The Dependence of Type Ia Supernova Luminosities on their Host Galaxies

Abstract

(Abridged) Precision cosmology with Type Ia supernovae (SNe Ia) makes use of the fact that SN Ia luminosities depend on their light-curve shapes and colours. Using Supernova Legacy Survey (SNLS) and other data, we show that there is an additional dependence on the global characteristics of their host galaxies: events of the same light-curve shape and colour are, on average, 0.08mag (~4.0sigma) brighter in massive host galaxies (presumably metal-rich) and galaxies with low specific star-formation rates (sSFR). SNe Ia in galaxies with a low sSFR also have a smaller slope ("beta") between their luminosities and colours with ~2.7sigma significance, and a smaller scatter on SN Ia Hubble diagrams (at 95% confidence), though the significance of these effects is dependent on the reddest SNe. SN Ia colours are similar between low-mass and high-mass hosts, leading us to interpret their luminosity differences as an intrinsic property of the SNe and not of some external factor such as dust. If the host stellar mass is interpreted as a metallicity indicator, the luminosity trends are in qualitative agreement with theoretical predictions. We show that the average stellar mass, and therefore the average metallicity, of our SN Ia host galaxies decreases with redshift. The SN Ia luminosity differences consequently introduce a systematic error in cosmological analyses, comparable to the current statistical uncertainties on parameters such as w. We show that the use of two SN Ia absolute magnitudes, one for events in high-mass (metal-rich) galaxies, and one for events in low-mass (metal-poor) galaxies, adequately corrects for the differences. Cosmological fits incorporating these terms give a significant reduction in chi^2 (3.8-4.5sigma). We conclude that future SN Ia cosmological analyses should use a correction of this (or similar) form to control demographic shifts in the galaxy population. Comment: Accepted for publication in MNRAS.

Full text

Page 1

arXiv:1003.5119v1 [astro-ph.CO] 26 Mar 2010
Mon. Not. R. Astron. Soc. 000, 1–23 (0000)Printed 29 March 2010(MN LATEX style file v2.2)
The Dependence of Type Ia Supernova Luminosities on
their Host Galaxies
M. Sullivan1⋆, A. Conley2, D. A. Howell3,4, J. D. Neill5, P. Astier6, C. Balland7,8,
S. Basa8, R. G. Carlberg9, D. Fouchez10, J. Guy6, D. Hardin6, I. M. Hook1,11,
R. Pain6, N. Palanque-Delabrouille12, K. M. Perrett9,13, C. J. Pritchet14,
N. Regnault6, J. Rich12, V. Ruhlmann-Kleider12, S. Baumont15,6, E. Hsiao16,
T. Kronborg6, C. Lidman17, S. Perlmutter18,16, E. S. Walker19,1
1Department of Physics (Astrophysics), University of Oxford, DWB, Keble Road, Oxford OX1 3RH, UK
2Department of Astrophysical and Planetary Sciences, University of Colorado, Boulder, CO 80309-0391, USA
3Las Cumbres Observatory Global Telescope Network, 6740 Cortona Dr., Suite 102, Goleta, CA 93117, USA
4Department of Physics, University of California, Santa Barbara, Broida Hall, Mail Code 9530, Santa Barbara, CA 93106-9530, USA
5California Institute of Technology, 1200 E. California Blvd., Pasadena, CA 91125, USA
6LPNHE, Universit´ e Pierre et Marie Curie Paris 6, Universit´ e Paris Diderot Paris 7, CNRS-IN2P3, 4 place Jussieu, 75252 Paris Cedex 05, France.
7University Paris 11, Orsay, F-91405, France
8LAM, Pole de l’Etoile Site de Chateau-Gombert, 38 rue Frederic Joliot-Curie, 13388 Marseille Cedex 13, France
9Department of Astronomy and Astrophysics, University of Toronto, 50 St. George Street, Toronto ON M5S 3H4, Canada
10CPPM, Aix-Marseille Universit´ e, CNRS/IN2P3, 13288 Marseille Cedex 9, France
11INAF - Osservatorio Astronomico di Roma, via Frascati 33, 00040 Monteporzio (RM), Italy
12CEA/Saclay, DSM/Irfu/Spp, 91191 Gif-sur-Yvette Cedex, France
13Network Information Operations, DRDC Ottawa, 3701 Carling Avenue, Ottawa, ON, K1A 0Z4, Canada
14Department of Physics and Astronomy, University of Victoria, P.O. Box 3055 STN CSC, Victoria BC V8T 1M8, Canada
15LPSC, CNRS-IN2P3, 53 rue des Martyrs, 38026 Grenoble Cedex, France
16Lawrence Berkeley National Laboratory, Mail Stop 50-232, Lawrence Berkeley National Laboratory, 1 Cyclotron Road, Berkeley, CA 94720, USA
17Anglo-Australian Observatory, P.O. Box 296, Epping, NSW 1710, Australia
18Department of Physics, University of California, 366 LeConte Hall MC 7300, Berkeley, CA 94720-7300, USA
19Scuola Normale Superiore, Piazza dei Cavalieri 7, 56126 Pisa, Italy
29 March 2010
c ? 0000 RAS

Page 2

2
M. Sullivan et al.
ABSTRACT
Precision cosmology with Type Ia supernovae (SNe Ia) makes use of the fact that
SN Ia luminosities depend on their light-curve shapes and colours. Using Supernova
Legacy Survey (SNLS) and other data, we show that there is an additional dependence
on the global characteristics of their host galaxies: events of the same light-curve
shape and colour are, on average, 0.08mag (≃ 4.0σ) brighter in massive host galaxies
(presumably metal-rich) and galaxies with low specific star-formation rates (sSFR).
These trends do not depend on any assumed cosmological model, and are independent
of the SN light-curve width: both fast and slow-declining events show the same trends.
SNe Ia in galaxies with a low sSFR also have a smaller slope (“β”) between their
luminosities and colours with ∼2.7σ significance, and a smaller scatter on SN Ia Hubble
diagrams (at 95% confidence), though the significance of these effects is dependent on
the reddest SNe. SN Ia colours are similar between low-mass and high-mass hosts,
leading us to interpret their luminosity differences as an intrinsic property of the SNe
and not of some external factor such as dust. If the host stellar mass is interpreted as
a metallicity indicator using galaxy mass–metallicity relations, the luminosity trends
are in qualitative agreement with theoretical predictions. We show that the average
stellar mass, and therefore the average metallicity, of our SN Ia host galaxies decreases
with redshift. The SN Ia luminosity differences consequently introduce a systematic
error in cosmological analyses, comparable to the current statistical uncertainties on
parameters such as w, the equation of state of dark energy. We show that the use
of two SN Ia absolute magnitudes, one for events in high-mass (metal-rich) galaxies,
and one for events in low-mass (metal-poor) galaxies, adequately corrects for the
differences. Cosmological fits incorporating these terms give a significant reduction in
χ2(3.8–4.5σ); linear corrections based on host parameters do not perform as well. We
conclude that all future SN Ia cosmological analyses should use a correction of this
(or similar) form to control demographic shifts in the underlying galaxy population.
Key words: supernovae: general – cosmology: observations – distance scale.
1INTRODUCTION
As calibrateable standard candles, Type Ia supernovae (SNe
Ia) provide a direct route to understanding the nature of
the dark energy that drives the accelerated expansion of the
Universe. Yet, the relationships that allow the calibration
of their peak luminosities, and hence permit their cosmo-
logical use, remain purely empirical. Relations between the
width of the SN Ia light curve and peak luminosity (Phillips
1993) and between the SN Ia optical colours and luminosity
(e.g. Riess et al. 1996; Tripp 1998) reduce the scatter in their
peak magnitudes to ∼0.15mag (Jha et al. 2007; Guy et al.
2007; Conley et al. 2008). As the available SN Ia samples
increase in both size and quality, and the dark energy con-
straints they provide become correspondingly more statisti-
cally precise, it is increasingly important that the validity of
these calibrating relationships is robustly examined.
The observed properties of SNe Ia are known to cor-
relate with the physical parameters defining their host
galaxy stellar populations. SNe Ia are more than an or-
der of magnitude more common (per unit stellar mass) in
actively star-forming or morphologically late-type galaxies
than in passive or elliptical systems (Mannucci et al. 2005;
Sullivan et al. 2006). SNe Ia in elliptical or passively evolv-
ing systems are also intrinsically fainter, with narrower,
faster (or lower “stretch”), light curves (Hamuy et al. 1995,
⋆E-mail: sullivan@astro.ox.ac.uk
1996b; Riess et al. 1999; Hamuy et al. 2000; Sullivan et al.
2006). Though this effect is corrected for by the light-curve
shape correction, the amount of star formation activity in
the universe increases with redshift, and these differences
lead to an observed “demographic shift” in mean SN Ia
properties. A greater fraction of intrinsically luminous, wider
light-curve events in the distant universe are seen compared
to that observed locally (Howell et al. 2007). These photo-
metric differences are also partially reflected in SN Ia spec-
tra, with SNe Ia in spiral galaxies showing weaker inter-
mediate mass element line strengths than those in ellipti-
cal galaxies (Bronder et al. 2008; Balland et al. 2009), and
a corresponding evolution in the mean SN Ia spectrum with
redshift (Sullivan et al. 2009).
There are suggestions that these effects may be the
result of multiple astrophysical channels capable of pro-
ducing SN Ia explosions (e.g. Scannapieco & Bildsten 2005;
Mannucci et al. 2006). In particular, delay-time distribu-
tions with distinct “prompt” and “delayed” components,
or with a wide range of delay-times, match most obser-
vational datasets well (Mannucci et al. 2006; Sullivan et al.
2006; Pritchet et al. 2008; Totani et al. 2008), though the
minimum age for the prompt systems remains controver-
sial (Aubourg et al. 2008; Raskin et al. 2009) with some
evidence that “prompt” SNe Ia occur more frequently in
metal-poor systems (Cooper et al. 2009). The use of SNe
Ia as precision cosmological probes therefore depends on
establishing that the demographic shifts, or existence of
c ? 0000 RAS, MNRAS 000, 1–23

Page 3

SN Ia host galaxies
3
multiple channels to a SN Ia, do not impact on the light-
curve-width/colour/luminosity relationships. If these rela-
tions show environmental dependence, then the task of cal-
ibrating SNe Ia for cosmology becomes substantially more
challenging (e.g. Sarkar et al. 2008; Kelly et al. 2009).
A second complication arises from the (poorly under-
stood) colour corrections applied to SN Ia luminosities. Red-
der SNe Ia appear fainter than their bluer counterparts,
but the slope of the relationship between SN Ia colour and
magnitude is inconsistent with the ratio of total-to-selective
absorption appropriate for the diffuse interstellar medium
of the Milky Way (RV = 3.1). Multiple studies of differ-
ent SN Ia samples indicate that the effective RV inferred
from normal SNe Ia is smaller than 3.1 (e.g., Tripp 1998;
Astier et al. 2006; Krisciunas et al. 2006), and the assump-
tion of RV = 3.1, even after light curve shape corrections,
leads to serious systematic error problems such as a spurious
“Hubble Bubble” (Jha et al. 2007; Conley et al. 2007). The
reason for this low effective RV is not well understood. Al-
though uncorrected intrinsic variations in the SN Ia popula-
tion could play a role (e.g. Kasen et al. 2009), some dust ex-
tinction must also affect the SN Ia luminosities and colours,
and this may vary by environment. Furthermore, the exact
value of RV obtained is sensitive to method used to deter-
mine it, with lower RV obtained when fitting linear relations
between SN Ia luminosities, colours, and light curve widths
(Folatelli et al. 2010), perhaps due to intrinsic variation in
SN Ia colours that correlates with luminosity but not light-
curve width. Current knowledge of SNe Ia is not sufficient
to separate and correct for both intrinsic colour–luminosity
and dust-induced colour–luminosity effects in cosmological
SN Ia samples.
Examining how SN Ia luminosities vary with envi-
ronment after light curve shape and colour corrections
can place constraints on the degree of these possible
variations. Early studies showed little evidence that cor-
rected SN Ia luminosities varied with host galaxy mor-
phologies (e.g., Perlmutter et al. 1999; Riess et al. 1999;
Sullivan et al. 2003; Williams et al. 2003; Gallagher et al.
2005), though these tests used relatively small samples of
events (? 50), in some cases from the first-generation of
SN Ia cosmological samples before dense multi-colour light
curves were routinely obtained.
More recent analyses,
samples, have shown tentative evidence for variation.
Hicken et al. (2009b) found ≃ 2σ evidence that SNe Ia in
morphologically E/S0 galaxies are brighter than those in
later-type spirals after light-curve shape and colour correc-
tions. Extending beyond simple host galaxy morphologies to
more physically motivated variables gives further tantalising
suggestions of variation. Gallagher et al. (2008) found evi-
dence for a correlation between Hubble diagram residual and
host galaxy stellar metallicity in a sample of 17 local SNe
Ia located in E/S0 galaxies, in the sense that fainter SNe
Ia after correction were found in metal poor systems (note
this is the reverse of the originally published trend due to
an error in the original analysis; P. Garnavich, private com-
munication). Howell et al. (2009) used 55 SNe Ia from the
first year of the SNLS and showed no significant correla-
tion between Hubble residual and host galaxy metallicity,
albeit using host gas-phase metallicities inferred from av-
erage galaxy stellar-mass–metallicity relations, a less direct
using larger, well-observed
measure of metallicity. Kelly et al. (2009) have shown a re-
lation between host galaxy stellar mass and Hubble residual,
in the sense that more massive systems host brighter SNe
Ia after light curve shape and colour corrections. Under the
assumption that more massive galaxies are metal rich, this
trend is consistent with the revised Gallagher et al. (2008)
result.
In this paper, we use a sample of 282 high redshift SNe
Ia discovered and photometrically monitored by the Canada-
France-Hawaii Telescope (CFHT) as part of the Supernova
Legacy Survey (SNLS), and which form the SNLS “three-
year” sample (SNLS3). Using deep optical imaging of their
host galaxies taken over the duration of the survey, we place
constraints on their recent star-formation activity, stellar
masses (and hence inferred metallicity), and compare to the
photometric properties of the SNe Ia that they host. In par-
ticular, we search for evidence that the corrected SN Ia lumi-
nosities correlate with these host properties, indicating pos-
sible systematic errors in the light curve fitting framework
that underpins their cosmological use. We compare with the
properties of a sample of lower-redshift SNe Ia taken from
the literature.
A plan of the paper follows. In § 2 we introduce the SN
Ia sample and the data available on their host galaxies. § 3
investigates how the SN Ia light curve widths and colours of
these SNe Ia varies according to their host galaxy proper-
ties, and in § 4 we compare their corrected luminosities to the
host properties. We discuss the results, including the cosmo-
logical implications, in § 5, and conclude in § 6. Throughout,
where relevant we assume a flat ΛCDM cosmological model
with ΩM = 0.256 (the reason for this non-standard choice
is explained in § 4) and H0=70kms−1Mpc−1assumed in
all quoted absolute magnitudes. All magnitudes are given
on the Landolt (1992) photometric system as described in
Regnault et al. (2009).
2 TYPE IA SUPERNOVA AND HOST
GALAXY DATA
We begin by introducing the SN Ia samples that we will use
in this paper, and the associated data available for their host
galaxies.
2.1The SN Ia samples
The high-redshift SN Ia data is taken from the Super-
nova Legacy Survey (SNLS). This used optical imaging data
taken as part of the deep component of the five-year Canada-
France-Hawaii Telescope Legacy Survey (CFHT-LS) us-
ing the square-degree “MegaCam” camera (Boulade et al.
2003), located in the prime focus environment “MegaPrime”
on the CFHT. The “deep” component of CFHT-LS con-
ducted repeat imaging of 4 low Galactic-extinction fields,
time-sequenced with observations conducted every 3–4
nights in dark and grey time. Four filters, gMrMiMzM,
were used allowing the construction of high-quality multi-
colour SN light curves; uM data were also taken but are
not time-sequenced. On each night of observations, the data
were searched using two independent pipelines, and an amal-
c ? 0000 RAS, MNRAS 000, 1–23

Page 4

4
M. Sullivan et al.
gamated candidate list produced1(see Perrett et al. 2010).
Spectroscopic follow-up confirmed SN types and measured
redshifts to allow SNe Ia to be placed on a Hubble diagram.
In this paper we use SNe Ia belonging to the three-
year sample, SNLS3; this includes all SNe Ia discovered up
until the end of July 2006. The light curves and other de-
tails for these SNe can be found in Guy et al. (2010), and
spectroscopic information is taken from Howell et al. (2005),
Bronder et al. (2008), Ellis et al. (2008), Balland et al.
(2009), and Walker et al. (2010). The full SNLS3 sample
consists of 282 SNe Ia, however we exclude some of these
events from some of our analyses:
(i) Only SNLS SNe Ia with an identified (§ 2.3) host
galaxy (272 events) are considered (for a discussion of the
identification procedures, see Sullivan et al. 2006, hereafter
S06),
(ii) Only SNe Ia passing light curve quality cuts (e.g.
Conley et al. 2008; Guy et al. 2010) are used – there must
be sufficient photometric coverage to derive reliable peak
luminosities, light curve widths, and colours (see details in
Guy et al. 2010),
(iii) We only consider normal SNe Ia with light-curve pa-
rameters in the range considered for cosmological fits – our
motivation in this paper is to assess the cosmological im-
pact of any host galaxy dependent trends. Specifically, we
require that the stretch of the SN be greater than 0.80, and
that the colour be less than 0.30 (see § 2.2 for a discussion
of the meaning of these parameters). We also discard > 3σ
outliers from the best fitting cosmological model. 231 events
pass the light-curve coverage and SN parameter cuts.
(iv) In analyses where we search for trends in the SNLS
data, we use only SNe Ia at redshift z ? 0.85 (195 events).
At these lower redshifts, both the SN and host galaxy pho-
tometry are higher signal-to-noise and their photometric pa-
rameters better measured. The SNLS Malmquist biases are
also smaller (Perrett et al. 2010).
Where relevant, we also use samples of SNe Ia from
the literature. We construct a sample of low-redshift SNe Ia
from the compilation of Conley et al. (2010), which includes
SNe Ia from a variety of sources (primarily Hamuy et al.
1996a; Riess et al. 1999; Jha et al. 2006; Hicken et al. 2009a;
Contreras et al. 2009). We apply bulk-flow peculiar veloc-
ity corrections to the SN magnitudes and redshifts, placing
the redshifts in the CMB-frame (zcmb) following Neill et al.
(2007), but with updated models (Conley et al. 2010). The
accuracy of these corrections is estimated to ±150 km s−1,
which we propagate into the SN magnitude errors in cosmo-
logical fits. We only use SNe Ia in the smooth Hubble flow,
here defined as zcmb? 0.01, and apply the same light curve
quality cuts as for the SNLS sample. There are 110 low-
redshift objects in total. We also use the HST-discovered
sample of Riess et al. (2004) and Riess et al. (2007), here-
after the R07 sample. We select 24 SNe Ia at z > 0.9 from
this sample to increase our redshift lever arm above z = 1.
1Candidates
http://legacy.astro.utoronto.ca/
canbe foundat
2.2SN Light curve fitting
In its current application, SN Ia cosmology depends on two
corrections to “raw” SN Ia peak luminosities that when
applied reduce the dispersion in their peak magnitudes.
The first is the light-curve-shape/luminosity relation (e.g.
Phillips 1993): brighter SNe Ia tend to have wider, longer-
duration light curves (higher stretch) than their fainter
counterparts. The second is a colour correction: brighter SNe
Ia tend to have bluer colours, whilst fainter SNe tend to be
redder (Riess et al. 1996; Tripp 1998). Together the applica-
tion of these corrections can yield distance estimates precise
to ≃ 6 per cent. These corrections are applied in different
ways depending on whether a technique is a distance es-
timator (e.g., MLCS2k2; Jha et al. 2007) or a light curve
fitter (e.g. SALT or SiFTO; Guy et al. 2007; Conley et al.
2008), though the underlying principle is the same in both
approaches.
In this paper, we primarily use the SiFTO light curve
fitter (Conley et al. 2008) and compare our results to SALT2
(Guy et al. 2007) where appropriate. In general SALT2 and
SiFTO give very similar results when trained on the same
SN sample – a full discussion can be found in Guy et al.
(2010). Both fitters have been retrained and improved since
the original published versions using SNLS and low-redshift
data. The product of both fitters for each SN is a rest-frame
B-band apparent magnitude (mB), a stretch (s) measure-
ment, and a colour estimate (C), together with associated
errors and covariances (SALT2 uses the broadly equivalent
x1 parameter in place of s). Throughout, the SN Ia colour,
C, is defined as the rest-frame (B − V ) colour of the SN at
the time of maximum light in the rest-frame B-band. We
refit all SNe Ia using these light curve fitters to ensure that
the different samples can be placed on the same distance
scale. A full discussion of their application to the current
dataset, together with their tabulated output, can be found
in Guy et al. (2010) and Conley et al. (2010).
2.3 SN host galaxy photometry
Our SNLS host galaxy photometry comes in the optical from
the CFHT-LS (uMgMrMiMzM) and in the near infra red
(IR) from the WIRcam Deep Survey (WIRDS; Bielby et al.
in prep.) of a sub-section of the CFHT-LS fields (J, H, Ks).
The identification procedure for the SNLS SN Ia host galax-
ies is the same as that in S06. Photometry is performed by
SExtractor (Bertin & Arnouts 1996) using FLUX AUTO
photometry running in dual-image mode, detecting from the
deep iM stack and measuring from each of the optical and
near-IR filters in turn (60% of our SN Ia hosts have near-IR
data). Each stack has a similar image quality and hence the
same physical aperture is used in each filter. In about ≃3% of
cases no host galaxy can be identified. This could be because
the SN lies far from the centre of its host galaxy leading to
ambiguity in the correct choice of host, or because the host
lies below the CFHT-LS flux limits. We discard these ob-
jects. Weight maps are used to determine the measurement
errors, and in the optical, the photometric zeropoints are
generated using a comparison to the tertiary standard star
lists of Regnault et al. (2009). No SN light is present in the
optical stacks which are constructed on a per season basis
(S06).
c ? 0000 RAS, MNRAS 000, 1–23

Page 5

SN Ia host galaxies
5
For the low-redshift sample, we use host galaxy photom-
etry recently compiled by Neill et al. (2009, hereafter N09),
including ultraviolet data from the GALEX (Galaxy Evo-
lution Explorer) satellite, and optical photometry from the
third reference catalog of bright galaxies (RC3; Corwin et al.
1994) and the Sloan Digital Sky Survey (SDSS). The pho-
tometry for this sample was carefully performed in matched
apertures with foreground contaminating stars masked.
Though these data span a different observed wavelength
range compared to the SNLS host photometry, in the rest-
frame the wavelength range covered is reasonably similar:
the maximal range is 1500–9000˚ A for the low-redshift sam-
ple (though frequently the available data span a smaller
range than this), and 2400–13000˚ A for the SNLS sample (at
z = 0.6). Note that not all the low-redshift SN hosts have
publicly available host photometry: only 69 (of 110; 63%)
low-redshift SNe Ia have sufficient data and are carried for-
ward in the analysis. The missing low-redshift SNe Ia are
due to incomplete GALEX and SDSS coverage, rather than
the host galaxies being too faint to be detected by the two
surveys.
For the R07 sample, we use photometry taken from
the Great Observatories Origins Deep Survey (GOODS)
HST data (Giavalisco et al. 2004), taken with the Advanced
Camera for Surveys (ACS). Data are available in four fil-
ters: F435W (broadly equivalent to a B filter), F606W (V ),
F814W (i′) and F850LP (z′). All of the R07 SNe have ACS
coverage, and again SExtractor FLUX AUTO photometry
is used. We also use J, H and K imaging taken as part of
the GOODS NICMOS survey (e.g. Buitrago et al. 2008), as
well as ground-based data (Retzlaff et al. 2009).
2.4 Host galaxy parameter estimation
The
the
S06
PEG
upon
Fioc & Rocca-Volmerange 1997). We use an expanded
set of 15 exponentially declining star formation histories
(SFHs) with SFR ∝ exp−t/τ, where t is time and τ is
the e-folding time, each with 125 age steps. The internal
P´EGASE.2 dust prescriptions are not used, and instead a
foreground dust screen varying from E(B − V ) = 0 to 0.30
in steps of 0.05 is added. With the 7 different foreground
dust screens, this gives a total of 105 unique evolving
spectral energy distributions (SEDs). The metallicity of
the stellar population evolves consistently following the
standard P´EGASE.2 model with an initial value of 0.0004,
and the standard P´EGASE.2 nebular emission prescription
is added.
Z-PEG is used to locate the best-fitting SED model
(in a χ2sense), with the redshift fixed at the CMB-frame
redshift of the SN. Only solutions younger than the age of
the Universe at each SN redshift are permitted. The current
stellar mass in stars (Mstellar, measured in M⊙), the recent
star formation rate (SFR, in M⊙yr−1, averaged over the last
0.25Gyr before the best fitting time step), and the specific
star formation rate (sSFR, the SFR per unit stellar mass
with units of yr−1, e.g. Guzman et al. 1997) are all recorded.
Error bars on these parameters are taken from their range in
method
SN
which
(Le Borgne & Rocca-Volmerange
the P´EGASE.2
to estimate
galaxies
the
physical
is
parameters
to
redshift
2002)
synthesis
of
in
Z-
Ia host
used
similar that
codephotometric
based
(e.g.,spectral code
the set of solutions that have a similar χ2(as in S06). Note
that we do not measure the instantaneous SFR as we only
fit broad-band photometry. We refit all the N09 photometry
to ensure the exact same library SEDs are used for all hosts.
We use a Rana & Basu (1992) initial mass function
(IMF), the P´EGASE.2 default, throughout. Our results are
not sensitive to this choice – we have repeated our analysis
in full with both the more standard Salpeter (1955) IMF,
and with a Baldry & Glazebrook (2003) IMF, and find sim-
ilar results, though the Mstellar and SFRs derived for the
host galaxies have (expected) small mean offsets when using
different IMFs. In detail, the use of a Salpeter IMF gives sys-
tematically larger host masses by 0.04 dex (smaller masses
by 0.16 dex for the B&G IMF), and smaller SFRs by 0.04
dex (smaller by 0.16 dex for B&G), with scatter of around
0.1dex in each comparison. These differences are not a func-
tion of Mstellar, SFR or redshift and so have a negligible
impact on our conclusions.
As only ≃60% of our SNLS SN Ia hosts have observer-
frame near-IR data, we compare the derived properties with
and without these data to check for potential biases in the
remaining 40% of objects. The mean difference in Mstellar
(defined as MOPT
in the recent SFR the mean difference (SFROPT-SFRIR) is
−0.18 dex (r.m.s. 0.44), in the sense that excluding the IR
data leads to smaller SFRs. Thus we find no evidence that
the near-IR data leads to systematically different Mstellar,
and mild evidence that including these data leads to larger
SFRs. The differences in SFR do not follow a Gaussian dis-
tribution; instead the difference is centred around zero but
with a long tail to negative differences; therefore we choose
not to apply the mean offset to the 40% of hosts with no
IR data. There is no evidence for any redshift dependent
trend. Information on the derived properties for the SNLS,
low-redshift and R07 hosts can be found in Table 1.
The Mstellar and SFR distributions for the SNLS and
low-redshift samples are shown in Fig. 1, together with the
distribution of galaxies in the SFR–Mstellar plane. As might
be expected, galaxies with the smallest sSFRs tend to be the
most massive systems, with the lowest mass systems univer-
sally consistent with strong star formation activity. As pre-
viously highlighted by N09, the SNLS and low-redshift hosts
show quite different distributions in Mstellar (and to a lesser
extent SFR): The low-redshift SNe are drawn from more
massive host galaxies. This is almost certainly due to selec-
tion biases. SNLS is a rolling search which will locate SNe
Ia in any type of host galaxy in which they explode, and,
modulo any small spectroscopic follow-up bias, this range
will be reflected in the cosmological sample. At low-redshift,
most SNe Ia are drawn from galaxy-targeted searches which
search known (and typically bright/massive) galaxies, conse-
quently the most massive systems will be over-represented.
Following Howell et al. (2009), we convert the Mstellar
mass estimates into metallicities using average Mstellar–
metallicity (Mstellar–Z) relations. As the universe ages,
galaxies will become more massive via merging processes,
and more metal rich following chemical enrichment and de-
creased metal loss. We use a relation between gas-phase
metallicity, explicitly the nebular oxygen abundance relative
to hydrogen, O/H, and Mstellar derived from SDSS galax-
ies (Tremonti et al. 2004). We use units of 12 + log(O/H),
where the solar abundance is ≃8.69. This Mstellar–Z rela-
stellar-MIR
stellar) is 0.001 dex (r.m.s. 0.15) and
c ? 0000 RAS, MNRAS 000, 1–23