The nuclear region of NGC 613 – II. Kinematics and stellar archaeology

In order to study the gas emission from the data cubes and also to analyse the stellar archaeology, spectral synthesis was applied, using the starlight software (Cid Fernandes et al., 2005), with a base created from the Medium-resolution Isaac Newton Telescope Library of Empirical Spectra (MILES; Sánchez-Blázquez et al. 2006) in the optical wavelengths (GMOS and SIFS data cubes). The spectral synthesis is applied to each spectrum of the data cube, and is performed by a linear combination of the base spectra, in order to fit the observed spectra. With that, it is possible to create flux maps of the detected stellar populations, to obtain histograms of the flux fractions of those stellar populations in a determined region of the FOV of the data cube (which may be the whole FOV or a specific area) and to remove the stellar continuum of the data cube to study the gas emission (as done in Paper I). This last process corresponds to creating a synthetic stellar data cube, using the synthetic spectra obtained from the spectral synthesis, and then by subtracting this data cube from the original one. We obtain a cube with mainly gas emission, which we call gas data cube. In this work, the gas data cube was used to study the gas kinematics.

The penalized pixel fitting method (pPXF; Cappellari & Emsellem 2004) was used in the SINFONI data cube. This method consists of a spectral synthesis with the base spectra convolved with Gauss–Hermite functions. The base used covers only the $K$ band (Winge et al. 2009). In this case, the pPXF was used to obtain a $K$-band gas data cube, to study the stellar kinematics in the optical (SIFS and GMOS data cubes) and in the determination of the stellar velocity dispersion with the SINFONI data cube. For more details, see section 5.

In section 4, we present the results of the stellar archaeology obtained from the spectral synthesis applied to the extracted spectra from different regions of the data cubes (GMOS and SIFS). We characterized the detected stellar populations by obtaining histograms showing the flux fractions associated with each stellar population. In order to estimate the uncertainty of this process, first of all, we determined the medians of the ages of the detected stellar populations in all these regions of the data cubes. Such medians were calculated using graphs of the accumulated flux fractions as a function of the ages of the stellar populations and taking the ages at which the accumulated flux fractions were equal to 50%. The uncertainties of these medians were obtained with the following Monte Carlo procedure: First, we constructed, for each extracted spectrum, a histogram representing the spectral noise and fitted a Gaussian function to this histogram. After that, we obtained Gaussian distributions of random noise with the same width of the Gaussian fitted to the initial histogram. Such noise distributions were added to the synthetic stellar spectrum provided by the spectral synthesis of the original spectrum. Finally, we applied the spectral synthesis to all the obtained “noisy” spectra and calculated the median of the ages of the detected stellar populations. The uncertainty of the median of the ages was taken as the standard deviation of all the median values obtained with this process. The representative uncertainty of the process was taken as the highest value of the uncertainties determined for the median ages. The result obtained is 0.02 dex.

When we take the results from the spectral synthesis it is important to notice that there are degeneracies in the method. Different linear combinations of the base spectra could result in the same continuum. However, in many cases, it is possible to notice the consistence of the spectral synthesis, since it can differentiate spectra with deeper stellar absorptions (associated with old stellar populations) from spectra with flatter stellar absorptions (associated with younger stellar populations) with good precision. It is not possible to determine an uncertainty to those degeneracies, and we are in need of more precise stellar bases and methods to perform this analysis.

In Paper I, we came out with a general scenario to the main structures of the NGC 613 nucleus in terms of gas emission. There is an optical double emission, separated by a stream of dust, which we called N1 and N2. We believe that the AGN, identified as a variable point-like source (V1), is located between N1 and N2. The eight Hii regions in the circumnuclear ring orbit around this nuclear structure (composed by N1, N2 and V1). As we will see in this work, N1 and N2 are part of the same structure, as we stated in Paper I. The stream of dust that divides this structure is probably associated with the nuclear spiral that is also located between N1 and N2.

The centroid of the [O i]$\lambda$6300 emission, [O i]_C, which is a line typical of regions of partial ionization, is not located at the position of V1 (the AGN). The projected distance between V1 and [O i]_C is $\sim$ 0.24 arcsec. This difference may be due to the ionization cone being partially obscured by dust near the AGN. In Paper I, [O i]_C was a reference point to the images superpositions, to compare directly the data from different instruments, since it was the most point-like source in the GMOS data cube (see section 3 of Paper I, for more details). In the images of this work, we indicate the position of [O i]_C and V1.

In this section, we will study the gas dynamics to determine, with better precision, how the structures behave in the nuclear environment of NGC 613. The study is made according to the FOV of each instrument. Since GMOS has the inner central FOV, we are able to study the inner central gas kinematics. In the cases of the SIFS and SINFONI FOVs, we have, besides the central, the circumnuclear gas kinematics in greater detail.

The GMOS data cube has a FOV that shows the regions that are internal to the circumnuclear ring. Channel maps of emission lines are a good tool to visualize in greater detail the gas kinematics. Fig. 1 shows the channel maps of the H$~{}\alpha$ emission line. As presented in Appendix A of Paper I, the spectra of the central region of NGC 613 have blended lines, so we have to consider that the first and last channels might be contaminated by the [N ii]$\lambda$6748 and [N ii]$\lambda$6784 emission lines, respectively. Between the second and eighth channel maps, we see that there is a gas rotation of the regions at the edges of the FOV, which are the H ii regions of the circumnuclear ring. We also notice a central emission with a defined orientation between the 8th and 10th channel. This emission is in redshift and probably represents an outflow of gas that comes from V1. In order to represent this outflow, we took the channel with v= 306 km s^-1 and the PA of this emission is $\sim$ 17^∘.

The channel maps of the [O iii]$\lambda$5007 emission line of the GMOS data cube are very noisy and hard to interpret due to the low signal-to-noise ratio (S/N) in the blue region of the optical spectra. However, we notice a very clear emission when we observe the channel with –650 $\lesssim$ v $\lesssim$ –590 km s^-1 (see Fig. 2). This emission has a preferred direction, whose PA $\sim$ 43^∘ is not compatible with the one estimated for the outflow observed in the H $\alpha$ channel maps, neither is the velocity, since it is in blueshift.

The SINFONI data cube shows the positions of the H ii regions with great precision when we look to the Br $\gamma$ emission line images (Böker et al., 2008; Falcón-Barroso et al., 2014). Thus, we made an RGB composition of this line to determine the gas kinematics in the circumnuclear ring (see the left-hand panel of Fig. 3). We notice that there is a gas rotation of the H ii regions around the AGN: The regions located to the west present gas emission in redshift and the gas with velocities close to zero seems to be located to the east, together with the gas with emission in blueshift. The [Fe ii]$\lambda$16436 RGB (right-hand panel of Fig.  3) presents also the same trend; however, the granularity of the image of the H ii regions is higher than in the Br $\gamma$ image.

Besides the RGB compositions of [Fe ii]$\lambda$16436 and Br $\gamma$, we also created RGBs of the H${}_{2}\lambda$21218 emission line (Fig. 4 panels a and c). In the inner centre of the FOV, we see that the gas with emission in blueshift is located to the north, whereas the gas with emission in redshift is located to the south (Fig. 4a). That might indicate a rotation of molecular gas around the AGN. When we look at the RGB composition of the ALMA data cube of the inner centre (Fig. 4b), we see that the nuclear spiral has an opposite pattern in the nucleus: blueshift to the south and redshift to the north (the same trend was detected by Audibert et al. 2019). We see also that the nuclear spiral is connected to the bar and its arms have the same kinematics, which suggest that the nuclear spiral is bringing gas and dust to the centre, feeding the AGN. This process has no connection with the molecular gas observed in the H${}_{2}\lambda$21218; therefore, we are presenting two different phenomena. In the ring (panels c and d of Fig. 4), the molecular gas shows emission (in the circumnuclear ring) in redshift to the west and emission in blueshift to the east, as in the previous RGBs. However the low-velocity gas is located to the west, together with the gas in redshift, the opposite of the RGBs of [Fe ii]$\lambda$16436 and Br $\gamma$. In this case, we could not separate completely the images representing higher positive velocities (red colour in the RGBs) from the images corresponding to lower moduli of velocities, due to the insufficient spectral resolution.

One of the most relevant spectral features in the SIFS data cube is the [O iii]$\lambda$5007 emission line. As we know that there is an AGN (V1), we can associate this emission with an ionization cone that comes from it (see Paper I). The [O iii]$\lambda$5007 kinematics was also studied here with the channel maps of Fig. 5. The morphology of the ionization cone can be seen in the channels with low velocities (–170 and 9 km s^-1). The [O iii]$\lambda$5007 emission line has a profile with a prominent blue wing, indicating outflows with high negative velocities. The outflow appears clearly in the channel with v = –889 km s^-1 and PA= -10°. When we compare the channels with velocities –530 and 429 km s^-1, we see that there might be a rotation of gas in the inner central region, although we cannot exclude the possibility of other gas outflows. There is no correlation between the [O iii]$\lambda$5007 emission and the star-forming regions, except that there is a region of a break in the ring that might be associated with the emission of the ionization cone.

We studied the H $\alpha$ emission-line kinematics of the SIFS data using the channel maps of Fig. 6. As in the GMOS data cube, we have to consider that, in the inner regions of the FOV (mostly in the regions closer to the AGN), the H $\alpha$ line is blended with the [N ii]$\lambda\lambda$6548, 6584 lines, so the first and last channels might be contaminated with this emission. When we look at the channel maps of Fig.6, we see that the kinematics of the H ii regions in the circumnuclear ring (which we cannot spatially separate, due to the low spatial resolution) is consistent with what we see in the SINFONI and ALMA data (see Figs. 3 and 4). The H ii regions to the south-east of V1 show emission in blueshift and those to the north-west show emission in redshift.

N1 and N2 spectra were extracted from the GMOS data cube according to Fig. 10 in Paper I before creating the gas data cube. The diameter of the extraction region was equal to the FWHM of the PSF of the data.

From the results of the spectral synthesis, we made histograms of the flux fraction (normalized in each region) of the detected stellar populations. In both regions, the young stellar populations (10⁷ yr) with high metallicity (0.02 and, mostly, 0.05) are responsible for the highest flux fractions. There are a few differences between regions N1 and N2: the presence of the 10^8.5-yr stellar populations with 0.02 metallicity and the old stellar populations (10¹⁰ yr) with intermediate metallicity (0.008) in the N2 region (Fig. 7).

When we compare these data with the SIFS data (see section 4.2), we see that regions N1 and N2 show flux fractions associated with mainly young stellar populations, while, in the ring, we see a balance between old and young stellar populations.

Fig. 8(a) shows the flux map of the populations of 10⁹ yr detected with the spectral synthesis applied to the entire FOV of the GMOS data cube. The map includes the flux of the 10⁹-yr stellar populations with all metallicity values, although the dominant metallicity is high (0.02 and 0.05). We notice, by looking at the histograms of Fig. 7 and this map, that the concentration of this stellar population is completely circumnuclear and resembles a ring of radius $\sim$ 1 arcsec intermediate between the nucleus and the previously known circumnuclear ring (see section 6.1.1).

The histogram of Fig. 8(b) shows that the 10⁹-yr stellar populations are responsible for the largest flux fraction in the inner region of the nucleus of NGC 613, followed by the 10⁷-yr stellar populations. It also shows how relevant is the inner circumnuclear ring of this intermediate-age stellar populations. A comparison between this histogram and the one in Fig. 7 reveals that the 10-Gyr stellar populations are responsible for a larger flux fraction in the circumnuclear region when compared to regions N1 and N2.

It is already known in the literature that NGC 613 has a circumnuclear star-forming ring (Böker et al., 2008; Falcón-Barroso et al., 2014). So, from the SIFS data cube, whose FOV covers all the area of the ring (besides the two additional regions detected in this work, see Fig. 10 from Paper I), we applied a spectral synthesis with the starlight software to the extracted spectra of the H ii regions named from 1 to 10 (whose extraction areas are shown in Fig. 10 of Paper I).

We created a histogram from the results of the spectral synthesis applied to the spectrum of each region, as shown in Fig. 9. We notice that all regions have mainly both young (10⁷ yr) and old ($\sim$ 10¹⁰ yr) stellar populations, whereas some regions do not show flux fractions associated with intermediate ages. Regions 7, 9, and 10 show most (or all, in the case of region 9) of their flux fractions attributed to high-metallicity stars (0.02 and 0.05). The other regions present, apparently, flux fractions divided between high and low (10^-4 and $4\times 10^{-4}$) metallicities.

The stellar velocity map of Fig. 10(a) is very irregular. One can note a rotation pattern in the inner central 0.5 arcsec. Despite the irregularities, we notice that there is a predominance of negative velocities to the east/south-east and positive velocities to the west/north-west of [O i]_C. The kinematic centre was defined as the point, along the line of nodes, with a velocity value equal to the average between the maximum and minimum velocity values. This average velocity was subtracted from the velocity map, so Fig. 10(a) actually shows the radial velocities relative to the kinematic centre, which is between [O i]_C and N2 and closer to N2. However, considering the uncertainty, the kinematic centre is also compatible with V1, [O i]_C, or even with the N1 centroid, which is not represented in the image.

The stellar velocity dispersion curve has two peaks (Fig.10b), one at $\sim$ 1 arcsec and the other at $\sim$ –1 arcsec, and the values decrease towards the centre and towards the edges. There are also perturbations in the velocity curve, with values lower than expected, in the case of a Keplerian motion. The positions of such perturbations are coincident with the ones of the velocity dispersion peaks and also with the possible inner circumnuclear ring of 10⁹ yr stellar populations. The velocity dispersion at the centre of the curve is $\sim$ 126 km s^-1 and, at the edge, is $\sim$ 100 km s^-1 (Fig.10b).

In the more peripheral areas of the $h_{3}$ map, we can see a possible anticorrelation with the velocity map (Fig. 10c). The uncertainties in those regions are high, and as the velocity map itself has irregularities, we cannot draw conclusions from this result.

Since it has a larger FOV, the stellar velocity map from the SIFS data cube reveals more clearly the stellar rotation that we could not see in the GMOS results (Fig. 11a). The kinematic centre is not compatible with V1 in this case, being at a distance of 0.66 $\pm$ 0.11 arcsec. The stars have negative velocities to the south-east and positive velocities to the north-west. The PA of the line of nodes is –70^∘ $\pm$ 1^∘, compatible at the 3$\sigma$ level with the PA of the line of nodes of the GMOS results, which is –62^∘ $\pm$ 3^∘.

The velocity dispersion curve of Fig.11(b) shows, as in the GMOS velocity dispersion curve, two peaks at the same position of the perturbations in the velocity curve, although, in the SIFS results, those peaks and perturbations are diluted due to the lower spatial resolution of these data. Those peaks are not located at the same positions of the peaks observed in the GMOS data cube. In fact, their locations are consistent with the circumnuclear ring (not the inner ring as in GMOS). The value of the velocity dispersion at the centre is $\sim$ 123 km s^-1.

The $h_{3}$ map (Fig. 11c) has irregularities but, in general, shows an anticorrelation with the velocity map in the region of the ring, which may indicate that the stars of the ring rotate superposed to a stellar background with velocities close to zero towards the line of sight.

Although the SINFONI data cube could have a significant amount of information related to the stellar kinematics, the low S/N prevented the construction of pPXF parameters maps that could be interpreted. In order to obtain, at least, more reliable information about the stellar velocity dispersion, the pPXF method was applied to spectra extracted from concentric rings around V1, with 3 spaxels of thickness. With that, we could minimise the spectral noise and obtain the curve of Fig. 12.

The uncertainties of the values of the velocity dispersion were obtained by applying a Monte Carlo method to each of the extracted spectra. First, for each spectrum, we subtracted the stellar continuum, using the corresponding pPXF fitting. Then, we created a histogram with the values of the spectrum (after the stellar continuum subtraction) in many spectral intervals, without emission lines. We fitted a Gaussian function to this histogram and created Gaussian distributions of random values with the same width of the Gaussian fitted to the histogram. Those distributions were added to the synthetic stellar spectrum provided by the pPXF, resulting in “synthetic noisy spectra”. At the end, the pPXF was applied to each one of those “synthetic noisy spectra” and the final uncertainty of the stellar velocity dispersion was taken as the standard deviation of all the obtained values.

Unlike what we see in the previous results in the optical, we see here (Fig. 12) a peak of the stellar velocity dispersion at the centre, with a value of $\sim$ 144 km s^-1. The stellar velocity dispersion decreases with the radius and reaches a nearly constant value of $\sim~{}95$ km s^-1. At $\sim$ 1.8 arcsec from V1, there appears to be a second peak of velocity dispersion, with a value of $\sim~{}121$ km s^-1; however, considering the uncertainties of the values, it is not possible to confirm the existence of this second peak.

In order to obtain a stellar velocity dispersion value representing the central region of NGC 613, we calculated a weighted mean of the values of the map obtained with the pPXF from the SINFONI data cube, taking as weights the integrated fluxes of each spaxel. The value obtained was 92 $\pm$ 3 km s^-1. The uncertainty was calculated after the creation of two additional maps, one with the sum of the uncertainty map and the other with the subtraction of the uncertainty map. Then, we calculated the weighted means of the dispersion of the obtained maps. The uncertainty was the standard deviation of those values.

N1 and N2 are two regions with significant stellar emission, so it was convenient to apply the spectral synthesis to their spectra (Fig. 4.1). The flux fractions of those two regions are due to, essentially, the same stellar populations, without significant differences. We did not detect any region of the data cube with a significant contribution of the featureless continuum of the AGN (represented by a power law with spectral index of 1.5). We actually performed a few tests, applying the spectral synthesis with power laws with different spectral indexes (with values between 1.0 and 1.9); however, we did not find credible evidence for the presence of a featureless continuum, which might indicate that this AGN is so obscured that we cannot detect such an emission. This result is compatible with the conclusions of Paper I, in which, by analysing the map of D_CO created by Falcón-Barroso et al. (2014), we could not see the effect of the featureless continuum that would create a decrease of the values of this map towards the centre (or where the AGN is located). Another hypothesis is related to the fact that the AGN might be a variable source (V1), as we claimed in Paper I. The absence of a featureless continuum might be due to a low state of activity of the AGN at the epoch of observation.

The results of the stellar archaeology do not show a continuous star formation in the central region of NGC 613, as only four events of star formation were detected, with most of the H ii regions showing evidence of just two or three of such events. In the cases of N1 and N2, evidence of two and three star formation events, respectively, was detected. This suggests that the influence of the bar on the star formation in the central region may not be significant, or just might be very recent (10 million years). The absence of a continuous star formation in the nucleus of NGC 613 is not in disagreement with the fact that this galaxy hosts a pseudo-bulge. Actually, as shown by Breda & Papaderos (2018), the bulge formation process in a galaxy probably does not follow only two possible routes: one involving a quasi-monolithic gas collapse, which would result in a classical bulge, and the other involving secular processes with continuous star formation, which would form a pseudo-bulge. Instead, the authors conclude that bulges and discs evolve simultaneously in a process that results in a continuous sequence of observed properties, such as stellar age, metallicity, and surface brightness. In addition, it is worth mentioning that studies have shown that the properties of certain pseudo-bulges, specially in isolated galaxies, suggest that most of their mass was formed in an early epoch, without much subsequent star formation (Fernández Lorenzo et al., 2014).

The circumnuclear star-forming ring kinematics reveals the same pattern in all the spectral bands and emission lines analysed in this work. The regions located to the east have negative velocities and those to the west have positive velocities (see Figs. 3 and 4). This result is also compatible with the one obtained by Miyamoto et al. (2017, 2018) using velocity maps of molecular gas.

We notice that the nuclear spiral (Audibert et al., 2019) has its own kinematics that suggests feeding of the central region. The gas emission of this nuclear spiral to the north-west is in blueshift and that to the south-east in redshift. This same pattern was detected in the H${}_{2}\lambda$21218 emission (see Fig. 4), which suggests that such a pattern may also be related to the feeding of the central region.

The H $\alpha$ channel maps of the GMOS data cube reveal the gas kinematics of the inner central region (Fig.1). We notice, at the edges of this FOV, the ring kinematics observed in NIR and ALMA data. Besides that, the outflow we detected, with v = 306 km s^-1, is located at a position compatible with V1 and has a PA $\sim$ 17^∘, compatible with the PA of the radio jet observed by Hummel & Jorsater (1992) (see also Fig. 16 of Paper I). The rotation pattern detected with the H $\alpha$ channel maps of the SIFS data cube (Fig. 6) is consistent with what we saw in the GMOS, SINFONI, and ALMA data cubes. The outflow of gas, detected in the GMOS data cube, though, cannot be seen here in the SIFS data, due to the lower spatial resolution.

The ionization cone detected in some of the [O iii]$\lambda$5007 channel maps (v = –170 and 9 km s^-1) is mostly located to the north-east of V1. The emission detected to the south-west, with velocities higher than –170 km s^-1, could be the other side of the ionization cone. The size of the “break” in the circumnuclear ring, detected in the Br $\gamma$, H${}_{2}\lambda$21218, and [Fe ii]$\lambda$16436 images (Falcón-Barroso et al., 2014), north-east of V1, is compatible with the apparent size of the ionization cone. This might indicate that this region is being ionized, along the ionization cone, by the radiation coming from the AGN. However, one of the detected outflows in [O iii]$\lambda$5007 has a PA that is compatible with the axis of the ionization cone (PA $\sim$ 55^∘ and v = –710 km s^-1, observed with the SIFS data cube, and PA $\sim$ 43^∘ and –650 $\lesssim$ v $\lesssim$ –590 km s^-1, observed with the GMOS data cube) and, therefore, within this ”break” region. The outflow might be pushing the gas out of the ring and possibly preventing star formation in this region or the ionization coming from the AGN is heating the gas at some point where it is not possible for the gas to cool down and form stars. This seems to be just a local impact of the feedback of this AGN in this galaxy, but Davies et al. (2017) found evidence that the outflowing material from the ionization cone might be interacting with the interstellar medium at larger scales by shocks.

There is a second outflow of gas observed in the [O iii]$\lambda$5007 channel with velocity of v = –889 km s^-1, with PA $\sim$ –10^∘. Though this outflow has a PA compatible with the one observed in the H$~{}\alpha$ channel map and the radio jet, its velocity is very different, so it must be an isolated event or associated with other phenomenon that we did not detect.

According to Hopkins & Quataert (2010), inflows of gas in a galaxy start at kpc-scales and may be driven by major mergers or even by secular processes (involving the formation of bars). At distances between 1 kpc and 10 pc from the central supermassive black hole (which correspond to the region analysed in this work), the inflow continues, but the gas can have many spatial morphologies, such as spirals, rings, clumps or even secondary bars. The observed properties of the central region of NGC 613 are consistent with this scenario, as the line-emitting regions show different morphologies, such as a nuclear spiral and a circumnuclear ring. As discussed in Section 6.1, this inflow of gas may also result in a gas reservoir for the star formation.

The stellar kinematic pattern detected in this work is essentially the same of the one observed for the gas kinematics along the circumnuclear ring (Figs. 3 and 4): blueshift to the south-east and redshift to the north-west of V1 in the velocity maps obtained from the pPXF of the SIFS data cube (the only cube that allowed the determination of the kinematics along the entire ring extension – see Fig. 11a). However, in order to establish a more precise comparison between the stellar and gas kinematics along the circumnuclear ring, we made a radial velocity map of the gas by fitting a Gaussian function to the H $\alpha$ emission line. We verified that the PA of the line of nodes and also the curve extracted along the line of nodes for both the stellar and gas kinematics are consistent. This indicates that the stars and the gas along the circumnuclear ring share the same kinematics.

The velocity maps obtained with the GMOS and SIFS data cubes (Figs. 10a and 11a, respectively) show evidence of perturbations (similar to "depressions"), which are probably related to the two peaks observed in the velocity dispersion curves (Figs. 10b and 11b). As said previously, the positions of these peaks coincide with the positions of the inner circumnuclear ring (in the GMOS results) and of the circumnuclear ring (in the SIFS results). A possible explanation for this behaviour is that the observed rings have lower radial velocities when compared to the bulge stars. Therefore, the superposition of the spectra of the stellar populations in the bulge and in the rings results in the increase of the velocity dispersion values.

The stellar velocity dispersion maximum determined from the curve obtained from the SINFONI data cube ($\sim~{}144$ km s^-1) is compatible with the ones obtained by Schechter (1983) and Batcheldor et al. (2005) (using the maximum penalized likelihood method), while our average value of the stellar velocity dispersion taken from the SINFONI data cube (92 $\pm$ 3 km s^-1) is compatible with the value obtained by Batcheldor et al. (2005) using the cross-correlation method. By observing the velocity dispersion curve of SINFONI, we can see that there is one central peak, followed by a higher value at $\sim$0.3 arcsec and a series of values that, within the uncertainties, are constant. We do not see clearly the perturbations mentioned before and it is probably due to the fact that those stellar populations have some different effects when we look at the NIR spectrum.

Comparisons between active and non-active galaxies suggest that active galaxies show, in their central regions, lower values of stellar velocity dispersion than non-active galaxies. Hicks et al. (2013) found in their sample of active galaxies that the values of the stellar velocity dispersion did not exceed 100 km s^-1 (see fig. 17 from Hicks et al. 2013). This might be due to the presence of young stars that inherited the cold kinematics of their parent clouds, which probably constitute a gas reservoir for the star formation and also for the accretion onto the central supermassive black hole. In this case, the central region of NGC 613 shows stellar velocity dispersion values more compatible with the ones of active galaxies (in the Hicks et al. 2013 sample) and also the presence of young stellar populations. In addition, the presence of the circumnuclear rings and the evidence of inflows of gas help to corroborate this idea.

The portion of the ring with negative velocities has positive values of $h_{3}$, while the positive velocity regions of the ring have negative values of $h_{3}$. This anticorrelation of velocity and values of $h_{3}$ could indicate that the stars of the ring rotate around the nucleus superposed to an stellar background with stars with velocities close to zero. Such anticorrelation is not seen to the north-west, which might indicate deviations of a simple stellar rotation around the nucleus.

By using the stellar velocity map obtained from the SIFS data cube, we could estimate a representative velocity for the stars of the circumnuclear ring, v $\sim$ 100 km s^-1, rotating around the nucleus. Considering this velocity and the average values of the ring radius and inclination taken from Table 4 of Paper I ($\bar{R}$ = 249 pc and $\bar{i}$ = 57^∘), we conclude that the time necessary for a group of stars to complete a turn around the ring is $\sim$ 8 $\times~{}10^{6}$ yr. The temporal resolution of the spectral synthesis method is lower than this value. The scenario ”pearls on string” (Böker et al., 2007, 2008; Falcón-Barroso et al., 2014) predicts that there are two gas concentrations at opposite points of the ring that form H ii regions. Such H ii regions rotate in the ring and this rotation generates an age gradient along the ring. Considering the estimated time to a full turn of a group of stars along the ring and the temporal resolution of the spectral synthesis, we conclude that more precise methods are necessary in order to determine stellar ages in this case, to confirm this scenario.