3.3.3. Changing Yields
If stellar nucleosynthesis yields change as a function of metallicity, then this will have a resultant effect on relative abundances. For example, recent stellar evolution models including mass loss via winds predict much larger yields of C relative to O (Maeder 1992, Woosley et al. 1993). If stellar winds are primarily radiatively driven, and dependent on opacity, then the stellar mass loss rates should depend on metallicity. Maeder (1992) presents an analysis of the expected yields from stellar evolution models which incorporate metallicity-dependent stellar mass loss. These models result in an increasing yield of carbon, and a decreasing yield of oxygen, with increasing metallicity.
Another possibility for changing yields would result from a variable IMF. In the present climate, it is not fashionable to consider IMF variations. First off, the IMF police will come and take you away, and no one wants that! But, seriously, there are many good observational arguments supporting an environment and metallicity independent IMF (Massey et al. 1995a, b, and Phil's chapter in this volume). However, a variable stellar initial mass function has been invoked to explain some properties of giant H II regions and starburst galaxies (e.g., Melnick, Terlevich, & Eggleton 1985). If one wanted to consider the effect of a metallicity dependent IMF (this is all hypothetical, of course), then it is easy to see that an increase in the proportion of high mass stars or an increase in the upper mass cut-off would lead to more heavy element production from a given population of stars. If the IMF is weighted toward more massive stars at low metallicities, then an increasing C/O ratio with metallicity would result from a change in the relative contributions of massive and intermediate mass stars. Now, I'm not suggesting that you consider this, but I thought I should mention it for completeness.