Differences in carbon and nitrogen abundances between field and cluster early-type galaxies
ABSTRACT Central line-strength indices were measured in the blue spectral region for a sample of 98 early-type galaxies in different environments. For most indices (Mgb and in particular) ellipticals in rich clusters and in low-density regions follow the same index-sigma relations. However, striking spectral differences between field ellipticals and their counterparts in the central region of the Coma cluster are found for the first time, with galaxies in the denser environment showing significantly lower C4668 and CN2 absorption strengths. The most convincing interpretation of these results is a difference in abundance ratios, arising from a distinct star formation and chemical enrichment histories of galaxies in different environments. An scenario in which elliptical galaxies in clusters are fully assembled at earlier stages than their low-density counterparts is discussed. Comment: 12 pages, including 3 figures, accepted for publication in ApJL
arXiv:astro-ph/0305244v2 20 May 2003
Differences in carbon and nitrogen abundances between field and
cluster early-type galaxies
S´ anchez–Bl´ azquez P., Gorgas J., Cardiel N.1and Cenarro J.
Departamento de Astrof´ ısica, Facultad de F´ ısicas, Universidad Complutense de Madrid,
28040 Madrid, Spain
Gonz´ alez J.J.
Inst. de Astronom´ ıa, Universidad Nacional Aut´ onoma de M´ exico, Apdo-Postal 70-264,
M´ exico D.F, M´ exico
Central line-strength indices were measured in the blue spectral region for a
sample of 98 early-type galaxies in different environments. For most indices (Mgb
and ?Fe? in particular) ellipticals in rich clusters and in low-density regions follow
the same index-sigma relations. However, striking spectral differences between
field ellipticals and their counterparts in the central region of the Coma cluster
are found for the first time, with galaxies in the denser environment showing
significantly lower C4668 and CN2absorption strengths. The most convincing
interpretation of these results is a difference in abundance ratios, arising from
a distinct star formation and chemical enrichment histories of galaxies in differ-
ent environments. An scenario in which elliptical galaxies in clusters are fully
assembled at earlier stages than their low-density counterparts is discussed.
Subject headings: galaxies: abundances — galaxies: evolution — galaxies: clus-
ters — galaxies: formation — galaxies: stellar content
Despite the observational and theoretical efforts of the last decades, the evolutionary
status of early-type galaxies is still an unsolved problem. The stellar populations of nearby
1Also at Calar Alto Observatory, CAHA, Apdo. 511, 04044, Almer´ ıa, Spain
– 2 –
ellipticals preserve a record of their formation and evolution. In particular, the study of their
element abundance ratios should be a powerful discriminant between different star formation
histories (e.g. Worthey 1998). However, this last approach is still at its infancy. The pio-
neering works of late 1970s already revealed that abundance ratios in early-type galaxies are
often non-solar (O’Connell 1976; Peterson 1976). Since then, several studies have provided
compelling evidence of Mg/Fe overabundances in massive ellipticals as compared with the so-
lar ratio (Worthey, Faber, & Gonz´ alez 1992; Peletier 1989; Vazdekis et al. 1997), which have
been interpreted in the light of several possible scenarios based on the understanding that
Mg is mainly produced in type II supernovae (Faber, Worthey, & Gonz´ alez 1992; Worthey,
Faber, & Gonz´ alez 1992; Matteucci 1994). However, and in contrast with the above find-
ings, another alpha element, namely Ca, seems to be underabundant in ellipticals (O’Connell
1976; Saglia et al. 2002; Cenarro et al. 2003; Thomas, Maraston, & Bender 2003), challenging
current chemical evolution models (Matteucci 1994; Moll´ a & Garc´ ıa–Vargas 2000).
Several authors (Worthey 1998;Vazdekis et al. 2001) have also noted a strengthening
in other absorption line-strengths, in particular in the IDS/Lick C4668 and CN2 indices,
when compared with stellar population models predictions. The variations of these two
indices are mainly driven by C and N (in the case of CN) abundances (Tripicco & Bell
1995), which could suggest a possible enhancement of these two elements relative to Fe when
compared with the solar values. In contrast with Mg, C and N are mainly produced in
low- and intermediate-mass stars (Renzini & Voli 1981; Chiappini, Romano, & Matteucci
2003, although there are recent suggestions that most of the C should come from massive
stars). During the AGB phase, these stars eject into the ISM significant amounts of4He,
12C,13C and14N, enriching the medium from which new stars will be formed. Therefore,
it seems difficult to simultaneously reproduce the abundances of all these elements with a
simple chemical evolution scenario.
The problem of C and N abundances has been more thoroughly studied in the field of
globular clusters. An interesting puzzling problem is the existence of a CN dichotomy in
Galactic and M31 globular clusters (Burstein et al. 1984). Although this is a controversial
issue, recent works (Harbeck, Smith, & Grebel 2003) tend to favor the scenarios of different
abundances in the parental clouds against the ones that predict abundance changes produced
internally by the evolution of the stars (see Cannon et al. 1998 for a review).
Given the expected sensitivities of relative abundances to the star formation history of
ellipticals, their study in galaxies within different environments should help to discriminate
between different formation and evolution models. For instance, hierarchical scenarios pre-
dict that ellipticals in rich clusters assembled completely at high redshift (z>3), whereas
field ellipticals may have experienced an elapsed and more complex star formation history
– 3 –
(Kauffmann & Charlot 1998). However, very little is known about the dependence of the
relative abundances on environment. One piece of information is that there is no difference
in the [Mg/Fe] ratio between cluster and field elliptical galaxies (Jørgensen 1999; Kuntschner
et al. 2002). In this letter, we study the behaviour of several Lick/IDS indices (see Worthey
et al. 1994 for definition) in a sample of low and high density environment galaxies (LDEG
and HDEG respectively) and, surprisingly, we do find systematic differences in the strength
of C and CN features.
2.Observations and data analysis
Long–slit spectra of 98 early–type galaxies in different environments were taken in four
observing runs with two different telescopes. The sample comprises 59 galaxies from the
field and the Virgo cluster (LDEG), and 34 galaxies from the central region of the Coma
cluster (HDEG), spanning a wide range of absolute magnitudes (−22.5 < MB < −16.5,
using H0= 75 km s−1Mpc−1), and central velocity dispersions 40 < σ < 400 km s−1(from
dwarf ellipticals, in Virgo and Coma, to giant galaxies).
In the first two runs (1998 January and 1999 August) we used the 3.5m telescope at Calar
Alto Observatory (Almer´ ıa, Spain), employing the Twin Spectrograph. The observations of
the third and fourth runs (1999 March and 2001 April) were carried out with the 4.2m WHT
at the Roque de los Muchachos Observatory (La Palma, Spain) using the ISIS spectrograph.
Spectral resolutions range from 2.6˚ A and 4.0˚ A (FWHM) for LDEG to 8.6˚ A for HDEG, in
a spectral range λλ3600–5400˚ A. Exposures times of 1200–3600 secs per galaxy allowed us
to obtain central spectra with signal-to-noise (S/N) ratios (per˚ A) ranging from 25 to 250.
We also observed several galaxies in common between runs to ensure that the measurements
were in the same system. 85 stars from the IDS/Lick library were included to transform the
measured line-strength indices to the Lick system.
Standard data reduction procedures (flat-fielding, cosmic ray removal, wavelength cal-
ibration, sky subtraction and fluxing) were performed with REDuc
allowed a parallel treatment of data and error frames and provided an associated error file
for each individual data spectrum.
mE (Cardiel 1999), which
– 4 –
2.2.Velocity dispersion and line-strength indices
For each galaxy, central spectra were extracted by coadding within a standard met-
ric aperture size corresponding to 4 arcsec projected at the distance of the Coma cluster
(simulating in this way a fixed linear aperture of length 1.8 kpc in all the galaxies). Veloc-
ity dispersions were determined using the MOVEL and OPTEMA algorithms described in
Gonz´ alez (1993). We measured the Lick/IDS line-strengths indices, although only CN2, Mgb,
C4668 and <Fe> are presented here (the rest of the indices, with a more detailed explanation
of data handling, will be presented in a forthcoming paper). All the indices were transformed
to the Lick spectrophotometric system using the observed stars and following the prescrip-
tions in J. Gorgas, P. Jablonka, & P. Goudfrooij (2003, in preparation); see also Worthey &
Ottaviani (1997). Using galaxies in common with Trager (1997) and with repeated observa-
tions between runs, we double-checked that there are no systematic errors in the indices of
galaxies observed in different runs. All the indices presented in this work are transformed
into magnitudes, following Kuntschner (1989): I′(mag) = −2.5log(1−I(˚ A)/∆λ), where ∆λ
is the width of the index bandpass.
In Fig. 1 we present the CN2, C4668′, Mgb′, and ?Fe?′indices versus velocity dispersion
(σ) for the 98 galaxies of the sample. The spectra of three dwarf galaxies from the Coma
sample did not have enough S/N to derive a reliable measurement of σ and a typical value
of 40 km s−1was assumed. Also, for the rest of the galaxies with σ < 60 km s−1, we adopted
σ values from Pedraz et al. (2002) and Guzm´ an et al. (2003). It is apparent from this figure
that galaxies located in low and high density environments show systematic differences in
C4668′and CN2, being the indices in HDEG systematically lower than in LDEGs. On the
other hand, both galaxy subsamples seem to follow similar relationships in the ?Fe?′and
Mgb′versus σ diagrams.
To quantify the possible systematic differences, we have performed a linear least-square
fit to the LDEG subsample, and have measured the mean offsets (weighting with errors) of
the Coma galaxies (HDEG) from the fits. These differences and their errors are indicated
within each panel, confirming the high significance of the systematic offsets in CN2 and
These systematic differences are also visible directly in the spectra of the galaxies. For
illustration, Fig. 2 shows the coadded spectra of LDEG (thin line) and HDEG (thick line)
in the range 150 < σ < 250 km s−1, previously shifted to the same radial velocity and
– 5 –
broadened to the maximum σ. The S/N ratios of the two combined spectra are above 300.
It is evident that the offsets found in Fig. 1 are due to real enhancements of the CN band
at λ4177˚ A and the C2Swan band at 4735˚ A in LDEG compared to HDEG. Note that this
effect is also quite remarkable in the CN band around λ3865˚ A (not included in the Lick
Prior to interpret the systematic differences in CN2and C4668′as variations on elements
abundances, and since the indices are also sensitive to other physical parameters, we have
explored further possibilities:
(1) Given the gravity dependences of the CN2and C4668′indices (both are stronger in
giant than in dwarf stars; Worthey et al. 1994), a decrease in the giant/dwarf ratio in HDEG
(with respect to LDEG) would lead to lower index values. Using the models by Vazdekis et
al. (1996), we have checked that a change in the exponent of the IMF from 0.80 to 2.80 would
decrease CN2and C4668′by 0.033 and 0.009 mag respectively, while the expected changes in
Mgb′and <Fe> would be of 0.005 and 0.003 in the opposite sense (assuming an age of 10 Gyr
and solar metallicity). These predictions are marginally consistent with our results (note
that the above offsets for Mgb′and <Fe> are compatible with the measurements within the
uncertainties), thus we cannot reject this possibility. However, this result is in contradiction
with other studies (Rose et al. 1994) which suggest a decrease in the giant/dwarf ratio
in LDEG compared to HDEG, and would imply important changes in other observables.
Measurements of the Ca triplet in the near-infrared (with a high sensitivity to the IMF; see
Cenarro et al. 2003) should help to discard or confirm this scenario.
(2) A difference in (luminosity-weighted) mean age between HDEG and LDEG could
also introduce systematic offsets in CN2and C4668′between both galaxy samples. Under
this scenario, and using the models by V96, the observed offsets could be accounted for
if HDEG were about 8 Gyr younger than LDEG. This possibility can be rejected since it
would imply a decrease in Mgb′and <Fe> of 0.026 and 0.014 in HDEG compared with
LDEG, which is not observed. Furthermore, it contradicts previous findings which suggest
that HDEG are, in any case, older than LDEG (Kuntschner et al. 2002).
Thus, the most plausible explanation of our results is that variations in the relative
abundances of C and N with respect to Mg and Fe are responsible for the observed offsets
between galaxies in different environments. Fig. 3 compares the observed CN2and C4668′
with the predictions of stellar population models. This figure clearly shows that the previ-
– 6 –
ously found overabundances of C and N only stand for LDEG, while HDEG tend to exhibit
relative abundances closer to the solar partition.
The C4668′index is extremely sensitive to carbon changes, so little variations in carbon
abundance can change this index dramatically (Tripicco & Bell 1995). However, the vari-
ations of CN are mostly controlled by N, because extra C is readily incorporated into CO
but extra N makes more CN molecules. Therefore, a change in both C and N abundances
is required to explain our results. Besides, if extra carbon is easily incorporated into CO,
one should detect an enhancement in the strength of the CO bands in LDEG compared
to HDEG. This effect has indeed been found by Mobasher & James (1996) comparing the
CO band at 2.3µm in galaxies from the field and from the Pisces and Abell 2634 clusters.
They interpreted this difference as an evidence of intermediate-age stellar population in
LDEG (through a major contribution of AGB stars). Although we do not discard a larger
contribution of younger populations in LDEG compared to HDEG, our results in the blue
spectral range imply that their observations can be solely explained by a relative abundance
difference. In any case, the observed offsets can not be due to an age effect alone (see above).
The most plausible scenario to explain the differences in relative abundances is the one
in which LDEG and HDEG have experienced different star formation histories. In particular,
since the ISM is progressively enriched in C and N by stars of low and intermediate mass
stars, HDEG should have been fully assembled before the massive release of these elements.
The hierarchical clustering paradigm currently predicts that galaxies in clusters formed at
different epochs than LDEGs. If the time elapsed between the assembling of the former
and the later is enough to permit the C and N enrichment of the ISM of the pre-merging
building blocks in LDEG, the stars formed in these galaxies during the merging events will
be C and N enhanced. The constancy of the iron-peak elements could be understood if,
in both environments, the mergers take place before type-I SN can significantly pollute the
ISM of the pre-merging blocks. Additionally, HDEG could have experienced a truncated star
formation and chemical enrichment history compared to a more continuous time-extended
history for their counterparts in low density environments. However, under this hypothesis,
there should be an increase of magnesium (produced by type-II SN) in LDEG which is
not detected. One way to understand the constancy of the Mgb′index could be to invoke
the IMF-metallicity relationship suggested by Cenarro et al. (2003), in which the succesive
episodes of star formation in LDEG would take place with lower giant-to-dwarf ratios.
To conclude, we have noted for the first time a systematic difference in the element
abundance ratios of galaxies situated in different environments. These differences impose
strong constraints to models of chemical evolution and galaxy formation. It is clear that more
work is still needed to completely understand the causes of the differences. In particular, it
– 7 –
would be very interesting to compare the CN2and C4668’ indices between the central regions
and the outskirts of the Coma cluster, where Mobasher & James (2000) found differences in
the strength of the near-IR CO molecule. Also, the detailed study of other dense clusters is
highly needed to confirm whether this effect is particular of the Coma cluster.
We are indebted to R. Guzm´ an for providing us with a catalogue of dwarf elliptical
galaxies in the Coma cluster. The WHT is operated on the island of La Palma by the Royal
Greenwich Observatory at the Observatorio del Roque de los Muchachos of the Instituto de
Astrof´ ısica de Canarias. The Calar Alto Observatory is operated jointly by the Max-Planck-
Institute f¨ ur Astronomie, Heidelberg, and the Spanish Comisi´ on Nacional de Astronom´ ıa.
This work was supported by the Spanish research project No.AYA2000-977.
– 8 –
Burstein, D., Faber, S., Gaskell, C. M., & Krumm, N. 1984, ApJ, 287, 586
Cannon, R. D., Croke, B. F. W., Bell, R. A., Hesser, J. E, & Stathakis, R. A. 1998, MNRAS,
Cardiel, N. 1999, PhD thesis, Universidad Complutense, Madrid
Cenarro, A. J., Gorgas, J., Vazdekis, A., Cardiel, N., & Peletier, R. F. 2003, MNRAS, 339,
Chiappini, C., Romano, D., & Matteucci, F. 2003, MNRAS, 339, 63
Faber, S. M., Worthey, G., & Gonz´ alez, J. J. 1992, in IAU Symp. 149, Stellar Populations
of Galaxies, ed. B . Barbuy ,& A. Renzini (Dordrecht: Kluwer), 255
Gonz´ alez, J. J. 1993, PhD thesis, University of California, Santa Cruz.
Guzm´ an, R., Graham, A., Matkovic, A., Vass, I., Gorgas, J., & Cardiel, N. 2003, preprint
Harbeck, D., Smith, G., & Grebel, E. 2003, AJ, 125, 197
Jørgensen I. 1999, MNRAS, 306, 607
Kauffmann, G., & Charlot, S. 1998, MNRAS, 294, 705
Kuntschner, H. 1989, PhD thesis, University of Durham, UK.
Kuntschner, H., Smith, R., Colles, M., Davies, R., Kaldare, R., & Vazdekis, A. 2002, MN-
Matteucci, F. 1994, A&A, 288, 57
Mobasher, B., & James, P. A. 1996, MNRAS, 280, 895
Mobasher, B., & James, P. A. 2000, MNRAS, 316, 507
Moll´ a, M., & Garc´ ıa–Vargas, M. L. 2000, A&A, 359, 18
O’Connell, R. W. 1976, ApJ, 206, 370
Pedraz, S., Gorgas, J., Cardiel, N., S´ anchez–Bl´ azquez, P., & Guzm´ an, R. 2002, MNRAS,
– 9 –
Peletier, R. F. 1989 PhD thesis, Univ. of Groningen
Peterson, R. C. 1976, ApJ, 210, L123
Renzini, A., & Voli, M. 1981, A&A, 94, 175
Rose, J. A., Bower, R., Caldwell, N., Ellis, R., Sharples, R. M., & Teague, P. 1994, AJ, 108,
Saglia, R. P., Maraston, C., Thomas, D., Bender, R., & Colles, M. 2002, ApJ, 579, L13
Thomas, D., Maraston, C., & Bender, R. 2003, MNRAS, in press (astro-ph/0303615)
Trager, S. C. 1997, PhD tehsis, University of California, Santa Cruz.
Tripicco, M. J., & Bell, R. A. 1995, AJ, 110, 3035
Vazdekis, A., Casuso, E., Peletier, R. F., & Beckman, E. 1996, ApJS, 106, 307
Vazdekis, A., Peletier, R. F., Beckman, J.E & Casuso, E. 1997, ApJS, 111, 203
Vazdekis, A., Kuntschner, H., Davies, R., Arimoto, N., Nakamura, O., & Peletier, R. 2001,
ApJ, 551, 127
Worthey, G. 1998, PASP, 110, 888
Worthey, G., & Ottaviani, D. L. 1997, ApJS, 111, 377.
Worthey, G., Faber, S. M., & Gonz´ alez, J. J. 1992, ApJ, 398, 69
Worthey, G., Faber, S. M., Gonz´ alez, J. J., & Burstein, D. 1994, ApJS, 94, 687
This preprint was prepared with the AAS LATEX macros v5.0.
– 10 –
Fig. 1.— Relations of the line-strength indices analyzed in this paper with central velocity
dispersion for HDEG and LDEG. LDEG are represented with open symbols whereas filled
symbols correspond to galaxies from the Coma cluster. Dwarfs ellipticals are plotted with
triangles, S0 with squares and E galaxies with circles. Lines represent error-weighted least-
squares linear fits to the LDEG subsample. Typical errors in the indices and σ’s are included
at the bottom-left of each panel. Labels indicate the mean offsets, and their corresponding
errors, of the HDEG with respect to the fits.
– 11 –
Fig. 2.— Coadded spectra for LDEG and HDEG, represented with thin and thick lines
respectively, in our spectral range. See text for details. The position of the CN band around
λ4177˚ A, and the central bandpasses of the CN2and C4668′indices are indicated. Besides
these bands, a very consistent difference in the CN violet system (3850-3890˚ A) is also
– 12 –
Fig. 3.— C4668′line-strengths versus CN2for the complete samples of HDEG and LDEG.
Overplotted are models by Vazdekis (1996). Lines of constant [Fe/H]= -1.68,-0.68,-0.38, 0.0,
+0.2 are shown by solid lines. Dashed lines represent models of constant ages, from 1 Gyr
to 17.78 Gyr, increasing from left to right. Symbols are the same as in Fig. 1.