The MAGellanic Inter-Cloud (MAGIC) project – II. Slicing up the Bridge Free

Abstract

The origin of the gas in between the Magellanic Clouds (MCs), known as the Magellanic Bridge, has always been the subject of controversy. To shed light into this, we present the results from the MAGellanic Inter-Cloud II (MAGIC II) project aimed at probing the stellar populations in 10 large fields located perpendicular to the main ridge-line of H i in the Inter-Cloud region. We secured these observations of the stellar populations in between the MCs using the WFI (Wide Field Imager) camera on the 2.2 m telescope in La Silla. Using colour–magnitude diagrams, we trace stellar populations across the Inter-Cloud region. In good agreement with MAGIC I, we find significant intermediate-age stars in the Inter-Cloud region as well as young stars of a similar age to the last pericentre passage in between the MCs (∼200 Myr ago). We show here that the young, intermediate-age and old stars have distinct spatial distributions. The young stars correlate well with the H i gas suggesting that they were either recently stripped from the Small Magellanic Cloud (SMC) or formed in situ. The bulk of intermediate-age stars are located mainly in the Bridge region where the H i column density is higher, but they are more spread out than the young stars. They have very similar properties to stars located ∼2 kpc from the SMC centre, suggesting that they were tidally stripped from this region. Finally, the old stars extend to some 8 kpc from the SMC supporting the idea that all galaxies have a large extended metal-poor stellar halo.

INTRODUCTION

A key challenge in modern astrophysics is to understand how galaxies form and evolve. In particular, interactions between galaxies are an important driver of galaxy formation and evolution (e.g. Larson & Tinsley 1978; Barnes & Hernquist 1992). These encounters reshape galaxies by transferring mass, energy, gas and metals between them, creating new sites of star formation both within the galaxies and in stripped gas amidst them. Important clues about the galaxy formation process lie in the faint edges of galaxies. Galaxy formation simulations suggest that almost all galaxies, even the smallest, should contain an extended old, metal-poor, stellar halo that is primarily composed of the detritus from galactic mergers (e.g. Searle & Zinn 1978; White & Rees 1978; Bullock & Johnston 2005; Read et al. 2006; Cooper et al. 2010). This is supported by a host of observations from large spirals (e.g. Roman 1954; Eggen, Lynden-Bell & Sandage 1962; Chapman et al. 2006; Kalirai et al. 2006; McConnachie et al. 2006; Gilbert et al. 2012; Monachesi et al. 2013; Trujillo & Bakos 2013; Deason et al. 2014; Greggio et al. 2014) to dwarf galaxies (e.g. Aparicio, Tikhonov & Karachentsev 2000; Minniti et al. 2003; Noël & Gallart 2007; Pejcha & Stanek 2009; Strader, Seth & Caldwell 2012). Such haloes retain a fossil record of past mergers and interactions in the form of extended tidal debris (e.g. Searle & Zinn 1978; Martínez-Delgado et al. 2001; Muñoz et al. 2006; Sohn et al. 2007; Amorisco, Evans & van de Ven 2014).

The distribution and extent of such debris, or a characteristic underlying old stellar halo are both sensitive to the total mass and size of a galaxy (e.g. Noël & Gallart 2007; Noël et al. 2013a; Deason et al. 2014). Thus, probing the faint edges of galaxies can also give us constraints on their dark matter content. Observing these remote edges of galaxies is a complicated task due to the faintness of these outlying regions, typically μ ≥ 28 mag arcsec⁻² (e.g. Noël & Gallart 2007). Reliable surface brightness measurements at these faint levels require accurate flat-field division, sky determination, bright star masking, among others (e.g. Barker et al. 2009).

Our nearest irregular neighbours, the Magellanic Clouds (MCs) constitute a unique opportunity to study a gas-rich interacting system close-up and to investigate faint outskirts of galaxies. First, their close proximity (∼50–70 kpc away) enables us to resolve their individual stars, which allows us to directly probe fainter surface brightness levels placing constraints on ages and metallicities without the need to use stellar population synthesis (e.g. Noël et al. 2013b). Secondly, due to their mutual association over the past few Gyr (as determined from their proper motions: Kallivayalil, van der Marel & Alcock 2006; Besla et al. 2007, 2012; Piatek, Pryor & Olszewski 2008; Costa et al. 2011; Kallivayalil et al. 2013; van der Marel & Kallivayalil 2014; Cioni et al. 2014; and numerical simulations: Bekki 2011; Diaz & Bekki 2012; Besla, Hernquist & Loeb 2013). Thirdly, because of their significant gas reservoirs (e.g. Stanimirovic et al. 2000; Kemper et al. 2010; Bolatto et al. 2011; Matsuura, Woods & Owen 2013). Indeed various gaseous features were created by the interactions of the MCs with each other and with the Milky Way (MW) such as the Magellanic Bridge (MB; Hindman, Kerr & McGee 1963), a common envelope of H i spanning ∼13 kpc in which both the Large Magellanic Cloud (LMC) and the Small Magellanic Cloud (SMC) are embedded; the Magellanic Stream, a pure gaseous stream trailing the MCs as they orbit the MW (Mathewson, Cleary & Murray 1974; Nidever et al. 2010) connecting both of the MCs with the MW, and the Leading Arm, a mass of high-velocity clouds leading the MCs (Putman 2000).

Since its discovery, the MB was believed to have a tidal origin, forming during an encounter between the MCs (Mathewson 1985). The SMC Wing – the eastern protuberance of the SMC main body – extends towards the LMC into the Bridge (Irwin, Demers & Kunkel 1990) indicating that this tidal interaction could have formed a disc around the SMC that was later on stripped to form the MB (Grondin, Demers & Kunkel 1992). Part of the strong observational evidence of the tidal stripping scenario includes the following.

• Carbon stars also in the MB area (e.g. Kunkel, Demers & Irwin 2000).

• Intermediate-age stellar populations in the Southern SMC (Noël & Gallart 2007).

• Kinematically peculiar stars in the LMC that might be evidence of being captured from the SMC (Olsen et al. 2011).

• Possible evidence of extra-tidal stars at ∼10| $_{.}^{\circ}$|6 from the SMC centre (Nidever et al. 2011).

• Old stars found in the MB (Bagheri, Cioni & Napiwotzki 2013).

• The red clump stars found in the OGLE IV fields¹ (Skowron et al. 2014).

• The stellar structure in the eastern SMC identified by Nidever et al. (2013). These authors suggest that this is a stellar counterpart of the H i MB that was tidally stripped from the SMC around 200 Myr ago (during a close encounter with the LMC). This implies an increment in the self-lensing of the SMC stars, in line with the tidal origin for the microlensing events reported towards the LMC by Besla et al. (2013).

• The intermediate-age stars in the MAGIC I (Noël et al. 2013a) in a field closer to the SMC make up 28 per cent of all stars in that region. These are not present in fields in a direction pointing away from the LMC. This provides potential evidence that these intermediate-age stars could have been tidally stripped from the SMC.

With the advent of the new STEP² survey (Ripepi et al. 2014) that will cover the main body of the SMC (32 deg²), the MB (30 deg²) and a small part of the Magellanic Stream (2 deg²), evidence will likely continue to mount.

In spite of the above striking evidence, there still remain some results at odds with the tidal scenario. Demers & Battinelli (1998) carried out the first extensive study of the Inter-Cloud stellar population finding only young stars (∼10–25 Myr old). A similar result was also found by Harris (2007) who studied twelve fields in the MB finding only old stars east of α ∼ 03^h 18′ and δ ∼ −74° and only young stars in their Inter-Cloud fields along the MB's main ridge-line, with no evidence of an intermediate-age component. As they did not find evidence of tidally stripped stars, Harris (2007) concluded that the material stripped from the MCs into the MB must be almost purely gaseous, while any young stars must have formed in situ in the MB.

With the purpose of shedding light on the stellar content of the MCs’ Inter-Cloud area, we have initiated the MAGIC (MAGellanic Inter-Cloud) project, aimed at disentangling the population age and distribution in the Inter-Cloud region as well as probing the faint outskirts of the MCs in the direction of the Bridge. As stated above, in MAGIC I (Noël et al. 2013a) we performed a quantitative colour–magnitude diagram (CMD) fitting analysis of two large Inter-Cloud fields, finding that some ∼28 per cent of the stars in one of the fields have ages between ∼1 and 10 Gyr old (intermediate-age population). Harris (2007) likely missed this intermediate-age population because their CMDs were substantially shallower than the MAGIC CMDs that reach the oldest MS turn-offs, of key importance for spotting intermediate-age stars (see e.g. Noël 2008). The results from MAGIC I provide more evidence that these intermediate-age stars could have been stripped from the SMC, supporting the tidal scenario.

In this paper, we focus on the stellar content of 10 fields strategically located perpendicular to the MAGIC I field where intermediate-age population was found represented by a purple triangle in Fig. 1 (field B2, located at ∼8 kpc from the SMC centre in the Inter-Cloud region). Five of the fields are radially away from the MB towards the north and five are radially away from the MB towards the south of field B2 in MAGIC I. This arrangement allows us to differentially compare stellar populations inside the Bridge region with directly neighbouring fields, allowing us to separate distant SMC stars from those possibly stripped from the SMC's interior. Our goal here is to probe the outskirts of the SMC in the MB direction.

This paper is organized as follows. In Section 2, we describe the observations, data reduction and photometry. The analysis and results are discussed in Section 3. Finally, in Section 4 we discuss our results and present our conclusions.

OBSERVATIONS, DATA REDUCTION AND PHOTOMETRY

Following our experience with previous observations for MAGIC I (Noël et al. 2013a), we obtained B- and R-band images of 10 fields between the nights of 2011 December 18–26 using the Wide Field Imager (WFI) on the 2.2 m telescope in La Silla. The WFI consists of eight chips covering a field of view of approximately 30 arcmin × 30 arcmin and has a pixel scale of 0.24 arcsec pixel⁻¹. Each B-band exposure was 1200 s with 900 s for the R band. An overview of the field positions can be seen in Fig. 1 with the MAGIC II fields depicted as squares showing the Northern and Southern extensions from our base position. The Northern Arm extends radially away from the SMC, while the Southern Arm extends perpendicular from the MB (we will return to this figure later). In Table 1, we present the field name, RA/DEC centres of the observed fields, observation date, seeing estimate, filter, airmass and the 50 per cent completeness level of the data. N represent the fields located north of B2 and S those located south of B2, the increasing numbers in S (S1–S5) represent fields of increasing declination from the Bridge ridge-line while the increasing numbers in N (N1–N5) represent fields of decreasing declination. The overlaps between the MAGIC I field (B2) and the first northern field is 16.8 and 14.3 arcmin for the southern field. Each subsequent field has an overlap of approximately 2–3 arcmin. For the overlap with MAGIC I this corresponds to at 2300 stars and for the overlaps between fields we obtain an average of 780 stars to determine the photometric offsets.

The data reduction steps performed here followed the same procedure as in MAGIC I. The basic reduction was performed by the Cambridge Astronomical Survey Unit pipeline (casu; Irwin & Lewis 2001) with nightly master bias and master flat frames from adjacent nights were combined to build rolling-average master frames for each night of data. In the past, the ESO/WFI instrument has had problems related to scattered light resulting in the need to apply a zero-point correction map to the data, as per Monelli et al. (2003). With the installation of the baffle in 2006, this has dramatically reduced the scattered light incident on the camera. Attempts to measure the scattered light in WFI in 2012, using a specially designed survey 086.A-9021(A), did not recover a scattered light signature as seen by previous authors. For this reason, we have not applied any correction of this manner. The casu pipeline was then used to generate aperture photometry and to determine the astrometric solution by cross-matching with the 2MASS point source catalogue. As per MAGIC I, the astrometric solution is generally better than 0.2 arcsec. The final photometric catalogue and artificial star tests were undertaken using point spread function photometry pipeline dolphot³ (Dolphin 2000). dolphot extracts the sources and determines the magnitudes for the two bands, B and R, simultaneously. This was run on all of the fields creating a catalogue of sources per field in both filters. The output files of dolphot include detailed information for each star including the position and global solution such as: χ, sharpness, roundness and object type. For our purposes, we selected a value for χ ≥ 3.5, reasonable for good stars within relatively crowded fields, a value of (sharpness)² ≤ 0.1 for a perfectly-fit star, and object type (which ranges from 1 to 5) of 1 indicating a ‘good star’. These parameters allow us to remove spurious objects such as extended objects and cosmic rays from the photometry. Typically, this corresponds to ∼3000 objects removed per field from our area of interest in CMD space.

Artificial star tests were used to determine the photometric completeness of the observations and consisted of ∼48 300 fake stars per chip. dolphot places each artificial star iteratively into the science image and then attempts to extract each of them along with the real stars. Each success or failure of this process allows us to build up a profile of the photometric completeness. Following the procedure in MAGIC I, we fitted the ratio of the detected fake stars to total number of fake stars per magnitude bin with a logistic function and from this we were able to determine the 50 per cent completeness level of each frame. This information is then utilized in the CMD fitting routine match which will be discussed in a later section.

To calibrate the MAGIC II fields, we use the photometric solution for MAGIC I and from the overlaps between each successive field determine the correction for both the Northern and the Southern Arms. In this process, we tie the instrumental magnitudes of all fields in each arm to the N1 and S1 field, respectively, as this only requires a simple offsets between fields. From this solution, we then apply the remaining colour terms and offsets determined from the overlap with MAGIC I to correct the photometry to the same Johnson–Cousins system as MAGIC I. We utilized the Markov Chain Monte Carlo code emcee (Foreman-Mackey et al. 2013) to simultaneously calculate the best offsets in both B and R for all fields when determining their relation to N1 and S1. Similarly, when determining the colour term and offset between MAGIC I and N1/S1 we also used emcee. In both cases, we typically used over 1000 walkers with more than 5000 steps each. The solutions for the fits can be found in Table 1. The errors in the offsets, as determined using emcee, are passed on in quadrature to each successive field. The evolution of the photometric error with magnitude for both the Northern and Southern Arms are shown in Fig. 2. The spread in errors is greater for the B band as it has a much larger colour term than the R band, which is essentially zero. This has the effect of broadening the error profile. As expected though, at the bright end, the minimum error for each field is greater than the previous field. The resulting CMDs are shown in Figs 3 and 4, where the CMDs corresponding to the northern and southern fields (with respect to B2 of MAGIC I) are presented. The red dashed lines in Fig. 3 show the region used for our analysis here. In order to avoid the contaminants, such as background galaxies, at the faint end of the magnitude limit we only took the stars up to R = 22.3.

CMD fitting

The CMD fitting technique was carried out using the software package match (Dolphin 2002) as explained in Noël et al. (2013a) where the reader should refer for more details. In summary, the observed CMDs are converted into Hess diagrams and compared with synthetic CMDs of model populations from Girardi et al. (2004). Theoretical isochrones are then convolved with a model of the completeness and photometric accuracy in order to create the synthetic CMDs. We also obtained a foreground estimation based on a Galactic structure model given by match that is included in the software as an extra model population. The software uses a maximum-likelihood technique to find the best linear combination of population models, resulting in an estimate of the star formation history (SFH) and the age–metallicity relation (AMR). In order to obtain the SFHs for the Inter-Cloud fields analysed here, we used the colour range −0.5 ≤ (B − R) ≤ 1.2 and the magnitude range R ≤ 22.3, as marked by the red dashed lines in Fig. 3. In this way, we avoid the inclusion of red low mass foreground stars in the fitting. To further test the influence of unidentified background galaxies contaminating our sample in the region between 21 ≤ R ≤ 22.3, we calculated the ratio of objects in the red clump to the number of objects at the lower edge of the CMD (21 ≤ R ≤ 22.3). We find these ratios to be 10–12 per cent. This is comparable with the ratio of ∼11 per cent for the wing SMC field from Noël et al. (2007) (qj0111, see grey small diamond in Fig. 1), where the contamination is known to be very low. This, together with the constraints given by dolphot described above, assures us that any background contamination has been minimized for our data set.

We recovered the SFH and metallicity for all the fields using age bins in the range from 10 Myr to 13 Gyr and metallicities in the range −2.4≤ [Fe/H] ≤ 0 in bins of width 0.2 dex. We used a Salpeter initial mass function (IMF; since these fields contain predominately intermediate- and old-age stars, the choice of IMF is not critical as shown in de Jong et al. 2010) and assumed a 30 per cent binary fraction (see Noël et al. 2007). The fitting software then finds a linear combination of model CMDs that best fit the observed CMD. The resulting relative star formation rates (SFRs) and the [Fe/H] as a function of time for the southern and northern fields are presented in Figs 5 and 6, respectively.

RESULTS

The SFHs in the MAGIC II fields

The upper panels of Figs 5 and 6 show the SFHs for all fields. The SFR for each bin is the average of the values given by the fits at the different distances and the error bars represent the complete range of SFR values found. As in the case of the MAGIC I fields, we find here a conspicuous old population (older than ∼10 Gyr old) present in all fields. It is striking to note that the bulk of intermediate-age and young(er) stars increase as we move closer to field B2 of MAGIC I located in the vicinity of the MB main ridge-line. Very little star formation took place in the past 10 Gyr in the southernmost and northernmost fields. These results agree with the proposed numerical scenario where the MB is the result of a recent tidal interaction between the MCs and the MW (Muller & Bekki 2007). The amount of H i (as seen from Fig. 1) is similar in all the Northern Arm fields as in the southernmost field S5.

The lower panels show the AMRs. The metallicities are only plotted for age bins with significant star formation (SFR ≥ 0.2). The metallicity of each bin is the average of the fits weighted by the SFH. Error bars in the AMR represent the standard deviation. Horizontal bars in both the SFH and the AMR plots represent the width of the age bins used. The AMRs are consistent with being constant on average at [Fe/H] ∼ −0.8 in agreement with the results of MAGIC I (Noël et al. 2013a) and with the calcium triplet metallicities obtained for the outskirts of the SMC (Carrera et al. 2008) and – within the quoted uncertainties – the LMC (Carrera et al. 2011). Recently, Dobbie et al. (2014) found similar results using a factor 10 of more stars. Note that these AMRs obtained from CMD fitting should only be taken as indicative of the metallicity range.

As shown in Noël et al. (2013a), valuable quantitative information on the stellar distribution is given by the cumulative SFH that provides the fraction of stellar mass formed prior to a given time. In Figs 7 and 8, we show these cumulative SFHs for our 10 MAGIC II fields. The upper y-axis shows the redshift as a function of time⁴ with the blue dashed lines mark the time when 50 per cent of the stars were formed. Although this varies across the fields, most of the fields had 50 per cent of their stellar population before 10–12 Gyr ago.

The previously introduced Fig. 1 gives a visual picture that shows the density H i map using data from the LAB survey of Galactic H i (Kalberla et al. 2005). The colour contours denote the H i emission levels integrated over the velocity range 80 < v [km s⁻¹] < 400. Each contour represents the H i column density twice as large as the neighbour contour (Skowron, private communication). Our MAGIC II fields are overlapped and depicted as bright magenta, pink and off-white squares (these differences are explained below). MAGIC I field B2 is also shown in Fig. 1 as a purple triangle. Harris (2007) fields, Bica et al. (2008) associations and clusters in the relevant right ascension (2^h–4^h), and OGLE IV fields 102 up to 125 from Skowron et al. (2014) (without numbers for clarity purposes) are also overlapped.

Southern Arm fields

Here we concentrate on disentangling the age distribution in the Southern Arm fields S1–S5 as seen in Fig. 1. The MAGIC II fields in bright magenta indicate that we find a fair amount of intermediate-age stars there (between ∼27 and ∼48 per cent). Fields in pink show substantial amount of intermediate-age population (between ∼18 and ∼22 per cent) while off-white fields indicate that very little intermediate-age population is found there. The areas with larger amounts of intermediate-age stars coincide with the higher H i density column but they are more spread out than the H i. In these regions, there are a greater amount of young stars, stellar clusters and associations (Bica et al. 2008). As we move away from the higher H i areas, the amount of intermediate-age population diminishes until there are very few intermediate-age stars in the southernmost field S5. The latter are somehow puzzling since there are young associations found in those parts of the MB. However, our results are in agreement with fields 110, 113 and 114 from Skowron et al. (2014) (the OGLE IV fields overlapping our Southern Arm fields) who do not find young or intermediate-age component in those areas. Note that the MAGIC I field B2 presents some ∼28 per cent of intermediate-age stars.

From the cumulative SFHs shown in Fig. 7 we find that fields S1 and S2 show a significant amount of stars formed between 10 and 1 Gyr ago. The remaining stars formed in the last Gyr with a conspicuous increment (20 per cent of the stars) in the past few Myr. This recent burst is the only clear star burst associated with a pericentric passage between the Clouds (∼0.2 Gyr ago; Kallivayalil et al. 2006). The arrows in S1 and S2 in Fig. 7 show the last two pericentric passage between the LMC and the SMC ≤2.5 Gyr ago and ∼250 Myr ago, respectively, from Diaz & Bekki (2012). In their best model, Diaz & Bekki (2012) find that the SMC disc remains intact until the dynamical coupling of the MCs ∼2.4 Gyr ago, at which point the Magellanic Stream and the Leading Arm are violently torn away by the strong tidal forces of the LMC. They claim that a second tidal encounter occurring ∼250 Myr ago is the responsible for pulling the MB from the SMC disc. From our qualitative comparison, there seems to be a correlation between this last pericentric passage between the MCs but no clear connection can be made between the other enhancements in the SFR and the pericentric encounters between the MCs, in agreement with the previous results from Noël et al. (2009) and MAGIC I.

Northern Arm fields

We focus here in the age distribution in the Northern Arm fields N1–N5 from Figs 6 and 8. As in the case of the Southern Arm fields, there is a bulk of intermediate-age stars in these fields. From Fig. 8 it is clear that almost no stars younger than 1 Gyr old were formed, ruling out the in situ formation for the Northern Arm.

The density column of H i in the Northern Arm fields is considerable lower than in the Southern Arm. However, we still see intermediate-age stars indicating that the young stars follow the H i gas but the intermediate-age population is considerably more spread out.

Disentangling the age spread

We now turn our attention to the correlation between young-, intermediate- and old-age stars in the MAGIC II fields and the H i gas. Without velocity information, we cannot determine conclusively if such correlations exist, but marginalizing over the unknown stellar velocities we can ask if such correlations are at least supported by the data. In Fig. 9, we look for evidence for a correlation between the H i gas and young stars in our fields. We sum up the number of young MS stars in each pixel of the SMC gas map from Muller et al. (2003). Stepping through each velocity slice of these radio data, we multiplied the star and gas densities and summed all positive pixels. The sum of this convolution as a function of the velocity slice is shown in Fig. 9. We find three velocity peaks where the stars and gas density correlate strongly, suggesting that it is at least possible for the young stars and H i gas to be highly correlated.

Unfortunately, we cannot repeat the above analysis for our intermediate-age and old stars since these cannot be so easily identified by their colour and magnitude alone. Instead, we consider how the intermediate-age and old stars relate spatially to our young star population at the spatial resolution of each MAGIC II field. This is shown in Fig. 10. The grey-scale contour shows the percentage of young- (left), intermediate- (middle) and old-age (right) stars in each of our MAGIC II fields. These are overlaid on gas density contours taken from Skowron et al. (2014) as in Fig. 1.

Note that the young stars are clustered strongly around our MAGIC I field B2 (marked by the purple triangle in Fig. 1). By contrast, the intermediate-age stars are more extended to the north and south, while the old stars are spread with almost constant density across all fields. Thus, if the young stars correlate strongly with the H i gas then the intermediate- and old-age stars do not. We discuss the implications of this in Section 4.

Surface brightness profiles

We derived the surface brightness for the ten fields analysed here using the selection box shown in Fig. 3 as red dashed lines. The surface brightness is measured in the B band as a function of distance from the mass centre of the SMC. We integrated the flux in the CMD subtracting the flux contribution from foreground MW halo giants and metal-poor MW dwarfs and from background galaxies. In order to estimate and remove the contamination, we used the predictions from the trilegal Galaxy model (Girardi et al. 2005) scaling the number of predicted MW stars to the number observed in our fields in those parts of the CMDs that were clearly devoid of MB Inter-Cloud stars (see Gallart et al. 2004 and Noël & Gallart 2007 for more details).

We obtained two separate fluxes: the total flux from stars older than 10 Gyr old (considered the old population) and the total flux from intermediate-age stars (between 1 and 10 Gyr old). The results are shown in Fig. 11 where the squares represent the southern fields and the circles depict the northern fields. Open symbols denote old populations while filled ones denote the intermediate-age populations. Error bars represent the difference between the values of the foreground flux in the fiducial CMD area as predicted by the trilegal code and after scaling by the ratio of observed and predicted fluxes (see Noël & Gallart 2007). As seen from the figure, both profiles (from the old- and from the intermediate-age populations) are relatively flat within the errors instead of the expected exponential at inner radii from the SMC centre. This is in agreement with the density profile from Nidever et al. (2011) who found that at these remote distances the star counts profile deviates from the exponential fit, showing a ‘break’ in the population that extends up to 10| $_{.}^{\circ}$|6 from the SMC centre. Nidever et al. (2011) argue that the origin of this break population is either a tidal debris or a ‘classical’ bound stellar halo similar to the one around the MW and M31. We cannot, however, directly compare their analysis with ours since Nidever et al. (2011) deliberately removed their outer fields in the direction towards the LMC.

The most plausible explanation is that the close pericentre passage between the LMC and the SMC ∼200 Myr ago (e.g. Besla et al. 2007; Noël et al. 2009), believed to have stripped H i to form the MB, could have also tidally stripped stars of all ages from the SMC as indicated by the similar spatial distribution of the old- and intermediate-age profiles in Fig. 11. The fact that intermediate-age stars show a brighter surface brightness profile could mean that the number of these intermediate-age stars is larger than the old one or that the intermediate-age stars are brighter than the old stars.

Finally, it is important to note that the intermediate-age and old stars in Fig. 11 are similarly distributed, while in Fig. 10 they have clearly different spatial distributions. This apparent contradiction arises because in Fig. 10 we see that the young-, intermediate- and old-age stars are distributed differently with respect to the H i gas in the MB. By contrast, in Fig. 11 we consider instead how they are distributed with respect to distance from the SMC. Over the small radial range of 6.5–8.5 kpc from the SMC centre probed by our MAGIC II fields, we are unable to detect a difference in the radial profile of the intermediate- and old-age populations (within our quoted uncertainties). It is likely, however, that over large radial ranges and/or with more complete spatial coverage we would see rather different distributions for the intermediate-age and old-age stars, as we do in Fig. 11.

DISCUSSION AND CONCLUSIONS

We have presented the stellar content and distribution of a transversal strip perpendicular to the main H i ridge of the ‘Inter-Cloud’ region. The fields were chosen deliberately to allow us to trace the stellar populations of the Inter-Cloud radially out from the SMC since thus the contaminating halo Galactic stars can be removed and allowing us to probe the outskirts of the SMC in the MB direction.

We confirm here the results of MAGIC I, finding that the stellar populations and mean metallicities of the fields are similar to fields to the south and west of the SMC but at smaller radii from the SMC centre (∼2.5 kpc as in the cases of Noël et al. 2009). The similarity between the MAGIC I and MAGIC II fields to some SMC fields that lie much closer in projection to the SMC is an indication that these MB fields contain tidally stripped SMC stars.

In good agreement with MAGIC I and previous studies, we found significant intermediate-age stars in the Bridge region and young stars of a similar age to the last pericentre passage in between the clouds ∼200–250 Myr ago, as determined from their proper motions and radial velocities (see Fig. 7). These recent enhancements are likely associated with the recent pericentric passage between the clouds (Kallivayalil et al. 2006). However, it is interesting to note that we find no such prominent bump in the star formation at earlier times. This could owe to the difficulty of determining orbits backwards in time (Lux, Read & Lake 2010), or the poorer temporal resolution of the SFH backwards in time (Dolphin 2012), or could indicate that the two clouds have only just interacted for the very first time (Diaz & Bekki 2012).

We showed here for the first time that the young, intermediate-age and old stars have distinct spatial distributions with respect to the H i gas in the MB. Our key results are as follows.

• The young stars correlate well with the H i gas suggesting that they were either recently stripped from the SMC or formed in situ. Unfortunately, at our current CMD–age resolution these two options are degenerate. Based on their proper motions, the clouds likely interacted just ∼200 Myr ago (Besla et al. 2012) which is comparable to the recent star formation burst (see Fig. 7). Thus, these stars could have formed in the body of the SMC and had time to be tidally torn into the MB (along with the Bridge gas); or they could have formed in situ. We note that both possibilities could act in tandem, while both lead to young stars that are highly correlated with dense gas.

• The intermediate-age stars (1-10 Gyr old) are located mainly in the Bridge region but are more spread out than the young stars, as seen in Fig. 1.

• The intermediate-age population has very similar properties to stars located ∼2 kpc from the SMC centre, suggesting they were tidally stripped from this region.

• Finally, the old stars extend to some ∼9 kpc from the SMC supporting the idea that all galaxies have a large extended metal-poor stellar halo.

The work presented here and in MAGIC I represent a complex picture for the MB region. There appears to be young stars that correlate with H i gas; a spatially distinct intermediate-age component; and a possibly smoother old-age component. It is important to emphasize that this study shows the power of deep resolved CMDs as a tool for empirically unravelling the past history of even rather complex galactic systems like the MCs.

In a future paper, we will use stellar kinematics in order to pin down whether the gas and young stars perfectly correlate; to explore the role of ram pressure on the gas; to test whether the young stars formed in situ; and to further probe the likely tidally-stripped intermediate-age stars.

We thank the anonymous referee for their helpful comments that enhanced the manuscript. We would like to thank Jan Skowron for his invaluable help in producing Fig. 1. We also thank O. Agertz, C. Gallart, N. Martin and M. Monelli for useful discussions. NEDN would like to thank the Department of Physics at the University of Surrey. BCC thanks Gemini Observatory and the Max Planck Institute for Astronomy and the Alexander von Humbodlt Foundation for their support during this project. JIR would like to acknowledge support from STFC consolidated grant ST/M000990/1 and the MERAC foundation. RC acknowledges financial support provided by the Spanish Ministry of Economy and Competitiveness under grant AYA2010-16717.

REFERENCES