THE GALEX/S⁴G UV–IR COLOR–COLOR DIAGRAM: CATCHING SPIRAL GALAXIES AWAY FROM THE BLUE SEQUENCE

Using the asymptotic photometry described above, we construct the (FUV–NUV) versus (NUV–[3.6]) color–color diagram (Figure 1). We find a clear bimodal pattern: a very tight blue sequence, which we call the GALEX blue sequence (GBS), and a less tight red sequence we call the GALEX red sequence (GRS). The distribution of the GRS is perpendicular to that seen in optical color–color diagrams (e.g., Strateva et al. 2001), likely due to the mass and metallicity dependence of the UV upturn (Boselli et al. 2005). The more massive and metal rich a galaxy is, the more important the contribution from the UV upturn to the UV emission and bluer the (FUV–NUV) color. Alternative uses of this color are still being explored (e.g., Hernández-Pérez and Bruzual 2014; Zaritsky et al. 2014, 2015).

Zoom In Zoom Out Reset image size

To quantify the GBS and the GRS, we split the color–color plane by choosing a line that passes through coordinates (2,3) and (5,1) (left panel of Figure 1). We fit the bluer half with a simple error-weighted linear regression (the subsample size here is 1787 galaxies), giving us a slope of 0.12 ± 0.01 and a y intercept at 0.16 ± 0.01. The standard deviation of the sample over this range is rms_GBS = 0.20, while a Gaussian fit to the residuals yields an even smaller value of σ_GBS = 0.10.

We also fit the redder half (144 galaxies) with a simple error-weighted linear regression (in the (NUV–[3.6]) direction), giving us a slope of −4.21 ± 1.62 and an rms_GRS = 0.45. The location of galaxies in this diagram is strongly correlated with the Hubble type (see also Figure 4). There are only two true ellipticals in the GBS, ESO 548-023 and NGC 855. Some of the ellipticals found in the GBS are actually BCDs morphologically misclassified as ellipticals. The latter plus the 31 later-typed ETGs (E-S0,S0) located in the GBS either could be in the tail of star formation (Bresolin 2013) or could be rejuvenated systems (Kannappan et al. 2009; Thilker et al. 2010). Finally, early-type spiral galaxies (S0-a, Sa, Sab) are populating the GBS, the GRS, and the region in between (which we name GALEX green valley, GGV), which is consistent with a scenario where these galaxies are transitional objects.

Figure 1 (right) shows the residuals of the UV color to the GBS linear fit obtained in our UV–IR color–color diagram (left), (FUV–NUV)–(FUV–NUV)_GBS, as a fraction of the absolute magnitude at 3.6 μm, M_[3.6], and stellar mass measured assuming a ${{\Upsilon }_{3.6}}$ of $0.6{{M}_{\odot }}/{{L}_{\odot }}$ (Meidt et al. 2014; B. R. Röck et al. 2015, in preparation). Stellar mass is not the driver for the evolution of these systems in or out the GBS. The distribution of our samples of ETGs (our data and those from Zaritsky et al. 2015) is similar, meaning that the bias against HI-poor ETGs in S⁴G is not driving our results.

In Figure 2, we show the optical CMD, commonly used to study galaxy evolution (e.g., Faber et al. 2007). We show the position of our GBS, GRS, and GGV galaxies using blue, red, and green points, respectively (Figure 1).

Zoom In Zoom Out Reset image size

Despite the low number of GGV galaxies overall, they fall either on the optical red sequence (RS) or on the (ill-defined) optical green valley (GV), with only a very few being optically blue galaxies. This result indicates that GGV galaxies are evolving off the GBS much slower in (NUV–[3.6]) than in (FUV–NUV), while the optical colors are reddened by more than [0.4 mag] 0.2 mag in [(U–B)] (B–V). Alternatively, the GGV galaxies are optically red systems that are growing a disk (toward the GBS), in which case the optical reddening is less than [0.3 mag] 0.2 mag in [(U–B)] (B–V).

It is interesting to note that the majority of the galaxies in the optical GV belong to the GBS, i.e., they are star-forming systems to be classified as such at UV wavelengths. The same can be said about the blue dots found in Figure 2 near the optical RS. This means that either they are in the tail end of star formation or they are regrowing a disk. In particular, a disk regrowth is unlikely, given that the GGV galaxies are found preferentially in dense environments where such a process may be more difficult than in the outskirts of a cluster or a lower-density environment (Section 4.3).

By using the classification of interacting systems within S⁴G given by Knapen et al. (2014), we do not find a clear difference between the fraction of close interactions in the three regions of our color–color diagram. When splitting the sample by level of interaction and morphological type, low number statistics hamper determining, for example, whether or not E/S0 in the GGV are mergers or post-starbursts.

In Figure 3 (left), we show single stellar population (SSP) models (initial mass function, IMF: Chabrier) of S. Charlot & G. Bruzual (2007, private communication) for different metallicities and ages. For all cases, the (FUV–NUV) color becomes redder until it reaches a peak (at about 1 Gyr in the case of solar metallicity), and then becomes bluer again (UV-upturn) as the models age. For these SSP models, the color transition from GBS to GRS takes less than a few 10⁸ yr.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

To better understand what drives the position of galaxies in the color–color diagram, we use a grid of disk galaxy models from Boissier & Prantzos (2000) with various circular velocities V_c and spin parameters λ (Figure 3, right). These models span the entire GBS due to their variation in the star formation timescale with V_c and λ. Besides, the reddening arrow points along the GBS, so internal extinction contributes little to its spread since the star formation timescale and dust attenuation are nearly degenerate in these colors (Section 4.2). The tightness of the GBS is also due to the fact that the effects of an episodic SFH in the FUV–NUV color (compared to the NUV–[3.6]) are only expected at very low masses (typically rather blue NUV–[3.6] colors). Noteworthy examples of this are PGC 065367 or ESO 245-007. To produce more realistic galaxy colors, we add a bulge to the Boissier & Prantzos (2000) disks using SSP predictions with ages between 10⁹ and ${{10}^{10.3}}$ Gyr and a different formation timescale τ. We adopt a bulge model with (NUV–[3.6]) = 5.3 mag and (FUV–NUV)color = 2.4 mag.

Three of these "disk + bulge" models are shown in Figure 3 (right) for various bulge-to-total mass ratios $B/T$. Each colored line represents the colors of the resulting galaxy model for $B/T$ from 0.0 to 0.5 in steps of 0.1. Adding a BC03 SSP-based bulge to any of our disk models does not reproduce the reddening in (FUV–NUV) observed in numerous early-type spirals unless we assume a very high $B/T$ (typical observed $B/T$ of S0 and S0-a galaxies are below 30%; Laurikainen et al. 2007). This shows that the position of transitional early-type spirals is not only due to their (red) massive bulges but also to their disks being red compared to the disks of galaxies in the GBS. This result explains the red (FUV–NUV) colors of some (transitional) disk galaxies. Whether those disk galaxies could evolve into systems with properties similar to those of current-day ETGs cannot be determined from these data alone.

To analyze the transformation timescale of these disks under a more realistic scenario, we use a model that includes ram-pressure stripping of the gas (hereafter RPS) as a potential mechanism, albeit not the only (potential) one, to redden a galaxy disk. This RPS model starts from an unperturbed disk with V_c = 270 km s⁻¹ and λ = 0.04 but with additional ram-pressure to fit the properties of the Virgo galaxy NGC 4569 (Boselli et al. 2006). The points represent the epochs when the peak of RPS occurred, with t = 0 corresponding to the present. Both the disk models and the peak RPS models are shifted redward by 0.3 mag in (FUV–NUV) to match the observed color–color distribution. The need for an offset has been discussed in Muñoz-Mateos et al. (2011) and is believed to be due to uncertainties in the calibration of the stellar atmospheres used in the spectral synthesis. This latter RPS disk model fits the observed distribution well, especially with a moderate bulge. The transition timescale derived from the GBS to the GRS in this case would again be of the order of a few 10⁸ yr.

To analyze the effects of internal dust, we show the reddening in both colors in Figure 1 (left) for A_V = 0.5 mag, assuming the attenuation law of Calzetti et al. (1994). This law reproduces well the relation between IR excess and UV color (Gil de Paz et al. 2007). Interestingly, the reddening arrow indicates an almost complete degeneracy in these colors between reddening and a change in the star formation timescale (as illustrated by the models of Boissier & Prantzos 2000). We find no change in the distribution of offsets from the GBS as a function of the galaxies' axial ratio (not shown), so the (small) spread of the GBS is unlikely to be due to inclination effects.

In Figure 4, we show, superposed on the number density color–color diagram, the distribution of galaxies that are in Virgo or high-density regions (Laine et al. 2014). The relative abundance of transitional galaxies in the Virgo cluster is higher for the majority of morphological types, suggesting that an environmental effect is present. There are some indications that environment plays a role in determining whether galaxies lie in the GBS, GGV, or GRS. For example, 43% (12 of 28) of Sa galaxies classified as either GRS or GGV are in the Virgo cluster, while only 12% (16 of 137) of those classified as GBS are in Virgo. The relative preference of GGV and GRS galaxies for the Virgo environment is found for each morphological class separately (Table 1); although in some cases, the results are not statistically significant. Finally, although the presence of early-type spirals in the GGV might be related to environment, this might not be simply due to RPS; there might be other, more fundamental drivers (that also depend on environment) that determine the position of galaxies in this color–color diagram and in the optical CMDs. One possibility might be the degree of uniformity of the angular momentum of accreted material, which would segregate galaxies by morphological type, mass, and color in different environments (less uniform accretion having likely occurred in denser environments).