Chemical evolution models, differing in the nucleosynthesis prescriptions (yields) for carbon, nitrogen and oxygen, have been computed for the Milky Way and Andromeda (NGC 224). All models fit the observed O/H gradients well and reproduce the main characteristics of the gas distributions, but they are also designed to do so. The N/O gradient for NGC 224 cannot be reproduced without ad hoc modifications to the yields and a similar result is obtained for the Milky Way N/O gradient, although in the latter case the slopes of the gradients obtained with unmodified yields are consistent with the observed gradient. For the C/O gradients (obtained from B stars) the results are inconclusive. The C/Fe, N/Fe, O/Fe versus Fe/H, as well as C/O versus O/H trends predicted by the models for the solar neighbourhood were compared with stellar abundances from the literature. For O/Fe versus Fe/H, all models fit the data, but for C/Fe, N/Fe versus Fe/H and C/O versus O/H, only modified sets of yields provide good fits. Since in the best-fit model, the yields were modified such that carbon should be primarily produced in low-mass stars, it is quite possible that in every environment where the peak of star formation happened a few Gyr back in time, the winds of carbon stars are responsible for most of the carbon enrichment, although models with a significant contribution from high-mass stars cannot be ruled out. In the solar neighbourhood, almost two-thirds of the carbon in the interstellar medium may come from carbon stars. Finally, the challenges met by stellar evolution and nucleosynthesis modelling due to this 'carbon star hypothesis' for the origin of carbon are discussed. It is suggested that a mass-loss prescription where the mass-loss rate depends on the carbon excess may act as a self-regulating mechanism for how much carbon a carbon star can deliver to the interstellar medium.
2008. Vol. T133, 014027- p.