledge of the density-dependence of age-specific reproduction and mortality rates.
To test the above models, genetically identical individuals of Simocephalus vetulus (Müller) were reared in a density gradient; density-dependence in the per capita rate of increase was shown to be mainly due to a reduction of the juvenile growth rate, thereby selecting for larger offspring; offspring size at birth appeared to be phenotypically plastic and to increase with density.
Models were therefore qualitatively supported. However, a discrepancy occurred in quantitative predictions; offspring were produced larger than predicted. Field and laboratory studies are suggested to address this.