Emission lines in normal galaxies are powerful tracers of the physical processes associated with the galaxy interstellar medium and the formation of massive stars. They allow quantitative estimates of the star formation rates, as well as of the chemical abundances and physical conditions of the gas in galaxies (McCall et al. 1985; Gallagher et al. 1989; Kennicutt et al. 1994; Zaritsky et al. 1994; Barbaro & Poggianti 1997).
A census of emission line properties in integrated spectra of normal galaxies is particularly interesting in view of studies of galaxies at large redshifts, too distant to be spatially resolved. It provides useful information also for nearby galaxies, as it allows investigating the global properties of their emission. It is well known that emission lines tend to become more prominent as one goes from early to late galaxy types, and the degree to which the spiral arms are resolved into individual HII regions (where most of the emission lines in normal galaxies are produced) is one of the most important criteria for discriminating among the spiral types in the Hubble system (Hubble 1936; Sandage 1961). Kennicutt (1992a, hereafter K92a) has already discussed some properties of the emission lines in integrated spectra, mainly evaluating their reliability as quantitative tracers of the total massive star formation rate in galaxies. Lehnert & Heckman (1994) have examined the location of integrated spectra of galaxies in the standard emission line diagnostic diagrams used to classify emission line objects (Baldwin et al. 1981; Veilleux & Osterbrock 1987), concluding that the integrated emission in star-forming galaxies has a substantial contribution of a diffuse component, with physical properties different from those found in high surface brightness HII regions. Most studies, however, have focused on the properties of emission lines produced either in HII regions in the spiral arms of disk galaxies (e.g. McCall et al. 1985; Belley & Roy 1992; Oey & Kennicutt 1993; Zaritsky et al. 1994; Kennicutt & Garnett 1996; Roy & Walsh 1997) or in galactic nuclei (Heckman et al. 1980; Keel 1983; Ho et al. 1997). In particular, Zaritsky et al. (1994) have investigated the oxygen abundance properties of a sample of disk galaxies from the spectra of individual HII regions located at various galactocentric radii. They found that the characteristic abundance of the galaxies correlates well with both their morphological type and their luminosity.
In this paper we examine the trends of emission line properties in the integrated spectra of 15 normal, nearby spiral galaxies. The spectra come from Kennicutt's (1992b, hereafter K92b) spectro-photometric atlas.
The galaxies are ordered by galaxy spectral type (hereafter ST), obtained using Principal Component Analysis of the integrated properties of the continua and absorption features of 23 normal galaxies of all morphologies, in a manner similar to the one introduced by Sodré & Cuevas (1994, 1997) and several other authors (e.g., Connolly et al. 1995; Zaritsky et al. 1995; Folkes et al. 1996; Galaz & de Lapparent 1998). Such a procedure allows to define a spectral classification that correlates well with Hubble morphological types and present some advantages over the usual morphological classification. Firstly, it provides quantitative, continuous, and well defined types, avoiding the ambiguities of the intrinsically more qualitative and subjective morphological classification. Secondly, the sequence of galaxies obtained in this way is easier to model than the Hubble sequence, because the physical process behind galaxy spectra are better understood than those needed to explain galaxy morphologies. Thirdly, this classification can be applied to redshift surveys where no information is available on the galaxy morphologies, allowing to "recycle" data obtained for other purposes (de Lapparent et al. 1998; Bromley et al. 1998).
We will show that the patterns of variation of emission line properties along the sequence of normal galaxies are much more regular when considering the galaxy spectral type sequence than when using the morphological sequence.
This paper is organized as follows. In Sect. 2 we introduce the sample of normal galaxies, briefly describe the procedure used to obtain spectral types from integrated galaxy spectra, and present the relevant data for the emission lines in the blue/visible region of the spectra. In Sect. 3 we present the trends of equivalent widths and emission line ratios with galaxy spectral types. In Sect. 4 we interpret these trends, after a brief reminder on emission line theory and a comparison of the loci of integrated galaxy spectra with those of giant HII regions in classical diagnostic diagrams. Finally, in Sect. 5 we summarize our results.
© European Southern Observatory (ESO) 1999
Online publication: April 19, 1999