SDSS-IV MaNGA: Probing the Kinematic Morphology–Density Relation of Early-type Galaxies with MaNGA

MaNGA will ultimately obtain integral-field spectroscopy of 10,000 nearby galaxies with the 2.5 m Sloan Foundation Telescope (Gunn et al. 2006) and the BOSS spectrographs (Smee et al. 2013). Fibers are joined into 17 hexagonal fiber bundles to perform a multi-object IFS survey (Drory et al. 2015). Each fiber has a diameter of 2'', while the bundles range in diameter from 12'' to 32'' with a 56% filling factor. The BOSS spectrographs have a wavelength coverage of 3600–10300 Å and a spectral resolution of ${\sigma }_{r}\approx 70$ km s⁻¹. The relative spectrophotometry is accurate to a few percent (Yan et al. 2016b). The survey design is described in Yan et al. (2016a), the observing strategy in Law et al. (2015), and the data reduction pipeline in Law et al. (2016).

The MaNGA sample is selected from the NASA-Sloan Atlas (NSA; Blanton M., http://www.nsatlas.org) with redshifts mostly from the SDSS Data Release 7 MAIN galaxy sample (Abazajian et al. 2009). The sample is built in i-band absolute magnitude (M_i)-complete shells, with more luminous galaxies observed in more distant M_i shells such that the spatial coverage (in terms of R_e) is roughly constant across the sample (Yan et al. 2016a; Wake et al. 2017). We focus on the combination of the Primary sample ($\sim 50 \%$ of the total sample) and Secondary sample ($\sim 40 \%$ of the total sample), selected such that 80% of the galaxies in each M_i shell can be covered to $1.5{R}_{e}$ ($2.5{R}_{e}$) by the largest MaNGA IFU for the Primary (Secondary) sample, respectively. To compare different stellar mass bins, it is necessary to reweight the galaxy distributions in the sample to account for the different volumes probed by each mass shell. We apply these volume weights whenever population means are presented.

We work with the MaNGA data derived from the data-release pipeline (DRP) v2.0.1 (MaNGA Product Launch (MPL) 5) sample that are also in the Yang et al. (2007, hereafter Y07) group catalog (updated to DR7). Y07 use an iterative, adaptive group finder to assign galaxies to halos. Briefly, they first use a friends-of-friends algorithm to identify potential groups, followed by abundance matching to assign likely halo masses to each group based on the total galaxy luminosity. They iterate their group selection based on the initial estimate for halo mass. We select the central galaxy as the most luminous one, while all other galaxies are designated as satellite galaxies.

We select central and satellite galaxies that reside in halos more massive than ${10}^{12.5}\,{h}^{-1}\,{M}_{\odot }$ (where the group catalog is complete; Yang et al. 2009) and additionally require that the satellite galaxies have stellar masses ${M}_{* }\gt {10}^{10}\,{h}^{-2}\,{M}_{\odot }$. We also visually remove all galaxies with spiral structure, focusing only on early-type (E/S0) galaxies (for details see Paper I). The final sample comprises 379 early-type centrals with a minimum stellar mass of $3\times {10}^{10}$ ${M}_{\odot }$ and a median stellar mass of 10¹¹ ${M}_{\odot }$. There are 159 early-type satellite galaxies with a minimum stellar mass of 10¹⁰ ${M}_{\odot }$ and a median stellar mass of $3\times {10}^{10}$ ${M}_{\odot }$.

The kinematic measurements we use to derive ${\lambda }_{{\rm{R}}}$ come from the MaNGA Data Analysis Pipeline (DAP) and are measured on Voronoi-binned data (Cappellari & Copin 2003) with a signal-to-noise ratio of at least 10 per 70 km s⁻¹ spectral pixel. The kinematics are measured using the penalized pixel-fitting code pPXF (Cappellari & Emsellem 2004), with emission lines masked. The stellar templates are drawn from the MILES library (Sánchez-Blázquez et al. 2006) and are convolved with a Gaussian line-of-sight velocity distribution to derive the velocity and velocity dispersion of the stars. The velocity dispersions are reliable above $\sigma \gt 40$ km s⁻¹ (Penny et al. 2016), while the small number of unresolved spaxels are removed from analysis. The central σ values range from 50 to 400 km s⁻¹, with only 43 galaxies having dispersions $\lt 100$ km s⁻¹ for the sample, so we are not working in this low dispersion regime. An eighth-order additive polynomial is included to account for flux calibration and stellar population mismatch. We adopt the "DONOTUSE" flags from the DAP (meaning that there are known catastrophic problems with these data) and flag all bins with $\sigma \gt 500$ km s⁻¹ or $V\gt 400$ km s⁻¹. We only keep galaxies for which at least 50% of their spaxels are unflagged.

We calculate ${\lambda }_{{\rm{R}}}$ within elliptical isophotes. We adopt the position angle and galaxy flattening $\epsilon =1-b/a$ from the NASA-Sloan Atlas (NSA; Blanton et al. 2011).²³ While it is standard in the literature to compare galaxies at ${\lambda }_{{{\rm{R}}}_{e}}$, we show in Paper I that at the spatial resolution of MaNGA these measurements can be biased by 10%–50%, depending on the input ${\lambda }_{{{\rm{R}}}_{e}}$, with lower values of ${\lambda }_{{{\rm{R}}}_{e}}$ suffering more severely. As described in detail in Paper I, we adopt the outermost measurement of ${\lambda }_{{\rm{R}}}$ (${\lambda }_{\mathrm{out}}$). ${\lambda }_{\mathrm{out}}$ matches λ($1.5{R}_{e}$) with $\sim 20 \%$ scatter and no bias, but allows us to include galaxies with limited radial coverage. Finally, the inner 2'' of data are excluded, since the low spatial resolution of MaNGA tends to lower ${\lambda }_{R}$. In Paper I, we use simulations to show that excluding the central region brings the measured ${\lambda }_{R}$ value closer to the true value, and we estimate that the residual impact of low spatial resolution leads to at most a systematic increase in slow-rotator fraction of 10% from our measured values.

To determine whether a galaxy is a slow or fast rotator requires further comparison with the intrinsic shape of the galaxy, since the amount of rotation needed to support an oblate galaxy rises with ellipticity. Emsellem et al. (2007) report an empirically motivated division between slow and fast rotators of ${\lambda }_{{{\rm{R}}}_{e}}$ $\lt \,0.31\sqrt{\epsilon }$. For measurements at λ($1.5{R}_{e}$), we simply scale by the typical ratio of ${\lambda }_{{{\rm{R}}}_{e}}$/λ($1.5{R}_{e}$) from our data to define slow rotators as those with ${\lambda }_{\mathrm{out}}$ $\,\lt 0.35\sqrt{\epsilon }$.

In Paper I, we examined ${\lambda }_{\mathrm{out}}$ in bins of M_200b, finding no evidence for any residual dependence of ${\lambda }_{\mathrm{out}}$ on halo mass M_200b at fixed M_*. Nor was there any significant evidence for differences between central and satellite galaxies at fixed mass. Therefore, in this Letter, we do not reconsider M_200b or the satellite/central distinction. We do note, however, that there is considerable scatter in the halo masses and the central designations in group catalogs (e.g., Campbell et al. 2015). It is useful, therefore, to consider local overdensity measures as a complement to the group-based measures of environment.

The first papers investigating links between ${\lambda }_{{\rm{R}}}$ and environment focused on a local overdensity measurement (e.g., Dressler 1980). The local overdensity measurement employed here (${{\rm{\Sigma }}}_{n}$) is simply the number of galaxies (in our case, n = 5) with ${M}_{r}\gt -20.3$ mag, divided by the volume required to enclose that number of neighbors above the magnitude limit. We adopt the ${{\rm{\Sigma }}}_{5}$ measurements from Goddard et al. (2017; as defined by Etherington & Thomas 2015).

${{\rm{\Sigma }}}_{n}$ is a complex measure of environment. For large overdensities, with more than three to five members, the region of measurement falls within the parent halo. However, when the number of members of a group approaches the overdensity measure, the distance to the nth nearest neighbor may fall in a different halo. In this case, the measure becomes more sensitive to very large scale clustering than to local overdensity (Muldrew et al. 2012; Woo et al. 2013). Furthermore, the physical processes impacting galaxies at group centers (e.g., merging) may be different than those dominating the satellite galaxies living further from the group center (e.g., stripping). We therefore prefer using the radial group position, $R/{R}_{200b}$, which ensures only intra-group comparisons are made. For comparison with previous literature, we examine both measures.

We employ a simple mass-weighting scheme to ensure flat mass distributions in all environmental bins. In each environment bin, we inversely weight each galaxy based on the number of galaxies at that mass, such that the final mass distribution is flat. We do this in two mass bins, divided at the median stellar mass. We do not employ the MaNGA weights here, since we are looking at renormalized mass distributions.

Throughout the rest of the Letter, we will display the weighted slow-rotator fractions as a function of ${{\rm{\Sigma }}}_{n}$ and $R/{R}_{200b}$. In these figures, the bin sizes are chosen to ensure a constant number of objects per bin and are plotted at the weighted bin center. The shaded regions indicate the error in the mean, derived via bootstrapping.

We first examine whether there is a trend between ${\lambda }_{\mathrm{out}}$ and local overdensity at fixed M_*. Figure 1 (top left) presents two clusters that illustrate the range of results from individual studies of clusters, and the slow-rotator fraction is shown for our full sample of satellite and central galaxies. There is no strong trend with local overdensity in our data, which are roughly consistent with the Virgo results in a similar Σ range (although the Virgo results employ ${{\rm{\Sigma }}}_{3}$ rather than ${{\rm{\Sigma }}}_{5}$). We see a rising slow-rotator fraction at low ${{\rm{\Sigma }}}_{5}$. This rising fraction reflects the increasing dominance of higher-mass central galaxies at low overdensity in our sample. Figure 1 (top right) shows that there is a trend (albeit weak) between local overdensity and stellar mass as satellites grow more dominant at higher overdensity. To garner reasonable results, we must examine the trends with overdensity at fixed mass. Note the full relationship between stellar mass and local overdensity is complicated for satellites (e.g., Woo et al. 2013) and is not fully probed by our early-type, high-mass sample.

In Figure 1 (bottom left), we impose mass-weighting. While higher-mass galaxies show a higher slow-rotator fraction, no residual trend is seen with ${{\rm{\Sigma }}}_{5}$ for either bin. Since slow-rotator fraction is assigned as a binary (noisy) division, it is useful to examine the full distribution of ${\lambda }_{\mathrm{out}}$/$\sqrt{\epsilon }$ (Figure 1, bottom right). Again, we see a split in samples based on stellar mass, but no residual trend in ${\lambda }_{\mathrm{out}}$/$\sqrt{\epsilon }$ as a function of ${{\rm{\Sigma }}}_{5}$. We thus turn to examine ${\lambda }_{\mathrm{out}}$ as a function of $R/{R}_{200b}$.

$R/{R}_{200b}$ is complementary to ${{\rm{\Sigma }}}_{5}$, since the radial distance is always measured within the group halo, while the local overdensity can extend to neighboring halos at low density. Figure 2 (top left) presents the slow-rotator fraction as a function of $R/{R}_{200b}$ for all satellites in the sample. The central galaxies by construction are found at the group center and have a 40% slow-rotator fraction averaged over all masses. There is no compelling trend in slow-rotator fraction as a function of radius. In terms of mass, the central galaxies are more massive than the satellites, but there is not a strong mass segregation within the satellite galaxies (Figure 2, top right). The true radial mass trend is washed out by our narrow range in stellar mass and morphology (e.g., Woo et al. 2013).

Zoom In Zoom Out Reset image size

We then consider the mass-weighted slow-rotator fraction as a function of radius (Figure 2, bottom left). Again, the high- and low-mass–weighted samples have higher and lower slow-rotator fractions, respectively, but there is no compelling additional evidence for a radial trend in slow-rotator fraction when we control the distribution in stellar mass. We similarly see a very flat mean ${\lambda }_{\mathrm{out}}$/$\sqrt{\epsilon }$ as a function of radius (Figure 2, bottom right).

In short, there is no evidence that large-scale or local environment plays a driving role in the distribution of ${\lambda }_{\mathrm{out}}$, for an array of environmental measures. Of course, it is possible that a stronger radial trend might be apparent if we could consider only the most massive halos; such analysis required larger samples and should be possible with MaNGA soon.