A principal component analysis approach to the star formation history of elliptical galaxies in compact groups

Ignacio Ferreras 1 2 Ofer Lahav 2 0 Max-Planck-Institut fu r Astronomie , Koenigstuhl 17, D-69117 Heidelberg , Germany 1 Department of Physics, King's College London , Strand, London WC2R 6LS 2 Department of Physics and Astronomy, University College London , Gower Street, London WC1E 6BT 3 Instituto de Astrof sica de Canarias , Calle V a La ctea, E-38200 La Laguna, Tenerife , Spain 4 INPE/MCT, Avenida dos Astronautas 1758, Sa o Jose dos Campos , SP 12227-010 , Brazil A B S T R A C T Environmental differences in the stellar populations of early-type galaxies are explored using principal component analysis (PCA), focusing on differences between elliptical galaxies in Hickson Compact Groups (HCGs) and in the field. The method is model-independent and purely relies on variations between the observed spectra. The projections (PC1, PC2) of the observed spectra on the first and second principal components reveal a difference with respect to environment, with a wider range in PC1 and PC2 in the group sample. We define a spectral parameter ( 0.36PC1 PC2) which simplifies this result to a single number: field galaxies have a very similar value of , whereas HCG galaxies span a wide range in this parameter. The segregation is found regardless of the way the input spectral energy distributions (SEDs) are presented to PCA (i.e. changing the spectral range; using uncalibrated data; subtracting the continuum or masking the SED to include only the Lick spectral regions). Simple models are applied to give physical meaning to the PCs. We obtain a strong correlation between the values of and the mass fraction in younger stars, so that some group galaxies present a higher fraction of them, implying a more complex star formation history in groups. Regarding 'dynamically related' observables such as a4 or velocity dispersion, we find a correlation with PC3, but not with either PC1 or PC2. PCA is more sensitive than other methods based on a direct analysis of observables such as the structure of the surface brightness profile or the equivalent width of absorption lines. The latter do not reveal any significant variation between field and compact group galaxies. Our results imply that the presence of young stars only amounts to a fraction of a per cent in its contribution to the total variance, reflecting the power of PCA as a tool to extract small variations in the spectra from unresolved stellar populations. - The connection between mass assembly and star formation provides the best approach to building a comprehensive picture of galaxy formation. In this respect, massive elliptical galaxies are ideal probes since their light is mostly dominated by old stellar populations, whereas their mass and morphology require late assembly. Both old ages and late assembly can be reconciled within the standard concordance cosmology. However, we currently lack a method to give an accurate estimate of the star formation history (SFH) from photospectroscopic data. There are various factors that contribute to decouple the evolution of the total and stellar mass content of a galaxy. Most notably, the effect of feedback on star formation can make these two components follow divergent paths: for instance, a highly efficient mechanism of star formation at early stages can lead to dry mergers at later times (Bell et al. 2004). Indeed, the so-called downsizing trend (see e.g. Cowie et al. 1996; Treu et al. 2005) in which lower mass galaxies contain most of the global star formation at later times supports this effect. The environment is another factor which can play an important role in shaping the SFHs of galaxies. Clusters of galaxies trace high-density regions that collapsed earlier and present galaxies with a hostile environment in which gas can be removed by, for instance, ram pressure, galaxy harassment and tidal interaction (e.g. Haynes 1989). Furthermore, the high virial velocities in clusters prevent mergers from taking place. This effect is dependent on the cosmic epoch and might be a non-trivial one to accurately estimate. The association between compact groups (CGs) and clusters has been investigated by Rood & Struble (1994) who found that 75 per cent of Hicksons groups seem to be associated to structures such as loose groups and clusters, indicating that CGs are actually part of the same observed hierarchy from isolated galaxies to superclusters. More recently, de Carvalho et al. (2005) examining a sample of CGs at a slightly larger redshift than Hicksons found that there is an excess of CGs within one Abell radius of the nearest cluster, over a random distribution (32 per cent). Moreover, they found a marginal excess of CGs related to rich clusters relative to poor clusters, which might be important for considering properly the environment where CGs live in. Therefore, it is of paramount importance to establish the connection between CGs and the large-scale structure. In this context, it is important to remember that groups are the dominant type of structure found in the Universe (Nolthenius & White 1987) making them especially important from the cosmological viewpoint. The effects of the environment on the galaxian properties have been studied over more than two decades (Guzman et al. 1992; de Carvalho & Djorgovski 1992) and the issue is still debatable. Recent works regarding differences between early-type galaxies with respect to environment have found controversial results. While some indicate that environment does play some role (de Carvalho & Djorgovski 1992; Ziegler et al. 2005), others find no difference at all (e.g. Bernardi et al. 1998; de la Rosa, de Carvalho & Zepf 2001a). Also, studies of the stellar population of early-type systems indicate that they are older and more metal poor in CGs and clusters than in the field (Rose et al. 1994; Proctor et al. 2005; de la Rosa et al. 2001b; Mendes et al. 2005), suggesting that Es in dense systems experience a truncated period of star formation, which ultimately affects their chemical enrichment history, as opposed to their counterparts in the low-density regime. This paper presents an alternative approach to the analysis of unresolved old stellar populations, focused towards a comparison of the SFH between CGs and the field. The current epoch of large surveys provides a fertile ground for statistical techniques such as PCA (Madgwick et al. 2002), ICA (Lu et al. 2006), IB (Slonim et al. 2001) or ANNs (Naim et al. 1995). We show here the capabilities of PCA to disentangle the SFH of early-type galaxies. The structure of this paper is as follows. Section 2 presents the sample and Section 3 explains the technique of principal component analysis (PCA) applied to spectra. Section 4 gives the results for our sample, which are put in context with physical models in Section 5, followed by a discussion in Section 6. 2 T H E S A M P L E Our sample comprises 30 elliptical galaxies split into 18 located in the cores of Hickson Compact Groups (...truncated)