International Scholarly Research Notices

International Scholarly Research Notices / 2011 / Article

Research Article | Open Access

Volume 2011 |Article ID 958973 | https://doi.org/10.5402/2011/958973

Ing贸lfur 脕g煤stsson, Tereasa G. Brainerd, "Locations of Satellite Galaxies in the Two-Degree Field Galaxy Redshift Survey", International Scholarly Research Notices, vol. 2011, Article ID 958973, 11 pages, 2011. https://doi.org/10.5402/2011/958973

Locations of Satellite Galaxies in the Two-Degree Field Galaxy Redshift Survey

Academic Editor: W. W. Zeilinger
Received24 Jun 2011
Accepted08 Aug 2011
Published24 Oct 2011

Abstract

We compute the locations of satellite galaxies in the Two-Degree Field Galaxy Redshift Survey using two sets of selection criteria and three sources of photometric data. Using the SuperCOSMOS 饾憻饾惞 photometry, we find that the satellites are located preferentially near the major axes of their hosts, and the anisotropy is detected at a highly significant level (confidence levels of 99.6% to 99.9%). The locations of satellites that have high velocities relative to their hosts are statistically indistinguishable from the locations of satellites that have low velocities relative to their hosts. Additionally, satellites with passive star formation are distributed anisotropically about their hosts (99% confidence level), while the locations of star-forming satellites are consistent with an isotropic distribution. These two distributions are, however, statistically indistinguishable. Therefore it is not correct to interpret this as evidence that the locations of the star-forming satellites are intrinsically different from those of the passive satellites.

1. Introduction

The existence of massive halos of dark matter around large, bright galaxies is well accepted. However, at present there are relatively few direct observational constraints on the sizes and shapes of these dark matter halos. The most popular theory for structure formation in the universe, known as Cold Dark Matter (CDM), predicts that the dark matter halos extend to radii that are at least an order of magnitude greater than the radii of the visible galaxies (see, e.g., [1] and references therein). In addition, CDM predicts that the dark matter halos of galaxies are not spherical; instead they are triaxial in shape (e.g., [26]).

In principle, the locations of small, faint satellite galaxies, measured with respect to the major axes of the large, bright, 鈥渉ost鈥 galaxies that they orbit, have the potential to provide strong constraints on the dark matter halos that surround the hosts, as well as on the relationships of the luminous hosts to their dark matter halos. Recent studies of satellite galaxies from modern redshift surveys have shown that, when their locations are averaged over the entire population, the satellites of relatively isolated host galaxies have a preference for being located near the major axes of their hosts (e.g., [711]). The observed locations of the satellite galaxies in the Sloan Digital Sky Survey (SDSS; [12]) are also known to depend upon various physical properties of the hosts and satellites (e.g., [9, 11, 13]). The satellites of the SDSS host galaxies that have the reddest colors, highest stellar masses, and lowest specific star formation rates (SSFRs) show a pronounced tendency for being located near the major axes of their hosts. (Note: the SSFR has units of yr1 and is defined to be the ratio of the star formation rate in the galaxy to its stellar mass; see [11].) On the other hand, the satellites of the SDSS host galaxies that have the bluest colors, lowest stellar masses, and highest SSFR are distributed isotropically around their hosts. The SDSS satellite galaxies that have the reddest colors, highest stellar masses, and lowest SSFR also show a strong preference for being located near the major axes of their hosts, while the SDSS satellite galaxies that have the bluest colors, lowest stellar masses, and highest SSFR show little to no anisotropy in their locations. The alignment of the satellites of relatively isolated SDSS host galaxies is also known to be similar to the alignment of satellites with the central galaxies of relatively isolated SDSS galaxy groups, where the strongest alignment is found for red central galaxies and their red satellites, while no significant satellite alignment is detected for groups that have blue central galaxies (e.g., [14]).

From a theoretical standpoint, one would expect that if the dark matter halos of large, bright galaxies consist of CDM, then the locations of the satellite galaxies should reflect the deviations of the halo potentials from pure spherical symmetry. Simulations of structure formation in 螞CDM universes have shown that, in projection on the sky, the locations of the satellite galaxies trace the shapes of their hosts鈥 dark matter halos rather well (e.g., [15]). However (and crucially), from an observational standpoint, the expected nonspherical distribution of satellite galaxies will only manifest in an observational data set if mass and light are reasonably well aligned within the hosts. In other words, the satellites should trace the dark mass associated with their hosts, but not necessarily the luminous mass associated with their hosts (i.e., since the dark mass exceeds the luminous mass by ~2 orders of magnitude).

If the halos of the hosts are triaxial, and if one could simply use the symmetry axes of the hosts鈥 dark matter halos (as projected on the sky) to define the geometry of the problem, one would naturally expect to observe an anisotropy in the locations of satellite galaxies such that the satellites are found preferentially close to the major axes of their hosts鈥 dark matter halos. If there is a substantial misalignment between the projected major axes of the luminous host galaxies and their dark matter halos, however, one would expect to observe little to no anisotropy in the locations of the satellites. Using simple prescriptions for embedding luminous host galaxies within their dark matter halos [11] showed that the observed dependences of SDSS satellite locations on various host properties can be easily reproduced if mass and light are aligned in the elliptical hosts (i.e., luminous ellipticals are effectively miniature versions of their dark matter halos), while the disk hosts are instead oriented such that their angular momentum vectors are aligned with the net angular momentum vectors of their halos. The angular momentum alignment for the disk hosts and their halos introduces a significant misalignment of mass and light (e.g., [16]), resulting in the satellites of disk hosts being distributed much more isotropically than the satellites of elliptical hosts.

One of the difficulties with observational samples of host galaxies and their satellites is the presence of 鈥渋nterlopers鈥 (i.e., 鈥渇alse鈥 satellites) in the data. Since the distances to the galaxies are generally unknown, hosts and satellites are selected from redshift surveys via a set of redshift space proximity criteria. Typically, satellite galaxies must be located within a projected distance 饾憻饾憹500鈥塳pc of their host, and the line of sight velocity difference between a host and its satellite must be |饾浛饾懀|500鈥塳m鈥塻鈭1. From simulations in which hosts and satellites were selected using criteria that are identical to the redshift space criteria used for observational data sets, it is known that the majority of objects that are selected as satellite galaxies are, in fact, located physically nearby a host galaxy. However, a substantial number of objects that are selected as satellites are located physically far away from a host galaxy and are, therefore, interlopers (i.e., not genuine satellites; see [11]). When investigating the properties of the satellite population, the interlopers are a source of noise and ideally one would eliminate them from the sample if at all possible. This can be done in a simulation since the 3-dimensional locations of all of the objects are known, but it is not obvious how or if this can be accomplished in an observational data set.

So far, the only direct attempt to eliminate interlopers from an observational study of the locations of satellite galaxies is the work of [7]. In their study, [7] computed the locations of the satellites of relatively isolated host galaxies in the Two-Degree Field Galaxy Redshift Survey (2dFGRS; [17, 18]). In order to address the detrimental effects of interloper contamination, [7] argued that if they divided their host-satellite sample by relative line of sight velocity, |饾浛饾懀|, the set of host-satellite pairs that had the largest observed values of |饾浛饾懀| should suffer greater interloper contamination than the set of host-satellite pairs that had the smallest observed values of |饾浛饾懀|. That is, [7] anticipated that the peak of the observed relative velocity distribution, 饾憙(饾浛饾懀), would be dominated by genuine satellites, while the tails of the distribution would be dominated by interlopers (for which the observed values of |饾浛饾懀| would be largely attributable to the Hubble flow). Therefore, [7] divided their sample of hosts and satellites into a 鈥渓ow relative velocity鈥 sample (|饾浛饾懀|<160km鈥塻鈭1) and a 鈥渉igh relative velocity鈥 sample (|饾浛饾懀|>160 km鈥塻鈭1), expecting that the low relative velocity sample would suffer much less interloper contamination in comparison to the high relative velocity sample. Within both the full sample and the high relative velocity sample, [7] found no evidence for any anisotropy in the locations of the satellite galaxies. However, in the sample with |饾浛饾懀|<160km鈥塻鈭1, [7] reported a preference for the satellites to be located near the major axes of their hosts (see [10], the erratum to [7]). Within this low relative velocity sample, [7] found that the ratio of 鈥減lanar鈥 (饾湙<30) to 鈥減olar鈥 (饾湙>60) satellite locations was 饾憮=饾憗<30/饾憗>60=1.250.06 and that the distribution of satellite locations was fitted well by a double cosine function with amplitude 饾惔=0.120.04.

In their analysis, the authors in [7] did not directly determine whether the results for the satellite locations in the low velocity sample were statistically distinct from the results for the satellite locations in the high velocity sample. That is, given the small number statistics with which the authors in [7] were working, it is entirely possible that the distribution of satellite locations in their high velocity sample was consistent with being drawn from the same parent population as the distribution of satellite locations in the low velocity sample. Therefore, it is not clear that their result should be interpreted as evidence that the satellites in the high velocity sample are distributed isotropically about their hosts, while the satellites in the low velocity sample are distributed anisotropically about their hosts. Rather, all that can be concluded about the high relative velocity sample in [7] is that the null hypothesis of a uniform distribution could not be ruled out. The fact that the null hypothesis could not be ruled out may be due to the locations of the satellites in the high velocity sample having an intrinsically isotropic distribution. On the other hand, it could also be due to the size of the sample being too small to detect an intrinsic anisotropy in the presence of a significant amount of noise (i.e., this is a pair-counting problem that is dominated by 饾憗 statistics).

At the time the authors of [7] were doing their work, little was known about the distribution of the interlopers relative to the host galaxies and, for the most part, interlopers were simply assumed to be a population of objects that were selected at random (see, e.g., [1921]). However, careful analysis of the interloper population from simulations has shown that the interlopers are far from being a random population. Instead, along the line of sight, most interlopers are located within a distance of 2Mpc of a host (i.e., a distance far less than the ~7鈥塎pc one would expect from the Hubble flow, given a maximum host-interloper velocity difference of |饾浛饾懀|=500鈥塳m鈥塻鈭1; see [22]). In addition, the probability distribution of relative velocities, 饾憙(饾浛饾懀), for the hosts and interlopers reaches a maximum at 饾浛饾懀=0 (e.g., [22, 23]). The distribution of relative velocities for host-interloper pairs is, in fact, quite similar to the distribution of relative velocities for pairs of hosts and their genuine satellites. Therefore, interlopers are almost as likely to have low velocities relative to the host galaxies as are the genuine satellites. In retrospect, then, it is not clear that the original velocity cut that [7] imposed in their analysis is well motivated, nor that there should be any significant difference in the locations of satellites with low velocities relative to their hosts and the locations of satellites with high velocities relative to their hosts.

Here we revisit the question of the locations of satellite galaxies in the 2dFGRS. We first adopt the selection criteria of [7] to obtain a host-satellite sample, and we compute the satellite locations using three different sets of photometry for the galaxies. We next adopt the selection criteria that we used in a previous study of the locations of satellite galaxies in the SDSS (e.g., [11]), and we focus our analysis on the hosts and satellites found using the SuperCOSMOS scans of the 饾憻饾惞 plates. In all cases we determine whether the satellite locations in a low relative velocity sample of the data can be distinguished from the satellite locations in a high relative velocity sample. Finally, using the sample obtained with the SDSS selection criteria, we investigate the effect of star formation rate on the observed locations of the 2dFGRS satellites. Throughout we adopt cosmological parameters 饾憵0=0.25, 0=0.75, and 饾惢0=73km鈥塻鈭1Mpc1.

2. Two-Degree Field Galaxy Redshift Survey

The Two-Degree Field Galaxy Redshift Survey (http://msowww.anu.edu.au/2dFGRS/) is a publicly-available redshift survey that covers ~5% of the sky. The target objects in the survey were selected in the 饾憦饾惤 band from the Automated Plate Measuring (APM) galaxy survey and extensions to the survey (see [24, 25]). The photometry of the APM galaxy survey was based on scans of the UK Schmidt Telescope photographic survey plates obtained in blue (饾憦饾惤) and red (饾憻饾惞) spectral bands. Although the APM did not complete the scans of the 饾憻饾惞 plates, the SuperCOSMOS measuring machine was ultimately used to make independent scans of both the 饾憦饾惤 and 饾憻饾惞 plates (see [26, 27]); (http://www-wfau.roe.ac.uk/sss/). In addition to providing photometry in two spectral bands, [28] reports that the SuperCOSMOS scans yielded improved linearity and smaller random errors in comparison to the original APM scans. The final data release of the 2dFGRS contains 245鈥591 galaxies, of which 233鈥251 have good quality spectra (饾憚3). Here we use the final 2dFGRS data release, and, specifically, we use the data for the best spectrum of each object (i.e., the ASCII catalog 鈥best.observations.idz,鈥 which contains the 2dFGRS spectral information, as well as the photometric information from the APM scans, for 231,178 sources with 饾憚3 and extinction-corrected magnitudes 饾憦饾惤19.45). Additionally, we use the 2dFGRS database to retrieve the apparent magnitudes, the galaxy shape parameters (semimajor and semiminor axes), and the galaxy position angles for the SuperCOSMOS scans of the 饾憦饾惤 and 饾憻饾惞 plates.

Spectral types for the galaxies are quantified by the parameter 饾渹, which can be interpreted as an indicator of the amount of star formation in the galaxy (e.g., [29]). Rest-frame colors for the 2dFGRS galaxies can be obtained by using the relationship 饾憦饾惤饾憻饾惞0=饾憦饾惤饾憻饾惞饾憦饾惥饾惤饾憻+饾惥饾惞,(1) where 饾惥(饾憦饾惤) and 饾惥(饾憻饾惞) are color-dependent K-corrections from [30].

3. Locations of Satellites: Sample 1

We begin by obtaining hosts and satellites from the 2dFGRS using selection criteria that are identical to the criteria used by [7]. In selecting this host-satellite sample, we use the photometric parameters from the APM scans of the 饾憦饾惤 plates, as did [7]. Here host galaxies must have redshifts 饾懅<0.1, absolute magnitudes 饾惖饾惤<18, and image ellipticities 饾憭=1饾憦/饾憥0.1. In addition, host galaxies must be relatively isolated within their local regions of space. In order for a host to qualify as being relatively isolated, its 饾惖饾惤 magnitude must be at least one magnitude brighter than any other galaxy that is found within a projected radius of 饾憻饾憹<700kpc and line of sight velocity difference |饾浛饾懀|<1000鈥塳m鈥塻鈭1. Satellites are galaxies that have absolute 饾惖饾惤 magnitudes that are at least two magnitudes fainter than their host, are found within projected radii 饾憻饾憹<500kpc, and have line of sight velocity differences |饾浛饾懀|<500km鈥塻鈭1 relative to their hosts. In order to exclude host-satellite systems that are likely to be groups of galaxies, we reject all host-satellite systems that meet the above criteria, but which contain 5 or more satellites (see also [7]). After all of the above restrictions are imposed, our first sample consists of 1鈥725 hosts and 2鈥594 satellites. The size of our sample is slightly larger than that of [7] (which had 1鈥498 hosts), and the difference in sample size is likely attributable to small differences in the implementation of the selection criteria.

We define the location of a satellite, 饾湙, to be the angle between the major axis of its host galaxy and the direction vector on the sky that connects the centroids of the host and satellite. Since we are only interested in determining whether the satellites are found preferentially close to either the major or minor axes of their hosts, we restrict 饾湙 to the range 0饾湙90. Therefore, 鈥減lanar alignment鈥 corresponds to a mean satellite location 饾湙<45 and 鈥減olar alignment鈥 corresponds to a mean satellite location 饾湙>45.

Shown in the top panels of Figure 1 are the differential and cumulative probability distributions for the satellite locations in our first sample ((a) and (b), resp.). Here the centroids of the hosts and satellites, as well as the position angles of the host galaxies, are taken from the APM scans of the 饾憦饾惤 plates. Error bars for 饾憙(饾湙) were computed using 1,000 bootstrap resamplings of the data. Also shown in the top panels of Figure 1 are the mean satellite location, the median satellite location, the confidence level at which the 饾湌2 test rejects a uniform distribution for 饾憙(饾湙), and the confidence level at which the Kolmogorov-Smirnov (KS) test rejects a uniform distribution for 饾憙(饾湙饾湙max). From Figures 1(a) and 1(b), then, the satellite locations in our first sample are consistent with an isotropic distribution. Following [7] we also compute the planar-to-polar ratio, 饾憮=饾憗<30/饾憗>60=1.080.05, which again is consistent with an isotropic distribution.

Next, and again following [7], we divide our first sample into a 鈥渓ow relative velocity鈥 sample (|饾浛饾懀|<160鈥塳m鈥塻鈭1; 1,209 hosts and 1,514 satellites) and a 鈥渉igh relative velocity鈥 sample (|饾浛饾懀|>160鈥塳m鈥塻鈭1; 855 hosts and 1,080 satellites), and we repeat the analysis above. Unlike [7], however, we find no statistically significant indication that the satellites in either velocity sample are distributed anisotropically around their hosts. Further, a two-sample KS test that compares 饾憙(饾湙饾湙max) for the low relative velocity sample to 饾憙(饾湙饾湙max) for the high relative velocity sample finds that the two distributions are statistically identical. That is, the two-sample KS test rejects the null hypothesis that the two distributions are drawn from the same parent distribution at a confidence level of 18%. We summarize our results in lines 1鈥3 of Table 1, where 饾憙KS is the confidence level at which the KS test rejects a uniform distribution for the satellite locations, 饾湙 is the mean satellite location, 饾湙med is the median satellite location, and 饾憮 is the planar-to-polar ratio. The error bound on 饾湙 is the standard deviation in the mean, and the error bounds on 饾湙med and 饾憮 are 68% confidence bounds obtained from 2,000 bootstrap resamplings of the data.


Sample Scan | 饾浛 饾懀 | 饾憙 K S 饾湙 (degrees) 饾湙 m e d (degrees) 饾憮

1 APM 饾憦 饾惤 <500鈥塳m s鈭1 89% 4 4 . 3 0 . 5 4 4 . 2 + 0 . 7 1 . 1 1 . 0 8 0 . 0 5
1 APM 饾憦 饾惤 <160鈥塳m s鈭1 91% 4 4 . 0 0 . 7 4 3 . 9 + 0 . 9 1 . 6 1 . 1 1 0 . 0 7
1 APM 饾憦 饾惤 >160鈥塳m s鈭1 9% 4 4 . 7 0 . 8 4 4 . 7 + 1 . 2 1 . 8 1 . 0 4 0 . 0 8
1 SC 饾憦 饾惤 <500鈥塳m鈥 s 1 86% 4 4 . 3 0 . 5 4 4 . 0 + 1 . 0 0 . 9 1 . 0 8 0 . 0 5
1 SC 饾憦 饾惤 <160鈥塳m鈥 s 1 90% 4 3 . 8 0 . 7 4 3 . 3 + 0 . 9 1 . 4 1 . 1 3 0 . 0 7
1 SC 饾憦 饾惤 >160鈥塳m鈥 s 1 6% 4 4 . 9 0 . 8 4 5 . 3 + 2 . 0 1 . 3 1 . 0 0 0 . 0 8
1 SC 饾憻 饾惞 <500鈥塳m鈥 s 1 90% 4 4 . 4 0 . 5 4 4 . 1 + 0 . 9 1 . 0 1 . 1 0 0 . 0 5
1 SC 饾憻 饾惞 <160鈥塳m鈥 s 1 83% 4 4 . 4 0 . 7 4 3 . 8 + 1 . 3 1 . 0 1 . 1 0 0 . 0 7
1 SC 饾憻 饾惞 >160鈥塳m鈥 s 1 31% 4 4 . 5 0 . 8 4 4 . 3 + 1 . 6 1 . 3 1 . 0 8 0 . 0 8
2 SC 饾憻 饾惞 <500鈥塳m鈥 s 1 99.6 % 4 3 . 9 0 . 4 4 3 . 2 + 0 . 7 0 . 5 1 . 1 0 0 . 0 4
2 SC 饾憻 饾惞 <160鈥塳m鈥 s 1 99.2% 4 3 . 7 0 . 5 4 2 . 7 + 0 . 9 1 . 0 1 . 1 2 0 . 0 5
2 SC 饾憻 饾惞 >160鈥塳m鈥 s 1 87% 4 4 . 3 0 . 6 4 3 . 6 + 1 . 0 0 . 7 1 . 0 7 0 . 0 6

Lastly, we repeat our analysis using the same hosts and satellites as above, but we now obtain the host galaxy position angles from the SuperCOSMOS scans of the 饾憦饾惤 and 饾憻饾惞 plates. Using the host position angles from the SuperCOSMOS scans has no affect on our conclusions above; in all cases the locations of the satellites are consistent with an isotropic distribution. We summarize our results for the locations of the satellites from the SuperCOSMOS scans in Table 1, lines 4鈥6 (饾憦饾惤) and lines 7鈥9 (饾憻饾惞). A two-sample KS test that compares 饾憙(饾湙饾湙max) for the low relative velocity sample to 饾憙(饾湙饾湙max) for the high relative velocity sample finds that, for the 饾憦饾惤 SuperCOSMOS scans, the two distributions are statistically indistinguishable (KS rejection confidence level of 63%). Similarly, a two-sample KS test that compares 饾憙(饾湙饾湙max) for the low relative velocity sample to 饾憙(饾湙饾湙max) for the high relative velocity sample finds that, for the 饾憻饾惞 SuperCOSMOS scans, the two distributions are also statistically indistinguishable (KS rejection confidence level of 7%).

Therefore, at least in case of the selection criteria adopted by [7], our analysis finds that there is no statistically significant evidence that the 2dFGRS satellites are distributed anisotropically around their hosts. Further, we find that there is no statistically significant evidence that dividing the host-satellite sample by relative velocity (i.e., low versus high) results in different conclusions about the locations of the satellites.

4. Locations of Satellites: Sample 2

In order to compare most directly with our previous work using SDSS galaxies, we next obtain a host-satellite sample from the 2dFGRS using the selection criteria from [11]. Since the SDSS results are based upon 饾憻-band imaging and also because the shapes of galaxies are generally smoother at longer wavelengths than they are at shorter wavelengths (i.e., the position angles of the host galaxies may be more accurate when measured at longer wavelengths), here we restrict our analysis to the SuperCOSMOS scans of the 饾憻饾惞 plates. The selection criteria that we adopt are similar to the selection criteria that we used to obtain our first sample, but here they are somewhat more relaxed. Host galaxies must have 饾憻饾惞 magnitudes that are at least one magnitude brighter than any other galaxy that is found within a projected radius 饾憻饾憹700鈥塳pc and a line of sight velocity difference |饾浛饾懀|1,000鈥塳m鈥塻鈭1. Satellite galaxies are objects that, relative to their hosts, are found within projected radii 饾憻饾憹500鈥塳pc, have line of sight velocity differences |饾浛饾懀|500km鈥塻鈭1, and have 饾憻饾惞 magnitudes that are at least two magnitudes fainter than their host. In addition, the luminosity of each host must exceed the sum total of the luminosities of its satellites, each host may have at most 9 satellites, and hosts are restricted to the redshift range 0.01饾懅0.15. We place no restrictions either on the ellipticities of the hosts' images or on their absolute magnitudes. However, we do require that the hosts and satellites have good quality spectra (饾憚3) and that the hosts have well-defined spectral parameters (饾渹99.9). The latter constraint helps to insure that the host galaxies have fairly regular shapes. This results in 2,947 host galaxies and 4,730 satellites in our second sample (i.e., 80% larger than our first sample above).

We assign rest-frame colors to the 2dFGRS hosts and satellites using (1) above. Following [30] we define red galaxies to be those with rest-frame colors (饾憦饾惤饾憻饾惞)01.07 and blue galaxies to be those with rest-frame colors (饾憦饾惤饾憻饾惞)0<1.07. Following [29] we use the value of 饾渹 as a measure of the star formation rate within a galaxy, from which we define galaxies with 饾渹>1.4 to be 鈥渟tar-forming鈥 and galaxies with 饾渹1.4 to be 鈥減assive.鈥 Although rest-frame color and star formation rate are strongly correlated (i.e., red galaxies tend to have low star formation rates, while blue galaxies tend to have high star formation rates) these two parameters are not identical. Within our sample, 11% of the 鈥減assive鈥 hosts have blue rest-frame colors and 28% of the 鈥渟tar-forming鈥 hosts have red rest-frame colors. Of the 4,332 satellites that have well-defined spectral parameters, 13% of the 鈥減assive鈥 satellites have blue rest-frame colors and 10% of the 鈥渟tar-forming鈥 satellites have red rest-frame colors.

Figure 2 summarizes the basic statistical properties of the SuperCOSMOS 饾憻饾惞 host-satellite sample obtained using the selection criteria of [11]. The different panels of Figure 2 show probability distributions for (a) the number of satellites per host, (b) the redshifts of the hosts, (c) the 饾憻饾惞 apparent magnitudes of the hosts and satellites, (d) the 饾憻饾惞 absolute magnitudes of the hosts and satellites, (e) the rest-frame colors of the hosts and satellites, and (f) the spectral types of the hosts and satellites. From Figure 2, then, our host sample is dominated by red, passive galaxies while our satellite sample is dominated by blue, star-forming galaxies. This is in good agreement with our previous results for the SDSS (e.g., [11]). For comparison, Figure 3 shows the basic statistical properties for the hosts and satellites of Sample 1 (see Section 3). Aside from differences that are due to different imposed cutoffs (i.e., maximum number of satellites and host galaxy redshift range), the statistical properties of the hosts and satellites are very similar for our two samples.

Probability distributions for the locations of all of the satellites in our second sample are shown in the bottom panels of Figure 1. The differential probability distribution, 饾憙(饾湙), is shown in Figure 1(c), along with the mean satellite location and the confidence level at which the 饾湌2 test rejects a uniform distribution for the satellites. The cumulative probability distribution, 饾憙(饾湙饾湙max), is shown in Figure 1(d), along with the median satellite location and the confidence level at which the KS test rejects a uniform distribution for the satellites. From Figures 1(c) and 1(d), then, the 2dFGRS satellites in our second sample are distributed anisotropically about their hosts, and the sense of the anisotropy is the same as the anisotropy of the SDSS satellites: when averaged over the entire sample, the satellites are located preferentially close to the major axes of their hosts. The significance of our detection of the anisotropy is, however, less for the locations of the 2dFGRS satellites (饾湌2 and KS rejection confidence levels of 99.7% and 99.6%, resp.) than it was for the locations of the SDSS satellites in our previous study (饾湌2 and KS rejection confidence levels >99.99%; see [11]). This is likely due to a combination of effects. First, the host-satellite sample in [11] is ~50% larger than the one we have used here (4,487 SDSS hosts and 7,399 SDSS satellites), which simply results in better statistics. Second, although the SDSS and 2dFGRS hosts have very similar redshift distributions, the images of the SDSS galaxies are somewhat better resolved than the images of the 2dFGRS galaxies (pixel size of 0.40 arcsec in the SDSS versus pixel size of 0.67 arcsec for the SuperCOSMOS scans). This could lead to more accurate centroids for the SDSS galaxies, as well as more accurate position angles for the SDSS hosts. In addition, the rms velocity error in the 2dFGRS is ~85鈥塳m鈥塻鈭1 (e.g., [17]), which is significantly greater than the ~30鈥塳m鈥塻鈭1 rms velocity error in the SDSS (e.g., [31]). As a result, it would not be surprising if the 2dFGRS sample contains a larger fraction of interlopers than does the SDSS sample. The effect of interlopers is to reduce the observed anisotropy in the satellite locations (e.g., [11]). Hence, due to the smaller size of the 2dFGRS sample, the greater interloper contamination of the 2dFGRS sample, and the more accurate photometric parameters of the SDSS, we would naturally expect to find somewhat less anisotropy in the locations of the 2dFGRS satellites than in the locations of the SDSS satellites.

Next, using our second 2dFGRS host-satellite sample we again investigate whether dividing the sample into host-satellite pairs with low relative velocities (|饾浛饾懀|<160鈥塳m鈥塻鈭1; 1,988 hosts and 2,633 satellites) and high relative velocities (|饾浛饾懀|>160鈥塳m鈥塻鈭1; 1,512 hosts and 2,097 satellites) affects our ability to detect the anisotropy in the satellite locations. We summarize our results in Table 1, lines 10鈥12, from which it is clear that the anisotropy in the satellite locations is detected for the host-satellite pairs with low relative velocities (although, due to the smaller number of satellites, the significance is lower than it is for the full sample). In the case of the host-satellite pairs with high relative velocities, the satellite locations are consistent with an isotropic distribution (KS rejection confidence level of 87%). However, it is important to note that this alone does not constitute proof that the locations of the satellites in the high relative velocity sample are intrinsically different from the locations of the satellites in the low relative velocity sample (e.g., as might be expected if the high relative velocity sample contained a much larger fraction of interlopers than the low relative velocity sample).

In order to determine whether the satellite locations in the high relative velocity sample are truly different from those in the low relative velocity sample, we again compute a two-sample KS test. When we compare 饾憙(饾湙饾湙max) for the high relative velocity sample to 饾憙(饾湙饾湙max) for the low relative velocity sample, we find that the two distributions are statistically indistinguishable; the two-sample KS test rejects the null hypothesis that the two distributions are drawn from the same distribution at a confidence level of 54%. Therefore, it is not correct to conclude that dividing our second sample by relative velocity yields one set of satellites that are distributed anisotropically about their hosts (i.e., the low relative velocity sample) and another set of satellites that are distributed isotropically about their hosts (i.e., the high relative velocity sample).

At least for the rather small host-satellite sample that can be obtained from the 2dFGRS, it does not appear that dividing the sample by host-satellite relative velocity yields a substantial reduction in the effects of interloper contamination on the observed locations of the satellite galaxies. In other words, since 饾憙(饾湙饾湙max) for the high relative velocity sample is consistent with being drawn from the same distribution as 饾憙(饾湙饾湙max) for the low relative velocity sample, there is no statistically significant evidence that the satellites in the high relative velocity sample are distributed much more uniformly around their hosts than are the satellites in the low relative velocity sample. Both theoretically (e.g., [11, 22, 23]) and observationally, then, dividing the sample by host-satellite relative velocity does not obviously provide a significant reduction of the effects of interlopers on the observed locations of satellite galaxies.

If, however, we consider the star formation rates of the satellites, it could in principle be possible to identify a sample of satellites that contains both the smallest level of interloper contamination and the greatest degree of intrinsic anisotropy in the locations of the genuine satellites. From the theoretical work by [11], we know that the selection criteria that we have adopted here yield host galaxies that reside at the dynamical centers of large dark matter halos. The satellites are noncentral galaxies (i.e., 鈥渟ubstructure鈥) that orbit within their hosts鈥 large dark matter halos. Prior to being accreted into the dark matter halo of its host galaxy, a satellite galaxy would have grown and evolved within its own dark matter halo. After accretion, the satellite would have ceased growing in mass and may have even lost mass (e.g., by tidal stripping when passing near the center of its hosts' halo or by interactions with other subhalos). Star formation within the satellite would have been severely quenched after accretion by the host galaxy because the satellite loses most of its cold gas reservoir to the warmer, larger halo of its host. The higher the redshift at which a satellite was accreted, then, the lower should be its star formation rate at the present day, and the more likely its orbit will reflect the (nonspherical) gravitational potential of its host's dark matter halo.

From [22], we know that by the present day (i.e., 饾懅=0), only ~40% of the genuine satellite galaxies in the Millennium Run Simulation (i.e., [1]) that have blue SDSS colors, (饾憯饾憻)0<0.7, have completed at least one orbit of their host galaxy. In contrast, 86% of the genuine satellite galaxies with red SDSS colors, (饾憯饾憻)00.7, have completed one or more orbits of their host galaxy by the present day. In addition, [11] found that when our selection criteria above were applied to the Millennium Run Simulation, only 42% of the objects with blue SDSS colors that were selected as satellites were, in fact, genuine satellites. However, [11] also found that 81% of the objects with red SDSS colors that were selected as satellites were actually genuine satellites. All in all, then, we would expect that an observational sample of satellite galaxies with low star formation rates and red SDSS colors should suffer the least amount of interloper contamination, while also exhibiting the greatest amount of intrinsic anisotropy in their locations relative to their hosts (i.e., since they are relatively 鈥渙ld鈥 satellites that have been within their hosts' halos for a considerable length of time).

Due to the very small overlap of the 2dFGRS and the SDSS, SDSS colors are not available for more than a few objects in our sample. However, using the parameter 饾渹 we can investigate the effects of star formation rate on the observed locations of the 2dFGRS satellites. If we classify the satellites with 饾渹>1.4 as 鈥渟tar forming鈥 (3201 satellites) and 饾渹1.4 as 鈥減assive鈥 (1131 satellites), we then find that the cumulative probability distribution for the locations of the passive satellites is inconsistent with an isotropic distribution (KS rejection confidence level of 99%), while the cumulative probability distribution for the locations of the star-forming satellites is consistent with an isotropic distribution (KS rejection confidence level of 89.4%). In the case of the passive satellites, 饾湙=43.00.8 and 饾湙med=42.3+1.21.1, while for the star-forming satellites 饾湙=44.20.5 and 饾湙med=43.50.7. As with the above results for the locations of satellites with high and low velocities relative to their hosts, however, this result should not be interpreted as evidence that the star-forming satellites are distributed isotropically about their hosts, while the passive satellites are distributed anisotropically about their hosts. Rather, a two-sample KS test finds that 饾憙(饾湙饾湙max) for the star-forming satellites is statistically indistinguishable from 饾憙(饾湙饾湙max) for the passive satellites (KS rejection confidence level 88.8%).

Finally, it is worth noting that, unlike our first sample, in our second sample we find a statistically significant detection of anisotropic satellite locations when we use the locations of all of the satellites in the analysis; that is, in our first sample, the satellite locations are consistent with an isotropic distribution. Given that our second sample is almost twice as large as our first sample, it is tempting to attribute the difference in the results from the two samples solely to improved statistics. However, the increase in the sample size does not appear to be the primary cause of the increased signal-to-noise. Instead, the selection of the hosts and satellites specifically using the SuperCOSMOS photometry seems to be the source of the improved signal-to-noise in our second sample.

If we simply restrict the analysis of our second sample to only those hosts that have 饾懅0.1, K-corrected SuperCOSMOS absolute magnitudes 饾惖饾惤<18, ellipticities 饾湒>0.1 as measured from the 饾憻饾惞 SuperCOSMOS photometry, and fewer than 5 satellites (i.e., to effectively match the selection criteria used to obtain our first sample), our second sample is substantially reduced in size: 2,089 hosts and 3,056 satellites. This restricted version of our second sample is only ~20% larger than our first sample, which was selected using the APM scans of the 饾憦饾惤 plates. This smaller, restricted 饾憻饾惞 sample is substantially different from our 饾憦饾惤 sample in Section 3 in that it includes only 1272 of the 1725 hosts in the 饾憦饾惤 sample and only 1835 of the 2594 satellites in the 饾憦饾惤 sample. Therefore, ~40% of the hosts and satellites in the restricted 饾憻饾惞 sample are not present in the 饾憦饾惤-selected sample from Section 3, and ~25% of the hosts and satellites in the 饾憦饾惤-selected sample are not present in the restricted 饾憻饾惞 sample.

When averaged over all satellites in the restricted version of our second sample, the locations of the satellites are still inconsistent with an isotropic distribution (KS rejection confidence level of 99.9% and 饾湌2 rejection confidence level of 99.6%). Therefore, using the SuperCOSMOS scans of the 饾憻饾惞 plates allows a detection of the anisotropic distribution of satellite galaxies that was not possible with the original APM scans of the 饾憦饾惤 plates.

5. Summary

We have computed the locations of satellite galaxies in the 2dFGRS using two sets of selection criteria, and we have investigated whether dividing the sample by host-satellite relative velocity provides a significant reduction of the effects of interlopers on the observed locations of the satellites. When we adopt the selection criteria used by the authors of [7] in their study of the locations of 2dFGRS satellites, we find no statistically significant evidence that the satellites are distributed anisotropically about their hosts. This result is independent of the photometric catalogs that we use (APM scans of the 饾憦饾惤 plates and SuperCOSMOS scans of the 饾憦饾惤 and 饾憻饾惞 plates), as well as the velocities of the satellites relative to their hosts. Our result is in contrast to the original study of [7], which found that the 2dFGRS satellites in the low relative velocity sample are distributed anisotropically around their hosts. The cause of this discrepancy is not clear, but it may lie in the fact that our samples are not truly identical or perhaps in differences in the way that the satellite locations were calculated in our independent analyzes.

We obtain a second host-satellite sample by applying a set of selection criteria that are based upon the criteria we used in a previous study of the locations of satellite galaxies in the SDSS. Further, our second sample is obtained using the 饾憻饾惞 SuperCOSMOS photometry instead of the 饾憦饾惤 APM photometry. Using our second sample, we find that the satellites are anisotropically distributed about their hosts at a statistically significant level (KS rejection confidence level of 99.6%). The sense of the anisotropy is in agreement with previous studies; when averaged over the entire population, the satellites have a preference to be found near the major axes of their host galaxies. When we divide our second sample into host-satellite pairs with low relative velocities (|饾浛饾懀|<160鈥塳m鈥塻鈭1) and host-satellite pairs with high relative velocities (|饾浛饾懀|<160鈥塳m鈥塻鈭1), we find that the satellites with low relative velocities are anisotropically distributed about their hosts at a statistically significant level, while an isotropic distribution cannot be ruled out for the locations of the satellites with high relative velocities. However, this result should not be interpreted as evidence that the distribution of the satellites in the low relative velocity sample is intrinsically different from that of the satellites in the high relative velocity sample. When we compare the distributions of the satellites in the low and high relative velocity samples, we find that they are statistically indistinguishable. As a result, it is not clear that dividing the sample by host-satellite relative velocity is a direct means of eliminating the effects of interlopers on the observed locations of satellite galaxies.

Although the selection criteria that we use to obtain our second sample result in a sample that is nearly twice as large as our first sample, the increase in the sample size is not the primary reason that the anisotropy in the satellite locations can be detected in the second sample, but not the first. Instead, it is the improved photometry from the SuperCOSMOS scans of the 饾憻饾惞 plates that leads to the increased signal-to-noise. If we restrict the analysis of our second sample to a set of host-satellite systems whose properties match those of our first sample, the second sample is only ~20% larger than the first sample, yet the anisotropy of the satellite locations is detected at a highly significant level (KS rejection confidence level of 99.9%).

Finally, in an attempt to isolate a population of satellites that are likely to have the least interloper contamination, as well as the greatest degree of anisotropy in the locations of the genuine satellites, we investigated the effects of star formation rate on the locations of the 2dFGRS satellites. In our second sample, we find that passive satellites (which constitute only 26% of the satellites with well-defined spectral parameters) are distributed anisotropically around their hosts with high statistical significance (KS rejection confidence level of 99%). An isotropic distribution cannot be ruled out for the locations of the star-forming satellites. However, as with our result for dividing the sample by relative velocity, this should not be interpreted as evidence that the locations of the star-forming satellites around their hosts are intrinsically different from the locations of the passive satellites. Rather, we find that the two distributions of satellite locations are statistically indistinguishable in our 2dFGRS sample. Although the star formation rates are quantified differently in the SDSS than they are in the 2dFGRS (i.e., star formation is quantified by SSFR, not 饾渹, in the SDSS), this last result is in reasonable agreement with our previous results for satellite galaxies in the SDSS. That is, the SDSS satellites with the lowest SSFR show a pronounced tendency to be located near the major axes of their hosts, and the SDSS satellites with the highest SSFR show little anisotropy in their locations. However, the mean satellite locations, 饾湙, for the SDSS satellites with the highest SSFR and the lowest SSFR agree with each other at the 2饾湈 level (see [11, Figure鈥夆1(c)]). Therefore, it is not clear that dividing the sample by the star formation rates of the satellites is sufficient to largely eliminate the effects of interlopers on the observed locations of satellite galaxies.

Acknowledgments

The authors are pleased to thank all of those who were involved with the 2dFGRS. This work was supported by the National Science Foundation under NSF Contract AST-0708468.

References

  1. V. Springel, S. D. M. White, A. Jenkins et al., 鈥淪imulations of the formation, evolution and clustering of galaxies and quasars,鈥 Nature, vol. 435, no. 7042, pp. 629鈥636, 2005. View at: Publisher Site | Google Scholar
  2. M. S. Warren, P. J. Quinn, J. K. Salmon, and W. H. Zurek, 鈥淒ark halos formed via dissipationless collapse. I. Shapes and alignment of angular momentum,鈥 Astrophysical Journal, vol. 399, no. 2, pp. 405鈥425, 1992. View at: Google Scholar
  3. Y. P. Jing and Y. Suto, 鈥淭riaxial modeling of halo density profiles with high-resolution N-body simulations,鈥 Astrophysical Journal, vol. 574, no. 2, pp. 538鈥553, 2002. View at: Publisher Site | Google Scholar
  4. J. Bailin and M. Steinmetz, 鈥淚nternal and external alignment of the shapes and angular momenta of CDM halos,鈥 Astrophysical Journal, vol. 627, no. 2, pp. 647鈥665, 2005. View at: Publisher Site | Google Scholar
  5. S. F. Kasun and A. E. Evrard, 鈥淪hapes and alignments of galaxy cluster halos,鈥 Astrophysical Journal, vol. 629, no. 2, pp. 781鈥790, 2005. View at: Publisher Site | Google Scholar
  6. B. Allgood, R. A. Flores, J. R. Primack et al., 鈥淭he shape of dark matter haloes: dependence on mass, redshift, radius and formation,鈥 Monthly Notices of the Royal Astronomical Society, vol. 367, no. 4, pp. 1781鈥1796, 2006. View at: Publisher Site | Google Scholar
  7. L. Sales and D. G. Lambas, 鈥淎nisotropy in the distribution of satellites around primary galaxies in the 2dF Galaxy Redshift Survey: the Holmberg effect,鈥 Monthly Notices of the Royal Astronomical Society, vol. 348, no. 4, pp. 1236鈥1240, 2004. View at: Google Scholar
  8. T. G. Brainerd, 鈥淎nisotropic distribution of SDSS satellite galaxies: planar (not polar) alignment,鈥 Astrophysical Journal, vol. 628, no. 2, pp. L101鈥揕104, 2005. View at: Publisher Site | Google Scholar
  9. M. Azzaro, S. G. Patiri, F. Prada, and A. R. Zentner, 鈥淎ngular distribution of satellite galaxies from the Sloan Digital Sky Survey Data Release 4,鈥 Monthly Notices of the Royal Astronomical Society, vol. 376, no. 1, pp. L43鈥揕47, 2007. View at: Publisher Site | Google Scholar
  10. L. Sales and D. G. Lambas, 鈥淓rratum: anisotropy in the distribution of satellites around primary galaxies in the 2dF Galaxy Redshift Survey: the Holmberg effect,鈥 Monthly Notices of the Royal Astronomical Society, vol. 395, no. 2, p. 1184, 2009. View at: Publisher Site | Google Scholar
  11. I 脕g煤stsson and T. G. Brainerd, 鈥淎nisotropic locations of satellite galaxies: clues to the orientations of galaxies within their dark matter halos,鈥 Astrophysical Journal, vol. 709, no. 2, pp. 1321鈥1336, 2010. View at: Google Scholar
  12. K. N. Abazajian, J. K. Adelman-Mccarthy, M. A. Ag眉eros et al., 鈥淭he seventh data release of the Sloan Digital Sky Survey,鈥 Astrophysical Journal, Supplement Series, vol. 182, no. 2, pp. 543鈥558, 2009. View at: Publisher Site | Google Scholar
  13. R. J. Siverd, B. S. Ryden, and B. S. Gaudi, 鈥淕alaxy orientation and alignment effects in the SDSS DR6,鈥 submitted to The Astrophysical Journal, http://arxiv.org/abs/0903.2264. View at: Google Scholar
  14. Y. Wang, X. Yang, H. J. Mo et al., 鈥淧robing the intrinsic shape and alignment of dark matter haloes using SDSS galaxy groups,鈥 Monthly Notices of the Royal Astronomical Society, vol. 385, no. 3, pp. 1511鈥1522, 2008. View at: Publisher Site | Google Scholar
  15. I 脕g煤stsson and T. G. Brainerd, 鈥淭he locations of satellite galaxies in a CDM universe,鈥 Astrophysical Journal, vol. 650, no. 2, pp. 550鈥559, 2006. View at: Publisher Site | Google Scholar
  16. P. Bett, V. Eke, C. S. Frenk, A. Jenkins, J. Helly, and J. Navarro, 鈥淭he spin and shape of dark matter haloes in the Millennium simulation of a 螞 cold dark matter universe,鈥 Monthly Notices of the Royal Astronomical Society, vol. 376, no. 1, pp. 215鈥232, 2007. View at: Publisher Site | Google Scholar
  17. M. Colless, G. Dalton, S. Maddox et al., 鈥淭he 2dF Galaxy Redshift Survey: spectra and redshifts,鈥 Monthly Notices of the Royal Astronomical Society, vol. 328, no. 4, pp. 1039鈥1063, 2001. View at: Publisher Site | Google Scholar
  18. M. Colless, B. A. Peterson, C. Jackson et al., 鈥淭he 2dF Galaxy Redshift Survey: final data release,鈥 . preprint, http://arxiv.org/abs/astroph/0306581. View at: Google Scholar
  19. T. A. McKay, E. S. Sheldon, D. Johnston et al., 鈥淒ynamical confirmation of Sloan Digital Sky Survey weak-lensing scaling laws,鈥 Astrophysical Journal, vol. 571, no. 2, pp. L85鈥揕88, 2002. View at: Publisher Site | Google Scholar
  20. T. G. Brainerd and M. A. Specian, 鈥淢ass-to-light ratios of 2dF galaxies,鈥 Astrophysical Journal, vol. 593, no. 1, pp. L7鈥揕10, 2003. View at: Publisher Site | Google Scholar
  21. F. Prada, M. Vitvitska, A. Klypin et al., 鈥淥bserving the dark matter density profile of isolated galaxies,鈥 Astrophysical Journal, vol. 598, no. 1, pp. 260鈥271, 2003. View at: Publisher Site | Google Scholar
  22. I 脕g煤stsson, Satellite galaxies as probes of dark matter halos, Ph.D. thesis, Boston University, 2011.
  23. F. C. van den Bosch, P. Norberg, H. J. Mo, and X. Yang, 鈥淧robing dark matter haloes with satellite kinematics,鈥 Monthly Notices of the Royal Astronomical Society, vol. 352, no. 4, pp. 1302鈥1314, 2004. View at: Publisher Site | Google Scholar
  24. S. J. Maddox, G. Efstathiou, W. J. Sutherland, and J. Loveday, 鈥淭he APM galaxy survey. I鈥擜PM measurements and star-galaxy separation,鈥 Monthly Notices of the Royal Astronomical Society, vol. 243, pp. 692鈥712, 1990. View at: Google Scholar
  25. S. J. Maddox, G. Efstathiou, and W. J. Sutherland, 鈥淭he APM Galaxy Survey鈥攑art two鈥攑hotometric corrections,鈥 Monthly Notices of the Royal Astronomical Society, vol. 246, no. 3, pp. 433鈥457, 1990. View at: Google Scholar
  26. N. C. Hambly, H. T. MacGillivray, M. A. Read et al., 鈥淭he SuperCOSMOS Sky Survey鈥擨. Introduction and description,鈥 Monthly Notices of the Royal Astronomical Society, vol. 326, no. 4, pp. 1279鈥1294, 2001. View at: Google Scholar
  27. N. C. Hambly, M. J. Irwin, and H. T. MacGillivray, 鈥淭he SuperCOSMOS Sky Survey鈥擨I. Image detection, parametrization, classification and photometry,鈥 Monthly Notices of the Royal Astronomical Society, vol. 326, no. 4, pp. 1295鈥1314, 2001. View at: Google Scholar
  28. S. Cole, W. J. Percival, J. A. Peacock et al., 鈥淭he 2dF Galaxy Redshift Survey: power-spectrum analysis of the final data set and cosmological implications,鈥 Monthly Notices of the Royal Astronomical Society, vol. 362, no. 2, pp. 505鈥534, 2005. View at: Publisher Site | Google Scholar
  29. D. S. Madgwick, O. Lahav, I. K. Baldry et al., 鈥淭he 2dF Galaxy Redshift Survey: galaxy luminosity functions per spectral type,鈥 Monthly Notices of the Royal Astronomical Society, vol. 333, no. 1, pp. 133鈥144, 2002. View at: Publisher Site | Google Scholar
  30. V. Wild, J. A. Peacock, O. Lahav et al., 鈥淭he 2dF Galaxy Redshift Survey: stochastic relative biasing between galaxy populations,鈥 Monthly Notices of the Royal Astronomical Society, vol. 356, no. 1, pp. 247鈥269, 2005. View at: Publisher Site | Google Scholar
  31. M. A. Strauss, D. H. Weinberg, R. H. Lupton et al., 鈥淪pectroscopic target selection in the Sloan Digital Sky Survey: the main galaxy sample,鈥 Astronomical Journal, vol. 124, no. 3, pp. 1810鈥1824, 2002. View at: Publisher Site | Google Scholar

Copyright 漏 2011 Ing贸lfur 脕g煤stsson and Tereasa G. Brainerd. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.


More related articles

 PDF Download Citation Citation
 Download other formatsMore
 Order printed copiesOrder
Views542
Downloads290
Citations

Related articles

We are committed to sharing findings related to COVID-19 as quickly as possible. We will be providing unlimited waivers of publication charges for accepted research articles as well as case reports and case series related to COVID-19. Review articles are excluded from this waiver policy. Sign up here as a reviewer to help fast-track new submissions.