3.1 Stellar Evolution and Nucleosynthesis

As the IGM condenses into galaxies it contains the primordial abundances of H, D, ³He, ⁴He and ⁷Li. One of the beauties of some star-forming dwarf galaxies is that they can be so metal-poor that their abundance analysis bear strong cosmological implications. The current wisdom is that most of the element production in the Universe, apart from the early ``Big Bang'' nucleosynthesis, occurs in stellar interiors. Part of the products of stellar nucleosynthesis are released when stars die or in stellar winds from evolved stars, while another part remains locked up in stellar remnants: white dwarfs, neutron stars and black holes. These remnants may not be completely sterile since in a binary system, accretion onto a white dwarf may lead to a type Ia supernova.

The end products that enrich the ISM can be predicted from models of stellar evolution and nucleosynthesis. The returned mass in metals as a function of the initial mass is referred to as the stellar yield (Maeder 1992). The absolute and relative yield for different elements depend on stellar mass in a non-linear fashion. Moreover, the yields for the so called ``secondary'' elements depend themselves on the initial composition of the star. An example is ¹⁴N, although observations strongly indicate the need also for primary production. Furthermore the mass-loss history of stars needs to be accounted for, since elements that would have been subject to further nucleosynthesis might have been ejected, and also the effects of stellar binarity. Unfortunately, the yields for many elements are still very uncertain, also for important elements like C and N (Prantzos, 1998), and this has to be kept in mind when interpreting abundances and abundance ratios. The oxygen yield is among the best determined ones, still the uncertainty is around a factor of two.

Low mass stars (M < 1 M) are long lived and simply lock up part of the gas, since their lifetimes are longer than or comparable to the age of the Universe. On the other hand they contribute to light and have imprinted in their chemical composition the conditions of the ISM at the time and place they were formed. Intermediate mass stars (between 1 and 8 M) undergo dredge-up processes that significantly affect the C and N abundances and even the ⁴He abundance, and such stars are important contributors to these elements. Massive stars (M > 8 M) are short-lived and complete their evolution in less than 5 x 10⁷ years. As such a star collapses and becomes a neutron star its envelope is ejected in a supernova explosion, carrying away the earlier nucleosynthesis products and the ones resulting from explosive nucleosynthesis in the inner layers. This picture is still uncertain since the initial masses for which this is valid are not well known and it has been argued that some of the most massive stars may collapse into black holes, without an associated SN explosion. Before their dramatic end, massive stars undergo mass-loss processes via stellar winds as exemplified by Wolf-Rayet stars and other variable stars (e.g. luminous blue variables) that can carry away some CNO processed material, reducing the yield of O and increasing that of C, N and He. The efficiency of stellar winds depend strongly on their chemical composition (Maeder 1992) since metals increase the opacity. At low metallicity the mass loss is small and the resulting He, C and O production is insensitive to the initial stellar composition, although it can be affected by the tendency of mass loss to increase with metallicity. Hence one expects the yields of these elements to be metallicity dependent. In addition to stars losing part of their mass, a galaxy as a whole may be subject to mass loss, which will influence the ISM abundances, (see Sect. 3.3). Since different elements are produced in stars of different mass, they enrich the ISM on different timescales. Massive stars constitute the main source of oxygen and other -elements, thus these elements are ejected on short timescales. Also significant amounts of carbon and nitrogen are produced in massive stars. For iron, massive stars dominate, but on long timescales the contribution from SN-Ia produced in binary systems may be important.

The initial mass function (IMF, the distribution of stellar masses in a population of newly borne stars) is a critical issue. To predict the element production of a population of stars, the stellar yields have to be convolved with the IMF to form the net yield, defined as the mass of newly synthesised elements per unit mass locked up in remnants and long lived stars. Unfortunately, the IMF is not yet well determined even locally, and it is uncertain whether it is universal or depends on environment and metallicity. In most extragalactic studies the IMF is assumed constant in time. This assumption needs to be examined with care although no observational evidence has convincingly contradicted it by now. The strongest claim by Terlevich and Melnick (1981) for IMF variations with metal content of gas has never been compelling. There have been some suggestions that starburst galaxies should have an IMF biased towards massive stars or deficient in low mass stars (cf. Scalo 1990). However there is no direct evidence for this or for a low mass cut-off in giant H II regions like 30 Doradus or in globular clusters. Marconi et al. (1994) argue that the chemical evolution of starburst galaxies is well understood with a normal Salpeter-like IMF. Currently it seems that, for massive and intermediate massive stars, the IMF is reasonably well described with a power-law and a slope close to Salpeter's (1955) original value while it flattens, but does not cut-off, at masses below 1 M. For a discussion on the IMF, its derivation, and possible variations, see Scalo (1998). If the total mass involved in star formation at each instant is modest, the high mass part of the IMF will be badly populated and concequently predictions for nucleosynthesis and spectral evolution will be sensitive to statistical fluctuations.