The Faint End of the Luminosity Function and Low Surface Brightness Galaxies

C/2

X:

THE FAINT END OF THE LUMINOSITY FUNCTION AND
LOW SURFACE BRIGHTNESS GALAXIES

Margaret J. Geller

(N

O

C^ I mgeller@cfa.harvard.edu

Smithsonian Astrophysical Observatory,
60 Garden St., Cambridge, MA 02138

Antonaldo Diaferio

Dipartimento di Fisica Generate 'Amedeo Avogadro',
Universitd degli Studi di Torino, via P. Giuria 1, 10125 Torino, Italy

a^:

U;

i-C ■ INFN, Sezione di Torino, via P. Giuria 1,

df ^ > ^

I ■ 10125 Torino, Italy

H

Harvard-Smithsonian Genter for Astrophysics,

60 Garden Street, Cambridge, MA 02138

^ I adiaf erioQcf a . harvard . edu

> .

o

rn ■ Michael J. Kurtz

a^

. ■ Smithsonian Astrophysical Observatory,

^- 60 Garden St., Cambridge, MA 02138

mkurtzOcf a . harvard . edu

Ian P. Dell 'Antonio

Department of Physics, Brown University,
Box 1843, Providence, RI 02912

ianShet . brown . edu

Daniel G. Fabricant

Smithsonian Astrophysical Observatory,
60 Garden St., Cambridge, MA 02138

df abricantOcf a . harvard . edu

-2-

ABSTRACT

SHELS (Smithsonian Hectospec Lensing Survey) is a dense redshift survey
covering a 4 square degree region to a limiting R = 20.6. In the construction of
the galaxy catalog and in the acquisition of spectroscopic targets, we paid careful
attention to the survey completeness for lower surface brightness dwarf galaxies.
Thus, although the survey covers a small area, it is a robust basis for computation
of the slope of the faint end of the galaxy luminosity function to a limiting
Mr, = —13.3+5logh. We calculate the faint end slope in the R-band for the subset
of SHELS galaxies with redshifts in the range 0.02 < z < 0.1, SHELSq.i. This
sample contains 532 galaxies with R< 20.6 and with a median surface brightness
within the half light radius of SBso,/? = 21.82 mag arcsec"^. We used this sample
to make one of the few direct measurements of the dependence of the faint end
of the galaxy luminosity function on surface brightness. For the sample as a
whole the faint end slope, a = —1.31 ±0.04, is consistent with both the Blanton
et al. (2005b) analysis of the SDSS and the Liu et al. (2008) analysis of the
COSMOS field. This consistency is impressive given the very different approaches
of these three surveys. A magnitude limited sample of 135 galaxies with optical
spectroscopic redshifts with mean half-light surface brightness, SBsq,/? > 22.5
mag arcsec"^ is unique to SHELSq.i- The faint end slope is 0:22.5 = —1.52 ±0.16.
SHELSq.i shows that lower surface brightness objects dominate the faint end
slope of the luminosity function in the field, underscoring the importance of
surface brightness limits in evaluating measurements of the faint end slope and
its evolution.

Subject headings: galaxies: distances and redshifts; galaxies: luminosity function,
mass function; galaxies: dwarf; galaxies: fundamental parameters

1. Introduction

The faint end of the galaxy luminosity function is a fundamental constraint on theories of
galaxy formation. All determinations of the low luminosity slope are dramatically shallower
than the predicted mass function of dark matter halos. Baryonic physics appears to be the
key to resolving this discrepancy.

Physical processes possibly relevant to the faint end slope include the gas cooling time
(White & Rees 1978), suppression by photoionization of low-mass galaxy formation (Benson
et al. 2002), "feedback" mechanisms (Benson et al. 2003), merging and tidal stripping.

-3-

Benson et al (2002; 2003) show that various combinations of these processes lead to very
different faint end slopes. These differences can be a function of galaxy environment.

Measuring the slope of the faint end of the galaxy luminosity function, a, remains an
unresolved observational challenge. For the "field" luminosity function deeper redshift sur-
veys covering increasing areas of the sky provide a route to better and better constraints.
However, in any magnitude limited redshift survey the least luminous galaxies occupy a small
volume. Because luminosity and surface brightness are strongly correlated, neither the de-
tection nor the spectroscopy of the lowest luminosity galaxies is trivial. Thus incompleteness
at the faint end is a frustrating and serious issue.

Disney & Phillipps (1983) and McGaugh (1996) emphasized the systematic biases re-
sulting from failure to include low surface brightness galaxies in the determination of the
luminosity function. Sprayberry et al. (1997) made an early measurement of the impact of
low surface brightness galaxies on the determination of the field luminosity function. Their
analysis, based on a catalog of low surface brightness galaxies derived from APM (Automatic
Plate Measuring) scans (Impey et al. 1996), demonstrates that the inclusion of low surface
brightness galaxies substantially steepens the field luminosity function. They obtained a
faint end slope a = —1.46 in the B-band. For active star-froming galaxies in the 2dF red-
shift survey based on based on bj photometry, Madgwick et al. (2002) obtained a faint end
slope of q; = —1.5

Also in the B-band, Driver et al. (2005) analyze the carefully constructed Millenium
Galaxy Catalog and derive a faint end slope for the global luminosity function of a =
— 1.13 ± 0.02. They examine the relationship between luminosity and surface brightness
for their sample and conclude that the surface brightness distribution is broader for less
luminous objects. Driver et al. (2005) show that the faint end slope of the luminosity
function is sensitive to the limiting surface brightness of the survey. Earlier B-band work by
Cross & Driver (2002) had yielded an even shallower faint end slope and indicated robustness
to surface brightness issues.

Blanton et al (2005b) made a major step toward measuring and understanding the be-
havior of the field luminosity function at low luminosity across all of the SDSS photometric
bands. They construct and analyze a low redshift sample of galaxies from the Sloan Digital
Sky Survey (SDSS hereafter) to place constraints on a. Blanton et al (2005b) caution that
their sample is not necessarily complete at least in part because the SDSS was not opti-
mized for this application and they carefully simulate their surface brightness completeness.
They measure a ~ —1.3 in the r-band. Interestingly, this result is similar to the faint end
slope of spectroscopically determined cluster and group luminosity functions extending to
comparably low luminosities (e.g. Mahdavi et al. 2005; Rines & Geller 2010). They also

-4-

argue that missing low surface brightness galaxies could steepen the slope to a ~ —1.5. Like
Sprayberry et al. (1997), Blanton et al. (2005b) conclude that a majority of the faint galaxy
population is blue, low concentration, and low surface brightness.

Here we use a deep complete redshift survey with R< 20.6, the Smithsonian Hectospec
Lensing Survey (SHELS hereafter; Geller et al. 2005; Geller et al. 2010) covering 4 square
degrees of the sky to constrain the value of a. The galaxy catalog derives from the Deep
Lens Survey (DLS hereafter; Wittman et al. 2006) which reaches a la surface brightness
limit of /iij = 28.7 magnitudes per square arcsecond. We constrain a for redshifts z < 0.1
focusing on the contribution of blue, low surface brightness galaxies.

We review the SHELS survey in Section 2. Section 2 contains a discussion of the
survey completeness as a function of surface brightness. We discuss the magnitude - surface
brightness relation for the survey. In Section 3, we display some of the lowest luminosity
objects in our sample and we derive the luminosity function for the sample segregated by
surface brightness. In Section 4 we compare the faint-end slope of our luminosity function
with the COSMOS faint-end slope based on photometric redshifts (Liu et al. 2008). We
conclude in Section 5.

2. The Data

We use two ambitious surveys to explore the faint end of the galaxy luminosity function.
Generally low luminosity galaxies have low surface brightness. We take advantage of the
properties of the two surveys that particularly enable access to these low surface brightness
objects.

The DLS (Wittman et al. 2006) is an NOAO key program covering 20 square degrees
in five separate fields; we use the four square degree F2 field at a = 09'^19'"32.4* and 5 =
+30°00'00". The DLS photometric data were acquired in a 5-hour integration on the Mayall
4-meter in < 0.9" seeing and reaching a 1 a limit for source detection in R of 28.7 magnitudes
arcsec"^ is a good basis for identifying low surface brightness galaxy candidates. We describe
our approach to this issue in Section 12.11

SHELS (Geller et al. 2005; Geller et al. 2010) is a redshift survey covering the F2 field
to a limiting apparent magnitude R = 20.6. SHELS is 98% complete to R = 20.3, 96%
complete to R = 20.6. SHELS contains 541 galaxies with R < 20.6 at z < 0.1 where we
can examine the behavior of the luminosity functions for the lowest luminosity galaxies. We
made a concerted attempt to measure a redshift for each of the lowest surface brightness
candidates in the photometric catalog. We describe the completeness of the entire SHELS

-5-

redshift survey as a function of surface brightness in Section 12.21

SHELS covers a small field with deep photometry and spectroscopy. The SDSS covers a
very wide field to a much brighter limiting apparent magnitude. We contrast the two surveys
in Section [2731

2.1. Photometry

Photometric observations of F2 were made with the MOSAIC I imager (Muller et al.
1998) on the KPNO Mayall 4m telescope between November 1999 and November 2004. The
DLS selected all fields including F2 to exclude apparently bright nearby galaxies and to avoid
known rich clusters with redshift z < 0.1. Even though rich clusters are rare, this selection
biases the density at 2 < 0.1 toward values below the average for the local universe as a
whole. We revisit this issue in Section HJ

The R band exposures are the basis for the galaxy catalog in F2. The effective exposure
is 14,500 seconds. Wittman et al (2006) describe the imaging reduction pipeline. The la
limit for source detection in R is 28.7 magnitudes arcsec"^. To this limit there are 45 sources
per square arcminute. The automatic object identification algorithm produces a complete
catalog of objects with surface brightness /i5o,_R < 27.0 within the half-light radius.

We constructed a galaxy catalog from the R-band source list; we base our luminosity
function computation on this R-band catalog. We selected galaxy candidates with Kron-
Cousins R < 20.6 for spectroscopic observation. The magnitudes are extrapolated total
magnitudes; they are extrapolated from isophotal magnitudes within the limiting 28.7 mag-
nitudes arcsec"^ isophote.

Nearly all of the galaxy candidates with R< 20.6 from the DLS also have SDSS photom-
etry; there are only 104 DLS galaxy candidates without SDSS photometry. Many of these
objects are not resolved by the SDSS and thus they have a broad range in surface brightness.
We do spectroscopy for these objects as for all other galaxy candidates. When we compare
the DLS with the SDSS, we estimate r-band magnitudes from the R-band DLS photometry
and from our spectroscopy.

Figure d] shows the classification diagram we use to select galaxies from the DLS object
list. The upper left hand panel shows all of the 302,574 objects in F2 with R < 22.5. We use
the difference between the magnitude within our 1.5" fiber, R1.5", and the total magnitude R
as a discriminant. The upper right panel shows (dark dots) all of the SHELS galaxies with
redshifts. There are 15,652 redshifts for galaxies with R< 21 and 12,783 for galaxies with

-6-

R< 20.6 (vertical green line).

Objects with the smallest R1.5//— R are mostly stars; objects with larger Ri 5//— R that
lie above the dense galaxy locus generally result from background fluctuations and various
artifacts around bright stars and/or diffraction spikes.

To construct the catalog for SHELS spectroscopy, we examined all of the 33,038 objects
with R< 20.6 visually to remove the obvious artifacts. The initial catalog of 33,038 objects
is complete to /i5o,_R = 27.0 at the expense of including artifacts. We conservatively included
some apparently stellar objects in the observing list, including any of these objects classifled
as a galaxy by the SDSS; our spectroscopy then showed that ~ 5% of the spectroscopically
observed objects with R> 19.5 are, in fact, stars.

Very few of the objects that lie above the dense locus of SHELS points are candidate
galaxies; we included all objects that were not obviously noise or other artifacts in the
spectroscopic observing list. The galaxy candidates are all small compared to the largest
angular size objects contained in the catalog. The galaxy candidates are well away from the
photometric thresholds in Figure [H implying little or no bias against detection of large low
surface brightness objects in the photometric catalog.

2.2. Spectroscopy

We acquired spectra for the objects with the Hectospec (Fabricant et al. 1998, 2005)
on the MMT from April 13, 2004 to April 20, 2007. The Hectospec observation planning
software (Roll et al. 1998) enables efficient acquisition of a magnitude limited sample. We
made a concerted effort to acquire spectra for the lowest surface brightness objects.

The SHELS spectra cover the wavelength range 3,700 — 9,100 A with a resolution of
~6 A. Exposure times ranged from 0.75 to 2 hours. The two hour exposures are adequate to
yield a redshift even for the lowest surface brightness objects. The lowest surface brightness
objects in the survey required the longer integrations. We reduced the data with the standard
Hectospec pipeline (Mink et al. 2007) and derived redshifts with RVSAO (Kurtz & Mink
1998) with templates constructed for this purpose (Fabricant et al. 2005). Our 1468 unique
pairs of repeat observations imply a mean internal error of 56 km s~^ for absorption- line
objects and 21 km s~^ for emission-line objects (see also Fabricant et al. 2005).

For each spectrum we compute the stellar population age indicator, D„4000. This
indicator is the ratio of flux in the 4000-4100A band to flux in the 3850-3950A band (Balogh
et al. 1999); it is a measure of the strength of the 4000A break. The rms scatter in our

-7-

measurement of D„4000 is 0.086. The internal error in D„4000 is only 4.5% based on our
1468 repeat measurements. A comparison of overlapping spectra with the SDSS yields a
median ratio of 1.00 (Fabricant et al. 2008). Following Woods et al. (2010) and Kauffmann
et al. (2003), we use this indicator to segregate galaxies dominated by old and young stellar
populations.

SHELS includes 13,362 galaxies to the limiting apparent magnitude, R = 20.6. The
integral completeness of the redshift survey to this limit is 96%. Geometric constraints are
responsible for the 579 objects without redshifts; they are mostly near the survey corners
and edges. On average, Hectospec positionings revisit every region within the DLS field
(except for the corners and edges) more than a dozen times. Thus we are minimally biased
against close pairs and satellite galaxies.

The upper right-hand panel of Figure [T] shows R1.5//— R as a function of total magnitude
R for all objects with redshifts and R< 21 (black points). The lower left-hand panel shows
the objects with redshifts as yellow points and galaxy candidates without redshifts as black
points. The concentration of objects without spectra toward faint magnitudes is obvious:
there are 4238 objects with R < 21 but only 579 with R< 20.6.

The lower right-hand panel of Figure [1] shows the sample we use to study the faint end
of the galaxy luminosity function, SHELSq.i. SHELSq.i includes all SHELS galaxies with
z < 0.1. There are 541 objects with R < 20.6 and 482 with R < 20.3. These galaxies tend
to have larger R1.5//— R because they are nearby and hence larger on the sky. Most of the
objects in the sample with large R1.5'/— R are in this low redshift subsample.

Table [1] contains the redshifts and R-band magnitudes for the SHELSq.i sample. The
Table includes redshifts we obtained for some galaxies fainter than the survey limit R = 20.6
but with redshift z < 0.1. The Table lists the SHELS identification (column 1), the SDSS
objID (column 2), the right ascension (column 3), the declination (column 4), the R-band
total magnitude (column 5), the source for the magnitude (column 6), the redshift (column
7), the error in the redshift in km s~^ (column 8) , and the observed mean R-band surface
brightness within the half-light radius, /X5o,_r (column 9). Among these redshifts, 18 are from
the SDSS; the rest are new Hectospec measurements.

Figure |2] shows the observed mean surface brightness within the half-light radius, /i^o^r
as a function of the discriminant we use to construct the galaxy catalog, Ris//— R (Figure
H]). Gray points indicate all of the objects in the catalog. We also show (black points) the
galaxies in the SHELS redshift survey (including galaxy candidates with or without a redshift
and with 19 < R < 20.6.). We can readily detect objects with /i5o,i? > 24 mag arcsec"^ and
we are complete to /i5o,ij = 27 mag arcsec"^, but galaxy candidates among these objects

are rare in this local region of the universe. The lowest surface brightness galaxies we find
have /i5o,_R ~ 26 mag arcsec"^. It is interesting to note that all of the known dwarfs in the
Local Group with M/j < — 13 (the low luminosity limit of our luminosity function; Section
in]) have mean surface brightnesses within the range we can sample (e.g. Mateo 1998; Grebel,
Gallagher & Harbeck 2003).

Figure |3] shows the completeness of the entire SHELS redshift survey as a function
of observed mean surface brightness within the half light radius /X5o,_r- The completeness
is simply the fraction of galaxies in the photometric catalog with measured redshifts as a
function of observed /i5o,_R- The dashed histogram shows the fractional completeness for the
sample of galaxies with R< 20.3; the solid histogram refers to the sample with R< 20.6. We
took substantial care in repeat Hectospec fields to obtain the high level of completeness for
R< 20.3. Most of the incompleteness results from geometric constraints; we fail to sample
the corners and edges of the field as well as we sample the central portion.

2.3. Comparison of SHELSq i with the SDSS

Blanton et al. (2005b) provide a benchmark for measurement of the faint end of the
galaxy luminosity function in the local universe in the r-band. They carefully analyze the
extensive NYU-VAGC catalog (Blanton et al. 2005a) of SDSS galaxies with r < 17.77. Here
we compare the central average surface brightness distributions and the surface brightness-
luminosity relations for the NYU-VAGC and SHELSq.i. Because Blanton et al. (2005) study
the r-band luminosity function, we focus on this comparison.

We first compare the surface brightness distribution of the SHELSq.i sample with the
SDSS sample. Figure Hlshows the distribution of observed mean SDSS r-band surface bright-
ness, /i5o,r, for the galaxies in the SDSS sample with z < 0.05 (solid gray histogram). The
heavy solid line shows the observed r-band surface brightness distribution for the SHELSq.i
galaxies with 0.02 < z < 0.05. We scale the SHELSq.i histogram by the relative area of the
two surveys.

Figure m demonstrates two important aspects of SHELSq.i. For observed fi^o^r around
the peak SDSS sensitivity, SHELSq.i has systematically fewer objects than the SDSS sam-
ple. This deficit in the SHELSq.i survey refiects the a priori selection against nearby galaxies
and clusters; the SHELSq.i region is underdense by selection. For observed fi^o^r ^ 22, the
SHELSq.i survey contains a relatively much larger number of objects than the SDSS. This
difference results from the SHELSq.i redshift survey fainter magnitude limit and complete-
ness for mostly blue low surface brightness objects. The surface brightness distribution for

-9-

the NYU-VAGC declines steeply over the surface brightness range 23-24 mag arcsec ^; the
SHELSq.i distribution over this range is essentially flat.

Figure shows the fraction of blue objects with g — r < 0.4 and with z < 0.05 as
a function of /i5o,r for the SDSS (gray histogram) and for SHELSq.i; the blue fraction is
substantially larger in SHELSq.i.

Because the SDSS has a substantial (known) incompleteness for /i5o,r ^ 23, Blanton et
al. (2005b) model the incompleteness based on an extrapolation of the relationship between
surface brightness and absolute magnitude for galaxies with M^ < —18.0. We can examine
the relationship between the rest frame surface brightness SBso,^ within the half-light radius
and the absolute magnitude M^ in SHELSq.i and thus test the Blanton et al. (2005b)
relationship over a larger range in absolute magnitude. For comparison with Blanton et al.
(2005b), the median R-r = 0.279 for SHELSq.i galaxies.

The absolute R-band magnitude is

MR = mR- 5\ogdL - 25.0 - kniz)

where di is the luminosity distance in Mpc for a Hubble constant H = lOOh km s~^ Mpc~^
and kji{z) is the R-band k-correction derived according to the procedure in Westra et al.
(2010). We use the Hubble constant normalized to 100 for easy comparison with Blanton et
al. (2005b) and others.

The rest frame surface brightness is

SB^o^R = H5o^R - 10/o5fiQ(l + z) - kR{z).

This equation implies that, for example, a typical survey galaxy with /isq^^ = 25.0 (at the
low surface brightness limit of Figure [5]) has SBsq/j ~ 24.3 at the z = 0.1 survey limit.

Figure E] shows the relation between SBsq^j and absolute magnitude M^. There is only
one galaxy in the survey that lands outside the plot limits. It has SBsq^j = 25.1 and is
obviously contaminated by a nearby bright star; we thus omit it.

We use a Bayesian approach to quantify the correlation between SBsq^^ and absolute
magnitude M^. This approach is similar to the one used by Blanton et al. (2005b) but not
identical. SBsq^/j and IsAr clearly do not have a one-to-one relation, but, at fixed M/j, SBsq^/j is
distributed according to some probability density distribution (PDF), p{SB^q^r\Mr). In the
Bayesian approach, we can infer the parameters of this PDF. We do not need to assume that

-10-

the spread originates from random measurement errors around the ideal relation SB^qj^ =
a + 6Mr, as in a standard linear regression. In addition, unlike the usual fitting technique,
we can model the uncertainties in the individual measures as random variates.

We assume that SB^q^r is normally distributed around the mean {SB^o^n) = a + bM^
with variance af^^. We then need to determine the three parameters a, b, and (Tint and their
PDFs. We assume flat priors for both a and b. For the inverse of the variances of the
individual measures, l/asB^oR"^ ^^^ ^/<^Mu^, we adopt the usual assumption that they are
random variates drawn from a Gamma distribution with large variance (e.g. Andreon and
Hurn 2010). This choice is appropriate for quantities that are positively defined and provides
basically flat priors for the uncertainties.

We use the free software JAGS developed by Martyn Plummeio to run Markov Chain
Monte Carlo simulations. We estimate the PDFs of our three parameters by running 3 x 10^
iterations. For the full sample SHELSq.i we find a = 31.08+|J:^?, b = 0.538+[J;|5^?, ajnt =
0.7911o;q25- Figure [6] shows the sample and the result of the Bayesian analysis of the entire
sample (solid line).

To compare more closely with the analysis of Blanton et al. (2005b), we examine
the SB5o,h-Mh relation omitting the redder objects. Rather than explicitly using color to
segregate the galaxy populations, we use the redshift independent spectroscopic indicator
D„4000. Woods et al. (2010) and Kauffman et al. (2003) show that the distribution
of this indicator is bimodal and that it can be used to segregate galaxies with younger
stellar populations (generally blue objects) from those dominated by an old stellar population
(generally red objects). At low redshift segregation by D„4000 is essentially equivalent to
segregation by g — r. Woods et al (2010) divide their sample at the local minimum between
the two peaks, D„4000 = 1.44.

For the 431 blue objects with D„4000< 1.44, we find a = 29.92^°:^°, b = OAeitoml
c"int = 0.742]'^q;q25- Figure [6] shows the 431 objects as open circles, the solid points represent
galaxies with D„4000> 1.44, and the dashed line represents the result of the Bayesian anal-
ysis. The slope b is nearly identical to the value, 0.45, obtained by Blanton et al. (2005b)
for SDSS galaxies with Sersic index n < 2 and M^ < —18.

In their model for p(5'i?5o,_R|MR), Blanton et al. (2005b) allow for an increase in cjint
for less luminous galaxies (for results in the B-band see e.g. de Jong & Lacey (2000); Cross
& Driver (2002); and Driver et al. (2005)). Our data do not support such an increase. For
galaxies with M/j < —18 we find (Tint = O-^SGlQ'Qlg, very similar to the result for the sample

i-www-fis.iarc.fr/'^martyn/software/jags/

-li-

as a whole. The value of a-mt is, however, consistent with the SDSS value at M^ ~ —18. The
overall consistency of the SDSS results with the DLS is remarkable.

2.4. SHELSo.i Redshift Survey Completeness

The completeness of SHELSq.i to the limiting apparent magnitude may, in principle,
differ from the survey as a whole. Figure [7] shows our approach to estimating the completeness
of the SHELSq.i sample. We use a combination of color and surface brightness as a proxy
for redshift to evaluate the completeness for SHELSq.i (Kurtz et al. 2007).

The three panels in the left-hand column of Figure [7] refer to the entire SHELS sample
with R< 20.3; the right-hand panels show the SHELS sample with 20.3 < R < 20.6. All
of the panels show the observed SDSS g — r color as a function of the mean DLS surface
brightness within the half-light radius, fi^o.R. The gray points in the bottom panels represent
the entire photometric SHELS sample including galaxies with and without a redshift; the
black points in the upper panels show the total number of these galaxies that lack a redshift.
The red points in the central panel indicate galaxies with redshift z > 0.5 and the blue
points indicate galaxies with z < 0.1, the SHELSq.i sample. Not surprisingly, galaxies with
20. 3< R < 20.6 overlap substantially with the z > 0.5 portion of the redshift survey.

The green line in the central and upper panels is an arbitrary delineation of the boundary
of the SHELSq.i sample in magnitude-color space. 93% of the apparently brighter SHELSq.i
galaxies with R< 20.3 with redshifts are blueward of the line and 87% of the fainter galaxies
with 20.3 < R < 20.6 are blueward of the line. We use this admittedly arbitrary line to
estimate the incompleteness of SHELSq.i. The black points in the upper panels show all
of the SHELS galaxies without redshifts. To estimate the completeness of SHELSq.i we
estimate the fraction of these black points that are probably galaxies with z < 0.1. Note
that most of the black points, regardless of apparent magnitude correspond to red galaxies,
most probably at high redshift.

Table [2] lists the galaxy candidates below the green line and without redshifts; these
are the objects most likely to be missing from our low redshift sample. For R< 20.3, 21%
of the galaxies with redshifts and below the green line are at z < 0.1; for 20. 3> R < 20.6
this fraction is 11%. Note that the gray points below the green line in the middle and
upper panels have redshifts between 0.1 and 0.5. Only the black points lack a redshift and
statistically we expect that most of these are at z > 0.1. In fact, assuming that the fraction
with z < 0.1 is the same among the galaxies without measured redshifts (black points), we
expect that only ~ 14 of the 86 objects in the Table are at z < 0.1. We note that many of the

-12-

objects in Table [2] are near the edges of the field (particularly the higher surface brightness
objects).

To estimate the completeness of SHELSq.i, we compute the fraction of SHELS redshifts
with 2; < 0.1 both above and below the fiducial green line. We note that these estimates
are insensitive to the exact position of the green line. We then assume that these same
fractions of the galaxies without redshifts are probably at z < 0.1. For R< 20.3, SHELSq.i
is 98% complete. The differential completeness in the interval 20.3 < R < 20.6 for SHELSq.i
is 92%, greater than the 89% for the entire SHELS sample because most of the objects
without redshifts are small, faint, and red. Photometric redshifts from SDSS substantiate
this analysis. In summary, we estimate that SHELSq.i is 97% complete to R = 20.6; we are
missing only 14±4 objects. Because our samples are substantially complete, we make no
corrections for incompleteness in our calculation of the luminosity function.

3. The Galaxy Luminosity Function

The SHELSq.i survey probes a small region of the universe to a faint, uniform limiting
observed surface brightness. Figure [8] shows the redshift of each survey galaxy as a function
of the observed DLS R-band surface brightness within the half light radius, /i5Q,/j. Large-
scale structure in the region is obvious in the highly clustered redshift distribution. The
distribution of observed surface brightness reaches the survey limit at every redshift.

Figure [8] also provides some insight into the galaxy populations. Again, we segregate
the galaxy populations based on the spectroscopic indicator D„4000. Open circles indicate
galaxies with a predominantly young population and with D„4000< 1.44; the solid circles
denote galaxies with D„4000> 1.44.

In SHELSq.i galaxies dominated by an old population are rare and they tend to be
higher surface brightness objects. They appear predominantly in the densest structure in
the survey at 2; ~ 0.062, a reflection of the standard morphology- density relation. The
general absence of low surface brightness red objects does not result from selection. In fact,
the R-band DLS photometric data are actually more sensitive to these objects than to low
surface brightness blue objects.

The lowest surface brightness galaxies are generally also the lowest luminosity objects
(Figure [6]). Thus Figure [8] underscores the result previously obtained by Blanton et al.
(2005b): the faint end of the luminosity function is dominated by low surface brightness
galaxies dominated by a young stellar population. These galaxies are usually blue. It is
interesting that even though we select our galaxies at R, there are no galaxies with pre-

-13-

dominantly old populations and yUso,/? ^ 23.5 mag arcsec"^. As a result of (1 + z)"^ and
K-dimming, the lowest surface brightness objects in Figure 2 are at 2; > 0.1 and thus do not
appear in Figure 8.

We compute the luminosity function for SHELSq.i with 0.02 < 2 < 0.1 (532 galaxies)
and for three subsets of this sample separated by rest-frame surface brightness within the
half-light radius, SBsq^^r. The restriction to 2; > 0.02 reduces the effect of peculiar velocities
on the determination of the luminosity function and sets a low luminosity limit on the
luminosity function of M,. = —13.3 + 51og/i.

The lowest luminosity galaxies in the SHELSq.i are at z < 0.015 and do not enter into
the luminosity function calculation. Figure [9] shows the five lowest luminosity galaxies listed
in Table [TJ Galaxies b)-e) all have R< 20.6 and their rest frame mean surface brightness
within the half light radius SBso,/? > 22.5 mag arcsec"^. Galaxy a), the lowest luminosity
galaxy in our sample, is just fainter than the SHELSq.i magnitude limit; it has R = 20.69.
Not surprisingly, these galaxies are blue. Many of the spectra show Balmer absorption
characteristic of a predominantly young stellar population. Most of the spectra show Ha
emission.

Table [3] gives the number of galaxies in each of the samples we analyze. The HSB
(high surface brightness) sample includes all galaxies with SBso,^ < 21.82 mag arcsec"^, the
median for the sample. The LSB sample includes the galaxies with SBso^ij > 21.82 mag
arcsec"^. We note that Blanton et al. (2005b) explore the impact of dividing their sample
by surface brightness (Figure 21); they divide their sample at /i5o,r = 21 ~ SB^q^r = 20.7
mag arcsec"^, a higher surface brightness by nearly a magnitude arcsec"^ than the median
surface brightness for SHELSq.i.

To explore the dependence on surface brightness more fully we also compute the lumi-
nosity function for galaxies with SBsq^/j > 22.5 mag arcsec"^. This subsample, SBsq^/j > 22.5
mag arcsec"^, satisfies the useful definition of low surface brightness galaxies given by O'Neil
(2002): the central surface brightness is a magnitude fainter than the night sky. To compute
our limit we took the 20% percentile darkest night sky brightness at Gemini, R = 20.4 mag
arcsec"^ (lhttp://www.gemini.edu/sciops/telescopes-and-sites/observing-condition-constraints/optical-sky-
as our fiducial value. We model the objects as pure exponential disks to compute our low
surface brightness limit, SB5o,ij > 22.5 mag arcsec"^. In this simple exponential disk model,
the mean surface brightness within the half-light radius (the quantity we use in this study)
is 1.1 magnitudes fainter than the central surface brightness.

We apply the well-known SWML (step-wise maximum likelihood) technique (Efstathiou,
Ellis & Peterson 1988) to compute the luminosity function. We also applied the less widely

-14-

used (but also non-parametric) C^ (Lynden-Bell 1971) and LCCP (Takeuchi, Ishikawa &
Ichii 2000) techniques to the data; the results are indistinguishable from the SWML results
and for simplicity and clarity, we do not report them here.

Figures [10] and [12] show the results of the luminosity function calculation for the entire
sample and for the three subsamples. Figures [11] and [13] show the corresponding confidence
contours of the luminosity function parameters. The points in the luminosity function plots
show the SWML results. We use the bootstrap method to compute the uncertainty at each
point. We resample the galaxy sample 50 times for each luminosity function computation.

Of course, the SWML technique does not assume a form for the luminosity function.
We represent the SWML results with a fit to a single Schechter (1976) function:

(l)iM)dM = O.4/nlO0*lO°-^("+i)(*^*-^^)exp[-lO°-^(*^*-^^)]rfM

where a is the faint end slope, M* is the characteristic magnitude, and 0* is the normalization.
With their much larger dataset Blanton et al. (2005) fit a double Shechter function. We
obtain reasonable x^ fo'^ the single Schechter function fits (Table [3]).

Table [3] lists the luminosity function parameters for the entire SHELSq.i sample and
for each of the three subsamples. Table [3] also lists the luminosity density for each sample.
Application of the parametric STY method (Sandage, Tammann & Yahil 1979) to our data
yields similar results.

Because the volume of SHELSq.i is small, the bright end of the luminosity function and
the values of M*^ are poorly constrained and cannot be compared with other estimates; the
values of M*^ and of the luminosity density, C, are only useful for comparison of different
subsamples of the SHELSq.i data. Our intent here is to focus on the faint end of the
luminosity function and on the contribution of low surface brightness galaxies to the slope.

For the sample as a whole the faint end slope, a = —1.31 ± 0.04, is reasonably well-
determined. This result is the same as the r-band faint-end slope Blanton et al (2005)
obtained without correction for missing low surface brightness objects (a2 = —1.34 ± 0.01).
This faint end slope is also consistent with other measurements of field and cluster luminosity
functions derived for low redshift samples with complete spectroscopy (e.g. Christlein &
Zabludoff 2003; Mahdavi et al. 2005; Rines & Geller 2010).

The luminosity density for the SHELSq.i sample is lower than the value obtained by
Blanton et al (2005b) and others as expected based on the selection of the region. We also
note that the xl P^r degree of freedom is reasonable for the single Schechter function fit;
the sample is too small to support the more complex approach of fitting a double Schechter

-15-

function as in Blanton et al (2005b).

The luminosity functions for the HSB and LSB subsamples demonstrate that the lower
surface brightness galaxies dominate the faint end of the luminosity function. The slope
for the HSB subsample is quite shallow, uhsb = —0.69 ± 0.07. The faint-end slope for the
LSB sample is aisB = —1.57 ± 0.09. This value is somewhat steeper than the fit Blanton
et al. (2005b) obtain for their r-band luminosity function corrected for missing low surface
brightness galaxies (^2 = —1.40 ± 0.01).

Because there is little correction for surface brightness incompleteness to our limiting
magnitude, the steep faint end slope is an empirical determination of the impact of low
surface brightness galaxies on the luminosity function. We make an additional empirical
test of this conjecture by extracting a subsample of SHELSq.i that lies within 1° of the field
center where the integral completeness of SHELS to R = 20.6 is 97% (rather than 96% in
the full area). In this region there are 416 galaxies with z < 0.1 and we estimate that the
completeness of SHELSq.i is 99% according to the technique demonstrated in Figure 7. In
other words we are missing 6±2 galaxies. When we recompute the luminosity functions for
all of the subsamples in surface brightness considered here, the faint end slopes agree with
those in Table 3 to within la for the original sample.

The value of M|j for the LSB sample is fainter than for the HSB sample as expected
from the correlation between surface brightness and luminosity. It is interesting that the
fraction of the luminosity density contributed by the LSB half of the sample is only 16% to
the limiting surface brightness and absolute magnitude we sample.

Our lowest surface brightness sample SBso,/? > 22.5 mag arcsec"^ contains 135 galaxies
and thus the error in a is large. However, there is no other published luminosity function
derived from a highly complete redshift survey of such low surface brightness galaxies. The
faint end slope is 0:22.5 = —1.52 ± 0.16, consistent with the slope we obtain for the larger
LSB sample.

4. Discussion

Determination of the faint end slope of the galaxy luminosity function is sensitive to
the inclusion of low surface brightness galaxies. Although these galaxies make a relatively
small contribution to the total luminosity density, they dominate the count of objects at low
luminosity.

The identification of low surface brightness galaxies from a photometric survey is a

-16-

challenging problem in itself and the acquisition of a spectroscopic redsliift for the lowest
surface brightness objects is time-consuming even with a large telescope. One alternative
approach is the use of photometric redshifts. Here we compare our results with the luminosity
function obtained by Liu et al. (2008) for the redshift range z = 0.02 — 0.1 in the 2 deg^
COSMOS field. In this redshift bin they compute the faint end slope to a limiting My ~
— 12.8, comparable with our R-band limit of ~ —13.0 for the same redshift interval.

Liu et al. (2008) use photometric redshifts computed according to the method of
Mobasher et al. (2007). To compute the luminosity function, Liu et al. (2008) treat each
galaxy as a weighted probability-smoothed luminosity distribution. They apply a modified
version of the 1/Vmax method and use numerical simulations of their procedure to assess the
biases and random errors in their technique.

For their entire sample, Liu et al. (2008) obtain a faint-end slope, acosMOS = —1-24 ±
0.07, remarkably consistent with our a = — 1.31±0.04. Liu et al. (2008) segregate their sam-
ple according to spectral energy distributions roughly corresponding to various morphological
types. Their ScD+lrr bin is probably the most comparable with the low surface brightness
portion of our sample. For this subsample, they obtain acosMos,ScD+irr = —1.46 ± 0.07 in
essential agreement with our aisB = —1.57 ± 0.09. It is impressive that two very different
techniques yield such similar results over the same redshift range.

Liu et al. (2008) comment that the faint-end slope in their z = 0.02 — 0.1 bin may
be "qualitatively" dominated by low surface brightness dwarfs that are not detected in
their survey at higher redshift. The SHELSq.i faint-end slope is dominated by low surface
brightness dwarfs. Thus the essential agreement of the COSMOS low redshift faint-end
slope with SHELSq.i argues strongly that low surface brightness dwarfs (and not evolution)
account for the relatively steeper faint end slope at the lowest redshifts in the COSMOS
sample. The comparison of SHELSq.i with the COSMOS results underscores the importance
of cleanly defined and measured surface brightness limits in surveys addressing the galaxy
luminosity function and its evolution.

5. Conclusion

Measurement of the faint end slope of the galaxy luminosity function requires attention
to lower surface brightness objects that dominate the count at lower luminosities. This issue
is, of course, important at all redshifts and failure to understand surface brightness limits may
lead to apparent evolution of the faint end of the luminosity function with redshift. We use
the SHELS redshift survey of one of the DLS survey fields to evaluate the faint end slope at

-17-

low redshift and to examine its sensitivity to surface brightness. The DLS photometry offers
the possibihty of identifying galaxies with lower surface brightness (the photometric survey
is complete to a mean surface brightness within the half-light radius of 27.0 mag arcsecond"^
at R) and, in carrying out the spectroscopic survey, we paid attention to acquiring redshifts
for these lower surface brightness objects. The lowest surface brightness objects we identify
by careful inspection of all of the galaxy candidates in the field are above the photomettric
detection limit. Two- hour integrations with Hectospec on the MMT are adequate to acquire
a redshift even for the lowest surface brightness objects in the photometric catalog.

We calculate the faint end slope in the R-band for the subset of SHELS galaxies with
redshifts in the range 0.02< z < 0.1, SHELSq.i. This sample contains 532 galaxies with
R< 20.6 with a median surface brightness within the half light radius of SBsq,/? = 21.82 mag
arcsec"^. We estimate that there are only 14± 4 objects missing from this sample. We used
this sample to make one of the few direct measurements of the dependence of the faint end
of the galaxy luminosity function on surface brightness.

We compare the properties of the SHELSq.i galaxies with the much larger, but shallower
NYU-VAGC derived from the SDSS (Blanton et al. 2005a, 2005b). We show that SHELSo.i
has a fainter median observed surface brightness /i5o,r within the half light radius. The
SHELSq.i sample thus enables a strong test of the relation between absolute magnitude and
surface brightness that Blanton et al. (2005b) use to correct their luminosity function for
missing low surface brightness objects. We use a Bayesian approach to evaluate the correla-
tion between magnitude and surface brightness; we derive a slope of 0.46±0.02, essentially
identical to the slope derived by Blanton et al (2005b) for galaxies with M,. < — 18 + 5logh.
Our R-band relation extends to Mr ~ — 14-|-51og/i with essentially constant variance around
the mean relation.

We compute the faint end slope of the luminosity function for the SHELSq.i sample as a
whole and for three subsamples segregated by mean surface brightness within the half light
radius. For the sample as a whole the faint end slope a = —1.31 ± 0.04, is consistent with
both the Blanton et al. (2005b) analysis of the SDSS and the Liu et al. (2008) analysis
of the COSMOS field. This consistency is impressive given the very different approaches of
these three surveys: SDSS is a large area shallow survey with spectroscopy; COSMOS is a
deep photometric survey with an analysis based on photometric redshifts; and SHELSq.i is a
dense spectroscopic survey with a 6.5-meter telescope of a photometric catalog derived from
5 hours of integration in better than 0.9" seeing on a 4-meter telescope.

A magnitude limited sample of 135 galaxies with optical spectroscopic redshifts and
with mean surface brightness, SBsq^/? > 22.5 mag arcsec"^ is unique to SHELSq.i. The faint
end slope is 0:22.5 = —1.52 ± 0.16, consistent with previous limits on similarly low surface

-18-

brightness populations from independent samples. Because SHELSq.i samples a low density
region of the universe by construction, these low surface brightness objects are predominantly
blue. Surveying a larger volume to the depth of SHELSq.i would be an important basis for
exploring the dependence of the faint end of the luminosity function on environment and
galaxy type.

We thank Scott Kenyon for many insightful discussions. We thank Beth Willman for
a discussion about Local Group dwarfs. We also thank Michael Strauss and Marijn Franx
for encouraging us to use the SHELS survey as a basis for studying the local luminosity
function. We thank Michael Blanton and his collaborators for providing the NYU-VAGC
on line and we thank Martyn Plummer for his publicly available JAGS code. Perry Berlind
and Michael Calkins operated the Hectospec efficiently and with extraordinary expertise.
We thank Susan Tokarz for her dedication to careful data reduction. Nelson Caldwell deftly
manages Hectospec queue scheduling for optimal scientific results. We thank the anonymous
referee for urging us to improve the clarity of the presentation here. This paper makes use of
data from the Sloan Digital Sky Survey. The Smithsonian Institution partially supported this
research. NSF grant AST-0708433 supports Ian Dell'Antonio's research at Brown University.
INFN grant PD51 and the PRIN-MIUR-2008 grant 2008NR3EBK_003 "Matter-antimatter
asymmetry, dark matter and dark energy in the LHC era" partially support Antonaldo
Diaferio.

Facilities :MMT{}lectospec)

-19-

REFERENCES

Andreon, S., & Hum, M. A. 2010, MNRAS, 404, 1922

Balogh, M. L., Morris, S. L., Yee, H. K. C, Carlberg, R. G., & Ellingson, E. 1999, ApJ, 527,
54

Benson, A. J., Lacey, C. G., Baugh, C. M., Cole, S., & Frenk, C. S. 2002, MNRAS, 333, 156

Benson, A. J., Bower, R. G., Frenk, C. S., Lacey, C. G., Baugh, C. M., & Cole, S. 2003,
ApJ, 599, 38

Blanton, M. R., et al. 2005a, AJ, 129, 2562

Blanton, M. R., Lupton, R. H., Schlegel, D. J., Strauss, M. A., Brinkmann, J., Fukugita,
M., & Loveday, J. 2005b, ApJ, 631, 208

Christlein, D., & Zabludoff, A. I. 2003, ApJ, 591, 764

Cross, N., & Driver, S. R 2002, MNRAS, 329, 579

de Jong, R. S., & Lacey, C. 2000, ApJ, 545, 781

Disney, M., & Phillipps, S. 1983, MNRAS, 205, 1253

Driver, S. R, Liske, J., Cross, N. J. G., De Propris, R., & Allen, R D. 2005, MNRAS, 360,
81

Efstathiou, G., Ellis, R. S., & Peterson, B. A. 1988, MNRAS, 232, 431

Fabricant, D. G., Hertz, E. N., Szentgyorgyi, A. H., Fata, R. G., Roll, J. B., & Zajac, J. M.
1998, Proc. SPIE, 3355, 285

Fabricant, D., et al. 2005, PASP, 117, 1411

Fabricant, D. G., Kurtz, M. J., Geller, M. J., Caldwell, N., Woods, D., & DelPAntonio, I.
2008, PASP, 120, 1222

Geller, M. J., DelPAntonio, I. P., Kurtz, M. J., Ramella, M., Fabricant, D. G., Caldwell, N.,
Tyson, J. A., & Wittman, D. 2005, ApJ, 635, L125

Geller, M. J., Kurtz, M. J., DelLAntonio, I. P., Ramella, M., & Fabricant, D. G. 2010, ApJ,
709, 832

Grebel, E. K., Gallagher, J. S., HI, & Harbeck, D. 2003, AJ, 125, 1926

-20-

Impey, C. D., Sprayberry, D., Irwin, M. J., & Bothun, G. D. 1996, ApJS, 105, 209

Kauffmann, G., et al. 2003, MNRAS, 341, 33

Kurtz, M. J., Geller, M. J., Fabricant, D. G., Wyatt, W. F., & DelFAntonio, I. P. 2007, AJ,
134, 1360

Kurtz, M. J., & Mink, D. J. 1998, PASP, 110, 934

Liu, C. T., Capak, P., Mobasher, B., Paglione, T. A. D., Rich, R. M., Scoville, N. Z.,
Tribiano, S. M., & Tyson, N. D. 2008, ApJ, 672, 198

Lynden-Bell, D. 1971, MNRAS, 155, 95

Madgwick, D. S., Lahav, O., Baldry, I. K., et al. 2002, MNRAS, 333, 133

Mahdavi, A., Trentham, N., & Tully, R. B. 2005, A J, 130, 1502

Mateo, M. L. 1998, ARA&A, 36, 435

McGaugh, S. S. 1996, MNRAS, 280, 337

Mink, D. J., Wyatt, W. F., Caldwell, N., Conroy, M. A., Furesz, G., & Tokarz, S. P. 2007,
Astronomical Data Analysis Software and Systems XVI, 376, 249

Mobasher, B., et al. 2007, ApJS, 172, 117

MuUer, G. P., Reed, R., Armandroff, T., Boroson, T. A., & Jacoby, G. H. 1998, Proc. SPIE,
3355, 577

O'Neil, K. 2002, Extragalactic Gas at Low Redshift, 254, 202

Roll, J. B., Fabricant, D. G., & McLeod, B. A. 1998, Proc. SPIE, 3355, 324

Rines, K., & Geller, M. J. 2008, AJ, 135, 1837

Sandage, A., Tammann, G. A., & Yahil, A. 1979, ApJ, 232, 352

Schechter, P. 1976, ApJ, 203, 297

Sprayberry, D., Impey, C. D., Irwin, M. J., & Bothun, G. D. 1997, ApJ, 482, 104

Takeuchi, T. T., Yoshikawa, K., & Ishii, T. T. 2000, ApJS, 129, 1

Westra, E., Geller, M. J., Kurtz, M. J., Fabricant, D. G., & Dell' Antonio, I. 2010, PASP,
122, 1258

-21 -

White, S. D. M., & Rees, M. J. 1978, MNRAS, 183, 341

Wittman, D., Dell' Antonio, I. P., Hughes, J. P., Margoniner, V. E., Tyson, J. A., Cohen,
J. G., & Norman, D. 2006, ApJ, 643, 128

Woods, D., Geller, M. J., Kurtz, M. J., Westra, E., Fabricant, D. G., & DelPAntonio, I.
2010, AJ, 139, 1857

This preprint was prepared with the AAS E^TJtjX macros v5.2.

Table 1. SHELSm Redshifts^

SHELS id

SDSS objID

i'a2ooo

dec

2000

R

R Source

z Source''

(5z

M50,fl

mag arcsec"

138.7082030+30.
138.7106146+30.
138.7245438+30.
138.7254512+30.
138.7268120+30.
138.7271334+29.
138.7289058+30.
138.7325010+30.
138.7353477+30.
138.7358290+30.

4863253 587738947740697041

1474856 588017978876101272

2541523 588017978876166576

6434571 588017979413103093

4084400 587738947740696966

2781541 588017977802228132

2024933 588017978876166301

0934152 587738947203760418

4341614 587738947740696611

2827235 588017978876166629

9:14:49.969

30:29:10.771

19.5084

DLS

0.0218412

MMT

32.3

22.5807

9:14:50.547

30:08:50.948

19.6469

DLS

0.0373713

MMT

47.1

24.1041

9:14:53.891

30:15:14.948

18.8388

DLS

0.0236177

MMT

32.1

22.1238

9:14:54.108

30:38:36.446

18.7908

DLSm

0.0238024

MMT

81.9

23.4231

9:14:54.435

30:24:30.384

20.9377

DLS

0.0628633

MMT

40.9

22.8286

9:14:54.512

29:16:41.355

19.4561

DLS

0.0206206

MMT

50.6

22.7012

9:14:54.937

30:12:08.976

15.7759

DLS

0.0232745

MMT

18.4

20.67t^

9:14:55.800

30:05:36.295

18.3344

DLS

0.0238612

MMT

43.6

23.ic|p
21.69

9:14:56.483

30:26:02.981

17.0162

DLS

0.0222926

MMT

34.3

9:14:56.599

30:16:57.805

20.6605

DLS

0.0222285

MMT

47.6

23.1852

^Table 1 is published in its entirety in the electronic version of the Astronomical Journal. The first ten lines are shown here for guidance regarding its
form and content.

''The R-band magnitude source is either the DLS or the SDSS. We translate SDSS r to R using the redshift and the SDSS colors. DLSm or SDSSm
indicates that the magnitudes required a detailed calculation outside the pipeline for the survey generally as a result of a nearby star and/or an artifact
in the imaging data.

''The redshift z is from the Hectospec on the MMT (MMT) or from the SDSS. The error Sz is also from the redshift source.

-23-

Table 2. Low Redshift Candidate List^

SHELS id

SDSS objID

ra2ooo

dec2ooo

R

mag arcscc^^

g-r^

138.7089504+30.9843187

587738948277698932

9:14:50.148

30:59:03.547

20.5556

23.3638717

0.521

138.7255980+30.0542613

587738947203760545

9:14:54.144

30:03:15.341

20.5392

23.4122087

0.507

138.7264737+30.3206758

588017978876166652

9:14:54.354

30:19:14.433

19.1392

23.6712299

0.265

138.7291346+30.4367746

587738947740697008

9:14:54.992

30:26:12.389

20.3627

23.0114264

0.612

138.7487849+29.4746605

587738946666758726

9:14:59.708

29:28:28.778

19.9382

24.0572414

0.498

138.7539734+29.7394633

588017978339164239

9:15:00.954

29:44:22.068

20.0661

22.5720023

0.563

138.7563217+31.0219574

587738948277699044

9:15:01.517

31:01:19.047

20.3730

21.775373

0.385

138.7582448+30.9727119

587738948277698992

9:15:01.979

30:58:21.763

20.1276

23.2405294

0.295

138.7623082+29.7296108

588017978339164237

9:15:02.954

29:43:46.599

19.5096

22.6763703

0.389

138.7648117+31.0037925

587738948277698637

9:15:03.555

31:00:13.653

20.5552

22.4070289

0.622

''Table 2 is published in its entirety in the electronic version of the Astronomical Journal. The first ten lines are
shown here for guidance regarding its form and content.

^The g

color is the SDSS fiber color.

-24-

Table 3. Luminosity Functions

All

HSB

LSB

5^50,^ > 22.5

N

532

266

266

135

a

-1.31 ±0.04

-0.69 ±0.07

-1.57±0.09

-1.52±0.16

M^ - 5\ogh

-21.32 ±0.30

-20.42 ±0.15

-18.72 ±0.34

-17.32 ±0.34

0710-3 mag-i h^ Mpc-3

4.22 ±0.96

11.7±1.6

4.7 ±2.0

8.5 ±4.4

xl

1.28

0.71

1.34

0.85

V

16

14

11

9

C/lO^hLQ Mpc-3

0.96 ±0.35

0.80 ±0.25

0.15 ±0.08

0.07 ±0.04

-25-

QC

M

(a

B -

" 1 '«' 14: '-:'',' W;.' ' ' 1 '

— F 1 \ —

•^ '"i'ty^m^^'.

tC^

, y

^ ^ j^''^ ''^i^^^^^^'^ndBNB^^H

■A.. «x ^{ ^"^ "^ '^t^MBuCSBiSB^^^^H

^Hft>- ..

vk. y;i^ ,^ " * ^:^^B^^^

^^^^ki"^'

'^^f^^%^^M|

^H

^^H

^^^H

^^M

"^^H^^^l

^H

^^^^^1

^ ^^^m^^^^^^^^^^^^^^i

^^^H

, r . . . 1 'f^S^^S^B

^^H

8

14

15

18

20

22

§ -

T — I — I — I — I — I — I — I — n — I — r

14 JB

22

I ' ' ',■!'": -'^'!V;| ' ' ' I

»*s.

22

8

6 -

a -

-I— I — I — I — I — I — I — t — I — 1 — I — I — I — I — I — I — r

14 IB 18 20 22

R

Fig. 1. — SHELS galaxy selection from DLS photometry. R1.5" is the DLS magnitude within
the 1.5" Hectospec fiber aperture. R is the total DLS apparent magnitude. The vertical
green line shows the SHELS survey limit R = 20.6. Gray points show all of the objects
detected in the DLS survey; the black points (upper right panel) show all of the galaxies
with SHELS spectra and R< 21. Most of the "objects" at large R1.5" — R (above the main
locus of objects with SHELS spectra) and with R< 20.6 are artifacts. In the lower two
panels, yellow points represent all galaxies with measured redshifts. On the lower left, black
points represent all galaxy candidates with R< 21 and without a measured redshift. In the
lower right, black points represent SHELSq.i galaxies with redshift z < 0.1, blue points are
galaxy candidates without a redshift.

-26-

O

28

1

—

■. .= ■ ■■.:.■ ■.vvV:;:;'=;:.-:^?-'=;?\--.v-.. V - . ■

-

"■ ■ ■■■■ i^^.;; v^=■■■;%■^v■>■;^^^^^^ ■■

-

• ■ ■■■■•/■ ■'■■:'^: r''r}:,':^'~'.S/Cri..\-. ■ ■ ■ .■

26

—

—

-

■ • . rV..- ■■■■; ^V ■■■i:» •"£:■'.;;->■;■'' ^^ ■■•■■.■■.■■■ ■ ' ..

■ ' ■ ' -

24

—

■' ■♦'■■i-iltV '■■ .i^ ■';■'-/. J'- '■'-.'W ■.:■'■'. .

—

■; ■

. J^^^^^K* ' ' -' .

-

j^^^Bp^'

22

— A

—

'*''| 4II

Hr

- ' *^«i^|

Hi*

-

p

■

on

■■■■.> '"^P

1 1 1 1 1 1 1 1 1

,

Ri.5" - R

Fig. 2. — DLS observed mean surface brightness within the half-hght radius, /iso,/? as a
function of the magnitude difference R1.5" — R in the classification diagram of Figure [1]
Gray points denote all of the DLS objects in the apparent magnitude range 19 < R < 20.6.
Black points denote all of the galaxy candidates (with or without a redshift). Note that the
lowest surface brightness galaxies are all above the survey limit and that most of the low
surface brightness objects are artifacts.

-27-

0.8 -

" 0.6

c

0)

Si
"a

S
o
u

0.4 -

0.2 -

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

1 1 1 1 1 1 1 1 1 1 1 1

II,,
II,,

1 1 1 1 1 1 1 1 1 1 1 1

1

1"

1
1
1

1

ij

, 1
, 1
, 1
, 1

—1 J

1

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

18

20

22

24

26

Fig. 3. — SHELS redshift survey completeness as a function of observed surface brightness,
Ai50,i? for galaxies with R< 20.3 (dashed histogram) and for galaxies with R< 20.6 (solid
histogram)

-28-

10000 -

1000

B

G

100 -

10 -

18

20

22

24

26

^'■SO.r

Fig. 4. — Comparison of the observed surface brightness distribution for the SDSS sample
(sohd gray histogram) of Blanton et al. (2005) with galaxies in SHELSq.i that have z < 0.05
(heavy black line), /^so.r is the SDSS observed mean r-band surface brightness within the
half-light radius. The vertical axis shows the actual number of SDSS galaxies; we scale
the SHELSq.i z < 0.05 sample by the relative areal coverage of the two surveys. Note the
relatively larger representation of low surface brightness galaxies in the SHELSq.i z < 0.05
sample .

-29-

1 -

0.8 -

c

o
•■—I

u
Id

3

0.6 -

0.4 -

0.2 -

18

20

22

24

Mso.r

Fig. 5. — Fractions of galaxies with g — r < 0.4 in Figure HI The gray histogram shows
the SDSS blue fraction; the black histogram shows the fraction for SHELSo.i galaxies with
0.02 < z < 0.05. Note the greater blue fraction in the SHELSo.i z < 0.05 sample.

-30-

24 -

22

O

QQ
CO

20 -

18

-

1 1 1 1

r

1 1 1 1 1

1 1 1 1 1 1 1 1 1

o

-

~~

° o o o /

y

-

o

8o o o (90 Oo° ° o>^^
°° ao o o°° °o.>^o

o ^°^* «:^ oX^° • °

O0^°o ol§^^ .0 o^ °

-

-

o

o o
o o

o°

-

—

o°o o''o®o5%02r^cp8o_^OQ30oO o

O o ^ (5 0>^ C)9c)^ "DO ^ o

^o o 9f°^/V^ ^ oVp o

—

o °o /<^

i3 ^.SJO o .
' 8 o o o

o

A

-^ „/i^ •Or

° ''o ^ o '^
/''o./o o o

^0 o;/ *o°oo° Si

>^%^ Oo* \

r ^'^ @ Q.o^

o V o °

o

y

,o^

/ / °

•

•

• o

/

" "X *

• •

-

^

•

•

•f,* o o

/

•

o

• •

• o

-

o

••
, 1 , ,

•
1

1 , , , 1

1 1 1 1 1 1 1 1 1

-22

-20 -18

Mr - Slog h

-16

-14

Fig. 6. — Mean rest frame surface brightness, SBso^ij as a function of absolute magnitude,
Mi^ for SHELSo.i- The sohd hne shows the relation between the two quantities derived from
a Bayesian analysis for the full SHELSo.i sample; the dashed line shows the relation for
the galaxies with D„4000< 1.44. Solid dots denote galaxies with D„4000>1.44; open circles
denote objects with D„4000 < 1.44.

-31-

2.5
2

1.5
1

0.5

-i 1 1 1 1 1 1 1 1 1 r-

2.5
2

1.5
1

0.5

-1 1 1 1 1 1 1 1 TT p 1 1-

- -■* '■ '
■■ ■ ..'j-.V- ■ ■

-I I I-

.'f ■".'j'-t*!"':""*- .

'.. '/■■•<£'.■'■■■. /"■ ■ 3

_i I i_

_l I L_

18

20

22

24

I

2.5
2

1.5
1

0.5

-i 1 r-

-i 1 r-

-i 1 r-

■"^>^^^-'

i ■ . ■ £ ■ ... f'i.4K?-'l /'

_I I L_

18

■'^.■■^■■'iiC*:.\3^-.<H- ■ ■■
.-. r:' /?"is^.r> ;■■■■■■ '^.-

-I I I_

20

-I I l_

22

24

2.5

1 1

1 1 1 1 1 1 1 1 1 .1

T 1

2.5

-

■J .

-

2

;.

■ kidA^^kj. ' ' ■'

—

2

-

. ..aUaWflKSElit' ■ ■

.-

1.5

|-

yi^lSSK^^^^KKm^-' *■'

-_

1.5

1

|-

■f-

''ii^^^HJi^^^^Hs^'' .

1

Oft

I-

.■

—

0.5

_

'." r^^^Sran^^Rf.ij^V.'^ ■'■ '

_

-

■

' '■■ ^\'i '?:,■■■.■ .''■ ■ ■I*"".

-

—

■ . ' . ' ■

—

1

18

20

22

24

1 1 1

1 ' ' ' . 1 " ■ '. . ' 1

1

'-

-j

'-

■■^^Hk"'- "

■ -_

'-

Hp>

•.'r.

\

:W^t^-

-_

1 1 . . ' 1 . . 1 1

18

A'-SO.R

20

22

24

Fig. 7. — SDSS g — r color as a function of DLS observed surface brightness, fi5o,R for SHELS.
This plot is the basis for an estimate of the completeness of SHELSq.i- The left column applies
to galaxies with R< 20.3.; the right-hand column applies to galaxies with 20.3 < R < 20.6.
The bottom panels show all galaxies in the magnitude range (gray points). The central
panels show the low and high redshift portions of the sample: blue points represent galaxies
with z < 0.1 and red points represent galaxies with z > 0.5. Not surprisingly the galaxies
in the 20.3 < R < 20.6 interval are mostly at z > 0.1 (gray and red points in the upper
panel). Most of the low redshift sample lies below the green line. Black points in the upper
panels show all of the objects in the photometric survey without a redshift; only a few points
overlap the color-surface brightness range spanned by the low redshift sample indicated in
blue.

0.1

0.08 -

0.06 -

N

0.04 -

0.02 -

18

-32-

o»
o

-| U wl I* • 1 UTC CXTTJ c 1 1— CT

o o^ o oo o

OO ^ ^ Qo O

o ^Jf^oo OOW®"^ @3"6'o§*'o qOOoo q

o

o

oo f^ o o -.o i^9, ° oo ^

o

o

o • c^ o
o •

O © o o

005 o OqCxS'O o Qd o

J» o

, o o

o

o oo

o

oo
o

o " o a3°oOo o I

oo

O^'irS ° O

'o

O

o o

o^«

o^^ ^o'»,o Q^ o aooo „

o o
o
o

o

o

o

^-o°„%°„ °^

o
o ^„o„ . r^^O

O o° Q) O O O

o

oo
o

8>°p

°« ° ° o o«§°o §0^ ^^J'So^

o o
o

20

22

24

/^50,R

Fig. 8. — Mean observed surface brightness, fi5o,R, as a function of redshift for SHELSq.i
galaxies. Solid points indicate galaxies with D„4000 > 1.44; open circles indicate objects with
D„4000 < 1.44 This division corresponds well with division by color or spectroscopic type.
Note that the large D„4000 (redder) galaxies are predominantly in the densest structure at
z ~ 0.06.

-33-

ld)-1

-BS.O.DlSa

if'WvvA,^

Fig. 9. — The five lowest luminosity galaxies in SHELS ordered in R-band total luminosity.
Galaxy a) has an apparent magnitude R = 20.69, fainter than the SHELSq.i limit. The
bold numbers give the absolute R-band luminosity {h = 1) and the redshift; the yellow
numbers in the 3 arcminute square images give the right ascension and declination of the
galaxy. Galaxies b)-e) have R < 20.6. The objects are all blue and their spectra show Balmer
absorption and/or Ha emission.

-34-

-s-

10°

I
O

CO

7 10"

CO

£ 10"

10"" —

10

-10

— ^ — , — I — , — , — , — , — f
z = [0.02,0.10], N

532

^ 40 ^

-25

-20 -15

MR-51ogh

-25

-20 -15

MR-51ogh

10

10

Fig. 10. — SHELSo.i luminosity functions for all galaxies (solid squares and solid line), for
HSB galaxies with SBso,/? < 21.82 mag arcsec"^ (x's and dashed line), and for LSB galaxies
with SBso^i? > 21.82 mag arcsec^^ (open squares and dotted line). The symbols show the
SWML values; the lines are the best fit Schechter function. The histograms show the number
of galaxies in each 0.5 magnitude bin for each luminosity function.

-35-

-2.5

-2.0

1.5
a

1.0

-0.5

Fig. 11. — Confidence contours for the SHELSo.i luminosity function parameters for all
galaxies (solid squares and solid line), for HSB galaxies with SB5o,_r < 21.82 mag arcsec"^ (x
and dashed line), and for LSB galaxies with SBsq^r > 21.82 mag arcsec"^ (open square and
dotted line). The symbols show the best fit Schechter parameters. The contours indicate
the 1,2, and Sa limit for the parameters.

-36-

10°

CO

o

laa*f^

^""^

Gh

10"^

^..-.-"gSBT

^

m

—

/^ ,ff"

—

1

10-^

iafl

CO

s

SWML

10"^

/

■ —

all

g

10"^

/

□ ....

- SB50R>22.5

t:>-

/

in-i°

1 1 1 1 / 1

1 1 1 ,' 1 1 1 1 1

1 1 1 1 1 1

-25

-20 -15

MR-51ogh

-25

-20 -15

MR-51ogh

10

-10

Fig. 12. — SHELSo.i luminosity functions for the entire SHELSq.i sample (solid squares)
compared with the SBso,/? > 22.5 mag arcsec"^ subset (open squares), an unusually low sur-
face brightness sample. The histograms show the number of galaxies in each 0.5 magnitude
bin of the respective luminosity functions.

-37-

K

S

14

16

18

-20

-22

-24

"1 I I I I I I I I I I I I I I I I I I I I I I r

D

SWML
all

n SB5o_R>22.5

J I I I I I I I I I I I I I I I I I I I I I I L

-2.5

-2.0

1.5 -1.0

a

-0.5

0.0

Fig. 13. — Confidence contours for the sample from Figure [I2] with SBso,/? > 22.5 mag
arcsec"^ (dotted contours) and for the entire SHELSq.i sample (solid contours). For the low
surface brightness sample the error ellipses admit a wide range of values of a and do not
exclude a = —2. The best fit is a = —1.52 ± 0.15. These low surface brightness galaxies
dominate the determination of the faint end slope for the luminosity function of the sample
as a whole.