4.4.4 The Wolf-Rayet galaxies

Wolf-Rayet (WR) stars are considered to be highly evolved descendants of the most massive O-stars. They are extreme Population I stars and have spectra characterised by broad emission lines resulting from dense, high-velocity winds. These stars are detectable in external galaxies by their prominent emission lines at around 4650-4690 Å (the ``Wolf-Rayet bump''). This bump has been detected in many emission line galaxies (Allen et al. 1976, Kunth & Sargent 1981, Kunth & Joubert 1985, Conti 1991, Vacca & Conti 1992; see Schaerer et al. 1999 for the latest updated catalogue), providing a new insight on the process of massive star formation in metal-poor galaxies. Arnault et al. (1989), Cerviño & Mas-Hesse (1994), Meynet (1995) and Schaerer & Vacca (1998) have discussed the dependence of the WR bump strength on the parameters that define the star-forming episodes (metallicity, age, IMF slope, etc.). The two most interesting properties of the Wolf-Rayet bump is its strong dependence on metallicity and the constraints it can impose on the age of the cluster. Since the WR phenomenon is tightly coupled to the generation of strong stellar winds, its incidence decreases significantly with decreasing metallicity, so that at Z = 1/20 Z only very massive stars (initial mass > 80 M) might become WR stars. This small mass range implies that the detection of the WR bump in low metallicity galaxies can provide important information on the upper mass limit of the IMF.

The relative population of WR to O stars is usually measured through the L(WR) / L(H) ratio, the luminosity of the WR-bump over the H-luminosity. To compare with model predictions it is necessary to integrate over the whole ionised region which poses some observational technical problems. The measurements of this ratio might also be strongly affected by differential extinction. Since L(WR) is of stellar origin, it should be affected by the same extinction as the stellar continuum. On the other hand, L(H) is of nebular origin and might suffer from a larger amount of extinction. Ignoring this effect may lead to a significant overestimation of the L(WR) / L(H) ratio. Schmutz and Vacca (1999) have questioned the use of the 4640Å emission feature which may not entirely be due to WR stars but to large numbers of O stars or contamination from other nebular lines. Observations show that very short bursts are compatible in general with a Salpeter IMF and a large upper mass cut-off. Recent results by Cerviño (1998) show that if a significant fraction of massive stars are formed in binary systems, mass transfer episodes can lead to the formation of WR stars during longer periods of time than predicted by models based on the evolution of single stars alone. Therefore, age calibrations through WR features has to be taken with caution.

Schaerer (1996) has evaluated the effect of the evolution of the WR stars population on the He II narrow emission line at 4686 Å , by combining model atmospheres accounting for stellar winds with evolutionary tracks. He concludes that for metallicities in the range Z = 1/5 Z to Z, a strong nebular He II emission line should originate in early WR phases when WC stars begin to appear. The He II emission line is indeed detected in a few objects with very young stellar populations, below 3 Myr, and therefore starting to produce WR stars (NGC 2363 and Mrk 36), in good agreement with the scenario proposed by Schaerer. This is also supported by the detection of WR stars in IZw18 (that came as a surprise in view of its very low metallicity) (see Fig 7 ; Legrand et al. 1997b, Izotov et al. 1997a). A last argument is the spatial distribution of the nebular HeII lines that follows that of the WR features (Legrand et al. 1997b, Maiz-Apellaniz et al. 1998, and De Mello et al. 1998). Other broad emission lines from C IV at 5805Å originating from WC stars (representing more evolved phases than WN stars) are currently observed (Schaerer et al. 1997, 1999). This indicates that bursts cannot be completely instantaneous in order for both WR types to be present. The observations of WR features in low metallicity objects is indeed a challenge to our understanding of the WR phases and forces to reassess the metallicity dependence of stellar winds, the binary channel for WR production and the effect of rotation onto the evolution of massive stars (Maeder and Meynet, private communication).